- Email: [email protected]
Gene–Gene Interaction between PPARδ and PPARγ Is Associated with Abdominal Obesity in a Chinese Population
Available online at www.sciencedirect.com
Journal of Genetics and Genomics 39 (2012) 625e631
JGG ORIGINAL RESEARCH
GeneeGene Interaction between PPARd and PPARg Is Associated with Abdominal Obesity in a Chinese Population Yi Ding a, Zhi-Rong Guo a,*, Ming Wu b, Qiu Chen a, Hao Yu b, Wen-Shu Luo c a
Department of Epidemiology, School of Public Health, Soochow University, Suzhou 215123, China b Center for Diseases Control of Jiangsu Province, Nanjing 210009, China c Center for Disease Control of Changzhou, Changzhou 213002, China Received 22 May 2012; revised 31 July 2012; accepted 16 August 2012 Available online 2 November 2012
ABSTRACT The peroxisome proliferator-activated receptors (PPARs) -a, -d/b and -g are the ligand-activated transcription factors that function as the master regulators of glucose, fatty acid and lipoprotein metabolism, energy balance, cell proliferation and differentiation, inflammation, and atherosclerosis. The objective of the current study was to examine the main and interactive effect of seven single nucleotide polymorphisms (SNPs) of PPARd/g in contribution to abdominal obesity. A total of 820 subjects were randomly selected and no individuals were related. The selected SNPs in PPARd (rs2016520 and rs9794) and PPARg (rs10865710, rs1805192, rs709158, rs3856806, and rs4684847) were genotyped. Mean difference and 95% confident interval were calculated. Interactions were explored by the method of generalized multifactor dimensionality reduction. After adjustment for gender, age, and smoking status, it was found that the carriers of the C allele (TC þ CC) of rs2016520 were associated with a decreased risk of abdominal obesity compared to the carriers of the TT genotype (mean difference ¼ 2.63, 95% CI ¼ 3.61e1.64, P < 0.0001). A significant two-locus model (P ¼ 0.0107) involving rs2016520 and rs10865710 and a significant three-locus model (P ¼ 0.0107) involving rs2016520, rs9794, and rs1805192 were observed. Overall, the three-locus model had the highest level of testing accuracy (59.85%) and showed a better cross-validation consistency (9/10) than two-locus model. Therefore, for abdominal obesity defined by waist circumference, we chose the three-locus model as the best interaction model. In conclusion, the C allele in rs2016520 was significantly associated with a lower abdominal obesity. Moreover, an interaction among rs2016520, rs1805192, and rs9794 on incident abdominal obesity could be demonstrated. KEYWORDS: PPARs gene; Polymorphism; Abdominal obesity; Interaction
1. INTRODUCTION Obesity is a complex metabolic disorder that affects a growing number of patients worldwide (Hossain et al., 2007). The fraction of the population variation explained by genetic factors (heritability) has been considered in a large number of twin, adoption, and family studies. In general, 40%e60% of the variation in obesity-related phenotypes, such as body mass index (BMI), has been estimated to be heritable (Qi and Cho, 2008). Although in the minority of cases, a single * Corresponding author. Tel: þ86 512 6588 0079, fax: þ86 512 6588 4830. E-mail address: [email protected] (Z.-R. Guo).
mutation leads to an obese phenotype, evidence suggests that in most obese patients, it is a large number of genes that are involved (Yu et al., 2011). Abdominal adiposity measured by waist circumference (WC) is a significant predictor of morbidity (Janssen et al., 2004) and mortality (Bigaard et al., 2003), independently of BMI. Moreover, WC provides a unique indicator of body fat distribution, which can identify patients who are at increased risk for obesity-related cardiometabolic disease, above and beyond the measurement of BMI (Zhu et al., 2002). The peroxisome proliferator-activated receptors (PPARs) are orphan nuclear receptors belonging to the steroid, retinoid and thyroid hormone receptor superfamily of ligand-activated
1673-8527/$ - see front matter Copyright Ó 2012, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, and Genetics Society of China. Published by Elsevier Limited and Science Press. All rights reserved. http://dx.doi.org/10.1016/j.jgg.2012.08.005
626
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631
transcription factors. Three distinct receptor types, PPARa (NR1C1), PPARd/b (NR1C2) and PPARg (NR1C3) have been cloned and characterized. Many studies suggest that PPARd is critically involved in the regulation of lipid, lipoprotein, and glucose metabolism in multiple tissues including adipose tissue, skeletal muscle, and heart (Evans et al., 2004). PPARg is highly expressed in adipose tissue, where it plays an indispensable role in the regulation of adipocyte differentiation, lipid storage, glucose metabolism, and the transcriptional regulation of a number of genes involved in these metabolic processes (Bensinger and Tontonoz, 2008). Most importantly, PPARg is a well-recognized cellular target for the anti-diabetic drugs thiazolidinediones, which sensitize cells to insulin and improve insulin sensitivity and action (Yki-Jarvinen, 2004). Some interplay among PPAR isoforms was suggested for the repression of the PPARd- and PPARa-mediated activation of target gene expression after PPARd activation and for PPARd-dependent PPARg-activation (Shi et al., 2002; Consilvio et al., 2004). These results indicate that a functional crosstalk between PPARd and PPARg might exist concerning the control of their expression levels. In addition, some studies have examined the relationship between several polymorphisms in the PPARg isoform and obesity (Evans et al., 2001; Bosse et al., 2003; Fornage et al., 2005; Cecil et al., 2006; Rhee et al., 2006). In contrast, limited studies have focused on PPARd and the results are also inconsistent (Shin et al., 2004; Aberle et al., 2006; Gallicchio et al., 2009). Therefore, the aim of this study was to examine the main effects of both single-locus and multi-locus interactions among genetic variants of PPARd and PPARg in 820 individuals, and to test the hypothesis that single nucleotide polymorphisms (SNPs) of PPARs may contribute to the etiology of abdominal obesity independently and/or through such complex interactions.
alcohol use, family disease history, and metabolic variables. Blood samples of the 820 subjects were collected as a baseline and were used to genotype analysis. Abdominal obesity was defined as WC 85 cm for males and 80 cm for females (Chen, 2008) at the terminal. Each participant signed an informed consent form at the interview.
2. MATERIALS AND METHODS
2.4. Genotyping
2.1. Subjects
Genomic DNA from the participants was extracted from EDTA-treated whole blood using the DNA Blood Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. All SNPs were detected using fluorescent TaqMan probes (TaKaRa, Dalian, China). The restriction enzyme Taq I (TaKaRa) was used to identify and cut specific sequences, and PCR was then performed using the forward primer 50 -ACAATCACTCCTTAAATATGGTGG-30 and the reverse primer 50 -AAGTAGGGACAGACAGGACCAGTA-30 . The PCR conditions included an initial denaturation for 3 min at 95 C, followed by 40 cycles of denaturation for 10 s at 95 C, annealing for 30 s at 63 C, and extension for 30 s at 72 C. ABI Prism7000 software and the allelic discrimination procedure were used for genotyping of the seven SNPs. The probe sequence for the TaqMan fluorescence probe analysis is presented in Table S1. A 25 mL reaction mixture including 1.25 mL of 20 SNP Genotyping Assays, 12.5 mL of 2 Genotyping Master Mix, and 20 ng DNA was used. The conditions included an initial denaturation for 10 min at 95 C, followed by 50 cycles of denaturation for 15 s at 92 C, and
The study participants were recruited within the framework of the PMMJS (Prevention of Multiple Metabolic Disorders and Metabolic Syndrome in Jiangsu Province) (Zhou et al., 2010) cohort population study, which was initiated from April of 1999 to June of 2004, and 5-year follow-up data from total 4582 subjects were obtained between March of 2006 and October of 2007. A total of 4083 participants (89.11%) received follow-up examination (the baseline characteristics of participants who attended the follow-up examinations were similar to those missing examinations; P > 0.05). After excluding subjects who had experienced stroke or exhibited cardiovascular disease (n ¼ 36, 11 of whom died), type 2 diabetes (n ¼ 289, 31 of whom died), missing data (n ¼ 133) or a BMI < 18.5 kg/m2 (n ¼ 27, 2 of whom died), a total of 820 unrelated individual subjects (270 males, 550 females) were selected from the remaining 3731 cases using simple random sampling. The selected subjects were similar to those who were not selected in terms of age, sex, smoking status,
2.2. Phenotypic measurement Body weight, height, and WC were measured following standard procedures (Lohman et al., 1988), and BMI was calculated as the weight in kilograms divided by the square of the height in meters. Blood samples were collected in the morning after at least 8 h of fasting. All plasma and serum samples were frozen at 80 C until laboratory testing. Plasma glucose was measured using the oxidase enzymatic method. The concentrations of high-density lipoprotein cholesterol (HDL-C) and triglycerides (TG) were assessed enzymatically in an automatic biochemistry analyzer (Hitachi, Japan). All analyses were performed by the same laboratory. The investigation method applied at the follow-up exam was the same as that administered for the baseline. 2.3. Selection of SNPs We selected seven SNPs within the PPARd and g genes according to associations with obesity risk factors reported, based on known heterozygosity and a minor allele frequency (MAF) > 0.05, and based on location in a gene fragment that could have functional effects. PPAR polymorphism selected for analysis were rs2016520 (PPARd exon4), rs9794 (PPARd exon 9), rs3856806 (PPARg exon 6), rs10865710 (PPARg exon A2), rs4684847 (PPARg intron 3), rs709158 (PPARg intron 2), and rs1805192 (PPARg exon B).
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631
annealing and extension for 90 s at 60 C. For quality control purpose, approximately 10% of the samples were regenotyped in a blind fashion, and the same results were obtained. 2.5. Statistics The chi-square test (c2) was used to examine differences in the categorical data distribution. The clinical characteristics of the continuous variables were expressed as the mean SD and were tested using a two-sample t-test or ANOVA. A value of P < 0.05 for two-sided tests was considered statistically significant. A linear regression model was used to analyze the relationship of WC as a continuous variable to the predictors of the investigated genotypes and confounders. The mean differences were adjusted for potential confounding effects including sex, age, and smoking status. For quality control purpose, HardyeWeinberg equilibrium (HWE) test was used to detect genotype typing errors by Fisher’s exact test. 2.6. Generalized multifactor dimensionality reduction (GMDR) Previously, geneegene interaction analysis was conducted by using logistic regression model or generalized linear model, but the traditional methods are typically inadequate because of the problem called the “curse of dimensionality”. Recent combinatorial approaches, such as the multifactor dimensionality reduction (MDR) method (Ritchie et al., 2001), could resolve the problem of dimensionality. However, the existing approaches have several limitations, such as not allowing for covariates, which restrict their practical use. GMDR method (Lou et al., 2007) permits adjustment for covariates and provides a number of output parameters, including crossvalidation consistency, the testing balanced accuracy and the sign test, to assess each selected interaction. The crossvalidation consistency score is the measure of the degree of
627
consistency with which the selected interaction is identified as the best model among all possible two- to seven-locus models that potentially have an influence on the risk of abdominal obesity. The testing balanced accuracy is the measure of the degree to which the interaction accurately predicts WC value with scores between 0.50 (indicating that the model predicts no better than chance) and 1.00 (indicating perfect prediction). Finally, a sign test or a permutation test (providing empirical P values) for prediction accuracy can be used to measure the significance of an identified model. In this study, we analyzed the interaction among seven SNPs by using GMDR model after adjustment for gender, age, and smoking status. 3. RESULTS 3.1. Study population The baseline characteristics of the study sample are shown in Table 1. Males were significantly more likely than females to report being a current smoker. Additionally, males exhibited a significantly higher mean systolic blood pressure (SBP), diastolic blood pressure (DBP), and WC than females. In contrast, females were more likely to present higher HDL-C levels than males. There was no significant difference in the baseline age, BMI, total cholesterol (TC), TG, and fasting plasma glucose levels between the male and female subjects. 3.2. Characterization of the seven SNPs Seven SNPs within the PPARd/g genes are shown in Table S2. No significant deviation from HardyeWeinberg equilibrium was detected for the studied polymorphisms. 3.3. Single-locus analysis As shown in Table 2, there was a significant difference in rs2016520 allele distributions between abdominal obesity and
Table 1 The clinical and biochemical characteristics of the study participants separated by gender Total
Male
Female
P valuea
Number
820
270
550
e
Age (years)
50.05 9.41
50.70 9.74
49.73 9.23
SBP (mm Hg)
121.72 17.60
125.19 19.10
120.02 16.57
<0.001 <0.001
0.164
DBP (mm Hg)
77.76 9.65
79.95 10.54
76.68 8.99
TC (mmol/L)
4.90 1.12
5.00 1.14
4.86 1.10
0.080
HDL-C (mmol/L)
1.29 0.30
1.24 0.33
1.31 0.27
0.003
TG (mmol/L)b
1.27 (1.01e1.62)
1.27 (1.03e1.81)
1.27 (1.00e1.56)
0.051
Fasting plasma glucose (mmol/L)
5.01 0.75
5.02 0.77
5.01 0.74
0.807
BMI (kg/m2)
22.96 3.12
23.00 2.94
22.93 3.21
0.653
WC (cm)
77.62 9.05
80.84 8.51
76.05 8.88
<0.001
Current smokers (%)
199 (24.3)
168 (62.2)
31 (5.6)
<0.001
P values in bold indicate statistical significance (P < 0.05). The values shown are the means SD for age, SBP, DBP, TC, HDL-C, FPG, BMI, and WC. a P value indicates male vs. female; b Median and inter-quartile for TG.
628
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631
Table 2 Distributions of the seven SNPs alleles frequencies SNPs
Alleles
Frequencies (%)
P value
Total (n ¼ 820)
Non-abdominal obese (n ¼ 416)
Abdominal obese (n ¼ 404)
1278 (77.9)
652 (78.4)
626 (77.5)
PPARd rs9794
C G
362 (22.1)
180 (21.6)
182 (22.5)
rs2016520
T
1142 (69.6)
551 (66.2)
591 (73.1)
C
498 (30.4)
281 (33.8)
217 (26.9)
rs10865710
C
1099 (67.0)
576 (69.2)
523 (64.7)
G
541 (33.0)
256 (30.8)
285 (35.3)
rs3856806
C
1162 (70.9)
592 (71.2)
570 (70.5)
T
478 (29.1)
240 (28.8)
238 (29.5)
rs709158
A
1154 (70.4)
584 (70.2)
570 (70.5)
486 (29.6)
248 (29.8)
238 (29.5)
rs1805192
Pro
1206 (73.5)
613 (73.7)
593 (73.4)
Ala
434 (26.5)
219 (26.3)
215 (26.6)
0.664
0.002
PPARg
G
rs4684847
C
1295 (79.0)
661 (79.4)
634 (78.5)
T
345 (21.0)
171 (20.6)
174 (21.5)
0.052
0.786
0.876
0.895
0.626
P values in bold indicate significance (P < 0.05).
non-abdominal obesity participants. Additionally, the linear regression analysis in Table 3 shows the allelic distributions of the seven SNPs in subjects exhibiting normal WC versus abdominal obesity. After adjustment for gender, age, and
smoking status, it was found that the carriers of the C allele (TC þ CC, dominant model) of rs2016520 was associated with a decreased risk of abdominal obesity compared to the carriers of the TT genotype (mean difference ¼ 2.63, 95%
Table 3 Distributions of the selected SNPs genotypes and the results of a linear regression analysis for abdominal obesity SNPs
Genotype
Mean differencea (95% CI)
Genotype frequencies Abdominal obesity (n ¼ 404)
P value
Non-abdominal obesity (n ¼ 416)
PPARd rs9794
rs2016520
CC
244
254
0
CG þ GG
160
172
0.54 (1.56e0.47)
TT
211
177
0
TC þ CC
193
239
2.63 (3.61e1.64)
0.29 <0.0001
PPARg rs10865710
CC
175
192
0
CG þ GG
229
224
0.38 (0.62e1.38)
CC
206
212
0
CT þ TT
198
204
0.86 (0.13e1.85)
rs709158
AA
204
206
0
AG þ GG
200
210
0.19 (0.80e1.18)
rs1805192
PP
235
224
0
PA þ AA
169
192
0.31 (0.69e1.31)
rs4684847
CC
250
269
0
CT þ TT
154
147
0.32 (0.71e1.35)
rs3856806
a
Adjustment for gender, age, and smoking status. P values in bold indicate significance (P < 0.05).
0.46
0.09
0.71
0.54
0.54
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631
CI ¼ 3.61e1.64, P < 0.0001). However, the other six SNPs did not exhibit any significant association with abdominal obesity before or after covariates adjustment. 3.4. Interaction analysis by GMDR We employed the GMDR analysis to assess the impact of the interaction among the seven SNPs, after adjustment for covariates including gender, age, and smoking status. Table 4 summarizes the results obtained from GMDR analysis for twoto seven-locus models for abdominal obesity defined by WC value. As shown in Table 4, there were a significant two-locus model (P ¼ 0.0107) involving rs2016520 and rs10865710 and a significant three-locus model (P ¼ 0.0107) involving rs2016520, rs9794, and rs1805192. Overall, the three-locus model had the highest level of testing accuracy (59.85%) and showed a better cross-validation consistency (9/10). Therefore, for abdominal obesity defined by WC value, we chose the three-locus model as the best GMDR model, indicating a potential geneegene interaction among rs2016520, rs9794, and rs1805192. 4. DISCUSSION In the present study, we investigated the role of multiple common variants of PPARd and their interaction with PPARg polymorphisms on the risk of abdominal obesity. We found that rs2016520 showed a significant association with abdominal obesity risk. Besides, we observed a statistically significant interaction among three SNPs (rs2016520, rs9794, and rs1805192). Taken together, these findings support our hypothesis that minor gene (even the main effect approaches zero) in single-locus analysis can also contribute to the etiology of abdominal obesity due to the existence of geneegene interaction. This is, to the best of our knowledge, the first study to assess the association among a broad spectrum of genetic variants individually of the PPARd/g genes and abdominal obesity risk. Our findings are supported by the biology of PPARs and relevant observations. Unsaturated fatty acids and their derivatives are endogenous ligands of PPARs but are not isotypespecific (Michalik et al., 2006). After binding to ligand, PPARs form heterodimers with the retinoid X receptor and subsequently bind to PPAR response elements in the regulatory
629
region of target genes (Michalik et al., 2006; Ricote and Glass, 2007). The canonical view is that the selectivity in the action of PPARs in different tissues depends on the isotype-specific tissue expression, interactions with different co-regulator complexes, and the presence of different spectra of endogenous PPAR ligands (Michalik and Wahli, 2008). PPAR isotypes have no DNA binding specificity (Ricote and Glass, 2007) and may compete for one DNA site (Lemay and Hwang, 2006). Therefore, a functional crosstalk among PPARs might exist concerning the control of their expression levels. For example, Skogsberg et al. (2003) showed that there is an interaction between the polymorphisms in PPARa and PPARd on HDL-C, TC, and low-density lipoprotein cholesterol (LDL-C). Similarly, Aberle et al. (2006) have detected a significant effect of a geneegene interplay of PPARa and PPARd on body weight and BMI (P ¼ 0.02) as well as a similar trend between PPARd and PPARg genotypes (P ¼ 0.07). Recently, one study investigated the possibility of the existence of a crosstalk between PPARd and PPARg on obesity through a polygenic approach. This study was conducted by Saez et al. (2008) in a sample of 1953 individuals in relation to the absence or presence of obesity and was found no interaction between PPARd and PPARg gene variants. Unfortunately, the three aforementioned studies did not find a significant interaction between PPARd and PPARg polymorphisms on obesity. Therefore, we cannot compare our results. Single-locus analysis of the present work showed that rs2016520 influences the risk of abdominal obesity and associates with lower WC for the rare C allele. Current evidence suggests an important role of PPARd in energy storage and dissipation. In animal model, Wang et al. (2003) showed that activation of PPARd through a selective agonist reduced fatty acid storage in adipocytes and prevented development of obesity under high-fat diet. Oliver et al. (2001) suggested that treatment of obese rhesus monkeys with GW501516, a selective PPARd agonist, led to a significant improvement of metabolic traits characterized by a rise of HDL-C and a decrease of LDL-C, TC, and insulin. Robitaille et al. (2007) found an association between the C allele and traits of the metabolic syndrome including abdominal obesity. In our study, the consistent results with the above-mentioned studies were obtained. We have not found the association of the PPARg selected polymorphism with abdominal obesity in this population.
Table 4 Best geneegene interaction models identified by the GMDR Locus No.
Best combination
Cross-validation consistency
Testing accuracy
P valuea
2
rs2016520, rs10865710
5/10
0.5423
0.0107
3
rs2016520, rs9794, rs1805192
9/10
0.5985
0.0107
4
rs2016520, rs10865710, rs709158, rs4684847
3/10
0.5027
0.6230
5
rs2016520, rs9794, rs1805192, rs1805192, rs3856806
4/10
0.5438
0.1719
6
rs2016520, rs10865710, rs1805192, rs709158, rs4684847, rs3856806
8/10
0.5262
0.6230
10/10
0.5171
0.3770
7
rs2016520, rs9794, rs1805192, rs709158, rs4684847, rs3856806, rs10865710 a
Adjustment for gender, age, and smoking status.
630
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631
Some studies found that PPARg is clearly associated with a reduced risk for the development of type 2 diabetes mellitus (Altshuler et al., 2000). The rs1805192 polymorphism in PPARg has been investigated intensively. Considering the key role of PPARg in the differentiation of adipocytes and the moderate reduction of transcriptional activity associated with the A allele, the reduction of obesity could occur in its mutant genotype. However, the results are inconsistent, presumably because the improvement of insulin sensitivity is transmitted primarily through the expression of adipocytokines (Stumvoll and Haring, 2002). As seen in our study, rs1805192 may reflect a reduction in abdominal obesity by interacting with rs2016520 of PPARd which is critically involved in the regulation of lipid, lipoprotein and glucose metabolism in multiple tissues. Therefore, we cannot rule out PPARg gene as a genetic factor influencing abdominal obesity risk in our population. Several previous studies have assessed single variants in PPARs for associations with diseases, but mostly focused on only one or two variants. As abdominal obesity is a multifactorial disease probably involving multiple SNPs in a variety of genes, we evaluated a broader spectrum of PPARs variants individually and collectively, which may be more powerful than the analysis of a single locus. In addition, it is more reasonable that we use continuous instead of categorical measures of WC in this analysis. Given that WC is naturally a continuous variable, it is artificially dichotomized for analyses which unnecessarily discarded a great deal of phenotypic information. For example, a male with WC of 86 cm is treated the same as a male with WC of 130 cm. This forced dichotomization likely limits the ability to detect relevant associations, if they exist. Furthermore, using WC as a continuous variable for single SNP association tests would require linear regression methods and GMDR is capable of handling continuous phenotypes. Limitation of this study should also be considered. The main limitation of the present study is its modest sample size. This limitation would therefore weaken the power of the applied analysis in detecting significant genetic associations. Despite this, the existing sample size has an 80% power to detect associations with abdominal obesity (with 95% CI and an odds ratio of 2.0) for any SNP with an MAF of 0.05, which was applicable for the three studied SNPs (i.e., MAF in Table 2). Moreover, the present study did not collect information regarding the use of PPARd/g agonists such as thiazolidinedione, which may potentially intensify the effects of PPARd/g. Finally, the information on physical activity factors, which might interact with PPARd/g genotypes or act as potential confounding factors, was not available in our study. Possible interactions between PPARd/g genotypes and these risk factors should be thoroughly investigated in future studies. In conclusion, our study has tested the association between PPARd gene polymorphisms and abdominal obesity based on single-locus and multi-locus analyses. Our findings support the hypothesis that the SNPs from PPARd/g genes contribute to the risk of abdominal obesity independently or in an interactive manner. Independent replications in large sample sizes are
needed to confirm the role of the polymorphisms found in this study for abdominal obesity. ACKNOWLEDGEMENTS This study was supported in part by the grants from the Scientific Research Fund of National Ministry of Health (WKJ 2004-2-014) and the Priority Academic Program Development of Jiangsu Higher Education Institutions.
SUPPLEMENTARY DATA Table S1. Probe sequences for the seven SNPs used for TaqMan fluorescent probe analysis. Table S2. Information about the seven genotyped SNPs within the PPARd/g genes. Supplementary data related to this article can be found online at http://dx.doi.org/10.1016/j.jgg.2012.08.005. REFERENCES Aberle, J., Hopfer, I., Beil, F.U., Seedorf, U., 2006. Association of peroxisome proliferator-activated receptor d þ294T/C with body mass index and interaction with peroxisome proliferator-activated receptor a L162V. Int. J. Obes. (Lond.) 30, 1709e1713. Altshuler, D., Hirschhorn, J.N., Klannemark, M., Lindgren, C.M., Vohl, M.C., Nemesh, J., Lane, C.R., Schaffner, S.F., Bolk, S., Brewer, C., Tuomi, T., Gaudet, D., Hudson, T.J., Daly, M., Groop, L., Lander, E.S., 2000. The common PPARg Pro12Ala polymorphism is associated with decreased risk of type 2 diabetes. Nat. Genet. 26, 76e80. Bensinger, S.J., Tontonoz, P., 2008. Integration of metabolism and inflammation by lipid-activated nuclear receptors. Nature 454, 470e477. Bigaard, J., Tjonneland, A., Thomsen, B.L., Overvad, K., Heitmann, B.L., Sorensen, T.I., 2003. Waist circumference, BMI, smoking, and mortality in middle-aged men and women. Obes. Res. 11, 895e903. Bosse, Y., Despres, J.P., Bouchard, C., Perusse, L., Vohl, M.C., 2003. The peroxisome proliferator-activated receptor a L162V mutation is associated with reduced adiposity. Obes. Res. 11, 809e816. Cecil, J.E., Watt, P., Palmer, C.N., Hetherington, M., 2006. Energy balance and food intake: the role of PPARg gene polymorphisms. Physiol. Behav. 88, 227e233. Chen, C.M., 2008. Overview of obesity in Mainland China. Obes. Rev. 9 (Suppl. 1), 14e21. Consilvio, C., Vincent, A.M., Feldman, E.L., 2004. Neuroinflammation, COX2, and ALS e a dual role? Exp. Neurol. 187, 1e10. Evans, D., Aberle, J., Wendt, D., Wolf, A., Beisiegel, U., Mann, W.A., 2001. A polymorphism, L162V, in the peroxisome proliferator-activated receptor a (PPARa) gene is associated with lower body mass index in patients with non-insulin-dependent diabetes mellitus. J. Mol. Med. (Berl.) 79, 198e204. Evans, R.M., Barish, G.D., Wang, Y.X., 2004. PPARs and the complex journey to obesity. Nat. Med. 10, 355e361. Fornage, M., Jacobs, D.R., Steffes, M.W., Gross, M.D., Bray, M.S., Schreiner, P.J., 2005. Inverse effects of the PPAR(g)2 Pro12Ala polymorphism on measures of adiposity over 15 years in African Americans and whites. The CARDIA study. Metabolism 54, 910e917. Gallicchio, L., Chang, H.H., Christo, D.K., Thuita, L., Huang, H.Y., Strickland, P., Ruczinski, I., Clipp, S., Helzlsouer, K.J., 2009. Single nucleotide polymorphisms in obesity-related genes and all-cause and causespecific mortality: a prospective cohort study. BMC Med. Genet. 10, 103. Hossain, P., Kawar, B., El Nahas, M., 2007. Obesity and diabetes in the developing world e a growing challenge. N. Engl. J. Med. 356, 213e215.
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631 Janssen, I., Katzmarzyk, P.T., Ross, R., 2004. Waist circumference and not body mass index explains obesity-related health risk. Am. J. Clin. Nutr. 79, 379e384. Lemay, D.G., Hwang, D.H., 2006. Genome-wide identification of peroxisome proliferator response elements using integrated computational genomics. J. Lipid. Res. 47, 1583e1587. Lohman, T., Roche, A., Martorel, R. (Eds.), 1988. Standarizaton of anthropometric measurements. The Airlie (VA) Consensus Conference. Human Kinetics Publishers, Champaign, IL, The USA, pp. 20e37. Lou, X.Y., Chen, G.B., Yan, L., Ma, J.Z., Zhu, J., Elston, R.C., Li, M.D., 2007. A generalized combinatorial approach for detecting gene-by-gene and gene-by-environment interactions with application to nicotine dependence. Am. J. Hum. Genet. 80, 1125e1137. Michalik, L., Auwerx, J., Berger, J.P., Chatterjee, V.K., Glass, C.K., Gonzalez, F.J., Grimaldi, P.A., Kadowaki, T., Lazar, M.A., O’Rahilly, S., Palmer, C.N., Plutzky, J., Reddy, J.K., Spiegelman, B.M., Staels, B., Wahli, W., 2006. International Union of Pharmacology. LXI. Peroxisome proliferator-activated receptors. Pharmacol. Rev. 58, 726e741. Michalik, L., Wahli, W., 2008. PPARs mediate lipid signaling in inflammation and cancer. PPAR Res. 2008, 134059. Oliver Jr., W.R., Shenk, J.L., Snaith, M.R., Russell, C.S., Plunket, K.D., Bodkin, N.L., Lewis, M.C., Winegar, D.A., Sznaidman, M.L., Lambert, M.H., Xu, H.E., Sternbach, D.D., Kliewer, S.A., Hansen, B.C., Willson, T.M., 2001. A selective peroxisome proliferator-activated receptor d agonist promotes reverse cholesterol transport. Proc. Natl. Acad. Sci. USA 98, 5306e5311. Qi, L., Cho, Y.A., 2008. Geneeenvironment interaction and obesity. Nutr. Rev. 66, 684e694. Rhee, E.J., Oh, K.W., Lee, W.Y., Kim, S.Y., Oh, E.S., Baek, K.H., Kang, M.I., Kim, S.W., 2006. Effects of two common polymorphisms of peroxisome proliferator-activated receptor-g gene on metabolic syndrome. Arch. Med. Res. 37, 86e94. Ricote, M., Glass, C.K., 2007. PPARs and molecular mechanisms of transrepression. Biochim. Biophys. Acta. 1771, 926e935. Ritchie, M.D., Hahn, L.W., Roodi, N., Bailey, L.R., Dupont, W.D., Parl, F.F., Moore, J.H., 2001. Multifactor-dimensionality reduction reveals highorder interactions among estrogen-metabolism genes in sporadic breast cancer. Am. J. Hum. Genet. 69, 138e147. Robitaille, J., Gaudet, D., Perusse, L., Vohl, M.C., 2007. Features of the metabolic syndrome are modulated by an interaction between the
631
peroxisome proliferator-activated receptor-d-87T>C polymorphism and dietary fat in French-Canadians. Int. J. Obes. (Lond.) 31, 411e417. Saez, M.E., Grilo, A., Moron, F.J., Manzano, L., Martinez-Larrad, M.T., Gonzalez-Perez, A., Serrano-Hernando, J., Ruiz, A., Ramirez-Lorca, R., Serrano-Rios, M., 2008. Interaction between calpain 5, peroxisome proliferator-activated receptor-g and peroxisome proliferator-activated receptor-d genes: a polygenic approach to obesity. Cardiovasc. Diabetol. 7, 23. Shi, Y., Hon, M., Evans, R.M., 2002. The peroxisome proliferator-activated receptor d, an integrator of transcriptional repression and nuclear receptor signaling. Proc. Natl. Acad. Sci. USA 99, 2613e2618. Shin, H.D., Park, B.L., Kim, L.H., Jung, H.S., Cho, Y.M., Moon, M.K., Park, Y.J., Lee, H.K., Park, K.S., 2004. Genetic polymorphisms in peroxisome proliferator-activated receptor d associated with obesity. Diabetes 53, 847e851. Skogsberg, J., Kannisto, K., Cassel, T.N., Hamsten, A., Eriksson, P., Ehrenborg, E., 2003. Evidence that peroxisome proliferator-activated receptor d influences cholesterol metabolism in men. Arterioscler. Thromb. Vasc. Biol. 23, 637e643. Stumvoll, M., Haring, H., 2002. The peroxisome proliferator-activated receptor-g2 Pro12Ala polymorphism. Diabetes 51, 2341e2347. Wang, Y.X., Lee, C.H., Tiep, S., Yu, R.T., Ham, J., Kang, H., Evans, R.M., 2003. Peroxisome-proliferator-activated receptor d activates fat metabolism to prevent obesity. Cell 113, 159e170. Yki-Jarvinen, H., 2004. Thiazolidinediones. N. Engl. J. Med. 351, 1106e1118. Yu, Y.G., Zhu, H.T., Miller, D.T., Gusella, J.F., Platt, O.S., Wu, B.L., Shen, Y.P. on behalf of the Children’s Hospital Boston Genotype Phenotype Study Group, 2011. Age- and gender-dependent obesity in individuals with 16p11.2 deletion. J. Genet. Genomics 38, 403e409. Zhou, H., Guo, Z.R., Yu, L.G., Hu, X.S., Xu, B.H., Liu, H.B., Wu, M., Zhou, Z.Y., 2010. Evidence on the applicability of the ATPIII, IDF and CDS metabolic syndrome diagnostic criteria to identify CVD and T2DM in the Chinese population from a 6.3-year cohort study in mid-eastern China. Diabetes Res. Clin. Pract. 90, 319e325. Zhu, S., Wang, Z., Heshka, S., Heo, M., Faith, M.S., Heymsfield, S.B., 2002. Waist circumference and obesity-associated risk factors among whites in the third National Health and Nutrition Examination Survey: clinical action thresholds. Am. J. Clin. Nutr. 76, 743e749.
Journal of Genetics and Genomics 39 (2012) 625e631
JGG ORIGINAL RESEARCH
GeneeGene Interaction between PPARd and PPARg Is Associated with Abdominal Obesity in a Chinese Population Yi Ding a, Zhi-Rong Guo a,*, Ming Wu b, Qiu Chen a, Hao Yu b, Wen-Shu Luo c a
Department of Epidemiology, School of Public Health, Soochow University, Suzhou 215123, China b Center for Diseases Control of Jiangsu Province, Nanjing 210009, China c Center for Disease Control of Changzhou, Changzhou 213002, China Received 22 May 2012; revised 31 July 2012; accepted 16 August 2012 Available online 2 November 2012
ABSTRACT The peroxisome proliferator-activated receptors (PPARs) -a, -d/b and -g are the ligand-activated transcription factors that function as the master regulators of glucose, fatty acid and lipoprotein metabolism, energy balance, cell proliferation and differentiation, inflammation, and atherosclerosis. The objective of the current study was to examine the main and interactive effect of seven single nucleotide polymorphisms (SNPs) of PPARd/g in contribution to abdominal obesity. A total of 820 subjects were randomly selected and no individuals were related. The selected SNPs in PPARd (rs2016520 and rs9794) and PPARg (rs10865710, rs1805192, rs709158, rs3856806, and rs4684847) were genotyped. Mean difference and 95% confident interval were calculated. Interactions were explored by the method of generalized multifactor dimensionality reduction. After adjustment for gender, age, and smoking status, it was found that the carriers of the C allele (TC þ CC) of rs2016520 were associated with a decreased risk of abdominal obesity compared to the carriers of the TT genotype (mean difference ¼ 2.63, 95% CI ¼ 3.61e1.64, P < 0.0001). A significant two-locus model (P ¼ 0.0107) involving rs2016520 and rs10865710 and a significant three-locus model (P ¼ 0.0107) involving rs2016520, rs9794, and rs1805192 were observed. Overall, the three-locus model had the highest level of testing accuracy (59.85%) and showed a better cross-validation consistency (9/10) than two-locus model. Therefore, for abdominal obesity defined by waist circumference, we chose the three-locus model as the best interaction model. In conclusion, the C allele in rs2016520 was significantly associated with a lower abdominal obesity. Moreover, an interaction among rs2016520, rs1805192, and rs9794 on incident abdominal obesity could be demonstrated. KEYWORDS: PPARs gene; Polymorphism; Abdominal obesity; Interaction
1. INTRODUCTION Obesity is a complex metabolic disorder that affects a growing number of patients worldwide (Hossain et al., 2007). The fraction of the population variation explained by genetic factors (heritability) has been considered in a large number of twin, adoption, and family studies. In general, 40%e60% of the variation in obesity-related phenotypes, such as body mass index (BMI), has been estimated to be heritable (Qi and Cho, 2008). Although in the minority of cases, a single * Corresponding author. Tel: þ86 512 6588 0079, fax: þ86 512 6588 4830. E-mail address: [email protected] (Z.-R. Guo).
mutation leads to an obese phenotype, evidence suggests that in most obese patients, it is a large number of genes that are involved (Yu et al., 2011). Abdominal adiposity measured by waist circumference (WC) is a significant predictor of morbidity (Janssen et al., 2004) and mortality (Bigaard et al., 2003), independently of BMI. Moreover, WC provides a unique indicator of body fat distribution, which can identify patients who are at increased risk for obesity-related cardiometabolic disease, above and beyond the measurement of BMI (Zhu et al., 2002). The peroxisome proliferator-activated receptors (PPARs) are orphan nuclear receptors belonging to the steroid, retinoid and thyroid hormone receptor superfamily of ligand-activated
1673-8527/$ - see front matter Copyright Ó 2012, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, and Genetics Society of China. Published by Elsevier Limited and Science Press. All rights reserved. http://dx.doi.org/10.1016/j.jgg.2012.08.005
626
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631
transcription factors. Three distinct receptor types, PPARa (NR1C1), PPARd/b (NR1C2) and PPARg (NR1C3) have been cloned and characterized. Many studies suggest that PPARd is critically involved in the regulation of lipid, lipoprotein, and glucose metabolism in multiple tissues including adipose tissue, skeletal muscle, and heart (Evans et al., 2004). PPARg is highly expressed in adipose tissue, where it plays an indispensable role in the regulation of adipocyte differentiation, lipid storage, glucose metabolism, and the transcriptional regulation of a number of genes involved in these metabolic processes (Bensinger and Tontonoz, 2008). Most importantly, PPARg is a well-recognized cellular target for the anti-diabetic drugs thiazolidinediones, which sensitize cells to insulin and improve insulin sensitivity and action (Yki-Jarvinen, 2004). Some interplay among PPAR isoforms was suggested for the repression of the PPARd- and PPARa-mediated activation of target gene expression after PPARd activation and for PPARd-dependent PPARg-activation (Shi et al., 2002; Consilvio et al., 2004). These results indicate that a functional crosstalk between PPARd and PPARg might exist concerning the control of their expression levels. In addition, some studies have examined the relationship between several polymorphisms in the PPARg isoform and obesity (Evans et al., 2001; Bosse et al., 2003; Fornage et al., 2005; Cecil et al., 2006; Rhee et al., 2006). In contrast, limited studies have focused on PPARd and the results are also inconsistent (Shin et al., 2004; Aberle et al., 2006; Gallicchio et al., 2009). Therefore, the aim of this study was to examine the main effects of both single-locus and multi-locus interactions among genetic variants of PPARd and PPARg in 820 individuals, and to test the hypothesis that single nucleotide polymorphisms (SNPs) of PPARs may contribute to the etiology of abdominal obesity independently and/or through such complex interactions.
alcohol use, family disease history, and metabolic variables. Blood samples of the 820 subjects were collected as a baseline and were used to genotype analysis. Abdominal obesity was defined as WC 85 cm for males and 80 cm for females (Chen, 2008) at the terminal. Each participant signed an informed consent form at the interview.
2. MATERIALS AND METHODS
2.4. Genotyping
2.1. Subjects
Genomic DNA from the participants was extracted from EDTA-treated whole blood using the DNA Blood Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. All SNPs were detected using fluorescent TaqMan probes (TaKaRa, Dalian, China). The restriction enzyme Taq I (TaKaRa) was used to identify and cut specific sequences, and PCR was then performed using the forward primer 50 -ACAATCACTCCTTAAATATGGTGG-30 and the reverse primer 50 -AAGTAGGGACAGACAGGACCAGTA-30 . The PCR conditions included an initial denaturation for 3 min at 95 C, followed by 40 cycles of denaturation for 10 s at 95 C, annealing for 30 s at 63 C, and extension for 30 s at 72 C. ABI Prism7000 software and the allelic discrimination procedure were used for genotyping of the seven SNPs. The probe sequence for the TaqMan fluorescence probe analysis is presented in Table S1. A 25 mL reaction mixture including 1.25 mL of 20 SNP Genotyping Assays, 12.5 mL of 2 Genotyping Master Mix, and 20 ng DNA was used. The conditions included an initial denaturation for 10 min at 95 C, followed by 50 cycles of denaturation for 15 s at 92 C, and
The study participants were recruited within the framework of the PMMJS (Prevention of Multiple Metabolic Disorders and Metabolic Syndrome in Jiangsu Province) (Zhou et al., 2010) cohort population study, which was initiated from April of 1999 to June of 2004, and 5-year follow-up data from total 4582 subjects were obtained between March of 2006 and October of 2007. A total of 4083 participants (89.11%) received follow-up examination (the baseline characteristics of participants who attended the follow-up examinations were similar to those missing examinations; P > 0.05). After excluding subjects who had experienced stroke or exhibited cardiovascular disease (n ¼ 36, 11 of whom died), type 2 diabetes (n ¼ 289, 31 of whom died), missing data (n ¼ 133) or a BMI < 18.5 kg/m2 (n ¼ 27, 2 of whom died), a total of 820 unrelated individual subjects (270 males, 550 females) were selected from the remaining 3731 cases using simple random sampling. The selected subjects were similar to those who were not selected in terms of age, sex, smoking status,
2.2. Phenotypic measurement Body weight, height, and WC were measured following standard procedures (Lohman et al., 1988), and BMI was calculated as the weight in kilograms divided by the square of the height in meters. Blood samples were collected in the morning after at least 8 h of fasting. All plasma and serum samples were frozen at 80 C until laboratory testing. Plasma glucose was measured using the oxidase enzymatic method. The concentrations of high-density lipoprotein cholesterol (HDL-C) and triglycerides (TG) were assessed enzymatically in an automatic biochemistry analyzer (Hitachi, Japan). All analyses were performed by the same laboratory. The investigation method applied at the follow-up exam was the same as that administered for the baseline. 2.3. Selection of SNPs We selected seven SNPs within the PPARd and g genes according to associations with obesity risk factors reported, based on known heterozygosity and a minor allele frequency (MAF) > 0.05, and based on location in a gene fragment that could have functional effects. PPAR polymorphism selected for analysis were rs2016520 (PPARd exon4), rs9794 (PPARd exon 9), rs3856806 (PPARg exon 6), rs10865710 (PPARg exon A2), rs4684847 (PPARg intron 3), rs709158 (PPARg intron 2), and rs1805192 (PPARg exon B).
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631
annealing and extension for 90 s at 60 C. For quality control purpose, approximately 10% of the samples were regenotyped in a blind fashion, and the same results were obtained. 2.5. Statistics The chi-square test (c2) was used to examine differences in the categorical data distribution. The clinical characteristics of the continuous variables were expressed as the mean SD and were tested using a two-sample t-test or ANOVA. A value of P < 0.05 for two-sided tests was considered statistically significant. A linear regression model was used to analyze the relationship of WC as a continuous variable to the predictors of the investigated genotypes and confounders. The mean differences were adjusted for potential confounding effects including sex, age, and smoking status. For quality control purpose, HardyeWeinberg equilibrium (HWE) test was used to detect genotype typing errors by Fisher’s exact test. 2.6. Generalized multifactor dimensionality reduction (GMDR) Previously, geneegene interaction analysis was conducted by using logistic regression model or generalized linear model, but the traditional methods are typically inadequate because of the problem called the “curse of dimensionality”. Recent combinatorial approaches, such as the multifactor dimensionality reduction (MDR) method (Ritchie et al., 2001), could resolve the problem of dimensionality. However, the existing approaches have several limitations, such as not allowing for covariates, which restrict their practical use. GMDR method (Lou et al., 2007) permits adjustment for covariates and provides a number of output parameters, including crossvalidation consistency, the testing balanced accuracy and the sign test, to assess each selected interaction. The crossvalidation consistency score is the measure of the degree of
627
consistency with which the selected interaction is identified as the best model among all possible two- to seven-locus models that potentially have an influence on the risk of abdominal obesity. The testing balanced accuracy is the measure of the degree to which the interaction accurately predicts WC value with scores between 0.50 (indicating that the model predicts no better than chance) and 1.00 (indicating perfect prediction). Finally, a sign test or a permutation test (providing empirical P values) for prediction accuracy can be used to measure the significance of an identified model. In this study, we analyzed the interaction among seven SNPs by using GMDR model after adjustment for gender, age, and smoking status. 3. RESULTS 3.1. Study population The baseline characteristics of the study sample are shown in Table 1. Males were significantly more likely than females to report being a current smoker. Additionally, males exhibited a significantly higher mean systolic blood pressure (SBP), diastolic blood pressure (DBP), and WC than females. In contrast, females were more likely to present higher HDL-C levels than males. There was no significant difference in the baseline age, BMI, total cholesterol (TC), TG, and fasting plasma glucose levels between the male and female subjects. 3.2. Characterization of the seven SNPs Seven SNPs within the PPARd/g genes are shown in Table S2. No significant deviation from HardyeWeinberg equilibrium was detected for the studied polymorphisms. 3.3. Single-locus analysis As shown in Table 2, there was a significant difference in rs2016520 allele distributions between abdominal obesity and
Table 1 The clinical and biochemical characteristics of the study participants separated by gender Total
Male
Female
P valuea
Number
820
270
550
e
Age (years)
50.05 9.41
50.70 9.74
49.73 9.23
SBP (mm Hg)
121.72 17.60
125.19 19.10
120.02 16.57
<0.001 <0.001
0.164
DBP (mm Hg)
77.76 9.65
79.95 10.54
76.68 8.99
TC (mmol/L)
4.90 1.12
5.00 1.14
4.86 1.10
0.080
HDL-C (mmol/L)
1.29 0.30
1.24 0.33
1.31 0.27
0.003
TG (mmol/L)b
1.27 (1.01e1.62)
1.27 (1.03e1.81)
1.27 (1.00e1.56)
0.051
Fasting plasma glucose (mmol/L)
5.01 0.75
5.02 0.77
5.01 0.74
0.807
BMI (kg/m2)
22.96 3.12
23.00 2.94
22.93 3.21
0.653
WC (cm)
77.62 9.05
80.84 8.51
76.05 8.88
<0.001
Current smokers (%)
199 (24.3)
168 (62.2)
31 (5.6)
<0.001
P values in bold indicate statistical significance (P < 0.05). The values shown are the means SD for age, SBP, DBP, TC, HDL-C, FPG, BMI, and WC. a P value indicates male vs. female; b Median and inter-quartile for TG.
628
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631
Table 2 Distributions of the seven SNPs alleles frequencies SNPs
Alleles
Frequencies (%)
P value
Total (n ¼ 820)
Non-abdominal obese (n ¼ 416)
Abdominal obese (n ¼ 404)
1278 (77.9)
652 (78.4)
626 (77.5)
PPARd rs9794
C G
362 (22.1)
180 (21.6)
182 (22.5)
rs2016520
T
1142 (69.6)
551 (66.2)
591 (73.1)
C
498 (30.4)
281 (33.8)
217 (26.9)
rs10865710
C
1099 (67.0)
576 (69.2)
523 (64.7)
G
541 (33.0)
256 (30.8)
285 (35.3)
rs3856806
C
1162 (70.9)
592 (71.2)
570 (70.5)
T
478 (29.1)
240 (28.8)
238 (29.5)
rs709158
A
1154 (70.4)
584 (70.2)
570 (70.5)
486 (29.6)
248 (29.8)
238 (29.5)
rs1805192
Pro
1206 (73.5)
613 (73.7)
593 (73.4)
Ala
434 (26.5)
219 (26.3)
215 (26.6)
0.664
0.002
PPARg
G
rs4684847
C
1295 (79.0)
661 (79.4)
634 (78.5)
T
345 (21.0)
171 (20.6)
174 (21.5)
0.052
0.786
0.876
0.895
0.626
P values in bold indicate significance (P < 0.05).
non-abdominal obesity participants. Additionally, the linear regression analysis in Table 3 shows the allelic distributions of the seven SNPs in subjects exhibiting normal WC versus abdominal obesity. After adjustment for gender, age, and
smoking status, it was found that the carriers of the C allele (TC þ CC, dominant model) of rs2016520 was associated with a decreased risk of abdominal obesity compared to the carriers of the TT genotype (mean difference ¼ 2.63, 95%
Table 3 Distributions of the selected SNPs genotypes and the results of a linear regression analysis for abdominal obesity SNPs
Genotype
Mean differencea (95% CI)
Genotype frequencies Abdominal obesity (n ¼ 404)
P value
Non-abdominal obesity (n ¼ 416)
PPARd rs9794
rs2016520
CC
244
254
0
CG þ GG
160
172
0.54 (1.56e0.47)
TT
211
177
0
TC þ CC
193
239
2.63 (3.61e1.64)
0.29 <0.0001
PPARg rs10865710
CC
175
192
0
CG þ GG
229
224
0.38 (0.62e1.38)
CC
206
212
0
CT þ TT
198
204
0.86 (0.13e1.85)
rs709158
AA
204
206
0
AG þ GG
200
210
0.19 (0.80e1.18)
rs1805192
PP
235
224
0
PA þ AA
169
192
0.31 (0.69e1.31)
rs4684847
CC
250
269
0
CT þ TT
154
147
0.32 (0.71e1.35)
rs3856806
a
Adjustment for gender, age, and smoking status. P values in bold indicate significance (P < 0.05).
0.46
0.09
0.71
0.54
0.54
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631
CI ¼ 3.61e1.64, P < 0.0001). However, the other six SNPs did not exhibit any significant association with abdominal obesity before or after covariates adjustment. 3.4. Interaction analysis by GMDR We employed the GMDR analysis to assess the impact of the interaction among the seven SNPs, after adjustment for covariates including gender, age, and smoking status. Table 4 summarizes the results obtained from GMDR analysis for twoto seven-locus models for abdominal obesity defined by WC value. As shown in Table 4, there were a significant two-locus model (P ¼ 0.0107) involving rs2016520 and rs10865710 and a significant three-locus model (P ¼ 0.0107) involving rs2016520, rs9794, and rs1805192. Overall, the three-locus model had the highest level of testing accuracy (59.85%) and showed a better cross-validation consistency (9/10). Therefore, for abdominal obesity defined by WC value, we chose the three-locus model as the best GMDR model, indicating a potential geneegene interaction among rs2016520, rs9794, and rs1805192. 4. DISCUSSION In the present study, we investigated the role of multiple common variants of PPARd and their interaction with PPARg polymorphisms on the risk of abdominal obesity. We found that rs2016520 showed a significant association with abdominal obesity risk. Besides, we observed a statistically significant interaction among three SNPs (rs2016520, rs9794, and rs1805192). Taken together, these findings support our hypothesis that minor gene (even the main effect approaches zero) in single-locus analysis can also contribute to the etiology of abdominal obesity due to the existence of geneegene interaction. This is, to the best of our knowledge, the first study to assess the association among a broad spectrum of genetic variants individually of the PPARd/g genes and abdominal obesity risk. Our findings are supported by the biology of PPARs and relevant observations. Unsaturated fatty acids and their derivatives are endogenous ligands of PPARs but are not isotypespecific (Michalik et al., 2006). After binding to ligand, PPARs form heterodimers with the retinoid X receptor and subsequently bind to PPAR response elements in the regulatory
629
region of target genes (Michalik et al., 2006; Ricote and Glass, 2007). The canonical view is that the selectivity in the action of PPARs in different tissues depends on the isotype-specific tissue expression, interactions with different co-regulator complexes, and the presence of different spectra of endogenous PPAR ligands (Michalik and Wahli, 2008). PPAR isotypes have no DNA binding specificity (Ricote and Glass, 2007) and may compete for one DNA site (Lemay and Hwang, 2006). Therefore, a functional crosstalk among PPARs might exist concerning the control of their expression levels. For example, Skogsberg et al. (2003) showed that there is an interaction between the polymorphisms in PPARa and PPARd on HDL-C, TC, and low-density lipoprotein cholesterol (LDL-C). Similarly, Aberle et al. (2006) have detected a significant effect of a geneegene interplay of PPARa and PPARd on body weight and BMI (P ¼ 0.02) as well as a similar trend between PPARd and PPARg genotypes (P ¼ 0.07). Recently, one study investigated the possibility of the existence of a crosstalk between PPARd and PPARg on obesity through a polygenic approach. This study was conducted by Saez et al. (2008) in a sample of 1953 individuals in relation to the absence or presence of obesity and was found no interaction between PPARd and PPARg gene variants. Unfortunately, the three aforementioned studies did not find a significant interaction between PPARd and PPARg polymorphisms on obesity. Therefore, we cannot compare our results. Single-locus analysis of the present work showed that rs2016520 influences the risk of abdominal obesity and associates with lower WC for the rare C allele. Current evidence suggests an important role of PPARd in energy storage and dissipation. In animal model, Wang et al. (2003) showed that activation of PPARd through a selective agonist reduced fatty acid storage in adipocytes and prevented development of obesity under high-fat diet. Oliver et al. (2001) suggested that treatment of obese rhesus monkeys with GW501516, a selective PPARd agonist, led to a significant improvement of metabolic traits characterized by a rise of HDL-C and a decrease of LDL-C, TC, and insulin. Robitaille et al. (2007) found an association between the C allele and traits of the metabolic syndrome including abdominal obesity. In our study, the consistent results with the above-mentioned studies were obtained. We have not found the association of the PPARg selected polymorphism with abdominal obesity in this population.
Table 4 Best geneegene interaction models identified by the GMDR Locus No.
Best combination
Cross-validation consistency
Testing accuracy
P valuea
2
rs2016520, rs10865710
5/10
0.5423
0.0107
3
rs2016520, rs9794, rs1805192
9/10
0.5985
0.0107
4
rs2016520, rs10865710, rs709158, rs4684847
3/10
0.5027
0.6230
5
rs2016520, rs9794, rs1805192, rs1805192, rs3856806
4/10
0.5438
0.1719
6
rs2016520, rs10865710, rs1805192, rs709158, rs4684847, rs3856806
8/10
0.5262
0.6230
10/10
0.5171
0.3770
7
rs2016520, rs9794, rs1805192, rs709158, rs4684847, rs3856806, rs10865710 a
Adjustment for gender, age, and smoking status.
630
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631
Some studies found that PPARg is clearly associated with a reduced risk for the development of type 2 diabetes mellitus (Altshuler et al., 2000). The rs1805192 polymorphism in PPARg has been investigated intensively. Considering the key role of PPARg in the differentiation of adipocytes and the moderate reduction of transcriptional activity associated with the A allele, the reduction of obesity could occur in its mutant genotype. However, the results are inconsistent, presumably because the improvement of insulin sensitivity is transmitted primarily through the expression of adipocytokines (Stumvoll and Haring, 2002). As seen in our study, rs1805192 may reflect a reduction in abdominal obesity by interacting with rs2016520 of PPARd which is critically involved in the regulation of lipid, lipoprotein and glucose metabolism in multiple tissues. Therefore, we cannot rule out PPARg gene as a genetic factor influencing abdominal obesity risk in our population. Several previous studies have assessed single variants in PPARs for associations with diseases, but mostly focused on only one or two variants. As abdominal obesity is a multifactorial disease probably involving multiple SNPs in a variety of genes, we evaluated a broader spectrum of PPARs variants individually and collectively, which may be more powerful than the analysis of a single locus. In addition, it is more reasonable that we use continuous instead of categorical measures of WC in this analysis. Given that WC is naturally a continuous variable, it is artificially dichotomized for analyses which unnecessarily discarded a great deal of phenotypic information. For example, a male with WC of 86 cm is treated the same as a male with WC of 130 cm. This forced dichotomization likely limits the ability to detect relevant associations, if they exist. Furthermore, using WC as a continuous variable for single SNP association tests would require linear regression methods and GMDR is capable of handling continuous phenotypes. Limitation of this study should also be considered. The main limitation of the present study is its modest sample size. This limitation would therefore weaken the power of the applied analysis in detecting significant genetic associations. Despite this, the existing sample size has an 80% power to detect associations with abdominal obesity (with 95% CI and an odds ratio of 2.0) for any SNP with an MAF of 0.05, which was applicable for the three studied SNPs (i.e., MAF in Table 2). Moreover, the present study did not collect information regarding the use of PPARd/g agonists such as thiazolidinedione, which may potentially intensify the effects of PPARd/g. Finally, the information on physical activity factors, which might interact with PPARd/g genotypes or act as potential confounding factors, was not available in our study. Possible interactions between PPARd/g genotypes and these risk factors should be thoroughly investigated in future studies. In conclusion, our study has tested the association between PPARd gene polymorphisms and abdominal obesity based on single-locus and multi-locus analyses. Our findings support the hypothesis that the SNPs from PPARd/g genes contribute to the risk of abdominal obesity independently or in an interactive manner. Independent replications in large sample sizes are
needed to confirm the role of the polymorphisms found in this study for abdominal obesity. ACKNOWLEDGEMENTS This study was supported in part by the grants from the Scientific Research Fund of National Ministry of Health (WKJ 2004-2-014) and the Priority Academic Program Development of Jiangsu Higher Education Institutions.
SUPPLEMENTARY DATA Table S1. Probe sequences for the seven SNPs used for TaqMan fluorescent probe analysis. Table S2. Information about the seven genotyped SNPs within the PPARd/g genes. Supplementary data related to this article can be found online at http://dx.doi.org/10.1016/j.jgg.2012.08.005. REFERENCES Aberle, J., Hopfer, I., Beil, F.U., Seedorf, U., 2006. Association of peroxisome proliferator-activated receptor d þ294T/C with body mass index and interaction with peroxisome proliferator-activated receptor a L162V. Int. J. Obes. (Lond.) 30, 1709e1713. Altshuler, D., Hirschhorn, J.N., Klannemark, M., Lindgren, C.M., Vohl, M.C., Nemesh, J., Lane, C.R., Schaffner, S.F., Bolk, S., Brewer, C., Tuomi, T., Gaudet, D., Hudson, T.J., Daly, M., Groop, L., Lander, E.S., 2000. The common PPARg Pro12Ala polymorphism is associated with decreased risk of type 2 diabetes. Nat. Genet. 26, 76e80. Bensinger, S.J., Tontonoz, P., 2008. Integration of metabolism and inflammation by lipid-activated nuclear receptors. Nature 454, 470e477. Bigaard, J., Tjonneland, A., Thomsen, B.L., Overvad, K., Heitmann, B.L., Sorensen, T.I., 2003. Waist circumference, BMI, smoking, and mortality in middle-aged men and women. Obes. Res. 11, 895e903. Bosse, Y., Despres, J.P., Bouchard, C., Perusse, L., Vohl, M.C., 2003. The peroxisome proliferator-activated receptor a L162V mutation is associated with reduced adiposity. Obes. Res. 11, 809e816. Cecil, J.E., Watt, P., Palmer, C.N., Hetherington, M., 2006. Energy balance and food intake: the role of PPARg gene polymorphisms. Physiol. Behav. 88, 227e233. Chen, C.M., 2008. Overview of obesity in Mainland China. Obes. Rev. 9 (Suppl. 1), 14e21. Consilvio, C., Vincent, A.M., Feldman, E.L., 2004. Neuroinflammation, COX2, and ALS e a dual role? Exp. Neurol. 187, 1e10. Evans, D., Aberle, J., Wendt, D., Wolf, A., Beisiegel, U., Mann, W.A., 2001. A polymorphism, L162V, in the peroxisome proliferator-activated receptor a (PPARa) gene is associated with lower body mass index in patients with non-insulin-dependent diabetes mellitus. J. Mol. Med. (Berl.) 79, 198e204. Evans, R.M., Barish, G.D., Wang, Y.X., 2004. PPARs and the complex journey to obesity. Nat. Med. 10, 355e361. Fornage, M., Jacobs, D.R., Steffes, M.W., Gross, M.D., Bray, M.S., Schreiner, P.J., 2005. Inverse effects of the PPAR(g)2 Pro12Ala polymorphism on measures of adiposity over 15 years in African Americans and whites. The CARDIA study. Metabolism 54, 910e917. Gallicchio, L., Chang, H.H., Christo, D.K., Thuita, L., Huang, H.Y., Strickland, P., Ruczinski, I., Clipp, S., Helzlsouer, K.J., 2009. Single nucleotide polymorphisms in obesity-related genes and all-cause and causespecific mortality: a prospective cohort study. BMC Med. Genet. 10, 103. Hossain, P., Kawar, B., El Nahas, M., 2007. Obesity and diabetes in the developing world e a growing challenge. N. Engl. J. Med. 356, 213e215.
Y. Ding et al. / Journal of Genetics and Genomics 39 (2012) 625e631 Janssen, I., Katzmarzyk, P.T., Ross, R., 2004. Waist circumference and not body mass index explains obesity-related health risk. Am. J. Clin. Nutr. 79, 379e384. Lemay, D.G., Hwang, D.H., 2006. Genome-wide identification of peroxisome proliferator response elements using integrated computational genomics. J. Lipid. Res. 47, 1583e1587. Lohman, T., Roche, A., Martorel, R. (Eds.), 1988. Standarizaton of anthropometric measurements. The Airlie (VA) Consensus Conference. Human Kinetics Publishers, Champaign, IL, The USA, pp. 20e37. Lou, X.Y., Chen, G.B., Yan, L., Ma, J.Z., Zhu, J., Elston, R.C., Li, M.D., 2007. A generalized combinatorial approach for detecting gene-by-gene and gene-by-environment interactions with application to nicotine dependence. Am. J. Hum. Genet. 80, 1125e1137. Michalik, L., Auwerx, J., Berger, J.P., Chatterjee, V.K., Glass, C.K., Gonzalez, F.J., Grimaldi, P.A., Kadowaki, T., Lazar, M.A., O’Rahilly, S., Palmer, C.N., Plutzky, J., Reddy, J.K., Spiegelman, B.M., Staels, B., Wahli, W., 2006. International Union of Pharmacology. LXI. Peroxisome proliferator-activated receptors. Pharmacol. Rev. 58, 726e741. Michalik, L., Wahli, W., 2008. PPARs mediate lipid signaling in inflammation and cancer. PPAR Res. 2008, 134059. Oliver Jr., W.R., Shenk, J.L., Snaith, M.R., Russell, C.S., Plunket, K.D., Bodkin, N.L., Lewis, M.C., Winegar, D.A., Sznaidman, M.L., Lambert, M.H., Xu, H.E., Sternbach, D.D., Kliewer, S.A., Hansen, B.C., Willson, T.M., 2001. A selective peroxisome proliferator-activated receptor d agonist promotes reverse cholesterol transport. Proc. Natl. Acad. Sci. USA 98, 5306e5311. Qi, L., Cho, Y.A., 2008. Geneeenvironment interaction and obesity. Nutr. Rev. 66, 684e694. Rhee, E.J., Oh, K.W., Lee, W.Y., Kim, S.Y., Oh, E.S., Baek, K.H., Kang, M.I., Kim, S.W., 2006. Effects of two common polymorphisms of peroxisome proliferator-activated receptor-g gene on metabolic syndrome. Arch. Med. Res. 37, 86e94. Ricote, M., Glass, C.K., 2007. PPARs and molecular mechanisms of transrepression. Biochim. Biophys. Acta. 1771, 926e935. Ritchie, M.D., Hahn, L.W., Roodi, N., Bailey, L.R., Dupont, W.D., Parl, F.F., Moore, J.H., 2001. Multifactor-dimensionality reduction reveals highorder interactions among estrogen-metabolism genes in sporadic breast cancer. Am. J. Hum. Genet. 69, 138e147. Robitaille, J., Gaudet, D., Perusse, L., Vohl, M.C., 2007. Features of the metabolic syndrome are modulated by an interaction between the
631
peroxisome proliferator-activated receptor-d-87T>C polymorphism and dietary fat in French-Canadians. Int. J. Obes. (Lond.) 31, 411e417. Saez, M.E., Grilo, A., Moron, F.J., Manzano, L., Martinez-Larrad, M.T., Gonzalez-Perez, A., Serrano-Hernando, J., Ruiz, A., Ramirez-Lorca, R., Serrano-Rios, M., 2008. Interaction between calpain 5, peroxisome proliferator-activated receptor-g and peroxisome proliferator-activated receptor-d genes: a polygenic approach to obesity. Cardiovasc. Diabetol. 7, 23. Shi, Y., Hon, M., Evans, R.M., 2002. The peroxisome proliferator-activated receptor d, an integrator of transcriptional repression and nuclear receptor signaling. Proc. Natl. Acad. Sci. USA 99, 2613e2618. Shin, H.D., Park, B.L., Kim, L.H., Jung, H.S., Cho, Y.M., Moon, M.K., Park, Y.J., Lee, H.K., Park, K.S., 2004. Genetic polymorphisms in peroxisome proliferator-activated receptor d associated with obesity. Diabetes 53, 847e851. Skogsberg, J., Kannisto, K., Cassel, T.N., Hamsten, A., Eriksson, P., Ehrenborg, E., 2003. Evidence that peroxisome proliferator-activated receptor d influences cholesterol metabolism in men. Arterioscler. Thromb. Vasc. Biol. 23, 637e643. Stumvoll, M., Haring, H., 2002. The peroxisome proliferator-activated receptor-g2 Pro12Ala polymorphism. Diabetes 51, 2341e2347. Wang, Y.X., Lee, C.H., Tiep, S., Yu, R.T., Ham, J., Kang, H., Evans, R.M., 2003. Peroxisome-proliferator-activated receptor d activates fat metabolism to prevent obesity. Cell 113, 159e170. Yki-Jarvinen, H., 2004. Thiazolidinediones. N. Engl. J. Med. 351, 1106e1118. Yu, Y.G., Zhu, H.T., Miller, D.T., Gusella, J.F., Platt, O.S., Wu, B.L., Shen, Y.P. on behalf of the Children’s Hospital Boston Genotype Phenotype Study Group, 2011. Age- and gender-dependent obesity in individuals with 16p11.2 deletion. J. Genet. Genomics 38, 403e409. Zhou, H., Guo, Z.R., Yu, L.G., Hu, X.S., Xu, B.H., Liu, H.B., Wu, M., Zhou, Z.Y., 2010. Evidence on the applicability of the ATPIII, IDF and CDS metabolic syndrome diagnostic criteria to identify CVD and T2DM in the Chinese population from a 6.3-year cohort study in mid-eastern China. Diabetes Res. Clin. Pract. 90, 319e325. Zhu, S., Wang, Z., Heshka, S., Heo, M., Faith, M.S., Heymsfield, S.B., 2002. Waist circumference and obesity-associated risk factors among whites in the third National Health and Nutrition Examination Survey: clinical action thresholds. Am. J. Clin. Nutr. 76, 743e749.