1 Introduction

The prevalence of obesity is rising at an alarming rate worldwide, including Malaysia, where its 4th National Health and Morbidity Survey (NHMS IV) in 2011 reported the prevalence of overweight and obesity at 29.4% and 15.1%, respectively (Institute of Public Health Malaysia, 2011). Obesity is a multifactorial disorder which involves interplay between genetic and environmental factors (Kopelman 2000). The current obesity pandemic is the result of an obesogenic environment with energy-dense foods and lack of physical activity in individuals who have a genetic susceptibility towards developing obesity (Hurt et al. 2011). There are more than 120 candidate genes which have been linked with obesity-related phenotypes in the 2005 version of the human obesity gene map (Rankinen et al. 2006), and the uncoupling protein (UCPs) genes are a family of them.

UCPs are approximately 32 kDa proteins that belong to a family of mitochondrial carrier proteins present in the inner mitochondrial membrane (Yonezawa et al. 2009). UCPs dissipate the proton gradient by allowing the re-entry of protons into the mitochondrial matrix during oxidative ATP generation, resulting in the uncoupling of the respiratory chain and heat production (Aquila et al. 1985). The human UCP2 gene is mapped to chromosome 11q13 and the protein has the widest tissue expression of all UCPs – in white adipose tissue, skeletal muscle, spleen, lung, pancreatic β-cells, and isolated macrophages (Fleury et al. 1997; Gimeno et al. 1997; Arsenijevic et al. 2000). UCP2 plays a role in regulating ATP synthesis, generation of reactive oxygen species and glucose-stimulated insulin secretion in β-cells (Ricquier and Bouillaud, 2000).

Several gene polymorphisms of UCP2 have been reported in human studies and the association between UCP2 locus and vulnerability for obesity and type 2 diabetes have been investigated with specific attention being paid to -866G/A (rs659336) polymorphism in the promoter region; Ala55Val (C/T; rs660339) polymorphism in exon 4 and 45-bp insertion/deletion (I/D variant in the 3′-untranslated region of exon 8 (reviewed in Brondani et al. 2014). These three UCP2 common polymorphisms have been variably associated with altered body mass index (BMI), changes in energy expenditure, and maintenance of body weight after overfeeding (Dalgaard and Pederson, 2001; Esterbauer et al. 2001). Previous studies showed that carriers of UCP2 45-bp I allele had significantly higher BMI and risk of developing obesity compared to D allele (Cassell et al. 1999; Evans et al. 2000; Marti et al. 2004; Lee et al. 2008; Brondani et al. 2014).

We previously performed a pilot study investigating the association of UCP2 45-bp I/D polymorphism with overweight among university students who were majority of the Chinese ethnicity (n =256), and found no association (Yiew et al. 2010). Due to the limitations of the study (small sample size, lack of other ethnic groups, lack of obese subjects), we expanded the study to a larger and more age- and ethnically-diverse sample population, in multiple cohorts of urban (Klang Valley) and sub-urban (Kampar, Perak) dwellings in Malaysia. Therefore, this study aimed to determine the prevalence of the UCP2 45-bp I/D polymorphism and its possible association with obesity (assessed by BMI), overall adiposity (assessed by total body fat percentage, TBF) and central adiposity (assessed by waist-to-hip ratio, WHR) in a representative sample of the multi-ethnic Malaysian population.

2 Materials and methods

2.1 Subjects

A convenience sampling method was adopted for this study. Questionnaire and sample collection was carried out among unrelated and non-overlapping 926 subjects comprising four cohorts: (1) 256 students of Universiti Tunku Abdul Rahman (UTAR) and Kolej Tunku Abdul Rahman (KTAR) at their Setapak, Kuala Lumpur campuses were recruited from Feb – Apr 2009, as described in our previous study (Yiew et al. 2010); (2) 180 UTAR Setapak Campus students and residents of Setapak and Petaling Jaya were recruited from Oct 2009 – Feb 2010 [83 M/97 F; 43 Malays(Y)/84 ethnic Chinese(C)/53 ethnic Indians (I); mean age 26.27 ±11.93 years]; (3) 194 patrons of the Kampar Health Clinic were recruited from June – Dec 2011 (75 M/119 F; 59 Y/97 C/38 I; mean age 54.29 ±13.61 years); (4) 296 UTAR Perak Campus students in Kampar were recruited from Jan – Apr 2013 (118 M/178 F; 250 C/46 I; mean age 21.34 ±1.57 years). The ethnicities of the subjects were self-identified. All subjects were pooled together for data analysis. This study has received ethical approvals either from the UTAR Scientific and Ethical Review Committee (SERC) or the Medical Research and Ethnics Committee, Ministry of Health Malaysia (NMRR-09-826-4266). All subjects signed informed consent forms, and the study was conducted in accordance with the Declaration of Helsinki (amended in Brazil, 2013).

2.2 Questionnaire and anthropometric measurements

Clinical and anthropometric measurements namely systolic blood pressure (SBP), diastolic blood pressure (DBP), pulse rate, waist circumference (WC), hip circumference (HC), waist-to-hip ratio (WHR), weight, height, body mass index (BMI) and total body fat (TBF) were measured as described in our previous studies (Yiew et al. 2010; Chan et al. 2011). The cut-off points for obesity, overall adiposity (TBF) and central adiposity (WHR) were ≥25 kg/m2 (WHO/IOTF/IASO, 2000), 20% (males) or 30% (females) (Omron, n.d.) and 0.90 (males) or 0.85 (females)(WHO, 2011), respectively.

2.3 DNA extraction and genotyping

The DNA extraction from mouthwash samples and Polymerase Chain Reaction- restriction fragment length polymorphism (PCR-RFLP) for genotyping of the UCP2 45-bp ID polymorphism was carried out according to our previous study (Yiew et al. 2010).

2.4 Statistical analysis

The IBM SPSS Statistics software was used to analyze the data of the study. Allelic frequencies were estimated by gene counting and the distribution of genotypes was tested for Hardy-Weinberg equilibrium using the Chi-square (χ2) test. Data for continuous variables were presented as means ± standard deviations (SD) or adjusted means ± standard error of the mean (SEM) and as frequency for categorical variables. The normality of distributions of continuous variables was tested with the Kolmogorov-Smirnov test and variables that were not distributed normally were log-transformed prior to statistical analysis. Genotype and allele frequencies of the polymorphism were assessed for association with demographic and anthropometric classes using Pearson’s χ2 test in both overall and stratified analysis based on gender and ethnicity. Logistic regression analysis (enter method) was performed with adjustment to age and ethnicity to evaluate if the polymorphism could predict the risk of obesity and increased overall adiposity. Analysis of covariance using the univariate General Linear Model with adjustment for covariates (age and ethnicity) was performed for anthropometric measurements and blood pressures. A p-value of less than 0.05 was considered as statistically significant.

3 Results

Table 1 shows the demographic and anthropometric characteristics of the subjects. More than half of the subjects were females, where the distributions of subjects under different BMI and WHR classes were not significantly different among genders. However, there were more females from all the ethnicities, age groups and TBF classes (p<0.05).

Table 1 Demographic and anthropometric characteristics of the subjects according to gender
Full size table

The distributions of all UCP2 45-bp I/D genotypes and alleles according to ethnicities and genders are as shown in table 2, which did not deviate from the Hardy-Weinberg equilibrium. The overall minor allele frequency (MAF) for UCP2 45-bp I/D was 0.14, while according to Malay/Chinese/Indian and non-obese/obese categories, their MAFs were 0.17/0.12/0.21 and 0.13/0.17, respectively. The allele distribution of UCP2 45-bp I/D was significantly associated with ethnicity (p<0.001) – MAF of Indians being significantly higher. The genotype and allele distribution was also significantly associated with obesity status and overall adiposity (TBF) status, but not central adiposity (WHR) status. The allele distribution of the polymorphism was also not significantly different among gender (p=0.35).

Table 2 Association of UCP2 45-bp I/D polymorphism genotype and allele distribution with demographic and anthropometric classes
Full size table

Due to the high heterogeneity of the sampled subjects, their allele frequencies were also analysed separately by Pearson’s χ2 test based on gender and ethnicity (Table 3). Within males, the genotype and allele distribution was not associated with ethnicity and anthropometric classes. However, within females, significantly more Indians, obese and those with high TBF (all p<0.001) carried the I allele, indicating that the UCP2 45-bp I/D polymorphism was significantly associated with ethnicity, BMI and TBF. In ethnically-stratified analysis, only TBF class was significantly associated with the UCP2 allele among Malays (p=0.03), and with the UCP2 genotype among Chinese (p=0.02). Meanwhile, a further stratified analysis based on both gender and ethnicity (male – Malay, male – Chinese, female – Malay, etc.) revealed that only TBF was significantly associated with UCP2 allele distribution among female Chinese subjects only (χ2=7.57, p=0.01; data not shown). These results indicate that the female gender had the main effect in associating obesity and adiposity with the UCP2 45-bp I/D polymorphism and hence, analyses involving only females were conducted (tables 4 and 5).

Table 3 Stratified analysis of UCP2 45-bp I/D polymorphism genotype and allele distributions based on genders and ethnicities
Full size table
Table 4 Logistic regression analysis for the association of UCP2 45-bp I/D polymorphism with obesity and adiposity status among females
Full size table
Table 5 Adjusted means of anthropometric measurements for different genotypes and alleles of UCP2 45-bp I/D polymorphism among females
Full size table

Table 4 shows that females with ID genotype and I allele were 2.10 times and 1.87 times, respectively, more likely to be obese; and 2.34 times and 2.19 times, respectively, more likely to have high TBF/overall adiposity. However, when adjusted for age and ethnicity, this association of UCP2 45-bp I/D polymorphism with overall obesity was abolished (p=0.10) but not so for high adiposity (p=0.01). Indeed, covariate analysis of variance after controlling for age and ethnicity, also showed similar result, where females carrying ID genotype or I allele had significantly higher TBF than the rest (both p=0.006) (table 5). Interestingly, SBP, DBP and PR were also significantly different among the genotypes (all p<0.001), with DD, ID and II having the highest SBP, DBP and PR, respectively (table 5).

4 Discussion

In this study, the MAF of ethnic Chinese was quite similar to that of a previous Chinese study (Liu et al. 2012). As for ethnic Indians, their MAF was higher than other ethnic groups at 0.21; in agreement with 0.25 in a study among South Indians of Chennai, Tamil Nadu (Cassell et al. 1999). Indeed, majority of Malaysian Indians in this study comprised Tamils and Telugus with ancestries tracing back to the South Indian state (Teo et al. 2009). Therefore, the similar MAF found between both ethnic groups in Malaysia and those in China and India reflect the genetic ancestral origins based on their migration history. With regard to Malays, observation cannot be made as no published data is available for comparison. The I/I genotype was detectable at the prevalence of 1.9% in this Malaysian population, whereas Tongan (Duarte et al. 2003) and French (Otabe et al. 1998) populations reported the absence of this genotype.

A recent meta-analysis to assess the effect of the three common UCP2 polymorphisms – Ala55Val, 45-bp I/D, and -866G/A on the risk of overweight and obesity using in both Asian and European populations found that the UCP2 I allele was associated with increased BMI in Asians (Brondani et al. 2014). In this study, we found that the polymorphism was associated with obesity in the overall Malaysian population. However, stratified analysis revealed that this significance was only found among females but was further abolished after controlling for age and ethnicity. Therefore, UCP2 I/D polymorphism was not associated with obesity regardless of gender and Malay/Chinese/Indian ethnicity, similar with a previous Chinese study (Liu et al. 2012).

As BMI is just one phenotype of obesity, there is possibility that the UCP2 I/D polymorphism could have different effects on other phenotypes, such as TBF, WC, WHR, subcutaneous fat and visceral fat. Indeed, the UCP2 I/D polymorphism was associated with overall adiposity (assessed by TBF via bio-impedence), whether in the overall Malaysian population or in female-stratified analysis (even after controlling for age and ethnicity). Females carrying the ID genotype or I allele had almost two-fold risk of having higher overall adiposity, and I allele carriers had 2.05% higher TBF compared to D allele. However, central obesity (assessed by WHR) was not affected by UCP2 I/D polymorphism across gender and ethnicities. Consistently, Yanowski et al. (2000) also found that TBF (assessed via skinfold thicknesses and dual-energy X-ray absorptiometry) was significantly associated and greater in ID subjects compared to II subjects, while WHR was not significantly different African American, white, and Asian ethnicities (Yanowski et al. 2000). Also, no association between WC and waist-to-thigh ratio with this polymorphism was found among Pima Indians (Walder et al. 1998).

The lack of association of this polymorphism with BMI but with TBF may be attributed to the low BMI and high TBF paradox among Asians (including Malays, Chinese and Indians) compared to Caucasians; thus, obesity should be regarded as an excess of body fat and not as an excess of weight (increased BMI) (Deurenberg-Yap et al. 2000).

At present, there is scarce information about the effect of I/D polymorphism on UCP2 function, although its location in the 3′-untranslated region of UCP2 exon 8 suggests potential involvement in mRNA processing or in the stability of the transcript. However, it has been reported that this polymorphism had no apparent effect on UCP2 mRNA levels in the skeletal muscles of Pima Indians (Walder et al. 1998), or in the adipose tissue of Caucasian, Asians, and African Americans (Wang et al. 2004). Furthermore, the ratio of I to D mRNA expression was highly variable in the adipose tissue of subjects heterozygous for 45-bp I/D (Esterbauer et al. 2001). The findings suggested an independent role for this polymorphism in mRNA stability, possibly by just simply acting as a marker for another genetic variant in this region which affects adipogenesis and energy balance.

Limitations of the present study include the small sample size of Malay and Indian subjects, and the lack of other indigenous ethnic groups especially from East Malaysia (Sabah and Sarawak); hence the results from this study may not be fully representative of the general Malaysian population. The case–control design in this study also does not allow for a causality conclusion to be made. Also as only one polymorphism of UCP2 was evaluated in this study, it is unclear whether other UCP2 or other UCP families (UCP1, UCP3) polymorphisms in tight linkage disequilibrium with I/D polymorphism instead might have association with obesity. Therefore, population-based studies with a larger and more ethnically-diversed sample size and an investigation on the effects of environmental and lifestyle factors (like physical activity and dietary habits) are necessary in order to clarify the possible gene-environment interaction that causes inconsistent findings in different populations. UCP2 mRNA and protein amounts in both adipose tissue and skeletal muscle of persons with the three possible genotypes could also be examined to test the direct functional effect of this polymorphism.