Effects of a PPARG gene variant on obesity characteristics in Brazil

The contribution of genetic factors to the development of obesity has been widely recognized, but the identity of the genes involved has not yet been fully clarified. Variation in genes involved in adipocyte differentiation and energy metabolism is expected to have a role in the etiology of obesity. We assessed the potential association of a polymorphism in one candidate gene, peroxisome proliferator-activated receptor-gamma (PPARGγ), involved in these pathways and obesityrelated phenotypes in 335 Brazilians of European descent. All individuals included in the sample were adults. Pregnant women, as well as those individuals with secondary hyperlipidemia due to renal, liver or thyroid disease, and diabetes, were not invited to participate in the study; all other individuals were included. The gene variant PPARG Pro12Ala was studied by a PCR-based method and the association between this genetic polymorphism and obesity-related phenotypes was evaluated by analysis of covariance. Variant allele frequency was PPARG Ala12 = 0.09 which is in the same range as described for European and European-derived populations. No statistically significant differences were observed for mean total cholesterol, LDL cholesterol, HDL cholesterol, or triglyceride levels among PPARG genotypes in either gender. In the male sample, an association between the PPARG Pro12Ala variant and body mass index was detected, with male carriers of the Ala variant presenting a higher mean body mass index than wild-type homozygotes (28.3 vs 26.2 kg/m2, P = 0.037). No effect of this polymorphism was detected in women. This finding suggests that the PPARG gene has a gender-specific effect and contributes to the susceptibility to obesity in this population. Correspondence


Introduction
Human obesity is a multifactorial disease with a significant heritable component.The energy imbalance that results in human obesity occurs as a result of genetic or acquired changes in eating behavior, physical activity, and efficiency of fuel storage and metabolism (1).Therefore, genes involved in energy metabolism and storage are good candidates as determinants of susceptibility to obesity.
Peroxisome proliferator-activated receptor-gamma (PPARγ) is a transcription factor that plays a key role in activation of adipocyte differentiation and is an important modulator of gene expression in a number of specialized cell types, including adipocytes, where it acts by regulating the transcription of numerous target genes.Human PPARγ expression was first described in hematopoietic cells (2) and later also in spleen, liver, testis, skeletal muscle, and brain, in addition to fat (3).PPARγ-signaling pathways affect both cellular and systemic lipid metabolism and have links to obesity, diabetes and cardiovascular disease (4).PPARγ exists as two isoforms, γ1 and γ2, generated by alternative promoters and differential splicing of at least three different transcripts from the PPARG gene on chromosome 3p25 (3,5).PPARγ2 is the most important isoform in adipose tissue, where it is almost exclusively expressed.A C>T polymorphism located in the exon B of the PPARG gene, which codes the aminoterminal polypeptide that defines the PPARγ2 isoform, results in a Pro12Ala substitution.The Ala allele was shown to have reduced efficiency in transactivating responsive promoters ( 6) and a reduced ability to stimulate adipogenesis in response to thiazolidinedione activation (7).Nonetheless, results of studies on the association with this variant in man have been variable, both regarding the ability to detect an effect on obesity or glucose homeostasis and the direction of such effect (8).
Since the actual importance of the above gene variant to the development of obesity in human populations is a matter of controversy, the aim of the present study was to determine whether the Pro12Ala polymorphism of the PPARG gene contributes to variation in obesity-related phenotypes in a European-derived Southern Brazilian population, and to examine these results in the context of current findings related to the functions proposed for this protein.

Subjects
The population studied, anthropometric measurements and biochemical analyses have been fully described elsewhere (9).Briefly, our sample comprised 335 individuals (153 men and 182 women) with body mass index (BMI, weight in kg/(height in m) 2 ) ranging from 16.2 to 42.4 kg/m 2 (mean BMI: 26.1 kg/ m 2 ).All individuals were of European descent and resided in Porto Alegre city, capital of Brazil's southernmost state.
Data were collected regarding smoking habits, alcohol consumption, medication intake, regular engagement in leisure-time physical activity, and menopausal status.Weight, height and waist and hip circumferences were measured as previously described (9).The characteristics of the individuals investigated are shown in Table 1.To avoid the potential confounding effects of diabetes mellitus and its treatment on obesity-related phenotypes, individuals with a previous history of diabetes or fasting glucose levels over 7.0 mmol/L were excluded from the sample.All subjects signed an informed consent to participate in the study.
On the basis of the World Health Organization (10) criteria for BMI, 45.5% of the subjects were of normal weight, 36.6% were overweight and 17.9% were obese.

Biochemical and DNA analyses
Blood samples were collected from subjects after a 12-h fast.Total cholesterol, HDL cholesterol, triglycerides, and glucose

Statistical analysis
Allele frequencies were estimated by gene counting.The agreement of genotype frequencies with Hardy-Weinberg expectations was tested using the goodness of fit chisquare test.
The association of the polymorphism with quantitative variables was tested by analysis of co-variance using general linear models.Since the other biological and environmental variables assessed (cigarette smoking, leisure-time physical activity, alcohol consumption, oral contraceptive use, hormone replacement therapy) made little or no contribution to the anthropometric variables (BMI and waist circumference), only age was included in these analyses as a co-variable.For lipid levels, the co-variables were age and BMI.Triglyceride levels were log transformed prior to analysis to remove skewness.Association analyses were performed separately for each gender.All analyses were performed using the SPSS 8.0.0 statistical package.

Results
The distribution of the Pro12Ala sequence variant among the individuals studied was 84.5% Pro/Pro, 14.0% Pro/Ala, and 1.5% Ala/Ala.Genotype frequencies were as expected according to Hardy-Weinberg equilibrium.When compared with those observed in other studies, the frequency for the variant PPARG Ala-allele (0.09) was in the same range as described for European and European-derived populations (13).
The genotype and allele frequencies of the normal weight, overweight, and obese groups were compared by the chi-square test, but no statistically significant differences were observed (data not shown).
Association analyses of the PPARG Pro12Ala variant with BMI and waist circumference are presented in Table 2.In the male sample, carriers of the Ala allele presented higher mean BMI than wild-type Pro/ Pro homozygotes (28.3 vs 26.2 kg/m 2 , P = 0.037), a feature that was not observed in women.
Since PPARγ plays a key role in adipogenesis and lipogenesis and treatment with PPARγ-agonists alters plasma lipid profiles (14)(15)(16), the possible association between the Pro12Ala polymorphism and plasma lipid levels was also evaluated.However, no statistically significant differences were observed for mean total cholesterol, LDL cholesterol, HDL cholesterol, or triglyceride levels among PPARG genotypes in either gender (Table 2).

Discussion
Data regarding the association of the PPARG gene with human obesity have been conflicting.To clarify this issue, we investigated the Pro12Ala polymorphisms in a Brazilian population and searched for associations of this gene variant with obesity-related traits.Three previous studies investigated this polymorphism in the Brazilian population.Caramori et al. (17) concluded that the presence of the Ala allele may confer protection from diabetic nephropathy in patients with type 2 diabetes, whereas Tavares et al. (18) suggested that subjects with diabetes carrying the Ala12 allele could be more sensitive to insulin than those homozygous for Pro12.A third study (19) with 85 Parkateje Indians from the Brazilian Amazon region described an association between the Pro 12Ala variant and features of metabolic syndrome.Almost all individuals in their sample were classified as having normal glucose tolerance.The present study focused only on non-diabetic individuals to avoid these potential confounding effects of diabetes on BMI.
The primary effect of PPARG seems to be on body weight; at least 10 studies (13) have shown an association between the Ala allele and higher BMI or obesity, although many negative results have also been reported (13), as well as six studies associating the Ala allele and a decrease in BMI (13).Despite the fact that a number of studies may represent type I or type II errors due, for instance, to small sample size or population stratification, there are a growing number of better designed studies with at least moderate sample sizes, showing that statistical errors may not be sufficient to explain discrepancies among association studies.Furthermore, three longitudinal studies consistently suggested greater weight gain in carriers of the Ala allele (20)(21)(22).A meta-analysis concerning the effect of the Pro12Ala variant on BMI concluded that Ala12 allele carriers have a significantly higher BMI than noncarriers, but this effect is restricted to individuals with a BMI of 27 kg/m 2 or higher (23).An explanation for these inconsistencies was provided by a recently published meta-analysis that showed that the effect of the Pro12Ala polymorphism was restricted to Caucasian populations and obese subjects.In that analysis the Ala12 carriers presented significantly increased BMI (24).
In the present study, the effect of the Pro12Ala variant was restricted to males.Male-restricted associations of PPARG with obesity-related phenotypes have also been found in other studies, together with femalespecific effects (13).PPARγ has been shown to inhibit the expression of the aromatase gene, a key enzyme in estrogen biosynthesis (25).Conversely, a recent study demonstrated that estrogens can induce expression of the small heterodimer partner, a nuclear receptor that can modulate the activity of several transcription factors, including PPARα and PPARγ (26).Hence, hormonal factors may potentially modulate the action of PPARG variants and underlie these gender-specific associations.Notice, however, that not only differences in hormones and fat-distribution patterns among genders, but also sex-specific lifestyle components such as differences PPARG and obesity in Brazil www.bjournal.com.br in smoking habits, physical activity, diet or alcohol consumption may act as modifiers of the effect of gene variants in determining the risk of obesity.
Therefore, although in vitro assays suggest that the Pro12Ala variant would reduce PPARγ ability to stimulate adipogenesis, in vivo studies indicate a more complex scenario, with multiple interactions among biological and environmental factors in the determination of its biological role.For instance, Cardona et al. (27) recently demonstrated that patients with the metabolic syndrome who are Ala12 carriers are more prone to postprandial hypertriglyceridemia. Furthermore, these investigators found a significant interaction between the Pro12Ala and apolipoprotein E genotypes in the determination of this phenotype.The modulation of PPARγ activity by several co-factors (28) makes this scenario even more complex; it seems that the effect of PPARγ is highly context dependent, a feature that must be further explored in other studies.On the other hand, it is important to keep in mind that the Pro12Ala variant could be in linkage disequilibrium with an unknown variant elsewhere in the gene, which could explain the discrepancies among results of association studies in different populations.

Table 1 .
Anthropometric and lifestyle characteristics of 153 men and 182 women analyzed.
*Data are reported as mean ± SEM or as percent.*P < 0.0001 compared to men (unpaired t-test).**P < 0.0011 compared to men (chisquare test).

Table 2 .
Anthropometric measurements and lipid levels as a function of PPARG genotypes in men and women.
a Tests were performed on log-transformed values.*P = 0.037 compared to body mass index (BMI) of Pro/Pro subjects (ANCOVA).