Obesity is one of the most common disorders in clinical practice and is closely associated with a number of pathological disorders, such as non—insulin-dependent diabetes, hypertension, cancer, gallbladder disease, and atherosclerosis (1). Obesity is a multifactorial disease with genetic and environmental origins (2), resulting from an imbalance between energy intake and expenditure. A large number of genetic association studies have suggested that polymorphic variants in genes that regulate energy metabolism, such as proliferator-activated receptor-γ, β3-adrenergic receptor, leptin receptor, glucocorticoid receptor, and tumor necrosis factor, may predispose individuals to become obese (3). Additionally, obesity induces insulin resistance, and a high-fat diet is the major cause of obesity, as well as insulin resistance. This suggests a close association between insulin signaling and obesity (4).
The binding of insulin to its receptor initiates signaling toward various signaling pathways, and of these, the phosphoinositide 3-kinase—Akt/protein kinase B signaling pathway seems to be crucial in the ability of insulin to regulate metabolic homeostasis and protection from cell death (5). Activation of Akt/protein kinase B results in the phosphorylation of many proteins involved in insulin-dependent cellular responses, including the regulation of glucose metabolism, protein translation, protection from cell death, and the forkhead transcription factors of the FOXO subfamily (FOXO1a, FOXO3a, and FOXO4). FOXO proteins regulate the transcription of target genes that are involved in metabolism and mediate the survival-factor function of growth factors (6). FOXO proteins, therefore, play roles in a variety of cellular functions, such as differentiation, metabolism, proliferation, and survival. FOXOs mediate the effects of insulin on the expression of genes that affect glucose metabolism, including phosphoenolpyruvate carboxykinase (7), glucose 6-phosphatase (8), and pyruvate dehydrogenase kinase-4 (9). FOXO1a promotes the transcription of genes that increase glucose production, acting in concert with the proliferator-activated receptor-γ coactivator PGC1α (10). Moreover, because diabetes increases oxidative stress through the generation of reactive oxygen species (11), it is also possible that an increase in FOXO-dependent transcription mediates the deleterious effects of hyperglycemia (so-called “glucose toxicity”). FOXOs are also involved in lipid metabolism through the regulation of mitochondrial 3-hydroxy-3-methylglutaryl-CoA synthase (12) and sterol carrier protein gene expression (13). The mRNA levels of three FOXO isoforms in rat livers were altered in response to fasting and refeeding, which suggests that the genes respond differently to nutritional and hormonal factors (14).
Despite potentially important roles in metabolic diseases, no study of these genes has yet been performed in humans [although no previous evidence exists to link them to obesity, other FOXO genes (FOXO3a and FOXO4) were also evaluated in this study as possible candidate genes for obesity, based on an assumption of similar function to that of FOXO1]. In an effort to identify genetic polymorphisms in potential candidate genes that may be associated with BMI, we sequenced all exons (including 5′ and 3′ untranslated regions), exon—intron boundaries (±50 bp), and promoter regions (∼1.5 kb) from 24 unrelated Korean DNA samples, and examined the association with BMI in a Korean population (n = 734).
Sixteen polymorphisms were identified in FOXO genes: three in FOXO1a, seven in FOXO3a, and six in FOXO4. The locations and allele frequencies of identified polymorphic sites are shown in Figure 1. By pairwise linkage analysis with 24 Korean DNA samples, which were used for direct sequencing, we found that three sets of single nucleotide polymorphism (SNPs)1 in FOXO3a (c.−343–1852T>C: c.−343–1565G>A, c.−343–1852T>C: c.621 + 101A>T, and c.1100T>C [p.L367P]: c.1113C>T [p.D371 days]) were in absolute linkage disequilibrium (LD) (D′ = 1 and r2 = 1; Figure 1).
Among the 16 identified polymorphisms, seven sites (FOXO1a_c.2074T>C, FOXO1a_c.3222G>A, FOXO3a_c.−343–1582C>T, FOXO3a_c.1100T>C [p.L367P], FOXO4_c.144G>C [p.Q48H], FOXO4_c.1497 + 41T>G, and FOXO4_c.2808C>T) were selected for larger-scale genotyping based on their locations (SNPs in exons were given preference), LDs (only one SNP if there were absolute LDs [r2 = 1]), and frequencies (>0.05) among 24 individuals.
The minor allele frequencies of the polymorphisms are shown in Figure 1 and Supplemental Table 1 (available at the Obesity web site, www.obesityresearch.org). Genotype distributions of all loci, except for SNPs in FOXO4 on chromosome X, were in Hardy-Weinberg equilibrium (p > 0.05; Supplemental Table 1). Among the haplotypes identified in the Korean population, only those with frequencies >0.05 (Figure 1) were used for further haplotype association analysis. FOXO1a-ht2, FOXO1a-ht3, and FOXO4-ht3 were not used because they are equivalent with single SNPs (tagged by single SNPs; see Fig. 1) based on the results of larger-scale genotyping (n = 734).
Associations of FOXO gene polymorphisms with BMI were analyzed using multiple regression, adjusting for age and sex as covariates. One promoter SNP in the 5′ flanking region of FOXO3a showed significant association with BMI (Table 1), e.g., lowest BMI (23.3 ± 2.69 kg/m2) in individuals who were carrying T/T, intermediate BMI (26.6 ± 3.14 kg/m2) in heterozygous individuals (C/T), and highest BMI (27.2 ± 3.47 kg/m2) in individuals who were homozygous for the major allele (C/C; p = 0.01).
|Gene||Locus||Position||C/C*||C/R*||R/R*||p||Power (effect size)|
|FOXO1a||c.2074T>C||3′ untranslated region||561 (27.2 ± 3.40)||173 (27.0 ± 3.54)||12 (25.3 ± 3.73)||0.42||0.472 (0.55)|
|c.3222G>A||3′ untranslated region||331 (27.0 ± 3.40)||317 (27.1 ± 3.36)||94 (27.1 ± 3.86)||0.77||0.057 (0.03)|
|ht1||349 (26.9 ± 3.34)||213 (27.3 ± 3.31)||171 (27.1 ± 3.79)||0.84||0.095 (0.12)|
|FOXO3a||c.−343–1582C>T||Promoter||668 (27.2 ± 3.47)||77 (26.6 ± 3.14)||5 (23.3 ± 2.69)||0.01||0.713 (1.13)|
|c.1100T>C (p.L367P)||Exon2||326 (27.1 ± 3.52)||318 (27.1 ± 3.37)||89 (26.9 ± 3.57)||0.77||0.077 (0.06)|
|ht1||317 (27.0 ± 3.33)||280 (27.3 ± 3.58)||119 (26.7 ± 3.56)||0.18||0.128 (0.17)|
|ht2||324 (27.1 ± 3.53)||312 (27.1 ± 3.38)||80 (27.0 ± 3.58)||0.88||0.056 (0.03)|
|FOXO4||c.144G>C (p. Q48H)||Exon1||716 (27.0 ± 3.45)||5 (27.1 ± 4.37)||4 (27.8 ± 2.75)||0.70||0.075 (0.23)|
|c.1497 + 41T>G||Intron1||440 (27.1 ± 3.39)||121 (26.6 ± 3.96)||160 (27.4 ± 3.16)||0.53||0.157 (0.23)|
|c.2808C>T||3′ untranslated region||645 (27.1 ± 3.51)||34 (26.8 ± 3.20)||42 (27.5 ± 2.80)||0.74||0.113 (0.20)|
|ht1||162 (27.4 ± 3.15)||120 (26.6 ± 4.02)||423 (27.0 ± 3.38)||0.52||0.242 (0.20)|
|ht2||501 (27.1 ± 3.31)||91 (26.4 ± 4.24)||113 (27.3 ± 3.32)||0.86||0.086 (0.23)|
The promoter SNP, c.−343–1582C>T, was analyzed for a putative transcription factor-binding site using TFSEARCH software (Searching Transcription Factor Binding Sites V1.3, Parallel Application TRC Laboratory, RWCP, Tokyo, Japan; putative score > 0.90). Results indicated that c.−343–1582C>T was not located in any putative mammalian transcription factor-binding sites.
In contrast to Caenorhabditis elegans, mammals have three main FOXO isoforms, and some share common downstream transcriptional targets in vitro. However, the expressional pattern of each FOXO gene is remarkably different in tissues (15), and it changes under a variety of conditions (14). Furthermore, the phenotypes of mice completely lacking each of the FOXO genes are quite diverse, which suggests that the physiological roles of FOXO genes are quite diverse in mammals (16, 17). Now that we have determined that the c.−343–1582C>T polymorphism of FOXO3a shows significant association with BMI, further study is needed to investigate the roles of the c.−343–1582C>T polymorphism of the FOXO3a gene in glucose and lipid metabolism.
In this study, we did not find the effect of the FOXO3a genetic polymorphism on BMI to be very powerful. Therefore, it may be argued that Bonferroni correction should be applied to the obtained p values. If the Bonferroni correction were strictly adopted, the associated p value of c.−343–1582C>T would not retain significance (the threshold of significance would be 0.007; seven polymorphisms). However, although there is a chance of type 1 error caused by multiple comparisons, and considering that the comparisons were not wholly independent of one another because of tight LDs among SNPs, the significance of the associations might still be noticeable. Moreover, through genome-wide linkage studies (18, 19), chromosome 6q21 (where FOXO3a lies) has been identified as a region that may be linked to obesity.
When considering the mild strength of association and the limitation of our study, e.g., the analysis was based on blood samples collected from study participants at a single point in time, representing an isolated snapshot of biological activity, further biological and/or functional evidence would be needed to confirm the association of FOXO3a polymorphisms with BMI that has been suggested in this study.
In summary, we identified 16 polymorphic sites in the FOXO genes and have examined the genetic association with BMI in a Korean population (n = 734). Statistical analyses revealed that one promoter SNP in the FOXO3a gene was associated with BMI. The polymorphism/association information identified in this study would be useful for further genetic studies of other metabolic diseases.