Genetic Variation in the Human Winged Helix/Forkhead Transcription Factor Gene FOXC2 in Pima Indians
- From the Department of Health and Human Services, Phoenix Epidemiology and Clinical Research Branch, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Phoenix, Arizona
FOXC2 is a winged helix gene that has been shown to counteract obesity, hypertriglyceridemia, and diet-induced insulin resistance in rodents. Therefore, FOXC2 was analyzed as a candidate gene for susceptibility to type 2 diabetes in Pima Indians. Four variants were identified by sequencing the coding region, as well as 638 bp of the 5′ region and 300 bp of the 3′ region of the gene. Two single nucleotide polymorphisms (SNPs) were found in the putative promoter region, a C-512T transition and a G-350T. In addition, two SNPs were found in the 3′ region, a C1548T and a C1702T. The G-350T and the C1702T variants were in complete linkage disequilibrium, and the C1548T variant was relatively rare; therefore, only the C-512T and G-350T variants were additionally genotyped in 937 full-blooded Pima Indians. Neither of these polymorphisms were associated with type 2 diabetes; however, the C-512T variant was associated with BMI (P = 0.03) and percentage of body fat (P = 0.02) in male and female Pima subjects, as well as with basal glucose turnover and fasting plasma triglycerides in women. Our data indicate that variation in FOXC2 may have a minor role in body weight control and seems to be involved in the regulation of basal glucose turnover and plasma triglyceride levels in women, but this gene does not significantly contribute to the etiology of type 2 diabetes in Pima Indians.
The forkhead box C2 (FOXC2; also known as homologue-like 14 or mesenchyme forkhead 1) is a winged helix transcription factor whose clinical relevance was first recognized in the context of hereditary lymphedema. Several loss-of-function mutations were initially identified in families with lymphedema distichiasis (1), a condition characterized not only by lymphedema of the limbs but also by double rows of eyelashes (distichiasis), cardiac defects, cleft palate, extradural cysts, and photophobia, illustrating the developmental pleiotropy of this transcription factor.
The role of FOXC2 as a key regulator of adipocyte metabolism has recently been described (2). In mice overexpressing FOXC2 in adipocytes, the intra-abdominal white adipose tissue depot was reduced and had acquired a histology similar to brown adipose tissue, whereas interscapular brown adipose tissue was hypertrophic. Increased adipocyte FOXC2 expression had a pleiotropic effect on the expression of genes influencing cellular differentiation and metabolism, insulin action, β-adrenergic sensitivity, and intracellular signaling. The net effect of these FOXC2-related alterations was consistent with protection against obesity. Furthermore, increased FOXC2 expression induced by a high-fat diet seemed to counteract most of the symptoms associated with obesity, including hypertriglyceridemia and diet-induced insulin resistance, suggesting a protective effect also against type 2 diabetes. These findings in mice suggest that FOXC2 is a plausible candidate gene for obesity, insulin resistance, and type 2 diabetes in humans.
The prevalence of obesity and type 2 diabetes in the Pima Indian population is among the highest in the world. Both of these diseases have a substantial genetic component (3). To determine whether variation in the FOXC2 contributes to the genetic basis of either type 2 diabetes or obesity in humans, we screened the gene for prevalent and functionally relevant variants and genotyped two informative single nucleotide polymorphisms (SNPs) in a large Pima cohort for association analyses.
The single exon of the FOXC2 gene, as well as an additional 638 bp of the flanking 5′ region and 300 bp of the 3′ region, was sequenced in 24 DNA samples from full-blooded Pima Indians. Four SNPs were identified (Fig. 1). Two SNPs were found in the putative promoter region, a C-512T transition (allelic frequency T = 23%) and a G-350T (allelic frequency T = 28%). In addition, two SNPs were found in the 3′ region, a C1548T (allelic frequency T = 4.5%) and a C1702T (allelic frequency T = 28%). The G-350T and the C1702T variants were in complete linkage disequilibrium, and the C1548T variant was rare; therefore, only the C-512T and the G-350T variants were additionally genotyped in 937 Pima Indians. The genotype distributions for all SNPs were consistent with Hardy-Weinberg equilibrium (P > 0.13 vs. expected frequency, χ2 test).
There was no association of either the C-512T or the G-350T with type 2 diabetes in 937 full-blooded Pima Indians (611 diabetic and 326 nondiabetic subjects), which is consistent with a recent Japanese study (4). However,the C-512T SNP was associated with BMI in 644 Pimas with normal glucose tolerance who were at least 18 years of age. Individuals homozygous for the −512T allele (T/T) had a lower BMI than the homozygous C/C and heterozygous C/T individuals (35.4 ± 0.4 kg/m2 in C/C, 35.9 ± 0.5 kg/m2 in CT, and 32.4 ± 0.9 kg/m2 in TT, P = 0.03 or P = 0.009 under a recessive model, adjusted for age, sex, date of birth, and family membership). The association with BMI was substantiated by an association with percent body fat (P = 0.02, adjusted for age, sex, and family membership) in 215 Pima Indians with normal glucose tolerance who had undergone further metabolic testing to exam body composition (Table 1). However, the C-512T variant was not associated with an insulin secretory defect or with insulin resistance—metabolic characteristics previously determined to be risk factors for type 2 diabetes in this population. To estimate the likelihood of erroneously assuming no difference, a power calculation was performed for insulin action (glucose disposal for low-dose insulin clamp). Based on the measured variance of this parameter, the power to detect a 20 or 15% difference between two genotype groups was 0.83 or 0.61, respectively.
In a study from a Swedish cohort, the T allele at position −512 was associated with enhanced insulin sensitivity (based on homeostasis model analysis) in 82 female sibpairs discordant for the variant (5). The same researchers reported that muscle FOXC2 mRNA levels increased in response to insulin in subjects carrying the protective T allele, while no significant change was seen in carriers of the C allele during a hyperinsulinemic-euglycemic clamp. Therefore, in contrast to our results in Pima Indians, the Swedish data clearly suggest an involvement of FOXC2 in insulin resistance, as well as suggest a potential mechanism of differential expression of FOXC2. Since the Swedish study provided evidence for female-specific effects of the −512 polymorphism, we analyzed men and women separately (Table 2). Basal glucose turnover, which in the steady state is equal to glucose disposal, was significantly greater in women with the −512T allele as compared with women with the −512C allele. This difference was not observed in men. In the presence of similar fasting insulin, a difference in basal glucose turnover may suggest greater insulin sensitivity only detectable at low insulin levels, in agreement with the Swedish report, but not during experimental hyperinsulinemia. Also consistent with the Swedish study, we found a sex-specific association of fasting plasma triglycerides with the C-512T variant, where women carrying the −512T allele had lower plasma triglycerides. No difference in triglycerdies was seen in Pima Indian men (Table 2). The G-350T variant also showed a sex-specific effect in that there was a significant association with plasma free fatty acids in men only (P = 0.04) (Table 2). Although both our study and the Swedish study identified sex-specific effects, the physiologic basis of these sex differences remains unknown. It is also noteworthy that these sex differences were not observed when we performed association analyses in men and women together, and simply adjusted for sex.
In Pima Indians with normal glucose tolerance, the G-350T variant was associated with fasting plasma glucose concentrations (P = 0.006) (Table 1). However, we view this association with caution, since none of the traits implicated in the regulation of fasting glucose, such as insulin action, insulin secretion, glucose production, or 2-h plasma glucose, significantly differed between the genotypic groups (Table 1). Moreover, the association with fasting plasma glucose concentrations was only due to the difference in the mean from a small group of T/T homozygotes (n = 23). Therefore, we cannot exclude the possibility that the association of G-350T with fasting plasma glucose concentrations is due to a statistical type 1 error.
In conclusion, variation in the FOXC2 does not appear to have an important role in the etiology of type 2 diabetes in Pima Indians. However, the C-512T variant is associated with obesity and percent body fat in Pima Indians and seems to be involved in the regulation of basal glucose turnover and plasma triglyceride levels in women.
RESEARCH DESIGN AND METHODS
Subjects and clinical characteristics.
The 937 subjects are participants of our ongoing longitudinal study of the etiology of type 2 diabetes among the Gila River Indian Community in Arizona. Every 2 years, all individuals ≥5 years of age are invited to participate in a standardized health examination. To determine diabetes status, a 75-g orally administered glucose tolerance test is given and the results are interpreted according to the criteria of the World Health Organization (6).
For detailed metabolic testing, nondiabetic individuals are admitted to our clinical research ward for 7–10 days, and only individuals found to be healthy by medical history, physical examination, and routine laboratory tests, and not taking medications, are studied. Oral glucose tolerance is measured after 2–3 days on a weight-maintaining diet of mixed composition. Subjects ingest 75 g glucose, and blood for plasma glucose and insulin measurements is drawn before ingesting the glucose and at 30, 60, 120, and 180 min thereafter. On a different day, subjects also receive a 25-g intravenous injection of glucose over 3 min to measure the acute insulin response. Blood samples are collected before infusion and at 3, 4, 5, 6, 8, and 10 min after infusion for determination of plasma glucose and insulin concentrations. The acute insulin response is calculated as half the mean increment in plasma insulin concentrations from 3 to 5 min. The homeostatis model assessment for insulin resistance (HOMA-IR) is calculated as fasting plasma glucose times fasting serum insulin divided by 22.5 (7).
The hyperinsulinemic-euglycemic clamp technique is used to determine basal glucose appearance and insulin-stimulated glucose disappearance (uptake) rates (8). Briefly, insulin is infused to achieve physiologic and maximally stimulating plasma insulin concentrations (137 ± 3 and 2,394 ± 68 μU/ml, respectively) for 100 min for each step. Plasma glucose concentrations are held constant at ∼100 mg/dl by a variable 20% glucose infusion. Tritiated glucose is infused for 2 h before the insulin infusion to calculate rates of postabsorptive glucose appearance and glucose disappearance during the lower dose of insulin infusion. Body composition was estimated by underwater weighing until January 1996 and thereafter by dual-energy X-ray absorptiometry (DPX-1; Lunar Radiation, Madison, WI). A conversion equation derived from comparative analyses is used to make estimates of body composition equivalent between methods (9). Fasting plasma triglycerides were measured by an automated enzymatic method (Express Plus; Ciba-Corning, Irvin, CA), and free fatty acids were measured by a colorimetric assay (Wako Chemicals, Richmond, VA). All studies were approved by the Tribal Council and the Institutional Review Board of the National Institutes of Diabetes and Digestive and Kidney Diseases.
Sequencing of the FOXC2 and genotyping of SNPs.
The entire FOXC2 coding region (1,506 bp), the putative promoter region (638 bp upstream of the exon), and 300 bp of the 3′ region were sequenced in DNA samples from 24 non-first-degree-related Pima Indians. Sequencing was performed using the Big Dye Terminator (Applied Biosystems) on an automated DNA capillary sequencer (model 3700; Applied Biosystems). Sequence information for all oligonucleotide primers used for variant screening is provided in Table 3. Sequencing was also used for genotyping of selected SNPs in 937 Pima Indians.
Statistical analyses were performed using the statistical analysis system of the SAS Institute (Cary, NC). For continuous variables, the general estimating equation procedure was used to adjust for the covariates, including family membership, since some subjects were siblings. The genotype was entered as a class variable. Plasma insulin concentrations, rates of glucose disappearance during the low-dose insulin infusion, and acute insulin response were log transformed before analyses to approximate a normal distribution. The association of SNP genotypes with diabetes was assessed by analysis of contingency tables. A P value <0.05 was considered statistically significant.
Address correspondence and reprint requests to Leslie J. Baier, Clinical Diabetes and Nutrition Section, NIDDK, NIH, 4212 N. 16th St., Phoenix, Arizona 85016. E-mail:.
Received for publication 21 November 2002 and accepted in revised form 10 February 2003.
SNP, single nucleotide polymorphism.