A Novel Missense Substitution (Val1483Ile) in the Fatty Acid Synthase Gene (FAS) Is Associated With Percentage of Body Fat and Substrate Oxidation Rates in Nondiabetic Pima Indians

The subjects were participants of our ongoing longitudinal study of the etiology of type 2 diabetes among the Gila River Indian Community in Arizona (3,4). The 216 individuals selected for the association analyses were all nondiabetic subjects who had been admitted to our clinical research ward for 7–10 days for detailed metabolic studies. All were found to be healthy by medical history, physical examination, and routine laboratory tests and were not taking medications. Oral glucose tolerance was measured after 2–3 days on a weight-maintaining diet of mixed composition. Subjects ingested 75 g of glucose, and blood for plasma glucose and insulin measurements was drawn before ingesting the glucose and at 30, 60, 120, and 180 min thereafter (13). Body composition was estimated by underwater weighing until January 1996 and by dual-energy X-ray absorptiometry (DPX-1; Lunar Radiation, Madison, WI) thereafter. A conversion equation derived from comparative analyses was used to make estimates of body composition equivalent between methods (6). The measurements of energy expenditure and substrate oxidation in the respiratory chamber have been described previously (6). Briefly, volunteers entered the chamber after an overnight fast and remained in the chamber for 23 h. Subjects were fed calories to maintain energy balance according to previously determined equations, and the rate of energy expenditure was measured continuously, calculated for each 15-min interval within the chamber and then extrapolated to 24 h. Carbon dioxide production (Vco₂) and oxygen consumption (Vo₂) were calculated for every 15-min interval. The 24-h respiratory quotient was calculated as the ratio of 24-h Vco₂ and 24-h Vo₂. Based upon the 24-h respiratory quotient, 24-h metabolic rate, and 24-h urinary nitrogen excretion, the 24-h oxidation rates of fat, carbohydrate, and protein were determined (6). All studies were approved by the Tribal Council and the Institutional Review Board of the National Institutes of Diabetes and Digestive and Kidney Diseases.

To identify genetic variants, all 43 exons, including intron/exon splicing sites, the 5′ region (1,958 bp upstream of the first translation initiation site), and the 3′ untranslated region (UTR) were sequenced in DNA samples from 32 full-blooded, non–first degree related Pima Indians. Sequencing was performed using the Big Dye Terminator (Applied Biosystems, Foster City, CA) on an automated DNA capillary sequencer (ABI Prism 3700; Applied Biosystems). Sequence information for all oligonucleotide primers used for variant screening is available upon request.

Genotyping of selected SNPs in a cohort of nondiabetic full-blooded Pima Indians (n = 216) representing a subset of individuals from the sibling-based linkage analysis, which is described elsewhere (5), was done using the TaqMan assay (Applied Biosystems) for the −605C/T variant and by direct sequencing (as described above) for the −1597C/T, Val1483Ile, 7795C/T, and 8271G/A variants. The TaqMan genotyping reaction was amplified on a GeneAmp PCR system 9700 (50°C for 2 min, 95°C for 10 min, 95°C for 15 s, and 62°C for 1 min, for 38 cycles), and fluorescence was detected on an ABI Prism 7700 sequence detector (Applied Biosystems).

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, because some subjects were siblings. Plasma insulin concentrations were log transformed before analyses to approximate a normal distribution.

The FAS was sequenced in 32 Pima Indian DNA samples to detect variation, and 13 SNPs were identified (Fig. 1). Two SNPs were in the 5′ region, a −1597C/T (allele frequency of the T allele = 0.16) and a −605C/T (T = 0.23). Six SNPs were found in the exons, where four predicted silent substitutions and two predicted missense substitutions. SNPs predicting silent substitutions were in exon 4 (684C/T→Asn189; T = 0.06), exon 18 (3060G/A→981Glu; A = 0.02), exon 20 (3384G/C→Ala1089; C = 0.19), and exon 26 (4827C/T→1570Ser; T = 0.02). SNPs predicting missense substitutions were in exon 8 (1403G/A→Arg429His; A = 0.02) and exon 25 (4564G/A→Val1483Ile; A = 0.10). Two SNPs were in introns: intron 26 (20G/A; A = 0.06) and intron 37 (12C/T; T = 0.16; rs4246445 in the National Center for Biotechnology Information SNP Database) and three SNPs were in the 3′UTR (7718G/A, A = 0.02; 7795C/T, T = 0.16; and 8271G/A, A = 0.22).

Based on the genotypic concordance, several of these SNPs were in strong linkage disequilibrium. Among the more common variants (minor allele frequency ≥0.10), the 12C/T variant in intron 37 was in 100% linkage disequilibrium with 7795C/T in the 3′ UTR. Similarly, the −605C/T variant in the promoter was in nearly complete linkage disequilibrium with the 3384G/C variant (Ala1089). The remaining three common SNPs appeared to be genotypically unique. Therefore, five representative common SNPs were selected (−1597C/T, −605C/T, Val1483Ile, 7795C/T, and 8271G/A) for additional genotyping for association analyses. The results of a Hardy-Weinberg equilibrium test were as follows: −1597C/T (P = 0.77), −605C/T (P = 0.39), Val1483Ile (P = 0.02), 7795C/T (P = 0.02), and 8271G/A (P = 0.53).

The Val1483Ile polymorphism was associated with percentage of body fat, 24-h carbohydrate oxidation rates, and 24-h respiratory quotient as measured in a respiratory chamber. Compared with Val homozygotes, subjects with Ile/x had a lower mean percentage of body fat (30 ± 1 vs. 33 ± 1%, P = 0.002; adjusted for age, sex, and family membership) and a lower mean 24-h carbohydrate oxidation rate (983 ± 41 vs. 1,094 ± 19 kcal/day, P = 0.03), which resulted in a lower mean 24-h respiratory quotient (0.845 ± 0.01 vs. 0.850 ± 0.01 kcal/day, P = 0.04; both adjusted for age, sex, family membership, percentage of body fat, and energy balance) (Table 1).

The SNP −1597C/T was associated with 24-h metabolic rate (P < 0.05 after adjusting for age, sex, family membership, fat mass, and fat-free mass) (Table 2). The SNP −605C/T was modestly associated with fasting plasma glucose (P < 0.05), but none of the other variants were associated with any measure of plasma glucose or insulin concentrations that would be predictive of type 2 diabetes (Tables 1 and 2).

In addition, among the five different SNPs genotyped, we identified five common haplotypes (frequency >0.05) that accounted for 88% of the observed haplotypes. These common haplotypes could be defined by four of the SNPs ([−1597C/T]-[−605C/T]-[7795C/T]-[8271G/A]). Each of the common haplotypes was analyzed for its association with metabolic characteristics. To accomplish this, a modification of the zero-recombination haplotyping approach was used (14–16). However, analyses of these haplotypes did not reveal any strong associations beyond those expected from analysis of individual SNPs.

FAS was analyzed as a candidate gene for human obesity, and a novel SNP predicting a Val1483Ile substitution was identified that was associated with percentage of body fat and substrate oxidation rates in nondiabetic Pima Indians. In addition, the Val1483Ile variant appears to contribute to the linkage at 17q25, but by itself does not explain all of the linkage at this position, as a Haseman-Elston test for linkage following adjustment for this SNP drops the LOD score from 1.9 to 1.5. Although no reports on genetic variation in human FAS have been previously published, there are a number of recent studies from rodents indicating that FAS may be involved in obesity through regulation of feeding behavior. Treatment of mice with FAS inhibitors (cerulenin, C75) resulted in reduced food intake and substantial body weight loss (8). The FAS inhibitor, C75, acts both centrally to reduce food intake and peripherally to increase fatty acid oxidation, leading to rapid and profound weight loss, loss of adipose mass, and resolution of fatty liver in diet-induced obese mice (12). If the Val/Ile substitution functions to reduce the activity of FAS, it might be predicted to be associated with decreased fat in humans. The Val1483Ile variant is positioned within the interdomain region of the FAS. Although the interdomain region has no catalytic activity itself, it appears to have an important role in dimer formation. Both human and animal FAS proteins are dimers, composed of two identical multifunctional subunits arranged in an antiparallel configuration, which generates two active centers for fatty acid synthesis separated by the interdomain regions. Therefore, alteration of the interdomain region could affect dimer formation and consequently alter the configuration necessary for catalytically active FAS (17–20). However, further studies are needed to determine whether the Val/Ile substitution indeed structurally alters the interdomain, which in turn reduces FAS activity.

The physiologic mechanism by which the Val1483Ile substitution results in the phenotypic and metabolic changes observed in Pima Indians also requires further study. Our data indicate that the higher percentage of body fat in subjects with the Val/Val genotype results from their higher carbohydrate oxidation rates. As indicated by their higher 24-h respiratory quotient, these Val/Val subjects have an increased ratio of carbohydrate to lipid oxidation rates, i.e., they preferentially burn carbohydrates over lipids, which results in an increased body lipid mass. In mice, inhibition of FAS by C75 leads to profound weight loss via an increase in peripheral fatty acid oxidation, but this effect is achieved by a C75-specific pharmacological stimulation of carnitine palmityltransferase-1, which controls fatty acid entry into the mitochondria for oxidation (12). It is unlikely that the Val1483Ile substitution affects the substrate oxidation rates via a similar mechanism. However, FAS inhibitors also act centrally through inhibition of the orexigenic system of neuropeptide Y and agouti-related protein, leading to a reduced food intake and weight loss in rodents. Therefore, future studies could examine possible relationships between the Val1483Ile variant and food intake in humans, which could partially explain the reduced body fat seen in carriers of the 1483Ile variant. It has also been reported (9) that in lean individuals, FAS inhibitors reduce body weight independent of food intake or hypothalamic neuropeptide Y, suggesting that other uncharacterized neuropeptide pathways may mediate the effects of FAS inhibitors on metabolic rate. Therefore, changes in FAS activity could potentially affect substrate oxidation rates, which would result in the decrease in body fat observed in the Pima Indians. It is also noteworthy that FAS inhibitors are more potent in obese compared with lean mice, suggesting that FAS inhibition may be enhanced by the effects of factors such as insulin, glucose, and fatty acids, which are elevated in obese individuals (9). Therefore the effect of FAS inhibition may be more marked in Pima Indians, an extremely obese population, compared with leaner populations.

Based on our data, we suggest that the human FAS may have a role in the regulation of body weight in Pima Indians. Although these findings support previous studies in rodents, replication studies in humans and functional studies with the Val1483Ile substitution are necessary to confirm these findings and define the relative importance of this novel substitution in the pathogenesis of human obesity.