点击显示 收起
1 From the Department of Nutrition (JSF), the Rowe Program in Genetics (CHW), the Department of Pediatrics (CHW), and the Section of Neurobiology, Physiology, and Behavior (CHW), University of California, Davis, Davis, CA
2 Reprints not available. Address correspondence to CH Warden, Rowe Program in Genetics, University of CA, Davis, Davis, CA 95616. E-mail: chwarden{at}ucdavis.edu.
See corresponding article on page 26.
In the 12 y since the cloning of the ob gene (1), which coded for the adipose-generated hormone leptin, there has been a continuing search for other genes whose polymorphisms contribute to obesity. Whereas the identification of single-gene mutations causing obesity has been relatively straightforward, the search for genes contributing to common obesity is proving more difficult. Mapping quantitative trait loci by linkage in families of humans and other species has identified many chromosomal regions with obesity-causing genes (2). Although a recent study (3) reported the positional cloning of the quantitative trait from a human obesity quantitative trait locus, identification of quantitative trait genes has been slow.
Association studies provide a mechanism for testing the hypothesis that alleles of specific genes are associated with obesity and obesity-related phenotypes. The Québec Family Study (QFS) is a long-running study of the genetics and phenotypes related to obesity (4). A great strength of the QFS, begun in 1978 by Bouchard et al, is the extensive phenotyping of the study subjects. The QFS data set has produced >30 association publications, the latest of which, the report by Loos et al (5) on the association of obesity phenotypes with gene variants in adiponectin and its receptors, appears in this issue of the Journal.
Adiponectin is a hormone, secreted by adipocytes and detectable in serum, that sensitizes the body to insulin (6). It is secreted as several different multimer complexes. The functions of adiponectin are mediated through 2 receptors, ADIPOR1 and ADIPOR2, that are variably expressed and that differ in sensitivity to the adiponectin complexes.
The significant actions of adiponectin are those of sensitizing tissues to insulin and increasing fatty acid oxidation. Lower adiponectin concentrations in humans are associated with greater visceral obesity, insulin resistance, and incidence of type 2 diabetes in certain populations (6). Studies in mice show that adiponectin increases fatty acid oxidation and energy consumption and reduces tissue triacylglycerol content, changes that lead to an increase in insulin sensitivity (7). Some investigators have proposed that low adiponectin concentrations that result from the interaction of genetic factors and environment play a causal role in the development of insulin resistance, type 2 diabetes, and the metabolic syndrome (6). Because obesity is a significant component in the development of insulin resistance, it is pertinent to question whether polymorphisms in the genes coding for adiponectin or its receptors are associated with obesity phenotypes.
Significant associations (P = 0.05 to 0.001) between various measures of obesity (eg, overall obesity, body mass index, and visceral fat) and polymorphisms in the gene coding for adiponectin (ADIPOQ) have been reported in 14 studies: 3 of these studies were reported as original research (8-10), and 11 earlier studies were reviewed (2). No reports of associations between polymorphisms in ADIPOQ and energy expenditure or fat oxidation are available. Likewise, no studies of associations between polymorphisms in either ADIPOR1 or ADIPOR2 and any measure of adiposity or energy metabolism are available.
Loos et al (5) state that a polymorphism in ADIPOQ is associated with overall adiposity (P = 0.006 for percentage body fat; P = 0.0002 for the sum of 6 skinfold thicknesses) in the QFS. The reported associations with measures of abdominal fat, however, are not as strong (P = 0.02 and 0.04). In addition, 2 separate polymorphisms in ADIPOR1 and ADIPOR2 are associated with respiratory quotient (RQ) in lean subjects only (P = 0.018 for ADIPOR1; P = 0.003 for ADIPOR2). In addition to this obesity-gene interaction, gene-gene interactions between single-nucleotide polymorphisms (SNPs) in the promoters of ADIPOQ and ADIPOR1 affect RQ and both overall and abdominal adiposity (P = 0.02 to 0.0002). These P values are certainly as strong as the P values recently reported in the QFS for association of polymorphisms of plasminogen activator inhibitor 1 gene with obesity (11), of leptin and leptin receptor genes with metabolic rate and RQ (12), and of melanocortin 4 receptor gene with physical activity (13).
Gene-gene interactions contribute considerably to phenotypic variations in obesity (14, 15). Understanding this network of additive genetic effects and interactions is essential to a determination of the genetic contributions to obesity. It is important to note that the gene-gene interactions reported by Loos et al (5) were found only because those investigators chose to examine polymorphisms in the adiponectin receptor genes as well as in ADIPOQ. This finding highlights a significant weakness of association studies to dissect the genetic factors of obesity.
Recent years have witnessed the beginning of another approach to association studies, one in which whole-genome SNP maps are used to identify genes influencing traits anywhere in the genome (16, 17). Whole-genome association studies are made possible by technologies that can be used to genotype hundreds of thousands of SNPs across the genome at costs of <$1000 per subject (17). The advantages of this approach are that no prior assumptions about the identity of the underlying genes need be made and that the overall data can be evaluated to identify the most significant results, including interactions.
Does the introduction of whole-genome association studies mean that the QFS approach of studying one gene or a few candidate genes at a time is a thing of the past? For several reasons, we do not think so. Whole-genome studies necessarily make assumptions about the number of SNPs to genotype per gene, whereas a more limited association study can investigate a greater fraction of the SNPs in target genes. Genotyping a greater fraction of SNPs in any one target gene can be important because associations may be observed with some but not other SNPs. A significant disadvantage of whole-genome association studies lies in the statistical problems resulting from multiple testing (16, 17). Ultimately, however, investigations in well-phenotyped cohorts, such as the cohorts available in the QFS, may use these validated sets of genetic variants to help identify disease-promoting genotypes. Because many interacting genes are components of any one pathway, the examination of multiple genes within a pathway could mitigate the problem of finding interacting genes.
ACKNOWLEDGMENTS
Neither of the authors had a personal or financial conflict of interest with respect to the study by Loos et al.
REFERENCES
Related articles in AJCN: