点击显示 收起
1 From the Department of Genetics, Southwest Foundation for Biomedical Research, San Antonio, TX.
2 Address reprint requests to AG Comuzzie, Department of Genetics, Southwest Foundation for Biomedical Research, PO Box 760549, San Antonio, TX 78245-0549. E-mail: agcom{at}darwin.sfbr.org.
See corresponding article on page 881.
The old adage that "we are what we eat" holds, and, certainly from a scientific perspective, we recognize dietary composition as a fundamental and potent factor that not only influences patterns of weight gain but also acts as a mediator of susceptibility to and severity of a host of chronic metabolic diseases (eg, hypertension and type 2 diabetes). It is clear that a variety of exogenous environmental and lifestyle factors (eg, dietary composition and quantity) play a role in the development of obesity and its related comorbidities (1-3), but there is also substantial evidence of a significant genetic component to these conditions (4-6). In fact, over the past decade, significant progress was made in mapping genes that influence a variety of obesity-related traits, particularly those phenotypes related to adipose tissue accumulation (5). It is interesting that much of this work points to genes involved in appetite regulation, such as POMC and MCR4 (7-11). In this issue of the Journal, Collaku et al (12) report the first significant efforts in humans to identify the genes that are directly involved in the nutrient intake side of the equation.
In contrast to simple Mendelian traits that result from polymorphisms in a single gene, variation in complex phenotypes, such as those represented by measures of nutrient and energy intakes, arises from the action of multiple genes and their interaction with environmental factors (eg, fat intake and adiponectin expression; 13). The identification of the genes that influence complex phenotypes, however, presents several conceptual and analytic challenges that must be addressed if such efforts are to be successful (5). Fortunately, over the course of the past decade, several advances in statistical genetic methods and molecular genetic technology occurred to effectively meet these challenges (5). Indeed, Collaku et al clearly show that the genetic analysis of highly complex traits, represented in their work by quantitative measures of nutrient intake, can prove tractable and highly informative under the appropriate study designs (5).
Initial efforts to identify the specific genes that influence complex traits, such as those associated with a variety of metabolic diseases, relied on the use of a priori selected candidate genes. Such candidate genes are selected on the basis of their preconceived roles or functions in biochemical pathways that are relevant to the specific phenotype of interest (eg, the structural gene for a circulating protein). Whereas the seductive simplicity of the "traditional" candidate gene approach arises mostly from this reassuring use of genes that are already felt to be important contributors to the phenotypes of interest, the sad truth is that this approach has not proven to be all that successful with respect to complex phenotypes. In addition, the use of such a priori candidate genes can, at best, only confirm previously held ideas regarding a gene’s contribution, but cannot identify previously unidentified or unsuspected genes. There are many potential explanations as to why the traditional candidate gene approach has not met with great success in identifying genes for complex phenotypes: perhaps our knowledge of the biology of the focal phenotypes is insufficient or even incorrect, which greatly diminishes our ability to select relevant genes a priori. Another issue that could account for the lack of significant results from traditional candidate gene analyses may be the fact that most of these studies were performed with the use of inefficiently collected or relatively small numbers of samples, which resulted in a significant loss of statistical power to detect these genetic effects.
As a result of the general malaise produced by traditional candidate gene studies, we have seen in recent years a growing trend toward the systematic screening of the entire genome to identify the genes that influence the expression of complex phenotypes. The principal distinction between genome-scanning efforts and those based on the traditional candidate gene approach is the fact that no a priori assumptions concerning the potential importance of genes or chromosomal regions are made before the scan is started, but rather the variation across the entire genome is examined. In a genome scan, linkage analysis is conducted by using a series of anonymous polymorphisms, scattered across the entire genome, to identify quantitative trait loci (QTLs)—defined regions of the genome containing a gene or genes that make a significant contribution to the observed variation in the phenotype. As a result, a genome scan can be used to identify positional candidate genes, which then become the foci of more intensive follow-up analyses (14). A positional candidate gene differs from a traditional candidate gene in that it is considered as a candidate only after the establishment of its proximity to a QTL that was identified by prior linkage analysis in a genome screen. Thus, the genome scan approach offers the potential of identifying novel or previously unsuspected genes (or both types) that influence the phenotype of interest.
Collaku et al reported their results from the application of a genome-wide screening approach to detect genes that influence variations in dietary energy intake in participants in the Health, Risk Factors, Exercise Training, and Genetics (HERITAGE) Family Study. Whereas previous work in rodents had found a genetic contribution to the preference for various dietary components (15), this is the first work to detect significant evidence that a QTL influences such traits in humans. On the basis of that work, Collaku et al identified several chromosomal regions throughout the genome that influence variations in nutrient intakes, and they proposed several positional candidate genes for more detailed follow-up. This work is of interest not only for its specific findings, but also because it shows the utility of this type of genetic approach in the study of nutrient intake in free-living human populations, and it thereby opens the door for additional research in this promising area.
The results of such genome-scanning efforts are impressive, but they really represent only the necessary first steps on the path to gene identification. The true value of such work as that presented by Collaku et al is that it serves to focus our attention on the most promising areas of the genome by allowing us to select "positional" candidate genes, rather than following preconceived notions of which genes might be important, for the more intensive (and costly) follow-up analyses that are required to identify the specific polymorphisms involved. Whereas the ultimate goal of genetic analysis is to identify the specific variation in a gene that is responsible for the variation observed in a phenotype, a critical first step remains the systematic narrowing of the search from the entire genome (estimated to include between 30 000 and 40 000 genes) to a few genomic regions that may contain just a few hundred genes, and this is what Collaku et al have accomplished. Certainly the task remaining is not trivial, but there are promising new advances, both technical and analytic, that make feasible the sequencing and analysis of large spans of DNA in the region of these QTLs. Indeed, given the continued development of technologic and analytic advances, we are moving rapidly toward the ultimate goal of identifying the variations in specific genes that influence the expression of such complex phenotypes as nutrient intakes.
REFERENCES
Related articles in AJCN: