FTO: the first gene contributing to common forms of human obesity
Ruth Loos, MRC Epidemiology Unit, Institute of Metabolic Science, Addenbrooke's Hospital, Box 285, Hills Road, Cambridge CB2 0QQ, UK.
Genome-wide association, the latest gene-finding strategy, has led to the first major success in the field of obesity genetics with the discovery of FTO (fat mass and obesity associated gene) as an obesity–susceptibility gene. A cluster of variants in the first intron of FTO showed a strong and highly significant association with obesity-related traits in three independent genome-wide association studies, a finding that has been replicated in several other studies including adults and children of European descent. Homozygotes for the risk allele weigh on average 3–4 kg more and have a 1.67-fold increased risk of obesity compared with those who did not inherit a risk allele. We are still at an early stage in our understanding of the pathways through which FTO confers to increased obesity risk. Studies in humans and rodents have suggested a central role for FTO through regulation of food intake, whereas others have proposed a peripheral role through an effect on lipolytic activity in adipose tissue. There is no doubt that many more obesity–susceptibility loci remain to be discovered. Progress on this front will therefore require major collaborative efforts and pooling of compatible datasets. We stand to learn a lot about the genetic architecture of human obesity in the coming years. The expectations are high but many challenges remain. Among the latter, translating new advances into useful guidelines for prevention and treatment of obesity will be the most demanding.
The steady increase in the prevalence of obesity over the last three decades appears to be driven by a changing environment that is characterized by excessive calorie intake, lower physical activity levels and various other obesogenic factors (1,2). However, not all of us in the present-day sedentary, food-abundant societies become overweight or obese, highlighting the possibility that variation in genetic susceptibility may influence the outcome. Indeed, it appears that obesity is a common, multi-factorial disease that arises through the joint actions of multiple genetic and environmental factors. Family and twin studies have shown that genetic factors contribute 40–70% to the variation in common obesity (3,4), although significantly higher and lower heritability values have also been reported. Gene discovery efforts for common obesity have had only limited success, and progress in the field has been slow.
Genome-wide association – the latest gene-discovery tool
For the last two decades, candidate gene and genome-wide linkage studies have been the two main approaches used to search genes and genetic variants for common diseases and traits. Although a large number of genetic variants and quantitative trait loci that potentially predispose to obesity-related traits have been identified, only few of these variants and loci have been convincingly confirmed (5) (book). This has been caused mainly by the fact that most studies were weak in terms of study design, phenotype assessment, marker coverage of genes and statistically underpowered sample sizes.
Genome-wide association, the latest gene-finding strategy, has changed the pace of gene discoveries and has already proven to work in the context of complex human diseases. Similar to genome-wide linkage approach, genome-wide association is a hypothesis-free approach requiring the screening of the whole genome with the aim of identifying new, unanticipated genetic variants associated with a given disease/trait. While linkage studies rely on co-segregation of chromosomal loci with a disease within families, genome-wide association entails simple associations between hundreds of thousands of genetic variants, generally single nucleotide polymorphisms (SNPs) and a trait or disease of interest. As it does not rely on familial relatedness, genome-wide association studies can simply use large cohorts of cases and controls and thus be based on much larger sample sizes than typical family-based studies. Two major advances in genetics have enabled this new paradigm. First, the completion of the Human Genome project (6) and, more recently, of the International HapMap (7) has considerably increased our knowledge of genetic heterogeneity. Second, substantial progress in high-throughput genotyping technology has made it possible to genotype more than 1 million genetic variants in a single experiment. Together, these two breakthroughs have allowed the production of smartly designed SNP chips that can capture more than 80% of the common genetic variation reported in the HapMap (8). This change in genotyping capacity has facilitated progress in gene-hunting for complex traits and is set to fundamentally improve our understanding of the biology of many traits and diseases at the most fundamental level.
Discovery of FTO as an obesity–susceptibility gene
In recent months, genome-wide association studies have led to a spectacular series of genetic discoveries for various common diseases and traits (9) such as type 1 and type 2 diabetes, prostate and breast cancer, Crohn's disease, rheumatoid arthritis, coronary artery disease, multiple sclerosis, body height and blood lipid levels.
Also we have witnessed the first major success in the field of obesity genetics with the discovery of FTO (fat mass and obesity associated gene) in two independent studies (10,11). Interestingly, the first study, by Frayling et al.(10), identified FTO through a genome-wide association study for type 2 diabetes. A cluster of common SNPs in the first intron of FTO showed a strong and highly significant association with type 2 diabetes. After adjusting for body mass index (BMI), the association with type 2 diabetes was completely abolished, suggesting that the FTO– type 2 diabetes association – was mediated through BMI. Subsequently, the association with BMI and obesity was unequivocally replicated in 13 cohorts comprising more the 38 000 individuals. The second study, by Scuteri et al.(11), performed the first large-scale high-density genome-wide association study of BMI in more than 4000 Sardinians. In their initial analyses, variants in the FTO and PFKP (phosphofructokinase, platelet) gene showed the strongest association, but only those in FTO were significantly replicated in European Americans and Hispanic Americans. A third study, published at the same time as the first two studies, identified FTO while testing for population stratification (12).
FTO is a very large gene whose nine exons span more than 400 kb on chromosome 16. The SNPs identified by these three studies are located in the first intron of the gene, a region where the sequence is strongly conserved across species. They represent a cluster of at least 40 SNPs that are highly correlated (linkage disequilibrium r2 > 0.80 in CEU of the HapMap) in Caucasian populations.
Confirmation of FTO as an obesity–susceptibility gene
As its discovery, the role of FTO as a gene predisposing to obesity in Caucasian populations has been established more firmly (13–16). Besides association with BMI and the augmented risk of overweight and obesity, FTO SNPs have also shown association with various obesity-related traits such as body weight (11,13,16), leptin levels (16), subcutaneous fatness (10,13), fat mass (10,13,16), hip (11) and waist circumference (16), but not with lean mass or body height (10,13,16).
The importance of FTO as an obesity–susceptibility gene was highlighted by a genome-wide association study that compared 487 extremely obese young individuals and 442 healthy lean controls (15). Of the more than 440 000 SNPs that were tested for association with early onset extreme obesity, 15 SNPs, of which six were located in the FTO gene, reached genome-wide significance. Of these 15 SNPs, only the six FTO SNPs showed subsequently association with obesity in 644 nuclear families with at least one obese offspring, whereas none of the other nine SNPs were replicated (15).
The influence of FTO on body composition and on the risk of obesity and overweight is observed already in childhood and persists into adolescence (10,12). While no association is observed with birth weight or with the ponderal index at birth (10,16,17), already at the age of 2 weeks FTO SNPs are associated with increased weight and ponderal index (17).
Of particular interest is the fact that the FTO variants do not seem to affect BMI or the risk of obesity in African Americans (11), Chinese Hans (18), Japanese (19) or Oceanic populations (Melanesians, Micronesians and Polynesians) (20). The minor allele frequency in these populations is less than half of that reported for populations of European descent and the patterns of linkage disequilibrium are also distinct. These observations suggest that these population differences have arisen through evolutionary divergence, perhaps as a result of some negative selection pressure against the FTO risk alleles in African and East Asian populations.
The clinical relevance of FTO
The absolute influence of FTO on BMI and on the risk of obesity and overweight is relatively modest but consistent across studies performed on Caucasians. Each risk allele increases the BMI by ∼0.10–0.13 Z-score units, which is equivalent to ∼0.40–0.66 kg m−2 in BMI or ∼1.3–2.1 kg in body weight for a person 1.80 m tall (10,11,16). The risk of overweight and obesity increases by ∼1.2 and ∼1.3 odds, respectively, for each additional risk allele (10,13,16). In the aggregate, homozygotes for the risk allele weigh about 3–4 kg more and have a 1.67-fold increased risk of obesity compared with those who did not inherit a risk allele (10,11). Similar effect sizes were observed in children and adolescents in whom each FTO risk allele increased the BMI by 0.08–0.12 Z-score units, while the risk of being overweight and obese increased ∼1.27- and ∼1.35-fold with each additional risk allele (10).
Despite these modest effect sizes, the importance of FTO should not be underestimated. The frequency of the FTO risk alleles is high in populations of European descent, with ∼63% of the population carrying at least one risk allele and 16% being homozygous for the risk allele. Furthermore, the population attributable risk (PAR) of FTO for obesity was estimated at ∼20%, which is comparable to the PAR of TCF7L2 variants for type 2 diabetes. This implies that from a population perspective, 20% of the obese cases among Caucasians could be prevented if the negative effects of the FTO risk allele were eliminated (10,12). The PAR percentage for overweight was estimated at ∼13%. It should be noted that PAR does not indicate how many individuals would need to be treated to obtain a 20% reduction in obesity prevalence. More valuable risk estimates for clinical practice could come from prospective cohort studies, in which the predictive value of the FTO SNPs would be determined. Such data would thus indicate whether knowing an individual's FTO genotype could improve the prediction of obesity events. Such information is currently lacking, although we have learned already from recent cross-sectional studies that FTO variants explain only a sobering 1%–1.3% of the variance in BMI (10,11).
One lesson to be derived from the research reported to date is that carrying FTO risk alleles does not imply that one is destined to become obese. This was well illustrated by a recent study showing in more than 6000 middle-aged Danes that the genetic susceptibility of individuals that carry two FTO risk alleles could be suppressed through physical activity. Thus homozygotes for the FTO risk allele were two BMI units heavier if they were physically inactive whereas physically active homozygotes for the same allele had the same BMI as the non-carriers for the FTO risk allele (16).
Physiological role of FTO
Understanding the biology of FTO and its product may shed additional light on the complex system governing the regulation of energy balance. Even though we are still in the early stage of the research on the biology of the FTO gene, new findings have begun to accumulate.
FTO is well conserved across vertebrates and algae, but not in invertebrate animals, fungi and green plants (21,22). Two recent studies showed that FTO is a member of the non-heme dioxygenase superfamily, encodes a 2-oxoglutarate-dependent nucleic acid demethylase, and localizes to the nucleus (23,24). Studies in rodents indicated that Fto mRNA is not only abundant in the brain, particularly in the hypothalamic nuclei governing energy balance, but also in many peripheral tissues (21,23). Furthermore, Fto mRNA in the arcuate nucleus of mice was shown to be up-regulated by feeding and down-regulated by fasting (23,25), but opposite expression patterns were observed in rats (21). Others have supported a central role for FTO through an effect on cerebrocortical insulin sensitivity as individuals homozygous for the risk allele have a reduced insulin response in the brain (26). A peripheral role for FTO was proposed by a study in healthy women showing that FTO mRNA levels in adipose tissue increase with BMI, and carriers of the risk-allele had reduced lipolytic activity, independent of BMI (27).
Let us assume for a moment that the true heritability of BMI among Caucasians is about 50% of the age and sex adjusted variance. What has been accomplished through the series of studies on the FTO gene is that we now have a consensus to the effect that FTO susceptibility alleles account for about 1% of the BMI genetic component. One obvious conclusion is that there are many more obesity–susceptibility loci to be uncovered. However, the results from the first two high-resolution genome-wide association studies for BMI (11) and early onset obesity (15) indicate that this will not be an easy task. Both studies could identify only FTO variants as consistently associated with BMI and obesity. There is no doubt that many more obesity–susceptibility loci remain to be discovered, but these loci are likely to also have small effect sizes, perhaps even smaller than FTO. These other genes remain currently hidden amid false positive or negative results mainly because of limited power of the single genome-wide association study. Progress on this front will therefore require major collaborative efforts and pooling of compatible datasets. One example of such an effort is the so-called ‘Genetic Investigation of Anthropometric Traits’ consortium, which brings together eight research groups from across Europe and the USA. This consortium aims to identify new genetic variants for various obesity-related traits by combining data of 15 genome-wide association studies including more than 32 000 individuals from European ancestry.
Large sample sizes are absolutely key to the success of these gene discoveries enterprises. However, it is also important to expand the studies to more sophisticated phenotype panels. Obesity is a multi-factorial trait that results from a complex interplay between genes and environment. The pathways that lead to obesity are numerous and heterogeneous. It will therefore be essential to undertake genome-wide association studies for indicators of total adiposity, of abdominal adiposity, metabolic rates, metabolic fuel partitioning, dietary intake behaviour, level of physical activity and other determinants of energy balance.
Once a new genetic variant is identified, replicated and tested in different ethnic groups, a number of epidemiological, genetic and physiological studies are needed to clarify the exact nature of the association. Studies are required to refine the strength and exact nature of the genetic signal and to determine which of the variants within a haplotype cluster is functionally related to obesity or a related trait. If the evidence is solid and consistent, then cellular and physiological studies are clearly warranted to shed light on the mechanisms potentially underlying the association between the new gene and its allelic versions with obesity traits.
Genome-wide association has brought considerable excitement to the study of complex human diseases. However, one should keep in mind that well-executed, large-scale candidate gene studies will remain important to the discovery process in the years ahead. Despite the sharp decrease in the cost of SNP chips, not all scientists will be able to afford genome-wide genotyping of their samples. Furthermore, candidate gene studies may be able to evaluate specific genes more in-depth through a number of assets that include large sample sizes, strong panel of well-characterized phenotypes, careful selection of tagSNPs or sequencing the gene or the region of interest. The current Genome-Wide Association (GWA) chips are based on the data reported in the HapMap, which captures the majority, but no all, common variants. Recently, the 1000 Genomes Project was launched which aims to produce a catalogue of variants that are present at 1% or higher frequency across most of the human genome, and down to 0.5% or lower within genes by sequencing the 1000 individuals of various ethnicities. This project will lead to new GWA chips that more fully capture the genetic variation in humans. This will allow the identification of rare variants with potentially large effects.
There is no doubt, however, that we have entered a new era in the search for the genetic determinants of common chronic diseases and other complex human traits. The availability of large panels of markers allowing capturing in excess of 80% of the DNA sequence variation in the human genome offers unprecedented opportunities. We stand to learn a lot about the genetic architecture of human obesity in the coming years. The expectations are high but many challenges remain. Among the latter, translating new advances on the genetic predisposition to obesity into useful guidelines for prevention and treatment will be the most demanding.
C. Bouchard is partially supported by the George A Bray Chair in Nutrition and by grants from the National Institutes of Health.
Conflict of Interest Statement
No conflict of interest was declared.