Scientists involved in the study of the causes of human obesity have become optimistic about the possibility of identifying the genes associated with the predisposition to this disease. A growing understanding of the human genome, the high degree of homology between humans and common laboratory mammals for a large number of genes and chromosomal regions, and the availability of a whole variety of technologies and tools to study and manipulate DNA in the laboratory are among the most important reasons for the present level of hope in the obesity research community.

It is quite common to observe in population data that the prevalence of obesity reaches its maximum at ~50-60 y of age. After these ages, it often begins to decline. If we assume that 60 is the age at which we have the maximum prevalence of obesity, it is not surprising that a major imbalance between energy intake and energy expenditure has occurred so frequently by that age. Indeed, people of these ages have had a lifetime ingestion of ~55 million calories. Presumably, they have expended about the same amount of calories, but as the population data indicate, positive energy balance is extremely frequent in our environment.

Why is the chronic imbalance so frequent? One of the hypotheses is that the genes are involved. I will briefly review the status of the research regarding the genetic and molecular basis of obesity. But let me make one point clear at the onset: I believe that the current epidemic of obesity in the developed countries is not caused primarily by our genes. It is caused by a lifestyle that favors positive energy balance as a result of the easy access to a plentiful food supply and technological advances that are making us more and more sedentary. The preceding does not eliminate the possibility that there are people who are more susceptible than others, in the circumstances that we live in, to being in positive energy balance. Indeed, there were obese people early in this century and in past centuries. However, there were fewer of them.

HERITABILITY LEVEL

These results have been confirmed by a recent reanalysis of the data from the Danish Adoption Study (Vogler et al. 1995), whose findings had been published earlier by Stunkard and his Danish colleagues (Stunkard et al. 1986). The new assessment, based on a path analysis approach, concluded that the heritability of BMI reached 34% (Vogler et al. 1995), a value quite compatible with the estimates reported from the Nord-Trøndelag Norwegian Study (Tambs et al. 1991) and the Quebec Family Study (Bouchard et al. 1988).

THE SINGLE-GENE HYPOTHESIS

Several complex segregation analysis studies performed in various populations have reported that a single major gene for high body mass was segregating from the parents to their offspring. However, three studies did not find support for Mendelian transmission unless age and/or gender variations in the major gene were taken into account. From this small body of data, the trend seems to be for a major recessive gene accounting for ~20-25% of the variance, but with age-associated effects, with a gene frequency of ~0.2-0.3. These results must be viewed with great caution, however, because they are based only on the unmeasured genotype approach and the gene(s) has(have) not yet been identified (Bouchard 1994).

OVERVIEW OF METHODS TO IDENTIFY OBESITY-RELATED GENES

Human studies are currently dominated by two types of strategies. The first is the candidate gene approach and is based on the current understanding of the physiopathology of obesity, most commonly pathways related to the regulation of energy balance or to adipose tissue biology. Genes specifying molecules such as receptors, hormones, transporters and other proteins that are key elements of the pathways become candidate genes. These genes are studied for the presence of mutations or some kind of polymorphisms that are then used in association and linkage studies to test whether they provide useful information. A major limitation of the strategy is our fragmented understanding of the physiopathology of obesity at present.

Beside these candidate gene models, there is also available a strong and powerful method commonly used to identify a quantitative trait locus (QTL). The approach is based on the crosses of informative inbred strains and it allows identification of areas of the genome in which genes affecting the phenotype, in this case body fat, reside. It does not yield, by itself, the identity of the gene, but it implicates a chromosomal region. A QTL can signal the beginning of the search for a gene influencing a complex multifactorial trait such as obesity. After such a genomic region has been identified, the QTL rodent linkage group, including markers in the 5' and the 3' areas on both sides of the putative gene, is defined, and a search for a human equivalent can be undertaken in the human genomic data base.

Briefly, the 23 pairs of chromosomes are covered with a set of markers, mostly repetitive sequences that vary in length, used in about the same way as you would use kilometer posts to locate your position on a highway. It takes ~400 polymorphic markers to cover the whole human genome at ~10-cM intervals. Linkage analysis is then undertaken between these markers and the phenotype of interest. If strong evidence for linkage is found within a region of a given chromosome, then it becomes useful to screen that region with a much denser map of markers. If the signal remains strong, the relevant human chromosomal region can be scrutinized with a variety of techniques. More refined positional cloning could be undertaken and DNA sequencing would then follow. In many ways, animal and human research methods are interactive and complementary. Thus far, at least 24 QTL have been reported for body mass or body fat (Chagnon and Bouchard 1996, Pérusse et al. 1997).

Here, we are interested in the role of the specific DNA sequence variations that are co-segregating with obesity phenotypes. A summary of the linkage data indicates that there are now several positive findings. Unfortunately, the vast majority of these positive results have not been confirmed in independent studies. The only genome-wide search that has been completed to date for obesity-related genes was undertaken in pairs of siblings from the Pima Indian community. Persuasive evidence for linkage was observed only with markers on chromosome 11 (11q21-q22) and suggestive evidence on chromosome 3 (3p24.2-p22) (Norman et al. 1997). The scan was accomplished with ~600 markers. Markers on 6p21 were also found to be linked to obesity phenotypes in Pima Indians. Indeed, in 304 pairs of siblings, a fairly strong linkage with a marker ~10 cM away from tumor necrosis factor- was reported (Norman et al. 1995). In the Quebec Family Study, significant linkages were found with DIS200, a marker on chromosome 1p32, and body fat (Chagnon et al. 1997).

There are now at least four linkage studies suggesting that the OB gene or a gene nearby on 7q31.3 might be involved in human obesity. Several hundred human LEP genes have been sequenced, and the OB mutation appears to be extremely infrequent in humans if it exists at all. However, linkage studies suggest that, in the vicinity of LEP, there are markers that are linked to obesity. The first of these studies is our Quebec Family Study in which ~400 pairs of siblings were typed for blood groups and red blood cell enzymes at entry into the study ~20 y ago. A strong linkage was found with the KELL blood group (Borecki et al. 1994). The KELL blood group is about 10-15 cM away from LEP. The strongest linkages were with BMI and the sum of six skinfolds. Over the last several months, three more studies have been reported on this topic.

A group in Paris typed eight microsatellites in pairs of siblings and found evidence of linkage, but only when the siblings had a BMI of at least 35. In these cases, three markers gave significant linkages. They also found linkage disequilibrium between some of the alleles when the BMI was 35. When the BMI was <35, the evidence was negative (Clément et al. 1996). Another study came from the Pennsylvania group, which used 78 families and 8 markers on 7q. They found evidence for linkage with some of the markers only when the BMI was 40. They also looked at linkage disequilibrium among alleles; the BMI became significant only when that of the siblings reached 40 or more (Reed et al. 1996). Both of these studies are based on small sample sizes, but they both provide suggestive evidence of linkage.

Finally, the last paper is based on data from the San Antonio Family Diabetes Study. They relied on 545 pairs of siblings. They used the LOD score approach with several body fat phenotypes. Their strongest evidence for linkage was with an area close to the LEP gene but it was with the sum of five limb skinfolds, and not with the percentage of body fat or the BMI. Their maximum LOD score of 3.1 was with a marker about 10⁶ bp from LEP. In other words, the linkage is probably not with LEP but it could be with a gene that is co-expressed with LEP, or interacting with LEP, or influences the expression of LEP (Duggirala et al. 1996).

We therefore have evidence that becomes progressively more persuasive to the effect that several regions of the human genome might be involved in the susceptibility to obesity. At this time, the most important human chromosomal regions exhibiting linkage seem to be on 1p, 3p, 6p, 7q, and 11q. These results from human studies plus the evidence accumulated in the QTL studies strongly suggest that a good number of genes have the potential to influence energy balance and fat storage. There are probably other areas, but these genomic regions are among the most interesting to investigate at this time.

CONCLUSIONS

Several chromosomal areas are characterized by the presence of several positive findings. The chromosome arms with at least three entries in the map are the following: 1p, 1q, 3p, 4q, 6p, 7q, 8p, 8q, 11p, 11q, 15q, 20q and Xq. We expect that the number of genes and other markers associated or linked with human obesity phenotypes will increase dramatically in the coming years. The main task will remain that of identifying the genes and mutations that are contributing in a major way and under what circumstances.

ACKNOWLEDGEMENTS