International Journal of (2001) 25, Suppl 1, S10–S18 ß 2001 Nature Publishing Group All rights reserved 0307–0565/01 $15.00 www.nature.com/ijo PAPER The genetics of obesity: practical implications

J Hebebrand1*, C Sommerlad1, F Geller2,TGo¨rg2 and A Hinney1

1Clinical Research Group, Department of Child and Adolescent Psychiatry of the Philipps University, Marburg, Germany; and 2Institute of Medical Biometry and Epidemiology of the Philipps University, Marburg, Germany

Over the past years substantial progress has been made in the molecular elucidation of monogenic forms of obesity both in rodents and in humans. In addition, several quantitive trait loci have been mapped in mice. In humans, non-parametric linkage studies have led to the identification of relevant chromosomal regions, some of which have already been confirmed. In this review we focus on an interpretation of the heritability estimates obtained in twin, family and adoption studies. These estimates include both direct and indirect genetic effects. Non-additive genetic factors seemingly contribute even more than additive factors. The importance of the non-shared environment is stressed. GeneÂgene interactions need to be considered when interpreting recent molecular genetic results pertaining to haplo-insufficiency mutations in the melanocortin-4 receptor . We conclude by discussing the implications of the recent molecular findings in humans for phenotypical assessment in ongoing family studies. International Journal of Obesity (2001) 25, Suppl 1, S10 – S17

Keywords: energy expenditure; melanocortin-4 receptor; energy intake; phenotype; television; breast feeding

Introduction opment of novel drugs for treatment of obesity. In the light Excellent reviews pertaining to our current understanding of of these current and future developments we perceive the the genetics of obesity have been published recently,1–3 need to address some of the practical implications of past which document the tremendous pace with which this and ongoing genetic research related to weight regulation field is evolving. Large-scale efforts are being undertaken to and obesity. detect the involved in weight regulation, particularly in obesity. The major results over the past four years relate to the identification of monogenic forms of human obesity. As Heritability estimates: what do they imply? the sequencing of the human genome is virtually complete, Genetics vs environment the coming years will undoubtedly witness the identification It is worthwhile reviewing heritability estimates for obesity of several genes located within chromosomal linkage or in broader terms body weight in general. Twin studies regions. In contrast to the genes underlying monogenic have produced the most consistent and highest heritability forms of obesity, these future findings will have implications estimates for (BMI; kg=m2) in the range for substantial subgroups of the obese population. The between 0.6 and 0.9.4 These high estimates apply to twins respective findings are bound to not only have an impact reared both together and apart. Except for the newborn on the scientific community but also on society as a whole. period for which a lower heritability of 0.4 has been calcu- Because of the public interest in this phenotype the research lated5 age does not affect heritability estimates to a substan- results will find rapid entrance into modern day society, tial degree. Evidently, the influence of the intrauterine necessitating a communication process between researchers, environment on birth weight is strong. However, genes clinicians and the public to delineate potential implications. rapidly override this initial environmental effect, because Expectations are high that this genetic knowledge will allow in school-age children high heritabilities already apply. identification of new drug targets and the subsequent devel- Based on twin studies it has been suggested that the herit- ability of BMI is most pronounced during late childhood and adolescence.6 The genes relevant for weight regulation are apparently not the same throughout life, according to a large *Correspondence: J Hebebrand, Clinical Research Group, Department of longitudinal twin study7 only 40% of the genetic factors that Child and Adolescent Psychiatry of the Philipps University Marburg, Hans- Sachs-Str. 6, 35033 Marburg, Germany. influence the BMI at age 20 y continue to do so at age 48 y. E-mail: [email protected] After this age the same genes mostly remain operative. The genetics of obesity J Hebebrand et al S11 Adoption and family studies have derived considerably Table 1 Responses of mothers of extremely obese index patients aged lower heritability estimates, the lowest estimate being in the 6 – 23 y to the question ‘Has your child ever obtained food or sweets magnitude of 0.1 for BMI.8 However, a recent large family secretly on an almost daily basis for at least four weeks in a row?’ in relationship to age. Mothers were questioned retrospectively as to each study has for the first time come up with a heritability age period estimate of 0.67,8 which is in the same range as those derived from twin studies. Furthermore, a heritability estimate of Age n Yes (%) No (%) Don’t know (%) this magnitude indicates that the genetic component for First year of life 371 0.27 97.84 1.89 body weight is not that much below that for body height. In 1 – 6 y 376 7.71 90.69 1.60 their family study Maes et al8 discussed potential reasons 7 – 10 y 370 30.54 67.57 1.89 why the high heritability estimates in twin studies may be 11 – 14 y 315 48.89 47.94 3.17  15 y 109 44.04 52.29 3.67 better than those obtained in other types of family studies including a better control for age effects.

tains to breast-feeding which has been shown to protect Direct and indirect genetic effects from the development of . In a recent For an adequate interpretation of the high heritability esti- study the duration of breast-feeding predicted a lower rate mates obtained in twin studies it is noteworthy to point out of obesity at ages 5 and 6.10 The causual inference is that that both direct and indirect genetic effects are subsumed breast-feeding directly prevents the development of obesity. under the genetic component.9 Consider the following However, an alternative interpretation is that more hungry example: both infant twins of a monozygotic pair are fre- infants are switched to bottle feeding at an earlier age than quently irritable due to a biologically driven increased less hungry infants.11 Indeed, the most common reason (direct genetic effect). This hunger induces frequent mothers give for stopping breast-feeding is the impression feedings by the caretaker irrespective of his or her back- that their child does not get an adequate supply of milk.12 ground. Thus, even if the twins are separated at birth, the Thus, potentially the direct genetic effect (increased hunger) caretakers would respond similarly as soon as they had implies an earlier switch to bottle feeding (indirect genetic learned that providing food soothes the child. This indirect effect) in industrialized countries. Viewed in this context it is genetic effect actually represents the response of the envir- not surprising that the child bottle-fed earlier on turns out to onment to the genetically based excessive hunger. It is read- have a higher body weight in comparison to infants who are ily evident that in this case the indirect genetic effect is breast-fed for a prolonged period. crucial for the development of early onset obesity in both twins. If one of the caretakers of the reared apart twins systematically curtailed the infant’s energy intake (implying The importance of non-shared environment a willingness to deal with the resulting irritability), early Another interesting aspect of the twin studies, but also of onset obesity would ensue in only the other twin. adoption and family studies, has been the observation that It can be argued that the illustrated indirect genetic effect the non-shared environment explains considerably more is only important in infants and young children who cannot variance of the quantitative phenotype BMI than shared yet obtain food on their own. As the twins grow older their environment. In the large study of twins reared both relentless hunger would undoubtedly cause them to devise together and apart by Stunkard and coworkers13 shared ways and means to obtain food without direct parental environment basically did not explain variance, instead assistance. In addition, whereas the initial non-shared environment totally explained the environmen- might be perceived as an indication of good health, the tal component estimated at 30%. At first glance this is caretakers themselves would soon become aware of the difficult to comprehend because we intuitively attribute obesity and attempt to reduce the twins’ energy intake. obesity to familial eating and patterns and the Accordingly, it is of interest to note that almost 50% of the familial level of physical activity. However, based on the interviewed mothers of the extremely obese children and relative unimportance of shared environment we need to adolescents whom we have included in our molecular focus on what makes the BMI of twins or siblings dissimilar. genetic studies, reported secret eating episodes of their 11 Accordingly, it is not important what the mother buys to eat to 14-y-old children (Table 1). Evidently, the children learn or what she serves. Instead, variance can basically only be that their overeating is not socially acceptable and resort to accounted for by assessing what and how much each family eating secretly. Accordingly, in those individuals with a member actually eats.14 We need to overcome the notion genetically driven increased hunger who can by themselves that two obese children within the family developed their obtain access to food, the indirect genetic effect is accounted obesity because of their shared environment. Especially the for by modern day society with its ready availability of a (reared apart) twin studies indicate that this similarity has a large variety of inexpensive and highly palatable foods. strong genetic basis. In this context, therapeutic strategies This discussion has evident implications for assessing that depend on influencing shared environment are not environmental influences on body weight. An example per- incorporating the information stemming from these twin

International Journal of Obesity The genetics of obesity J Hebebrand et al S12 and family studies. We need to further devise ways and viewing and obesity does not amount to the identification of means to therapeutically target the non-shared environ- a causative environmental factor. Instead, based on our ment. Based on these considerations a list of candidate genotypes we react differentially towards this environmental environmental experiences believed to promote obesity in factor. This is just one out of potentially many different children has been outlined.15 genotypeÂenvironment interactions. Obviously, the com- plexity of the genetic basis of obesity applies both from a metabolic and behavioral perspective. GeneÂgene and geneÂenvironment interactions In the light of genotypeÂenvironment interactions the Additive and non-additive factors recent obesity epidemic is interesting to analyze. Because the Formal genetic studies have also revealed that both additive gene pool of a population cannot change within a genera- and non-additive gene effects are important.8 In their twin tion, environmental changes are presumed to be of eminent study Stunkard and coworkers13 estimated that in males 57% importance.23 Nevertheless, it should again be pointed out and 17% of the variance in BMI is due to non-additive and that these changes can only have a major impact because our additive gene effects, respectively. In females the respective genotypes render us especially obesity prone.24 Irrespective percentages were estimated at 37% and 31%. Thus, accord- of the assumed environmental basis of secular trends for ing to this study non-additive gene effects account for a both height and weight there is no indication that herit- larger proportion of the variance than additive. The discre- ability estimates in family studies are actually declining in pancy between the higher and lower heritability estimates the light of the obesity epidemic. calculated in the twin and family studies, respectively, is Despite the constancy of the gene pool a genetic contri- presumably also partially due to the strong effect of non- bution to epidemic obesity cannot be dismissed. Based on additive factors, which can be assessed more reliably in twin the high rate of parental obesity observed among parents of studies. In this situation different at different loci do extremely obese children we have hypothesized that the not simply affect the phenotype via mere additive effects. recent increase of social stigmatization of obese individuals Instead, 1 at locus A might only have an impact on might actually have led to an increase of assortative body weight if allele 1 at locus B is also inherited. The mating.25 Accordingly, having an obese partner would have importance of the genetic background including both addi- been more acceptable 50 y ago than currently or just one or tive and non-additive effects has repeatedly been documented two decades ago. If assortative mating for obesity indeed in inbred mice.16,17 increased over the last half of the twentieth century, children born of two obese parents would stand a substantial chance of inheriting predisposing gene variants from both parents Polygeny and genotypeÂenvironment interactions and of developing a severe form of obesity. This mechanism Whereas it is widely accepted that several different genes could contribute to epidemic obesity particularly by affect- contribute to obesity, the respective implications for our ing the upper tail of the BMI distribution. In this context it is understanding appear less clear. Evidently, these genes worthwhile pointing out that the most dramatic secular BMI would in some way influence energy intake and=or expen- increments in children and adolescents have been detected diture. Behavioral genetics has convincingly demonstrated in the and obese range.26,27 A part from obesity that approximately 50% of the variance of diverse complex per se assortative mating might also occur for specific life- quantitative behaviors is genetically determined.18 Accord- styles which predispose to weight gain. Assortative mating ingly, eating a high-fat or exercising too little cannot for leisure activities has been documented.28 According to only be viewed as having an environmental basis. Instead, this scenario obesity need not necessarily already be present gene variants predisposing an individual to choose such a when a couple first get acquainted. However, the preference diet or predisposing to physical inactivity need to be con- for (genetically co-determined) activities could entail an sidered. Accordingly, both the macronutrient intake19 and elevated risk for development of an elevated body weight activity levels20 have been shown to be genetically deter- in both partners. mined. Whereas many different studies have addressed the potential consequences of TV watching for childhood obe- sity,21 there is an absolute scarcity of studies addressing the Molecular genetic studies extent to which TV watching is heritable.22 In our own study Monogenic obesity we have found significant correlations between parents and The cloning of the agouti29 and gene30 in rodent their offspring for hours spent watching television, especially obesity models mark the initiation of the molecular genetics during the weekend (unpublished data). It appears that this of obesity. A success story has evolved since; in humans phenotype like many other behavioral phenotypes has a autosomal recessive mutations in the genes for leptin,31 – 33 heritable component. Thus, some children are genetically ,34 prohormone convertase 1 (PC1,35) and pro- prone to watch more TV than others. Such formal genetic opiomelanocortin (POMC,36) have been shown to lead to studies pertaining to this behavior are required to caution us early onset obesity. Not surprisingly, inbreeding was docu- that the mere delineation of the relationship between TV mented in the affected family members with leptin and

International Journal of Obesity The genetics of obesity J Hebebrand et al S13 leptin receptor gene mutations.31 – 34 Compound heterozyg- with a normal body weight.61 The weight curves of the osity has been detected for PC1 and POMC mutations.35,36 heterozygous MC4R knockout mice also clearly extend into All of the mutations apparently lead to additional pheno- the range formed by wild-type animals.65 The strong non- typical manifestations including adrenal insufficiency additive component determined in several formal genetic (POMC), red hair (POMC) and reduced or impaired fertility studies8,13 suggests that the reduced penetrance of a mela- (PC1, leptin and leptin receptor). Furthermore, seemingly all nocortin-4 receptor null mutation might be due to just this mutations lead to early onset extreme obesity induced by an effect. Whereas Vaisse et al61 did not find direct evidence for increased energy intake. A reduced energy expenditure as a modifying genes, such an effect can currently by no means contributing pathogenetic factor has not been documented. be excluded. Thus, such non-obese carriers might have Finally, all mutations are evidently exceedingly rare; muta- additionally inherited an allele(s) protecting them from tion screenings in the coding and promotor regions of the developing obesity. Further studies are required to assess respective genes have not revealed other mutations which the impact on body weight of individuals with the respective can functionally readily be linked to obesity.37 – 44 However, mutations. It is of interest to point out that recent evidence linkages or associations have been reported to markers sur- implicates a role of a missense mutation in the porcine rounding or within the leptin gene locus on 7q3145 – 47 and MC4R in fatness, growth and feed intake traits in this the leptin receptor locus on 1p32.42,48 – 50 In addition, linkage species.66 of serum leptin levels and fat mass has been described upon The MC4R mutations suggest that it will be especially use of markers that localize to the POMC region on chromo- difficult to pinpoint the effect of gene variants that exert some 2p51,52 and variation of leptin levels has been asso- only a minor influence on body weight. In association ciated with POMC polymorphisms.53 studies large case numbers will be required to firmly establish The first autosomal dominant form of human obesity due a role of the respective variants; a stratification according to to a missense mutation in the gene coding for the peroxi- appears reasonable. Thus, probands could be some-proliferator-activated receptor gamma2 (PPARg2) was categorized as , normal-weight, overweight and discovered in four out of 121 obese unrelated Germans,54 but obese; allele frequencies could accordingly be expected to not in any of the 237 normal weight controls. Analysis of the systematically increase. However, because the gene variants mutation by retroviral transfection and overexpression in predisposing to leaness might be non-allelic to variants murine fibroblasts revealed functional deficiencies. The same predisposing to overweight this systematic increase is prob- mutation has not been detected in any further obese indivi- able but not a precondition. Linkage analyses are hampered duals including obese German children, adolescents and by the fact that a priori the phenotype needs to be defined.60 adults.55,56 A low threshold (eg 75th BMI-percentile) will lead to inclu- sion of a substantial number of phenocopies, a high thresh- old (eg BMI > 97th BMI-centile) will potentially imply that Melanocortin-4 receptor gene allele carriers are not considered as obese. In this situation The most recent advance has been the identification of the transmission disequilibrium test67 is a viable option functionally relevant haploinsufficiency mutations in the provided that the allele frequency is not too low. melanocortin-4 receptor gene.57 – 62 Haploinsufficiency mutations have been detected in four out of 492 extremely obese German children and adolescents indicating that the Extensive or simple phenotypical assessment? frequency of these mutations is in the range of 1%.60 Several The detection of the first monogenic forms of obesity may missense mutations have also been detected in our own serve to critically reflect on standard procedures for the study group and in other samples.59,61,63,64 Assays have phenotypical assessment of obese individuals participating revealed that most of these mutations have a functional in genetic studies. Given our current knowledge the essential implication.61,64 Accordingly, the epidemiological signifi- features of the monogenic forms include red hair,36 an eating cance of functionally relevant melanocortin-4 receptor pattern similar to that observed in children with the Prader mutations is presumably in the magnitude of 3%.61 Willi syndrome,31,34 disturbances of the hypothalamic – It is presently unclear to what extent body weight pituary – gonadal axis,32 – 34 other endocrinological abnorm- is elevated by the respective mutations. The original alities including adrenal insufficiency,36 markedly elevated mutation screenings were performed in extremely obese or reduced serum leptin levels31,32,34 and elevated proinsulin patients.57 – 59,63,64 Systematic screenings in individuals with levels.35 Undoubtedly, obese individuals with any one or a overweight or obesity grade 1 have not been reported. How- combination of these abnormalities potentially have a ever, of the 15 haploinsufficiency carriers identified amongst monogenic form of obesity. No specific phenotypical the relatives of four index patients three had a current BMI abnormalities have been reported in carriers of MC4R hap- below the 95th centile. A gender difference was apparent:60 loinsufficiency mutations.57 – 62 Recently, indirect evidence similar to results obtained in MC4R knockout mice,65 female points to a potential role of leptin deficiency in reduced carriers were more obese than the males. In France single immune function.33 This wide range of phenotypical males with haploinsufficiency mutations were identified abnormalities observed in the rare autosomal recessive,

International Journal of Obesity The genetics of obesity J Hebebrand et al S14 quasi syndromal forms of obesity are definitely not repre- types to include. Thus, despite their evident potential impli- sentative of obese individuals in general. It is doubtful cation, behavioral phenotypes are not routinely included in whether standard phenotypic assessments used in genetic extensive phenotypical assessments to any significant research can include such a diverse range of phenotypes. extent. For example, in discordant sibs we have screened As for normal obesity it appears worthwhile to discuss for eating disorders and restrained eating using interviews phenotyping in genetically orientated research. Two extreme and questionnaires. Not unsurprisingly, amongst normal positions serve to illustrate different viewpoints. On the one weight or underweight sibs of obese index patients we hand it can be argued that due to the potentially enormous identified individuals with these problems (unpublished heterogeneity extensive phenotyping should be performed, data). A focus on diverse metabolic and physiological para- which actually serves two purposes: (1) phenotypically meters alone would not have cautioned us as to inclusion of homogeneous subgroups with an assumed common genetic these lower weight sibs in a discordant sib pair approach. basis can be identified; (2) apart from analyzing body weight In this context it is of interest to assess the subjective or BMI alone, additional phenotypes can be included in explanations obese and normal-weight individuals give for association and linkage studies. Obviously, extensive pheno- their current body weight. Despite a high familial loading for typing is expensive and additionally requires the proband’s obesity25 only a small minority of the German obese adoles- cooperation. Typically, current phenotypic assessment pro- cents and adults we have ascertained, spontaneously mention cedures include physiological and metabolic parameters, genetic factors as being important (unpublished data). In an many of which are correlated. attempt to systematically assess the subjective opinions we On the other hand a minimal phenotypic assessment questioned a large number of both extremely obese adoles- procedure would focus on current BMI. This approach cents and parents with obese offspring subcategorized accord- would allow the ascertainment of a considerably larger ing to the absence or presence of obesity (Figure 1). The number of individuals at the same cost as an extensive questions pertained to eating, smoking, alcohol intake, phenotyping of a considerably smaller number of indivi- weight watching behavior, emotional items and physical duals. In this scenario extensive phenotyping can be activity. Both the obese adolescents and the obese adults initiated once a relevant gene variant has been identified. particularly commonly endorsed statements pertaining to Contrasting these two extreme positions points to inher- eating, seemingly indicating that they consider their energy ent respective advantages and disadvantages. Aside from the intake as too high. For example ‘I consider the fact that I like to high costs per ascertained individual the first approach could eat as relevant to my current body weight’ was stated by almost be criticized for concentrating on specific phenotypes only. 90%, 75% and 90% of the obese fathers, mothers and adoles- Similar to the situation in the aforementioned monogenic cents, respectively. The comparison between obese and forms of obesity it is a priori difficult to decide what pheno- normal-weight parents revealed that mainly the items pertain-

Figure 1 (a) Affirmative responses of extremely obese male (n ¼ 81) and female (n ¼ 131) adolescents (mean age 16.44 y) to the question: ‘Is the following statement relevant for your current body weight?’

International Journal of Obesity The genetics of obesity J Hebebrand et al S15

Figure 1 (b) Affirmative responses of mothers (mean age 41.13 y) of extremely obese adolescents subdivided according to absence (BMI < 30, n ¼ 161) or presence (BMI  30, n ¼ 141) of obesity to the question: ‘Is the following statement relevant for your current body weight?’

Figure 1 (c) Affirmative responses of fathers (mean age 44.35 y) of extremely obese adolescents subdivided according to absence (BMI > 30, n ¼ 124) or presence (BMI  30, n ¼ 127) of obesity to the question: ‘Is the following statement relevant for your current body weight?’ ing to eating are more frequently endorsed by the obese. The more frequently perceived emotional aspects as being impor- responses to other statements only minimally differed tant. However, only the question related to eating and emo- between obese and normal-weight parents (eg ‘I have a physi- tions (‘I eat when frustrated’) separated the obese mothers from cally strenuous job’). Obese men especially also attributed their their normal-weight counterparts. obesity to a reduced energy expenditure (eg ‘My physical It is granted that these subjective opinions must be activity is low’). A comparison by gender reveals that females viewed with considerable caution. However, they do raise

International Journal of Obesity The genetics of obesity J Hebebrand et al S16 Table 2 Differences between current BMI and retrospectively recalled maximal and minimal BMI during adulthood in obese mothers (n=139) and fathers (n=126) of extremely obese index patients. The parents were subdivided according to their current grade of obesity. Parents were asked to recall both their maximal (max) and minimal (min) weights; BMI was calculated using current height. If recalled maximal weights were lower than measured current weight by more than 1.0 kg/m2 (n=35; 15 males and 20 females), parents were excluded. If the difference was less than 1.0 kg/m2, DBMI was set to zero. Not every parent was able to recall both minimal and maximal weight

Current Mean s.d. Min Max BMI Delta BMI n DBMI DBMI DBMI DBMI

Mothers 30 – 34.9 Max BMI minus current BMI 65 2.36 3.37 0.00 20.45 35 – 39.9 39 1.49 2.17 0.00 8.65  40 26 2.22 4.90 0.00 23.35

30 – 34.9 Current BMI minus min BMI 67 8.87 3.29 0.89 17.47 35 – 39.9 43 11.37 3.99 0.71 19.68  40 29 17.42 4.38 9.41 24.03

Fathers 30 – 34.9 Max BMI minus current BMI 73 1.73 2.17 0.00 10.98 35 – 39.9 26 1.39 1.88 0.00 7.42  40 17 2.62 5.34 0.00 16.14

30 – 34.9 Current BMI minus min BMI 74 7.58 2.79 0.00 13.62 35 – 39.9 33 10.13 3.87 3.40 19.52  40 19 15.75 4.20 7.12 24.90

the question as to whether extensive metabolic phenotypical also be argued that in the light of the high costs of both the assessments capture the essence of the biological mechan- ascertainment of probands and families and a genome scan, isms underlying obesity. What if the major causes for obesity different or related phenotypes should be analyzed simulta- are indeed simply an increased and a reduced level neously. Based on the current strong focus on metabolic of physical activity? The absence of assessment procedures to phenotypes we believe that the assessment procedure reliably quantify these basic behaviors in the large number of should also include behavioral phenotypes. Perhaps the individuals required to account for genetic heterogeneity complexity of the phenotype obesity can be overcome some- hampers genetic research. In addition, such an assessment what if different groups pursue different phenotypical assess- would have to take developmental aspects into account. ment procedures. It is readily apparent that reliably Thus, even if an elevated appetite can be measured reliably quantifiable phenotypes should be preferred. in an obese proband, does this also imply that this mechan- Currently available data from genome scans indicate that ism also led to the development of the elevated body weight? the use of the simple phenotype ‘current BMI’ does lead to Another problem arises from the fact that the co-analysis replicable results. Thus, the initial finding of linkage to of different phenotypes results in multiple testing, which chromosome 10p using current BMI as the relevant pheno- statistically should be corrected for, in order to render an type in a French study52 could be replicated in a German adequate interpretation of the results possible. Some studies study population.69 In addition, other linkage regions have do not seem to be focussed due to the inclusion of so many been identified upon use of the BMI.52,70 While these linkage different phenotypes. The intercorrelation of specific pheno- results might merely reflect the initial hits with many more types is an additional problem. Thus, in the light of the to come upon use of more differentiated phenotypes, they correlation between BMI, percentage body fat and serum do substantiate that the BMI is not too crude a measure to leptin levels68 the finding of linkage of serum leptin levels allow detection of linkage. unadjusted for BMI and fat mass to chromosome 2q51,52 is The published linkage findings also raise the question as difficult to comprehend. Finally, in the light of the cross- to the clinical significance of the gene variants within the sectional nature of most studies the rationale behind an identified linkage regions. Will these turn out to have a exact and expensive determination of fat or fat-free mass major effect on body weight? Or will the effect merely be can be questioned, given the commonly encountered undu- so small that the respective findings will turn out not to have lating weight course of extremely obese individuals (Table 2). a major impact on clinical practice? We expect that the If on the other hand obesity is such a genetically complex answer to this and related questions will be forthcoming phenotype, a crude phenotypical assessment procedure within the next few years. might not enable the identification of the respective genes. If indeed homogeneous subgroups can be defined which are Acknowledgements also more homogenous in genetic terms, this should greatly The Deutsche Forschungsgemeinschaft, the Bundesminister- facilitate the identification of the underlying genes. A ium fu¨r Bildung und Forschung, the International Associa- detailed phenotypical characterization might allow the iden- tion for the Promotion of Co-operation with Scientists from tification of putative genes predisposing to for instance both the New Independent States of the former Soviet Union obesity and non-insulin dependent mellitus. It can (INTAS) and the European Union (Framework V) generously

International Journal of Obesity The genetics of obesity J Hebebrand et al S17 support our research efforts. We thank the families that have 25 Hebebrand J, Wulftange H, Go¨rg T, Ziegler A, Hinney A, Barth N, Mayer H, Remschmidt H. Epidemic obesity: are genetic factors contributed to our genetic studies. involved via increased rates of assortative mating? Int J Obes Relat Metab Disord 2000; 24: 345 – 353. References 26 Troiano RP, Flegal KM, Kuczmarski RJ, Campbell SM, Johnson CL. 1 Barsh GS, Farooqi IS, O’Rahilly S. Genetics of body-weight regula- Overweight prevalence and trends for children and adolescents: tion. Nature 2000; 404: 644 – 651. the National Health and Nurtrition Examination Surveys, 1963 to 2 Chagnon YC, Pe´russe L, Weisnagel SJ, Rankinen T, Bouchard C. 1991. Arch Pediatr Adolesc Med 1995; 149: 1085 – 1091. The human obesity gene map: the 1999 update. Obes Res 2000; 8: 27 Barth N, Ziegler A, Himmelmann GW, Coners H, Wabitsch M, 89 – 117. Hennighausen K, Mayer H, Remschmidt H, Scha¨fer H, Hebebrand 3 Ravussin E, Bogardus C. (2000) Energy balance and weight reg- J. Significant weight gains in a clinical sample of obese children ulation: genetics versus environment. Br J Nutr 2000; 83(Suppl 1): and adolescents between 1985 and 1995. Int J Obes Relat Metab S17 – 20. Disord 1997; 21: 122 – 126. 4 Bouchard C, Perusse L. Genetic aspects of obesity. Ann NY Acad Sci 28 Houts RM, Robins E, Huston TL. Compatibility and the develop- 1993; 699: 26 – 35. ment of premarital relationships. J Marriage Family 1996; 58:7– 5 Vlietinck R, Derom R, Neale MC, Maes H, van Loon H, Derom C, 20. Thiery M. Genetic and environmental variation in the birth 29 Bultmann SJ, Michaud EJ, Woychik RP. Molecular characteriza- weight of twins. Behav Genet 1989; 19: 151 – 161. tion of the mouse agouti locus. Cell 1992; 71: 1195 – 1204. 6 Pietilainen KH, Kaprio J, Rissanen A, Winter T, Rimpela A, Viken 30 Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman RJ, Rose RJ. Distribution and heritability of BMI in Finish adoles- JM. Positional cloning of the mouse obese gene and its human cents aged 16 y and 17 y: a study of 4884 twins and 2509 homologue. Nature 1994; 372: 425 – 432. singletons. Int J Obes Relat Metab Disord 1999; 23: 107 – 115. 31 Montague CT, Farooqi IS, Whitehead JP, Soos MA, Rau H, Ware- 7 Fabsitz RR, Carmelli D, Hewitt JK. Evidence for independent ham NJ, Sewter CP, Digby JE, Mohammed SN, Hurst JA, Chee- genetic influences on obesity in middle age. Int J Obes Relat tham CH, Earley AR, Barnett AH, Prins JB, O’Rahilly S. Congenital Metab Disord 1992; 16: 657 – 666. leptin deficiency is associated with severe early-onset obesity in 8 Maes HH, Neale MC, Eaves LJ. Genetic and environmental factors humans. Nature 1997; 387: 903 – 908. in relative body weight and human adiposity. Behav Genet 1997; 32 Strobel A, Issad T, Camoin L, Ozata M, Strosberg AD. A leptin 27: 325 – 351. missense mutation associated with hypogonadism and morbid 9 Harris JR. The nurture assumption. Why children behave the way they obesity. Nature Genet 1998; 18: 213 – 215. do? Free Press: New York; 1998. 33 Ozata M, Ozdemir IC, Licinio J. Human leptin deficiency caused 10 von Kries R, Koletzko B, Sauerwald T, von Mutius E, Barnert D, by a missense mutation: multiple endocrine defects, decreased Grunert V, von Voss H. Breast feeding and obesity: cross sectional sympathetic tone, and immune system dysfunction indicate new study. Br Med J 1999; 319: 147 – 150. targets for leptin action, greater central than peripheral resistance 11 Hebebrand J. Breast feeding and obesity. Prolonging breast to the effects of leptin, and spontaneous correction of leptin- feeding to reduce obesity may be a burden. Br Med J 1999; 319: 1576. mediated defects. J Clin Endocrinol Metab 1999; 84: 3686 – 3695. 12 Graffy JP. Mothers’ attitudes to and experience of breast feeding: a 34 Cle´ment K, Vaisse C, Lahlou N, Cabrol S, Pelloux V, Cassuto D, primary care study. Br J Gen Pract 1992; 42: 61 – 64. Gourmelen M, Dina C, Chambaz J, Lacorte JM, Basdevant A, 13 Stunkard AJ, Harris JR, Pedersen NL, McClearn GE. The body mass Bougneres P, Lebouc Y, Froguel P, Guy-Grand B. A mutation in the index of twins who have been reared apart. New Engl J Med 1990; human leptin receptor gene causes obesity and pituitary dysfunc- 222: 1483 – 1487. tion. Nature 1998; 392: 398 – 401. 14 Hewitt JK. The genetics of obesity: what have genetic studies told 35 Jackson RS, Creemers JW, Ohagi S, Raffin-Sanson ML, Sanders L, us about the environment. Behav Genet 1997; 27: 353 – 358. Montague CT, Hutton JC, O’Rahilly S. Obesity and impaired 15 Faith MS, Johnson SL, Allison DB. Putting the behavior into the prohormone processing associated with mutations in the behavior genetics of obesity. Behav Genet 1997; 27: 423 – 439. human prohormone convertase 1 gene. Nature Genet 1997; 16: 16 Pomp D. Genetic dissection of obesity in polygenic animal 303 – 306. models. Behav Genet 1997; 27: 285 – 306. 36 Krude H, Biebermann H, Luck W, Horn R, Brabant G, Gruters A. 17 Ewart-Toland A, Mounzih K, Qiu J, Chehab FF. Effect of the Severe early-onset obesity, adrenal insufficiency and red hair genetic background on the reproduction of leptin-deficient pigmentation caused by POMC mutations in humans. Nature obese mice. Endocrinology 1999; 140: 732 – 738. Genet 1998; 19: 155 – 157. 18 Plomin R, DeFries JC, McClearn GE, Rutter M. Behavioral genetics. 37 Echwald SM, Rasmussen SB, Sorensen TI, Andersen T, Tybjaerg- Freeman: New York; 1997. Hansen A, Clausen JO, Hansen L, Hansen T, Pedersen O. Identi- 19 Reed DR, Bachmanov AA, Beauchamp GK, Tordoff MG, Price RA. fication of two novel missense mutations in the human OB gene. Heritable variation in food preferences and their contribution to Int J Obes Relat Metab Disord 1997; 21: 321 – 326. obesity. Behav Genet 1997; 27: 373 – 387. 38 Hinney A, Becker I, Heibu¨lt O, Nottebom K, Schmidt A, Ziegler A, 20 Pe´russe L, Tremblay A, Leblanc C, Bouchard C. Genetic Mayer H, Siegfried W, Blum WF, Remschmidt H, Hebebrand J. and environmental influences on level of habitual physical activity Systematic mutation screening of the pro-opiomelanocortin and participation. Am J Epidemiol 1989; 129: 1012 – 1022. gene: identification of several genetic variants including three 21 Gortmaker SL, Must A, Sobol AM, Peterson K, Colditz GA, Dietz different insertions, one nonsense and two missense point muta- WH. Television viewing as a cause of increasing obesity among tions in probands of different weight extremes. J Clin Endocrinol children in the United States, 1986 – 1990. Arch Pediatr Adolesc Metab 1998; 83: 3737 – 3741. Med 1996; 150: 356 – 362. 39 Hinney A, Bornscheuer A, Depenbusch M, Mierke B, To¨lle A, 22 Plomin R, Corley R, Carey G, DeFries JC, Fulker DW. Individual Middeke K, Ziegler A, Roth H, Gerber G, Zamzow K, Ballauff A, differences in television viewing in early childhood: nature as Hamann A, Mayer H, Siegfried W, Lehmkuhl G, Poustka F, well as nurture. Psychol Sci 1990; 1: 371 – 377. Schmidt MH, Hermann H, Herpertz-Dahlmann BM, Fichter M, 23 Taubes G. As obesity rates rise, experts struggle to explain why. Remschmidt H, Hebebrand J. No evidence for involvement of the Science 1998; 280: 1367 – 1368. leptin gene in , , underweight or 24 Neel JV, Weder AB, Julius S. Type II diabetes, essential hyperten- early onset extreme obesity: identification of two novel muta- sion, and obesity as ‘syndromes of impaired genetic homeostasis’: tions in the coding sequence and a novel polymorphism in the the ‘thrifty genotype’ hypothesis enters the 21st century. Perspect leptin gene linked upstream region. Mol Psychiat 1998; 3: 539 – Biol Med 1998; 42: 44 – 74. 543.

International Journal of Obesity The genetics of obesity J Hebebrand et al S18 40 Kalidas K, Dow E, Saker PJ, Wareham N, Halsall D, Jackson RS, 55 Hamann A, Mu¨nzberg H, Buttron P, Hinney A, Mayer H, Siegfried Chan SP, Gelding S, Walker M, Kousta E, Johnston DG, O’Rahilly W, Hebebrand J, Greten H. Missense variants in the human S, McCarthy MI. Prohormone convertase 1 in obesity, gestational peroxisome proliferator-activated receptor-g2 gene in lean and diabetes mellitus, and NIDDM: no evidence for a major suscept- obese subjects. Eur J Endocrinol 1999; 141: 90 – 92. ibility role. Diabetes 1998; 47: 287 – 289. 56 Evans D, Mann WA, Heer J, Michel U, Wendt D, Kortner B, Wolf 41 Maffei M, Stoffel M, Barone M, Moon B, Dammerman M, Ravus- A, Beisiegel U. Variation in the gene for human peroxisome sin E, Bogardus C, Ludwig DS, Flier JS, Talley M, Auerbach S, proliferator activated receptor gamma (PPAR gamma) does not Friedman JM. Absence of mutations in the human OB gene in play a major role in the development of morbid obesity. Int J Obes obese=diabetic subjects. Diabetes 1996; 45: 679 – 682. Relat Metab Disord 2000; 24: 647 – 651. 42 Roth H, Korn T, Rosenkranz K, Hinney A, Ziegler A, Kunz J, 57 Vaisse C, Cle´ment K, Guy-Grand B, Froguel P. A frameshift Siegfried W, Mayer H, Hebebrand J, Grzeschik KH. Transmission mutation in human MC4R is associated with a dominant form disequilibrium and sequence variants at the leptin receptor gene of obesity. Nature Genet 1998; 20: 113 – 114. in extremely obese German children and adolescents. Hum Genet 58 Yeo GS, Farooqi IS, Aminian S, Halsall DJ, Stanhope RG, O’Rahilly 1998; 103: 540 – 546. S. A frameshift mutation in MC4R associated with dominantly 43 Gotoda T, Manning BS, Goldstone AP, Imrie H, Evans AL, Stros- inherited human obesity. Nature Genet 1998; 20: 111 – 112. berg AD, McKeigue PM, Scott J, Aitman TJ. Leptin receptor gene 59 Hinney A, Schmidt A, Nottebom K, Heibu¨lt O, Becker I, Ziegler A, variation and obesity: lack of association in a white British male Gerber G, Sina M, Go¨rg T, Mayer H, Siegfried W, Fichter M, population. Hum Mol Genet 1997; 6: 869 – 876. Remschmidt H, Hebebrand J. Several mutations in the melano- 44 Matsuoka N, Ogawa Y, Hosoda K, Matsuda J, Masuzaki H, Miya- cortin-4 receptor gene including a nonsense and a frameshift waki T, Azuma N, Natsui K, Nishimura H, Yoshimasa Y, Nishi S, mutation associated with dominantly inherited obesity in Thompson DB, Nakao K. Human leptin receptor gene in obese humans. J Clin Endocrinol Metab 1999; 84: 1483 – 1486. Japanese subjects: evidence against either obesity-causing muta- 60 Sina M, Hinney A, Ziegler A, Neupert T, Mayer H, Siegfried W, tions or association of sequence variants with obesity. Diabetolo- Blum WF, Remschmidt H, Hebebrand J. Phenotypes in three gia 1997; 40: 1204 – 1210. pedigrees with autosomal dominant obesity caused by haploin- 45 Cle´ment K, Garner C, Hager J, Philippi A, Leduc C, Carey A, sufficiency mutations in the melanocortin-4 receptor gene. Am J Harris TJR, Jury C, Cardon LR, Basdevant A, Demenais F, Guy- Hum Genet 1999; 65: 1501 – 1507. grand B, North M, Froguel P. Indication for linkage of the human 61 Vaisse C, Clement K, Durand E, Hercberg S, Guy-Grand B, Froguel OB gene region with extreme obesity. Diabetes 1996; 45: 687 – P. Melanocortin-4 receptor mutations are a frequent and hetero- 690. geneous cause of morbid obesity. J Clin Invest 2000; 106: 253 – 46 Reed DR, Ding Y, Xu WZ, Cather C, Green ED, Price RA. Extreme 262. obesity may be linked to markers flanking the human OB gene. 62 Farooqi IS, Yeo GS, Keogh JM, Aminian S, Jebb SA, Butler G, Diabetes 1996; 45: 691 – 694. Cheetham T, O’Rahilly S. Dominant and recessive inheritance of 47 Roth H, Hinney A, Ziegler A, Barth N, Gerber G, Stein K, Bro¨mel T, morbid obesity associated with defi- Mayer H, Siegfried W, Scha¨fer H, Remschmidt H, Grzeschik K-H, ciency. J Clin Invest 2000; 106: 271 – 279. Hebebrand J. Further support for linkage of extreme obesity to 63 Gotoda T, Scott J, Aitman TJ. Molecular screening of the human the obese gene in a cohort of obese children and adolescents. Exp melanocortin-4 receptor gene: identification of a missense var- Clin Endocrinol Diabetes 1997; 105: 341 – 344. iant showing no association with obesity, plasma glucose, or 48 Thompson DB, Ravussin E, Bennett PH, Bogardus C. Structure insulin. Diabetologia 1997; 40: 976 – 979. and sequence variation at the human leptin receptor gene in lean 64 Gu W, Tu Z, Kleyn PW, Kissebah A, Duprat L, Lee J, Chin W, and obese Pima Indians. Hum Mol Genet 1997; 6: 675 – 679. Maruti S, Deng N, Fisher SL, Franco LS, Burn P, Yagaloff KA, 49 Chagnon YC, Wilmore JH, Borecki IB, Gagnon J, Perusse L, Nathan J, Heymsfield S, Albu J, Pi-Sunyer FX, Allison DB. Identi- Chagnon M, Collier GR, Leon AS, Skinner JS, Rao DC, Bouchard fication and functional analysis of novel human melanocortin-4 C. Associations between the leptin receptor gene and adiposity in receptor variants. Diabetes 1999; 48: 635 – 639. middle-aged Caucasian males from the HERITAGE family study. 65 Huszar D, Lynch CA, Fairchild-Huntress V, Dunmore JH, Fang Q, J Clin Endocrinol Metab 2000; 85: 29 – 34. Berkemeier LR, Gu W, Kesterson RA, Boston BA, Cone RD, Smith FJ, 50 Chagnon YC, Perusse L, Lamothe M, Chagnon M, Nadeau A, Campfield LA, Burn P, Lee F. Targeted disruption of the melanocor- Dionne FT, Gagnon J, Chung WK, Leibel RL, Bouchard C. Sug- tin-4 receptor results in obesity in mice. Cell 1997; 88: 131 – 141. gestive linkages between markers on human 1p32 – p22 and body 66 Kim KS, Larsen N, Short T, Plastow G, Rothschild MF. A missense fat and insulin levels in the Quebec Family Study. Obes Res 1997; variant of the porcine melanocortin-4 receptor (MC4R) gene is 5: 115 – 121. associated with fatness, growth, and feed intake traits. Mamm 51 Comuzzie AG, Hixson JE, Almasy L, Mitchell BD, Mahaney MC, Genome 2000; 11: 131 – 135. Dyer TD, Stern MP, MacCluer JW, Blangero J. A major quantitative 67 Spielman RC, McGinnis RE, Ewens WJ. Transmission test for trait locus determining serum leptin levels and fat mass is located linkage disequilibrium: the insulin gene region and insulin- on human chromosome 2. Nature Genet 1997; 15: 273 – 276. dependent diabetes mellitus (IDDM). Am J Hum Genet 1993; 52: 52 Hager J, Dina C, Francke S, Dubois S, Houari M, Vatin V, Vaillant 506 – 516. E, Lorentz N, Basdevant A, Cle´ment K, Guy-Grand B, Froguel P. A 68 Considine RV, Sinha MK, Heiman ML, Kriauciunas A, Stephens genome-wide scan for human obesity genes reveals a major TW, Nyce MR, Ohannesian JP, Marco CC, McKee LJ, Bauer TL, susceptibility locus on chromosome 10. Nature Genet 1998; 20: Caro JF. Serum immunoreactive-leptin concentrations in normal- 304 – 308. weight and obese humans. New Engl J Med 1996; 334: 292 – 295. 53 Hixson JE, Almasy L, Cole S, Birnbaum S, Mitchell BD, Mahaney 69 Hinney A, Ziegler A, Oeffner F, Wedewardt C, Vogel M, Wulftange MC, Stern MP, MacCluer JW, Blangero J, Comuzzie AG. Normal H, Geller F, Stu¨bing K, Siegfried W, Goldschmidt HP, Remschmidt variation in leptin levels in associated with polymorphisms in the H, Hebebrand J. Independent confirmation of a major locus for gene, POMC. J Clin Endocrinol Metab 1999; obesity on chromosome 10. Clin Endocrinol Metab (in press). 84: 3187 – 3191. 70 Lee JH, Reed DR, Li WD, Xu W, Joo EJ, Kilker RL, Nanthakumar E, 54 Ristow M, Muller-Wieland D, Pfeiffer A, Krone W, Kahn CR. North M, Sakul H, Bell C, Price RA. Genome scan for human Obesity associated with a mutation in a genetic regulator of obesity and linkage to markers in 20q13. Am J Hum Genet 1999; adipocyte differentiation. New Engl J Med 1998; 339: 953 – 959. 64: 196 – 209.

International Journal of Obesity