Leptin's Sexual Dimorphism Results from Genotype by Sex Interactions Mediated by Testosterone

Leptin’s Sexual Dimorphism Results fromGenotype by Sex Interactions Mediated byTestosteroneLisa J. Martin,* Michael C. Mahaney,* Laura Almasy,* Jean W. MacCluer,* John Blangero,*Cashell E. Jaquish,† and Anthony G. Comuzzie*

AbstractMARTIN, LISA J., MICHAEL C. MAHANEY, LAURAALMASY, JEAN W. MACCLUER, JOHN BLANGERO,CASHELL E. JAQUISH, AND ANTHONY G.COMUZZIE. Leptin’s sexual dimorphism results fromgenotype by sex interactions mediated by testosterone. ObesRes. 2002;10:14–21.Objective: Recent studies have reported the existence ofmarked sexual dimorphism in serum leptin levels in hu-mans, with women having approximately three times thelevels of men. As we have shown for other measures ofadiposity, such sexual dimorphism can arise from a specialcase of genotype by environment interaction, that of geno-type by sex interaction.Research Methods and Procedures: Using maximum like-lihood-based variance decomposition techniques, we exam-ined the genetic and environmental architecture of sexualdimorphism in serum leptin levels in 1147 Mexican Amer-icans from the San Antonio Family Heart Study.Results: Both the genetic and environmental variances forthis trait differed significantly between the sexes (p � 0.001and p � 0.01, respectively), with women displaying largervalues for both components. We found significant evidencethat different genes influence variation in serum leptin lev-els between the two sexes (p � 0.05). Furthermore, thispattern of sexual dimorphism in serum leptin levels per-sisted even after accounting for the effects of either thepercentage of body fat or total body fat. However, this

pattern of sexual dimorphism was eliminated after account-ing for the effects of testosterone.Discussion: These findings suggest that the sexual dimor-phism seen in leptin levels is not simply explained asdifferences in total adiposity between the sexes. We con-clude that the genes, which influence variation in serumleptin levels, are differentially expressed depending on sex,and that the sexes also show differences in response ofthe expression of this obesity-related trait to unmeasuredresidual effects.

Key words: Mexican Americans, variance decomposi-tion, body fat, sex hormones

IntroductionQuantitative traits, such as hormone levels or morpho-

metric variables, often exhibit genotype by environment(G � E) interaction. This interaction leads to differentialphenotypic expression based on the environment in whichthe genes are placed (1). A special case of G � E interactionis genotype by sex (G � S) (2,3), where intrinsic differencesbetween the sexes represent different environments inwhich genes are found. The same genetic background canyield a phenotype that is differentially expressed dependingon the sex of the individual; therefore, the expression ofsexual dimorphism in a phenotype can have a strong geneticcomponent. We have previously shown that a large portionof the sexual dimorphism associated with morphologicalmeasures of fat accumulation and topography (e.g., bodymass index [BMI], skinfolds, and circumference measure-ments) are the result of G � S interaction (4).

Work with rodent models of obesity led to the identifi-cation of leptin, a 16-kDa peptide hormone secreted byadipocytes and hypothesized to regulate feeding behaviorand energy expenditure (5–7). Studies in humans haveshown a strong phenotypic correlation between leptin levels

Submitted for publication August 20, 2001.Accepted for publication in final form October 24, 2001.*Southwest Foundation for Biomedical Research, San Antonio, Texas and the †NationalHeart Lung and Blood Institute, Bethesda, Maryland.Address correspondence to Dr. Lisa J. Martin, Department of Genetics, Southwest Foun-dation for Biomedical Research, Box 760549, San Antonio, TX 78245-0549.E-mail: [email protected] © 2002 NAASO

14 OBESITY RESEARCH Vol. 10 No. 1 January 2002

and measures of adiposity such as BMI and percentage ofbody fat (8,9). The structural locus for leptin, the OB gene,has been cloned and mapped to chromosome 7 in humans(10). Whereas variation at the OB gene accounts for a smallportion of the variation seen in leptin levels (�18%) (11),we have identified a quantitative trait locus on humanchromosome 2 that accounts for 47% of the total phenotypicvariation in levels of leptin in humans (12). As reported inprevious studies (8,9,13–16), there is pronounced sexualdimorphism in serum leptin levels in our sample of MexicanAmericans (12). Typically these sex-specific differenceshave been considered to reflect that women have a higherpercentage of total body fat and a different pattern of fatdistribution than do men (8,9,13). However, serum leptinlevels are higher in women compared with men, regardlessof body fatness (15). Furthermore, such a simplified expla-nation for the sexual dimorphism in this trait fails to addressthe real issue: even if leptin levels are a direct reflection oftotal adiposity, what is the underlying source of sexualdimorphism in adiposity-related traits in humans?

Baumgartner et al. (17) hypothesized that the sexualdimorphism of leptin may be attributable to sex hormones,specifically estrogen and testosterone. Previous research hasidentified a relationship between obesity and reproductivehormones, such as testosterone, estrogen, and sex hormone-binding globulin (SHBG) (18–22). Moreover, leptin may bedifferentially involved by sex in developmental (22–24) andreproductive pathways (25–29). Furthermore, leptin recep-tors are present in testicular tissue (19,30), whereas POMC,a candidate gene for the regulation of leptin, contains bothestrogen- and testosterone-responsive elements (31,32).However, these studies have not specifically examined therole of genetics in sexual dimorphism of leptin levels.

Therefore, we have posed three previously unexaminedquestions. What are the genetic and environmental contri-butions to the expression of sexual dimorphism in serumleptin levels? Is it possible that some portion of the geneticsex-specific variance in serum leptin levels may be inde-pendent of adiposity? Is it possible that some portion of thegenetic sex-specific variance in serum leptin levels may beindependent of reproductive hormones? To address thesequestions we used full pedigree-based variance decomposi-tion analysis to examine the contribution of both genetic andenvironmental factors to the expression of sexual dimor-phism in serum leptin levels, and to examine whether suchdimorphism exists independent of other measures of adi-posity and reproductive hormones.

Research Methods and ProceduresSubjects

Subjects included 1147 Mexican Americans (446 menand 701 women) distributed in 41 families (ranging in sizefrom 3 to 71 individuals). These families are among those

participating in the San Antonio Family Heart Study, abroader project designed to investigate the genetics of riskfactors for atherosclerosis, non-insulin-dependent diabetes(type 2 diabetes), and obesity (33). Probands and familymembers were ascertained without regard to obesity or anyother preexisting medical conditions. The ages ranged from18 to 92 years, with an average age of 40 years. Allprotocols were approved by the Institutional Review Boardat the University of Texas Health Science Center, SanAntonio, Texas.

Compared with non-Hispanic whites from the San Anto-nio area, the Mexican American population has a greatertendency to be overweight (34). Using a definition of over-weight based on BMI from the second National Health andNutrition Examination Survey (NHANES II; a BMI at age40 years �29.5 kg/m2 for men and �31.0 kg/m2 forwomen) (35), the prevalence of overweight individuals forthis Mexican American population was �55% for womenand 51% for men vs. 27% and 38% among non-Hispanicwhite women and men, respectively (34). Comparisons withNHANES II skinfold standards for U.S. whites and blacksconsistently placed the estimated means for both sexes inthe present sample between the 50th and 90th percentiles.

Phenotype and CovariatesLeptin was assayed using a commercially available ra-

dioimmunoassay kit (Linco Research, St. Charles, MO) (36)in serum samples collected after an overnight (�12-hour)fast. The within- and between-run coefficients of variation(CV) for this assay ranged from 3.4% to 8.3% and from3.6% to 6.2%, respectively. Covariates included were sex,age, diabetic status, use of diabetic medications, total fatmass (TFM; i.e., the absolute amount of body fat), percent-age of fat mass (FM; i.e., the relative amount of body fat pertotal body weight), serum testosterone, serum estradiol, andserum SHBG. Both TFM and percentage of FM were esti-mated by means of bioelectrical impedence using a Valhalla1990B body composition analyzer (Valhalla Scientific, SanDiego, CA) and Valhalla’s proprietary equations (37). Tes-tosterone, estradiol, and SHBG were assayed using com-mercially available RIA kits (Diagnostics Systems Labora-tory, Webster, TX). Means and SEs are reported in Table 1for the bioimpedence measures and the reproductive hor-mones.

Analytical MethodsUnivariate quantitative genetic analysis was done to par-

tition the phenotypic variance of serum leptin levels into itsadditive genetic and environmental variance componentsusing maximum likelihood variance decomposition meth-ods (38,39) implemented in the computer program SOLAR(Southwest Foundation for Biomedical Research, San An-tonio, TX) (40). This approach allows for the explicit test ofwhether correlations among family members for a given

Genetics of Sexual Dimorphism in Leptin Levels, Martin et al.

OBESITY RESEARCH Vol. 10 No. 1 January 2002 15

trait are solely the result of environmental factors, orwhether genetic effects are involved. A major strength ofthis approach is the ability to test for specific patterns ofinteraction, such as G � S.

The test for G � S interactions is based on hypothesesconcerning the nature of the variance–covariance relation-ship of a trait between male–female relative pairs(2,4,41,42). The expected genetic covariance between amale and female relative pair is defined as

COV�GM,GF� � 2��G�M,F��GM�GF

where � is the coefficient of kinship between the twoindividuals, �G(M, F) is the genetic correlation between theexpressions of the trait in the two sexes, and �GM and �GF

are the genetic SDs for men and women.In the absence of a G � S interaction (i.e., the null

hypothesis), the genetic correlation between male and fe-male relative pairs should be one (�G(M, F) � 1.0) and maleand female genetic SDs will be equivalent (�GM � �GF)(2,41). Conversely, if there is G � S interaction, the geneticcorrelation between the sexes, �G(M, F), will be signifi-cantly �1.0 and the genetic variances will not be equalbetween the sexes (�GM � �GF). Additionally, environ-ment by sex interaction (E � S) would be indicated bythe environmental variances not being equal between thesexes (�EM � �EF).

Using the likelihood ratio test, nested models are com-pared (43). Specifically, models in which �G is estimated,�GM and �GF are allowed to differ, and �EM and �EF areallowed to differ are compared with models that constrained�G to 1, �GM and �GF to be equal, and �EM and �EF to beequal, respectively. Three basic inferences concerning thenature of sex-based interactions can be made. Rejection ofthe model constraining the genetic correlation between thegroups to equal 1 (�G(M, F) � 1.0) implies that a differentgene or suite of genes contributes to the variance in leptinlevels in men and women. In contrast, rejection of the modelconstraining the genetic SDs of the groups to be equal(�GM � �GF) implies that the magnitude of the geneticeffect is different in the two sexes. Finally, rejection of the

model constraining the environmental SDs of the groups tobe equal implies that something in residual environmentinteracts with sex.

To address the issue of whether the sexual dimorphism seenin serum leptin levels is solely a function of women having onaverage more adipose tissue than men per body weight, weconducted analyses using percentage of FM and TFM asadditional covariates. The inclusion of these covariates yieldsparameter estimates for serum leptin levels that are conditionalon an individual’s percentage of FM or TFM.

To address whether the sexual dimorphism seen in serumleptin levels is a function of reproductive hormones, weconduced analyses using testosterone, estradiol, and SHBGas covariates. The inclusion of these covariates yields pa-rameter estimates for serum leptin levels that are conditionalon an individual’s level of testosterone, estradiol, or SHBG.

ResultsG � S Interaction with Age and Sex as Covariates

Table 2 provides the maximum likelihood estimates forthe mean leptin levels and their associated SEs for MexicanAmerican men and women. Leptin levels in women werenearly three times as high as those in men. In both sexes,there seemed to be a slight, but highly statistically signifi-cant (p � 0.01) age effect (�age) and age2 effect (�age2).We found no significant effect of diabetic status or use ofdiabetic medications on the expression of serum leptinlevels. Previous analyses of these data have not detected anyadditional environmental covariate effects, including, butnot limited to, smoking, menopausal status, use of oralcontraceptives, or hormone replacement therapy.

Table 3 provides the maximum likelihood estimates ofthe variance component parameters estimated with age andsex as covariates. As with the differences in the means,women displayed larger genetic (�G) and environmental(sE) SDs than men.

Table 4 summarizes the results of the likelihood ratiotests associated with the detection of sex-based interactions.The model, which held the genetic SDs to be equal (�GM ��GF), was rejected, resulting in an inference of G � Sinteraction. The model, which constrained the genetic cor-relation, �G(M, F) to 1, was rejected at p � 0.10, resulting inan inference of possible G � S interaction. Also the model,

Table 1. Means SE for body fat and reproductivehormones by sex

Men Women

Total body fat 74.50 0.93 63.85 0.43Percentage of body fat 21.09 0.48 35.65 0.34Testosterone 4.11 0.10 0.39 0.01Sex hormone-binding

globulin 22.31 1.01 42.25 1.29Estrogen 14.20 0.92 35.28 2.92

Table 2. Serum leptin levels and age and age2 effects(means SEs)

Men Women

Mean (ng/mL) 10.80 0.56 29.34 0.94�Age 0.08 0.02 0.14 0.04�Age2 0.004 0.002 0.004 0.002



which forced the environmental SDs in both sexes to beequal (�EM � �EF), was rejected, resulting in an inferenceof G � S interaction.

G � S Interaction with Age, Sex, and Body Fat asCovariates

Table 5 provides the maximum likelihood estimates ofthe variance component parameters for the models inwhich the percentage of FM and TFM are included ascovariates. Conditioning on either percentage of FM orTFM, men had lower genetic (�G) and environmental(sE) SDs than women.

Table 6 summarizes the results of the likelihood ratiotests associated with the detection of sex-based interactionswhen body fat covariates were included in the models.When either percentage of FM or TFM was included as acovariate, the models in which the genetic SDs of men andwomen were held equal (�GM � �GF) were not rejected.The models that constrained the genetic correlation betweenthe sexes, �G(M, F), to 1 were rejected, and in fact in bothcases the genetic correlation between the sexes, �G(M, F),was not significantly different from zero. Rejection of thegenetic correlation equaling 1 resulted in an inference ofG � S interaction. The models in which the sex-specificenvironmental SDs were forced to be equal (�EM � �EF) alsowere rejected, resulting in an inference of G � E interaction.

G � S Interaction with Age, Sex, and ReproductiveHormones as Covariates

Table 7 provides the maximum likelihood estimates ofthe variance component parameters for the models in which

testosterone, estradiol, and SHBG are included as covari-ates. Like the previous analyses, when conditioning oneither testosterone, estradiol, or SHBG, men had lowergenetic (�G) and environmental (�E) SDs than women.

Table 8 summarizes the results of the likelihood ratiotests associated with the detection of sex-based interactionswhen sex hormones were included as covariates. Whentestosterone was included as a covariate, neither the modelin which the genetic SDs of men and women were heldequal (�GM � �GF) or the model that constrained thegenetic correlation between the sexes, �G(M, F), to 1 wasrejected, resulting in an inference of no G � S interaction.The model in which the sex-specific environmental SDswere forced to be equal (�EM � �EF) was rejected, result-ing in an inference of E � S interaction. When conditioningon estrogen levels, the model forcing the genetic SDs ofmen and women to be equal (�GM � �GF) was not rejected,whereas the model that constrained the genetic correlationbetween the sexes (�G(M, F)) to 1 was rejected. Rejection ofthe genetic correlation equaling 1 resulted in an inference ofG � S interaction. The model in which the sex-specificenvironmental SDs were forced to be equal (�EM � �EF)was rejected, resulting in an inference of E � S interaction.When SHBG was included as a covariate, both the modelthat constrained the genetic SDs of men and women to beequal (�GM � �GF) and the model that constrained thegenetic correlation between the sexes, �G(M, F), to 1 wererejected, resulting in an inference of G � S interaction. Themodel that constrained the environmental SDs of men and

Table 3. Variance components (SE) for serum lep-tin levels

Men Women

�G 4.72 0.90 12.34 1.24�E 7.08 0.58 10.66 1.10h2 0.31 0.57�G(M,F) 0.64 0.21

Table 4. Test of sex-specific interactions in serumleptin levels

Model* p value Inference

�GM � �GF �0.00001 G � S interaction�EM � �EF 0.01537 E � S interaction�G(M,F) � 1 0.0517 G � S interaction

* All tests have 1 degree of freedom.

Table 5. Variance components with SE for serum leptin levels when the percentage of fat mass or total fat mass(TFM) is included as a covariate in the analysis (� percentage of fat mass, 0.41 0.03; �TFM, 0.36 0.02)

Percentage of fat mass TFM

Men Women Men Women

�G 4.86 1.02 7.60 1.46 4.02 0.88 6.58 1.44�E 6.64 0.70 10.66 0.94 6.20 0.58 10.00 0.86h2 0.35 0.34 0.30 0.30�G(M,F) 0.18 0.27 0.21 0.29



women to be equal (�EM � �EF) was rejected, resulting inan inference of E � S interaction.

DiscussionHumans, as well as most of the anthropoid primates,

exhibit pronounced sexual dimorphism in overall body size,with men typically being larger than women. However,there is also a pattern of sexual dimorphism in total fataccumulation (44,45), with women having a larger amountof fat relative to body weight than men (46,47). Addition-ally, there is marked sexual dimorphism in adipose tissuedistribution (45,46,48,49). In women, excess fat tends to beconcentrated on the lower body (e.g., hips and thighs),whereas in men it tends to accumulate on the upper body(e.g., abdomen) (45,46,49). We have previously reportedpronounced sexual dimorphism in fat accumulation anddistribution as assessed through a variety of morphometriccharacters in this sample of Mexican Americans and haveshown that the sexual dimorphism exhibited in these traitshas a substantial genetic component (4,37).

In this study, we have identified a strong genetic compo-nent in sexual dimorphism in leptin levels and have dem-onstrated that this genetic sexual dimorphism componentpersists even after conditioning on percentage of FM orTFM. Although several elaborate methods exist for theprecise measurement of body fat (e.g., computed tomogra-

phy or hydrostatic weighing), these generally are not feasi-ble for large-scale family-based studies. Traditionally, forlarge studies, body composition has been assessed usingindirect proxies, such as height, weight, and skinfolds. Bio-impedence measurements offer a more direct assessment ofan individual’s total body fat while still being suitable forsuch large studies. When either percentage of FM or TFMwas included as a covariate, we found clear evidence of bothG � S and E � S interactions determining serum leptinlevels. The existence of a significant G � S interactionindicates that the genes involved in determining leptin lev-els are expressed differently in men and women. In ananalogous manner, the detection of a significant E � Sinteraction is an indication that unmeasured effects differ-entially influence the expression of leptin in men andwomen. Additionally, the fact that the genetic correlationbetween the sexes, �G(M, F), is significantly less than 1,indicates that a subset of genes unique to one or both sexescontributes to the expression of variation in this trait. Theseresults point to a large genetic component in the expressionof sexual dimorphism in serum leptin levels, which arisesfrom differential contribution as well as expression ofgenes. In other words, this suggests that not only are dif-ferent genes involved in the expression of serum leptinlevels in the sexes, but also that genes common to leptinexpression in both sexes may be expressed differentially

Table 6. Test of sex-specific interactions in serum leptin levels corrected for percentage of fat mass (FM) or totalfat mass (TFM)

Model*

FM as a covariate TFM as a covariate

p value Inference p value Inference

�GM � �GF 0.1467 Male �G � female �G 0.1604 Male �G � female �G�EM � �EF 0.0016 E � S interaction 0.0011 E�Sinteraction�G(M,F) � 1 0.0060 G � S interaction 0.0119 G�Sinteraction


Table 7. Variance components with SE for serum leptin levels when testosterone, estrogen, or sex hormone-binding globulin (SHBG) is included as a covariate in the analysis (�)

Testosterone Estrogen SHBG

Men Women Men Women Men Women

�G 2.53 0.85 4.36 0.92 3.28 0.87 4.71 0.91 2.58 0.42 4.36 0.49�E 3.04 0.61 4.83 0.67 2.68 0.84 4.64 0.72 2.96 0.32 4.49 0.39h2 0.41 0.45 0.60 0.51 0.43 0.48�G(M,F) 0.68 0.32 0.53 0.26 0.62 0.18



depending on the individual’s sex. The existence of envi-ronment by sex interactions influencing the expression ofserum leptin levels, also suggests that men and women reactdifferentially to environmental factors. All of these findingssupport the fact that the sexual dimorphism seen in leptinlevels is not simply explained by differences in total adi-posity between the sexes, but rather is the result of differ-ential expression of genes by sex as well as the effects ofsets of sex-specific genes.

Given the increasing evidence of leptin’s potential role inreproduction, our results could reflect differences in geneticinfluences on the reproductive physiology in each sex.Therefore, we also tested for the significance of G � Sinteractions when testosterone, estradiol, or SHBG wereincluded as covariates. When estradiol or SHBG were in-cluded as covariates, we found clear evidence for G � Sinteractions determining serum leptin levels. However,when testosterone was included as a covariate, no signifi-cant G � S interaction was detected. This suggests that thesexual dimorphism seen in leptin levels may be attributed tosex differences in testosterone levels. Previous research hasdemonstrated an inverse relationship between leptin andtestosterone levels (7,17,18,50,51). Furthermore, severalstudies have implicated testosterone in the regulation ofPOMC expression (32,52–54), which has been identified asa potential quantitative trait locus for leptin levels (12).

Although previous studies have demonstrated that estra-diol also has a suppressive effect on POMC levels (31), ourresults suggest that estradiol is not the cause of the sexualdimorphism in leptin levels. These results are supported bythe fact that in postmenopausal women, leptin levels are stillhigher compared with those of men (17).

Sexual dimorphism in serum leptin levels has beenwidely noted. Given the ubiquitous nature of this observa-tion, it is reasonable to assume that the genotype-by-sexinteraction in serum leptin levels is not population-specific.Moreover, given the work by Elbers et al. (50) demonstratesthat suppression of testosterone in men substantially in-creases serum leptin levels, similar to what is seen inwomen, and it is likely that testosterone could mediate such

genotype-by-sex effects in other populations. However, ad-ditional studies are required to demonstrate the universalnature of such interactions. Future research should alsoidentify the genes responsible for influencing leptin levelsand determine the sex-specific factors with which theyinteract to produce the sexual dimorphism in this and per-haps other obesity-related traits.

AcknowledgmentsThis work was supported by National Institutes of Health

Grants GM15803, HL45522, DK44297, and MH59890. Wethank Edgar Rodriguez for technical assistance in conduct-ing the leptin assays.

References1. Falconer DS. Introduction to Quantitative Genetics. New

York, NY: John Wiley; 1981.2. Eisen EJ, Legates JE. Genotype-Sex interaction and the

genetic correlation between the sexes for body weight in Musmusculus. Genetics. 1966;51:611–23.

3. Griffing B. Influence of sex on selection: contribution ofautosomal genotypes having different values in the two sexes.Aust J Biol Sci. 1966;19:593–606.

4. Comuzzie AG, Blangero J, Mahaney MC, Mitchell BD,Stern MP, MacCluer JW. The quantitative genetics of sexualdimorphism in body fat measurements. Am J Hum Biol. 1993;5:725–34.

5. Campfield LA, Smith FJ, Guisez Y, Devos R, Burn P.Recombinant mouse OB protein: evidence for a peripheralsignal linking adiposity and central neural networks. Science.1995;269:546–9.

6. Halaas JL, Gajiwala KS, Maffei M, et al. Weight-reducingeffects of the plasma protein encoded by the obese gene.Science. 1995;269:543–46.

7. Pelleymounter MA, Cullen MJ, Baker MB, Hecht R,Boone T. Effects of the obese gene product on body weightregulation in ob/ob mice. Science. 1995;269:540–3.

8. Considine RV, Sinha MK, Heiman ML, et al. Serum im-munoreactive leptin concentrations in normal-weight andobese humans. N Engl J Med. 1996;334:292–5.

Table 8. Test of sex-specific interactions in serum leptin levels corrected for testosterone, estrogen, or sexhormone-binding globulin (SHBG)

Model*

Testosterone Estrogen SHBG

p value Inference p value Inference p value Inference

�GM � �GF 0.1324 �GM � �GF 0.2348 �GM � �GF 0.0051 G � S interaction�EM � �EF 0.0543 G � E interaction 0.0812 �EM � �EF 0.0032 E � S interaction�G(M,F) � 1 0.1862 �G(M,F) � 1 0.0212 G � S interaction 0.0188 G � S interaction




9. Maffei M, Halaas J, Ravussin E, et al. Leptin levels inhuman and rodent: measurement of plasma leptin and ob RNAin obese and weight-reduced subjects. Nat Med. 1995;1:1155–61.

10. Zhang Y, Proenca R, Barone M, Leopold L, Freidman JM.Positional cloning of the mouse obese gene and its humanhomologue. Nature. 1994;372:425–32.

11. Comuzzie AG, Braxton BD, Hixson JE, Dyer TD, Mac-Cluer JW, Blangero J. The ob locus influences serum leptinlevels in Mexican Americans. Diabetes. 1996;45:30A.

12. Comuzzie AG, Hixson JE, Almasy L, et al. A major quan-titative trait locus determining serum leptin levels and fat massis located on human chromosome 2. Nat Genet. 1997;15:273–6.

13. Haffner SM, Gingerich RL, Miettinen H, Stern MP. Leptinconcentrations in relation to overall adiposity and regionalbody fat distribution in Mexican Americans. Int J Obes RelatMetab Disord. 1996;20:904–8.

14. Ostlund RE Jr, Yang JW, Klein S, Gingerich R. Relationbetween plasma leptin concentration and body fat, gender,diet, age, and metabolic covariates. J Clin Endocrinol Metab.1996;81:3909–13.

15. Rosenbaum M, Nicolson M, Hirsch J, et al. Effects ofgender, body composition, and menopause on plasma concen-trations of leptin. J Clin Endocrinol Metab. 1996;81:3424–7.

16. Saad MF, Damani S, Gingerich RL, et al. Sexual dimor-phism in plasma leptin concentration. J Clin EndocrinolMetab. 1997;82:579–84.

17. Baumgartner RN, Waters DL, Morley JE, Patrick P, Mon-toya G, Garry PJ. Age-related changes in sex hormonesaffect the sex difference in serum leptin levels independentlyof changes in body fat. Metabolism. 1999;48:378–84.

18. Vettor R, De Pergola G, Pagano C, et al. Gender differencesin serum leptin in obese people: relationships with testoster-one, body fat distribution, and insulin sensitivity. Eur ClinInvest. 1997;27:1016–24.

19. Isidori AM, Caprio M, Strollo F, et al. Leptin and androgensin male obesity: evidence for leptin contribution to reducedandrogen levels. J Clin Endocrinol Metab. 1999;84:3673–80.

20. Denti L, Pasolini G, Sanfelici L, et al. Aging-related declineof gonadal function in healthy men: correlation with bodycomposition and lipoproteins. J Am Geriatr Soc. 2000;48:51–8.

21. Tchernof A, Despres JP. Sex steroid hormones, sex hor-mone-binding globulin, and obesity in men and women. HormMetab Res. 2000;32:526–36.

22. Apter D. Leptin in puberty. Clin Endocrinol (Oxf). 1997;47:175–6.

23. Macut D, Micie D, Pralong FP, Bischof P, Campana A. Isthere a role for leptin in human reproduction? Gynecol Endo-crinol. 1998;12:321–6.

24. Holness MJ, Munns MJ, Sugden MC. Current conceptsconcerning the role of leptin in reproductive function. MolCell Endocrinol. 1999;157:11–20.

25. Chehab FE, Lim ME, Lu RH. Correction of the sterilitydefect in homozygous obese female mice by treatment withthe human recombinant leptin. Nat Genet. 1996;12:318–20.

26. Chehab FF, Mounzih K, Lu RH, Lim ME. Early onset ofreproductive function in normal female mice treated withleptin. Science. 1997;275:88–90.

27. Masuzaki H, Ogawa Y, Sagawa N, et al. Nonadipose tissueproduction of leptin: leptin as a novel placental-derived hor-mone in humans. Nat Med. 1997;3:1029–33.

28. Schurbring C, Blum WF, Kratzsch J, Deutscher J, KiessW. Leptin, the ob gene product, in female health and disease.Eur Obstet Gynecol. 2000;88:121–7.

29. Watanobe H, Yoneda M, Kohsaka A, Kakizaki Y, Suda T,Schioth HB. Normalization of circulating leptin levels byfasting improves the reproductive function in obese OLETFfemale rats. Neuropeptides. 2001;35:45–9.

30. Caprio M, Isidori AM, Carta AR, Moretti C, Dufau ML,Fabbri A. Expression of functional leptin receptors in rodentLeydig cells. Endocrinology. 1999;140:4939–47.

31. Matera C, Wardlaw SL. Aromatization is not required forandrogen induced changes in the proopiomelanocortin gene ex-pression in the hypothalamus. Mol Brain Res. 1994;27:275–80.

32. Majdoubi ME, Ramaswamy S, Sahu A, Plant TM. Effectsof orchidectomy on levels of mRNAs encoding gonadotropin-releasing hormone and other hypothalamic peptides in theadult male rhesus monkey (Macaca mulatta). J Neuroendo-crinol. 2000;12:167–76.

33. Mitchell BD, Kammerer CM, Blangero J, et al. Genetic andenvironmental contributions to cardiovascular risk factors inMexican Americans—the San Antonio Family Heart Study.Circulation. 1996;94:2159–70.

34. Hazuda HP, Mitchell BD, Haffner S, Stern MP. Obesity inMexican American subgroups: findings from the San AntonioHeart Study. Am J Clin Nutr. 1991;53:1529S–34S.

35. Frisancho AR. Anthropometric Standards for the Assessmentof Growth and Nutritional Status. Ann Arbor, MI: Universityof Michigan Press; 1990.

36. Ma ZM, Gingerich RL, Santiago JV, Klein S, Smith CH,Landt M. Radioimmunoassay of leptin in human plasma. ClinChem. 1996;42:942–6.

37. Comuzzie AG, Blangero J, Mahaney MC, et al. Major genewith sex-specific effects influences fat mass in MexicanAmericans. Genet Epidemiol. 1995;12:475–88.

38. Hopper JL, Mathews JD. Extensions to multivariate normalmodels for pedigree analysis. Ann Hum Genet. 1982;46:373–83.

39. Lange K, Boehnke M. Extensions to pedigree analysis: co-variance components models for multivariate traits. Am J MedGenet. 1983;14:513–24.

40. Almasy L, Blangero J. Multipoint quantitative trait-linkageanalysis in general pedigrees. Am J Hum Genet. 1998;62:1198–211.

41. Robertson A. The sampling variance of the genetic correla-tion coefficient. Biometrics. 1959;15:469–85.

42. Towne B, Blangero J, Mott GE. Genetic analysis of sexualdimorphism in serum apo A1 and HDL-C concentrations inbaboons. Am J Primatol. 1992;27:107–17.

43. Self SG, Liang KY. Asymptotic properties of maximumlikelihood ratio tests under nonstandard conditions. J Am StatAssoc. 1987;82:605–10.

44. Dixon AK. Abdominal fat assessed by computed tomography:sex differences in distribution. Clin Radiol. 1983;34:189–91.



45. Vague J. La differenciation sexuelle, facteur determinant desformes de l’obesite. Presse Med. 1947;30:339–40.

46. Krotkiewski M, Bjorntorp P, Sjostrom L, Smith U. Impactof obesity on metabolism in men and women: importance ofregional adipose tissue distribution. J Clin Invest. 1983;72:1150–62.

47. Sjostrom L, Smith U, Krotkiewski M, Bjorntorp P. Cellu-larity in different regions of adipose tissue in young men andwomen. Metab Clin Exp. 1972;21:1143–53.

48. Bouchard C, Despres JP, Mauriege P. Genetic and nonge-netic determinants of regional fat distribution. Endocrinol Rev.1993;14:72–93.

49. Vague J, Vague P, Jubelin J, et al. Fat distribution, obesitiesand health: evolution of concepts. In: Bouchard C, JohnstonFE, eds. Fat Distribution during Growth and Later HealthOutcomes. New York, NY: Alan Liss; 1988, pp. 9–41.

50. Elbers JM, Asscheman H, Seidell JC, Frolich M,Meinders AE, Gooren LJ. Reversal of the sex differences

in serum leptin levels upon cross-sex hormone administra-tion in transsexuals. J Clin Endocrinol Metab. 1997;82:3267–70.

51. Soderberg S, Olsson T, Eliasson M, et al. A strong associ-ation between biologically active testosterone and leptin innon-obese men and women is lost with increasing (central)adiposity. Int J Obes Relat Metab Disord. 2001;25:98–105.

52. Blum M, Roberts JL, Wardlaw SL. Androgen regulation ofproopiomelanocortin gene expression and peptide content inthe basal hypothalamus. Endocrinology. 1989;124:2283–8.

53. Chowen-Breed J, Fraser HM, Vician L, Damassa DA,Clifton DK, Steiner RA. Testosterone regulation of proopio-melanocortin messenger ribonucleic acid in the arcuate nu-cleus of the male rat. Endocrinology. 1989;124:1697–702.

54. Hileman SM, Lubbers LS, Petersen SL, Kuehl DE, ScottCJ, Jackson GL. Influence of testosterone on LHRH release,LHRH mRNA and proopiomelanocortin mRNA in malesheep. J Neuroendocrinol. 1996;8:113–21.



Documents

Leptin's Sexual Dimorphism Results from Genotype by Sex Interactions Mediated by Testosterone