Free and receptor-bound leptin may be regulated by different mechanisms. Genes that influence the concentration of these fractions may have an important functional bearing. We determined circulating leptin receptor concentrations, bound as well as free leptin concentrations, and body composition in 24 monozygotic (MZ) and in 22 dizygotic (DZ) twin pairs. Bound leptin and leptin receptor concentrations were inversely correlated with body fat content. Free leptin concentrations were directly correlated with body fat content. The correlations in age- and sex-adjusted free leptin, bound leptin, and leptin receptor concentrations were higher between MZ twins than between DZ twins. Adjusted heritability (h2) estimates were 0.28 for free leptin, 0.73 for bound leptin, and 0.55 for leptin receptor. The genetic correlation with body fat was −0.58 for the leptin receptor, −0.20 for bound leptin, and 0.93 for free leptin. Our data are consistent with a strong genetic influence on leptin receptor and bound leptin and a weaker genetic influence on free leptin concentrations. The same genes that lower bound leptin and leptin receptor concentrations may increase fat mass or vice versa.
- twin study
- bound leptin
leptin decreases food intake through an increase in satiety and increased energy expenditure (22). The pivotal role of leptin for the maintenance of metabolic homeostasis is illustrated by patients with leptin deficiency or leptin receptor deficiency. Both conditions are associated with massive obesity (6, 17). Leptin deficiency or functional mutations of the leptin receptor gene are rare conditions in humans. Indeed, most people exhibit an appropriate increase in plasma leptin concentrations as body fat content increases (7).
Given the importance of leptin in body weight regulation, even subtle genetic alterations in the leptinergic system may be physiologically and clinically relevant. One mechanism that may influence leptin's biological activity is protein binding. Leptin circulates in the blood in a free and a protein-bound form. The major leptin binding protein is the soluble leptin receptor (14). The bound leptin concentration is related to the concentration of the soluble receptor. Free leptin, bound leptin, and the leptin receptor can be secreted from human subcutaneous adipocytes (4).
Usually, protein binding attenuates the activity of hormones and medications. The free concentration determines the biological effect. In contrast, binding to the circulating leptin receptor may augment leptin's biological activity. It has been suggested that binding to the circulating receptor might delay the clearance of leptin from the circulation (9). Another rather speculative explanation is that binding to the circulating receptor might alter the transport of leptin across the blood-brain barrier. Indeed, overexpression of the circulating leptin receptor in leptin-deficient mice increases the response to exogenous leptin (9). In humans, resting energy expenditure appears to be more closely related to bound leptin levels than to free leptin (3). We demonstrated that in normal weight men, resting sympathetic activity is correlated with bound leptin but not with free leptin concentrations (24). Thus genes that influence the concentration of the soluble leptin receptor may have an important functional bearing. However, the genetic variability of this measure is not known. We determined the heritability of circulating leptin receptor concentrations and bound as well as free leptin concentrations in a cohort of normal twins.
We studied 46 twin pairs; 24 twin pairs were monozygotic (MZ, age 35 ± 11 yr, 14 men, 34 women), and 22 twin pairs were dizygotic (DZ, age 30 ± 8 yr, 11 pairs female/male, 4 pairs male/male, 7 pairs female/female). The twin pairs were recruited from the Berlin Twin Register (2). We obtained a medical history and conducted a physical examination prior to study entry. Only healthy persons were included in the study. Persons ingesting medications except birth control pills were excluded from the study. Zygosity was determined using five microsatellite markers coamplified by polymerase chain reaction (1). The probability of a dizygotic twin pair to share all marker alleles by chance is 0.006. The overall rate of correct classification is >99%. Written informed consent was obtained before study entry as required by the institutional review board.
Subjects were evaluated at the Franz Volhard Clinical Research Center after an overnight fast. Height and weight measurements were taken. Skinfold thickness was measured with a caliper (triceps, biceps, back, hip) to estimate body fat content. Percent body fat was calculated by the following formulas(8). For women, and for men In addition, we assessed body composition with body impedance analysis (Quadscan 4000, Bodystat, Isle of Man, UK). Relative body fat content, relative lean body mass, and dry lean body mass were computed. Supine blood pressure and heart rate were determined after a resting phase of at least 15 min with an automated brachial blood pressure cuff (Dinamap). Venous blood samples were obtained from an antecubital venous catheter after an overnight fast. Before blood sampling, the subjects rested in the supine position for at least 30 min. All studies were conducted in a quiet room at 20°C during morning hours.
Free and bound leptin measurements.
Leptin receptor, free leptin, and bound leptin concentrations were measured by specific radioimmunoassay systems as described previously (15).
Statistics and quantitative genetics.
Statistical analysis was conducted using the SPSS program. All data are expressed as means ± SD. Relationship between parameters was assessed by correlation analysis. Differences of mean group values were tested with unpaired t-test. A value of P < 0.05 was considered to be statistically significant. Parameters of the quantitative genetic models were estimated by structural equation modeling using the Mx program developed by Neale (19). The variability of any given phenotype within a population can be decomposed into additive genetic influences (VaraddGen, A), environmental influences shared by the twins within a family (VarsharedEnv, C), and effects of random environment (Varenv, E): For MZ and DZ, the covariance (Cov) of their phenotype is given by: Heritability analysis in twin studies can estimate genetic variability as well as two environmental influences, shared and nonshared environmental influences. These values estimate the relative amount of the variable's influence on interindividual differences up to a sum of one. No distinction between additive and dominant genetic was made in this study, as a general proof of genetic variance rather than a precise estimate of such influences was the goal of this study. We were concerned that the sample size of our study may have been insufficient to distinguish between additive and dominant genetic influences. Accordingly broad sense heritability is reported only. The estimate was based on single estimates or the sum of potential additive and nonadditive genetic influences where appropriate. Genetic as well as environmental effects were estimated by the best fitting model as selected by the χ2 value. Nested models were tested, starting with the full ACE model; A and/or C were constrained to zero. The significance of a model component is tested by the drop of fit as measured by the difference in χ2 (20). Adjustments for age, body mass index, and blood pressure were done by multiple linear regression using unstandardized residuals. In case of significant deviations from normal distribution, the appropriate transformations were applied. To test if phenotypic correlations between different measures (e.g., body fat content and leptin receptor) are due to common underlying genetic or environmental factors, pairwise genetic correlations were computed based on estimates of genetic covariance from a bivariate cholesky model. The significance of genetic or environmental covariance was tested by the drop of model fit in a submodel constraining the respective estimate to zero. Genetic correlation (CorrGen)is defined by: where CovGen1Gen2 is the shared genetic influence, and VarGen1 and VarGen2 are the heritabilities of the traits separately.
As in conventional correlation analysis, this measure may vary between −1 and 1. It defines the subfraction of the genetic influence that is shared between traits; thus it can be greater than the heritability of either trait (20).
MZ and DZ twins had similar age, body mass index, resting blood pressure, and resting heart rate. Furthermore, free leptin, bound leptin, and leptin receptor concentrations were similar in both groups (Table 1). A larger proportion of the DZ twins was male compared with MZ twins. This difference is reflected in a significant difference in body composition between MZ and DZ twins. Bound leptin and leptin receptor concentrations were negatively correlated with relative body fat content as determined by impedance analysis (Table 2). In contrast, free leptin concentrations were positively correlated with body fat content and body mass index. The relationship between body mass index and free leptin concentrations is illustrated in Fig. 1. Body mass index and free leptin concentrations were correlated with each other (Table 2). However, at a given body mass index, free leptin concentrations were increased in women compared with men (Fig. 1, Table 3). Bound leptin and leptin receptor concentrations were slightly increased in men compared with women. Free leptin concentrations were increased in women. Age also influenced some leptin measurements (Table 2). A significant relation between height and free leptin as reported by Jenkins et al. (10) for total leptin concentrations disappeared after adjustment for sex.
As expected, we observed a genetic influence on body mass index and body composition. Given the sex- and age-related effect, we had to adjust our leptin measurements for gender and age. The correlation in age- and sex-adjusted bound leptin concentrations was higher between MZ twins than between DZ twins (Fig. 2). We observed a similar relationship for leptin receptor concentrations (Fig. 3). The heritability estimation suggested a strong genetic influence on bound leptin and leptin receptor concentrations (Table 1). The heritability estimate was weaker for unadjusted free leptin concentrations. Model fit for bound leptin, leptin receptor, and free leptin for the best fitting model was 0.01, 2.87, and 0.01, respectively, all with 2 degrees of freedom. The heritability estimate decreased further after adjustment for age and gender (Table 1, Fig. 4).
Leptin receptor and bound leptin showed significant heritability as well as strong correlation with the body fat content. Therefore, we tested the hypothesis of common underlying genes for leptin traits and impedance measurements. For the leptin receptor the genetic correlation with body fat was −0.58 (P < 0.01); for bound leptin we observed a strong trend toward common genes with a genetic correlation of −0.20 (P < 0.07). The genetic correlation between free leptin concentrations and body fat content was particularly strong (r = 0.93). Subject-specific environmental factors may also be correlated between body fat content and leptin traits. The environmental correlation of body fat content with free leptin was 0.93. No such correlation was found between body fat content and bound leptin or leptin receptor concentrations. For bound leptin and leptin receptor with a significant phenotypic correlation, we estimated genetic correlation at 0.52 (P < 0.01) with no significant environmental correlation. Given the absence of a phenotypical correlation between free leptin and bound leptin or leptin receptor concentrations, we did not calculate a genetic or environmental correlation between these measurements.
Twin studies have been used extensively to characterize the interaction of genetic and environmental factors on cardiovascular and metabolic phenotypes. We employed the twin approach to estimate the heritability of leptin receptor, free leptin, and bound leptin plasma concentrations. Because of a gender effect on all these measurements, we determined heritability before and after statistical adjustment for gender. Our data are consistent with a strong genetic influence leptin receptor and bound leptin concentrations. The genetic influence on free leptin concentrations may be weaker. Indeed, the 95% confidence interval for the heritability estimate of free leptin adjusted for gender included 0. We observed a strong gender difference in free leptin concentrations. The number of subjects in our study was too small to allow for a gender-specific calculation of heritabilities. We propose that the gender-specific estimation of free leptin heritability should be repeated in larger twin cohorts.
The existence of different “leptins” makes it difficult to compare our data with studies that did not differentiate bound and free leptin concentrations. Several studies assessed the heritability of total leptin concentrations in dizygotic and monozygotic twins (10, 11, 18, 25). These studies showed a strong genetic influence on leptin concentrations. Leptin concentrations were highly correlated with body mass index or fat mass. Heritability estimates for plasma leptin concentrations were attenuated after adjustment for body mass index (11, 18). These findings suggest a common genetic influence on total plasma leptin levels and body mass index. The issue was further elaborated in another study in monozygotic twins who were discordant for body weight (21). The average weight difference between siblings was 18 kg. Leptin levels were threefold greater in obese twins than in lean twins. The authors concluded that the increase in plasma leptin levels in obese persons is independent of the genetic background (21). One study used a physiological approach to study the heritability of leptin (10). The authors concluded that it is unlikely that leptin secretion rate is heritable after variability in fat mass and body size have been taken into account. Our interpretation of the data is that the genetic influence on total leptin is to a large part secondary to a genetic effect on body weight. Hence, interindividual differences in total leptin concentration are the consequence rather than the cause of body weight differences.
We hypothesized that in the general population, genetic mechanisms that modulate leptin's biological activity may have an important role in the regulation of leptinergic responses. Recent studies suggest that the soluble leptin receptor may be such a modulator. Indeed, overexpression of the soluble leptin receptor enhanced the response to exogenous leptin in leptin-deficient mice (9). The phenomenon may be explained in part by improved penetration of bound leptin across the blood-brain barrier (3). Improved penetration may explain why cerebrospinal bound leptin concentrations increase in parallel with increasing plasma concentrations. In contrast, cerebrospinal free leptin concentrations exhibit a saturation at higher plasma levels (3). Resting metabolic rate per kilogram body weight is more closely related to bound leptin concentrations than to free leptin concentrations (3). In a subsequent study in normal weight men we assessed the relationship between free and bound leptin concentrations and sympathetic vasomotor tone (24). Receptor-bound rather than free leptin levels were correlated with basal sympathetic vasomotor tone (24). The absence of a correlation with free leptin concentrations suggests that the findings are not secondary to interindividual differences in fat mass. Genes that influence the concentration of the circulating leptin receptor and, thus, bound leptin concentrations may be important for the regulation of basal metabolism and sympathetic activity. For that reason, we determined plasma concentrations of the circulating leptin receptor, free leptin, and receptor-bound leptin in monozygotic and in dizygotic twins.
Similar to previous studies (3), free leptin concentrations were positively correlated with body mass index and fat mass. In contrast, leptin receptor and bound leptin concentrations were negatively correlated with body mass index and fat mass. The inverse relationship between plasma levels and body composition between bound and free leptin might indicate different underlying regulatory mechanisms. An alternative explanation is that leptin downregulates the soluble leptin receptor (13). Leptin receptor, free leptin, and receptor-bound leptin were more closely correlated between monozygotic twins than they were between dizygotic twins. The heritability estimation indicated a genetic influence on leptin receptor, receptor-bound leptin, and to a lesser degree on free leptin.
To test whether phenotypic correlations between body composition and leptin measurements are due to common underlying genetic factors, we computed so-called genetic correlations. We found significant negative genetic correlations between body fat content and bound leptin and leptin receptor concentrations. Thus the same genes that lower bound leptin and leptin receptor concentrations may increase fat mass or vice versa. The genetic correlation between free leptin and fat mass was positive. Thus similarly to total leptin concentrations (10, 11, 18, 21), free leptin concentration may be determined by genes that influence fat mass. Studies in twins who are discordant for body weight might be useful to assess the causality of the relationship.
Our study necessarily has some limitations. One potential weakness is that we characterized genetic influences on leptin receptor, free leptin, and bound leptin concentrations in a cohort of healthy subjects rather than in patients with pathological adiposity. However, genes involved in Mendelian diseases were shown to act as quantitative trait loci in the general population, supporting the close relationship between physiological and pathological processes (5, 12). Furthermore, sample size was insufficient to obtain precise heritability estimates or study gender-specific heritability. The strong mean differences in bound and free leptin between male and female twins may reflect genuine effects of differential gene expression as well as secondary effects of the differences in body composition. Further studies are planned to address this problem. Another potential limitation is the fact that the diurnal variation of free leptin concentration is greater than the diurnal variation of bound leptin concentrations (16). The greater variability may lead to a systematic underestimation of the genetic effect.
Despite these issues we suggest that leptin receptor and bound leptin concentrations are inherited. The heritability estimate for free leptin concentration was borderline in significance after adjustment for gender. Because free and bound leptin may elicit different biological responses, our findings in healthy twins may have physiological and clinical implications. For example, insulin resistance and abdominal obesity are associated with low soluble leptin receptor concentrations and low ratio between bound leptin and free leptin independently of fat mass (23). Elucidation of the genes that influence the relative amount of free and receptor-bound leptin may provide important new insight in the pathophysiology of obesity and obesity-associated disorders.
J. Jordan is supported by a grant-in-aid from the Deutsche Forschungsgemeinschaft. A. Busjahn is supported by a grant-in-aid from the Bundesministerium für Bildung und Forschung (InnoRegio)
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society