In utero gender dimorphism of adiponectin reflects insulin sensitivity and adiposity of the fetus

Abstract

Circulating adiponectin reflects the degree of energy homeostasis and insulin sensitivity of adult individuals. Low abundance of the high molecular mass multimers (HMW), the most active forms mediating the insulin-sensitizing effects of adiponectin, is indicative of impaired metabolic status. The increase in fetal adiponectin HMW compared with adults is a distinctive features of human neonates. In order to further understand the functional properties of adiponectin during fetal life, we have evaluated the associations of adiponectin with insulin sensitivity, body composition and gender. Umbilical cord adiponectin, adiponectin complexes and metabolic parameters were measured at term by elective Cesarean section. The associations between adiponectin, measures of body composition and insulin sensitivity were evaluated in relation to fetal gender in 121 singleton neonates. Higher total adiponectin concentrations in females compared with male fetuses (34.3±9.5 vs 24.9±8.6, p<0.001) were associated with a 3.2-fold greater abundance in circulating HMW complexes (0.20±0.03 vs 0.08±0.03, p<0.001, n=9). Adiponectin was positively correlated with neonatal fat mass (r= 0.27, p< 0.04) and percent body fat in female fetuses (r= 0.28, p<0.03) and with lean mass in males (r= 0.28, p<0.03). There was no significant correlation between cord adiponectin and fasting insulin concentrations or fetal insulin sensitivity as estimated by HOMA-IR. The gender dimorphism for plasma adiponectin concentration and complex distribution first appears in utero. In sharp contrast to the inverse correlation found in adults, the positive relationship between adiponectin and body fat is a specific feature of the fetus.

INTRODUCTION

Adiponectin is a circulating hormone produced exclusively by adipose tissue in adult humans and rodents (1). It is secreted under trimeric subunits that can be separated by applying stringent reducing and denaturing conditions in vitro. Once released in the systemic circulation adiponectin aggregates into multimeric complexes of various molecular mass consisting of 3 to 18 monomers. The so-called low molecular weigh (LMW) hexamers and high molecular weight (HMW) 12–18 oligomers are the main circulating forms which may account for up to 80 percent of total adiponectin (2, 3). Each oligomeric isoform elicits distinct biological action(s) in the target tissues. The HMW oligomers are the major active form mediating the peripheral insulin-sensitizing effects, whereas the central actions are attributed primarily to the LMW and trimeric oligomers (4). A decrease in adiponectin concentrations is a hallmark of the metabolic transition between insulin sensitive and insulin resistant states. The insulin resistance of obesity, type 2 and gestational diabetes is associated with lower concentrations of total plasma adiponectin (5–7). Similarly, decreased adiponectin levels are observed in late pregnancy reflecting the maternal insulin resistance which is greatest during the third trimester (8). The lower abundance of the HMW complexes in several situations of insulin resistance has pointed to HMW as better markers of insulin resistance than total plasma adiponectin (9–11).

Changes in adiponectin concentrations in response to metabolic effectors are regulated at the molecular level. The production of adiponectin by the adipocytes is controlled at the level of gene expression through specific promoter regions (12,13). Additionally, complex post-translational modifications involving extensive hydroxylation and glycosylation reactions regulate the assembly and release of oligomers in the systemic circulation (14). However, the contribution of adipose tissue mass to the regulation of adiponectin production following changes in BMI, weight gain or body composition is not well understood. The inverse relationship between adiponectin concentrations and adipose tissue mass is paradoxical for an adipocyte derived hormone. Indeed, the negative correlation between adiponectin and adiposity is in sharp contrast to other adipocytokines such as leptin whose circulating concentrations are positively correlated with adipose tissue mass (15). The hormonal regulation of adiponectin secretion and circulating concentrations is mediated through insulin and testosterone action (16, 17). Testosterone has the potential for contributing to the gender difference in adiponectin with lower levels appearing in boys during the progression of puberty (18, 19).

A developmental regulation of adiponectin has also been suggested based on the higher adiponectin concentrations in early developmental stages. The concentration of adiponectin is 4-fold higher in umbilical cord plasma at birth and 2-fold higher in 2-month old neonates compared with adults (20–22). The adiponectin producing tissues have not been well characterized in newborns however, the expression of adiponectin in fetal white adipose tissue and vascular cells likely contributes to the high circulating concentrations at birth (20, 23). To gain further insight into the mechanisms of adiponectin action, we have characterized the relationships of adiponectin with regard to insulin sensitivity and adiposity in human fetuses and their association with gender.

METHODS AND PROCEDURES

Subjects

Volunteers gave signed informed consent in accordance with the Case Western Reserve University and MetroHealth Medical Center guidelines for the protection of human subjects. A total of 121 non-laboring women (100 with uncomplicated pregnancy and 21 with gestational diabetes) without evidence of clinical infection were recruited at the time of elective Cesarean section delivery after overnight fast. Gestational diabetes was diagnosed at 20–24 weeks gestation based on a positive 100 g oral glucose tolerance test according to the criteria defined by Carpenter and Coustan (24). Umbilical venous blood was obtained upon delivery of the placenta by puncture of the double clamped cord. Plasma was separated by centrifugation and kept frozen at −20°C for adiponectin, glucose, insulin and leptin assays.

Metabolic and anthropometric measurements

Gestational age was determined based on record of the last menstrual period and verified by ultrasound prior to week 20 of gestation. Placenta weight was recorded on a calibration scale after trimming of umbilical cord and fetal membranes. Neonatal length and weight were determined using measuring board and calibrated scale to the nearest of 0.1 cm and 10 g respectively. Neonatal body composition measurements were evaluated using skinfolds within 24 hours from delivery by one examiner experienced in technique (25). Two subjects were not included in final analysis based on their % neonatal body fat <2. The insulin resistance indexes were calculated according to the homeostasis model assessment (HOMA): HOMA = fasting plasma insulin (μU/ml) x fasting blood glucose (mmol/liter)/22.5.

Adiponectin electrophoretic measurements

Oligomeric adiponectin complexes in umbilical plasma were analyzed by western blot in five females and four males from the cohort of subjects matched for maternal age, gestational age and ethnicity. Four μl of a 1/10 plasma dilution (approx 12 μg protein) from male and female fetuses were electrophoresed under non-reducing and non-denaturing conditions and transferred to nitrocellulose membrane. Membranes were blocked overnight then incubated for 2 hours using a rabbit polyclonal anti-adiponectin antibody (Chemicon), and goat anti-rabbit IgG HRP-conjugated as secondary antibody. Immuno-complexes were visualized by chemiluminescence (Amersham, Piscataway, NJ). Abundance of high molecular weight adiponectin multimers were determined by densitometry with a gel-doc system (Biorad, Hercules, CA) and expressed as percent total adiponectin multimers. For determination of total adiponectin, all plasma samples were run in duplicate in a single assay using commercial ELISA kits with intra-assays coefficients of variation of 7.4 %.

Other plasma assays

Insulin was measured using radioimmunoassay (Linco, St Charles, MO). Glucose was assessed by the glucose oxidase method using a glucose analyser (Yellow Springs Instrument, Yellow Springs, OH). Leptin was measured using ELISA kits (Linco) with intra-assays coefficients of variation (CV) of 2.6–6.2%.

Data analysis

Values are presented as means ± SD. Differences between dependent variables were examined with one-way repeated measures analysis of variance (ANOVA). Statistical significant mean differences were identified with a Fisher’s PLSD post-hoc test. The relationships between fetal adiponectin metabolic and body composition parameters were estimated on bivariate correlation analysis. The correlations between adiponectin and variables of interest were evaluated by analysis of covariance and simultaneously adjusted for potential confounders (gestational age, maternal age, maternal pre-gravid BMI and ethnicity) using partial correlation. The data were analyzed using the Statview III statistical package (Abacus Concepts, Berkeley, CA) and statistics (Tallahassee, FL). P values were computed using the Fisher’s Z statistics. The level for statistical significance was set at 0.05.

RESULTS

Adiponectin association with gender

Demographic and metabolic variables of the neonates are presented in Table 1. The inclusion of pregnant women with and without gestational diabetes allowed us to analyze neonates with a wide range in neonatal adiposity at birth (4.4 to 22.8 % total body fat). The wide range of adipose tissue mass was reflected by the wide distribution of umbilical leptin concentrations from 3.4 to 96.6 ng/ml. There was a gender difference in birth length, birth weight and placental weight which were all higher in males. In contrast, the amount of adipose tissue mass was similar in male and female neonates at delivery (Table 1) as reported previously by our group. Total adiponectin measured in umbilical cord plasma at birth was higher in female as compared with male fetuses (34.3 ± 9.5 vs 29.4± 9.0 μg/ml, p< 0.004). The gender difference in total plasma adiponectin was associated with a 3.2-fold greater abundance (0.20±0.03 vs 0.08±0.03) of circulating HMW multimeric complexes (Figure 1) with no significant difference in LMW and trimers (p=0.1). There was no difference between plasma adiponectin in offspring of women with GDM and normal glucose tolerance either as a group or split by gender (29.4±9.0 vs 28.5±3.9 in males and 34.3±9.5 vs 35±11.9 μg/ml in females).

Table 1. Neonatal anthropometric and metabolic parameters at delivery.

n=60 females (including 11 from women with GDM) and 61 males (including 10 from women with GDM). Means ± SD.

	males	females	p-value
Maternal age (yr)	29.1 ± 5.6	26.5 ± 5.7	0.01
Gestational Age (wk)	38.7 ± 0.7	38.8 ± 0.6	0.26
Placental Weight (g)	691 ± 185	635 ± 134	0.01
Weight (g)	3406 ± 532	3199 ± 445	0.002
Length (cm)	49.6 ± 2.1	48.6 ± 2.0	0.0004
PI (g/cm3)	2.78 ± 0.27	2.79 ± 0.28	0.88
Fat Mass (g)	446 ± 194	407 ± 162	0.11
Lean Mass (g)	2951 ± 355	2783 ± 308	0.0002
Body Fat (%)	12.6 ± 3.8	12.4 ± 3.3	0.65
Leptin (ng/ml)	11.0 ± 9.8	17.3 ± 18.4	0.005
Insulin (μU/ml)	8.6 ± 5.5	10.3 ± 10.6	0.19
HOMA-IR index	1.3 ± 0.8	1.6 ± 1.5	0.09

GDM, gestational diabetes mellitus; HOMA-IR, homeostasis model assessment of insulin resistance; PI, pondral index

Figure 1. distribution of adiponectin oligomer complexes in umbilical circulation.

A: Distribution of HMW and LMW adiponectin complexes quantified by densitometry. B: western blot under non-reducing and non-denaturing SDS-PAGE conditions showing high molecular weight oligomers (HMW) migrating at approximately 340 kDa, low molecular weight hexamers (LMW) migrating at 160 kDa and trimers migrating at 67 kDa. Lanes 1–3: umbilical plasma (1, 2, and 4 ul per well) from one representative female fetus, lanes 4: molecular mass markers, lanes 4–6: umbilical plasma (1, 2 and 4 μl per well) from one representative male fetus. Data are means ± SE with n = 5 females and 4 males from the cohort of subjects described in Table 1 matched for maternal age, gestational age and ethnicity. C: Total adiponectin in umbilical plasma at delivery with n= 61 males and 60 females. *p<0.001 between males and females.

Adiponectin association with body composition

There was a positive correlation between umbilical cord adiponectin and fetal adiposity at birth estimated by percent body fat (r=0.19, p < 0.03) but not between adiponectin and lean body mass (r=0.10, p=0.26) (Figure 2). When analyzed by gender the association between adiponectin and percent body fat was significant only in females (r= 0.28, p<0.03) whereas the association between adiponectin and lean mass was significant only in males (r= 0.31, p<0.02) (Table 2).

Figure 2. gender specific relationship between fetal adiponectin and body composition.

Regression analyses show a significant positive correlation between umbilical adiponectin and percent total body fat in females (r=0.19, p=0.03) but not between adiponectin and lean body mass (r=0.10, p=0.25). Black circles: females, open circles, males.

Table 2. Sex discordant associations between umbilical cord plasma adiponectin concentrations and neonatal body composition at birth.

p and r values are provided without adjustment and after simultaneous adjustment for maternal age, maternal ethnicity, gestational age and pre-gravid BMI.

	Adiponectin in girls		Adiponectin in girls		Adiponectin in boys		Adiponectin in boys
	r	p	adjusted		r	p	adjusted
			r	p			r	p
HOMA-IR index	0.09	0.50	0.10	0.45	−0.02	0.90	−0.05	0.71
Insulin	0.15	0.25	0.15	0.27	−0.05	0.72	−0.04	0.75
Leptin	0.20	0.14	0.20	0.13	0.11	0.37	0.16	0.23
Neonatal lean mass	0.08	0.55	0.15	0.35	0.28	0.03	0.31	0.02
Neonatal fat mass	0.24	0.07	0.27	0.04	0.19	0.14	0.24	0.07
Percent body fat	0.28	0.03	0.28	0.03	0.13	0.31	0.21	0.21
Birth weight	0.13	0.33	0.20	0.14	0.26	0.03	0.29	0.02

HOMA-IR, homeostasis model assessment of insulin resistance.

Adiponectin association with insulin sensitivity

Neonatal insulin sensitivity at birth was estimated by calculating homeostasis model assessment of insulin resistance (HOMA-IR) indices in offspring of mothers fasted overnight prior to delivery. The mean value for HOMA–IR indices and insulin concentrations were higher in male fetuses suggesting that male fetuses are slightly less insulin sensitive than females (Table 1). However, there was no significant correlation between cord adiponectin and HOMA-IR or between adiponectin and fasting insulin concentrations (Figure 3, Table 2).

Figure 3. adiponectin is not correlated with insulin sensitivity of the fetus.

Regression analyses show an absence of significant correlation between umbilical cord plasma adiponectin concentrations, insulin sensitivity of the fetus estimated by HOMA-IR and fetal insulin concentrations. Black circles: females, open circles, males.

DISCUSSION

Gender dimorphism

Sex differences in total adiponectin with higher circulating concentrations in females as compared with males have been identified in adults, adolescents and children (18,26,27). The sexual dimorphism appears primarily related to the higher levels of the HMW complexes with no significant differences in lower molecular weight multimers (2, 28). We report that the gender difference in adiponectin already exists at birth with greater umbilical plasma concentrations as well as a greater proportion of HMW complexes in female neonates as compared with males. Our findings indicate that the abundance of adiponectin is regulated at early stages of human development, through mechanisms which reflect the gender specificity of the intrauterine environment. The HMW and LMW systemic complexes do not interchange in the circulation and thus directly reflect adiponectin secretion (29). Assuming that HMW abundance reflects the level of secretion by fetal tissues rather than a difference in clearance rates, the lower HMW in males may reflect regulated adiponectin production in vascular cells and adipocytes (20). Testosterone selectively reduces the HMW form by inhibiting its secretion from adipocytes (28, 30). Hence, the higher testosterone concentration in male as compared with female fetuses may contribute to the decreased HMW abundance we have observed in umbilical adiponectin (31). Androgens may also indirectly regulate adiponectin distribution through a decrease in adipocyte number and size in male fetuses (32, 33) providing that the pathways for hormonal regulation of adipose tissue growth are functional before birth. A potential reason why previous studies may have failed to document a sex difference in fetal/neonatal adiponectin concentrations is unclear (21,34,35). We regard the technique-inherent variability with large inter and intra-assay coefficients of variation of most commercial adiponectin assay kits as a potential source of bias.

Adiponectin and adiposity

The intuitive principle that the circulating concentration of a molecule secreted by adipose tissue varies accordingly to changes in adipose mass, which applies to most adipokines, is considerably more complex for adiponectin. The well established inverse correlation between adiponectin and body mass index in adults (36) has been questioned in young children and adolescents (19,27). In neonates, there is a positive association between umbilical adiponectin and birth weight (34,35,37,38) but the relationship with fat mass is unclear (39,40). In our cohort of 121 neonates, cord adiponectin at delivery was positively correlated with percent body fat in girls but not in boys reflecting a sexual dimorphism during in utero development. The weaker correlation between adiponectin and adiposity in fetal compared with adults (6,8) may reflect the multiple sites for adiponectin expression in the fetus (20). Our data provide experimental support to the hypothesis by Ong et al that the association between adiponectin and body size progressively switches from positive at birth to negative in adults (27). It further highlights that the association is based on adiposity rather than simply reflecting body weight. In adults, adiponectin secretion displays a depot difference with higher secretion by central/visceral as compared with subcutaneous fat (16,41). We thus speculate that the gender specificity that we have observed between adiponectin and adiposity reflects early specificities in adipose tissue development with greater visceral than subcutaneous fat mass in female fetuses. Additionally, our data suggest that the postnatal increase in adipose tissue mass contributes to production of inhibitors of adiponectin which either are not produced or are not functional in utero. The identification of such factors may provide valuable tools for further understanding adiponectin regulation.

Adiponectin and insulin sensitivity

Fetal insulin sensitivity was estimated by calculating HOMA-IR indices in our study cohort at birth. Insulin sensitivity was slightly lower in males compared with female fetuses, As previously reported the HOMA-IR indices were several orders of magnitude lower than in adults, suggesting a state of higher insulin sensitivity in utero (20). The absence of significant correlation between cord adiponectin concentration and HOMA-IR indices or umbilical insulin concentrations is in sharp contrast to the inverse relationship reported in adults (6,8,42). In agreement with estimates of insulin sensitivity in newborns, our data support the concept that early developmental stages are characterized by higher insulin sensitivity particularly of peripheral tissues (43). Taken together, these findings suggest that the combination of high insulin sensitivity and high adiponectin levels may have functional consequences in creating a favorable environment for rapid tissue growth during fetal life.

In conclusion, we report that the regulation of adiponectin in prenatal life displays gender characteristics. The sexual dimorphism between females and males suggests that adiponectin secretion by fetal tissues may be sensitive to the action of androgens as part of early metabolic programming. Our findings set the basis for further understanding the function of adiponectin during human development.