Insulin and Endothelin in the Acute Regulation of Adiponectin in Vivo in Humans


  • The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Indiana University School of Medicine, 975 West Walnut Street, IB424, Indianapolis, IN 46202. E-mail:


Objective: In vitro, insulin and endothelin (ET) both modulate adiponectin secretion from adipocyte cell lines. The current studies were performed to assess whether endogenous ET contributes to the acute action of insulin infusions on adiponectin levels in vivo in humans.

Research Methods and Procedures: We studied 17 lean and 20 obese subjects (BMI 21.8 ± 2.2 and 34.0 ± 5.0 kg/m2, respectively). Hyperinsulinemic euglycemic clamp studies were performed using insulin infusion rates of 10, 30, or 300 mU/m2 per minute alone or with concurrent infusion of BQ123, an antagonist of type A ET receptors. Circulating adiponectin levels were assessed at baseline and after achievement of steady-state glucose with the insulin infusion.

Results: Adiponectin levels were lower in obese than lean subjects (6.76 ± 3.66 vs. 8.37 ± 2.79 μg/mL, p = 0.0148 adjusted for differences across gender). Insulin infusions suppressed adiponectin by a mean of 7.8% (p < 0.0001). In a subset of 13 lean and 14 obese subjects for whom data with and without BQ123 were available, there was no evident effect of BQ123 to modulate clamp-associated suppression of adiponectin (p = 0.16). Surprisingly, there was no evident relationship between steady-state insulin concentrations and adiponectin suppression (r = 0.14, p = 0.30), and again no effect of BQ123 to modify this relationship was seen.

Discussion: Despite baseline differences in adiponectin levels, we observed equal suppression of adiponectin with insulin infusions in lean and obese subjects. ET receptor antagonism with BQ123 did not modulate this effect, suggesting that endogenous ET does not have a role in modifying the acute effects of insulin on adiponectin production and/or disposition.


Adiponectin is one of a number of adipokines that plays a role in modulation of glucose metabolism, insulin resistance, and inflammation (1). Recognized in 1995 as a novel protein produced by adipocytes (2), adiponectin has been observed to circulate in reduced concentrations in subjects with obesity (3) or type 2 diabetes (3,4,5).

The physiological regulation of adiponectin is unclear. Studies of mRNA levels in human adipose tissue have suggested reduced adiponectin gene expression in obese (6,7) and obese diabetic (8) subjects. However, in a similar study of obese women, adiponectin mRNA was not reduced in either subcutaneous or omental adipose tissue samples (9). In vitro studies have suggested roles for insulin (10,11,12), tumor necrosis factor α, interleukin-6, and glucocorticoids (11,13) in the regulation of adiponectin production. The role of insulin is unclear because both stimulation (10) and suppression (11) of adiponectin have been reported. An acute effect of endothelin (ET)-1 to stimulate adiponectin production has also been reported, mediated by the Type A ET (ETA)1 receptor (12).

Studies of in vivo regulation of adiponectin levels are few. A role for insulin in this capacity has been suggested and is supported by two reports using hyperinsulinemic euglycemic clamp protocols. These short-term elevations of circulating insulin levels acutely decreased serum adiponectin levels (5,14). In contrast, increases in serum adiponectin have been reported after exposure to peroxisome proliferator-activated receptor-γ agonists (5,15,16,17). The precise mechanism(s) of action of peroxisome proliferator-activated receptor-γ agonists and of insulin infusions to reduce adiponectin levels are uncertain. These effects could result from a change in adiponectin production, a change in adiponectin clearance, or both. Furthermore, it is unclear which mechanism(s) result in the observed reductions in circulating adiponectin levels in obese humans.

ET-1 is a peptide produced by the vascular endothelium (18) that has been shown to circulate in higher levels and to have increased activity in obese subjects with the metabolic syndrome and subjects with type 2 diabetes (19,20,21,22,23). Exogenous treatment with ET-1 can induce insulin resistance in healthy rats (24) and healthy humans (25) as measured by hyperinsulinemic euglycemic clamps. The mechanism of these effects on insulin sensitivity remains unclear. Studies suggest that ET-1 can modulate insulin signaling in smooth muscle cells (26) and in 3T3-L1 adipocytes (18,27,28) and that chronic exposure to ET-1 can desensitize cellular responses to insulin, including the effect of insulin on adiponectin production (12,18).

Individuals with insulin-resistant states have lower levels of adiponectin and higher endogenous ET-1 activity than insulin-sensitive individuals. Given that insulin modulates adiponectin levels in vitro and in vivo and that ET-1 modifies insulin signaling and adiponectin levels in vitro and induces insulin resistance in vivo, it is possible that ET-1 contributes to obesity-associated reductions in adiponectin, perhaps through effects on insulin action. We have, therefore, tested the hypothesis that endogenous ET-1 modulates the action of insulin to suppress adiponectin levels in human subjects.

Research Methods and Procedures

Thirty-seven individuals (17 lean and 20 obese) 18 to 46 years old volunteered to participate in the study. Subjects were categorized as lean or obese by gender-specific BMI cut-off points (women, BMI ≥ 28; men, BMI ≥ 26). Subjects were recruited through advertisement. All methods and procedures for the study were approved by the Institutional Review Board of the Indiana University Medical School and complied with the Declaration of Helsinki. Each participant provided written informed consent.

Subjects were screened with a medical history, physical exam, 2-hour oral glucose tolerance test, and fasting lipid measurements. Exclusion criteria were type 2 diabetes (ADA 1997 criteria), treated hypertension, blood pressure > 160/100 mm Hg, or total cholesterol > 240 mg/dL (6.2 mM).

Subjects were admitted to the Indiana University General Clinical Research Center the night before the study and fasted overnight. They underwent two hyperinsulinemic-euglycemic clamp studies at least 4 weeks apart, with adiponectin levels measured at baseline and at steady state (210 to 260 minutes of insulin infusion). Subjects received either a low dose of insulin (Humulin R; 10 mU/m2 per minute for lean individuals and 30 mU/m2 per minute for obese individuals; Eli Lilly, Indianapolis, Indiana) or a high dose of insulin (300 mU/m2 per minute for both lean and obese individuals) during the clamp, according to the method of DeFronzo (29). A given subject received the same insulin infusion rate for both study days. In random sequence, the two clamp studies were performed with and without a concomitant infusion of BQ123 (Clinalfa, Basel, Switzerland), a direct antagonist of the ET-1A (ETA) receptor (1000 nmol/min).

Our data set included 27 subjects with insulin studies both with and without BQ123 (i.e., paired studies), and 10 subjects who completed only one study (i.e., unpaired studies). Three individuals underwent a total of three studies: a complete pair under one insulin infusion rate and another study with or without BQ123 under a different insulin infusion rate.

Only the 27 subjects with insulin-clamp studies both with and without BQ123 (paired studies) were included in the main analyses of the effect of insulin and BQ123 on adiponectin levels. Analyses not dependent on BQ123 exposure status included all available insulin clamp studies.

Body Composition

Height (centimeters) and weight (kilograms) were measured the morning of each infusion study and used to calculate BMI as weight (kilograms)/height (meters squared). Percentage body fat was determined by DXA (GE Lunar DPX-L; General Electric, Madison WI).

Analysis of Blood Samples

Blood for serum glucose determinations was put in untreated polypropylene tubes and centrifuged using an Eppendorf microcentrifuge (Brinkman, Westbury, NY). The glucose concentration of the supernatant was then measured by the glucose oxidase method using a glucose analyzer (Model 2300; Yellow Springs Instruments, Yellow Springs, OH). Blood for determination of plasma insulin was collected in heparinized tubes, processed immediately, and frozen at −20 °C. Insulin determinations were made using a dual-site radioimmune assay, specific for human insulin and with cross-reactivity with proinsulin < 0.2%. The lower detection limit is 0.56 pM, and in our laboratory, the inter- and intraassay coefficients of variation are 4.1% and 2.6%, respectively. Standard methodologies for cholesterol and triglyceride determinations were performed through our local hospital's clinical laboratory. Blood for determination of adiponectin was collected in EDTA-treated tubes. Adiponectin was measured with a commercially available radioimmunoassay kit (Linco Research, St. Charles, MO). The limit of sensitivity of the adiponectin assay is 1 ng/mL with an intra- and interassay precision of 6.21% and 6.90%, respectively, at a sample concentration of 6 ng/mL. ET-1 levels were measured by RIA using a commercial kit (R&D Systems, Minneapolis, MN).

Statistical Analysis

Data that were not normally distributed were normalized through logarithmic transformations before analysis. Comparisons between and within groups were performed by Student's t tests, ANOVA, and repeated-measures ANOVA as appropriate. When significant differences were found by ANOVA, this was followed by post hoc pair-wise testing with the Student-Newman-Keuls test. Statistical significance was accepted at a level of p < 0.05. Results are presented as the mean ± SEM.


The demographic and metabolic characteristics of subjects are presented in Table 1. The baseline characteristics for the patients with only paired data were not significantly different from those of the entire group (p = not significant for all variables listed in Table 1).

Table 1. . Baseline characteristics
  1. Data are means ± SD. Sys, systolic; dias, diastolic.

N (women/men)3/148/120.30
Age (years)36.7 (6.5)32.9 (8.1)0.09
BMI (kg/m2)21.8 (2.2)34.0 (5.0)<0.001
Body fat (%)15.7 (5.5)36.2 (10.8)<0.001
Total cholesterol (mg/dL)155.2 (32.8)164.7 (33.5)0.16
Triglycerides (mg/dL)84.7 (27.2)114.7 (56.5)0.32
High-density lipoprotein (mg/dL)42.8 (9.8)44.8 (10.4)0.22
Low-density lipoprotein (mg/dL)96.6 (27.5)110.1 (31.1)0.19
Blood pressure (sys/dias)116/67 (17/11)122/68 (18/9)0.88/0.46
Fasting glucose (mg/dL)93.2 (7.5)97.6 (8.2)0.087
Fasting insulin (mU/L)8.4 (5.8)17.7 (7.8)0.009
ET-1 (pg/mL)2.13 (0.60)2.44 (1.08)0.48

Basal adiponectin levels were lower in men than in women (p = 0.012), and basal adiponectin levels were lower in obese than in lean subjects [6.8 (3.7) vs. 8.4 (2.8) μg/mL; mean (SE); p = 0.015 adjusting for the gender difference]. Figure 1 shows basal adiponectin in lean and obese patients divided by gender.

Figure 1.

Mean adiponectin levels in lean and obese subjects by gender.

Including data from all studies, the hyperinsulinemic euglycemic clamps were associated with a reduction in adiponectin levels. The overall mean basal level of 7.4 (0.7) fell to 6.8 (0.6) μg/mL at steady state (p < 0.0001), a reduction of 7.8%. This reduction was observed with all insulin infusion rates [10 mU/m2 per minute, 9.4 (1.3) to 8.4 (1.1), p = 0.009; 30 mU/m2 per minute, 8.2 (2.2) to 7.5 (2.1), p = 0.07; 300 mU/m2 per minute, 6.1 (0.6) to 5.8 (0.6) μg/mL, p = 0.01]. There was no difference in this effect across patient category (lean vs. obese) or gender category (p = 0.69 and 0.42, respectively). These aggregated results are presented in Figure 2.

Figure 2.

Reductions in adiponectin levels from basal to hyperinsulinemic steady state with and without concurrent infusions of BQ123. The reduction in response to insulin is highly significant (p < 0.0001). The response to insulin was not different across gender (p = 0.09). There was no effect of BQ123 to alter the suppression of adiponectin by the insulin infusion in either men or women (p = 0.45).

The coinfusion of BQ123 did not change the effect of the insulin infusion to acutely lower adiponectin levels during the hyperinsulinemic euglycemic clamp. Analyzing only paired studies, we observed a drop in adiponectin levels with the hyperinsulinemic euglycemic clamps that was unaffected by the coinfusion of BQ123 [from 7.4 (0.7) to 6.8 (0.6) μg/mL without BQ123 and from 7.9 (0.7) to 7.1 (0.6) μg/mL with BQ123; Figure 2]. Basal adiponectin levels were nonsignificantly higher in the studies in which BQ123 was infused (p = 0.15); correcting for this baseline difference did not materially alter the results. No difference in these effects was seen between lean and obese subjects (p = 0.69) or between men and women (p = 0.42).

The mean baseline ET-1 level did not differ between the two study days (2.2 ± 0.4 and 2.3 ± 0.3 pg/mL on the days with and without BQ123, respectively; p = not significant). Surprisingly, there was a fall in ET-1 levels in response to insulin, seen on both study days (to 1.7 ± 0.2 and 1.8 ± 0.3). This change did not achieve statistical significance (p = 0.2). There was no effect of BQ123 to modulate the effect of insulin on ET-1 (p = 0.9).

Clamp-associated suppression of adiponectin was seen at all infusion rates. Surprisingly, we saw no evidence of a dose-dependent effect of insulin on adiponectin levels. Despite baseline differences in adiponectin across the three infusion rate groups (p = 0.054), due to differences in the lean vs. obese composition of these groups, there was no difference across these groups in the relative reduction in adiponectin levels from baseline (11.1%, 6.1%, and 6.9%, respectively, for the 10, 30, and 300 mU/m2 per minute infusions, p = 0.38). Furthermore, linear regression analysis revealed no correlation between the insulin level achieved by the clamp and the associated reduction in adiponectin levels (r2 = 0.021, p = 0.30; Figure 3). There was no evident effect of BQ123 to modify this finding (with BQ123, r2 = 0.01, p = 0.61; without BQ123, r2 = 0.03, p = 0.42).

Figure 3.

Relationship between hyperinsulinemic steady-state insulin levels and observed reductions in adiponectin levels. r = 0.14, p = 0.30. Insulin levels were logarithmically transformed to normalize their distribution.


We found that circulating adiponectin levels are lower in obese than in lean individuals and that hyperinsulinemic euglycemic clamps acutely lower adiponectin levels. We have further shown that acute ETA receptor blockade does not change the adiponectin response to insulin infusion in lean or obese subjects, suggesting that ET-1 does not play a direct role in insulin infusion-mediated suppression of adiponectin levels.

It is reasonable to postulate a contribution of ET-1 to the regulation of insulin sensitivity and adiponectin metabolism in vivo. ET-1 is known to have increased activity through the ETA receptor on vascular endothelium in obese and diabetic individuals (22,23). Exogenous ET-1 induces insulin resistance in vivo in rats and humans (24,25). ET-1 is known to modulate insulin signaling in vascular smooth muscle cells (26). The effect observed differs by duration of exposure to ET-1. Acute exposure to ET-1 in vitro stimulates glucose transport at 2 hours through the insulin-sensitive glucose transporter GLUT4 in 3T3-L1 adipocytes (27,28). ET-1-mediated GLUT4 translocation occurs through the G protein Gαq/11 and phosphatidylinositol 3-kinase (27). At 6 hours, ET-1 stimulates glucose transport through coupling to Gαi, leading to activation of the Ras/mitogen-activated protein kinase pathway and enhanced expression of the constitutive glucose transporter GLUT 1 (28). However, chronic (24-hour) treatment with ET-1 in vitro in 3T3-L1 adipocytes leads to a subsequent decrease in insulin-stimulated glucose transport and ET-1-stimulated glucose transport, consistent with heterologous desensitization of insulin action and homologous desensitization of ET-1 action by prolonged ET-1 treatment (18). Similarly, in 3T3-L1 adipocytes, exposure to insulin and to ET-1 for 1 hour increased adiponectin concentrations in the 3T3-L1 media, but the effect of ET-1 and insulin to increase adiponectin concentrations acutely was significantly inhibited by 24-hour exposure to ET-1 (12).

The in vitro evidence of ET-1's involvement in insulin signaling and ET-1-associated reductions in adiponectin levels was the basis for our hypothesis that ET-1 would have an effect in vivo to modulate insulin's action to suppress adiponectin levels. However, we found that acute ETA antagonism with BQ123 did not modulate the effect of an insulin infusion to lower adiponectin levels.

Why Might BQ123 Have Failed to Modulate Adiponectin?

There are a number of possible explanations for the failure of BQ123 to acutely modulate the adiponectin response to hyperinsulinemia. The concurrent heterologous desensitization of insulin action and homologous desensitization of ET-1 action by prolonged exposure to ET-1 (12,18) could result in resistance to the effects of both insulin and ET antagonism. This phenomenon could explain our findings, assuming that 3T3-L1 adipocytes and human adipocytes respond similarly to ET-1 and insulin. This mechanism may be active in obese subjects, who are known to have increased circulating levels of ET-1 and increased ET-1 activity. Arguing against this explanation, BQ123 had no effect to change insulin's suppression of adiponectin in our lean subjects, in whom no chronic elevation of ET-1 activity is expected. Given the heterologous desensitization of insulin action induced by prolonged exposure to ET-1 in vitro, it is also possible that a more prolonged in vivo exposure to BQ123 than we undertook with the current study design would be needed to demonstrate an effect of ETA antagonism to modulate suppression of adiponectin levels during an insulin infusion.

It is unlikely that technical issues around dosing or administration of BQ123 account for the lack of effect of BQ123. BQ123 was administered in the same fashion as in a separate study with which we have previously demonstrated an effect of BQ123 to correct the baseline defect in endothelium-dependent vasodilation seen in obese and type 2 diabetic subjects (22). The ability of BQ123 to antagonize the effects of exogenous ET-1 has been repeatedly demonstrated in vitro (30,31) and in vivo in multiple laboratories (32,33,34,35,36); therefore, we have no reason to believe that the mechanism of action of this agent is in question. There was no change in ET-1 levels on exposure to BQ123, independent of insulin. This is consistent with the literature, which describes changes in ET-1 levels with ETB antagonism but not with ETA antagonism (34). The current observation that BQ123 failed to alter adiponectin levels in the setting of hyperinsulinemia argues against a separate, direct effect of BQ123 on adiponectin levels, although this question was not directly tested.

One important factor our studies did not address is the question of timing. Although we did not observe effects of ETA antagonism over the course of 4 hours of exposure to insulin, it remains possible that relevant in vivo interactions of ET and adiponectin occur on much longer time scales.

We hypothesized that endogenous ET exerted an effect on adiponectin biology. We did not specify the origin of the ET. Broadly, it could arise within adipose tissue by production from the local capillary network, or it could be carried to adipose tissue through the circulation from distant sites of production. In the absence of an observed effect, no comment can be made regarding paracrine or local tissue effects. If we postulate that actions of ET-1 on adipocytes are exerted by circulating ET-1, it is possible that we failed to detect an effect of ETA antagonism simply because the circulating levels of ET were not sufficiently elevated in our subjects to generate a detectable response to antagonism. Further studies in subjects with more profound derangements in ET biology and elevations in circulating ET levels (such as type 2 diabetes or congestive heart failure) will be required to address this question.

Insulin and Adiponectin

In vitro and in vivo evidence is accumulating in support of a role for insulin in the modulation of adiponectin levels. Insulin administration in rats and humans during hyperinsulinemic euglycemic clamps has been shown to decrease adiponectin levels acutely (5,14), although other authors have shown increased adiponectin secretion from omental fat cells exposed in vitro to insulin (10). Our data are in agreement with the findings of Yu et al. (5) and Mohlig et al. (14) that in vivo insulin administration in humans results in lower circulating adiponectin levels. How this decrease in circulating adiponectin occurs is not clear. Insulin could have a direct negative feedback mechanism in the release of adiponectin from the adipocyte. Alternatively, insulin could act through unknown pathways to accelerate the degradation of circulating adiponectin. Insulin could be acting on another tissue or pathway entirely, with the effect of insulin to lower adiponectin being an indirect result of changes in another metabolic, vascular, or neurological pathway. Given that BQ123, a highly specific ETA receptor antagonist, had no effect on a hyperinsulinemic euglycemic clamp's action to reduce circulating adiponectin levels, our data suggest that ET-1 is not a direct factor in insulin's regulation of adiponectin levels. Clearly, more research needs to be done to elucidate the pathways by which insulin affects circulating adiponectin levels.

There is a known inverse relationship between adiponectin and endogenous insulin levels (4,37,38). However, our study failed to show a dose-response effect of an exogenous insulin infusion on adiponectin levels. Longer durations of insulin exposure may be required to produce a dose-related effect, especially if the dose-related effect is mediated through modulation of gene expression in the adipocyte. The lack of a dose-response effect with exogenous insulin administration may also suggest that the acute effect of an insulin infusion to suppress adiponectin levels is not through a direct action on the adipocyte but through an as-yet unknown intermediary pathway.

In summary, despite baseline differences in adiponectin levels, we observed equal suppression of adiponectin levels with insulin infusions in lean and obese subjects. Antagonism of ETA receptors did not modify the effect of the insulin infusion on adiponectin levels, suggesting that ET-1 does not have a short-term physiological role in the regulation of adiponectin secretion and/or metabolism. There was no correlation between the level of insulin achieved at steady state of the hyperinsulinemic euglycemic clamp and the degree to which adiponectin was suppressed. Further studies are needed to better define the mechanisms by which insulin influences adiponectin levels.


Support for this project was provided by the National Institutes of Health Grant DK42469 (A.D.B.), the American Diabetes Association Junior Faculty Award (K.J.M.), and the American Diabetes Association Research Award (R.V.C.).


  • 1

    Nonstandard abbreviations: ET, endothelin; ETA, Type A ET.