Circulating adiponectin levels during human endotoxaemia

Abstract

Adiponectin, an adipocytokine secreted by fat tissue, may prevent development of diabetes type II, as high adiponectin levels are linked with insulin sensitivity. In contrast, tumour necrosis factor (TNF)-α, which is also produced by fat tissue, leads to insulin resistance and furthermore inhibits adiponectin mRNA production and secretion of the protein. However, adiponectin also negatively regulates TNF-α levels. Therefore, we set out to test whether an infusion of endotoxin would influence circulating adiponectin levels in healthy human subjects. Twenty-three healthy human subjects were injected with endotoxin (2 ng/kg body weight); eight of these subjects were also injected with saline and served as controls. Plasma levels of adiponectin, TNF-α and interleukin-6 were measured at 0, 1·5, 2, 4, 8 and 24 h. TNF-α and interleukin-6 levels peaked at 1·5 h and 2 h, respectively. Control subjects injected with saline showed a decrease in adiponectin plasma levels with time (P < 0·05) presumably owing to the effect of fasting or physical inactivity. However, there was no change in adiponectin plasma levels in endotoxin injected subjects, thus the effect of fasting was opposed. In conclusion, circulating adiponectin levels are reduced during a resting and fasting state, an effect reversed by endotoxin injection.

INTRODUCTION

Acute bacterial infections and human endotoxemia induce development of insulin resistance [1]. The mechanisms are not fully understood, but may involve increased levels of tumour necrosis factor (TNF)-α, which has been linked to insulin resistance [2]. TNF-α is a signalling molecule secreted by adipose tissue and interferes with insulin action in fat and muscle in rodents and humans [2,3], thus leading to insulin resistance.

Recently a newly identified adipocytokine, adiponectin, has also been suggested to be involved in glucose metabolism and insulin sensitization. Adiponectin is a peptide hormone produced by fat tissue. Adiponectin cDNA encodes a polypeptide of 247 amino acids and generates a protein of approximately 30 kDa in size [4]. Adiponectin is likely to be involved in fat and glucose metabolism in mice, as the globular domain of murine adiponectin reduces plasma levels of free fatty acids (FFA), glucose and triglycerides and accelerates the removal of FFA from plasma after intravenous lipid injection [5]. Moreover, adiponectin knock-out mice display delayed clearance of FFA in plasma and low levels of fatty-acid transport protein 1 mRNA in muscle [6]. Adiponectin treatment of obese, diabetic mice (KKA^y) increased FFA oxidation in skeletal muscle, and insulin resistance could be reversed partially when mice were treated with adiponectin [7]. Finally, adiponectin may be dysregulated in obese humans, because adiponectin mRNA is reduced by 50–80% in fat tissue samples from obese humans [4].

Although the primary structure of adiponectin and TNF-α are very different, certain domains display strikingly similar 3-D structures [8] and evidence exists that the levels of TNF-α are regulated by adiponectin and vice versa. Thus, adiponectin knock-out mice display high levels of TNF-α mRNA in adipose tissue and high concentrations in plasma, and supplementation of plasma adiponectin decreases TNF-α in the knock-out mice [6], suggesting that adiponectin inhibits TNF-α production. Moreover, adiponectin suppresses lipopolysaccharide-induced TNF-α mRNA expression in macrophages [9]. Thus, evidence exists that adiponectin exerts inhibitory effects on the production of TNF-α. However, incubation of differentiating human pre-adipocytes with TNF-α for 48 h resulted in a dose-dependent decrease of adiponectin expression in cells as well as a decrease in the adiponectin secretion to the media [10], suggesting that TNF-α inhibits the production of adiponectin. Taken together, these data suggest that adiponectin and TNF-α are regulated tightly and may inhibit each other's production mutually in adipose tissue via classical feedback mechanisms.

Experimental endotoxaemia in humans is a model of acute inflammation and has been well characterized. Endotoxin injection transiently increases plasma levels of a range of pro-inflammatory cytokines, including TNF-α [11]. Given (1) that endotoxaemia induces insulin resistance, (2) that high levels of TNF-α and low levels of adiponectin are linked with insulin resistance and (3) that the regulation of TNF-α and adiponectin are linked, we hypothesized that endotoxin injection in healthy humans would influence adiponectin plasma levels.

Therefore, levels of adiponectin, TNF-α and interleukin (IL)-6 were measured in healthy humans after an intravenous bolus of endotoxin and in subjects injected with saline.

MATERIALS AND METHODS

Endotoxin study

We included 23 healthy human subjects (median age 24, range 20–28 years, four females). All subjects underwent a physical examination prior to inclusion in the study and had no previous medical history. The subjects did not experience any signs or symptoms of infection in the fortnight preceding the endotoxin infusion. The regional scientific ethical committee approved the study, and written informed consent was obtained from each volunteer. Eight of the 23 subjects were asked to volunteer in two experiments, in which they received endotoxin injection in one experiment and saline infusion in the other. After an overnight fast, in total 23 subjects were given an intravenous bolus of Esherichia coli endotoxin (endotoxin, E. coli; lot EC-6, United States Pharmacopeia Convention, Rockville, MD, USA) at a dose of 2 ng/kg body weight. Endotoxin are generally administered at doses ranging from 2 to 4 ng/kg body weight; however, TNF-α levels increase to the same extent [11,12], thus the lower dose was chosen for this study. Subjects were kept in an intensive care unit, in supine rest, with continuous monitoring of their cardiovascular functions for 24 h. During the first 8 h of the experimental protocol, subjects were infused with an isotonic saline solution. Subjects were infused with 15 ml/kg/h during the first hour and then 7 ml/kg/h for the remaining 7 h. Subjects were allowed to eat after the 8-h sample was obtained. Blood samples for measurement of TNF-α, IL-6 and adiponectin were drawn at baseline and 1·5, 2, 4, 8 and 24 h after the endotoxin injection and at 2, 4, 8 and 24 h after a saline injection. Blood samples were drawn into tubes containing EDTA and trasylol and spun immediately at 3500 g for 15 min at 4°C in a centrifuge, following which plasma was stored at − 70°C until analysis. Plasma levels of endotoxin were not measured, as previous studies have shown that an endotoxin injection of 2 ng/kg gives a substantial rise in plasma, which is cleared from the bloodstream within 15 min [12].

Cytokine measurements

All cytokine measurements were run in duplicate and mean values calculated. Plasma levels of IL-6 and TNF-α were measured in 23 subjects receiving a bolus of endotoxin and in eight control subjects injected with saline using quantikine ELISA-kits from R&D systems (cat. nos HS600 and HSTA00C, respectively). The minimum detectable dose of IL-6 was 0·094 pg/ml, whereas for TNF-α it was 0·06 pg/ml. The samples for TNF-α were diluted 1 : 80 at the time-points 1·5, and 2 h, 1 : 40 at 4 h and 1 : 1 at the 8-h time-point. Samples for detection of IL-6 were diluted 1 : 1000 at 1·5, 2 and 4 h and 1 : 10 at 8 h. Adiponectin plasma levels were measured in 23 subjects receiving a bolus of endotoxin and in the eight control subjects injected with saline using a human adiponectin radioimmunoassay (RIA) kit from Linco (cat. no. HADP-61HK) with an assay range of 2·0–500 ng/ml. Samples to be detected for adiponectin protein levels by RIA were diluted 1 : 500 in assay buffer prior to measurements.

Statistical analyses

All data were ln-transformed in order to achieve normal distribution and the results are presented as reverse-transformed data expressed as geometric mean ± a 95% confidence interval (CI). A one-way analysis of variance (anova) for repeated measures was used to analyse for statistically significant changes in the plasma level of IL-6, TNF-α and adiponectin compared to resting levels. A two-way anova was used to analyse for the effect of time and treatment on adiponectin levels between controls and endotoxin injected subjects. Statistically significant changes were analysed using the Bonferroni t-test with an all-pairwise multiple comparison procedure. P-values < 0·05 were considered significant. Statistical calculations were performed using Sigma Stat 3·0 (SSPS Inc., Chicago, USA).

RESULTS

Rectal temperature increased approximately 2°C reaching maximal levels at 4 h after endotoxin injection (data not shown). Following endotoxin injection, TNF-α levels increased from 3·3 pg/ml to a peak of 840 pg/ml (95% CI; 545–972 pg/ml) at the 1·5-h time-point, whereas IL-6 levels increased from 1 pg/ml to a peak of 1690 pg/ml (95% CI; 1098–1924 pg/ml) 2 h following infusion ( Fig. 1a), which is in accordance with previous findings [11–14].

Fig. 1.

(a) Effect of endotoxin injection (n = 23) on TNF-α (black) and IL-6 (white). The data are presented as geometric mean ± 95% CI. The TNF-α level increased from 3·3 pg/ml to 840 pg/ml 1·5 h after an endotoxin bolus. As TNF-α induces IL-6, the IL-6 level increased from 1 pg/ml to 1690 pg/ml 2 h following injection. The peak of IL-6 protein was followed by a marked decrease in the TNF-α protein level. The TNF-α and IL-6 levels differed significantly at all time-points when compared to rest (one-way anova, P < 0·05). (b) Effect of a saline injection (n = 8) on TNF-α (black) and IL-6 (white). The data are presented as geometric mean ± 95% CI. A decrease (P < 0·05) in the TNF-α plasma level was detected at 4, 8 and 24 h when compared to 0 h. IL-6 plasma levels showed a modest increase (P < 0·05) at all time-points when compared to 0 h, showing an increase of 6 pg/ml at 4 h.

Control subjects displayed a decrease (P < 0·05) in TNF-α plasma levels during the time-course measured which was reduced significantly at 4, 8 and 24 h when compared to 0 h. IL-6 plasma levels showed a modest increase during the time-course measured (P < 0·05); however, this increase is not comparable to that seen in the endotoxin trial (Fig. 1b).

A two-way anova on adiponectin plasma levels showed a significant effect of time × treatment (P = 0·005), although no specific time-point of significance was detected by the post-hoc analysis. There was a significant decrease in adiponectin plasma levels over time after a saline injection (one-way anova, P = 0·007) with the post-hoc analysis showing an effect at time-points 8 and 24 h displaying, respectively, 48% (P = 0·008) and 45% (P = 0·031) decrease when compared to rest. However, there was no significant change in the level of adiponectin after endotoxin injection ( Fig. 2).

Fig. 2.

Effect of endotoxin (n = 23, black) or saline (n = 8, grey) injection on adiponectin plasma levels (*) denotes statistical significance of P < 0·05. The data are presented as geometric mean ± 95% CI. Adiponectin levels significantly decreased with time (one-way anova, P = 0·007) in the control group, whereas there was no change after endotoxin injection when compared to rest. A two-way anova showed an effect of treatment × time (two-way anova, P = 0·005), but the post-hoc test was unable to detect a specific time-point of significance.

DISCUSSION

The present study demonstrated that endotoxaemia did not change plasma levels of adiponectin, although endotoxaemia induced a marked increase in circulating levels of TNF-α and IL-6, as described previously [11,12]. As adiponectin levels decreased in the control experiment it seems, however, reasonable to conclude that the sole effect of endotoxaemia was a small relative increase in adiponectin levels.

The decrease in adiponectin levels in the control experiment may be ascribed to an effect of fasting or physical inactivity. As adiponectin has been suggested to have a role in the clearance of FFA [5], it seems logical that the production of adiponectin is down-regulated when the uptake of FFA is reduced, such as during a fasting state [15]. As the regulatory relationship between adiponectin and TNF-α is not clear, it was possible that endotoxin-induced elevated levels of TNF-α might either down-regulate adiponectin [10] or, in contrast, given that adiponectin down-regulates TNF-α production [6], adiponectin levels might increase as a negative feedback mechanism. In addition to the effects in adipose tissue, adiponectin and TNF-α have been shown to inhibit each other's function in other tissues. Thus, adiponectin inhibits TNF-α-mediated activation of NF-κB signalling in vascular endothelial cells [16,17]. Furthermore, adiponectin and TNF-α have opposite effect on the insulin stimulation of IRS-1-associated PI3-kinase, glucose uptake and FATP-1 mRNA expression in myocytes [6], suggesting an association between TNF-α and adiponectin.

The finding that adiponectin levels did not decrease in response to endotoxin administration to healthy humans suggest that adiponectin levels increase during endotoxaemia to counteract the metabolic effects of high levels of TNF-α.

Alternatively, catecholamines or other hormones may be involved in the regulation of adiponectin. During endotoxaemia, the concentration of adrenaline in blood is increased. In accordance, in a former study, we found that adrenaline increased following endotoxin infusion, but not after saline [18]. Given that adrenaline inhibits the endotoxin-induced increase in TNF-α levels [19], it is possible that it is involved in the interplay between TNF-α and adiponectin.

The decrease in adiponectin level during the control experiment (∼ 45%) may have important implications, as type II diabetic subjects display reduced plasma adiponectin levels, ranging from 50–84% of that of non-diabetic subjects [20,21].

In conclusion, we demonstrate for the first time that rest at a fasting state is followed by a decrease in adiponectin plasma levels, which may have important implications. This decrease is not observed in subjects receiving endotoxin. Thus, it is possible that an elevated level of TNF-α is a signal to the production of adiponectin.