Ghrelin Infusion in Humans Induces Acute Insulin Resistance and Lipolysis Independent of Growth Hormone Signaling

The studies were conducted in accordance with the Helsinki Declaration and following the approval by the local ethics committee, the Danish Medicines Agency, and the Good Clinical Practice (GCP) unit of Aarhus University Hospital. Both protocols were registered (Clinicaltrials.gov identification study 1: NCT00116025 and study 2: NCT00139945).

Synthetic human acylated ghrelin (NeoMPS, Strasbourg, France) was dissolved in isotonic saline and sterilized by double passage through a 0.8/0.2-μm pore-size filter (Super Acrodisc; Gelman Sciences, Ann Arbor, MI).

Six healthy men (aged 23 ± 1 years, BMI 23.5 ± 0.4 kg/m²) were examined as previously described (6). They received a constant infusion of saline or ghrelin (5 pmol · kg⁻¹ · min⁻¹) starting at 0 min. At 90 min, a muscle biopsy was obtained from the lateral vastus muscle with a Bergström biopsy needle (Fig. 1).

Eight hypopituitary men (aged 53 ± 4 years, BMI 31.6 ± 1.0 kg/m²) on stable replacement therapy with GH and hydrocortisone (for >3 months) participated. None of the patients had diabetes (A1C 5.7 ± 0.1% [range 4.9–6.0]) or any concomitant chronic disease. Each patient was studied on two occasions with 5-h infusions of saline or ghrelin (5 pmol · kg⁻¹ · min⁻¹) in a randomized double-blind, cross-over design. Both study days commenced at 0800 h after an overnight fast (>9 h), with the subjects remaining fasting.

One intravenous cannula was inserted in the antecubital region for infusion, and one intravenous cannula was inserted in a heated dorsal hand vein for sampling of arterialized blood. At t = 0 min, saline or a primed-continuous ghrelin infusion (5 pmol · kg⁻¹ · min⁻¹) was commenced. The bolus dose was estimated from the elimination rate constant of ghrelin (k₀₁) (6) and infused over a 20-min interval to avoid an overshoot of steady-state levels. Muscle biopsies were obtained at 120 min, as described above. A hyperinsulinemic-euglycemic clamp (insulin 0.6 mU · kg⁻¹ · min⁻¹; Actrapid, Novo Nordisk, Denmark) was performed from 120 to 300 min. Plasma glucose was clamped at 5.0 mmol/l by adjusting the rate of infusion of 20% glucose according to plasma glucose measurements every 10 min. Insulin sensitivity was calculated from the glucose infusion rate (GIR) during the clamp. The period from 0 to 120 min is referred to as the basal period and the period from 120 to 300 min as the clamp period. Blood samples were obtained as indicated in Fig. 2.

A primed-continuous infusion of [3-³H]-glucose (bolus 12 μCi, 0.17 μCi/min; NEN Life Science Products, Boston, MA) was initiated at t = 0 min and continued throughout. Glucose rate of appearance (R_a) was calculated at 10-min intervals from 90 to 120 min and 270 to 300 min using Steele's non–steady-state equations (9). During the clamp, endogenous glucose production was calculated by subtracting the GIR from R_a. Oxidative rates of glucose (G_OX) and lipids were calculated from indirect calorimetry (Deltatrac; Datex Instruments, Helsinki, Finland) after correction for protein oxidation, which was estimated from the urinary excretion of urea. Nonoxidative glucose disposal was calculated as whole-body glucose disposal (R_d) minus the rate of G_OX (10).

Microdialysis was performed and analyzed as described previously (8). Catheters (CMA 60, molecular cutoff of 20 kDa, membrane length 30 mm; CMA, Stockholm, Sweden) were placed in the lateral vastus muscle, in the subcutaneous adipose tissue, and in the femoral subcutaneous adipose tissue. The abdominal and femoral adipose tissue blood flow was estimated by Xe washout (11).

Plasma glucose was analyzed in duplicate using the glucose oxidase method (Beckman Instruments, Palo Alto, CA). Serum ghrelin (total levels) was measured in duplicate by an in-house assay (12). Serum GH, cortisol, and insulin were analyzed with a double monoclonal immunofluorometric assay (Delfia; Perkin Elmer, Wallac Oy, Turku, Finland). Serum free fatty acids (FFAs) were determined using a commercial kit (Wako Chemicals, Neuss, Germany). Plasma catecholamines were measured by liquid chromatography (13). Plasma glucagon was measured by radioimmunoassay (14). Glucose, glycerol, lactate, and urea in the microdialysis dialysate were measured in duplicate by an automated spectrophotometric kinetic enzymatic analyzer (CMA 600; CMA).

Muscle biopsies were homogenized as previously described (15). Aliquots of protein were resolved by SDS-PAGE, and proteins were electroblotted onto nitrocellulose membranes. Immunoblotting was performed using primary antibodies as follows: p signal transducers and activators of transcription (STAT5)a and -b, STAT5, pSTAT3, STAT3, p extracellular signal–regulated kinase (ERK)1 and -2, ERK-1 and -2, p AMP-activated protein kinase (AMPK)α, AMPKα-pan, p acetyal-CoA carboxylase (ACC), pAkt, pAkt substrate, and Akt substrate 160 (AS160). Membranes were incubated with horseradish peroxidase–coupled secondary antibodies, visualized by BioWest enhanced chemiluminescence (UVP LabWorks, Upland, CA) and quantified by the UVP BioImaging System. Densitometric measurements were adjusted to an internal control. Phosphatidylinositol 3-kinase (PI3K) activity was assessed, as previously described (15).

Results are expressed as means ± SE. Systemic levels of hormones, metabolites, and GIR were analyzed by two-way ANOVA. The interaction between time and treatment (time × treatment) was considered the term of interest. The Bonferroni correction was used to account for multiple comparisons when appropriate. Pairwise comparisons were carried out by Student's two-tailed paired t test when appropriate. P values <0.05 were considered significant. Statistical analysis was performed using SPSS version 14.0 for Windows.

Ghrelin infusion stimulated endogenous GH secretion, which peaked at t = 60 min (1.1 ± 0.9 μg/l [saline] vs. 33.3 ± 8.0 μg/l [ghrelin]; P = 0.008). A significant elevation in serum FFA levels was recorded (0.4 ± 0.04 μg/l [saline] vs. 1.0 ± 0.1 μg/l [ghrelin]; P = 0.003). The levels of serum cortisol (268 ± 24 nmol/l [saline] vs. 400 ± 57 nmol/l [ghrelin]; P = 0.06) and plasma glucose (5.2 ± 0.1 mmol/l [saline] vs. 5.5 ± 0.1 mmol/l [ghrelin]; P = 0.16) were similar. Western blots performed on skeletal muscle biopsies revealed distinct STAT5 phosphorylation in all six subjects 30 min after the endogenous GH burst (Fig. 3).

Pituitary surgery had been performed in all cases, and GH deficiency was documented by GH stimulation tests (insulin tolerance test [n = 7] or arginine test [n = 1]; means ± SE peak GH 0.3 ± 0.1 μg/l).

Hormones and metabolites are shown in Fig. 2. Serum ghrelin concentrations were similar at baseline on the 2 study days (0.51 ± 0.06 μg/l [saline] vs. 0.49 ± 0.06 μg/l [ghrelin]; P = 0.38) and correlated inversely with BMI (saline r = −0.83, P = 0.01; ghrelin r = −0.73, P = 0.04). Plasma levels of norepinephrine and epinephrine were comparable on both study days, and plasma levels of glucagon were also similar (10.9 ± 0.9 pmol/l [saline] vs. 9.3 ± 0.5 pmol/l [ghrelin], P = 0.10 at t = 120 min; and 8.6 ± 0.5 pmol/l [saline] vs. 7.1 ± 0.8 pmol/l [ghrelin], P = 0.14 at t = 300 min).

Data on resting energy expenditure and respiratory quotient (RQ) in study 2 are given in Table 1. Energy expenditure, RQ, or lipid oxidation were not significantly affected by ghrelin in the basal or in the clamp period. The increase in RQ (RQ_clamp − RQ_basal) during the clamp, however, was larger in the saline study (0.07 ± 0.01 vs. 0.03 ± 0.01 [ghrelin], P = 0.03).

Ghrelin infusion induced an 80% increase in FFAs to 0.62 ± 0.03 mmol/l at t = 120 (P < 0.05), followed by a return to placebo levels during the clamp period (Fig. 2C).

Ghrelin induced a rapid increase in plasma glucose levels with a peak value of 6.1 ± 0.2 mmol/l at t = 120 min (P = 0.009) (Fig. 2E). During the clamp, glucose levels gradually decreased toward postabsorptive levels on the ghrelin day, resulting in comparable glucose levels during the final 30 min of the clamp. The GIR was significantly decreased during ghrelin administration (Fig. 4A) (P < 0.01), and the corresponding M value was reduced by ∼60% (P < 0.001). Ghrelin did not significantly impact glucose metabolism in the basal state (Table 1 and Fig. 4C) but reduced the rates of oxidative, nonoxidative, and total glucose disposal during the clamp period.

Interstitial muscle glucose levels fluctuated in parallel with those in the circulation (Fig. 2F). By contrast, interstitial glucose in fat remained stable also during ghrelin infusion. Ghrelin did not significantly influence the levels of interstitial glycerol, lactate, or urea in either tissue (data not shown).

Densitometric quantitative bar graphs and representative Western blots from skeletal muscle biopsies are provided in Fig. 3. Administration of ghrelin translated into STAT5 phosphorylation in the healthy subjects but not in the hypopituitary patients. Ghrelin exerted no effects on total protein levels, and no effects were recorded with regard to AMPK, ACC, STAT3, ERK-1 or -2, Akt, or AS160 phosphorylation in study 2. Insulin receptor substrate–associated PI3K activity was also not modified by ghrelin infusion (data not shown).

We document for the first time that ghrelin induces peripheral insulin resistance and stimulates lipolysis in the absence of GH and cortisol release. This investigation is also the first to document that ghrelin-induced endogenous GH release translates into Janus kinase/STAT signaling in skeletal muscle.

In some human studies, ghrelin administration increases plasma levels of glucose and FFAs (6–8,16) and reduces glucose disposal (5,8), indicating insulin resistance. These effects are, however, partly attributable to GH and/or cortisol secretion (5–8). There are both clinical and in vitro data to suggest that ghrelin directly suppresses glucose-induced insulin secretion from the β-cell (17), whereas administration of a ghrelin antagonist does the opposite (18). Moreover, ghrelin knockout mice display enhanced glucose-induced insulin release from isolated islets (18,19) and exhibit increased peripheral insulin sensitivity (19). Ghrelin/GHS-R double knockout mice show lower glucose levels after a glucose tolerance test and a more rapid drop in plasma glucose levels after an insulin tolerance test (20).

Our study in hypopituitary men demonstrates that ghrelin in humans directly suppresses insulin-stimulated glucose disposal and stimulates FFA release. The observations at the whole-body level were corroborated by the demonstration of increased concentrations of glucose in skeletal muscle interstitial tissue. Previously, it has been demonstrated that in humans glucose disposal rates during hyperinsulinemic clamp conditions are predominantly determined by the rate of glucose uptake into skeletal muscle (21). By contrast, we observed no significant effect of ghrelin on hepatic insulin sensitivity. AMPK appears to be a key messenger in ghrelin signaling in several tissues (22–24), and AMPK is also a recognized cellular energy sensor (25). In our study, we did not record any significant effect of ghrelin on the activation of either AMPK or ACC in skeletal muscle. The biopsies, however, were only obtained in the basal period, which limits assessment of aberrations in insulin signaling pathways. The molecular mechanisms by which ghrelin causes insulin resistance in humans thus remain to be further ascertained and should be studied in both basal and insulin-stimulated conditions.

Serum FFA levels increased in response to ghrelin, which did not translate into increased lipid oxidation. It should be noted that hypopituitary adults are moderately obese, which is likely to influence any effect of ghrelin on FFA turnover. More importantly, infusion of ghrelin does not imitate the secretory pattern of endogenous ghrelin. Furthermore, this study was powered to detect effects of ghrelin on insulin sensitivity (8), and lack of significance with regard to lipid metabolism may be due to a β error.

In conclusion, we demonstrate for the first time in humans that ghrelin directly induces lipolysis and resistance to insulin-stimulated glucose disposal. We also demonstrate that ghrelin-induced GH release translates into GH signaling in skeletal muscle, emphasizing the significance of accounting for GH when evaluating any effects of ghrelin. The physiological significance of the direct metabolic effects of endogenous ghrelin remains unclear, but we propose that ghrelin in concert with GH partition substrate metabolism during conditions of energy shortage in such a way as to restrict glucose utilization to insulin-independent tissues such as the brain.

This study was supported by an unrestricted grant from Novo Nordisk as well as grants from the Novo Nordisk Foundation, the A.P. Moller Foundation, the World Anti-Doping Agency, and the FOOD Study Group/Ministry of Food, Agriculture, and Fisheries and Ministry of Family and Consumer Affairs. Microdialysis catheters were kindly supplied by Roche.

The GCP unit of Aarhus University Hospital is acknowledged for monitoring that GCP guidelines were followed. S. Sorensen, M. Moller, E. Carstensen, and E. Hornemann are acknowledged for excellent technical assistance.