Ghrelin, a novel GH-releasing peptide isolated from human and rat stomach, stimulates food intake and GH secretion. We determined plasma ghrelin concentrations in patients with simple obesity, anorexia nervosa, and type 2 diabetes mellitus by RIA. We also studied plasma ghrelin responses to glucose load and meal intake and obtained a 24-h profile of circulating ghrelin in humans.
Plasma ghrelin concentrations in patients with simple obesity and anorexia nervosa were lower and higher, respectively, than those of healthy subjects with normal body weight. Among those with type 2 diabetes mellitus, obese patients had lower and lean patients higher fasting plasma ghrelin concentrations than normal-weight patients. Fasting plasma ghrelin concentration was negatively correlated with body mass index in both nondiabetic and diabetic patients. Plasma ghrelin concentrations of normal subjects decreased significantly after oral and iv glucose administration; a similar response was also observed in diabetic patients after a meal tolerance test, reaching a nadir of 69% of the basal level after the meal. Circulating plasma ghrelin showed a diurnal pattern with preprandial increases, postprandial decreases, and a maximum peak at 0200 h. This study demonstrates that nutritional state is a determinant of plasma ghrelin in humans. Ghrelin secretion is up-regulated under conditions of negative energy balance and down-regulated in the setting of positive energy balance. These findings suggest the involvement of ghrelin in the regulation of feeding behavior and energy homeostasis.
MEMBERS OF THE family of seven-transmembrane, G protein-coupled cell surface receptors (GPCRs) respond to a wide variety of signals, including photons, amines, peptides, and proteases. Recent efforts in genomics research have identified a large number of cDNA sequences that encode orphan GPCRs, that is, putative GPCRs without known cognate ligands. We undertook a systematic biochemical search for endogenous peptide ligands of multiple orphan GPCRs, using a cell-based reporter system. These screening experiments led to the identification of a novel peptide that binds to a previously described orphan GPCR (1), the growth-hormone secretagogue receptor. This peptide, called ghrelin, contains 28 amino acids, including an N-octanoylated Ser 3, and was originally discovered in rat and human stomach (2). Ghrelin-producing endocrine cells, which are most abundant in the oxyntic mucosa of both species, account for about 20% of the oxyntic gland endocrine cell population (3, 4). Ghrelin was found to stimulate GH release in vivo and in vitro (2, 5–12). Ghrelin increased food intake and body weight in rodents when administered centrally and peripherally (11–16). In contrast, an intracerebroventricular administration of anti-ghrelin IgG robustly suppressed feeding (14). These results suggest a role for ghrelin in the regulation of feeding behavior. We established an RIA for ghrelin and found considerably high plasma concentrations of ghrelin in humans and rats (2, 17, 18). The plasma ghrelin concentration in rats increased upon fasting and returned to baseline after refeeding (4, 13, 18). In this study, to investigate the possible involvement of ghrelin in the regulation of metabolic balance, we measured plasma ghrelin concentrations in patients with simple obesity and anorexia nervosa, and lean and obese patients with type 2 diabetes mellitus (DM). We also studied plasma ghrelin responses to glucose load in normal subjects and to a test meal in diabetic patients. Finally, we investigated the 24-h profile of circulating ghrelin in humans.
Subjects and Methods
Subjects
The following groups were studied: 28 healthy controls [14 men and 14 women; mean age ± sem, 30.4 ± 4.1 yr; body mass index (BMI), 19.8–24.6, mean ± sem, 22.7 ± 0.4]; 17 patients with anorexia nervosa (1 man and 16 women; 22.2 ± 2.3 yr; BMI, 9.3–17.3, 14.2 ± 0.5); 11 patients with simple obesity (4 men and 7 women; 35.1 ± 3.7 yr; BMI, 26.3–40.5, 30.4 ± 1.2); and 42 patients with type 2 DM without nephropathy (18 men and 24 women; 58.5 ± 1.6 yr; BMI <18.5, n = 4; 18.5 ≤ BMI < 25, n = 19; BMI ≥25, n = 19). Among the anorexia nervosa patients, 9 were of the restricting type and 8 of the binge eating/purging type according to the Diagnostic and Statistical Manual of Mental Disorders, 4th edition, American Psychiatric Association, 1994. Among the diabetic patients, 12 were treated with diet and exercise, 22 with oral hypoglycemic agents, and 8 with insulin. All subjects were classified into lean (BMI <18.5), normal (18.5 ≤ BMI < 25), and obese (BMI ≥25) categories according to the criteria of the Japan Diabetes Society and Japan Society for the Study of Obesity. All subjects were clinically stable at the time of evaluation and had no evidence of gastrointestinal disease or cachectic states such as cancer, thyroid disease, liver disease, or infection. Patients with renal dysfunction (serum creatinine ≥1.5 mg/dl) were excluded. Blood was collected at 0800 h after an overnight fast.
Protocol
Healthy volunteers who were within 10% of ideal body weight (9 men and 10 women; 25.3 ± 2.5 yr) were given a 75 g/225 ml glucose solution orally. They also were given 225 ml distilled water orally on a different day. Blood was collected 0, 30, 60, and 120 min after administration. These subjects also were given 10 g/20 ml glucose iv for 2 min. Blood was collected 0, 3, 5, 10, 15, 30, and 60 min after the glucose injection. Seven diabetic patients without triopathy (3 men and 4 women; 52.6 ± 5.6 yr; BMI, 21.3–37.0, 26.0 ± 2.0) were given a test meal (450 cal, 50% carbohydrate, 30% fat, and 20% protein) at 0800 h. Blood was collected 0, 30, 60, and 120 min after eating. For the 24-h monitoring study, 10 healthy men of normal weight (29.6 ± 3.4 yr) were given meals at 0800, 1200, and 1800 h, and blood was collected every 2 h over that 24-h period. Plasma glucose was measured by the glucose oxidase method. Plasma insulin was determined by an RIA kit (Shionogi, Tokyo, Japan). The Institutional Committee of Miyazaki Medical College approved the protocol, and all subjects provided written informed consent before participation.
Ghrelin assay
Blood was drawn into chilled tubes containing EDTA·2Na (1 mg/ml) and aprotinin (500 U/ml). Plasma ghrelin was measured by an RIA developed in our laboratory (17). In brief, antiserum against the C-terminal region of human ghrelin was raised in New Zealand white rabbits that were immunized against synthetic human ghrelin [13–28] that had been coupled with maleimide-activated mariculture keyhole limpet hemocyanin. Human Tyr0-ghrelin [13–28] was radioiodinated by the lactoperoxidase method for use in the assay. Inter- and intra-assay variation was less than 8 and 6%, respectively. The limit of detection of this assay is 12 fmol per tube of human ghrelin. Two milliliters of plasma were diluted with 2 ml of 0.9% saline and applied to a Sep-Pak C-18 cartridge (Waters Corp., Milford, MA) pre-equilibrated with 0.9% saline. The cartridge was washed first with saline and then with a 0.1% trifluoroacetic acid solution, and peptides were eluted with a 60% acetonitrile (CH3CN) solution containing 0.1% trifluoroacetic acid. The eluate was evaporated, reconstituted with RIA buffer, and subjected to RIA analysis. A diluted sample or a standard peptide solution (100 μl) was incubated for 24 h with 100 μl of the antiserum diluent (final dilution, 1:20,000). The tracer solution (16,000 cpm/100 μl) was added, and the mixture was incubated for 24 h. Bound and free ligands were separated by the second antibody method. All procedures were performed at 4 C. Recovery of human ghrelin added to the plasma was 90.7 ± 4.0% (n = 6).
Statistical analyses
Groups of data (mean ± sem) were compared using ANOVA and a post hoc Fisher’s test. A P value of less than 0.05 was considered statistically significant. Correction coefficients were calculated by linear regression analysis.
Results
The antiserum raised against human ghrelin showed no cross-reactivity with human insulin, glucagon, somatostatin, pancreatic polypeptide, neuropeptide Y, or motilin (Fig. 1). A serial dilution curve of a human plasma sample was parallel to a standard curve for human ghrelin (Fig. 1).
Standard RIA curve for human ghrelin and cross-reactivity of antiserum. Inhibition of radiolabeled human ghrelin binding by serial dilutions of human ghrelin (○), plasma extract (•), and insulin, glucagon, somatostatin, pancreatic polypeptide, neuropeptide Y, and motilin (□). The numeral 1 in the plasma dilution curve denotes 2 ml.
There was no significant difference in plasma ghrelin concentration by gender or age in normal subjects studied (data not shown). Mean plasma ghrelin concentrations in patients with anorexia nervosa or simple obesity were 225% (P < 0.001) and 68% (P < 0.05), respectively, of the control value for healthy normal-weight subjects (132.4 ± 13.1 pm) (Fig. 2A). The mean fasting plasma ghrelin concentration in these three groups of subjects was negatively correlated with BMI (Fig. 2B). Mean plasma ghrelin concentrations in lean and obese type 2 diabetic patients were 190% (P < 0.005) and 66% (P < 0.05) of those found in corresponding normal-weight diabetics (129.3 ± 14.2 pm) (Fig. 3A). Also observed in diabetic patients was the negative correlation of fasting plasma ghrelin concentration and BMI (Fig. 3B). There was no significant difference in plasma ghrelin concentrations between normal-weight healthy subjects and normal-weight type 2 diabetics. Also in both lean and obese subjects, there was no statistical difference in plasma ghrelin concentrations between diabetic and nondiabetic groups.
A, Comparison of plasma ghrelin concentrations in anorexia nervosa patients (BMI <18.5), healthy controls (18.5 ≤ BMI < 25), and simple obesity patients (BMI ≥25). B, Negative correlation between plasma ghrelin concentration and BMI for subjects in A.
A, Comparison of plasma ghrelin concentration in lean, normal-weight, and obese type 2 diabetic patients. B, Negative correlation between plasma ghrelin concentration and BMI for diabetic patients in A.
The mean plasma ghrelin concentration in normal subjects decreased after administration of an oral glucose load, reaching a nadir of 66% (92.1 ± 19.2 pm) of the basal level 60 min after the glucose load, and increasing thereafter (Fig. 4A). Normal responses were observed for plasma glucose (Fig. 4A) and insulin levels (data not shown). The mean plasma ghrelin concentration in the same subjects did not change in response to an administration of the same volume of water (Fig. 4A), but decreased rapidly after an iv glucose administration, reaching a nadir of 68% (93.1 ± 10.4 pm) of the basal level 15 min after administration and increasing thereafter (Fig. 4B). In the meal tolerance test, the mean plasma ghrelin concentration for diabetics decreased, reaching a nadir of 69% of the basal level 60 min after eating and increasing thereafter (Fig. 5). A 24-h plasma ghrelin profile from 10 normal subjects is shown in Fig. 6. The circulating plasma ghrelin level showed a diurnal pattern with preprandial increases and postprandial decreases during the daytime and a maximum peak at 0200 h.
A, Plasma ghrelin responses to 75 g oral glucose tolerance test (○) and oral water load (▵) in normal subjects. *P < 0.0001 in plasma ghrelin vs. 0 min. B, Plasma ghrelin response to 10 g iv glucose tolerance test in normal subjects. *, P < 0.05, **, P < 0.005 in plasma ghrelin vs. 0 min. The results are represented as percentages of the basal level.
Plasma ghrelin response to a meal tolerance test in type 2 diabetic patients. *, P < 0.05 vs. 0 min.
A 24-h plasma ghrelin profile from normal subjects. Arrows denote meal times.
Discussion
Ghrelin mRNA expression in the rat stomach is up-regulated upon fasting and returns to the control value after refeeding (18). Furthermore, the plasma concentrations of ghrelin in the gastric and truncal veins of normal rats increase in response to fasting and decrease upon refeeding (4, 13, 18). Up-regulation of ghrelin expression under conditions of negative energy balance and down-regulation in the setting of positive energy balance appear to represent a negative feedback mechanism to maintain energy homeostasis. This study examines the effects of chronic and acute feeding states on plasma ghrelin concentration in humans. Two ghrelin-specific RIAs have been established; one recognizes the octanoyl-modified portion and another the C-terminal portion of ghrelin (17). In this study, the latter was used due to the instability of acylated ghrelin relative to its des-acylated counterpart, making its measurement from stored plasma samples unreliable.
Plasma ghrelin concentrations were higher in patients with anorexia nervosa and lower in patients with simple obesity compared with normal-weight control subjects. Tschöp et al. (19) have measured plasma ghrelin concentrations in normal and obese Caucasian and Pima Indian individuals. They reported that fasting plasma ghrelin concentration negatively correlates with percentage of body fat, and fasting insulin and leptin concentrations. We also found that fasting plasma ghrelin concentration in normal subjects and patients with anorexia nervosa, simple obesity, or type 2 DM correlated negatively with BMI within each group. If the action of ghrelin in humans is similar to that in rodents, ghrelin should stimulate feeding. Anorexia nervosa is an eating disorder in which patients have obsessive ideation about body weight. Reduced feeding in anorexia nervosa patients who have high plasma ghrelin concentrations suggests that they may have decreased sensitivity to circulating ghrelin, or that there may be a central system regulating energy homeostasis that overcomes the effect of ghrelin in such patients. Ghrelin stimulates GH release in humans (7, 8). Nutritional status plays an important role in the regulation of the GH axis, because GH secretion is markedly influenced by body composition (20–22). Although we did not investigate plasma GH levels in this study, the basal plasma ghrelin concentration probably correlates with that of GH in anorexia nervosa when we take into account the fact that plasma GH is elevated at baseline in this disorder (23, 24).
The stomach is the main source of circulating ghrelin in humans and rats. Ghrelin production in the stomach is localized to A (α)-like cells that do not have glucagon-immunoreactivity but share some morphological features with pancreatic α-cells, including the presence of compact and dense secretory granules (3). Oral and iv administration of glucose to normal subjects decreased their plasma ghrelin concentrations, whereas intake of an equivalent volume of water did not. Also in rats, stomach filling with water did not change plasma ghrelin level (13). Secretion of ghrelin is not affected by stomach expansion. Insulin-induced hypoglycemia up-regulates ghrelin mRNA expression in the rat stomach (18). Taken together, these results indicate that there may be a system in ghrelin-producing cells that responds to plasma glucose concentration. Molecular signals that regulate ghrelin secretion are not known. Further investigation is needed to define the receptors, transporters, and/or channels expressed in ghrelin-producing cells.
Glucose load and food intake lead to a rapid fall in plasma ghrelin concentration, suggesting that plasma ghrelin reflects an acute feeding state and may also serve as an indicator of short-term energy balance. Plasma ghrelin levels are low in human obesity, and are generally associated with increased weight. Ghrelin mRNA expression in the gastric fundi of db/db mice, an obese model characterized by a null mutation in the leptin receptor gene (25, 26), is down-regulated compared with control mice (18). These alterations of ghrelin expression may be a physiological adaptation to long-term positive energy balance.
The diurnal variation of ghrelin secretion in normal subjects appears to be entrained to meal-taking and sleep. The temporal pattern of plasma ghrelin concentration consists of a rise just before the onset of meals and a postprandial decline during the daytime, followed by a much greater increase culminating at 0200 h. The physiological signals that initiate eating in humans are poorly understood. A preprandial rise in plasma ghrelin concentration suggests that ghrelin may be a candidate for a meal-initiation signal. A 24-h profile of human plasma GH concentration (20, 27) resembles that of ghrelin. Human plasma GH also decreases after meals and peaks at 0200–0300 h. Additional studies are needed to examine the potential link between ghrelin and GH diurnal changes in humans.
In summary, the present study demonstrates that nutritional state is a determinant of plasma ghrelin concentration in humans. Ghrelin secretion is up-regulated under conditions of negative energy balance and down-regulated in the setting of positive energy balance. These findings shed new light upon the involvement of the novel gastrointestinal peptide, ghrelin, in the regulation of feeding behavior and energy homeostasis.
This study was supported in part by grants-in-aid from the Ministry of Education, Culture, Sports, Science and Technology, Japan; the Ministry of Health, Labor and Welfare, Japan; the NOVARTIS Foundation (Japan) for the Promotion of Science; the Society of Molecular Mechanism of the Digestive Tract; Mitsui Life Foundation; and The Foundation for Growth Science (to M.N.).
