Cholinergic Regulation of Ghrelin and Peptide YY Release May Be Impaired in Obesity

Eight obese (BMI >30 kg/m²) and eight normal-weight control subjects (BMI <25 kg/m²) were recruited from the obesity outpatient clinic and hospital staff. All participants were nonsmokers, none of the control subjects were taking any medication (with the exception of oral contraceptives), and one of the obese subjects was on antidepressant and antihypertensive therapy. All subjects underwent prestudy screening and had normal findings on laboratory measurements, electrocardiogram, and physical examination. All subjects underwent a 3-h oral glucose tolerance test (75 g glucose). The study protocol was approved by the ethics committee of the Medical University of Vienna, and all subjects gave informed consent before study entry.

The study was conducted with a prospective, randomized, single-blind, placebo-controlled crossover design. For each subject, 4 study days (A–D) were scheduled in randomized order with at least 3-day washout intervals. On study days, subjects arrived between 8:00 and 11:00 a.m. after an overnight fast. Studies were conducted in a quiet room with an ambient temperature of 22°C. Subjects abstained from alcohol and beverages containing caffeine 12 h before the study days.

A plastic cannula (Venflon) was inserted into an antecubital vein at time point −45 min. Blood samples were drawn at time points −30, −15, 0, +15, +30, +60, and +90 min for measurements of plasma ghrelin, PYY, insulin, and glucose. At the same time points (plus time points +45 and +75 min), subjects rated their hunger and satiety feelings on a 10-cm visual analog scale (VAS).

At time point −30, 1 mg atropine (atropimum sulfuricum, Nycomed; study days A and B) or placebo (isotonic saline; study days C and D) was given intravenously over 30 s. At time point 0, subjects received a standard breakfast consisting of two rolls with 15 g butter and 250 ml milk with 10 g commercially available cocoa mix (Benco, Suchard), total calorie content 590 kcal, total fat content 23 g, total carbohydrate content 75 g, and total protein content 18.5 g, on study days A and C only. Researchers and volunteers were blinded to atropine/placebo dosage, and, until time point 0, to breakfast/no breakfast. Blood pressure and pulse rate were monitored during the study period using automated devices and recorded at time points −30, −29, −27, −25, −20, −15, 0, +15, +30, +45, +60, +75, and +90 min.

Samples for plasma hormone measurements were centrifuged immediately at 4°C, and the supernatants were stored at −30°C until analysis. Insulin levels (in micro units per milliliter) were assayed by a commercially available RIA (Pharmacia-Upjohn, Uppsala, Sweden). Blood glucose was determined according to standard laboratory procedures.

Plasma ghrelin (in pico grams per milliliter) was measured with a commercial RIA (Peninsula Labs, San Carlos, CA) that uses I-125–labeled bioactive ghrelin as a tracer and polyclonal antibody raised in rabbits against the COOH-terminal end of human ghrelin. The inter- and intra-assay variations were both <10.9%. PYY (in pico grams per milliliter) was measured using a commercial RIA (Linco Research, St .Charles, MO) with inter- and intra-assay variations of 8.2 and 9%, respectively.

Homeostasis model assessment (HOMA) model index was calculated using HOMA calculator 2.2 (www.dtu.ox.ac.uk). Oral glucose sensitivity index (OGIS) was calculated according to a previously published formula (31) using the OGIS calculator (www.ladseb.pd.cnr.it/bioing/ogis/home.html). Hormone concentrations at single time points and Δhormone levels, Δheart rate, and ΔVAS values at the 4 different study days were compared with one-way ANOVA followed by multiple t tests with Bonferroni correction as post hoc statistics, where appropriate. Baseline parameters and heart rate response between the two groups were compared with unpaired student's t test. Linear regression analysis was performed to evaluate the association (or lack of) between parameters, as indicated. SPSS statistical software release 12.0.1 was used. P < 0.05 was considered statistically significant. Results are presented as means ± SEM.

Baseline characteristics of the subjects are shown in Table 1. The two groups were well matched for age, sex, and height. Apart from having a significantly higher BMI (group-defining criterion), obese subjects had significantly higher fasting glucose and insulin levels and were significantly more insulin resistant according to both HOMA and OGIS indexes. Three obese subjects had impaired glucose tolerance according to their 120-min glucose concentrations (range 143–155 mg/dl).

Ghrelin concentrations on the 4 study days are given in Fig. 1. In control subjects, both atropine and meal ingestion led to a decrease in ghrelin levels. There was a significant difference between ghrelin concentrations on the different study days at time points +30, +60, and +90 min (as compared using one-way ANOVA). When comparing differences between baseline and +90 min ghrelin concentrations (Δghrelin −30/+90), values of the 3 study days were significantly different from the placebo day without breakfast; the study days (A: atropine + breakfast, B: atropine alone, and C: breakfast alone) did not differ significantly from each other (Fig. 1C).

In obese subjects, atropine had no effect on ghrelin concentrations. When compared by ANOVA, ghrelin concentrations did not differ significantly at any single time point on the 4 different study days (Fig. 1B); when comparing Δghrelin −30/+90 values, only breakfast had a significant effect on ghrelin concentrations (Fig. 1D).

PYY concentrations on the 4 study days are given in Fig. 2. In the control group, atropine (study day B) led to a significant decrease of PYY concentrations at time point +90 min compared with breakfast alone (study day C). Using Δ−30/+90 values (differences between baseline and +90 min PYY concentrations) atropine lead to a significant decrease of PYY concentrations compared with all other study days, which did not differ significantly from each other (Fig 2C). In obese subjects, there were no significant differences between any of the study days, at single time points and between the Δ−30/+90 values (Fig. 2B and D, respectively).

Heart rates on the 4 study days are given in Fig. 3. Atropine led to a significant increase in heart rate in both obese and lean subjects (baseline values, 72.0 ± 3 min⁻¹ control vs. 74.5 ± 2 min⁻¹ obese, P = 0.5) with peak values at time point −25 min (5 min after atropine application, 100.3 ± 7 min⁻¹ control vs. 97.8 ± 2 min⁻¹ obese, P = 0.7). Thus, heart rate increase between baseline and time point −25 min (Δ−30/−25) was comparable in lean and obese subjects (23.25 ± 1.8 vs. 28.25 ± 5.8 min⁻¹, P = 0.42, by unpaired Student's t test).

Plasma glucose and insulin concentrations are given in Fig. 4. Atropine alone did not change plasma insulin or glucose concentrations compared with placebo (study days C and D). Meal-induced glucose increase was significantly affected by atropine only in obese subjects. Plasma glucose in control subjects between study days A and B were not significantly different at any single time points, as was the glucose difference between baseline and time point +30 min (Δ−30/+30) (+10.8 ± 4.5 vs. −1.8 ± 1.2 mg/dl, P = 0.178). Insulin concentrations between study days A and B were significantly different at time points +60 and +90 min, as was the difference between baseline and time point +30 min (Δ−30/+30) (23.3 ± 9.4 vs. −2.2 ± 0.9 μU/ml, P = 0.003).

In obese subjects, atropine significantly reduced meal-induced blood glucose at time point +60 min (P = 0.003). Glucose difference between baseline and time point +30 min (Δ−30/30) was not significantly different between study days A and B (5.1 ± 1.7 vs. −3.8 ± 1.7 mg/dl, P = 0.18). Insulin levels were significantly different between study days A and B at time point +60 min (P = 0.001). The Δ−30/+30 values were different between study days A and B, but this difference did not reach statistical significance in the ANOVA analysis (23.3 ± 9.4 vs. 0 ± 0.4 μU/ml, P = 0.133).

An overview of the time course of hunger and satiety VAS is given in Fig. 5. Meal ingestion alone and in combination with atropine led to sustained decreases in hunger ratings and increases in satiety ratings in lean individuals. Hunger scores tended to be lower and satiety scores higher on atropine alone versus placebo days, yet this difference did not reach statistical significance in the ANOVA analysis. When differences between baseline and time point +90 (Δ−30/+90) were compared (Fig. 5E), both hunger and satiety Δ−30/+90 values for all breakfast days were significantly different from all days with no breakfast, but atropine and placebo days did not significantly differ from each other.

In obese subjects, hunger ratings tended to be higher and satiety ratings lower on atropine compared with placebo days (Fig. 5B, again not statistically different for any single time point in the ANOVA analysis). Hunger scores on breakfast days started to slowly increase after reaching a nadir value directly after meal ingestion. Consequently, there were no statistical differences between any of the study days in Δ−30/+90 values. When comparing Δ−30/+90 values for satiety VAS, only study days B and C differed significantly from each other (P = 0.012), all other values were not statistically different from each other (Fig. 5F).

In lean subjects, there was a significant relationship between Δheart rate (−30/−25) and Δghrelin (−30/90), as revealed by linear regression analysis (R² = 0.373, P = 0.0002, Fig. 6A). There was also a significant relationship between Δheart rate (−30/−25) and ΔPYY (−30/90) (R² = 0.227, P = 0.0058, Fig. 6C). In obese subjects, there was no relationship between Δheart rate (−30/−25) and Δghrelin (−30/90) (R² = 0.042, P = 0.27, Fig. 6B) and a weak but significant relationship between Δheart rate (−30/−25) and ΔPYY (−30/90) (R² = 0.136, P = 0.041, Fig. 6D).

In lean subjects, Δghrelin (−30/90) showed a significant correlation with ΔVAS hunger (−30/90) (R² = 0.178, P = 0.016, Fig. 6E), and ΔPYY (−30/90) correlated significantly with ΔVAS satiety (−30/90) (R² = 0.294, P = 0.002, Fig. 6G), whereas there was no significant correlation in obese subjects (Fig. 6F and H, respectively). Insulin resistance as quantified by HOMA and OGIS indexes was not significantly correlated with ghrelin and PYY changes induced by atropine or breakfast (data not shown).

Several effects of ghrelin and PYY are conveyed by the cholinergic autonomous nervous system, and signaling from the gut to the brain is thought to be mediated in part by cholinergic fibers of the vagal nerve (32). We demonstrate in this study that in lean healthy humans, ghrelin and PYY release from the gut is also controlled by the cholinergic system. Ghrelin concentrations are suppressed by the unspecific muscarinic inhibitor atropine to the same amount as by meal ingestion in a nonadditive manner, and PYY release is significantly reduced after atropine application compared with placebo. This suppression of hormone release strongly correlates with the amount of heart rate increase induced by atropine.

In contrast, in obese subjects, there was no clear evidence of an influence of the cholinergic system on ghrelin or PYY release. There were no significant differences in hormone concentrations at any single time points or in Δ−30/+90 values and only a weak association between PYY and heart rate increase, whereas there was no correlation between heart rate and ghrelin differences, although all 4 study days were included in the regression analysis to enhance data spreading. Despite this lack of effect on gut hormone release, atropine-induced heart rate increases in obese subjects was comparable with those in lean subjects. The PYY data are somewhat limited by the higher variability of baseline levels between study days, and the correlation data are not as clear as those for ghrelin. Nevertheless, our data show that not the response to muscarinic inhibition per se but rather the cholinergic control of orexigenic ghrelin and (to a lesser extent) anorexic PYY release is disturbed in obesity. These results are well in line with the large body of evidence showing adrenergic sympathetic dysregulation of gut hormone, in particular PYY release in obesity (33).

Parallel to the lack of cholinergic gut control, there was also a disturbed sensing of hunger and satiety in response to meals in the obese subjects compared with their lean counterparts. VAS showed that the differences in hunger ratings between baseline and postprandial states were not statistically significantly different between any study days (although they tended to be lower on breakfast than on nonbreakfast days), whereas lean subjects reported a sustained suppression of hunger feelings at the end of the study period (90 min after breakfast) on breakfast compared with nonbreakfast days. To a lesser extent, the same could be observed for ratings of satiety, with statistically significant differences between all breakfast versus nonbreakfast days in lean and ratings being only different between study days B (atropine alone) and C (breakfast alone) in obese subjects (although they also tended to be lower on the other breakfast day than on nonbreakfast days). Moreover, in lean subjects, there was a correlation between Δ−30/90 values for hunger ratings and ghrelin and satiety ratings and PYY, respectively (as would be expected when all 4 study days are included in the analysis), but surprisingly again there was a complete lack of association in obese subjects.

Actually, only part of the circulating ghrelin seems to be associated with feelings of appetite, namely the active (acylated) ghrelin, which in circulation becomes rapidly degraded and inactive. Thus, since we measured only total (acylated plus deacylated) ghrelin, our results relate mainly to changes in ghrelin release, and the extension to feelings of hunger and satiety must be regarded with caution. It is possible that differences in ghrelin deacylation between lean and obese subjects rather than differences in ghrelin release were responsible for the observed lack of correlation in obese subjects. Nonetheless, ghrelin release—as shown by the changes in the surrogate parameter total ghrelin—seems to be remarkably different in lean and obese subjects.

It has been proposed that the blunted ghrelin response to meal ingestion could be responsible for the development of obesity in adolescents (34). Whereas in the absence of prospective data it is impossible to determine whether this blunted hormone response is cause or consequence of the disturbed eating behavior in obese subjects, it has been shown that improved sensing of meal-induced hunger suppression is associated with increased postprandial ghrelin drop in subjects losing weight (35). Similarly, it is impossible to know whether the disturbed cholinergic gut hormone regulation was present before the subjects studied here became obese or whether the deregulation of the system developed as a consequence of overeating. Even if cholinergic regulation was the primary factor in gut hormone release, feelings of hunger and satiety are of course not solely driven by changes in gut hormone levels alone: Although atropine caused a comparable drop in ghrelin levels and an increase in PYY levels as breakfast ingestion, there was no accompanying drop in hunger or increase in satiety ratings in lean subjects. Yet, even if a disturbed hormone regulation would lead to only a slight impairment in sensing of hunger and satiety, resulting in a small but daily additional calorie intake, this would ultimately considerably contribute to weight gain in the long term.

To our knowledge, there are no prospective data on the involvement of either gut hormone alterations or impairment of autonomic regulation in the etiology of obesity, and although it seems clear by now that the ample availability of energy dense food is part of the epidemic, it is currently unclear why some people contract obesity while others seem to be resistant. However, there are some data to support the hypothesis of cholinergic pathways being a crucial part of appetite regulation: M3 muscarinic receptor knockout mice are reported to be hypophagic and lean compared with their wild-type littermates (36). The M3 receptor is expressed in the lateral hypothalamus, and the melanin-concentrating hormone (MCH)-containing neurons of this area are apparently responsible for the M3 knockout phenotype. Interestingly, these same neurons also receive projections from AGRP/NPY neurons in the medial hypothalamus (37) (where ghrelin exerts its central effects on appetite).

The obese subjects of this study were also hyperinsulinemic and insulin resistant compared with their lean counterparts. It has been shown that cholinergic regulation is an important part of meal-induced insulin release (38). The M3 knockout mice mentioned above were reported to have improved glucose tolerance despite a blunted increase in serum insulin after oral glucose load that was only partly explained by their leanness; in vitro studies showed a lack of cholinergic stimulation of insulin release from pancreatic islets of these mice (39). A genetic variant of the M3 receptor has been associated with an increased risk for developing type 2 diabetes in Pima Indians (40). Atropine has been shown repeatedly to alter meal-induced glucose increase in humans (38,41–43), as was the case in the subjects studied here. One of these studies (42) reported a greater postprandial attenuation of insulin in obese compared with lean subjects. We report for the first time in this study that, while the alterations of meal-induced insulin and glucose release by atropine were largely comparable in lean and obese subjects, their atropine-induced regulation of gut hormone release was markedly different.

Insulin resistance, as quantified by HOMA and OGIS indexes, was not significantly correlated with ghrelin and PYY changes induced by atropine or breakfast in this study sample. In line with our data, it has been shown in lean but insulin-resistant Pima Indians that postprandial early-phase insulin release is inhibited by atropine but not to the same extent as by pancreatic polypeptide, and it was concluded that the hyperinsulinemia in this population was not due to increased vagal input to the pancreatic β-cell (43). Prolonged glucose infusion in lean healthy humans resulted in vagally mediated compensatory increase in C-peptide secretion but not in alterations of hunger ratings or food intake (44). These data argue against insulin resistance being the driving force of the disturbed cholinergic control observed in this group of obese subjects.

Taken together, the data presented here show that 1) in lean individuals the cholinergic system is involved in the regulation of gut hormone release, 2) gut hormone release is associated with subjective ratings of hunger and satiety, and 3) this regulatory system is markedly impaired in obese subjects. At present, however, it is not clear whether this impaired regulation of gut hormone release actually contributes to the impaired sensing of hunger and satiety seen in these obese subjects and may actively contribute to weight gain and/or hinder weight loss.