Characterization of GLP-1 Effects on β-Cell Function After Meal Ingestion in Humans

The study was undertaken in eight women, all 69 years of age, who were recruited from a larger cohort of women in the city of Malmö, Sweden and who participated in a population study (19). Because we aimed to establish the glucose dependency of the effects of GLP-1, we included subjects with a range of fasting glucose from 3.7 to 10.3 mmol/l. Six of the subjects were healthy, without diabetes or any known history of cardiovascular, liver, or kidney diseases, and they were not taking any medication known to affect carbohydrate metabolism. Two subjects had type 2 diabetes treated with diet for the past 6 and 10 months, respectively. These patients had no diabetic complications or any cardiovascular, liver, or kidney diseases, and they were not taking any medication known to affect carbohydrate metabolism. The mean (± SD) fasting plasma glucose level of the study subjects was 7.4 ± 2.4 mmol/l (range 3.7–10.3) and the BMI was 28.4 ± 7.6 kg/m² (range 22.4–43.9). The Ethics Committee of Lund University, Sweden, approved the study. All subjects gave written informed consent before entrance into the study.

Two meal tests were performed in each subject, in random order, with at least 3 weeks between. Intravenous catheters were inserted after an overnight fast into antecubital veins in both arms and baseline samples were collected at −5 and −2 min. Thereafter, synthetic GLP-1 (Clinalfa, Laeufelingen, Switzerland) was infused intravenously at a rate of 0.75 pmol · kg^–1 · min^–1 for 90 min. The dose of GLP-1 was selected from previous reports in which GLP-1 was shown to reduce prandial glucose (5,6). In the control test, saline was infused instead of GLP-1. At the start of GLP-1 or saline infusion, a breakfast was served and eaten within 5 min. The breakfast consisted of 450 kcal with 50% of energy as carbohydrate, 23% as fat, and 27% as protein. Additional samples were taken regularly for 120 min.

Blood samples for determination of insulin, C-peptide, and glucagon were collected in chilled tubes containing EDTA (7.4 mmol/l, final concentration) and aprotinin (500 kallikrein inhibitor units/ml blood; Novo Nordisk, Bagsværd, Denmark). They were immediately centrifuged at 4°C and plasma frozen at −20°C until analysis in duplicate. Plasma insulin and C-peptide concentrations were analyzed with double-antibody radioimmunoassay techniques (Linco Research, St. Charles, MO) using guinea pig anti-human insulin antibodies, human insulin standard, mono-¹²⁵I-Tyr–labeled human insulin, guinea-pig anti-human C-peptide antibody, human C-peptide standard, and ¹²⁵I-human C-peptide as tracer. Analysis of glucagon was performed with double-antibody radioimmunoassay using guinea-pig anti-glucagon antibodies specific for pancreatic glucagon, ¹²⁵I-glucagon as tracer, and glucagon standard (Linco). Blood samples for determination of GLP-1 were collected into chilled tubes containing EDTA and aprotinin as above with addition of diprotin A (0.1 mmol/l final concentration; Bachem, Bubendorf, Switzerland). They were kept on ice until centrifugation at 4°C. Plasma was separated and stored at −20°C until analysis for amino-terminal GLP-1 immunoreactivity by radioimmunoassay using antiserum 93242 (20), which has a cross-reactivity of ∼10% with GLP-1_1–36amide and of <0.1% with GLP-1_8–36amide and GLP-1_9–36amide. The detection limit is 1 pmol/l. High-performance liquid chromatography supports the use of radioimmunoassays with this specificity for determination of intact GLP-1 (21). The intra-assay coefficient of variation is <6%. It should be emphasized that the assay detects the active form of GLP-1, i.e., GLP-1_7–36amide, and not the total circulating pool of GLP-1, which to a large degree (>80%) includes the inactive form of GLP-1, GLP-1_9–36amide. Glucose was determined with the glucose oxidase technique.

Parameters of β-cell function were determined by reconstructing insulin secretion from C-peptide data by a deconvolution technique based on a two-compartment model of C-peptide kinetics (22), as previously described (17,18). The model describes the relationship between insulin secretion, S(t), and glucose concentration. Insulin secretion consists of two components, according to the equation S(t) = S_g(t) + S_d(t). The first component, S_g(t), represents the dependency of insulin secretion on the absolute glucose concentration (G) at any time point and is characterized by a dose-response function, f(G), relating these variables. Two characteristic parameters of the dose response are its mean slope (in the observed glucose range) and insulin secretion at a fixed glucose concentration of 7 mmol/l (approximately the mean basal glucose concentration). The dose response is modulated by a potentiation factor, P(t), which incorporates glucose-mediated and non–glucose-mediated potentiation (i.e., by non-glucose substrates, gastrointestinal hormones, and neurotransmitters), i.e., S_g(t) = P(t)f(G). The potentiation factor is, by definition, a positive function of time and averages 1 during the experiment (17,18). The second insulin secretion component, S_d(t), represents a dynamic dependency of insulin secretion on the rate of change of glucose concentration, expressed as S_d(t) = K_ddG(t)/dt if dG(t)/dt > 0 and S_d(t) = 0 if dG(t)/dt ≤ 0. This component is called the derivative component and is thus described by a single parameter (K_d). From these estimated model parameters, the dose-response f(G), and the potentiation factor P(t) (17,18), the total and basal insulin secretion were calculated. The excursion of the potentiation factor was quantified using ratios between mean values at time 60 and time 0, i.e., P(60)/P(0).

Data are reported as means ± SE if not stated otherwise. Paired Student’s t test was used to compare differences with versus without GLP-1.

Figure 1 shows the glucose, insulin, C-peptide, glucagon, and GLP-1 profiles before and after meal intake in the eight subjects during intravenous infusion of either saline or GLP-1. It is shown that glucose levels were lower during infusion of GLP-1 than during infusion of saline, whereas insulin and C-peptide levels were higher during GLP-1 than during saline. Glucagon levels were lower during infusion of GLP-1 than during saline. During saline infusion, GLP-1 levels increased from 1.9 ± 0.2 to a maximal level of 5.6 ± 0.3 pmol/l (P < 0.001). During infusion with GLP-1, the GLP-1 concentrations increased from 2.0 ± 0.3 to 27.5 ± 2.5 pmol/l (P < 0.001).

Figure 2 shows the modeled insulin secretion profile before and after the meal. Baseline insulin secretion was not different between the two tests. Meal ingestion alone induced a continuous increase in insulin secretion. GLP-1 markedly augmented this response. The augmentation by GLP-1 was particularly profound during the first 30 min after meal ingestion. Total insulin secretion during the test was higher with GLP-1 than with saline (P = 0.048; Table 1).

Figure 2 also shows the dose-response relationship between plasma glucose concentrations and insulin secretion as obtained in the model. The curves are obtained by reconstructing the dose-response relationship between glucose and insulin secretion for each subject in the presence and absence of GLP-1. The curves, therefore, represent the calculated insulin secretion as dependent on the glucose concentration without any potentiation. It is seen that the dose response is shifted upward by GLP-1; both insulin secretion at 7 mmol/l glucose and the mean dose-response slope are increased by GLP-1 (Table 1). This shows that for a given glucose concentration, insulin secretion is augmented by GLP-1. The parameter of insulin secretion representing the dynamic dependency on the rate of change of glucose concentration (K_d) was not significantly affected by GLP-1 (Table 1).

The potentiation factor obtained from the meal tests with or without infusion of GLP-1 is shown in Fig. 2. The potentiation factor was, by definition, constrained to be 1, on average, during the meal test (17,18). It is seen that during GLP-1 infusion, there was a marked and rapid increase in the potentiation factor already in the beginning of the test. The increase in the potentiation factor by GLP-1, quantified by the ratio of the potentiation factor at time 60 to time 0, was 0.97 ± 0.17 in the control test versus 2.71 ± 0.42 when GLP-1 was infused (P = 0.005). During GLP-1 infusion, the marked increase over the basal level of the potentiation factor (Fig. 2) paralleled that of the GLP-1 concentration (Fig. 1). By pooling the potentiation factor and GLP-1 concentration values of all subjects at all time points, the potentiation factor was positively correlated with GLP-1 levels (r = 0.53, P < 0.0001 after logarithmic transformation). Furthermore, by pooling all experiments, the average increment over basal of the potentiation factor (expressed as ratio) was positively correlated with the corresponding GLP-1 increment (Fig. 2; r = 0.76, P = 0.0006). From the observed nonlinear relationship, a twofold increment in potentiation is predicted for a 10-fold increment in GLP-1 concentration.

In an attempt to relate the effects of GLP-1 on β-cell function to clinical characteristics of the subjects, correlations were performed between the indexes reported in Table 1 with GLP-1 infusion on one hand and fasting plasma glucose or BMI, respectively, on the other hand. However, no significant correlation was found.

GLP-1 is known to augment insulin secretion in response to intravenous and oral glucose as well as to meal ingestion in humans (7–8,10–11,13–16). Besides the established knowledge that the effect of GLP-1 is dose dependent, the detailed mechanism of the β-cell actions of GLP-1 in humans is not established. Recently, it was shown that the prehepatic insulin secretion was dose dependently potentiated by GLP-1 because the peptide increased the linear slope between glucose concentration and insulin secretion when glucose was administered intravenously as a graded infusion (16). The present study examined the influences of GLP-1 on characteristics of β-cell function as obtained by analyses of the insulin secretion profile after meal ingestion, i.e., when GLP-1 is considered of most importance when evaluated as a therapeutic agent in type 2 diabetes. After meal ingestion, GLP-1 suppressed glucose levels, which is a combined effect of the peptide to stimulate insulin secretion, inhibit glucagon secretion, and inhibit gastric emptying as demonstrated before (1–11) and in regard to insulin and glucagon confirmed in the present study.

The different glucose profile after the meal ingestion when infusing saline versus GLP-1 requires modeling of the data for estimation of insulin secretion from the β-cells. In this study, we used a model of β-cell function, as recently described in detail (17,18). The modeling disclosed that GLP-1 markedly increased insulin secretion, despite glucose levels being lower, and that this was particularly powerful during the first 30 min after meal ingestion. After the initial 30 min, the action of GLP-1 on absolute insulin secretion was no longer apparent, but this was mainly a consequence of the lowered glucose levels. The main result of the study is that after meal ingestion in humans, the dose-response function representing the static relationship between glucose concentration and insulin secretion is markedly shifted upward by GLP-1, and that potentiation is strongly stimulated. Therefore, GLP-1 increases both glucose dose response and potentiation.

For appreciation of these findings, it is important to consider that the stimulatory effect of glucose on insulin secretion is not constant during the dynamic course of a meal test. The reason for this is, first, that the insulin response to a meal intake is dependent not only on glucose but also on other nutrients, neurotransmitters, and incretins. These factors contribute to the β-cell function by changing (mainly augmenting) the insulinotropic action of glucose. The model showed that GLP-1 augmented the effect of glucose to stimulate insulin secretion, i.e., for a given glucose concentration, the static release of insulin is increased on average during the test. This augmenting effect on glucose-stimulated insulin secretion is expressed as the potentiation factor, which was markedly increased above basal levels with GLP-1 infusion.

The potentiation factor was, by definition, constrained to be 1, on average, during the test (17,18). By doing this, the corresponding dose response represents the average dose response for the experiment and the reported potentiation factor represents a relative potentiation factor. A potentiation factor <1 means, therefore, a potentiation below average during the experiment and, consequently, a potentiation factor >1 means a potentiation above the average. Therefore, by inspecting Fig. 2, the potentiation factor in the GLP-1 experiment was lower in the basal state than in the control meal test, because GLP-1 increased potentiation. By this definition, it is the different shapes of the curves that illustrate the marked effect of GLP-1 rather than the actual potentiation factor. The relative change in potentiation factor can, however, be calculated by estimating the excursion of the potentiation factor by making a ratio of the factor at min 60 to baseline. This ratio was 2.71 ± 0.42 in the GLP-1 experiment versus 0.97 ± 0.17 in the control test, i.e., GLP-1 augments potentiation factor by more than 2.5-fold. Appreciating this, it is seen that GLP-1 throughout its infusion provided a high potentiation factor, resulting, therefore, in augmented insulin secretion, as reflected by the glucose dose-response relationship. In the previous study on the characteristics of the effects of GLP-1 during a graded intravenous infusion of glucose, it was found that an ∼40-fold increase in GLP-1 levels resulted in an ∼11-fold increase in potentiation (16). This higher efficiency when compared with the present study is probably explained by the prolonged (150-min) glucose infusion, which augments time-dependent potentiation. This effect is not seen in the present study because GLP-1 prevented glucose from being increased after ingestion of a meal. Therefore, the present study shows the efficiency of GLP-1 to augment potentiation of insulin secretion when glucose levels show the physiological perturbations after meal ingestion.

The results of the present study also allowed evaluation of a dose response for GLP-1 effects on insulin secretion, which is potentially interesting for GLP-1 dosage. Therefore, we found that a 10-fold increase in active GLP-1 levels resulted in a twofold increase in insulin secretion, as evaluated from the increase in the potentiation factor. Finally, we did not find evidence that the effect of GLP-1 on β-cell function is dependent on the fasting glucose or BMI, based on results in the subjects exhibiting a broad range of fasting glucose and BMI values.

In conclusion, we have characterized the effects of GLP-1 on β-cell function during a meal test in subjects with a broad range of circulating glucose by means of a model of insulin secretion. We found that GLP-1 markedly increased insulin secretion during the meal intake, despite circulating levels of glucose being reduced. This action of GLP-1 was mainly due to an increase in the average β-cell sensitivity to glucose, mediated by a potentiation effect related to GLP-1 levels.

This work was supported by grants from the Swedish Research Council (grant no. 6834), the Albert Påhlsson Foundation, the Swedish Diabetes Association, the Swedish Society for Medical Research, Lund University Hospital Research Funds, the Faculty of Medicine, Lund University, and the Danish Medical Research Council.

We thank Margaretha Persson and Kerstin Nilsson for nursing assistance during the study and Lilian Bengtsson for expert technical assistance in the analyses of samples.