Paradoxical Stimulation of Glucagon Secretion by High Glucose Concentrations

Islets of Langerhans were isolated with collagenase (Boehringer Mannheim, Mannheim, Germany) from C57 BL/6 mice. Local ethics committees approved the experimental procedures. The animals were killed by decapitation under anesthesia with CO₂. The peritoneal cavity was opened, and a collagenase solution was injected into the bile-pancreatic duct to expand the pancreas (glucagon secretion experiments). The pancreas was excised and cut into small pieces, which were digested with collagenase to obtain free islets of Langerhans. The lower duodenal part of the pancreas was rejected to avoid islets with cells producing pancreatic polypeptide (12). The freshly isolated islets were used for studies of glucagon and insulin secretion or preparation of free cells. Free cells were obtained by incubating the islets for 4 min at 37°C in Ca²⁺-deficient medium containing 0.5 mmol/l EDTA and 0.05% trypsin (Invitrogen, Carlsbad, CA) followed by brief shaking. The cells were suspended in RPMI 1640 medium with 5.5 mmol/l glucose (Gibco, Paisley, Scotland) supplemented with 10% FCS (Gibco), 100 IU/ml penicillin, 100 μg/ml streptomycin, and 30 μg/ml gentamicin. Small samples of this suspension (15 μl) were applied to the centers of poly-l-lysine-coated (Sigma Chemical, St. Louis, MO) circular 25-mm coverslips. The coverslips were then kept for 60 min in a culture incubator at 37°C with a humidified atmosphere of 5% CO₂ to allow cells to settle and begin attachment. More medium was then cautiously added, and the cells were cultured for 1–3 days.

Glucagon-secreting clonal hamster In-R1-G9 cells (13) were provided by Dr. Björn Olde (Lund University) with permission from Dr. Jacques Philippe (University of Geneva, Geneva, Switzerland). The cells were cultured on plastic dishes or culture bottles in GlutaMAX-containing RPMI 1640 medium with 11 mmol/l glucose (Gibco) supplemented with 10% FCS, 100 IU/ml penicillin, and 100 μg/ml streptomycin. For secretion studies, the In-R1-G9 cells were seeded at a density of 100,000 cells/well in 48-well plates and cultured as described above for 3 days.

The cells were loaded with the Ca²⁺ indicator fura-2 during a 40-min incubation at 37°C in a buffer containing 0.5 mg/ml BSA (Sigma), 125 mmol/l NaCl, 4.8 mmol/l KCl, 1.2 mmol/l MgCl₂, 1.28 mmol/l CaCl₂, 3 mmol/l glucose, 1 μmol/l fura-2 acetoxymethyl ester (Molecular Probes, Eugene, OR) (0.1% dimethylsulfoxide; Sigma), and 25 mmol/l HEPES (Sigma) with pH adjusted to 7.4 with ∼13 mmol/l NaOH. The coverslips with the attached cells were used as exchangeable bottoms of an open chamber. The chamber volume was 0.16 ml, and the cells were superfused at a rate of 1 ml/min. The superfusion chamber was placed on the stage of an inverted Nikon Diaphot microscope equipped with an epifluorescence illuminator and a 40× oil immersion fluorescence objective. The chamber holder and the objective were maintained at 37°C by custom-built thermostats. The epifluorescence illuminator was connected through a 5-mm-diameter liquid light guide to an Optoscan monochromator (Cairn Research, Faversham, U.K.) with rapid grating and slit width adjustment and a 150 W xenon arc lamp. The monochromator provided excitation light at 340 nm (2.8-nm half-bandwidth) and 380 nm (2.5-nm half-bandwidth), and emission was measured at >515 nm by an intensified CCD camera. The Metafluor software (Universal Imaging, Downingtown, PA) controlled the monochromator acquiring fluorescence images of 30 accumulated frames at 340 and 380 nm every 4 s. [Ca²⁺]_i images were calculated from 340-to-380 nm ratio images as previously described (14).

The α-cells were initially selected by their small size and [Ca²⁺]_i response to epinephrine (Sigma) (14,15), which is not shared by β-cells (14) and δ-cells (16). Only α-cells confirmed by immunostaining were included in the analyses (14).

Batches of 8–12 freshly isolated islets were preincubated for 30 min at 37°C in 1 ml Krebs-Ringer buffer (pH 7.4) supplemented with 10 mmol/l HEPES, 0.1% BSA, and 1 mmol/l glucose. Each incubation vial was gassed with 95% O₂ and 5% CO₂ to obtain constant pH and oxygenation. The islets were then incubated for 1 h at 37°C in the Krebs-Ringer buffer containing different glucose or 3-o-methyl glucose concentrations and 4.8 mmol/l K or 30 mmol/l K⁺ plus 250 μmol/l diazoxide. At the end of the incubation, aliquots of the medium were removed and frozen pending the radioimmunoassay of glucagon and insulin (17).

The 3-day cultured In-R1-G9 cells were carefully washed and preincubated for 30 min at 37°C in the Krebs-Ringer buffer containing 1 mmol/l glucose (1 ml/well). The medium was then replaced by one containing different glucose concentrations and incubated for 60 min at 37°C. Aliquots of the medium were removed and frozen pending assay of glucagon, insulin, and somatostatin (17,18).

Dose-response relationships for the effect of glucose on glucagon and insulin secretion were analyzed with ANOVA and paired Student’s t test. The reaction patterns in individual α-cells were analyzed with two-tailed Fisher’s exact test. All calculations were made by SigmaStat software (Systat Software, Erkrath, Germany).

Figure 1 shows how glucagon and insulin secretion from mouse islets and In-R1-G9 cells depend on the glucose concentration. In the mouse islets, glucagon secretion was more than half-maximally inhibited by 4 mmol/l glucose, and the maximal effect was reached at 7–8 mmol/l (Fig. 1, top left panel). With further rise of glucose to 16–20 mmol/l, there was a small but statistically significant reduction of this inhibition (P < 0.005–0.05). However, above 20 mmol/l, inhibition was dramatically reversed into a small stimulation at 25–30 mmol/l glucose. In parallel insulin measurements from the same islets, threshold stimulation of insulin release (18%) was obtained at 7 mmol/l glucose and maximal secretion at 25–30 mmol/l sugar (Fig. 1, bottom left panel). Stimulation of glucagon secretion at 25–30 mmol/l glucose could not be attributed to an unspecific action of the high sugar concentrations because 30 mmol/l nonmetabolizable glucose transport analog 3-o-methyl glucose was inhibitory (Fig. 1, top left panel). Separate control measurements under identical conditions revealed that somatostatin secretion was stimulated by glucose and significantly higher at 30 than at 20 mmol/l sugar (data not shown), indicating that glucose stimulation of glucagon secretion was not due to decreased release of somatostatin. There was a U-shaped dose-response relationship for glucose-regulated glucagon secretion also in In-R1-G9 cells with maximal inhibition at 7 mmol/l (Fig. 1, right panel). In this case, the stimulatory component was left-shifted with significant stimulation already at 12 mmol/l glucose. The In-R1-G9 cells did not secrete detectable amounts of insulin or somatostatin.

Rise of [Ca²⁺]_i is probably the most important trigger of pancreatic islet hormone secretion. Accordingly, the inhibition of glucagon secretion obtained with elevation of glucose was associated with lowering of [Ca²⁺]_i. In 57% (8 of 14; P < 0.01) of individual mouse α-cells selected for showing [Ca²⁺]_i oscillations at 1 mmol/l glucose, [Ca²⁺]_i was reduced to basal levels at 20 mmol/l sugar (Fig. 2). After further elevation to 30 mmol/l glucose, 86% (12 of 14; P < 0.001) of the α-cells were inhibited.

To clarify the role of [Ca²⁺]_i in glucose-regulated glucagon secretion, mouse islets were studied also under hyperpolarizing conditions, preventing voltage-dependent influx of the ion. This was achieved by exposure to 250 μmol/l diazoxide, which clamps the membrane potential close to the equilibrium potential for K⁺ by activating the ATP-sensitive K⁺ channels (K_ATP channels) that are present in all types of islet cells (12,14,16,19). Hyperpolarization in the absence of glucose had no effect on [Ca²⁺]_i of resting β-cells but lowered [Ca²⁺]_i of oscillating α-cells (not shown). In the presence of diazoxide, introduction of 7 mmol/l glucose resulted in a small reduction of basal [Ca²⁺]_i in 78% of the α-cells (14 of 18; P < 0.001) and all of 12 β-cells (P < 0.001; Fig. 3). Further elevation of glucose to 30 mmol/l had no effect on the α-cells but induced a more pronounced additional reduction of [Ca²⁺]_i in 23 of 25 β-cells (P < 0.001). These data are consistent with stimulated sequestration and outward transport of Ca²⁺ (14,20,21) being saturated at lower glucose concentrations in α-cells (14) than in β-cells (21). Measurements of hormone secretion under similar conditions demonstrated that hyperpolarization with diazoxide reduced glucagon release at 0–1 mmol/l glucose by 70–80% of the maximal glucose inhibition (Fig. 4, top panel). Nevertheless, the additional inhibition obtained by raising the sugar concentration to 7 mmol/l remained statistically significant. Glucagon secretion at higher glucose concentrations was not diminished during hyperpolarizing conditions and even slightly exceeded control secretion in the 12–30 mmol/l range. Basal insulin secretion at 0–7 mmol/l glucose was unaffected by the hyperpolarization (Fig. 4, bottom panel), whereas insulin secretion at 12–30 mmol/l glucose was reduced by ∼70% compared with control.

Glucose initiates insulin release by raising [Ca²⁺]_i in β-cells. The sugar also has an amplifying effect on secretion, which is apparent when [Ca²⁺]_i is held constant at an elevated level (22). This situation can be achieved by clamping the membrane potential at a depolarized level (30 mmol/l K⁺ plus diazoxide) to activate voltage-dependent Ca²⁺ influx. We used this approach to clarify whether glucose also has an amplifying effect on glucagon secretion. Figure 5 shows that depolarization by combining diazoxide with 30 mmol/l K⁺ resulted in rapid increases of [Ca²⁺]_i in both α- and β-cells. Subsequent introduction of 3 mmol/l glucose caused a small reduction of [Ca²⁺]_i in 3 of 19 α-cells, but there was no change of [Ca²⁺]_i when increasing the glucose concentration from 0 to 3, 7, and 30 mmol/l in the remaining 16 α-cells. However, 14 of 16 β-cells responded with temporary reduction of [Ca²⁺]_i after introducing at the highest glucose concentration (P < 0.001), supporting previous observations that stimulated Ca²⁺ sequestration and outward transport can transiently affect [Ca²⁺]_i in this cell type also when voltage-dependent influx is stimulated (23). Under these depolarizing conditions, basal insulin secretion increased 3.5-fold at 0–7 mmol/l (Fig. 6, bottom panel). Glucose did not affect the depolarization-induced insulin secretion in the 0–7 mmol/l range, but higher concentrations of the sugar had a pronounced amplifying effect. The corresponding glucagon data demonstrated that 30 mmol/l K⁺ plus diazoxide markedly stimulated glucagon secretion in the 0–30 mmol/l range (Fig. 6, top panel). Glucose elevation to 7 mmol/l failed to induce significant inhibition of secretion. Higher concentrations of the sugar still stimulated glucagon release, but this effect was not more pronounced than under control or hyperpolarizing conditions.

It is generally accepted that glucose is an inhibitor of glucagon secretion (1,2). However, its was evident from the present study of mouse pancreatic islets and clonal hamster In-R1-G9 glucagon-releasing cells that inhibition of glucagon secretion by intermediate glucose concentrations is followed by reversal of this effect and even stimulation at higher concentrations of the sugar. The U-shaped dose-response patterns are different from those obtained with perfused rat pancreas (7) and batch-incubated rat islets (8), which only show glucose inhibition of glucagon secretion. A recent observation that glucose stimulates glucagon secretion from purified rat α-cells was taken to indicate that this secretion is regulated in a β-cell–like manner and that the inhibitory effect of glucose requires release of paracrine factors from non–α-cells within the islets (8). The present observation of glucose-stimulated glucagon secretion from mouse islet is principally different from that in purified rat α-cells. Whereas the latter is attributed to a β-cell–like pathway with closure of K_ATP channels leading to depolarization and voltage-dependent Ca²⁺ influx, glucose stimulation of glucagon secretion from mouse islets occurred independent of K_ATP channels and did not require elevation of [Ca²⁺]_i. A reason for the different [Ca²⁺]_i responses is apparently that rat α-cells have a high density of K_ATP channels, exceeding that in rat β-cells (24), whereas mouse α-cells show little K_ATP channel expression (25,26), corresponding to 2% of that in mouse β-cells (27). Depolarization by tolbutamide-induced closure of the K_ATP channels thus increases the electrical activity (24), [Ca²⁺]_i (8), exocytosis (28), and secretion (8) in rat α-cells, but tolbutamide-induced elevation of [Ca²⁺]_i is observed only in a small minority of mouse α-cells (14). Moreover, whereas glucose increases the electrical activity and [Ca²⁺]_i of rat α-cells (24), the sugar hyperpolarizes mouse α-cells (14,27,29) and, as now confirmed, lowers [Ca²⁺]_i (14).

The present study of mouse α-cells and islets and clonal hamster glucagon-releasing cells provides arguments that glucose has direct effects on the α-cells in terms of inhibition and stimulation of glucagon release. Although the In-R1-G9 cells are derived from an insulin-secreting cell line also producing glucagon (13), the presently used cells did not release detectable amounts of insulin. Considering dilutions and detection limits of the assays, this means that maximal theoretical insulin secretion on a molar basis is lower than or similar to glucagon secretion compared with an ∼40-fold excess of insulin secretion in the mouse islets. Also, somatostatin secretion was undetectable, indicating that maximal theoretical secretion of this hormone corresponded to 0.5–4% of glucagon secretion, which is much lower than in the mouse islets (data not shown). It seems unlikely that phenotypic diversity among the In-R1-G9 cells explains inhibitory or stimulatory effects of glucose by paracrine routes, but an inhibitory effect of insulin is not entirely excluded. The observation that most inhibition of glucagon secretion from mouse islets is obtained at glucose concentrations, which fail to stimulate insulin release argues against the involvement of the proposed β-cell factors insulin (30), Zn²⁺ (30,31), and γ-aminobutyric acid (32,33). Moreover, glucose stimulation of glucagon secretion was not diminished, and some glucose inhibition of glucagon secretion remained under conditions expected to hyperpolarize the islet cells and reduce release of paracrine factors.

Three alternative mechanisms have been proposed to mediate a direct glucose inhibition of the α-cell. In conflict with the hyperpolarizing effect of glucose (14,27,29), one of these proposals assumes that glucose paradoxically inhibits glucagon secretion by depolarizing mouse α-cells (26,34). The other two alternatives involve glucose-induced hyperpolarization by activation of the electrogenic Na,K-ATPase (35) or inhibition of a depolarizing store-operated cation influx by glucose-stimulated Ca²⁺ sequestration in the endoplasmic reticulum (14).

Because the present and previous data (14) show that glucose lowers [Ca²⁺]_i in mouse α-cells, the question arises how high concentrations of the sugar stimulate glucagon secretion. In the insulin-releasing β-cells, glucose seems to stimulate insulin secretion by at least three mechanisms. Initiation of secretion is generally attributed to depolarization with increase of [Ca²⁺]_i after closure of K_ATP channels (19). Glucose also has an amplifying effect, which does not require further elevation of [Ca²⁺]_i (22). In addition, there is evidence that increased metabolism can stimulate insulin release independent of [Ca²⁺]_i elevation (36,37). We now confirm all of these actions by showing 1) stimulated insulin secretion after [Ca²⁺]_i elevation due to depolarization with 30 mmol/l K⁺ in the presence of diazoxide; 2) glucose amplification of this secretory response, although the sugar temporarily lowered [Ca²⁺]_i in isolated β-cells; and 3) maintenance of some glucose-stimulated insulin secretion under hyperpolarizing conditions when the sugar lowered [Ca²⁺]_i below the basal level.

The Ca²⁺-dependent and -independent pathways for stimulated secretion were present also in mouse α-cells. 1) Elevation of [Ca²⁺]_i in isolated α-cells correlated to enhanced glucagon release from pancreatic islets. 2) High glucose concentrations stimulated glucagon secretion under hyperpolarizing conditions, although [Ca²⁺]_i of the α-cells was lowered below basal levels. It was recently reported that glucose amplifies Ca²⁺-initiated glucagon secretion from purified rat α-cells (8). However, the presently observed stimulatory effect of glucose on glucagon secretion from mouse islets did not depend on the [Ca²⁺]_i level of the α-cells, indicating that glucose initiates rather than amplifies secretion. It may seem surprising that the α-cells can sense high concentrations of glucose because they express low K_m glucose transporter GLUT1 but lack high K_m GLUT2 (38). The situation is reminiscent of that in human β-cells, which, unlike the GLUT2-expressing rodent β-cells, predominantly express GLUT1 (39). Apparently glucose transport via GLUT1 is not rate limiting in these cases, and the sensing is explained by the presence of the high K_m glucokinase (39,40). The signal transduction of the intriguing glucose stimulation of glucagon secretion remains to be elucidated.

Hypersecretion of glucagon is an important feature of diabetes contributing to glucose dysregulation (1,2). The mechanism underlying this phenomenon is unclear. The present study of mouse islets and clonal hamster glucagon-releasing cells indicated that paradoxical hypersecretion of glucagon in hyperglycemia may depend on a U-shaped dose-response relationship for glucose regulation of secretion. A similar regulation of human α-cells may explain why diabetic patients with hyperglycemia display paradoxical hyperglucagonemia and why a further glucose challenge has been found to stimulate glucagon release (9–11).

This work has received support from the Swedish Medical Research Council (grant 12X-6240), the Swedish Diabetes Association, the Carl Trygger Foundation, Family Ernfors Foundation, Albert Påhlsson’s Foundation, the Crafoord Foundation, and the Scandinavian Physiological Society.