Insulin inhibits glucagon release by SGLT2-induced stimulation of somatostatin secretion

Abstract

Hypoglycaemia (low plasma glucose) is a serious and potentially fatal complication of insulin-treated diabetes. In healthy individuals, hypoglycaemia triggers glucagon secretion, which restores normal plasma glucose levels by stimulation of hepatic glucose production. This counterregulatory mechanism is impaired in diabetes. Here we show in mice that therapeutic concentrations of insulin inhibit glucagon secretion by an indirect (paracrine) mechanism mediated by stimulation of intra-islet somatostatin release. Insulin's capacity to inhibit glucagon secretion is lost following genetic ablation of insulin receptors in the somatostatin-secreting δ-cells, when insulin-induced somatostatin secretion is suppressed by dapagliflozin (an inhibitor of sodium-glucose co-tranporter-2; SGLT2) or when the action of secreted somatostatin is prevented by somatostatin receptor (SSTR) antagonists. Administration of these compounds in vivo antagonises insulin's hypoglycaemic effect. We extend these data to isolated human islets. We propose that SSTR or SGLT2 antagonists should be considered as adjuncts to insulin in diabetes therapy.

Conflict of interest statement

The authors declare no competing interests.

Figures

Fig. 1

Insulin stimulates somatostatin secretion. a Somatostatin release at 1, 4 and 10 mM glucose in the absence or presence of 100 nM insulin (n = 8–10 experiments using six male mice). †p < 0.05, †††p < 0.001 vs the same glucose concentration but in the absence of insulin, one-way ANOVA followed by Dunnet’s post hoc test. Glucose (4 and 20 mM) also stimulates somatostatin secretion compared with 1 mM in both the absence and presence of insulin (p < 0.05 or better, not indicated for clarity). b, c Lack of effect of insulin on somatostatin secretion elicited by high [K+]o (70 mM at 4 mM glucose; b) and by the KATP channel blocker tolbutamide (0.2 mM at 1 mM glucose; c). When [K+]o was increased, Na+ was correspondingly reduced to maintain iso-osmolarity. **p < 0.01 vs 1 mM glucose alone (n = 10 experiments/5 male mice), one-way ANOVA followed by Dunnet’s post hoc test. d Effects of insulin or IGF-1 on somatostatin secretion at 4 mM glucose in the absence and presence of the IGF-1R inhibitor PQ401 (n = 10 experiments using six male mice). ***p < 0.001 vs control (no insulin); †††p < 0.001 vs control in the presence of PQ401, one-way ANOVA followed by Dunnet’s post hoc test. e Effects of insulin and the insulin receptor antagonist S961 on somatostatin release in isolated mouse islets incubated at 4 mM glucose (n = 9 experiments/3 male mice). ***p < 0.001 vs no insulin. †††p < 0.001 vs 100 nM insulin in the absence of S961, one-way ANOVA followed by Dunnet’s post hoc test. Responses have been normalised to somatostatin secretion at 1 mM or 4 mM glucose as indicated. Data are presented as dot plots of individual experiments and/or mean values ± S.E.M. of all experiments with the experimental series

Fig. 2

Insulin exerts its glucagonostatic effect by somatostatin. a Effects of increasing concentrations of insulin (3 pM, 30 pM, 300 pM, 3 nM, 30 nM, 300 nM; logarithmic abscissa) on glucagon (red) and somatostatin (black) secretion in the presence of 4 mM glucose (n = 10 experiments/6 male mice). Arrow indicates IC50 for inhibitory effect of insulin on glucagon secretion (~100 pM). *p < 0.05, **p < 0.01 and ***p < 0.001 and †††p < 0.001 for the effects of insulin on somatostatin and glucagon secretion vs 4 mM glucose alone, respectively, one-way ANOVA followed by Dunnet’s post hoc test. b Somatostatin secretion at 4 mM glucose in the absence and presence of insulin and/or CYN154806 (n = 10 experiments//6 male mice). ***p < 0.001 vs 4 mM glucose; ††p < 0.005 vs 4 mM glucose and CYN154806, one-way ANOVA followed by Dunnet’s post hoc test. c As in b but glucagon is measured (n = 10 experiments/6 male mice). Secretion has been normalised to secretion rates at 4 mM glucose. *p < 0.05 and **p < 0.01 vs 4 mM glucose; ††p < 0.005 vs 4 mM glucose and insulin. Data are presented as dot plots of individual experiments and/or mean values ± S.E.M. of all experiments with the experimental series

Fig. 3

Insulin’s glucagonostatic effect mediated by InsR in δ cells. a, b Insulin tolerance test. Plasma glucose concentrations following injection of insulin (at t = 0; 1.5 U/kg) in female (a) and male (b) controls (black; n = 10 female and 6 male mice) and δ-cell-specific insulin receptor knockout (SIRKO, red; n = 9 female and 6 male mice) mice. *p < 0.05 vs control, two-way ANOVA followed by Sidak’s post hoc test. c, d Somatostatin (c) and glucagon secretion (d) at 4 mM glucose in the absence or presence of insulin (100 nM) in islets from control (black; n = 20/4 mice) or SIRKO (red: n = 18 with insulin and 20 without insulin/4 mice) mice. Responses have been normalised to secretion at 4 mM glucose. **p < 0.01, ***p < 0.001 vs 4 mM glucose; †p < 0.05 and †††p < 0.001 vs 4 mM glucose and 100 nM insulin in control islets, two-way ANOVA followed by Sidak’s post hoc test. Data are presented as dot plots of individual experiments and/or mean values ± S.E.M. of all experiments with the experimental series

Fig. 4

Effect of insulin on cytosolic Ca2+. a Effects of insulin on [Ca2+]i in δ cells in intact islets exposed to 4 mM glucose. Two representative traces are shown. b, c Frequency (b) and amplitude (c) of [Ca2+]i oscillations at 4 mM glucose in the absence and presence of insulin. *p < 0.05 vs 4 mM glucose (n = 54 δ cells in 11 islets from 8 mice of both sexes), Student’s t-test. d Effects of insulin (100 nM) on somatostatin secretion at 4 mM glucose in the absence and presence of the SERCA inhibitor thapsigargin (25 μM). ***p < 0.001 vs 4 mM glucose; ††p < 0.01 vs 4 mM glucose and insulin in the absence of thapsigargin (n = 10 experiments/4 male mice), one-way ANOVA followed by Sidak’s post hoc test. Data in b–d are presented as dot plots of individual experiments and/or mean values ± S.E.M. of all experiments with the experimental series

Fig. 5

Insulin activates SGLT2 in δ cells. a, b Somatostatin (a) and glucagon secretion (b) at 4 mM glucose in the absence or presence of insulin at normal or lowered (10 mM) extracellular Na+ ([Na+]o) with or without dapagliflozin as indicated. Responses have been normalised to secretion at 4 mM glucose. * p < 0.05, ***p < 0.001 vs 4 mM glucose alone; †p < 0.05, ††p < 0.05 †††p < 0.001 for comparisons with 4 mM glucose under control conditions (black bar). Data are based on 6–8 experiments using islets from three male mice). Statistical analyses were performed by one-way ANOVA followed by Dunnett’s post hoc test. c, d Insulin tolerance test. Plasma glucose (c) and glucagon (d) concentrations following intraperitoneal (ip) injection of insulin (at t = 0; 0.5 U/kg) in male mice (age: 9 wks) followed by injection of dapagliflozin (10 mg/kg, ip: red) or vehicle (5% DMSO in sterile PBS, ip; black). (see Supplementary Fig. 5 for corresponding data in female mice). In c, d, insets summarise effects of dapagliflozin on plasma glucose and glucagon expressed as the area under the curve (AUC) during the first 45 min. *p < 0.05 vs control (no dapagliflozin) by Student’s t-test (4 mice for each condition). e Effects of the SGLT2 inhibitor dapagliflozin (12.5 μM) on δ-cell action potential firing in the presence of 4 mM glucose and 100 nM insulin. Dapagliflozin was applied as indicated above the trace (representative of n = 4 δ cells in four islets from three mice of both sexes). f Effects of insulin in the presence of 19 mM of the non-metabolisable glucose analogue α-methyl-d-glucopyranoside (αMDG) on δ-cell action potential firing (n = 5 cells from three mice of both sexes). Red dotted line indicates membrane potential before addition of insulin. Data in a–d are presented as dot plots of individual experiments and/or mean values ± S.E.M. of all experiments with the experimental series

Fig. 6

Effects of insulin in the perfused mouse pancreas. a Glucagon secretion measured before and after addition of insulin using the perfused mouse pancreas preparation. For these experiments, wild-type mice were used. In each experiment, glucagon secretion was normalised to that at 4 mM glucose prior to the addition of insulin (100 nM in three experiments and 10 μM in two experiments: responses have been pooled for display; female mice were used). b As in a but the insulin receptor antagonist S961 was used instead of insulin (n = 7 female mice). c As in a but experiments were performed in hyperglycaemic/diabetic Fh1βKO mice and using insulin at a concentration of 100 nM (n = 8 mice of both sexes). In a-c, the black bars represent mean glucagon secretion measured during the 6 min preceding the addition of insulin or S961. d As in (c) but CYN154806 was included as indicated (n = 4 diabetic Fh1βKO mice of both sexes). Glucagon secretion has been normalised to that in the presence of CYN154806 (measured at steady-state, 4–8 min after the addition; t = 4–8 min). The red and bars represent glucagon secretion in the presence of insulin alone and following the addition of CYN154806, respectively. Statistical significances were analysed by Student’s t-test. *p < 0.05. Data are presented as dot plots of individual experiments and/or mean values ± S.E.M. of all experiments with the experimental series

Fig. 7

Effects of insulin in human pancreatic islets. a, b Somatostatin (a) and glucagon secretion (b) at 4 mM glucose in the absence and presence of insulin and S961 as indicated using islets from four donors (three males/one female) with four separate experiments for each donors. c, d As in a, b but testing the effects of CYN154806 as indicated (three female donors, three separate experiments for each donor). Note that CYN154806 does not affect insulin-induced somatostatin secretion but reverses the inhibitory effect of insulin on glucagon secretion. e, f As in a, b but instead testing the effect of dapagliflozin (100 nM; two female and one male donors, two separate experiments for each donor (different colours)). Note that dapagliflozin reduces the stimulatory and inhibitory effects of insulin on somatostatin and glucagon secretion, respectively. In a–f, data have been normalised to 4 mM glucose alone. *p < 0.01 vs 4 mM glucose; †p < 0.01 vs 4 mM glucose and insulin, one-way ANOVA followed by Sidak’s post hoc test. In all panels, data are presented as dot plots of individual experiments and/or mean values ± S.E.M. of all experiments with the experimental series

Fig. 8

Schematic summarising the effects of insulin in δ cells. Relationship between SGLT2-mediated glucose uptake and δ-cell electrical activity, Ca2+ entry, Ca2+-induced Ca2+ release (CICR) and somatostatin release. Insulin stimulates SGLT2 activity. The operation of SGLT2 is electrogenic Na+-dependent glucose uptake mediated by SGLT2 gives rise to a small depolarising current and may (in some cells with sufficiently high SGLT2 expression and/or low KATP channel activity) trigger electrical activity (action potential firing). The associated increase in cytoplasmic Ca2+ ([Ca2+]i) leads to intracellular Ca2+ mobilisation by activation of ryanodine receptors (RyRs) and Ca2+-induced Ca2+ release (CICR). In addition to SGLT2, the δ cells also express the glucose transporters GLUT1 and GLUT3. Glucose uptake via these transporters accounts for 99% of glucose uptake. Metabolism of glucose leads to KATP channel closure, membrane depolarisation, electrical activity and CICR