Steviol glycosides enhance pancreatic beta-cell function and taste sensation by potentiation of TRPM5 channel activity

Abstract

Steviol glycosides (SGs), such as stevioside and rebaudioside A, are natural, non-caloric sweet-tasting organic molecules, present in extracts of the scrub plant Stevia rebaudiana, which are widely used as sweeteners in consumer foods and beverages. TRPM5 is a Ca²⁺-activated cation channel expressed in type II taste receptor cells and pancreatic β-cells. Here we show that stevioside, rebaudioside A and their aglycon steviol potentiate the activity of TRPM5. We find that SGs potentiate perception of bitter, sweet and umami taste, and enhance glucose-induced insulin secretion in a Trpm5-dependent manner. Daily consumption of stevioside prevents development of high-fat-diet-induced diabetic hyperglycaemia in wild-type mice, but not in Trpm5^-/- mice. These results elucidate a molecular mechanism of action of SGs and identify TRPM5 as a potential target to prevent and treat type 2 diabetes.

Figure 1. TRPM5-mediated currents are potentiated by stevioside and rebaudioside A and their aglycon steviol.

(a) Structure of stevioside. (b) Time course of inward and outward currents from TRPM5 overexpressing HEK cells. To activate TRPM5, pipette solution contained 1 μM free Ca²⁺. Time 0 indicates break-in into the cell. Stevioside was applied as indicated. (c) Representative current traces from the time points indicated in b. (d) Left: the average±s.e.m. (n=13 cells, paired t-test, P<0.01) peak inward and outward current before and during the application of stevioside with 1 μM intracellular Ca²⁺. Right: the average±s.e.m. inward and outward current before and during the application of stevioside in the absence of free intracellular Ca²⁺ (n=10 cells, P>0.2, paired t-test). (e) Structure of rebaudioside A. (f) A time trace with inward and outward currents from Trpm5 overexpressing HEK cells. Rebaudioside A was applied as indicated. (g) Current traces upon the application of rebaudioside A as indicated in f. (h) Average±s.e.m. (n=14 cells, paired t-test P<0.05) inward and outward current before and during application of rebaudioside A. (i) Structure of steviol. (j) A time trace with inward and outward currents from Trpm5 overexpressing HEK cells. Steviol was applied as indicated. (k) Current traces upon the application of steviol as indicated in j. (l) Average±s.e.m. (n=12 cells, paired t-test P<0.001) inward and outward current before and during application of steviol. See also Supplementary Figs 1 and 2. Reb. A, rebaudioside A; Stev., stevioside; NS, not significant.

Figure 2. Patch-clamp recordings in isolated β-cells.

(a) The current recorded in primary WT β-cells evoked by 20 mM glucose or 20 mM glucose+10 μM stevioside (average±s.e.m.; n=5 cells). (b) Difference current of the traces in a, highlighting the stevioside contributed fraction of the total current. (c) Left: action potentials (APs) evoked in WT β-cells by 20 mM glucose (top) or 20 mM glucose+10 μM stevioside (bottom). Right: probability density function plot depicting change in AP firing frequency properties. AP interval (s) represents the duration of the interval between APs. Density represents the proportion of intervals between APs that lie within the bin duration, normalized to account for differing bin width. (d) As c but with 100 μM tolbutamide instead of 20 mM glucose. (e) APs evoked in Trpm5^−/− β-cells in the presence of 100 μM tolbutamide or 100 μM tolbutamide+10 μM stevioside, as in c (n=4 cells). NS, not significant.

Figure 3. Stevioside and steviol potentiate calcium oscillations in pancreatic islets.

(a) Representative time course of 10 mM glucose-induced Ca²⁺ oscillations in pancreatic islets isolated from WT mice, with the application of stevioside as indicated. (b) Same as in a, but from Trpm5^−/− islets. (c) Average±s.e.m. oscillation frequency of WT (P=1.6 × 10⁻⁶, n=71 islets from four mice, paired t-test) and Trpm5^−/− (P=0.15, n=68 islets from five mice) islets in the presence of 10 mM glucose alone or 10 mM glucose supplemented with 10 μM stevioside. Note the difference in the effect of stevioside on oscillation frequency between WT and Trpm5^−/− islets (P=2.63 × 10⁻⁴, two-sample t-test). (d) Wide-field image of WT pancreatic islets before the fluorescence measurement. (e) Image of Trpm5^−/− islets before the experiment. (f) Dose–response relation of different concentrations of stevioside on the potentiation of calcium oscillation frequency. The average±s.e.m. oscillations per minute during the application of 0, 1 nM to 100 μM stevioside in 10 mM glucose (269 islets from nine mice). The black line represents the logistic fit of the data, and the effector concentration for half-maximum response (EC₅₀)=690 nM. (g) Islet size of WT and Trpm5^−/− islets (n=90 WT and 102 Trpm5^−/− islets). (h) Time course of calcium oscillations in WT islets with perfusion of steviol. (i) Calcium oscillations in islets from Trpm5^−/− mice upon perfusion with steviol. (j) Average±s.e.m. oscillation frequency of WT and Trpm5^−/− islets (WT: P=5.5 × 10⁻¹⁵, paired t-test, n=150 islets from four mice; Trpm5^−/−: P=0.31, paired t-test, n=48 islets from two mice) in the presence of 10 mM glucose or 10 mM glucose supplemented with 10 μM steviol. (k) Steviol concentration in the plasma of mice exposed to 124 μM steviol in their drinking water for 3 weeks, as determined by high-performance liquid chromatography. (l) Example trace of a WT islet exposed to G10 and a physiological concentration of 400 nM steviol. (m) Calcium oscillation frequency in WT islets is significantly potentiated with 400 nM steviol (paired sample t-test). See also Supplementary Figs 3 and 4. NS, not significant.

Figure 4. Stevioside potentiates glucose-induced insulin secretion in vitro and in vivo.

(a) Insulin release measured in vitro on individual isolated islets of WT and (b) Trpm5^−/− mice (average±s.e.m. from n=24–30 islets per condition from four mice per genotype, two-sample t-test). The horizontal dashed line represents the level of the (10 mM) GIIS. (c) The effect of stevioside administration on glucose-induced insulin release in plasma from eight WT and nine Trpm5^−/− mice (average±s.e.m). (d) Example trace of insulin secretion during perifusion experiments without steviol (black) or with steviol supplementation at 10 min in 3 mM glucose (G3), 10 mM glucose (G10) and with 100 μM diazoxide (Dz) and 30 mM KCl (K30). (e) Total insulin secretion in different conditions relative to the K30 stimulus. First GIIS is the initial insulin peak between 20 and 30 min in c and the second phase between 30 and 50 min. Significant difference between control (n=6 experiments) and steviol conditions (n=6) of 866 islets from 22 WT mice with a two-way analysis of variance. See also Supplementary Fig. 5 and Supplementary Table 1. NS, not significant.

Figure 5. Two-bottle taste preference test.

Preferences indicated are measured against water for sweet (1% sucrose), umami (150 mM monopotassium glutamate—MKG), bitter (100 μM quinine—25 μM quinine for Tas1r2,Tas1r3^(−/−)2 in a) or sour (10 mM citric acid) with or without 124 μM steviol or SG supplemented to the drinking solution. (a) The bar graphs show either the preference (up bars) or avoidance (down bars) for the indicated compound in a two-bottle preference test during 48 h. There is clear preference for stevioside (P=2 × 10⁻²⁶, n=48 mice over four different experiments) and rebaudioside A (P=2 × 10⁻²¹, n=24) in WT animals, and indifference to steviol (P=0.24, n=24, one sample t-test versus 0.5). Steviol increases the perception of sweet (P=0.037), umami (P=0.001) and bitter (P=0.003, n=24, paired t-test) solution in WT animals, but not in Trpm5^−/− mice. In Tas1r2,Tas1r3^(−/−)2 mice bitter taste is potentiated by steviol (P=0.018) and stevioside (P=0.048, n=24, paired t-test). The indicated significances show a difference between the taste compound and the taste compound together with steviol. (b) Initial taste preference presented as normalized cumulative lick count for water versus water without (left) or with steviol (right) over 5 min after 23 h of water deprivation of 14 and 7 WT mice, respectively. (c) As b with sucrose, (d) MKG and (e) quinine. (b–e) Details from Supplementary Fig. 7a,b,e–j. (f) Preference from b–e represented as average±s.e.m. of relative lick count between mice for the bottle containing the tastant without or with steviol. Two-sample t-test (steviol/MKG) or paired sample t-test (sucrose/quinine). (g) Individual preferences from the data in f, indicating the variation between mice. The boxplot gives the 25–75% data interval with the mean indicated as a square and the median as a line. See also Supplementary Figs 6–9. Citr. A, citric acid; NS, not significant; Quin., quinine; Stev., stevioside; Sucr., sucrose.

Figure 6. Acute treatment of stevioside lowers the glycaemia of WT mice during a GTT in a β-cell-dependent way.

(a–f) Results of crossover tandem GTTs performed on the same mice with a 2 week interval. Mice received either 0.5 mg g⁻¹ stevioside applied per os 2 h before the GTT, or vehicle. Individual points represent the average±s.e.m. and the significances are the result of a paired sample t-test of the values at the same time point; non-significant differences are not indicated. (a) WT mice on a normal diet, (b) Trpm5^−/− mice on a normal diet, (c) WT mice 20 weeks on a HFD, (d) Trpm5^−/− mice 20 weeks on a HFD, (e) WT mice with WT islets transplanted under the kidney capsule and (f) WT mice with Trpm5^−/− islets transplanted under the kidney capsule. (g) Average±s.e.m. Area under the curve (AUC) from the experiments in a–f. Paired sample t-test within groups, two-sample t-test between groups. See also Supplementary Figs 10 and 11. NS, not significant.

Figure 7. Prevention of glucose intolerance with stevioside treatment during HFD.

(a) Results of i.p. GTT performed on eight WT mice at different time points after the start of HFD in control conditions. The indicated significance is the result of a paired sample t-test between week 0 and 20. Values are average±s.e.m. (b) Results from GTTs in eight WT mice, on HFD treated with a daily dose of 25 mg kg⁻¹ stevioside. The indicated significance is the result of a paired sample t-test between week 0 and 20. Values are average±s.e.m. (c) Area under the curve (AUC) from 0 to 120 min during the GTT on the WT mice from a,b. The solid and dotted lines represent the logistic fit for the respective data sets. The represented values are averages±s.e.m. from eight animals in every group, in two independent experiments, the significances are the result of a two-sample t-test between the untreated and stevioside-treated groups at the same week during the HFD. (d–f) As in a–c but from Trpm5^−/− mice. (g) Glycaemia (average±s.e.m.) during hyperinsulinemic euglycaemic clamp from WT and Trpm5^−/− animals. Infusion of steviol did not change the glycaemia during euglycaemic glucose infusion. (h) Glucose infusion rate during hyperinsulinemic euglycaemic clamp. No significant differences were detected between WT and Trpm5^−/− animals. (i) Glycaemia of WT mice on a HFD with or without stevioside (124 μM in drinking water) during a euglycaemic hyperinsulinemic clamp. Basal glycaemia is lower for the mice receiving stevioside, two-sample t-test between treatment groups on the average of −20 and −10 min glycaemic values. (j) Glucose infusion rates during euglycaemic hyperinsulinemic clamp experiments. Glucose infusion rate and hence insulin resistance was not different due to stevioside treatment during HFD. Acute i.v. infusion of 0.8 μmol kg⁻¹ steviol did not change the glycaemia during perfusion with the euglycaemic glucose infusion rate. See also Supplementary Figs 12 and 13 and Supplementary Table 2. NS, not significant.

Figure 8. Stevioside effects are rapidly reversible during HFD.

(a) Time line illustrating the course of the diet and stevioside treatment of three groups of mice. Time points of GTT are as indicated. (b) The results of the GTT after 10 weeks on HFD (black—17 weeks old) and after another 5 weeks (green—22 weeks old) of mice on HFD+stevioside (100 mg l⁻¹ in drinking water). (c) GTTs of mice on the same time points as (b) with stevioside treatment terminated at the age of 17 weeks. Individual points indicate the average±s.e.m. and the significances are the result of a paired sample t-test at the same time point from the two GTTs. (d) GTTs of mice in the untreated control group at the same time points. All values are averages±s.e.m. of n=8–10 mice. (e) Average±s.e.m. area under the curve (AUC) from the experiments in b–d, paired sample t-test, only significant differences are indicated. (f) The weekly average±s.e.m. resting glycaemia (spheres) and the average over 5 weeks (lines) of the mice in group 2 during (black) and after (green) the stevioside treatment. Two-sample t-test between the average glycaemia during and after stevioside treatment (lines). (g) Average±s.e.m. Body mass of the three groups at the age of 17 weeks (black) and 22 weeks (green) before the fasting for the GTT (two-sample t-test). See also Supplementary Fig. 14 and Supplementary Table 3. NS, not significant.

Figure 9. Schematic overview of intracellular pathways in type II taste receptor cells and pancreatic β-cells.

Important factors in the regulation of taste perception and insulin secretion, relevant for this work, are highlighted in this figure. The effects of stevioside through TRPM5 are highlighted with red arrows. In type II taste receptor cells, positive modulation of TRPM5 increases ATP release and afferent signalling from the receptor cell. In the pancreatic β-cell, upon glucose application β-cells display parallel V_m and [Ca²⁺]_cyt oscillations, which drive insulin secretion. TRPM5 is active during the lag phase in between bursts of action potentials, and determines the frequency of V_m and Ca²⁺ oscillations, which modulates insulin secretion: enhanced TRPM5 activity results in a higher oscillation frequency, which results in more insulin secretion. Note that steviol and its derivatives (for example, stevioside) only enhance TRPM5 activity and insulin secretion, but that glucose is the trigger for cell signalling. Instead, sulfonylureas such as glibenclamide (not depicted) block K_ATP channels directly and trigger Ca²⁺ signalling and insulin secretion independent of glucose transport.