The aglycone of ginsenoside Rg3 enables glucagon-like peptide-1 secretion in enteroendocrine cells and alleviates hyperglycemia in type 2 diabetic mice

Ginsenosides can be classified on the basis of the skeleton of their aglycones. Here, we hypothesized that the sugar moieties attached to the dammarane backbone enable binding of the ginsenosides to the sweet taste receptor, eliciting glucagon-like peptide-1 (GLP-1) secretion in the enteroendocrine L cells. Using the human enteroendocrine NCI-H716 cells, we demonstrated that 15 ginsenosides stimulate GLP-1 secretion according to the position of their sugar moieties. Through a pharmacological approach and RNA interference technique to inhibit the cellular signal cascade and using the Gαgust−/− mice, we elucidated that GLP-1 secreting effect of Rg3 mediated by the sweet taste receptor mediated the signaling pathway. Rg3, a ginsenoside metabolite that transformed the structure through a steaming process, showed the strongest GLP-1 secreting effects in NCI-H716 cells and also showed an anti-hyperglycemic effect on a type 2 diabetic mouse model through increased plasma GLP-1 and plasma insulin levels during an oral glucose tolerance test. Our study reveals a novel mechanism where the sugar moieties of ginsenosides Rg3 stimulates GLP-1 secretion in enteroendocrine L cells through a sweet taste receptor-mediated signal transduction pathway and thus has an anti-hyperglycemic effect on the type 2 diabetic mouse model.

The NCI-H716 cell line is derived from ascetic fluid of a 33 year old Caucasian male patient with a poorly differentiated cecal adenocalcinoma 12 . Culturing the cells with a specific extracellular matrix causes endocrine differentiation, leading the cells to express several neuroendocrine markers including chromogranin A, and on this basis they are a qualified enteroendocrine cellular model for studying the regulation of GLP-1 secretion 12 . We have reported that enteroendocrine L cells express taste receptors and their downstream signal elements, including a specific G protein, Gα -gustducin (Gα gust), and G protein-coupled sweet and bitter taste receptors, similar to their expression in the tongue 13,14 . T1R3 taste receptor, which is expressed by about 10-20% of taste cells, consists of a heterodimer T1R2 that recognizes a broad spectrum of sweet taste stimuli, including natural and synthetic sugar 15,16 . Previous reports suggest that the intracellular signal transduction pathway activated by sugar binding to taste receptors is mediated by the activation of Gα gust and a consequent signaling cascade including phospholipase Cβ 2 (PLCβ 2) and inositol 1,4,5-triphosphate (IP 3 ) 16,17 . This Gβ γ -subunit mediating the signaling cascade elicits the release of Ca 2+ from intracellular stores and subsequent Ca 2+ -dependent activation of a transient receptor potential channel M5 (TRPM5), leading to the membrane depolarization and further action potential generation in turn 18,19 .
In this study, we demonstrated the GLP-1 secreting effect of ginsenosides using the enteroendocrine NCI-H716 cell line. Rg3, a PPD group ginsenoside that is abundant in steamed ginseng, showed the strongest GLP-1 secreting effect. Using the cell line and Gα gust −/− mice, we investigated the cellular mechanism underlying the GLP-1 secreting effect of Rg3, and using db/db mice, we evaluated the possibility of exploiting the effect of Rg3 as a therapeutic agent for type 2 diabetes mellitus.
Ginsenoside Rg3 showed the strongest GLP-1 secreting effect in the NCI-H716 cells (Fig. 1). In this study, we performed further in vitro and in vivo studies using Rg3. We observed a dose-dependent GLP-1 secreting effect of Rg3 treatment in NCI-H716 cells (Fig. 2a). Rg3 did not affect the cell's viability below a concentration of 25 μ M (Fig. 2b).
To examine whether the GLP-1 secreting effect of Rg3 is mediated by sweet taste receptor activation, we transfected siRNAs targeting the T1R2 and/or T1R3 to the NCI-H716 cells and measured the GLP-1 levels stimulated by Rg3 treatment (Fig. 2c). We also measured glucose stimulated GLP-1 secretion to confirm our RNA interference sets (Fig. 2d). Rg3 (10 μ M) showed 2-fold GLP-1 secreting effect compare to the glucose (10%, w/v). One of the T1R2 or T1R3 siRNA transfection partly decreased the GLP-1 secreting effects of Rg3 and glucose, respectively (Fig. 2c,d). Glucose-stimulated GLP-1 secretion was completely blocked by the T1R2 and T1R3 double siRNA transfection while Rg3-stimulated GLP-1 secretion slightly remained. The siRNAs accurately knocked-down their targeting mRNA expressions (Fig. 2e). A human sweet taste receptor antagonist lactisole, which have reported to block T1R3, was pre-treated NCI-H716 cells and significantly inhibited Rg3 stimulated GLP-1 secretion (Fig. 2f)

Discussion
Panax ginseng is considered one of the most valuable medicinal plants in Asia and is largely consumed throughout the world. Ginsenosides, the saponins found in all parts of the ginseng plant, have been investigated for their various pharmacological effects on hyperglycemia, weight gain, neuroprotection, tumor cell growth, and hypertension. Regarding the diverse pharmacological aspects of ginsenosides, Attle et al. explained that the ginsenosides share structural similarities with steroid hormones, especially progesterone and pregnanolone, and thereby have numerous physiological activities 23 . Structural diversity including type-, number-, and site of attachment of sugar moieties also contributes to the diverse pharmacological effects of ginsenosides 23 . This structural diversity can be amplified through a steaming process and intestinal microbiota metabolism.
We focused on the chemical structure of dammarane family ginsenosides having various sugar moieties with their carbon-ring backbone and hypothesized that these sugar moieties act like ligands for the sweet taste receptor.
We demonstrated the GLP-1 secreting effect of 15-dammarane family ginsenosides in human enteroendocrine NCI-H716 cells. PPDs that have sugar moieties attached at C-3 and C-20 showed a GLP-1 secreting effect on the NCI-H716 cells. On the other hand, PPTs that have sugar moieties attached at C-6 did not show the GLP-1 secreting effect on the NCI-H716 cells while the ginsenoside Re and Rg1, which has a glucose residue at C-20, showed a moderate GLP-1 secreting effect. From the obtained results, we assume that the sugar moieties attached at the C-3 and/or C-20 contribute to the binding affinity of the dammarane family ginsenosides to the sweet taste receptor.
Interestingly, ginsenoside metabolites, the chemical structure of which was transformed through a steaming process or intestinal microbiota metabolism, showed the strongest GLP-1 secreting effect. Ginsenoside Rg3 showed the strongest GLP-1 secreting effect and we traced the intracellular mechanism using enteroendocrine NCI-H716 cells.
Structurally, Rg3 is a PPD with two D-glycopyranosyl moieties. Thus we have assumed that the sugar moieties of ginsenoside Rg3 provide a binding motif to the sweet taste receptor expressed on the enteroendocrine L cells. GLP-1 secreting effect of Rg3 was significantly decreased in each T1R2 or T1R3 siRNA transfected NCI-H716 cells, respectively. But the effect was not removed completely even both the siRNAs transfection. A considerable GLP-1 secretion in the both siRNAs transfected cells response to Rg3 stimuli suggesting the existence of multiple receptors for Rg3.   A recent study reported that a ginsenoside metabolite C-K stimulates GLP-1 secretion in NCI-H716 cells via binding of a bile acid receptor 5 . We assumed that the dammarane backbone of Rg3 may contribute a binding motif to the other receptors, such as a bile acid receptor or a cannabinoid receptor, and transfected the corresponding siRNAs to abolish the mRNA expression. The GLP-1 secreting effect of Rg3 was not mediated by the cannabinoidor bile acid receptors. Nevertheless, the GLP-1 secreting effect of Rg3 appears to be dependent on the Gα gust. Therefore, we assume participation of one or more bitter taste receptor activation in the GLP-1 secreting effect of Rg3 alongside sweet taste receptor.
We traced a common sweet taste modulatory cellular pathway, which is mediated by the Gβ 3 γ 13 -PLCβ 2-IP3 signal cascade. Through pharmacological approaches using corresponding inhibitors or antagonist, we determined that the GLP-1 secreting effect of Rg3 is mediated by the signal cascade, but also found that a considerable GLP-1 response remained.
Gα gust is expected to cause activation of PDE and thus decrease intracellular cAMP levels 24 . Indeed, we observed that the bitter tastant DB decreased intracellular cAMP levels during its GLP-1 secreting event responding to Gα gust activation 14 .
However, similar to studies showing that sugars increase intracellular cAMP levels, we also have found that Rg3 increases intracellular cAMP levels. We further showed the involvement of enzyme AC, which produces cAMP in response to the stimuli, in the GLP-1 secreting effect of Rg3 instead of PDE. Moreover, phosphorylations of PKA and CREB, which are activated in response to the increased intracellular cAMP level, are involved in the GLP-1 secreting effect of Rg3. PKA is also involved in the GLP-1 secreting effect of Rg3. ERK1/2, a mitogen-activated protein kinase (MAPK), is known to activate various transcription factors including CREB, in response to diverse extracellular stimuli such as forskolin 25 .
One of the interesting results in our study is that acesulfame K, an artificial sugar, did not affect the GLP-1 secreting effect of NCI-H716 cells. Since the expression of sweet taste receptors have been found in the enteroendocrine L cells along with its signal transduction elements artificial sugars had been convinced that they are able to activate the sweet taste receptors in the L cells as they do in the lingual tissues 13 . In contrast to the in vitro studies that show GLP-1 secreting effect of artificial sugars in the human and mouse enteroendocrine cells, in vivo studies using healthy human subjects failed to show the effects of artificial sugars on the GLP-1 secretion [26][27][28] .
However, a human study demonstrated that the sweet taste receptor inhibitor drastically blocked the GLP-1 and PYY secreting effect of glucose 29 . Perhaps, the GLP-1 and PYY secreting effect via activation of sweet taste receptor in the enteroendocrine L cells depends on the structural analogy to glucose than the sweetness itself. Our in vivo study elucidated the effects of ginsenoside Rg3 administration on hyperglycemia in type 2 diabetic mice. The therapeutic effects of Rg3 against metabolic disorder, such as obesity and hyperglycemia have been demonstrated. Park et al. reported enhanced glucose-stimulated insulin secretion via AMP-activated protein kinase (AMPK) activation upon Rg3 treatment in a hamster pancreatic β cell line 7 . The effects of Rg3 on AMPK activation suppressed adipocyte differentiation in mouse 3T3-L1 adipocyte and improved glucose uptake in the rat L6 myocyte 30,31 . Therefore, Rg3 has direct glucose lowering effect on hyperglycemia through enhanced glucose Data are mean ± s.e.m.; n = 6. Statistics, Mann-Whitney U test. * P < 0.05; *** P < 0.001 vs saline-treated group.
(c-f) The effect of Rg3 treatment on the plasma GLP-1 and plasma insulin levels in the C57 mice (c,e) after glucose gavage was abolished in the Gα gust −/− mice (d,f). Data are mean ± s.e.m.; n = 6. Statistics, Mann-Whitney U test. ** P < 0.01; *** P < 0.001 vs saline-treated group. uptake activity in the myoblast and also has an indirect effect through stimulation of insulin secretion in the pancreatic β cell.
In this study, Rg3 stimulated GLP-1 secretion by activating sweet taste receptor-and Gα gust-mediated signal transduction cascade in enteroendocrine cells and increased plasma GLP-1 and plasma insulin levels in db/db mice after glucose gavage. Using Gα gust −/− mice, we showed the stimulatory effect of Rg3 on plasma GLP-1 and that plasma insulin is Gα gust-dependent. The slightly lowered blood glucose level after Rg3 administration observed in the Gα gust −/− mice appears to demonstrate a direct effect of Rg3 on the blood glucose regulation 7 .
Our study elucidates a novel mechanism underlying the anti-diabetic effects of ginsenoside Rg3. The sugar moieties of Rg3 and other dammarane family ginsenosides enables their binding to the sweet taste receptor to stimulate GLP-1 secretion in intestinal L cells. The GLP-1 secreting effect lowers the blood glucose levels through insulinotropic action. Considering the diverse benefits of GLP-1 on hyperglycemia, food intake, and even β cell function, Rg3 has possibility to be developed as a novel therapeutic agent for type 2 diabetes and obesity. GLP-1 ELISA. The endocrine differentiated cell media was replaced with PBS containing 1 mM calcium chloride and different drug concentrations. The GPCR pathway inhibitors were pre-incubated for 30 min before the drug treatment. After incubation for 1 h in a CO 2 incubator, GLP-1 concentration was measured as previously described 14 . An active GLP-1 ELISA (EMD Millipore, Billerica, MA, USA) was performed as described in the manufacturer's instructions. The active GLP-1 concentrations in each sample were measured using a Fluoroskan Ascent FL machine (Thermo Fishwer Scientific, Vantaa, Finland). The lowest level of active GLP-1 that can be detected by the GLP-1 assay is 2 pM.  Table S1). A scrambled negative control siRNA was purchased from Bioneer. Endocrine differentiated NCI-H716 cells were transfected with the siRNA duplexes using Lipofectamine RNAiMAX reagent (Invitrogen).

Cell viability assay.
Real-time quantitative PCR. The expression of T1R2, T1R3, GNAT3, GPR119, and GPBAR1 after siRNA transfection was determined using a StepOne real-time PCR instrument (Applied Biosystems, Foster City, CA, USA). Total RNA isolation and subsequent cDNA hybridization were performed as previously described 32 . The expression levels of T1R2, T1R3, GNAT3, GPR119, and GPBAR1 in each type of siRNA-transfected cell were compared with the corresponding levels in the negative control siRNA-transfected cells, and the the 2 −ΔΔCt values were determined 33 . GAPDH was used as an endogenous control. The sense and antisense sequence of each primer were provided online (Supplementary Table S2).
Calcium imaging. NCI-H716 cells were seeded on a clear-bottom 96-well black plate (Corning, Tewksbury, MA, USA). After differentiation, the medium was replaced with PBS and the mixture was incubated for 30 min with fura-2 AM dye as described previously 14,34 . After 30 min, the medium was replaced with saline with or without 2.5 mM lactisole and incubated for a further 30 min. [Ca 2+ ] i were observed with a Nikon Eclipse TS 100 fluorescence imaging system (Nikon, Melville, NY, USA), and quantified and visualized with InCyt Im2 software (University of Cincinnati, Cincinnati, OH, USA). The number of cells observed was 10-20 per well. cAMP ELISA. Endocrine differentiated NCI-H716 cells were incubated with Rg3 or forskolin. The drug-treated cells were collected at 15 min intervals. Lactisole was pre-treated prior to Rg3 treatment. The collected cells were lysed using 0.1 M HCl, and the intracellular cAMP was assayed using ELISA (Enzo Life Sciences, Farmingdale, NY, USA) according to the manufacturer's instructions. The results were normalized to the protein concentration.