Flavone Hispidulin Stimulates Glucagon-Like peptide-1 secretion and Ameliorates Hyperglycemia in Streptozotocin-Induced Diabetic Mice

Abstract

Scope:

Loss of functional β-cell mass is central for the deterioration of glycemic control in diabetes. The incretin hormone glucagon-like peptide-1 (GLP-1) plays a critical role in maintaining glycemic homeostasis via potentiating glucose-stimulated insulin secretion (GSIS) and promoting β-cell mass. Agents that can directly promote GLP-1 secretion, thereby increasing insulin secretion and preserving β-cell mass, hold great potential for the treatment of T2D.

Methods and Results:

GluTag L-cells, INS832/13 cells, and mouse ileum crypts and islets were cultured for examining the effects of flavone hispidulin on GLP-1 and insulin secretion. Mouse livers and isolated hepatocytes were used for gluconeogenesis. Streptozotocin-induced diabetic mice were treated with hispidulin (20 mg/kg/day, oral gavage) for 6 weeks to evaluate its antidiabetic potential. Hispidulin stimulated GLP-1 secretion from L-cell line, ileum crypts, and in vivo. This hispidulin action was mediated via activation of cAMP/PKA signaling. Hispidulin significantly improved glycemic control in diabetic mice, concomitant with improved insulin release and β-cell survival. Additionally, hispidulin decreased hepatic pyruvate carboxylase expression in diabetic mice and suppressed gluconeogenesis in hepatocytes. Furthermore, hispidulin stimulated insulin secretion from β-cells.

Conclusion:

These findings suggest that hispidulin may be a novel dual-action anti-diabetic compound via stimulating GLP-1 secretion and suppressing hepatic glucose production.

Graphical abstract

The incretin hormone glucagon-like peptide-1 (GLP-1) plays a critical role in maintaining glycemic homeostasis via potentiating glucose-stimulated insulin secretion (GSIS) and promoting β-cell mass. In this study, we found for the first time that hispidulin, a naturally occurring flavone, is an inhibitor of phosphodiesterase (PDE) in intestinal L-cells, which results in intracellular cAMP accumulation and subsequent activation of PKA, leading to increased GLP-1 secretion. In addition, hispidulin suppressed gluconeogenesis in primary mouse hepatocytes. Oral administration of hispidulin significantly improved glycemic control in diabetic mice, concomitant with improved insulin release and β-cell survival. Thus, hispidulin may be a novel dual-action anti-diabetic compound via stimulating GLP-1 secretion and suppressing hepatic glucose production, given that loss of functional β-cell mass and excessive hepatic glucose production play critical roles for the development of T2D.

1. Introduction

Loss of β-cell mass and function is central to the deterioration of glycemic control over time in type 2 diabetes (T2D) [1]. Therefore, preservation or improvement of β-cell mass and its insulin secretory function could prevent and treat T2D [2, 3]. It has been well recognized that glucagon-like peptide-1 (GLP-1), which is an incretin hormone secreted from intestinal L-cells, plays a critical role in maintaining glycemic homeostasis via potentiating glucose-stimulated insulin secretion (GSIS) and promoting β-cell proliferation and survival [4]. GLP-1 also delays stomach emptying, induces satiety, and reduces body weight in animal models of obesity [5]. Secretion of GLP-1 from L-cells in the intestine is increased in response to ingested macronutrients, primarily fatty acids [6–8], although glucose [9, 10], some amino acids [11], and dietary fibers [12] may also induce GLP-1 release. In addition, a variety of neurotransmitters and neuropeptides released by the enteric nervous system and enteroendocrine cell types, such as acetylcholine [13] and gastrin-releasing peptide [14], have been implicated in the regulation of GLP-1 secretion. However, no therapeutic strategy based on these mechanisms has been successfully developed for treating T2D.

Recently, GLP-1-based drugs, including GLP-1 analogues and dipeptidyl peptidase-IV (DPP-IV) inhibitors that inhibit breakdown of GLP-1, have been developed for treating T2D [15–17]. However, patients given GLP-1-based drugs, which require injection and are expensive, suffer from side effects such as nausea and vomiting [18]. More importantly, recent studies have found that the use of incretin-based drugs raised substantial concerns of a possible increased risk of pancreatitis and C-cell hyperplasia [18–20]. Furthermore, DPP-IV is a ubiquitous enzyme involved not only in the enzymatic cleavage of numerous peptides, but also in non-enzymatic interactions with some proteins [21], highlighting a wide range of biological functions of DPP-IV. Therefore, non-selective inhibition of DPP-IV may affect a spectrum of important physiological functions of DPP-IV other than degradation of GLP-1. Given these substantial safety concerns, development of safer, cheaper, and more convenient agents that can directly promote GLP-1 secretion from L-cells, thereby improving β-cell mass and function, holds great potential for the treatment of T2D.

Recently, naturally occurring polyphenolic compounds such as flavonoids have drawn great interest due to their pharmacological implications, with considerable attention being devoted to their use in diabetes management [22]. Hispidulin (4′, 5, 7-trihydroxy-6-methoxyflavone) is a flavone (molecular formula‎: ‎C₁₆H₁₂O_6, mass: 300.26) found in some edible plants and medicinal herbs such as Petasites, Artemisia, and Salvia [23]. While studies regarding the health benefits of this compound are scarce, it has been reported that hispidulin has anti-oxidative, anti-inflammatory, and anti-cancer, and anti-seizure effects [23]. However, there is no published study, as to the best of our knowledge, investigated whether hispidulin possesses an anti-diabetic property, although it has been shown that salvia leaf extracts, which may contain hispidulin, exert metabolic beneficial effects in humans [24, 25]. Recent studies showed that some food-derived bioactive agents are GLP-1 secretagogues. For example, it was found that casein hydrolysates induced GLP-1 secretion from clonal L-cell [26]. In addition, there is evidence suggesting that non-nutritive compounds anthocyanins and grape seed extracts can increase GLP-1 secretion [27, 28]. Interestingly, it was found that hispidulin is an agonist of GABAα receptors [29], which are present in L-cells and β-cells involved in the GLP-1 and insulin secretion [30, 31]. In addition, computer modeling analysis showed that hispidulin may be an inhibitor of DPP-IV [32]. However, it is unclear whether hispidulin can stimulate GLP-1 and/or insulin secretion, thereby providing beneficial effects in prevention and treatment of T2D. In this study, we investigated the effect of hispidulin on GLP-1 secretion from L-cells and further determined the mechanism underlying this action. In addition, the anti-diabetic potential of hispidulin was explored by using streptozotocin (STZ)-induced diabetic mice and β-cells.

2. Experimental Section

2.1. Chemicals and Reagents

Apigenin, luteolin, kaempferol, naringenin, 3-isobutyl-1-methylxanthine (IBMX), forskolin, H89, bicuculline methiodide (BM), collagenase XI, and streptozotocin (STZ) were bought from Sigma-Aldrich (St. Louis, MO); pelargonidin, cyanidin chloride, and butin were obtained from Indofine chemical company (Hillsborough, NJ); genistein and formononetin were from Tocris Bioscience (Pittsburgh, PA); hispidulin was purchased from (Aktin Chemicals Inc, Chengdu, China). The purity of all flavonoids is ≥ 98% (HPLC). Cyclic AMP EIA kit were purchased from Cayman Chemical Co (Ann Arbor, MI); insulin assay kits were from Crystal Chemical Co (Elk Grove Village, IL) or Mecodia (Winston-Salem, NC); GLP-1 assay kit and horseradish peroxidase-conjugated goat anti-mouse antibody were from Millipore (Burlington, MA); Protein kinase A (PKA) assay kit was from Arbor Assay Inc (Ann Arbor, MI); and BD Matrigel™ matrix and rabbit caspase-3 antibody were purchased from BD Biosciences (Billerica, MA). Pyruvate carboxylase (PCB) antibodies were purchased from Santa Cruz Biotechnology (Dallas, Texas). Guinea pig insulin antibody and mouse β-actin antibody was purchased from Abcam (Cambridge, MA). Amplex Red glucose assay kit, horseradish peroxidase-conjugated goat anti-rabbit antibody, and SuperSignal™ West Femto substrate were from ThermoFisher Scientific (Waltham, MA).

2.2. Cell Culture

Mouse GluTag L-cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 5.5 mM glucose, 10% fetal bovine serum (FBS), and 1% penicillin/ streptomycin [33]. For GLP-1 secretion experiments, cells were seeded into poly-D-lysine coated 12-well plate (10⁵ cells/well). After 24 h, cells were washed with Krebs Ringer Buffer (KRB, 129 mM NaCl, 2.5 mM CaCl₂, 1.2 mM MgSO₄, 4.8 mM KCl, 1.2 mM KH₂PO₄, 10 mM HEPES, 5 mM NaHCO₃, pH 7.4) supplemented with 0.2% bovine serum albumin (BSA), and treated with various concentrations of hispidulin or with various flavonoids as indicated for 1 h. Active GLP-1 secreted in supernatants was measured using an ELISA kit. Cells were then lysed to measure protein content, and GLP-1 level was normalized to the cellular protein concentration from the same sample. INS1_832/13 cells (kindly provided by Dr. Christopher Newgard, Duke University) were cultured in complete RPMI-1640 medium as described [34]. Mouse islets were isolated from 8 to 10-week-old C57BL/6 male mice and maintained in complete RPMI1640 medium [34].

2.3. Isolation of Mouse Ileal Crypts and GLP-1 Secretion Assay

Mouse ileum crypts were isolated from 8 to14-wk-old C57BL6/J male mice as reported [35]. After euthanization of the mice, the entire small intestine was gently taken out, and whole segment of ileum was subsequently excised, which was then immediately flushed using ice-chilled PBS to clean intestinal contents. For isolating crypts, the ileal segments were minced into 1–2 mm pieces, digested with 0.3 mg/ml collagenase XI solution at 37ºC, and then centrifuged at 100 g for 3 min. The crypt fragments were collected and resuspended in DMEM supplemented with 25 mM glucose, 10% FBS, and 1% penicillin and streptomycin, and plated on Matrigel-coated 24-well plates. The crypts were maintained for 48 h, and then treated with various concentrations of hispidulin for 1 h at 37 ºC to measure GLP-1 secretion.

2.4. Intracellular cAMP and Adenylate Cyclase Activity Measurements

L-cells were incubated with various concentrations of hispidulin or with 1 μM forskolin for 20 min. Intracellular cAMP levels were measured by an EIA kit. For measuring adenylate cyclase activity, plasma membranes were isolated from cultured L-cells [36]. The membranes were then incubated with hispidulin or vehicle at 37^oC for 10 min, followed by measuring cAMP production as previously described [37].

2.5. PKA Activity Assay

L-cells were incubated with various concentrations of hispidulin for 30 min. In some experiments, cells were pretreated with PKA inhibitor H89 or vehicle for 20 min before addition of 10 μM hispidulin. The enzymatic activity of PKA in the cell lysates was measured using a PKA activity assay kit according to the manufacturer’s protocol.

2.6. DPP-IV Activity Measurement

The effect of hispidulin on DPP-IV activity was assessed as previously reported [38]. Stock hispidulin dissolved in DMSO was further diluted in 100 mM Tris buffer (pH 8) with DMSO concentration in each dilution normalized to 0.4%. For measuring DPP-IV activity, 50 μL human DPP-IV solution (2 mU/mL in 100 mM Tris buffer, pH 8.0) and 50 μL hispidulin were mixed in a 96-well plate and incubated at 37ºC for 10 min, followed by the addition of 100 μL 0.5 mM GPpNA chromogenic substrate dissolved in 100 mM Tris buffer (pH 8.0). For blank control, 50 μL 100 mM Tris buffer without DPP-IV was added in the reaction. As a positive control, DPP-IV enzyme was mixed with a DPPV-IV inhibitor, diprotin. The plate was further incubated at 37ºC. After 1 h, the reaction was terminated by the addition of 50 μL 3% acetic acid, and the absorbance was read at 405 nm. The DPP-IV activity was expressed as a percentage of the vehicle control.

2.7. Insulin Secretion Assay

INS1_382/13 cells (10⁵ cells/well) were seeded into 12-well plates. After 48 h, cells were starved with glucose-free RPMI-1640 medium containing 2% FBS for 2 h. Cells were then cultured with various concentrations of hispidulin in KRB containing 5 mM glucose or 5 μM hispidulin with various concentrations of glucose for 1 h. Cells cultured in KRB containing 1 mM glucose (1G) were used as the control. Mouse islets (40 islets/well) were treated with various concentrations of hispidulin or vehicle and 5.5 mM glucose for 1 h. Insulin secreted into supernatants was measured using ELISA kits. All data were normalized to the cellular protein concentrations.

2.8. Animal Experiments

Eight-wk-old male C57BL/6J mice were purchased from Jackson Laboratory (Bar Harbor, ME) and housed on a 12-h light/dark cycle from 7:00 am to 7:00 pm at constant temperature (22–25 ºC) in an animal room with ad libitum access to water and a standard rodent chow diet (D06072701, Research Diets, New Brunswick, NJ). All protocols for the following animal experiments were approved by the Institutional Animal Care and Use Committee at Virginia Tech (IACUC-18-115).

2.9. Dosage Information

Mice in all animal studies described below were received either 20 mg/kg/day hispidulin suspended in 2% methylcellulose or the same amount of vehicle (2% methylcellulose) via oral gavage between 3–4 o’clock in the afternoon. For testing the acute effects of hispidulin on GLP-1 secretion and food intake, hispidulin was only administered once. For examining its anti-diabetic effect, hispidulin or vehicle was given to diabetic mice once daily for 6 wks. This dose is similar or below those given to mice in previous studies [39, 40]. While no clinical study of this compound, to the best of our knowledge, has been reported, this dose of hispidulin, which is equivalent to only about 97 mg daily for a 60 kg human [41], is below the amount of some other flavonoids used in human studies [42, 43]. Therefore, this moderate amount of hispidulin could be realistically consumed by humans as a dietary supplement. In our in vitro studies, hispidulin at up to 100 μM, which is well beyond physiologically achievable doses via dietary intake, had no effect on cell viability (data not shown).

2.10. Evaluation of Hispidulin-Stimulated GLP-1 Secretion and Food Intake in vivo

For determining whether hispidulin stimulates GLP-1 secretion in vivo, first cohort of male mice (10 wks old) were randomly divided into two groups (n= 8/group). After fasted for 5 h, blood samples were collected in the pre-chilled tubes containing 50 μM diprotin A and 5 mM EDTA. Mice were then given either 1 g/kg glucose plus hispidulin or 1 g/kg glucose plus vehicle (2% methylcellulose) via oral gavage. After 20 min post-gavage, blood samples were drawn from tail vein, and the concentrations of total GLP-1 in plasma were measured using an ELISA kit following manufacturer’s protocol. For determining whether hispidulin affects food intake, a second cohort of male mice (9 wks old) were divided into two groups (n= 10 mice/group) with initial body weight balanced. After fasted for 5h, mice were given either hispidulin or vehicle via oral dose. Thereafter, pre-weighed food was provided, followed by recording food intake at 1, 3, 5, 8, 12, and 24 h. Body weight was measured at 24 h again post-hispidulin treatment.

2.11. Metabolic Effects of Hispidulin in Streptozotocin (STZ)-Induced Diabetic Mice

For evaluating the anti-diabetic effect of hispidulin, another cohort of mice were intraperitoneally (ip.) injected with STZ or vehicle at 45 mg/kg daily for 5 consecutive days. Ten days after STZ injection, mice with non-fasting blood glucose over 250 mg/dl in two consecutive days were divided into two groups based on their body weight and non-fasting blood glucose, and then received hispidulin or vehicle for 6 wks. Body weight, food intake, and blood glucose were measured weekly throughout the study. After 5 wks of treatment, mice were fasted for 16 h and then injected ip. with glucose (0.75 g/kg of body weight) for glucose tolerance test [34]. For determining GSIS, three days after glucose tolerance test, mice were fasted 16 h and then injected ip. with glucose (0.75 g/kg of body weight). Plasma samples were collected at 0 and 30 min after glucose administration. For measuring non-fasting blood insulin, non-fasting plasma samples were collected after 6-wk of treatment. Plasma insulin levels were measured using an ELISA kit.

2.12. Immunofluorescence and Histological Examinations of the Pancreas

At the end of the feeding study, mouse pancreata were dissected, weighed, and fixed in 4% paraformaldehyde. A series of tissue sections (5-μm thickness) were prepared by AML Laboratory (Jacksonville, FL). For detecting apoptotic β-cells, pancreatic sections were co-stained with antibodies against guinea pig anti-insulin (ab7842) and rabbit anti-caspase-3 (559565). Another set of tissue section was stained with hematoxylin and eosin (H&E) for evaluating islet morphology and mass. The β-cell area was measured using images acquired from insulin-stained pancreatic sections. The β-cell mass and percentage of apoptotic β-cells were calculated as previously described [34].

2.13. Western Blot Analysis

At the end of experiment, mouse liver was collected and snap frozen in liquid nitrogen for western blot analysis. Liver sample was lysed and loaded in 10% TGX™ FastCast™ precast gel (Bio-Rad Laboratories, Hercules, CA) and transferred to polyvinylidene difluoride (PVDF) blot. After blocking, blot was incubated with antibodies, pyruvate carboxylase (PCB, 1:1000) and β-actin (1:1000) at 4ºC for overnight. Afterwards, blot was incubated with horseradish peroxidase-conjugated anti-mouse or anti-rabbit antibodies for 1 h. The immunoreactive proteins were detected by using SuperSignal™ West Femto substrate. The optical density of protein band was quantitated with a software program (Image J, NIH) and normalized to that of the loading control.

2.14. Glucose Production Measurement

Primary hepatocytes were isolated from 10 wks old male mice as previous reported [44]. The cells were seeded into 12-well plates (10⁵ cells/well). For glucose production assay, cells were cultured in serum-free DMEM supplemented with 5.5 mM glucose and 1% penicillin streptomycin for 5 h, and then incubated in gluconeogenesis medium (glucose-free DMEM containing 20 mM sodium lactate and 2 mM sodium pyruvate) with various concentrations of hispidulin or forskolin (F, 10 μM) for 3 h. Glucose released in the media was measured using a glucose assay kit and normalized to protein content of the same samples.

2.15. Statistical Analysis

Data were analyzed by one-way ANOVA or student’s t-test using JMP 13 (SAS Institute Inc, Cary, NC) where appropriate. Treatment differences were subjected to Duncan’s multiple range test. A p-value < 0.05 was considered significant. Values are presented as mean ± standard error of mean (SEM).

3. Results

3.1. Hispidulin Stimulates GLP-1 Secretion

To test whether hispidulin stimulates GLP-1 secretion, we first treated GluTag L-cells or mouse ileal crypts with hispidulin (0.1, 1, 10, and 50 μM) for 1 h. We show that hispidulin promoted GLP-1 secretion from L-cells, with 1–50 μM concentrations inducing significant GLP-1 release (Fig.1A). Similarly, hispidulin treatment for 1 h increased GLP-1 secretion from crypts isolated from mouse ileum, with 1 μM eliciting 2.5-fold GLP-1 release over vehicle control (P< 0.05) (Fig. 1B), suggesting that hispidulin effect on L-cells is physiologically relevant.

Figure 1. Hispidulin stimulated GLP-1 secretion.

GluTag cells (10⁵ cells/well) (A) or isolated mouse ileal crypts (B) were treated with various concentrations of hispidulin or 10 μM forskolin and 10 μM IBMX (FI) in KRB containing 0.2% BSA. Cells and crypts treated with vehicle alone (0.05% DMSO) served as the control. After 1h, GLP-1 released into supernatant was measured and normalized to protein content in the same sample. The basal levels of active GLP-1 secretion from L-cells and crypts were 72.1 ± 5.4 pM/mg and 7.4 ± 1.4 pM/mg, respectively. Data are means± SEM from six experiments performed in duplicate each. * P < 0.05, ** P < 0.01, and *** P <0.001 vs. control.

3.2. Stimulation of GLP-1 Secretion by Hispidulin is Mediated Through the cAMP/PKA-Dependent Pathway.

As inhibition of DPP-IV could result in GLP-1 accumulation, we examined whether hispidulin induces GLP-1 secretion through inhibition of DPP-IV. We found that hispidulin in a concentration range of 1 nM to 100 μM had no significant effect on DPP-IV enzymatic activity, whereas diprotin A (1 nM to 100 μM) dose dependently inhibited DPP-IV activity, with 10 μM concentration reducing DPP-IV activity by 50% and 100 μM resulting in 90% inhibition (Fig. 2A). This result demonstrates that hispidulin action on GLP-1 secretion from L-cells is mediated through a DPP-IV-independent mechanism. It has been shown that hispidulin can modulate the activity of GABAα receptor [29], and activation of GABAα receptor is capable of stimulating GLP-1 secretion in GluTag cells [31]. However, the blockage of GABAα receptor with bicuculline methiodide (BM), a competitive inhibitor of GABAα receptor, increased GLP-1 secretion but failed to inhibit hispidulin-induced GLP-1 secretion (Fig. 2B), suggesting that GABAα receptor is not involved in hispidulin stimulation of GLP-1 secretion. However, incubation of GluTag cells with hispidulin increased intracellular cAMP accumulation (Fig. 3A), a dose-dependent pattern that is consistent with its effects on GLP-1 secretion. Since PKA is directly activated by cAMP, we next examined whether the elevated cAMP by hispidulin is sufficient to activate PKA in L-cells. As shown in Fig. 2C, hispidulin potently stimulated PKA activity. To examine whether the cAMP/PKA pathway mediates hispidulin induced GLP-1 secretion, we pre-incubated L-cells with the PKA inhibitor H89 (10 μM), and found that inhibiting the activity of PKA abrogated hispidulin-induced PKA activity (Fig. 2D) and GLP-1 secretion (Fig. 2E). These results indicate that hispidulin-induced GLP-1 secretion is mediated via a PKA-dependent mechanism. To examine how hispidulin elevates intracellular cAMP, we preincubated the cells with 0.2 mM IBMX, a potent phosphodiesterase (PDE) inhibitor, before addition of 10 μM hispidulin. As shown in Fig. 2F, hispidulin or IBMX alone increased cAMP levels over the control by 1.8-fold and 1.3-fold, respectively. However, hispidulin-stimulated cAMP production was no further augmented in the presence of IBMX. We then isolated cell plasma membranes and measured adenylate cyclase activity with or without hispidulin treatment. As shown in Fig, 2G, hispidulin had no stimulatory effect on adenylate cyclase activity (Fig.2H), suggesting that hispidulin-induced cAMP elevation was likely due to inhibition of PDE, thereby preventing cAMP from degradation.

Figure 2. Hispidulin-induced GLP-1 secretion was mediated via the cAMP-PKA pathway.

(A) The inhibitory effect of hispidulin and DPP-IV inhibitor, diprotin A, on DPP-IV activity was measured using an assay kit. The concentrations of the compounds are plotted on a logarithmic scale. Data are presented as mean ± SEM (n= 3). (B) GluTag cells (10⁵ cells/well) were preincubated in KRB containing 50 μM GABA receptor antagonist, bicuculline methiodide (BM), or vehicle for 20 min, followed by the addition of 10 μM hispidulin or vehicle (0.01% DMSO) in KRB buffer for 30 min. Supernatants were collected for measuring active GLP-1 concentrations. The basal GLP-1 secretion level was 46.0 ±1.4 pM/mg. Data are shown as means ± SEM (n=3). * P < 0.05, ** P < 0.01 vs. control cells. (C) GluTag cells (10⁵ cells/well) were treated with various concentrations of hispidulin or 1 μM cAMP inducer forskolin (Fos), in KRB for 20 min. Intracellular cAMP contents were measured using an ELISA kit. (D) GluTag cells (10⁵ cells/well) were treated with vehicle (C, 0.01% DMSO), hispidulin (10 μM), or forskolin (Fos, 1 μM) in KRB for 30 min. PKA activity in the cell lysates was measured using an assay kit. (E) GluTag cells (10⁵ cells/well) were pretreated with 10 μM PKA inhibitor, H89, or vehicle (0.1% DMSO) in KRB for 20 min, followed by addition of 10 μM hispidulin or vehicle (0.15% DMSO) for 30 min. Supernatants were then collected for measuring active GLP-1. (F) Cell lysates were used for PKA activity assay. (G) GluTag cells (10⁵ cells/well) were preincubated with 0.2 mM IBMX or vehicle (0.02 % DMSO) in KRB for 20 min, followed by addition of 5 μM hispidulin or vehicle (0.05% DMSO) at 37°C for another 20 min. Intracellular cAMP was then measured and normalized to protein content (H) Plasma membranes isolated from GluTag cells were incubated with adenylate cyclase assay mixture in the presence or absence of 10 μM hispidulin, 5 μM forskolin (Fos), or vehicle (0.05% DMSO) in KRB at 37°C for 15 min. Cyclic AMP production was then measured for assessing adenylate cyclase activity. Data are shown as means ± SEM (n=4–5). * P < 0.05, ** P < 0.01, and *** P <0.001 vs. vehicle alone treated cells.

Figure 3. The stimulatory effect of hispidulin and other flavonoids on GLP-1 secretion.

(A) Chemical structures of tested flavonoids. (B) GluTag cells (5 × 10⁴ cells/well) were treated with various flavonoids (10 μM), 1μM forskolin plus 1μM IBMX (FI), or vehicle (C, 0.05% DMSO) in KRB for 1 h. Active GLP-1 secreted into the supernatant was determined. The basal level of released GLP-1 from untreated control cells was 67.6 ± 5.7 pM/mg. Data are presented as mean ± SEM from three independent experiments performed in duplicates. *, P< 0.05 vs. C.

3.3. The Specificity of Hispidulin on GLP-1 Secretion

To determine the specificity of hispidulin action on GLP-1 secretion, we tested the effect of a cohort of structurally related flavonoid compounds in parallel to hispidulin on GLP-1 secretion. The result from this analysis showed that apigenin and luteolin, two other flavones, increased GLP-1 secretion by 31% and 37 %, respectively, a magnitude slightly less potent than that of hispidulin. However, no other subclasses of flavonoids, including isoflavones (genistein, formononetin), flavonol (kaempferol), flavonones (naringenin, butin), and anthocyanidins (pelargonidin, cyanidin) was active in stimulating GLP-1 secretion from L-cells (Fig. 3). These results demonstrate the unique effect of flavones on GLP-1 secretion that is possibly not shared by other subtypes of flavonoid compounds.

3.4. Hispidulin Induced GLP-1 Secretion in vivo

To determine whether hispidulin stimulates GLP-1 secretion in vivo, we orally gave 20 mg/kg hispidulin or vehicle to 5 h-fasted C57/BL6 mice (male, 10 wks old) and measured plasma total GLP-1 concentrations at 20 min after oral gavage. We found that acute administration of hispidulin significantly increased plasma GLP-1 levels in mice (Fig. 4A). Using another cohort of mice (male, 9 wks old), we determined whether hispidulin regulates food intake. In this regard, 5-h fasted mice were given 20 mg/kg hispidulin or vehicle via oral gavage, followed by measuring their food intake during 24-h period. As shown in Fig. 4B, administration of hispidulin significantly reduced food intake after 1h, 3h, and 5h of oral gavage as compared with that of control mice (p < 0.05), which is consistent with the established role of GLP-1. However, the 24-h cumulative food intake between two groups was not significantly different. In addition, the body weight of mice given hispidulin and the controls were comparable (data not shown).

Figure 4. Hispidulin improved glycemic control and insulin levels in STZ-induced diabetic mice.

(A) Healthy male C57BL/6J mice (10 wks old) was fasted 5h and then given 1 g/kg glucose plus 20 mg/kg hispidulin dissolved in 2% methylcellulose or 1 g/kg glucose plus vehicle (Control) via oral gavage. Blood samples were collected at 0 and 20 min after oral gavage, and plasma total GLP-1 levels were measured (n= 8 mice/group). A second cohort of male mice (9 wks old) was fasted 5 h and then given 20 mg/kg hispidulin or vehicle (Control) via oral gavage. Food intake was measured after 1, 3, 5, 8, 12, and 24 h of oral gavage (n=10 mice/group). STZ-induced diabetic mice were administered with hispidulin (20 mg/kg via oral gavage) or vehicle for 6 wks. Body weight (C) and non-fasting (D) and fasting (E) blood glucose levels were measured at indicated time points during hispidulin treatment. Glucose tolerance test were performed after 5-wk treatment (F), and the area under the curve was calculated shown on the right panel of F. GSIS (G) and non-fasting plasma insulin levels (H) were measured at the end of feeding experiment; livers were isolated and pyruvate carboxylase (PCB) protein levels in whole cell lysates were measured by immunoblotting and normalized to β-action contents (I). Data are means ± SEM (n=11–13 mice/group). (J) Freshly isolated hepatocytes from healthy male mice (10 wks old) were cultured in gluconeogenic medium containing various concentrations of hispidulin or forskolin (F: 10 μM) for 3 h. Glucose production was determined using an assay kit. Data are means ± SEM from 4 independent experiments. C: non-diabetic mice; STZ: STZ-induced diabetic mice treated with vehicle; STZ+H: STZ-induced diabetic mice treated with hispidulin. * P < 0.05, ** P < 0.01, and *** P <0.001 vs. C; # P < 0.05, ### P <0.001, and #### P <0.0001 vs. STZ alone.

3.5. Hispidulin Ameliorates Hyperglycemia in STZ-Induced Diabetic Mice

Next, we performed a mouse study to assess whether hispidulin has the anti-diabetic potential. To that end, STZ-induced diabetic mice with non-fasting blood glucose ≥250 mg/dl were treated with vehicle or hispidulin (20 mg/kg) once daily for 6 wks. During the treatment period, hispidulin-treated diabetic mice had a comparable body weight with that of vehicle-treated diabetic mice (Fig. 4C). Oral administration of hispidulin significantly ameliorated hyperglycemia in diabetic mice. Notably, hispidulin more effectively reduced non-fasting blood glucose levels in diabetic mice, which were already significantly lower as compared with those in untreated diabetic mice as early as 1 wk of treatment (Fig. 4D), while fasting blood glucose levels were gradually improved over the time (Fig. 4E). After 5 wks, hispidulin-treated diabetic mice displayed lower (P < 0.05) fasting blood glucose levels as compared with vehicle-treated diabetic mice. In addition, diabetic mice with hispidulin administration showed improved glucose tolerance (Fig. 4F), which were associated with better GSIS as compared with that of untreated diabetic mice (Fig. 4G). Consistently, hispidulin-treated diabetic mice also had higher non-fasting plasma insulin concentrations relative to those in control diabetic mice (Fig. 4H).

The fasting hyperglycemia is primarily caused by excessive glucose output from the liver [45]. As oral administration of hispidulin reduced fasting blood glucose, we examined whether hispidulin suppresses hepatic gluconeogenesis, which is largely responsible for excessive hepatic glucose output in diabetes [46]. We found that the protein levels of hepatic pyruvate carboxylase (PCB), a key enzyme for gluconeogenesis, were significantly increased in diabetic mice, but were normalized after treatment with hispidulin (Fig. 4I). We then isolated and cultured mouse hepatocytes with or without hispidulin treatment, and found that hispidulin directly suppressed glucose production in hepatocytes with the concentration as low as 1 μM already eliciting a significant effect (Fig. 4J), although it is unclear whether this concentration is readily achievable in vivo by dietary intake of this compound.

3.6. Hispidulin Improves Pancreatic β-Cell Survival in Diabetic Mice

To determine whether the increased GSIS is associated with the improved islet function, we evaluated pancreatic islet mass after H&E staining. As expected, STZ administration severely reduced islet mass (Fig. 5A and B). Hispidulin-treated diabetic mice had higher islet mass than that of vehicle-treated mice, although the difference was not statistically significant (p=0.08). As STZ induces diabetes by selectively destroying β-cells [47], we evaluated whether hispidulin treatment improved β-cell survival by performing double immunolabeling of insulin and the activated caspase-3, a key protease involved in the terminal steps of cell apoptosis. Consistently, diabetic mice treated with hispidulin displayed fewer apoptotic β-cells than vehicle-treated diabetic mice (Fig. 5C and D).

Figure 5. Hispidulin improved β-cell mass and survival in diabetic mice.

(A) Representative images from H&E staining of pancreatic sections are shown (scale bar=100 μm, magnification=20×). (B) Islet mass was quantitated as described in Materials and Methods. (C) Pancreatic sections were double immunostained for insulin and caspase 3 (Cas3) and then counterstained for nuclei with DAPI. Merged images are also shown (scale bar=50 μm, magnification=20×). (D) The number of apoptotic β-cells in each islet from merged images was counted and expressed as percentage of total insulin-positive cells. Data are means ± SEM (n= 4 mice). C: non-diabetic mice; STZ: STZ-induced diabetic mice treated with vehicle; STZ+H: STZ-induced diabetic mice treated with hispidulin. ***, P < 0.001, and ****, P< 0.0001 vs. C; ##, P < 0.01 vs. STZ alone.

3.7. Hispidulin Stimulates Insulin Secretion

As we found that hispidulin may also promote insulin secretion in response to glucose in diabetic mice, we then examined whether hispidulin directly enhances insulin secretion from β-cells. To that end, we cultured INS832/13 insulin secreting cells with various concentrations of hispidulin for 1 h, followed by measuring insulin released in the medium. As shown in Fig. 6, hispidulin dose-dependently augmented GSIS (p<0.05), with the lowest effective concentration at 1 μM (Fig. 6A and B). Consistently, hispidulin also augmented GSIS in mouse islets (Fig. 6C).

Figure 6. Hispidulin directly stimulated insulin secretion.

(A) INS832/13 cells (10⁵ cells/ well) were incubated with various concentrations of hispidulin in KRB containing 5 mM glucose (5G) for 1h. Cells incubated with KRB containing 1mM glucose (1G) was used as the control. (B) INS832/13 cells (5×10⁴ cells/ well) were incubated with vehicle (0.005% DMSO) or 5 μM hispidulin in KRB containing various concentrations of glucose for 1 h. (C) Freshly isolated mouse islets (40 islets/ well) were incubated in KRB containing 1 mM glucose or 5 mM glucose with or without indicated concentrations of hispidulin for 1h. Insulin released in the supernatants was measured using an ELISA kit. The basal insulin secretion level was 79.3 ± 8.1 μg/mg. Data are expressed as means ± SEM (n=3–5). *, P < 0.05, **, P < 0.01, ***, P <0.001 vs. 1G; #, P < 0.05, ###, P <0.001, ####, P <0.0001 vs. 5G.

4. Discussion

It is well established that GLP-1 plays a critical role in maintaining glycemic homeostasis via potentiating GSIS and promoting β-cell proliferation and survival [4, 48–50]. Therefore, GLP-1-based drugs, including GLP-1 analogues and DPP-IV inhibitors, have been developed for treating human T2D [15–17]. It has been shown that some flavones exert anti-diabetic action in animal models [51, 52]. However, the specific targets for these flavones to improve glycemic control are unclear. In this study, we reported for the first time that hispidulin, a flavone isolated from a medicinal Chinese herb, directly stimulated GLP-1 secretion from L-cells. Consistently, hispidulin treatment increased plasma GLP-1 concentrations and ameliorated hyperglycemia in STZ-induced diabetic mice, which was mediated at least partially by preserving β-cell mass and protecting its function. Loss of functional β-cell mass is central for overt development of diabetes [1, 53–56]. Therefore, hispidulin may be a plant-derived anti-diabetic compound via GLP-1-mediated promotion of β-cell mass and function, given that preservation or improvement of β-cell mass has been shown to be effective in preventing and treating diabetes [57, 58].

The result from comparing the effect of hispidulin on GLP-1 secretion with a cohort of structurally related flavonoids demonstrated that flavone, but not other structurally related flavonoids induce GLP-1 secretion from L-cells, which might suggest the structure specificity of stimulating GLP-1 secretion. Although the specific chemical structure on hispidulin responsible for its stimulatory action on GLP-1 secretion remains to be determined, its unique methyl group at C7 position may not be important for this action. Flavones have a double bond between C2 and C3 of the C-ring in the flavonoid skeleton, and an oxidized group at the C4 position of its B-ring, which may be crucial for this unique effect of flavone on GLP-1 secretion.

Secretion of GLP-1 from L-cells in the intestine is increased in response to ingested macronutrients, primarily fatty acids [6–8], although glucose [9, 10], some amino acids [11], and dietary fibers [12] may also induce GLP-1 release. In addition, a variety of neurotransmitters and neuropeptides released by the enteric nervous system and enteroendocrine cell types, such as acetylcholine [13], gastrin-releasing peptide [14], and GABA [31], have been implicated in the regulation of GLP-1 secretion. Hispidulin was shown to activate GABAα receptor [29], which is expressed on GluTag L-cells [31]. It has been shown that activation of GABAα receptor leads to GLP-1 secretion [30, 31]. In the present study however, we found that the blockage of GABAα receptor using its competitive inhibitor, which was shown to completely abolished GABA-induced GLP-1 secretion from L-cells [31], had no effect on hispidulin-induced GLP-1 secretion. Consistently, inhibition of GABA receptor was also ineffective in inhibiting hispidulin-stimulated insulin secretion from insulin secreting cells (data not shown). Thus, we believe that GABAα receptor is not involved in hispidulin-induced hormone secretion.

DPP-IV is a ubiquitous enzyme involved in the enzymatic cleavage of GLP-1 [21], and inhibition of DPP-IV prevents the breakdown of GLP-1, thereby elevating GLP-1 levels in the circulation [59]. However, our data did not show that hispidulin had significant effect on DPP-IV enzymatic activity. This result is in contrast with a previous report that hispidulin inhibited DPP-IV with IC50 of 0.49 μM [32], which is surprisingly more potent that diprotin A, a potent pharmaceutical inhibitor of DPP-IV with an IC50 of 3.7 μM [60]. This discrepancy is unclear, which could be due to the different species of DPP-IV. In the present study, we used human DPP-IV, while DPP-IV used in study by Bower et al. was from porcine [32]. Although amino acid sequence of porcine and human DPP-IV shares an overall 88% identity, it was found that two species are greatly different in susceptibility to certain inhibitors [61].

It is well established that an elevation of cAMP level and consequent activation of PKA in L-cells enhances GLP-1 secretion [62]. In the present study, we found that hispidulin-stimulated cAMP accumulation and activated PKA at the same concentration range as its induction of GLP-1 secretion. We further demonstrated that activation of PKA by hispidulin in response to elevated intracellular cAMP plays an essential role in mediating hispidulin-induced GLP-1 secretion since inhibition of PKA largely ablated hispidulin-stimulated GLP-1 secretion. The efficacy of the inhibitor was confirmed by demonstrating its ability to block hispidulin-stimulated PKA activity (data not shown). However, it is presently unclear how PKA is involved in hispidulin-stimulated GLP-1 secretion. PKA is a heterotetramer composed of a regulatory subunit dimer and a catalytic subunit dimer. Although it is known that elevation of cAMP causes translocation of PKACβ to the plasma membranes [63], where it may interact with L-type Ca²⁺ channel [64] and thereby may participate in the hormone exocytotic process, it is unclear whether PKACβ primarily regulates GLP-1 secretion, therefore mediating this hispidulin action. Adenylate cyclase and cAMP-specific PDE are the enzymes responsible for the regulation of cAMP. Our results suggest that hispidulin may elevate cAMP primarily via inhibiting PDE, as inhibition of PDE negated the hispidulin effect on intracellular cAMP accumulation. However, it remains to be determined how hispidulin inhibits PDE and which isoform is specifically targeted in L-cells.

Fasting hyperglycemia is primarily caused by excessive hepatic gluconeogenesis as a result of upregulated expression and activity of key gluconeogenic enzymes, including PCB, PEPCK, and G6Pase [65–68]. Our results show that PCB protein levels in the livers of diabetic mice significantly increased, but were normalized by hispidulin treatment. PCB, the initial and critical enzyme for gluconeogenic process, apparently plays a vital role in developing glycemia in humans [69], as inhibition of PCB greatly suppressed gluconeogenesis [70]. Thus, the improvement in fasting hyperglycemia by hispidulin treatment could be partially ascribed to its suppressing action on PCB expression, leading to reduced glucose production in the liver of diabetic mice. It is well established that insulin opposes glucagon action on hepatic glucose production, but this effect is diminished in STZ-induced diabetic mice due to insufficient insulin, which consequently causes excessive glucose production [71–73]. Therefore, we cannot exclude the possibility that hispidulin suppression of PCB expression and gluconeogenesis is partially secondary to its effect on insulin secretion, given that hispidulin treatment improved circulating insulin levels in diabetic mice. However, hispidulin might directly regulate PCB, as our data show that it suppresses gluconeogenesis in isolated hepatocytes. The mechanism by which hispidulin modulates the expression of hepatic PCB is still unknown. It was found that PCB expression is regulated by peroxisome proliferator-activated receptor-gamma (PPARγ) in adipose tissues in vivo and cultured preadpocytes [74], and hispidulin can inhibit PPARγ expression in preadipoctes [75]. These data suggest that hispidulin may reduce PCB expression via inhibiting PPARγ.

Consistent with the in vivo data, we demonstrated that hispidulin stimulated insulin secretion from cultured insulin secreting cell and pancreatic islets. These results obtained from both in vivo and in vitro experiments strongly suggest that hispidulin could also act directly on pancreatic β-cells to enhance its function, thereby contributing to its anti-diabetic effect. However, the mechanism for this action is unclear. It is well characterized that activation of cAMP signaling augments GSIS [76]. Our in vitro study shows that hispidulin treatment increased intracellular cAMP in L-cells. Therefore, it is likely that hispidulin stimulates insulin secretion through a cAMP-mediated mechanism, although it was not determined in the present study whether hispidulin also activates the cAMP/PKA pathway in β-cells.

Our results show that at least 1 μM hispidulin is needed for achieving significant stimulatory effect on insulin secretion in vitro. It is unclear whether this concentration in the circulation is achievable following dietary intake of hispidulin. The majority of flavonoids are poorly absorbed [77], and most ingested flavonoids pass to the colon where they are degraded by colonic bacteria into smaller molecules and phenolic acids [78, 79]. In the enterocytes, the absorbed flavonoids are metabolized by conjugation with glucuronic acid [80, 81]. Flavonoids that reach the liver are further conjugated with sulfate and/or methyl groups or excreted back with bile components [77, 82, 83]. Therefore, the concentrations of unmetabolized flavonoids in the circulation could be very low. Indeed, it was found that the maximum plasma concentrations of some flavones in mice and humans after dietary ingestion were less than 1 μM [84, 85]. However, there is no published study, as to the best of our knowledge, on the bioavailability of hispidulin.

In summary, we have found for the first time that hispidulin induces GLP-1 via the cAMP signaling pathway. Our animal study showed that hispidulin alleviated diabetes, which was associated with improved β-cell mass and function but reduced hepatic gluconeogenesis. These findings suggest that hispidulin could be a novel dual action, natural compound that exerts anti-diabetic action via promoting β-cell function and suppressing hepatic glucose production, given that loss of functional β-cell mass and excessive hepatic glucose production play critical roles for the development of T2D.

Abbreviations:

AC

adenylate cyclase

BM

bicuculline methiodide

BSA

bovine serum albumin

cAMP

cyclic adenosine monophosphate

DPP-IV

dipeptidyl peptidase-IV

FBS

fetal bovine serum

GLP-1

glucagon-like peptide-1

GSIS

glucose-stimulated insulin secretion

HBSS

Hanks Balanced Salt Solution

IBMX

3-Isobutyl-1-methylxanthine

KRB

Krebs Ringer Buffer

PDE

phosphodiesterase

PKA

protein kinase A

STZ

streptozotocin

T2D

type 2 diabetes