Involvement of cAMP/EPAC/TRPM2 Activation in Glucose- and Incretin-Induced Insulin Secretion

It has been long proposed that glucose-stimulated insulin secretion in pancreatic β-cells is initiated by closure of the ATP-sensitive K⁺ (K_ATP) channel, followed by membrane depolarization (1). In theory, closure of the K_ATP channel is an important process but is not sufficient to induce the shift of the membrane potential toward a threshold level, as membrane potential is determined by the overall balance of outward and inward currents. Modest constitutive opening of background inward current through nonselective cation channels (NSCCs) is crucial to facilitate depolarization after K_ATP channel closure (2). This idea suggests that further regulated opening of a class of NSCCs may bring about greater depolarization. However, whether glucose metabolism regulates NSCC activity remains unclear.

The incretin hormones, GLP-1 and glucose-dependent insulinotropic polypeptide (GIP), potentiate insulin secretion in association with increased intracellular Ca²⁺ concentrations at insulin-secreting glucose concentrations (3–5). GLP-1 fails to increase insulin secretion from the islets of mice deficient in transmembrane receptor potential (TRP) melastatin 2 (TRPM2) (6,7), a type of NSCC, suggesting that the TRPM2 channel is essential for GLP-1–induced potentiation of glucose-stimulated insulin secretion (8). GLP-1 depolarizes the plasma membrane by the opening of NSCC in β-cells (2). Several types of NSCC (TRPs) are expressed in insulin-secreting cells (9). The aims of the current study were to determine 1) the type of NSCC activation (through TRPM2 or other TRPs) that is crucial for signaling after stimulation of the incretin receptor, 2) whether the NSCC is modulated by glucose metabolism, and 3) the underlying mechanism involved in these physiologically and clinically fundamental stimuli.

Male Wistar rats and C57BL/6J mice (CLEA Japan, Inc.) were housed in accordance with the institutional guidelines for animal care in an air-conditioned room with a 12-h light/dark cycle, and food and water were available ad libitum. The study protocol was approved by the institutional animal ethics committee. Islets of Langerhans were isolated by collagenase digestion from male Wistar rats (age 8–12 weeks; weight ~200 g), C57BL/6J mice, and TRPM2-deficient mice (8 weeks old [20 g]) using a previously reported method (10,11). Briefly, animals were anesthetized by injection of pentobarbital (25–100 mg/kg i.p.) followed by injection of collagenase (1.05 mg/mL; Sigma-Aldrich, Tokyo, Japan) dissolved in a 5 mmol/L CaCl₂ solution containing HEPES-added Krebs-Ringer bicarbonate buffer (HKRB) (129 mmol/L NaCl, 5 mmol/L NaHCO₃, 4.7 mmol/L KCl, 1.2 mmol/L KH₂PO₄, 2 mmol/L CaCl₂, 1.2 mmol/L MgSO₄, and 10 mmol/L HEPES at pH 7.4 with NaOH) directly into the common bile duct. The pancreas was incubated at 37°C for 15 min in HKRB buffer. Collected islets and subsequently dispersed cells were used for insulin release and for electrophysiological experiments, respectively.

TRPM2-knockout (KO) mice were kindly provided by Dr. Y. Mori (Kyoto University) (12). TRPM2-KO mice were backcrossed with the C57BL/6J strain for at least nine generations.

Each batch of 10 islets was incubated for 30 min at 37°C in HKRB with 2.8 mmol/L glucose for stabilization, followed by incubation for 5, 15, or 60 min in HKRB with 2.8 mmol/L or 16.6 mmol/L glucose with and without 10⁻¹⁰ or 10⁻⁹ mol/L exendin (ex)-4 or 10 μmol/L 2-aminoethyl diphenylborinate (2-APB). Secreted insulin concentrations in supernatants of each test batch were determined with ELISA kits (Morinaga Institute of Biological Science, Yokohama, Japan).

Single β-cells were isolated from male C57BL/6J mice and plated on coverslips. Cytoplasmic Ca²⁺ concentration ([Ca²⁺]_i) in β-cells was measured as reported previously (10). Briefly, β-cells were superfused with HKRB at 36°C and [Ca²⁺]_i was measured by dual-wavelength fura-2 microfluorometry with excitation at 340/380 nm and emission at 510 nm by using a cooled charge-coupled device camera. Fluorescence ratio images were produced using an AquaCosmos system (Hamamatsu Photonics, Hamamatsu, Japan). Cells used for single-cell experiments fulfilled the morphological and physiological criteria for insulin-positive β-cells, including the diameter and responsiveness to glucose and tolbutamide. Effects of GLP-1 on [Ca²⁺]_i were investigated exclusively in the cells that responded to glucose with increases in [Ca²⁺]_i in a β-cell–specific manner and to tolbutamide at the end of recording.

Perforated whole-cell currents were recorded using a pipette solution containing amphotericin B (200 μg/mL) dissolved in 0.1% DMSO. Membrane currents, recorded by using an amplifier (Axopatch 200B; Axon, Foster, CA), were stored online in a computer with pCLAMP 10.2 software. Voltage clamp in perforated mode was considered to be adequate when the series resistance was <20 MΩ. Patch pipettes purchased from Narishige (Tokyo, Japan) and their resistances ranged from 3 to 5 MΩ when filled with pipette solution that contained 40 mmol/L K₂SO₄, 50 mmol/L KCl, 5 mmol/L MgCl₂, 0.5 mmol/L EGTA, and 10 mmol/L HEPES at pH 7.2 with KOH. For recording of the background current, β-cells were voltage clamped at a holding potential of −70 or −80 mV. In the presence of 100 μmol/L tolbutamide, which is a sufficient concentration to specifically block K_ATP channels, the residual current is background current corresponding to NSCC conductance. Membrane potential was stepped to test potentials from −100 to −20 mV every 10 s in the presence of 5.6 mmol/L glucose. After recording the control current, the bathing solution (HKRB) was changed to a solution containing test agent.

Measurement of membrane potentials was performed by switching from perforated whole-cell voltage-clamp mode to current-clamp mode. We demonstrated that the voltage-clamped cell was immunostained with anti-insulin antiserum and shown to express insulin-positive fluorescence (11). Electrophysiological experiments were performed at 27°C.

Data are presented as means (SEM) and were compared by Student t test performed with GraphPad Prism, version 5.0. P values of <0.05 were considered statistically significant.

For examination of whether exposure of rat β-cells to GLP-1 depolarizes the membrane and whether the change in membrane potential is a consequence of NSCC activation, membrane potential and whole-cell currents were recorded. During exposure to the GLP-1 receptor agonist, ex-4 at a concentration of 10⁻¹⁰ mol/L at 2.8 mmol/L glucose, the plasma membrane was depolarized (Fig. 1A and B). There has been a report suggesting that K_ATP channels are inhibited by ex-4 (5). To examine effects of ex-4 on NSCC current without influencing changes in activity of the K_ATP channel by ex-4, we voltage clamped the cell at −70 mV, which is close to potassium equilibrium potential, and used tolbutamide to inhibit the K_ATP channel. Under these conditions, ex-4 increased the inward background current in β-cells (Fig. 1C and Supplementary Fig. 1). When the subtracted currents sensitive to ex-4 were plotted against the membrane voltage for the current-voltage (I-V) relationship, the reversal potential was −19.2 mV, and it was shifted to −4.4 mV (Fig. 1D) by omitting external Ca²⁺. Thus, the ex-4–sensitive channel is permeable to Na⁺, K⁺, and Ca²⁺. The reversal potentials were consistent with NSCC reversal potentials in previous reports, in which GLP-1 elicited the NSCC current with reversal potentials ranging from −20 to 0 mV (2,13). The slope conductances in normal and Ca²⁺-free HKRB solutions were 82.3 and 25.2 pA/pF, respectively. These results indicate that the ex-4–sensitive current is in part Ca²⁺ permeable (Fig. 1D). Ex-4 increased the current in a concentration-dependent manner; the effect started at 0.01 nmol/L, peaked at 0.1–1 nmol/L, and declined at ≥10 nmol/L (Fig. 2).

Next, we confirmed that incretin hormone could increase first-phase insulin secretion and [Ca²⁺]_i response to glucose. Insulin secretion measured during 15-min (Fig. 3A-1) or 5-min (Supplementary Fig. 2) static incubation with 16.6 mmol/L glucose was increased by addition of ex-4. Increase in [Ca²⁺]_i is a primary context prior to initiation of insulin secretion. An increase in glucose concentration from 2.8 to 5.6 mmol/L induced first-phase increases in [Ca²⁺]_i, and a subsequent increase in glucose to 8.3 mmol/L induced further increases in [Ca²⁺]_i in an oscillatory pattern. The [Ca²⁺]_i oscillation declined with time during 20 min of exposure to 8.3 mmol/L glucose under control conditions, while it was maintained or even enhanced in the presence of GLP-1 (Fig. 3A-2). GLP-1 significantly increased the area under the curve of [Ca²⁺]_i oscillations (AUC) (Fig. 3A-3).

In order to clarify the mechanistic pathways in which incretin hormone facilitates the opening of NSCC, we tested compounds that mediated effects downstream of receptor stimulation. Both ex-4 and GLP-1, as well as another GLP-1 analog, liraglutide, significantly increased inward current at concentrations of 10⁻¹⁰ mol/L measured at −70 mV in β-cells (Fig. 3B–D). The K⁺ channel currents could be changed minimally, if at all, at this potential. K_ATP channels and voltage-gated potassium channels are inhibited by 100 μmol/L tolbutamide and are voltage-dependently inactivated, respectively. Therefore, throughout the subsequent experiments, we considered the inward current evoked at −70 mV to be that of NSCC. For further confirmation of this hypothesis, the membrane potential was held at −80 mV, which is closer to potassium equilibrium potential (−80 mV based on calculations from external and internal potassium concentrations, 5.9 and 130 mmol/L, respectively, at 27°C). Ex-4 at 10⁻¹⁰ mol/L elicited increases in inward current in the absence of tolbutamide (Supplementary Fig. 3A and B). Tolbutamide was without effect on current increased by ex-4 at 10⁻¹⁰ mol/L in the presence of 5.6 mmol/L glucose and was not able to induce NSCC current at −80 mV (Supplementary Fig. 3C and D). The ex-4–induced NSCC-current increase was attenuated in the presence of GLP-1 receptor antagonist ex-(9-39). This suggests that GLP-1 induces the current increase by interacting with GLP-1 receptor (Fig. 3E [see Supplementary Fig. 4A for current trace]).

There are several pieces of evidence indicating that GLP-1 stimulates glucose-induced insulin secretion via cAMP production (14–16). Consistently, exposure to 1 mmol/L dubutyryl cAMP (dbcAMP) evoked an NSCC current increase (Fig. 3F), and the protein kinase A (PKA) inhibitor H89 (Fig. 3G) did not prevent the increase in NSCC current induced by ex-4. Another potent PKA inhibitor, KT5720 (Supplementary Fig. 4B and C), had no effect on the ex-4–induced current increase in NSCC. The PKA activators 100 μmol/L 6-Benz cAMP (Fig. 3H) and 10 μmol/L 6-Phen cAMP (Fig. 3I) did not influence the amplitude of NSCC current (see Supplementary Fig. 4D and E for original current traces). The ex-4–induced NSCC current increase was mimicked by activator of exchange protein directly activated by cAMP (EPAC) (8-pCPT-2′-O-Me-cAMP [8-pCPT]) (Fig. 3J). An EPAC inhibitor (ESI-09; 3-(5-tert-butyl-isoxazol-3-yl)-2-[(3-chlorophenyl)-hydrazono]-3-oxo-propionitrile) (17) attenuated the ex-4–induced NSCC current increase (Fig. 3K and Supplementary Fig. 5A). Similarly, ex-4 did not increase NSCC current in the presence of RAP1 inhibitor, geranylgeranyltransferase I inhibitor GGTI-298 (Supplementary Fig. 5B) (18). These data suggest that GLP-1/ex-4–sensitive current is regulated by a cAMP/EPAC2 (or EPAC1) pathway and that RAP1 is involved. The ex-4–sensitive current increase was prevented by a nonspecific blocker of TRP channels, ruthenium red (not shown), and TRPM2-channel antagonist (2-APB) (19) (Fig. 3L and Supplementary Fig. 5C). GIP also increased NSCC current (Fig. 3M), and this current increase was attenuated by pretreatment of β-cells with 2-APB (Fig. 3N). GIP dose dependently increased NSCC current at concentration ranges similar to those of ex-4, with a peak effectiveness observed at 0.1 nmol/L GIP (Fig. 2B). The concentrations of GLP-1 or GIP that affected NSCC were in the suprapicomolar range, which is similar to the physiological plasma levels of these hormones after meal intake (20). Insulin secretion stimulated by 16.6 mmol/L glucose was potentiated by the presence of 0.1 or 1 nmol/L ex-4. This effect of ex-4 on the potentiation of glucose-stimulated insulin secretion was inhibited by 2-APB (Fig. 4).

EPAC2 serves as a major pathway for cAMP-mediated potentiation of glucose-stimulated insulin secretion (21). Glucose metabolism elevates cytoplasmic cAMP levels in β-cells (22–24). Hence, we tested whether glucose metabolism influences EPAC-regulated NSCC. We found that an increase in glucose concentration to 16.6 mmol/L reversibly increases the NSCC current, and the I-V relationship showed that the reversal potential and slope conductance were −15.4 mV and 82.6 pA/pF, respectively (Fig. 5A and B). These properties were consistent with those of the NSCC current induced by ex-4. Glucose at 2.8, 5.6, and 16.6 mmol/L dose dependently increased the NSCC current, and addition of ex-4 at each glucose concentration increased the current (Fig. 5C). These effects of glucose on current increases in NSCC were not the result of exposure to tolbutamide, as NSCC current was evoked by increasing glucose concentration to 16.6 mmol/L in the absence of tolbutamide and addition of tolbutamide did not influence the current level (Supplementary Fig. 3E and F). The amplitude of the current increase by ex-4 was greatest at 5.6 mmol/L and was lesser at 16.6 mmol/L glucose. Exposure of β-cells to an uncoupler of electron transport in mitochondria, p-trifluoromethoxyphenylhydrazone, or 2-APB attenuated the glucose-induced NSCC current increase. Sucrose (13.8 mmol/L) did not mimic the high glucose–induced NSCC current increase (Fig. 5D and Supplementary Fig. 6A and B). These results suggest that increased glucose metabolism and GLP-1/GIP receptor stimulation elicit increased activity in the same type of NSCC.

For examination of whether TRPM2 channels are involved in ex-4 and glucose-induced NSCC current increases, effects of these stimuli on NSCC current were studied in TRPM2-deficient mice. Ex-4, high glucose (Supplementary Fig. 7A), GIP, and 8-pCPT increased NSCC current to a degree similar to that seen in wild-type mice (Fig. 6A and Supplementary Figs. 7A and 8A). In TRPM2-KO mice (8,12), these effects were blunted (Fig. 6B and Supplementary Figs. 7B and 8B). The AUC of [Ca²⁺]_i response during exposure to 8.3 mmol/L glucose was attenuated in TRPM2-KO mice (Supplementary Fig. 9). The TRPM2 channel is a major physiological target of glucose metabolism–evoked and incretin hormone–potentiated insulin secretion.

8-pCPT-acetoxymethyl (8-pCPT-AM), an activator of EPAC with little activation effect on PKA at lower micromolar doses (25), enhanced the NSCC current increase at 2.8 mmol/L glucose more potently than ex-4 (Fig. 6C). Increasing glucose to 16.6 mmol/L did not overwhelm 8-cCPT-AM stimulation of NSCC at 2.8 mmol/L. ESI-09 added to 16.6 mmol/L glucose inhibited the increasing effects of glucose on NSCC. Accordingly, the pathway of EPAC activation to TRPM2 channel opening is commonly used by glucose metabolism and GLP-1 in these two important stimuli of insulin secretion.

Costimulation with 16.6 mmol/L glucose and tolbutamide induced greater depolarization than 2.8 mmol/L glucose and tolbutamide (Fig. 6D). The membrane potentials were changed from −66.0 ± 1.6 mV (left white bar) at 2.8 mmol/L glucose to −18.2 ± 1.8 mV (left black bar) at 16.6 mmol/L glucose with tolbutamide (100 μmol/L) and from −67.0 ± 0.6 mV (right white bar) in control at 2.8 mmol/L to −50.4 ± 0.9 mV (right black bar) after addition of tolbutamide. As tolbutamide-induced depolarization per se did not increase activity of the TRPM2 channel (compare amplitudes of white bars in Fig. 6A with corresponding white bars in Fig. 6B), high-dose glucose in the presence of tolbutamide produces greater depolarization than low-dose glucose with tolbutamide, presumably by activating NSCC (TRPM2) as an additional enhancing signal. These results were further confirmed in wild-type and TRPM2-KO mice (Supplementary Fig. 10). We observed that depolarization by tolbutamide was further enhanced after addition of EPAC activator, 8-pCPT-AM, at 2.8 mmol/L glucose in wild-type mice. This additional depolarization by 8-pCPT-AM was attenuated in TRPM2-KO mice.

Both glucose metabolism and GLP-1 receptor stimulation increased the activity of TRPM2 channels via the cAMP-EPAC pathway but not the PKA pathway. As increases in glucose concentration reportedly induce oscillations of cAMP in the cytoplasmic space in β-cells and these oscillations are preceded and potentiated by elevation of [Ca²⁺]_i (24), the activation of the cAMP/EPAC/TRPM2 channel by glucose metabolism may further facilitate glucose-induced depolarization. Although the depolarization initiated by glucose metabolism is believed to simply be a consequence of closure of the K_ATP channel, we now propose that the depolarization is a clear consequence of the simultaneous occurrence of both K_ATP channel closure and TRPM2 channel opening. Ex-4 or the EPAC activator, 8-pCPT-AM, inhibited K_ATP channels in whole-cell and inside-out patch experiments (5). Thus, ex-4 may stimulate β-cells cooperatively by inhibiting K_ATP channels and activating TRPM2 channels.

The glucose-induced membrane depolarization results from not only closure of the K_ATP channel but also opening of TRPM2 channels. Exposure to tolbutamide at 2.8 mmol/L glucose depolarized to −50.4 mV (Fig. 6D, right black bar), whereas it was more depolarized to −18.2 mV at 16.6 mmol/L glucose (left black bar). Most K_ATP channels are opened at 2.8 mmol/L glucose, and a small amount of TRPM2 channels are opened (Fig. 5C). The low-level openings of TRPM2 channels led membrane depolarization by only 17 mV after addition of tolbutamide (right bars in Fig. 6D). In contrast, when we used high glucose concentration (left bars in Fig. 6D), membrane was further depolarized by tolbutamide. As TRPM2 channels are further opened in high glucose as demonstrated in Fig. 5C, simultaneous stimulations with high glucose and tolbutamide are more effective for depolarization than tolbutamide alone at low glucose. In wild-type mice, 8-pCPT-AM further depolarized the membrane after addition of tolbutamide but not in KO mice. Activation of EPAC depolarizes the membrane through increases in activity of TRPM2 channels.

We have observed current increases in TRPM2 channels by ex-4 and high glucose regardless of the presence of tolbutamide. Similarly, amplitudes of currents without ex-4 in wild and KO mice were not different despite tolbutamide being present in both (white bars in Fig. 6A and B). These results suggest that tolbutamide alone did not influence TRPM2 channels at 2.8 mmol/L and 16.6 mmol/L glucose. Recently, EPAC2A was found to be a central mediator in GLP-1–stimulated insulin secretion (26). It is likely that TRPM2 is regulated by EPAC in our study. Interaction of sulfonylurea receptor 1/EPAC2A protein-protein (27–29) and increased EPAC2A activity by sulfonylureas (30) has been reported. PKA-independent potentiation of insulin secretion by cAMP (presumably EPAC2A-dependent pathway) was attenuated in sulfonylurea receptor 1–deficient islets (31). The functional relationship among these proteins, sulfonylureas, and TRPM2 channels remains to be elucidated. The observation that TRPM2 channels could not be activated by ex-4 in the presence of 10 μmol/L GGTI-298 RAP1 inhibitor suggests direct or indirect interaction of TRPM2 channels with RAP1 downstream of EPAC. The reasons for the difference between our results showing ineffectiveness of tolbutamide on TRPM2 channel current (Fig. 6A and B [white bars] and Supplementary Fig. 3) and the increase in activity of EPAC2A by sulfonylureas including tolbutamide (30,32) are unknown. It has been proposed that there are two binding sites for cAMP corresponding to high- and low-affinity sites for cAMP in EPAC2A (30). Sulfonylureas bind to the low-affinity site. If intrinsic cAMP levels produced by ex-4 or glucose are high enough to bind to both sites, tolbutamide can no longer bind to the low-affinity site. Under these conditions, tolbutamide may not influence TRPM2 channels stimulated by ex-4 or glucose. We used tolbutamide at 100 μmol/L, which may be an insufficient concentration to activate TRPM2. EPAC1 does not have a binding site to sulfonylureas (30). Further investigations are warranted.

The changes in TRPM2 channel current demonstrated in the current study might be dependent on indirect or direct protein-protein interactions among K_ATP/EPAC/TRPM2 proteins that combine in modulation of TRPM2 channel currents. The changes were observed under the conditions of little or no K_ATP channel current and therefore are not absolutely dependent on K_ATP channel current. Changes in protein-protein interactions of the K_ATP/EPACs/TRPM2 proteins could contribute to the results observed in the absence of K_ATP channel currents.

ADP ribose, cyclic ADP ribose (cADPR), and H₂O₂ are potent activators of TRPM2 (19,33). GLP-1 reportedly produces cADPR in mouse pancreatic islets (34), suggesting that cADPR plays a key role between EPAC and TRPM2 activation. Although TRPM2 activation by cADPR is dependent on temperature (>35°C) (33) and we performed the present experiments at 27°C, further investigation is required to elucidate whether cADPR is involved in the TRPM2 channel openings demonstrated in the current study.

In the current study, I-V relationship revealed that the GLP-1 and glucose-evoked current reversed at approximately −20 mV, although the I-V relationship of TRPM2 channels expressed in human embryonic kidney (HEK)293 cells showed a reversal at 0 mV in both single-channel and whole-cell analyses (33). The difference between these zero current potentials may be due to the distinct cell types, β-cell and HEK293 cells, and recording mode, perforated whole-cell clamp. HEK293 cells have a small amount of background current and voltage-dependent current system. Thus, in these cells, expressed TRPM2 channels should reverse at 0 mV because of its nonselectivity for cations. Similarly, as the TRPM2 channel also is permeable to Ca²⁺, an increase of Ca²⁺ influx through activated TRPM2 channels may cause an elevated cytosolic Ca²⁺ concentration in a local submembrane area. As a consequence of changes in cytosolic Ca²⁺ concentration due to weak buffer capacity in the perforated whole-cell mode, the Ca²⁺-activated and voltage-gated potassium channel (BK channel) may influence whole-cell current during glucose or GLP-1 stimulation. These effects may shift the reversal potential toward a negative direction upon TRPM2 channel activation by glucose or GLP-1. This hypothesis can be supported by the finding that I-V relations reversed at −4.4 mV in the absence of external Ca²⁺ (Fig. 1D).

The effects of ex-4 and GIP on TRPM2 channel current with regard to dose responsiveness showed a bell-shaped relationship (Fig. 2). The reduced effectiveness at these high doses of ligand can be explained by desensitization of the G-protein–coupled receptor signal (35). However, the effectiveness of the ligands on TRPM2 channel current was not zero, even in the nanomolar range. Thus, desensitization is not complete.

Our results show that glucose metabolism not only closes K_ATP channels but also opens TRPM2 channels and that these pivotal phenomena occur cooperatively. The two channel modulations work in concert to effectively depolarize the plasma membrane, serving as the triggering mechanism (Fig. 7). Stimulation of β-cells by GLP-1 or GIP potentiates glucose-induced TRPM2 channel openings resulting in easier and greater depolarization of the plasma membrane. β-Cells may be primed by the incretin-stimulated TRPM2 activation for glucose-induced insulin secretion, as the time required from the beginning of glucose or ex-4 stimulation to TRPM2 channel opening is rapid (<2 min) (Figs. 1A and 5A). The present findings place TRPM2 in the glucose and G-protein–coupled receptor signaling pathway in β-cells. TRPM2 channels as a novel partner with K_ATP channels are essential for priming, triggering β-cell insulin secretion. The current study, therefore, suggests TRPM2 as a potential target for the treatment of type 2 diabetes.

Acknowledgments. The authors are grateful to Drs. Wilfred Y. Fujimoto (University of Washington) and Akio Yoshida (Tottori University Graduate School of Medicine) for reading the manuscript and providing invaluable comments.

Funding. This work was supported by grants-in-aid for scientific research and priority areas from the Japan Society for the Promotion of Science (JSPS) (24591340 to M.Kak. and 24890219 to M.Y.) and Japanese Diabetes Foundation (to M.Kak. and T.Y.); by grants from the Salt Science Research Foundation, the Pharmacological Research Foundation (Tokyo, Japan) (to K.D.), and Takeda Science Foundation (to K.D. and T.Y.); by grants from the Support Program for Strategic Research Platform for Private University from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan, the Japan Diabetes Foundation, and the Uehara Memorial Foundation; and by a Basic Science Research Award from Sumitomo Foundation (to T.Y.). This work was partly supported by a grant-in-aid for scientific research (B) (23390044) from JSPS, the MEXT-Supported Program for the Strategic Research Foundation at Private Universities 2011–2015 and 2013–2017. This study was subsidized by the Japan Keirin Autorace (JKA) through its promotion funds from KEIRIN RACE (to T.Y.).

Duality of Interest. This work was also supported by an Insulin Research Award from Novo Nordisk (to T.Y.). No other potential conflicts of interest relevant to this article were reported.

Author Contributions. M.Y. researched data; wrote, edited, and reviewed the manuscript; and contributed to discussion. K.D. researched data and contributed to discussion. K.U., S.-eI., and H.S. reviewed and edited the manuscript. S.K., N.V.L., K.I., R.S.R., and H.Y. researched data. K.S. wrote, edited, and reviewed the manuscript. M.Kaw. reviewed and edited the manuscript. M.T. reviewed and edited the manuscript and contributed to discussion. T.Y. wrote and edited the manuscript. M.Kak. wrote and edited the manuscript and contributed to discussion. M.Y., K.D., T.Y., and M.Kak. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.