Masashi Yosida,1 Katsuya Dezaki,2 Kunitoshi Uchida,3 Shiho Kodera,1 Nien V. Lam,1 Kiyonori Ito,1

Rauza S. Rita,2 Hodaka Yamada,1 Kenju Shimomura,2 San-e Ishikawa,1 Hitoshi Sugawara,1

Masanobu Kawakami,1,4 Makoto Tominaga,3 Toshihiko Yada,2,3 and Masafumi Kakei1

Involvement of cAMP/EPAC/TRPM2 Activation in Glucose-and Incretin-Induced InsulinSecretionDiabetes 2014;63:3394–3403 | DOI: 10.2337/db13-1868

In pancreatic b-cells, closure of the ATP-sensitive K+

(KATP) channel is an initial process triggering glucose-stimulated insulin secretion. In addition, constitutiveopening of background nonselective cation channels(NSCCs) is essentially required to effectively evoke de-polarization as a consequence of KATP channel closure.Thus, it is hypothesized that further opening of NSCCfacilitates membrane excitability. We identified a classof NSCC that was activated by exendin (ex)-4, GLP-1, andits analog liraglutide at picomolar levels. This NSCC wasalso activated by increasing the glucose concentration.NSCC activation by glucose and GLP-1 was a conse-quence of the activated cAMP/EPAC-mediated pathwayand was attenuated in TRPM2-deficient mice. The NSCCwas not activated by protein kinase A (PKA) activatorsand was activated by ex-4 in the presence of PKA inhib-itors. These results suggest that glucose- and incretin-activated NSCC (TRPM2) works in concert with closureof the KATP channel to effectively induce membrane de-polarization to initiate insulin secretion. The current studyreveals a new mechanism for regulating electrical excit-ability in b-cells and for mediating the action of glucoseand incretin to evoke insulin secretion, thereby providingan innovative target for the treatment of type 2 diabetes.

It has been long proposed that glucose-stimulated insulinsecretion in pancreatic b-cells is initiated by closure of theATP-sensitive K+ (KATP) channel, followed by membrane

depolarization (1). In theory, closure of the KATP channelis an important process but is not sufficient to induce theshift of the membrane potential toward a threshold level,as membrane potential is determined by the overall ba-lance of outward and inward currents. Modest constitu-tive opening of background inward current throughnonselective cation channels (NSCCs) is crucial to facili-tate depolarization after KATP channel closure (2). Thisidea suggests that further regulated opening of a classof NSCCs may bring about greater depolarization. How-ever, whether glucose metabolism regulates NSCC activityremains unclear.

The incretin hormones, GLP-1 and glucose-dependentinsulinotropic polypeptide (GIP), potentiate insulin secre-tion in association with increased intracellular Ca2+ con-centrations at insulin-secreting glucose concentrations(3–5). GLP-1 fails to increase insulin secretion from theislets of mice deficient in transmembrane receptor poten-tial (TRP) melastatin 2 (TRPM2) (6,7), a type of NSCC,suggesting that the TRPM2 channel is essential forGLP-1–induced potentiation of glucose-stimulated insulinsecretion (8). GLP-1 depolarizes the plasma membrane bythe opening of NSCC in b-cells (2). Several types of NSCC(TRPs) are expressed in insulin-secreting cells (9). Theaims of the current study were to determine 1) the typeof NSCC activation (through TRPM2 or other TRPs) thatis crucial for signaling after stimulation of the incretinreceptor, 2) whether the NSCC is modulated by glucose

1Internal Medicine, Saitama Medical Center, Jichi Medical University, Saitama,Japan2Integrative Physiology, Jichi Medical University, Shimotsuke, Japan3National Institute for Physiological Sciences, Okazaki, Japan4Nerima Hikarigaoka Hospital, Nerima, Japan

Corresponding author: Masafumi Kakei, mkakei@jichi.ac.jp.

Received 11 December 2013 and accepted 3 May 2014.

This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1.

© 2014 by the American Diabetes Association. Readers may use this article aslong as the work is properly cited, the use is educational and not for profit, andthe work is not altered.

3394 Diabetes Volume 63, October 2014

ISLETSTUDIES

http://crossmark.crossref.org/dialog/?doi=10.2337/db13-1868&domain=pdf&date_stamp=2014-09-11mailto:mkakei@jichi.ac.jphttp://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1

metabolism, and 3) the underlying mechanism involved inthese physiologically and clinically fundamental stimuli.

RESEARCH DESIGN AND METHODS

Male Wistar rats and C57BL/6J mice (CLEA Japan, Inc.)were housed in accordance with the institutional guide-lines for animal care in an air-conditioned room witha 12-h light/dark cycle, and food and water were availablead libitum. The study protocol was approved by theinstitutional animal ethics committee. Islets of Langer-hans were isolated by collagenase digestion from maleWistar rats (age 8–12 weeks; weight ~200 g), C57BL/6Jmice, and TRPM2-deficient mice (8 weeks old [20 g])using a previously reported method (10,11). Briefly, ani-mals 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 ina 5 mmol/L CaCl2 solution containing HEPES-addedKrebs-Ringer bicarbonate buffer (HKRB) (129 mmol/LNaCl, 5 mmol/L NaHCO3, 4.7 mmol/L KCl, 1.2 mmol/LKH2PO4, 2 mmol/L CaCl2, 1.2 mmol/L MgSO4, and 10mmol/L HEPES at pH 7.4 with NaOH) directly into thecommon bile duct. The pancreas was incubated at 37°Cfor 15 min in HKRB buffer. Collected islets and subse-quently dispersed cells were used for insulin release andfor electrophysiological experiments, respectively.

TRPM2-knockout (KO) mice were kindly provided byDr. Y. Mori (Kyoto University) (12). TRPM2-KO micewere backcrossed with the C57BL/6J strain for at leastnine generations.

Measurement of Insulin SecretionEach batch of 10 islets was incubated for 30 min at 37°Cin HKRB with 2.8 mmol/L glucose for stabilization, fol-lowed by incubation for 5, 15, or 60 min in HKRB with2.8 mmol/L or 16.6 mmol/L glucose with and without 10210

or 1029 mol/L exendin (ex)-4 or 10 mmol/L 2-aminoethyldiphenylborinate (2-APB). Secreted insulin concentrationsin supernatants of each test batch were determined withELISA kits (Morinaga Institute of Biological Science,Yokohama, Japan).

Measurement of Cytoplasmic Ca2+ ConcentrationSingle b-cells were isolated from male C57BL/6J mice andplated on coverslips. Cytoplasmic Ca2+ concentration([Ca2+]i) in b-cells was measured as reported previously(10). Briefly, b-cells were superfused with HKRB at 36°Cand [Ca2+]i was measured by dual-wavelength fura-2microfluorometry with excitation at 340/380 nm andemission at 510 nm by using a cooled charge-coupleddevice camera. Fluorescence ratio images were producedusing an AquaCosmos system (Hamamatsu Photonics,Hamamatsu, Japan). Cells used for single-cell experimentsfulfilled the morphological and physiological criteria forinsulin-positive b-cells, including the diameter and re-sponsiveness to glucose and tolbutamide. Effects ofGLP-1 on [Ca2+]i were investigated exclusively in the cellsthat responded to glucose with increases in [Ca2+]i in

a b-cell–specific manner and to tolbutamide at the end ofrecording.

Electrophysiological ExperimentsPerforated whole-cell currents were recorded using a pi-pette solution containing amphotericin B (200 mg/mL)dissolved in 0.1% DMSO. Membrane currents, recordedby using an amplifier (Axopatch 200B; Axon, Foster, CA),were stored online in a computer with pCLAMP 10.2software. Voltage clamp in perforated mode was con-sidered to be adequate when the series resistance was,20 MV. Patch pipettes purchased from Narishige(Tokyo, Japan) and their resistances ranged from 3 to 5MV when filled with pipette solution that contained40 mmol/L K2SO4, 50 mmol/L KCl, 5 mmol/L MgCl2,0.5 mmol/L EGTA, and 10 mmol/L HEPES at pH 7.2with KOH. For recording of the background current,b-cells were voltage clamped at a holding potential of270 or 280 mV. In the presence of 100 mmol/L tolbu-tamide, which is a sufficient concentration to specificallyblock KATP channels, the residual current is backgroundcurrent corresponding to NSCC conductance. Membranepotential was stepped to test potentials from 2100 to220 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 performedby switching from perforated whole-cell voltage-clampmode to current-clamp mode. We demonstrated that thevoltage-clamped cell was immunostained with anti-insulinantiserum and shown to express insulin-positive fluores-cence (11). Electrophysiological experiments were per-formed at 27°C.

Statistical AnalysisData are presented as means (SEM) and were comparedby Student t test performed with GraphPad Prism, ver-sion 5.0. P values of ,0.05 were considered statisticallysignificant.

RESULTS

Ex-4 Depolarizes the Plasma Membrane in AssociationWith Increasing Background CurrentFor examination of whether exposure of rat b-cellsto GLP-1 depolarizes the membrane and whether thechange in membrane potential is a consequence ofNSCC activation, membrane potential and whole-cellcurrents were recorded. During exposure to the GLP-1receptor agonist, ex-4 at a concentration of 10210 mol/Lat 2.8 mmol/L glucose, the plasma membrane was depo-larized (Fig. 1A and B). There has been a report suggest-ing that KATP channels are inhibited by ex-4 (5).To examine effects of ex-4 on NSCC current withoutinfluencing changes in activity of the KATP channel byex-4, we voltage clamped the cell at 270 mV, which isclose to potassium equilibrium potential, and used tol-butamide to inhibit the KATP channel. Under these con-ditions, ex-4 increased the inward background current in

diabetes.diabetesjournals.org Yosida and Associates 3395

b-cells (Fig. 1C and Supplementary Fig. 1). When thesubtracted currents sensitive to ex-4 were plottedagainst the membrane voltage for the current-voltage(I-V) relationship, the reversal potential was 219.2 mV,and it was shifted to 24.4 mV (Fig. 1D) by omittingexternal Ca2+. Thus, the ex-4–sensitive channel is perme-able to Na+, K+, and Ca2+. The reversal potentials wereconsistent with NSCC reversal potentials in previousreports, in which GLP-1 elicited the NSCC current withreversal potentials ranging from 220 to 0 mV (2,13). Theslope conductances in normal and Ca2+-free HKRB solu-tions were 82.3 and 25.2 pA/pF, respectively. Theseresults indicate that the ex-4–sensitive current is in partCa2+ permeable (Fig. 1D). Ex-4 increased the current ina concentration-dependent manner; the effect started at0.01 nmol/L, peaked at 0.1–1 nmol/L, and declined at$10 nmol/L (Fig. 2).

Potentiation of First-Phase Insulin Secretion andCytosolic Ca2+ Caused by Increasing NSCC Current viacAMP/EPAC/NSCC Pathway by Ex-4Next, we confirmed that incretin hormone could increasefirst-phase insulin secretion and [Ca2+]i response to glu-cose. Insulin secretion measured during 15-min (Fig. 3A-1)or 5-min (Supplementary Fig. 2) static incubation with16.6 mmol/L glucose was increased by addition of ex-4.Increase in [Ca2+]i is a primary context prior to initiationof insulin secretion. An increase in glucose concentrationfrom 2.8 to 5.6 mmol/L induced first-phase increases in[Ca2+]i, and a subsequent increase in glucose to 8.3 mmol/Linduced further increases in [Ca2+]i in an oscillatory pat-tern. The [Ca2+]i oscillation declined with time during20 min of exposure to 8.3 mmol/L glucose under controlconditions, while it was maintained or even enhanced in thepresence of GLP-1 (Fig. 3A-2). GLP-1 significantly increased

Figure 1—Depolarization of the membrane potential in association with increased background current induced by ex-4. A: The membranepotential of a b-cell isolated from rat islets was recorded at 2.8 mmol/L glucose. The membrane was depolarized upon superfusion with10210 mol/L ex-4 in a reversible manner. B: Comparison of resting membrane potentials in the absence and presence of 10210 mol/L ex-4.Glucose concentration was 2.8 mmol/L. Membrane potentials in the absence and presence of ex-4 were264.2 6 1.3 mV and258.46 2.1mV, respectively. *P < 0.01, n = 9. C: Perforated whole-cell clamp experiment performed in the absence (control) and presence of 10210

mol/L ex-4 showed that ex-4 increased inward currents (see subtracted currents that were current components increased by ex-4). D:Mean current levels during test pulses of subtracted current traces were measured and plotted against corresponding voltage. The I-Vrelationship showed a slope conductance of 82.3 pA/pF (95% CI 67.3–97.3) and a reversal potential of 219.2 mV (225.2 to 211.0, n = 6[closed circles]) in the control; in the absence of Ca2+, the slope conductance was 25.2 pA/pF (12.0–38.4) and the reversal potential was24.4 mV (222.5 to 48.5, n = 6 [open circles]). Lines were drawn by using linear regression fit to the data points. Data are expressed asmean 6 SEM. Tolbutamide at 100 mmol/L was superfused in the experiments shown in C and D to inhibit KATP channels.

3396 TRPM2 Signal via cAMP/EPAC in Insulin Secretion Diabetes Volume 63, October 2014

http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1

the area under the curve of [Ca2+]i oscillations (AUC)(Fig. 3A-3).

In order to clarify the mechanistic pathways in whichincretin hormone facilitates the opening of NSCC, we

tested compounds that mediated effects downstreamof receptor stimulation. Both ex-4 and GLP-1, as well asanother GLP-1 analog, liraglutide, significantly increasedinward current at concentrations of 10210 mol/L measured

Figure 2—Dose-dependent increases of ex-4– (A) and GIP-sensitive (B) NSCC currents. The current was recorded according to the sameprotocol demonstrated in Fig. 1C and D. The membrane potential was held at 270 mV and the glucose concentration was 5.6 mmol/Lthroughout the experiments. The current was expressed by the current density (pA/pF). Data were collected from five to nine data points ineach experiment.

Figure 3—Ex-4 (10210 mol/L) increases the first-phase insulin secretion evoked by 16.6 mmol/L glucose with increased Ca2+ influx andincreased NSCC current via the cAMP/EPAC/NSCC pathway. A-1: Ex-4 increased the insulin secretion at 16.6 mmol/L glucose. *P < 0.05vs. 16.6 mmol/L glucose without ex-4. A-2: GLP-1 at 10210 mol/L increased the duration of Ca2+ oscillations. At the end of the experiment,100 mmol/L tolbutamide (Tolb) was added to confirm that the responsive cells were b-cells. G, glucose (mmol/L). A-3: AUC of oscillatoryCa2+ responses at 8.3 mmol/L glucose was increased by GLP-1. The results (A-2 and -3) were from mouse b-cells. HKRB solution with0.1% BSA was used for insulin secretion or Ca2+ measurements. B–N: Effects of the following agents on current are shown. GLP-1 (B),liraglutide (C), and ex-4 (D) at 100 pmol/L; ex-(9-39) (0.1 mmol/L) (E); dubutyryl cAMP (dbcAMP) (1 mmol/L) (F ); H89 (10 mmol/L) (G); PKAactivators 6-Benz cAMP (100 mmol/L) (H) and 6-Phen cAMP (10 mmol/L) (I); EPAC activator 8-pCPT (10 mmol/L) (J); EPAC inhibitor ESI-09(10 mmol/L) (K); TRPM2 inhibitor 2-APB (10 mmol/L) (L and N); and GIP (10210 mol/L) (M). In B–D, F, H–J, M, and N, effects of the indicatedagents on NSCC are shown. In E, G, K, and L, effects of 10210 mol/L ex-4 on NSCC in the presence of the indicated agents are illustrated.Rat islet b-cells were used and 100 mmol/L tolbutamide was present throughout the experiments of B–N. The number of data points was sixto nine. *P < 0.05 vs. control (cont) by paired test.

diabetes.diabetesjournals.org Yosida and Associates 3397

at 270 mV in b-cells (Fig. 3B–D). The K+ channel currentscould be changed minimally, if at all, at this potential. KATPchannels and voltage-gated potassium channels are inhibitedby 100 mmol/L tolbutamide and are voltage-dependentlyinactivated, respectively. Therefore, throughout the sub-sequent experiments, we considered the inward currentevoked at 270 mV to be that of NSCC. For further con-firmation of this hypothesis, the membrane potential washeld at 280 mV, which is closer to potassium equilibriumpotential (280 mV based on calculations from externaland internal potassium concentrations, 5.9 and 130mmol/L, respectively, at 27°C). Ex-4 at 10210 mol/L elic-ited increases in inward current in the absence of tolbu-tamide (Supplementary Fig. 3A and B). Tolbutamide waswithout effect on current increased by ex-4 at 10210 mol/Lin the presence of 5.6 mmol/L glucose and was not able toinduce NSCC current at 280 mV (Supplementary Fig. 3Cand D). The ex-4–induced NSCC-current increase wasattenuated in the presence of GLP-1 receptor antagonistex-(9-39). This suggests that GLP-1 induces the currentincrease by interacting with GLP-1 receptor (Fig. 3E [seeSupplementary Fig. 4A for current trace]).

There are several pieces of evidence indicating thatGLP-1 stimulates glucose-induced insulin secretion viacAMP production (14–16). Consistently, exposure to1 mmol/L dubutyryl cAMP (dbcAMP) evoked an NSCCcurrent increase (Fig. 3F), and the protein kinase A(PKA) inhibitor H89 (Fig. 3G) did not prevent the increasein NSCC current induced by ex-4. Another potent PKAinhibitor, KT5720 (Supplementary Fig. 4B and C), hadno effect on the ex-4–induced current increase in NSCC.The PKA activators 100 mmol/L 6-Benz cAMP (Fig. 3H)and 10 mmol/L 6-Phen cAMP (Fig. 3I) did not influencethe amplitude of NSCC current (see SupplementaryFig. 4D and E for original current traces). The ex-4–in-duced NSCC current increase was mimicked by activatorof exchange protein directly activated by cAMP (EPAC)(8-pCPT-29-O-Me-cAMP [8-pCPT]) (Fig. 3J). An EPACinhibitor (ESI-09; 3-(5-tert-butyl-isoxazol-3-yl)-2-[(3-chlorophenyl)-hydrazono]-3-oxo-propionitrile) (17) atten-uated the ex-4–induced NSCC current increase (Fig. 3Kand Supplementary Fig. 5A). Similarly, ex-4 did not in-crease NSCC current in the presence of RAP1 inhibitor,geranylgeranyltransferase I inhibitor GGTI-298 (Supple-mentary Fig. 5B) (18). These data suggest that GLP-1/ex-4–sensitive current is regulated by a cAMP/EPAC2 (orEPAC1) pathway and that RAP1 is involved. The ex-4–sensitive current increase was prevented by a nonspecificblocker of TRP channels, ruthenium red (not shown), andTRPM2-channel antagonist (2-APB) (19) (Fig. 3L and Sup-plementary Fig. 5C). GIP also increased NSCC current (Fig.3M), and this current increase was attenuated by pretreat-ment of b-cells with 2-APB (Fig. 3N). GIP dose dependentlyincreased NSCC current at concentration ranges similar tothose of ex-4, with a peak effectiveness observed at 0.1nmol/L GIP (Fig. 2B). The concentrations of GLP-1 or GIPthat affected NSCC were in the suprapicomolar range,

which is similar to the physiological plasma levels of thesehormones after meal intake (20). Insulin secretion stim-ulated by 16.6 mmol/L glucose was potentiated by thepresence of 0.1 or 1 nmol/L ex-4. This effect of ex-4 onthe potentiation of glucose-stimulated insulin secretionwas inhibited by 2-APB (Fig. 4).

NSCC Activation Is Evoked by Glucose Metabolismand Is Potentiated by Ex-4EPAC2 serves as a major pathway for cAMP-mediatedpotentiation of glucose-stimulated insulin secretion (21).Glucose metabolism elevates cytoplasmic cAMP levels inb-cells (22–24). Hence, we tested whether glucose metab-olism influences EPAC-regulated NSCC. We found that anincrease in glucose concentration to 16.6 mmol/L revers-ibly increases the NSCC current, and the I-V relationshipshowed that the reversal potential and slope conductancewere 215.4 mV and 82.6 pA/pF, respectively (Fig. 5A andB). These properties were consistent with those of theNSCC current induced by ex-4. Glucose at 2.8, 5.6, and16.6 mmol/L dose dependently increased the NSCC cur-rent, and addition of ex-4 at each glucose concentrationincreased the current (Fig. 5C). These effects of glucose oncurrent increases in NSCC were not the result of exposureto tolbutamide, as NSCC current was evoked by increasingglucose concentration to 16.6 mmol/L in the absence oftolbutamide and addition of tolbutamide did not influ-ence the current level (Supplementary Fig. 3E and F).The amplitude of the current increase by ex-4 was great-est at 5.6 mmol/L and was lesser at 16.6 mmol/L glucose.Exposure of b-cells to an uncoupler of electron transportin mitochondria, p-trifluoromethoxyphenylhydrazone, or

Figure 4—Glucose-induced insulin secretion at 11.2 mmol/L and16.6 mmol/L and potentiation by ex-4 followed by prevention in thesimultaneous presence of 2-APB. Batches containing 10 islets ineach were incubated for 1 h to evoke insulin secretion at indicatedglucose concentrations. Ex-4 potentiated insulin secretion at 0.1nmol/L and 1 nmol/L. Further addition of 2-APB at 10 mmol/Linhibited ex-4–induced potentiation of insulin secretion. **P <0.01 vs. 16.6 mmol/L glucose and #P< 0.05 vs. 2-APB by unpairedtest.

3398 TRPM2 Signal via cAMP/EPAC in Insulin Secretion Diabetes Volume 63, October 2014

http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1

2-APB attenuated the glucose-induced NSCC current in-crease. Sucrose (13.8 mmol/L) did not mimic the highglucose–induced NSCC current increase (Fig. 5D and Sup-plementary Fig. 6A and B). These results suggest thatincreased glucose metabolism and GLP-1/GIP receptorstimulation elicit increased activity in the same type ofNSCC.

Attenuation of Glucose-, Ex-4–, and GIP-ElicitedIncrease in NSCC Current in b-Cells From TRPM2-KOMiceFor examination of whether TRPM2 channels are involvedin ex-4 and glucose-induced NSCC current increases, effectsof 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 sim-ilar to that seen in wild-type mice (Fig. 6A and Supplemen-tary Figs. 7A and 8A). In TRPM2-KO mice (8,12), theseeffects were blunted (Fig. 6B and Supplementary Figs. 7Band 8B). The AUC of [Ca2+]i response during exposure to8.3 mmol/L glucose was attenuated in TRPM2-KO mice

(Supplementary Fig. 9). The TRPM2 channel is a majorphysiological target of glucose metabolism–evoked andincretin hormone–potentiated insulin secretion.

8-pCPT-acetoxymethyl (8-pCPT-AM), an activator ofEPAC with little activation effect on PKA at lowermicromolar doses (25), enhanced the NSCC current in-crease at 2.8 mmol/L glucose more potently than ex-4(Fig. 6C). Increasing glucose to 16.6 mmol/L did not over-whelm 8-cCPT-AM stimulation of NSCC at 2.8 mmol/L.ESI-09 added to 16.6 mmol/L glucose inhibited the in-creasing effects of glucose on NSCC. Accordingly, thepathway of EPAC activation to TRPM2 channel openingis commonly used by glucose metabolism and GLP-1 inthese two important stimuli of insulin secretion.

Costimulation with 16.6 mmol/L glucose and tolbuta-mide induced greater depolarization than 2.8 mmol/Lglucose and tolbutamide (Fig. 6D). The membrane poten-tials were changed from 266.0 6 1.6 mV (left white bar)at 2.8 mmol/L glucose to 218.2 6 1.8 mV (left black bar)at 16.6 mmol/L glucose with tolbutamide (100 mmol/L)

Figure 5—NSCC activation is evoked by glucose metabolism and potentiated by ex-4. A: The holding current at270 mV was increased bychanging the glucose concentration from 2.8 mmol/L to 16.6 mmol/L in the continuous presence of 100 mmol/L tolbutamide (Tolb). Thecurrent recording was interrupted as revealed by IV-a and IV-b for recording I-V relationship. The inwardly increased current with high-doseglucose was reversed when the glucose concentration was changed back to 2.8 mmol/L. Current traces indicated in IV-a and IV-b wereevoked by the same voltage protocol described in the legend to Fig. 1. Subtracted current traces are depicted (IV-b minus IV-a; IV-c).Current zero level is indicated by 0 with a short bar above the continuous current trace. B: The I-V relations of subtracted currents wereplotted by measuring mean subtracted currents as mentioned in the legend to Fig. 1D. The slope conductance was 82.6 pA/pF (95% CI65.3–99.9, n = 6), and the zero current potential was 215.4 mV (222.6 to 24.86, n = 6). C: The glucose-induced NSCC increase wasglucose-concentration dependent, and addition of 10210 mol/L ex-4 increased the current. D: Glucose-induced inward current wasattenuated by the presence of p-trifluoromethoxyphenylhydrazone (FCCP) (1 mmol/L) and 2-APB (10 mmol/L). Replacement of glucosewith 13.8 mmol/L sucrose (Sc) did not mimic the glucose-induced inward current increase. Experiments were performed in rat b-cells. *P <0.05. The number of experiments was six to nine. In C and D, 100 mmol/L tolbutamide was present.

diabetes.diabetesjournals.org Yosida and Associates 3399

http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1

and from 267.0 6 0.6 mV (right white bar) in control at2.8 mmol/L to 250.4 6 0.9 mV (right black bar) afteraddition of tolbutamide. As tolbutamide-induced depolar-ization per se did not increase activity of the TRPM2 chan-nel (compare amplitudes of white bars in Fig. 6A withcorresponding white bars in Fig. 6B), high-dose glucose inthe presence of tolbutamide produces greater depolariza-tion than low-dose glucose with tolbutamide, presumablyby activating NSCC (TRPM2) as an additional enhancingsignal. These results were further confirmed in wild-typeand TRPM2-KO mice (Supplementary Fig. 10). We ob-served that depolarization by tolbutamide was further en-hanced after addition of EPAC activator, 8-pCPT-AM, at2.8 mmol/L glucose in wild-type mice. This additional de-polarization by 8-pCPT-AM was attenuated in TRPM2-KOmice.

DISCUSSION

Both glucose metabolism and GLP-1 receptor stimulationincreased the activity of TRPM2 channels via the cAMP-

EPAC pathway but not the PKA pathway. As increases inglucose concentration reportedly induce oscillations ofcAMP in the cytoplasmic space in b-cells and these oscil-lations are preceded and potentiated by elevation of [Ca2+]i(24), the activation of the cAMP/EPAC/TRPM2 channelby glucose metabolism may further facilitate glucose-induced depolarization. Although the depolarization initi-ated by glucose metabolism is believed to simply bea consequence of closure of the KATP channel, we nowpropose that the depolarization is a clear consequence ofthe simultaneous occurrence of both KATP channel closureand TRPM2 channel opening. Ex-4 or the EPAC activa-tor, 8-pCPT-AM, inhibited KATP channels in whole-celland inside-out patch experiments (5). Thus, ex-4 maystimulate b-cells cooperatively by inhibiting KATP chan-nels and activating TRPM2 channels.

The glucose-induced membrane depolarization resultsfrom not only closure of the KATP channel but also open-ing of TRPM2 channels. Exposure to tolbutamide at 2.8mmol/L glucose depolarized to 250.4 mV (Fig. 6D, right

Figure 6—Attenuation of glucose-, ex-4–, and GIP-induced increases in NSCC current in b-cells from TRPM2-KO mice. A: Ex-4 (10210

mol/L), 16.6 mmol/L glucose, GIP (10210 mol/), and 8-pCPT (10 mmol/L) elicited NSCC current increase in wild-type (Wt) b-cells. B: NSCCcurrent increases by the compounds described above were not observed in TRPM2-KOmice. C: NSCC current was maximally activated by10 mmol/L 8-pCPT-AM at 2.8 mmol/L glucose to a degree similar to that evoked by 16.6 mmol/L glucose, and 10 mmol/L ESI-09 attenuatedthe current increase with high glucose. Membrane potential was held at 270 mV, and tolbutamide at 100 mmol/L was continuously presentwhen the current was compared between control and tested compounds in A–C. D: High glucose concentration (16.6 mmol/L) comparedwith 2.8 mmol/L glucose induces more depolarization despite the presence of the same concentration of tolbutamide (tolb) (100 mmol/L).Nitrendipine at 10 mmol/L was added to suppress the voltage-dependent Ca2+ channels for stable recording of resting membrane potentialwithout induction of electrical firing. *P < 0.05 by unpaired test. Number of experiments was six to nine. In C and D, experiments wereperformed in rat b-cells.

3400 TRPM2 Signal via cAMP/EPAC in Insulin Secretion Diabetes Volume 63, October 2014

http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1

black bar), whereas it was more depolarized to 218.2 mVat 16.6 mmol/L glucose (left black bar). Most KATP chan-nels are opened at 2.8 mmol/L glucose, and a smallamount of TRPM2 channels are opened (Fig. 5C). Thelow-level openings of TRPM2 channels led membrane de-polarization by only 17 mV after addition of tolbutamide(right bars in Fig. 6D). In contrast, when we used highglucose concentration (left bars in Fig. 6D), membranewas further depolarized by tolbutamide. As TRPM2 chan-nels are further opened in high glucose as demonstratedin Fig. 5C, simultaneous stimulations with high glucoseand tolbutamide are more effective for depolarization thantolbutamide alone at low glucose. In wild-type mice,8-pCPT-AM further depolarized the membrane after ad-dition of tolbutamide but not in KO mice. Activation ofEPAC depolarizes the membrane through increases in ac-tivity of TRPM2 channels.

We have observed current increases in TRPM2 chan-nels by ex-4 and high glucose regardless of the presence oftolbutamide. Similarly, amplitudes of currents withoutex-4 in wild and KO mice were not different despitetolbutamide being present in both (white bars in Fig. 6Aand B). These results suggest that tolbutamide alonedid not influence TRPM2 channels at 2.8 mmol/L and16.6 mmol/L glucose. Recently, EPAC2A was found tobe a central mediator in GLP-1–stimulated insulin secre-tion (26). It is likely that TRPM2 is regulated by EPAC inour study. Interaction of sulfonylurea receptor 1/EPAC2Aprotein-protein (27–29) and increased EPAC2A activity bysulfonylureas (30) has been reported. PKA-independentpotentiation of insulin secretion by cAMP (presumablyEPAC2A-dependent pathway) was attenuated in sulfonyl-urea receptor 1–deficient islets (31). The functional rela-tionship among these proteins, sulfonylureas, and TRPM2channels remains to be elucidated. The observation thatTRPM2 channels could not be activated by ex-4 in thepresence of 10 mmol/L GGTI-298 RAP1 inhibitor suggestsdirect or indirect interaction of TRPM2 channels withRAP1 downstream of EPAC. The reasons for the differ-ence between our results showing ineffectiveness of tol-butamide on TRPM2 channel current (Fig. 6A and B[white bars] and Supplementary Fig. 3) and the increasein activity of EPAC2A by sulfonylureas including tolbuta-mide (30,32) are unknown. It has been proposed thatthere are two binding sites for cAMP corresponding tohigh- and low-affinity sites for cAMP in EPAC2A (30).Sulfonylureas bind to the low-affinity site. If intrinsiccAMP levels produced by ex-4 or glucose are high enoughto bind to both sites, tolbutamide can no longer bind tothe low-affinity site. Under these conditions, tolbutamidemay not influence TRPM2 channels stimulated by ex-4 orglucose. We used tolbutamide at 100 mmol/L, which maybe 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 demon-strated in the current study might be dependent on

indirect or direct protein-protein interactions amongKATP/EPAC/TRPM2 proteins that combine in modula-tion of TRPM2 channel currents. The changes were ob-served under the conditions of little or no KATP channelcurrent and therefore are not absolutely dependent onKATP channel current. Changes in protein-protein inter-actions of the KATP/EPACs/TRPM2 proteins could con-tribute to the results observed in the absence of KATPchannel currents.

ADP ribose, cyclic ADP ribose (cADPR), and H2O2 arepotent activators of TRPM2 (19,33). GLP-1 reportedlyproduces cADPR in mouse pancreatic islets (34), suggest-ing that cADPR plays a key role between EPAC andTRPM2 activation. Although TRPM2 activation by cADPRis dependent on temperature (.35°C) (33) and we per-formed the present experiments at 27°C, further investi-gation is required to elucidate whether cADPR is involvedin the TRPM2 channel openings demonstrated in the cur-rent study.

In the current study, I-V relationship revealed that theGLP-1 and glucose-evoked current reversed at approxi-mately 220 mV, although the I-V relationship of TRPM2channels expressed in human embryonic kidney (HEK)293 cells showed a reversal at 0 mV in both single-channeland whole-cell analyses (33). The difference betweenthese zero current potentials may be due to the distinctcell types, b-cell and HEK293 cells, and recording mode,perforated whole-cell clamp. HEK293 cells have a smallamount of background current and voltage-dependentcurrent system. Thus, in these cells, expressed TRPM2channels should reverse at 0 mV because of its nonselec-tivity for cations. Similarly, as the TRPM2 channel also ispermeable to Ca2+, an increase of Ca2+ influx throughactivated TRPM2 channels may cause an elevated cyto-solic Ca2+ concentration in a local submembrane area.As a consequence of changes in cytosolic Ca2+ concentra-tion due to weak buffer capacity in the perforated whole-cell mode, the Ca2+-activated and voltage-gated potassiumchannel (BK channel) may influence whole-cell currentduring glucose or GLP-1 stimulation. These effects mayshift the reversal potential toward a negative directionupon TRPM2 channel activation by glucose or GLP-1.This hypothesis can be supported by the finding that I-Vrelations reversed at 24.4 mV in the absence of externalCa2+ (Fig. 1D).

The effects of ex-4 and GIP on TRPM2 channel currentwith regard to dose responsiveness showed a bell-shapedrelationship (Fig. 2). The reduced effectiveness at thesehigh doses of ligand can be explained by desensitization ofthe G-protein–coupled receptor signal (35). However, theeffectiveness of the ligands on TRPM2 channel currentwas not zero, even in the nanomolar range. Thus, desen-sitization is not complete.

Our results show that glucose metabolism not onlycloses KATP channels but also opens TRPM2 channelsand that these pivotal phenomena occur cooperatively.The two channel modulations work in concert to

diabetes.diabetesjournals.org Yosida and Associates 3401

http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db13-1868/-/DC1

effectively depolarize the plasma membrane, serving asthe triggering mechanism (Fig. 7). Stimulation of b-cellsby GLP-1 or GIP potentiates glucose-induced TRPM2channel openings resulting in easier and greater depolar-ization of the plasma membrane. b-Cells may be primedby the incretin-stimulated TRPM2 activation for glucose-induced insulin secretion, as the time required from thebeginning of glucose or ex-4 stimulation to TRPM2 channelopening is rapid (,2 min) (Figs. 1A and 5A). The presentfindings place TRPM2 in the glucose and G-protein–coupledreceptor signaling pathway in b-cells. TRPM2 channels as anovel partner with KATP channels are essential for priming,triggering b-cell insulin secretion. The current study, there-fore, suggests TRPM2 as a potential target for the treatmentof type 2 diabetes.

Acknowledgments. The authors are grateful to Drs. Wilfred Y. Fujimoto(University of Washington) and Akio Yoshida (Tottori University Graduate Schoolof Medicine) for reading the manuscript and providing invaluable comments.Funding. This work was supported by grants-in-aid for scientific researchand 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, thePharmacological Research Foundation (Tokyo, Japan) (to K.D.), and Takeda Sci-ence Foundation (to K.D. and T.Y.); by grants from the Support Program forStrategic Research Platform for Private University from the Ministry of Education,Culture, Sports, Science and Technology (MEXT) of Japan, the Japan DiabetesFoundation, and the Uehara Memorial Foundation; and by a Basic Science Re-search Award from Sumitomo Foundation (to T.Y.). This work was partly sup-ported by a grant-in-aid for scientific research (B) (23390044) from JSPS, theMEXT-Supported Program for the Strategic Research Foundation at Private Uni-versities 2011–2015 and 2013–2017. This study was subsidized by the JapanKeirin Autorace (JKA) through its promotion funds from KEIRIN RACE (to T.Y.).Duality of Interest. This work was also supported by an Insulin ResearchAward from Novo Nordisk (to T.Y.). No other potential conflicts of interest relevantto this article were reported.

Author Contributions. M.Y. researched data; wrote, edited, and re-viewed the manuscript; and contributed to discussion. K.D. researched dataand contributed to discussion. K.U., S.-eI., and H.S. reviewed and edited themanuscript. 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 manu-script. M.T. reviewed and edited the manuscript and contributed to discussion.T.Y. wrote and edited the manuscript. M.Kak. wrote and edited the manuscriptand contributed to discussion. M.Y., K.D., T.Y., and M.Kak. are the guarantors ofthis work and, as such, had full access to all the data in the study and takeresponsibility for the integrity of the data and the accuracy of the data analysis.

References1. Ashcroft FM, Rorsman P. Diabetes mellitus and the b cell: the last ten

years. Cell 2012;148:1160–11712. Leech CA, Habener JF. A role for Ca2+-sensitive nonselective cation

channels in regulating the membrane potential of pancreatic beta-cells. Diabetes

1998;47:1066–10733. Yada T, Itoh K, Nakata M. Glucagon-like peptide-1-(7-36)amide and a rise

in cyclic adenosine 39,59-monophosphate increase cytosolic free Ca2+ in rat

pancreatic beta-cells by enhancing Ca2+ channel activity. Endocrinology 1993;

133:1685–16924. Gromada J, Brock B, Schmitz O, Rorsman P. Glucagon-like peptide-1:

regulation of insulin secretion and therapeutic potential. Basic Clin Pharmacol

Toxicol 2004;95:252–2625. Leech CA, Dzhura I, Chepurny OG, et al. Molecular physiology of glucagon-

like peptide-1 insulin secretagogue action in pancreatic b cells. Prog Biophys Mol

Biol 2011;107:236–2476. Eisfeld J, Lückhoff A. Trpm2. Handbook Exp Pharmacol 2007;237–2527. Colsoul B, Vennekens R, Nilius B. Transient receptor potential cation

channels in pancreatic b cells. Rev Physiol Biochem Pharmacol 2011;161:87–

1108. Uchida K, Dezaki K, Damdindorj B, et al. Lack of TRPM2 impaired insulin

secretion and glucose metabolisms in mice. Diabetes 2011;60:119–1269. Uchida K, Tominaga M. The role of thermosensitive TRP (transient receptor

potential) channels in insulin secretion. Endocr J 2011;58:1021–102810. Dezaki K, Damdindorj B, Sone H, et al. Ghrelin attenuates cAMP-PKA sig-

naling to evoke insulinostatic cascade in islet b-cells. Diabetes 2011;60:2315–

2324

Figure 7—Schematic model of GLP-1/GIP-induced opening of TRPM2 channels and cooperative facilitation of depolarization with glucose-induced KATP channel closure.

3402 TRPM2 Signal via cAMP/EPAC in Insulin Secretion Diabetes Volume 63, October 2014

11. Yoshida M, Dezaki K, Yamato S, et al. Regulation of voltage-gated K+

channels by glucose metabolism in pancreatic beta-cells. FEBS Lett 2009;583:2225–223012. Yamamoto S, Shimizu S, Kiyonaka S, et al. TRPM2-mediated Ca2+influxinduces chemokine production in monocytes that aggravates inflammatoryneutrophil infiltration. Nat Med 2008;14:738–74713. Holz GG 4th, Leech CA, Habener JF. Activation of a cAMP-regulatedCa(2+)-signaling pathway in pancreatic beta-cells by the insulinotropichormone glucagon-like peptide-1. J Biol Chem 1995;270:17749–1775714. Damdindorj B, Dezaki K, Kurashina T, et al. Exogenous and endogenousghrelin counteracts GLP-1 action to stimulate cAMP signaling and insulin se-cretion in islet b-cells. FEBS Lett 2012;586:2555–256215. Holz GG, Heart E, Leech CA. Synchronizing Ca2+ and cAMP oscillations inpancreatic beta-cells: a role for glucose metabolism and GLP-1 receptors? Focuson “regulation of cAMP dynamics by Ca2+ and G protein-coupled receptors in thepancreatic beta-cell: a computational approach”. Am J Physiol Cell Physiol 2008;294:C4–C616. Shigeto M, Katsura M, Matsuda M, Ohkuma S, Kaku K. Low, but physio-logical, concentration of GLP-1 stimulates insulin secretion independent of thecAMP-dependent protein kinase pathway. J Pharmacol Sci 2008;108:274–27917. Almahariq M, Tsalkova T, Mei FC, et al. A novel EPAC-specific inhibitorsuppresses pancreatic cancer cell migration and invasion. Mol Pharmacol 2013;83:122–12818. Kogut MH, Genovese KJ, He H. Flagellin and lipopolysaccharide stimulatethe MEK-ERK signaling pathway in chicken heterophils through differential ac-tivation of the small GTPases, Ras and Rap1. Mol Immunol 2007;44:1729–173619. Togashi K, Inada H, Tominaga M. Inhibition of the transient receptor po-tential cation channel TRPM2 by 2-aminoethoxydiphenyl borate (2-APB). Br JPharmacol 2008;153:1324–133020. Holst JJ. The physiology of glucagon-like peptide 1. Physiol Rev 2007;87:1409–143921. Shibasaki T, Takahashi H, Miki T, et al. Essential role of Epac2/Rap1 sig-naling in regulation of insulin granule dynamics by cAMP. Proc Natl Acad Sci U S A2007;104:19333–1933822. Sharp GW. The adenylate cyclase-cyclic AMP system in islets of Langerhansand its role in the control of insulin release. Diabetologia 1979;16:287–296

23. Tian G, Sandler S, Gylfe E, Tengholm A. Glucose- and hormone-inducedcAMP oscillations in a- and b-cells within intact pancreatic islets. Diabetes2011;60:1535–154324. Dyachok O, Idevall-Hagren O, Sågetorp J, et al. Glucose-induced cyclic AMPoscillations regulate pulsatile insulin secretion. Cell Metab 2008;8:26–3725. Chepurny OG, Kelley GG, Dzhura I, et al. PKA-dependent potentiation ofglucose-stimulated insulin secretion by Epac activator 8-pCPT-29-O-Me-cAMP-AM in human islets of Langerhans. Am J Physiol Endocrinol Metab 2010;298:E622–E63326. Song WJ, Mondal P, Li Y, Lee SE, Hussain MA. Pancreatic b-cell responseto increased metabolic demand and to pharmacologic secretagogues requiresEPAC2A. Diabetes 2013;62:2796–280727. Eliasson L, Ma X, Renström E, et al. SUR1 regulates PKA-independentcAMP-induced granule priming in mouse pancreatic B-cells. J Gen Physiol 2003;121:181–19728. Holz GG. Epac: A new cAMP-binding protein in support of glucagon-likepeptide-1 receptor-mediated signal transduction in the pancreatic beta-cell.Diabetes 2004;53:5–1329. Seino S. Cell signalling in insulin secretion: the molecular targets of ATP,cAMP and sulfonylurea. Diabetologia 2012;55:2096–210830. Takahashi T, Shibasaki T, Takahashi H, et al. Antidiabetic sulfonylureas andcAMP cooperatively activate Epac2A. Sci Signal 2013;6:ra9431. Nakazaki M, Crane A, Hu M, et al. cAMP-activated protein kinase-independentpotentiation of insulin secretion by cAMP is impaired in SUR1 null islets. Diabetes2002;51:3440–344932. Zhang CL, Katoh M, Shibasaki T, et al. The cAMP sensor Epac2 is a directtarget of antidiabetic sulfonylurea drugs. Science 2009;325:607–61033. Togashi K, Hara Y, Tominaga T, et al. TRPM2 activation by cyclic ADP-riboseat body temperature is involved in insulin secretion. EMBO J 2006;25:1804–181534. Kim JW, Roberts CD, Berg SA, Caicedo A, Roper SD, Chaudhari N. Imagingcyclic AMP changes in pancreatic islets of transgenic reporter mice. PLoS ONE2008;3:e212735. Takeda Y, Amano A, Noma A, Nakamura Y, Fujimoto S, Inagaki N. Systemsanalysis of GLP-1 receptor signaling in pancreatic b-cells. Am J Physiol CellPhysiol 2011;301:C792–C803

diabetes.diabetesjournals.org Yosida and Associates 3403