NMDA receptors mediate leptin signaling and regulate potassium channel trafficking in pancreatic β-cells

NMDA receptors (NMDARs) are Ca2+-permeant, ligand-gated ion channels activated by the excitatory neurotransmitter glutamate and have well-characterized roles in the nervous system. The expression and function of NMDARs in pancreatic β-cells, by contrast, are poorly understood. Here, we report a novel function of NMDARs in β-cells. Using a combination of biochemistry, electrophysiology, and imaging techniques, we now show that NMDARs have a key role in mediating the effect of leptin to modulate β-cell electrical activity by promoting AMP-activated protein kinase (AMPK)-dependent trafficking of KATP and Kv2.1 channels to the plasma membrane. Blocking NMDAR activity inhibited the ability of leptin to activate AMPK, induce KATP and Kv2.1 channel trafficking, and promote membrane hyperpolarization. Conversely, activation of NMDARs mimicked the effect of leptin, causing Ca2+ influx, AMPK activation, and increased trafficking of KATP and Kv2.1 channels to the plasma membrane, and triggered membrane hyperpolarization. Moreover, leptin potentiated NMDAR currents and triggered NMDAR-dependent Ca2+ influx. Importantly, NMDAR-mediated signaling was observed in rat insulinoma 832/13 cells and in human β-cells, indicating that this pathway is conserved across species. The ability of NMDARs to regulate potassium channel surface expression and thus, β-cell excitability provides mechanistic insight into the recently reported insulinotropic effects of NMDAR antagonists and therefore highlights the therapeutic potential of these drugs in managing type 2 diabetes.

Insulin secretion by ␤-cells is under the control of a complex network of ion channels and signaling events (1). ATP-sensitive potassium (K ATP ) 5 channels composed of Kir6.2 and sulfonylurea receptor 1 (SUR1) subunits have a key role by coupling glucose metabolism to the ␤-cell membrane potential (2). Upon glucose stimulation, K ATP channels close in response to an increased intracellular ATP to ADP ratio, resulting in membrane depolarization, which activates voltage-gated calcium channels; the ensuing calcium influx then triggers insulin release (2,3). Cessation of insulin secretion occurs when the ␤-cell membrane potential returns to a hyperpolarized resting state. An important contributor of ␤-cell membrane repolarization is the voltage-gated delayed rectifier potassium channel Kv2.1. Reduction of Kv2.1 function in ␤-cells has been shown to enhance action potential duration, calcium influx, and insulin secretion (4,5). Recent studies showed that the density of K ATP and Kv2.1 channels in ␤-cells is dynamically regulated by metabolic or hormonal signals to modulate cell excitability (6 -10). In particular, leptin, a satiety hormone secreted by adipocytes to maintain energy and glucose homeostasis, was reported to promote trafficking of K ATP channels (6,8) and Kv2.1 channels (10) to the ␤-cell surface. Evidence suggests that leptin activates the AMP-activated protein kinase (AMPK) via its upstream kinase Ca 2ϩ -calmodulin-dependent protein kinase kinase ␤ (CaMKK␤) (8,10); however, the mechanism by which leptin activates the CaMKK␤-AMPK pathway in ␤-cells is unclear.
NMDA receptors (NMDARs) are ionotropic glutamate receptors whose activation requires the co-agonists glutamate and glycine as well as membrane depolarization, which removes external Mg 2ϩ block (11). NMDARs are Ca 2ϩ -permeable, which endows them the ability to trigger Ca 2ϩ -dependent signaling events. For example, in hippocampal neurons, Ca 2ϩ influx through NMDARs is coupled to activation of the Ca 2ϩdependent protein kinase CaMKK to induce long-term potentiation (12). Expression of NMDARs in ␤-cells has been reported since the mid-nineties (13)(14)(15). However, in contrast to their well-characterized functional role in the nervous system (16), the role of NMDARs in ␤-cells has remained elusive or even controversial. A recent study reported that inhibition of NMDARs in vitro and in vivo elicits increases in glucose-stimulated insulin secretion (GSIS) (17), but the underlying mechanism has yet to be elucidated.
In the present study, we demonstrate that NMDARs are expressed by ␤-cells and are required for leptin-induced calcium influx, AMPK activation, increased K ATP and Kv2.1 channel surface expression, and reductions in ␤-cell membrane excitability. Moreover, we show that activation of NMDARs alone induces channel trafficking and reduces ␤-cell membrane excitability. These findings reveal an important role of NMDARs in regulating ␤-cell excitability and provide a novel mechanistic paradigm for insulin secretion regulation.

NMDARs are expressed in pancreatic ␤-cells
We previously reported that leptin increases the surface density of K ATP and Kv2.1 channels in rat insulinoma INS-832/13 cells and human ␤-cells. In INS-832/13 cells, this increase is dependent upon activation of the AMPK, which is in turn dependent on its upstream effector, CaMKK␤ (6,10). Studies in hippocampal neurons have linked calcium influx through NMDARs to activation of the CaMKK␤-AMPK pathway (18,19). Furthermore, NMDAR stimulation has been shown to increase K ATP currents in an AMPK-dependent manner in subthalamic neurons (20,21). These reports prompted us to investigate whether NMDARs could be involved in the leptin signaling pathway that regulates surface expression of K ATP and Kv2.1 channels in ␤-cells.
Although expression of NMDARs and their functional roles have been studied in a number of rodent ␤-cell lines or primary islets by measuring mRNA, protein, or currents, the results vary and in some cases are controversial (13,15,17,(22)(23)(24). We first determined whether NMDARs are expressed by INS-832/13 cells, which were used in our previous studies. Immunoblotting was used to probe the NMDAR subunit GluN1, which is the mandatory subunit for all functional NMDARs (25), in INS-832/13 cell lysate. Although GluN1 protein was expressed by INS-832/13 cells, its expression level was less than that observed in whole brain homogenate (Fig. 1A). No expression was observed in COS cells, which lack NMDARs. Immunostaining illustrated that only 43% of insulin-positive INS-832/13 and 46% of human ␤-cells expressed detectable levels of GluN1 protein (Fig. 1B). Notably, although most GluN1-positive INS-832/13 and human ␤-cells showed low level of staining, some INS-832/13 cells showed intense GluN1 signals ( Fig.  1, C and D). In addition, we found that a small percentage of dissociated islet cells (23%), although positive for GluN1, were not identified as ␤-cells, implicating that other cell types within human islets also express NMDARs.
We next conducted whole-cell patch clamp recordings and used local pressure (puff) application of NMDA (1 mM) to assess NMDAR function. In 10 of 21 cells tested, puff application of NMDA induced inward currents (holding potential, Ϫ70 mV; no external Mg 2ϩ ) with a mean of 9.0 Ϯ 1.4 pA that was inhibited to 1.8 Ϯ 0.2 pA by the non-competitive NMDAR antagonist MK-801 (50 M; p Ͻ 0.001, n ϭ 10 by paired t test; Fig. 1E). Application of the competitive NMDAR antagonist D-APV (50 M) also reduced NMDAR currents (from 28.4 Ϯ 7.2 to 9.7 Ϯ 5.2 pA; p Ͻ 0.001, n ϭ 12 by paired t test; not shown). Consistent with immunostaining results, not all cells recorded had detectable NMDAR currents, and those that did displayed a range of amplitudes that reflected the heterogeneity in NMDAR expression (Fig. 1D). Importantly, NMDA-evoked currents were also observed in dispersed human ␤-cells and were reduced by MK-801 (from 21.3 Ϯ 8.9 to 6.3 Ϯ 3.0 pA; p Ͻ 0.001, n ϭ 5 by paired t test; Fig. 1F). Puff application of glutamate (1 mM), a physiological ligand of the NMDAR, also elicited outward currents when cells were held at a positive potential of 40 mV that were reversibly blocked by MK-801 (27.8 Ϯ 7.0 pA for glutamate, 2.3 Ϯ 1.4 pA for MK-801, and 23.5 Ϯ 6.7 pA for MK-801 washout; p Ͻ 0.01, n ϭ 5 by paired t test; Fig. 1G). Together, these results show that both INS-832/13 cells and human ␤-cells express functional NMDARs.

NMDARs are required for leptin-induced surface trafficking of K ATP and Kv2.1 channels
To test the role of NMDARs in leptin-induced surface trafficking of K ATP and Kv2.1 channels, we monitored surface expression of these channels using surface biotinylation following treatment of INS-832/13 cells with 0.1% DMSO, 10 nM leptin, or 50 M NMDA for 30 min in the absence or presence of MK-801. To test for specificity among ionotropic glutamate receptors, the effect of ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) (50 M), an agonist of the ionotropic AMPA receptor, which has been reported to be expressed in ␤-cells (26,27), was also examined in the absence or presence of MK-801. As expected (6), leptin induced a significant increase (ϳ2-fold) in surface SUR1 as compared with controls ( Fig. 2A). Strikingly, although treatment with NMDA induced a similar increase in biotinylated SUR1 protein, treatment with AMPA did not, suggesting that this is not a general response to ionotropic glutamate receptor activation ( Fig. 2A). Both leptin-and NMDA-induced increases in biotinylated SUR1 and Kv2.1 were inhibited by NMDAR antagonists (Fig. 2, A and C). Although a previous study attributed the leptin-induced trafficking of K ATP channels to activation of TRPC4 channels (8), we failed to see an effect of TRPC4 inhibition (ML204; 10 M) (28) on leptin-induced increases in surface expression of either K ATP or Kv2.1 channels (Fig. 2, B and D).
Previously, we and others showed that leptin increases AMPK phosphorylation at residue Thr-172 of the catalytic subunit, a signature of AMPK activation (8,10). The increase in phosphorylation was shown to be Ca 2ϩ -dependent and required the Ca 2ϩ -dependent kinase CaMKK␤. Importantly, activation of AMPK was required for the trafficking of K ATP and Kv2.1 channels (7). We found that, like leptin, NMDA also increased AMPK Thr(P)-172 and that the increase was blocked by MK-801 in both cases, implying that they activate similar signaling pathways (Fig. 3, A and B). Taken together, these results reveal a necessary role of NMDARs in mediating the effect of leptin on AMPK activation and the subsequent surface expression of K ATP and Kv2.1 channels.

NMDARs mediate leptin-induced hyperpolarization in INS-832/13 cells
Delivery of K ATP channels to the plasma membrane of INS-832/13 cells upon leptin signaling results in membrane hyperpolarization (10). We reasoned that this hyperpolarization should NMDA receptors regulate K ؉ channel trafficking in ␤-cells

NMDA receptors regulate K ؉ channel trafficking in ␤-cells
also be NMDAR-dependent. To monitor changes in membrane potential, we used cell-attached patch clamp recording, which provides an accurate measure of membrane potential, maintains cell integrity, and prevents dialysis of soluble factors that may be important for proper signaling (29). Membrane potential was recorded in response to vehicle or leptin in the absence or presence of tolbutamide, an inhibitor of K ATP channels, the competitive NMDAR antagonist D-APV, or the TRPC4 inhibitor ML204. In addition, dextromethorphan (DXM), an NMDAR antagonist recently shown to stimulate insulin secretion and improve glucose tolerance in type 2 diabetes patients (17), was also tested.
In the majority of cells tested (23 of 32 cells), bath application of leptin in the presence of 11 mM glucose induced a significant membrane hyperpolarization (⌬V m ϭ Ϫ0.2 Ϯ 0.7 and Ϫ23.1 Ϯ 3.5 mV for vehicle and leptin, respectively; p Ͻ 0.001) (Fig. 4). Leptin induced similar results when cells were recorded under whole-cell mode (⌬V m ϭ Ϫ26.6 Ϯ 3.6 mV; n ϭ 16), confirming the integrity of our cell-attached recordings. The onset of hyperpolarization occurred ϳ1-10 min after the start of leptin perfusion and was largely abolished by co-application with tolbutamide (30) (⌬V m ϭ Ϫ4.7 Ϯ 2.0 mV), indicating that the source of hyperpolarization was from K ATP channels. These results are in good agreement with those reported previously in a different ␤-cell line, CR1-G1 (31). By contrast, in the absence of leptin, tolbutamide alone did not significantly alter the depolarized baseline potential of INS-832/13 cells bathed in 11 mM glucose (baseline, 3.6 Ϯ 2.5 mV; tolbutamide, 7.9 Ϯ 3.3 mV; n ϭ NMDA receptors regulate K ؉ channel trafficking in ␤-cells 9), indicating that under this condition most existing K ATP channels are closed as expected. We next examined whether blockade of NMDARs could inhibit the ability of leptin to hyperpolarize the cell membrane. Application of either D-APV or DXM significantly reduced the level of hyperpolarization by leptin (⌬V m ϭ Ϫ6.2 Ϯ 2.9 and Ϫ5.9 Ϯ 2.1 mV, respectively; Fig.  4C). By contrast, cells receiving bath application of leptin with the TRPC4 inhibitor ML204 showed a similar level of hyperpolarization as with leptin alone (⌬V m ϭ Ϫ25.8 Ϯ 3.6 mV; Fig. 4, B and C), suggesting that TRPC4 channels do not contribute significantly to leptin-induced membrane hyperpolarization.

NMDAR activation is sufficient to induce membrane hyperpolarization in ␤-cells
Because NMDA is sufficient to enhance surface expression of K ATP channels, we asked whether it also regulates ␤-cell membrane excitability. In the majority of cells tested (ϳ80%), perfusion of NMDA (50 M) alone triggered a significant hyperpolarization in membrane potential in both INS-832/13 (⌬V m ϭ Ϫ20.7 Ϯ 4.4 mV; p Ͻ 0.01) and human ␤-cells (⌬V m ϭ Ϫ30 Ϯ 5.4 mV; p Ͻ 0.001) (Fig. 5C). The onset of hyperpolarization to NMDA treatment averaged ϳ7 min but ranged from 1 to 14 min in INS-832/13 cells and human ␤-cells after the start of perfusion. In some instances, NMDA induced a small membrane depolarization, indicative of NMDAR activation, prior to the onset of hyperpolarization (Fig. 5A, lower trace). Importantly, the NMDA-mediated hyperpolarization in both INS-832/13 and human ␤-cells was blocked by tolbutamide, suggesting that, like leptin, NMDA affects ␤-cell membrane potential via K ATP channels (Fig. 5C). These data indicate that activation of NMDARs is sufficient to induce trafficking of K ATP channels and subsequent membrane hyperpolarization in INS-832/13 and human ␤-cells.
Next, we tested whether leptin induces Ca 2ϩ influx through NMDARs. Bath application of leptin triggered a more sustained increase in intracellular Ca 2ϩ as compared with puff application of NMDA (Fig. 6D); however, similar sustained increases in Ca 2ϩ were observed when NMDA was bath-applied (see Fig.  6D, inset). On average, leptin induced a rise in intracellular Ca 2ϩ that was 2.48 Ϯ 0.15-fold higher than baseline. Importantly, this increase was markedly reduced by co-application of either D-APV (1.43 Ϯ 0.07-fold), MK-801 (1.50 Ϯ 0.08-fold), or DXM (1.28 Ϯ 0.10-fold) but little affected by ML204 (2.32 Ϯ 0.29-fold) (Fig. 6E). These data provide strong evidence that leptin triggers Ca 2ϩ influx in INS-832/13 cells predominantly through NMDARs.

Leptin potentiates NMDAR currents in ␤-cells
In the hippocampus, leptin has been reported to facilitate long-term potentiation by enhancing NMDAR function (32,33), raising the possibility that leptin may potentiate the function of NMDARs to trigger the downstream effects in ␤-cells.
To test this possibility, we conducted whole-cell voltage clamp experiments from individual ␤-cells and induced NMDAR currents by puff application of NMDA before and after the subsequent addition of leptin. We found that NMDAR currents evoked in INS-832/13 cells were significantly potentiated by leptin (Fig. 7A, left trace). On average, leptin increased NMDAR currents by 164.2 Ϯ 21.7% after 6 min (Fig. 7, A and B). The increase could be seen as quickly as 2 min following bath application of leptin and could be sustained for up to 30 min. In three  Fig. 2A). B, -fold increase in the mean pAMPK/AMPK ratio for each condition from three independent experiments. Error bars represent S.E. *, p Ͻ 0.05.

NMDA receptors regulate K ؉ channel trafficking in ␤-cells
of eight cells, leptin failed to potentiate NMDAR currents, indicating that potentiation did not result from repeated NMDA puff applications (see example in Fig. 7A, right trace) and suggesting surface leptin receptor heterogeneity among ␤-cells. Importantly, leptin induced a marked enhancement in NMDAevoked currents by 187.6 Ϯ 32.7% in human ␤-cells (Fig. 7C). Taken together, these data reveal a novel link between leptin signaling and the function of NMDARs in ␤-cells.

Discussion
Utilizing a combination of biochemical, electrophysiological, and imaging approaches, we show that NMDARs play a crucial role in mediating the effect of leptin in ␤-cells. Together with our previous studies (6,10), we propose a model by which leptin suppresses glucose-stimulated insulin secretion (Fig. 8). At 11 mM glucose under which condition ␤-cells are depolarized, NMDARs exposed to glutamate and glycine present in the extracellular milieu are able to conduct currents due to Mg 2ϩ unblock. Leptin potentiates NMDAR currents, thus increasing Ca 2ϩ influx to activate CaMKK␤, which then phosphorylates and activates AMPK to trigger PKA-dependent actin remodeling and trafficking of K ATP and Kv2.1 channels to the cell surface. The increased surface expression of Kv2.1 channels induced by NMDAR activation is expected to shorten action potentials (10), facilitating membrane repolarization, whereas increased K ATP channel abundance would facilitate membrane hyperpolarization to limit voltage-dependent Ca 2ϩ influx and inhibit insulin secretion. The expression of functional NMDARs was observed in both rodent and human ␤-cells, indicating that this mechanism is highly conserved and likely plays an important role in regulating ␤-cell function and insulin release.
A recent study by Marquard et al. (17) found that NMDARs were required to limit GSIS and that inhibition of NMDARs could serve as a potential antidiabetic treatment. However, how NMDARs regulate insulin secretion remains unknown. Interestingly, these authors showed that inhibition of NMDARs by DXM could enhance GSIS without affecting basal insulin secretion; moreover, the effect of NMDAR inhibition on insulin secretion was absent in Kir6.2 knock-out mice, implicating involvement of K ATP channels (17). Our findings provide a molecular mechanism for the observations made by Marquard et al. (17). Specifically, during low glucose, the negative resting ␤-cell membrane potential prevents NMDARs from opening due to blockade by external Mg 2ϩ (11), rendering NMDAR inhibition ineffective. When glucose is high, ␤-cells depolarize, removing the Mg 2ϩ block, allowing Ca 2ϩ to enter and trigger channel trafficking, increasing K ϩ efflux, and suppressing insulin release. Under this condition, inhibiting NMDARs would enhance insulin release by preventing K ϩ channel trafficking.
The co-trafficking of K ATP and Kv2.1 channels by leptin has been shown to require activation of CaMKK␤ and its down-

NMDA receptors regulate K ؉ channel trafficking in ␤-cells
stream effector AMPK (8,10), which brings into question the potential source of Ca 2ϩ influx upstream of CaMKK␤. Although it was reported previously that Ca 2ϩ influx through TRPC4 channels was the most likely candidate (8), we found that TRPC4 inhibition was unable to prevent recruitment of K ATP and Kv2.1 channels to plasma membrane. Rather, our data show that NMDARs are the primary Ca 2ϩ source mediating the effect of leptin as NMDAR antagonists (competitive and non-competitive) as well as a selective pore blocker inhibited leptin-induced Ca 2ϩ influx, activation of AMPK, K ATP and Kv2.1 channel trafficking, and membrane hyperpolarization.
Our finding that NMDAR activation triggers the co-regulation of K ATP and Kv2.1 channels highlights its importance in modulating the excitability of ␤-cells to control insulin secretion. This raises questions regarding how Kv2.1 and K ATP channels are sorted within ␤-cells. For example, are they

NMDA receptors regulate K ؉ channel trafficking in ␤-cells
sorted into specific secretory vesicles from the trans-Golgi network, and/or are they endocytosed together for subsequent recycling to the membrane? Knowing how Kv2.1 and K ATP channels are trafficked will undoubtedly provide new avenues for drug development to overcome diseases related to insulin and leptin misregulation.
There is increasing evidence that insulin secretion from ␤-cells is regulated by glutamate (34). In addition to expressing NMDARs and other glutamate receptors (13)(14)(15), pancreatic ␤-cells express vesicular glutamate transporters VGLUT1 and VGLUT3 (35), the excitatory amino acid transporter EAAT2 (36), and the glial glutamate transporter GLT1 (37), suggesting active regulation of intracellular and extracellular glutamate signals. Several possible physiological sources of glutamate for ␤-cells have been reported, including circulating plasma glutamate in the range of ϳ20 -30 M (38), which is well above the EC 50 of ϳ1 M for NMDARs (11); glutamate released by ␣-cells (39); glutamate released by ␤-cells that is not coupled to secretion (40); and finally glutamate co-released from insulin granules (36). The possibility that ␤-cells may co-release glutamate during insulin secretion is intriguing. Such a mechanism would allow NMDAR activation to be coordinated with periods of insulin release. Thus, NMDAR activation may provide autoinhibitory feedback to prevent the oversecretion of insulin per- NMDA receptors regulate K ؉ channel trafficking in ␤-cells haps even in the absence of leptin. In this regard, it is interesting to note that glucose stimulation has been reported to promote K ATP channel trafficking to the cell surface (9).
Our results show that leptin potentiates NMDAR currents. Potentiation of NMDAR currents by leptin has been observed in other cells such as hippocampal neurons (41) and cerebellar granule cells (42). How leptin potentiates NMDARs in ␤-cells is not clear, although involvement of Src kinases has been implicated in hippocampal neurons (41). More studies are needed to determine whether a similar mechanism is at play in ␤-cells and whether these signaling molecules form a complex to confer spatial and temporal specificity.
Analysis of data from immunocytochemistry, electrophysiology, and Ca 2ϩ imaging experiments revealed non-uniform NMDAR expression and responses to leptin. Studies of dispersed clonal or primary ␤-cells suggest that ion channel expression and composition in ␤-cells is heterogeneous (43,44), including a recent report of heterogeneous NMDAR expression in the BRIN-BD11 ␤-cell line (24). Such cell-to-cell variations may explain controversies regarding ␤-cell expression of NMDARs (45,46). Although heterogeneity among ␤-cells is becoming increasingly evident (47), how these subpopulations contribute to islet function and whether their expression patterns change in response to autocrine or paracrine signals are still open questions. Modeling studies have shown that heterogeneity in individual ␤-cell electrical properties is largely negated by the electrical coupling of ␤-cells via gap junctions to give rise to synchronized ␤-cell activity in islets (48,49). Thus, expression of NMDARs or leptin receptors in every ␤-cell may not be necessary for glutamate or leptin to exert a significant impact on the overall function of an islet. In this context, NMDAR-expressing ␤-cells may function to trigger waves of hyperpolarization throughout the islet to suppress insulin release. Our immunocytochemistry data on dispersed human islet cells also suggest expression of NMDARs in non-␤ cells, although the functional significance of this finding awaits further investigation.
In summary, we demonstrate that NMDARs are unequivocally expressed in pancreatic ␤-cells and contribute to Ca 2ϩ -and CaMKK␤-dependent trafficking of K ATP and Kv2.1 channels to the plasma membrane. The signaling pathway elucidated here provides a cellular mechanism linking glutamate signaling to NMDAR-dependent regulation of insulin secretion to explain the reported antidiabetic effects of NMDAR antagonists and  Binding of leptin to its receptor potentiates NMDAR currents, which leads to an increase in Ca 2ϩ influx, activation of CaMKK␤, and phosphorylation of Thr-172 on AMPK␣, which then cause PKA-dependent F-actin depolymerization to promote K ATP and Kv2.1 channel trafficking to the plasma membrane. The increased K ϩ efflux through increased surface density of Kv2.1 channels shortens action potentials and facilitates membrane repolarization, whereas the increased K ϩ efflux through increased surface K ATP channels leads to membrane hyperpolarization. Together, they reduce ␤-cell excitability and suppress insulin secretion.

NMDA receptors regulate K ؉ channel trafficking in ␤-cells
further reinforces the therapeutic potential of NMDAR in the treatment of diabetes.

Cell culture
INS-1 cells (clone 832/13; referred to herein as INS-832/13) were cultured in RPMI 1640 medium with 11.1 mM D-glucose (Invitrogen) supplemented with 10% fetal bovine serum (FBS), 100 units/ml penicillin, 100 g/ml streptomycin, 10 mM HEPES, 2 mM glutamine, 1 mM sodium pyruvate, and 50 M ␤-mercaptoethanol (50). Human ␤-cells were dissociated from human islets obtained through the Integrated Islets Distribution Program as described previously (6,10). Briefly, human islets were cultured in RPMI 1640 medium with 10% FBS and 1% L-glutamine. For recording, islets were dissociated into single cells by trituration in a solution containing 116 mM NaCl, 5.5 mM D-glucose, 3 mM EGTA, and 0.1% bovine serum albumin (BSA), pH 7.4. Dissociated cells were then plated on 0.1% gelatin-coated coverslips. For electrophysiological experiments, ␤-cells were identified using two separate criteria. The first criterion utilized the high autofluorescence signature of ␤-cells to 488-nm excitation as these cells have high concentrations of unbound flavin adenine dinucleotide (48), and the second criterion was that cells had an initial depolarizing membrane potential (ϳ0 mV) in response to 11 mM glucose (40). Donor information for specific experiments is provided in Table 1.

Drug treatments
Leptin, tolbutamide, glutamate, and dextromethorphan were from Sigma. NMDA, (R,S)-AMPA, MK-801, D-APV, STO-609, and ML204 were from Tocris Bioscience (Bristol, UK). For surface biotinylation experiments, INS-832/13 cells were incubated in regular RPMI 1640 medium without serum for 30 min before treatment with leptin or NMDA for 30 min. Where stated, pharmacological inhibitors were added 30 min before and during the addition of leptin and NMDA.

Immunoblotting
INS-832/13 cells were lysed in lysis buffer (50 mM Tris-HCl, 2 mM EDTA, 2 mM EGTA, 100 mM NaCl, 1% Triton X-100, pH 7.4, with Complete protease inhibitor) for 30 min at 4°C, and cell lysates were cleared by centrifugation at 21,000 ϫ g for 10 min at 4°C. Proteins were separated by SDS-PAGE (7.5-12.5%) and transferred onto nitrocellulose or PVDF membranes (Millipore). Membranes were incubated overnight at 4°C with a primary antibody diluted in TBST (Tris-buffered saline plus 0.1% Tween 20) followed by incubation with horseradish peroxidase-conjugated secondary antibodies in TBST for 1 h at room temperature. Antibodies against GluN1 and Kv2.1 (clone K89/34) were from NeuroMab (Davis, CA). Antibodies against AMPK and phosphorylated AMPK (pAMPK) were from Cell Signaling Technology (Danvers, MA) and Millipore, respectively. Antibody for SUR1 was generated in rabbit using a C-terminal peptide (KDSVFASFVRADK) of hamster SUR1 as described previously (51). Blots were developed using Super Signal West Femto (Pierce) and imaged with FluorChemE (Pro-teinSimple, San Jose, CA). Blots were stripped and reprobed with anti-tubulin (Sigma) as a control for loading. The blots were quantified with ImageJ (National Institutes of Health) and normalized to the corresponding controls.

Immunocytochemistry
INS-832/13 cells were fixed in 2% paraformaldehyde in phosphate-buffered saline (PBS) for 10 min at room temperature, permeabilized with 0.2% Triton X-100, and blocked for 60 min with 1% BSA in PBST (PBS ϩ 0.1% Tween 20) before being incubated overnight at 4°C with primary antibodies directed against GluN1 (NeuroMab) and insulin (Cell Signaling Technology). Proteins were visualized using Cy3-and Alexa Fluor 488conjugated secondary antibodies. Fluorescence images were acquired using a Zeiss LSM780 confocal microscope equipped with a 63ϫ oil immersion objective. Images were processed and analyzed using NIH ImageJ software.

Electrophysiology
Electrical recordings were performed using an Axon 200B amplifier (Molecular Devices, Sunnyvale, CA) and were filtered at 2 kHz and digitized and acquired at 20 kHz using pCLAMP software. For whole-cell patch clamp recordings, cells were held at Ϫ70 or 40 mV using micropipettes pulled from non- Cell-attached recording electrodes were pulled as described above and filled with 140 mM NaCl and had tip resistances of 2-6 megaohms. The liquid junction potential was calculated to be 0 mV. Seal resistances ranged from 1 to 5 gigaohms between the recording pipette and the cell membrane. Membrane potentials were recorded in current clamp mode. Signals were analyzed using Clampfit (pCLAMP).

Surface biotinylation
INS-832/13 cells were washed twice with cold PBS and incubated with 1 mg/ml EZ-Link Sulfo-NHS-SS-Biotin (Pierce) in PBS for 30 min on ice. The reaction was terminated by incubating cells for 5 min with PBS containing 50 mM glycine followed by three washes with cold PBS. Cells were then lysed in 300 l of lysis buffer as described above, and 500 g of total lysate was incubated with 100 l of an ϳ50% slurry of NeutraAvidin-agarose beads (Pierce) overnight at 4°C. Biotinylated proteins were eluted with 2ϫ protein loading buffer for 15 min at room temperature. Both eluent and input samples (50 g of total cell lysate) were analyzed by immunoblotting using anti-SUR1 or anti-Kv2.1 antibody described previously (6,10).

Calcium imaging
INS-832/13 cells were loaded in the dark with 2 M Fluo-4 AM (Thermo Fisher Scientific) according to the manufacturer's instructions. To prevent indicator extrusion by organic anion transporters, probenecid (2.5 mM) (Thermo Fisher Scientific) was added during loading (52). Cells were then imaged on an upright Leica microscope outfitted with a 40ϫ water immersion objective (0.8 numerical aperture) and a Polychrome IV monochromator light source (TILL Photonics, Munich, Germany) and continuously perfused with Tyrode's solution (with 0.1 mM glycine but no Mg 2ϩ or glucose) at room temperature (21-25°C). Note that glucose was not included to avoid glucose-induced Ca 2ϩ signals, which would make it difficult to discern NMDA-or leptin-evoked Ca 2ϩ signals; Mg 2ϩ was also not included to avoid blocking of NMDARs under no-glucose conditions. Fluo-4 fluorescence was excited at 480 nm, and fluorescence emission was filtered through a 525 (50)-nm singleband bandpass filter (Chroma Technology, Bellows Falls, VT). Images were acquired using a 12-bit ORCA-ER charge-coupled device camera (Hamamatsu, Japan) controlled by Metafluor image acquisition software (Molecular Devices). Images were acquired every 0.5 s and digitized. Fluorescence intensity was analyzed post hoc in regions of interest manually drawn around individual INS-832/13 cells using Metafluor. Changes in intracellular Ca 2ϩ concentration were normalized to baseline and expressed as -fold change in ⌬F/F 0 ϭ ((F Ϫ F 0 )/F 0 ) where F 0 is the average, background-subtracted baseline fluorescence and F is the fluorescence intensity immediately following the addition of NMDA or leptin.

Statistical analysis
Results are expressed as means Ϯ S.E. Differences were tested using one-way analysis of variance followed by the post hoc Dunnett's test for multiple comparisons. When only two groups were compared, unpaired or paired Student's t tests were used where indicated. The level of statistical significance was set at p Ͻ 0.05.
Author contributions-Y. W. designed and performed experiments, analyzed data, and edited the manuscript. D. A. F. designed and performed experiments, analyzed data, and wrote the manuscript. V. A. C. designed and performed calcium imaging experiments, analyzed data, and edited the manuscript. P.-C. C. conceived the project, designed and performed experiments, and analyzed data. S.-L. S. conceived the project, designed experiments, and wrote the manuscript. All authors have 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. All authors reviewed the results and approved the final version of the manuscript.