Leptin modulates pancreatic b -cell membrane potential through Src kinase – mediated phosphorylation of NMDA receptors

The adipocyte-derived hormone leptin increases trafficking of K ATP and Kv2.1 channels to the pancreatic b -cell surface, resulting in membrane hyperpolarization and suppression of insulin secretion. We have previously shown that this effect of leptin is mediated by the NMDA subtype of glutamate receptors (NMDARs). It does so by potentiating NMDAR activity, thus enhancing Ca 2 1 influx and the ensuing downstream signaling events that drive channel trafficking to the cell surface. How-ever, the molecular mechanism by which leptin potentiates NMDARs in b -cells remains unknown. Here, we report that leptin augments NMDAR function via Src kinase – mediated phosphorylation of the GluN2A subunit. Leptin-induced membrane hyperpolarization diminished upon pharmacological inhibition of GluN2A but not GluN2B, indicating involvement of GluN2A-containing NMDARs. GluN2A harbors tyrosine residues that, when phosphorylated by Src family kinases, potentiate NMDAR activity. We found that leptin increases phosphorylation of of a of leptin, a activator overexpression, show that and Tyr-1325 are for the effect of leptin. -cells from db/db a type 2 diabetes mouse from to reveals a signaling wherein

The adipocyte-derived hormone leptin increases trafficking of K ATP and Kv2.1 channels to the pancreatic b-cell surface, resulting in membrane hyperpolarization and suppression of insulin secretion. We have previously shown that this effect of leptin is mediated by the NMDA subtype of glutamate receptors (NMDARs). It does so by potentiating NMDAR activity, thus enhancing Ca 21 influx and the ensuing downstream signaling events that drive channel trafficking to the cell surface. However, the molecular mechanism by which leptin potentiates NMDARs in b-cells remains unknown. Here, we report that leptin augments NMDAR function via Src kinase-mediated phosphorylation of the GluN2A subunit. Leptin-induced membrane hyperpolarization diminished upon pharmacological inhibition of GluN2A but not GluN2B, indicating involvement of GluN2Acontaining NMDARs. GluN2A harbors tyrosine residues that, when phosphorylated by Src family kinases, potentiate NMDAR activity. We found that leptin increases phosphorylation of Tyr-418 in Src, an indicator of kinase activation. Pharmacological inhibition of Src or overexpression of a kinase-dead Src mutant prevented the effect of leptin, whereas a Src kinase activator peptide mimicked it. Using mutant GluN2A overexpression, we show that Tyr-1292 and Tyr-1387 but not Tyr-1325 are responsible for the effect of leptin. Importantly, b-cells from db/db mice, a type 2 diabetes mouse model lacking functional leptin receptors, or from obese diabetic human donors failed to respond to leptin but hyperpolarized in response to NMDA. Our study reveals a signaling pathway wherein leptin modulates NMDARs via Src to regulate b-cell excitability and suggests NMDARs as a potential target to overcome leptin resistance.
Leptin is an adipocyte-produced hormone that plays a key role in body weight regulation. The physiological actions and signaling mechanisms of leptin in the central nervous system have been extensively studied. Less well-understood is the function and mechanism of leptin signaling in peripheral tissues. In pancreatic islets, leptin was reported to down-regulate glucose-stimulated insulin secretion more than 2 decades ago (1)(2)(3)(4)(5)(6). Recent studies find that leptin stimulates potassium channel trafficking to the cell surface to reduce b-cell excitability (7)(8)(9). Specifically, leptin causes a transient increase in the number of ATP-sensitive potassium (K ATP ) channels and Kv2.1 channels in the b-cell plasma membrane (9). K ATP channels control b-cell resting membrane potential and couple blood glucose with insulin secretion, whereas Kv2.1 channels play a prominent role in action potential repolarization to stop insulin secretion (10,11). The increased surface abundance of these channels would reduce b-cell excitability and thus explain how leptin inhibits glucose-stimulated insulin secretion.
Leptin signaling is complex, and multiple signal transduction pathways have been described (12,13). The best-characterized is JAK2-dependent phosphorylation of STAT3 following activation of the ObRb receptor, but phosphatidylinositol 3-kinase/Akt, mitogen-activated protein kinase, AMPK, and Src family kinases (SFKs) are also possible downstream effectors (13,14). In addition, transactivation of other cytokine receptors via leptin signaling has also been implicated (13). Studies into the mechanisms by which leptin regulates K ATP and Kv2.1 channel trafficking so far have identified several key molecular players. These include the NMDA subtype glutamate receptors (NMDARs), calcium/calmodulin-dependent protein kinase kinase b (CaMKKb), AMPK, and PKA (15). Evidence that has emerged reveals a novel signaling pathway wherein leptin potentiates NMDAR function to increase Ca 21 influx, resulting in activation of CaMKKb, which phosphorylates and activates AMPK; AMPK in turn causes PKA-dependent actin depolymerization, culminating in increased trafficking of K ATP and Kv2.1 channels to the b-cell surface. A key question of how leptin potentiates NMDAR activity, however, has yet to be addressed.
NMDARs are calcium-permeant ionotropic glutamate receptors that are highly expressed in the brain and are important for learning and memory (16). Although they are extensively studied in the central nervous system, there is growing evidence that they are expressed in pancreatic b-cells and play a role in regulating insulin secretion (17)(18)(19). Functional NMDARs generally form as heterotetramers of two obligatory glycinebinding GluN1 subunits (also known as NR1) and two glutamate-binding GluN2 (GluN2A-D or NR2A-D) subunits (20). NMDARs containing GluN2A and -2B subunits are highly sensitive to blockade by extracellular Mg 21 under negative membrane potential and require membrane depolarization to remove Mg 21 block for activity, whereas those with GluN2C and -2D subunits are significantly less sensitive to external Mg 21 (16,20,21). Additionally, NMDAR activity can be modulated by post-translational modifications. In particular, GluN2A and GluN2B have long cytoplasmic tails that are known to be phosphorylated at serine/threonine and/or tyrosine residues by a variety of kinases, including CDK5, PKA, PKC, CaMKII, casein kinase II, and protein tyrosine kinases (22). In neurons, these phosphorylation modifications have been linked to regulation of NMDAR trafficking, localization, and function (reviewed in Ref. 23). For example, phosphorylation of several tyrosine residues in GluN2A by SFKs have been shown to enhance NMDAR activity (24,25). Of note, Src activation is one of the many downstream signaling events that have been reported following stimulation of the ObRb leptin receptor by leptin (14,26,27), raising the possibility that leptin may signal through SFKs to modulate NMDAR activity in pancreatic b-cells.
Here, we present evidence that leptin activates Src to phosphorylate GluN2A-containing NMDARs in pancreatic b-cells, which results in potentiation of NMDAR currents and increased trafficking of K ATP and Kv2.1 channels to the cell surface to hyperpolarize b-cell membrane potential. Importantly, we show that leptin fails to induce membrane hyperpolarization in b-cells from the leptin-resistant db/db mice and obese diabetic human donors. Interestingly, direct activation of NMDARs by NMDA was able to mimic the effect of leptin. As leptin resistance is frequently associated with obesity-related diabetes, NMDARs may be a potential target to overcome leptin resistance in diabetic b-cells.

Leptin hyperpolarizes pancreatic b-cells through GluN2A-containing NMDARs
In rodent pancreatic b-cell lines as well as primary human b-cells, leptin induces membrane hyperpolarization at glucose concentrations that depolarize b-cell membrane potential (3,(7)(8)(9)19). Recently, we showed that, both in rat insulinoma INS-1 832/13 cells and human b-cells, this effect is mediated by NMDARs (19). Leptin stimulation increased NMDA currents, leading to increased Ca 21 influx and increased surface density of K ATP and Kv2.1 channels (19). To determine the mechanism by which leptin increases NMDA currents, we began by characterizing the NMDARs that are expressed in b-cells. Functional NMDARs that are Ca 21 -permeant are tetramers of two GluN1 subunits and two of four different GluN2 subunits, GluN2A, -2B, -2C, and -2D (20). The GluN1 subunit is obligatory and is encoded by a single gene, Grin1. In contrast, GluN2 subunits are encoded by four different genes: Grin2a, Grin2b, Grin2c, and Grin2d for GluN2A-D (20). We first examined the expression of these subunits at the mRNA level in INS-1 832/13 cells by RT-PCR. Transcripts for Grin1, Grin2a, Grin2b, and Grin2d were clearly detected, but not Grin2c (Fig. 1A), suggesting that these cells express GluN1, GluN2A, GluN2B, and GluN2D, but not GluN2C.
NMDARs containing GluN2A or -2B are much more sensitive to external Mg 21 block than those containing GluN2C or -2D (20,28). We thus tested Mg 21 sensitivity of NMDA currents using whole-cell recording as an indicator of the GluN2 subunit composition. Puff application of NMDA (1 mM) at a holding potential of 270 mV in Mg 21 -free external solution elicited NMDA currents that were reduced by .80% upon the addition of MgCl 2 (100 mM) to the external solution (from 11.3 6 2.6 to 1.3 6 0.5 pA, n = 4; p , 0.05 by paired t test) (Fig.  1, B and C). The strong Mg 21 block observed is characteristic of NMDARs containing GluN2A and/or GluN2B subunits (21,28), suggesting that NMDARs at the surface of these cells are largely made up of GluN1 and GluN2A and/or GluN2B subunits. Consistently, in Western blotting experiments, GluN1 and GluN2A were readily detectable in total cell lysate and further enriched in the membrane fraction, whereas GluN2B was barely detectable in total cell lysate but clearly seen in the membrane fraction (Fig. 1D).
Because both GluN2A and -2B could potentially mediate the leptin response, we sought to determine their relative contributions using subunit-specific inhibitors: TCN-201 for GluN2A and Ro 25-6981 for GluN2B (20,29,30). Hyperpolarization of membrane potential in Tyrode's solution containing 11 mM glucose was used as a readout for leptin response. Because GluN2A inhibition by TCN-201 depends on glycine, which at higher concentrations renders the drug less effective (30), we reduced the glycine supplement in Tyrode's solution from 100 to 50 mM. These experiments were performed using cellattached current-clamp recording, which provides a robust assessment of changes in membrane potential while maintaining cell integrity and preventing dialysis of soluble factors that may be important for intracellular signaling (31,32). Bath application of leptin alone (10 nM) induced a mean membrane hyperpolarization of 246.8 6 8.1 mV, similar to that reported previously (9,19). However, co-application of the GluN2Aselective antagonist TCN-201 (50 mM) with leptin only induced a mean hyperpolarization of 214.4 6 3.9 mV, significantly less than that observed in cells treated with leptin alone (Fig. 1, E and F). By contrast, co-application of the GluN2B-selective antagonist Ro 25-6981 (33) had little effect on the ability of leptin to hyperpolarize INS-1 832/13 cells (Ro 25-6981 at 1 mM, DV m = 235.8 6 3.8 mV; Fig. 1, E and F). These results suggest that GluN2A is largely responsible for mediating the effect of leptin. Note that there is evidence that GluN2A can form diheteromeric GluN1/GluN2A or triheteromeric GluN1/GluN2A/ GluN2B complexes (20). Because the potency and efficacy of TCN-201 and Ro 25-6981 have been shown to be reduced for triheteromeric GluN1/GluN2A/GluN2B channels, especially for Ro 25-6981 (29,34), the lack of effect by Ro 25-6981 in our experiment does not allow us to exclude the involvement of GluN1/GluN2A/GluN2B heteromers, as both GluN2A and GluN2B (albeit less abundant compared with GluN2A based on Western blotting of total cell lysate) were detected in membrane fractions prepared from INS-1 832/13 cells by Western blotting (Fig. 1D).

Leptin regulates NMDAR activity via Src family kinases
In contrast to GluN1, which has a relatively small cytoplasmic domain of ;100 amino acids, GluN2A has a long cytoplasmic tail of ;600 amino acids that contains potential phosphorylation sites for a number of protein kinases, including PKA, PKC, CDK5, and the SFKs (22). Phosphorylation by PKC (35), CDK5 (36), and Src (25,37) in particular has been reported to potentiate NMDAR currents. We therefore monitored changes in INS-1 832/13 cell membrane potentials in response to leptin in the absence or presence of inhibitors for the various kinases. Neither roscovitine (Ros; 10 mM), an inhibitor of CDK5 kinase, nor Go 6983 (Go; 10 mM), a broad-spectrum PKC kinase inhibitor, had significant effects on leptin-induced hyperpolarization (DV m = 232. 4   Next, we determined whether activation of SFKs via an activating phosphopeptide EPQpYEEIPIYL (referred to as YEEI), which binds to the Src homology 2 domain to relieve kinase autoinhibition (39,40), could mimic the effects of leptin and induce membrane hyperpolarization. For these experiments, whole-cell current-clamp recordings of INS-1 832/13 cells were carried out to apply YEEI intracellularly through the patch pipette. Under whole-cell conditions with 5 mM ATP in the pipette solution (estimated intracellular [ATP] with 11 mM glucose in the bath solution), inclusion of YEEI phosphopeptide induced significant membrane hyperpolarization (DV m = 221.1 6 4.1 mV from break-in to steady state, n = 7) compared with the control without the peptide (DV m = 24.9 6 4.0 mV from break-in to steady state, n = 6) (Fig. 2, C and D). The result supports the notion that direct activation of SFKs is sufficient to mimic the effect of leptin and cause b-cell membrane hyperpolarization.
To directly test whether SFKs underlie the potentiation of NMDAR currents by leptin, we performed whole-cell recording of NMDAR currents and monitored how current amplitudes were affected by leptin and the SFK inhibitor AZD0530. In the absence of leptin, repeated puff applications of 1 mM NMDA at 1-min intervals elicited NMDAR currents of similar amplitudes. Bath application of 10 nM leptin potentiated NMDA-evoked currents within 5 min of leptin treatment from a baseline of 14.8 6 8.0 pA to 25.8 6 13.3 pA (p , 0.05 by Friedman's test; Fig. 3), as we reported previously (19). Subsequent co-application of 10 mM AZD0530 with leptin abolished the effect of leptin, with averaged NMDA currents of 15.5 6 8.7 pA comparable with baseline values (Fig. 3). These results provide direct evidence that leptin-mediated potentiation of NMDAR currents requires SFKs.
Leptin modulation of NMDAR activity requires phosphorylation of GluN2A at Tyr-1292 and Tyr-1387 SFKs have been reported to potentiate NMDAR currents by phosphorylating GluN2 (23,41,42). Specifically, three tyrosine residues in GluN2A (Tyr-1292, Tyr-1325, and Tyr-1387) have been identified as targets of SFK-mediated phosphorylation to enhance NMDAR function in different experimental systems (24,25,41,43). To determine whether these sites play a role in leptin-induced membrane hyperpolarization, we mutated the three tyrosine residues in GluN2A to phenylala-nines together GluN2A Y1292F,Y1325F,Y1387F as well as individually (GluN2A Y1292F , GluN2A Y1325F , and GluN2A Y1387F ). To facilitate visualization of expression, a GluN2A construct fused to GFP at its extracellular N terminus was used (44). INS-1 832/13 cells were then transfected to express WT or phosphomutants. To confirm that exogenously expressed GFP-GluN2A co-assembles with endogenous GluN1, we performed co-immunoprecipitation experiments using an anti-GFP nanobody. Both GluN1 and GFP-GluN2A were present in the immunoprecipitate as detected by anti-GFP, anti-GluN2A, and anti-GluN1 antibodies in Western blots (Fig.  4A), indicating association of transfected GluN2A with endogenous GluN1. Surface staining of the GFP tag further demonstrates that transfected WT and mutant GluN2A were expressed in the plasma membrane (Fig. 4B).
Changes in membrane potential following leptin treatment were again monitored using cell-attached current-clamp recording. We first compared cells expressing GFP-GluN2A WT with those expressing the GFP-GluN2A Y1292F,Y1325F,Y1387F triple phenylalanine mutant. GFP-negative cells on the same coverslip were also examined as untransfected controls. Representative membrane potential traces from untransfected cells and cells expressing GFP-GluN2A WT or triple phosphomutant GFP-GluN2A Y1292F,Y1325F,Y1387F are shown in Fig. 4C. Leptin application to untransfected cells induced a mean hyperpolarization of 251.1 6 5.6 mV (n = 10) that was not statistically significantly different from leptin responses in cells expressing GluN2A WT (248.0 6 6.5 mV, n = 13). By contrast, leptin had little hyperpolarizing effect in cells expressing the triple phosphomutant ( Fig. 4, C and D; GluN2A Y1292F,Y1325F,Y1387F DV m = 212.0 6 1.7 mV; n = 18; p , 0.0005 by unpaired t test compared with untransfected cells). The diminished response is not due to disruption of K ATP channel gating or expression as the K ATP activator diazoxide (200 mM) was fully able to hyperpolarize cells expressing GluN2A Y1292F,Y1325F,Y1387F (DV m = 252.7 6 4.9 mV; n = 14) (Fig. 4D). Moreover, we tested the response of cells expressing GluN2A Y1292F,Y1325F,Y1387F to NMDA, which we have shown hyperpolarizes cells in the absence of leptin (19). We found that NMDA hyperpolarized INS-1 832/13 cells by an average of 238.8 6 7.0 mV (n = 11), suggesting that the GluN2A phosphomutant also did not disrupt NMDAR expression or function (Fig. 4D). To determine the relative contributions of the three tyrosine residues, we next tested the membrane potential response to leptin in cells expressing Glu-N2A Y1292F , GluN2A Y1325F , or GluN2A Y1387F . In this cohort of cells, leptin application caused a mean hyperpolarization of 247 6 3.8 mV in untransfected cells (n = 26) (Fig. 4, E and F). Interestingly, leptin still hyperpolarized cells transfected with GluN2A Y1325F (DV m = 241.7 6 4.1 mV; n = 16), not significantly different from untransfected cells. By contrast, in cells expressing GluN2A Y1292F or GluN2A Y1387F , the response to leptin was significantly reduced (Fig. 4, E and F; DV m for GluN2A Y1292F = 217.4 6 4.1 mV, n = 19, and DV m for GluN2A Y1387F = 224.3 6 4.7 mV, n = 15; p , 0.0005 by unpaired t test). Taken together, these results identify the tyrosine phosphorylation sites Tyr-1292 and Tyr-1387 in GluN2A as responsible for mediating the leptin effect.

Leptin activates Src kinase in b-cells
The above results suggest that leptin likely activates SFKs to modulate NMDAR currents. Studies of GluN2A phosphorylation using in vitro or the HEK293 cell heterologous expression systems have found that both Src and Fyn can phosphorylate GluN2A (24). Src, when activated, autophosphorylates a highly conserved tyrosine residue, Tyr-418, in the catalytic domain of Src kinase, which is required for its full catalytic activity (45,46). Increased phosphorylation of Tyr-418 is therefore indicative of kinase activation. We used an antibody raised against a peptide containing Src-pY418 to monitor Src phosphorylation to directly test whether leptin activates Src. Immunocytochemistry experiments were carried out on INS-1 832/13 cells treated with or without leptin (10 nM for 10 min) in the presence or absence of the tyrosine kinase inhibitor dasatinib (50 mM). We found that leptin induced a significant increase in Src-pY418 staining as compared with matched controls with prominent staining at the cell periphery (Fig. 5, A and B) that was inhibited by co-application of dasatinib (Fig. 5, A and B). Further biochemical experiments using similar treatment conditions were carried out. In good agreement with the immunostaining results, leptin increased the ratio of phosphorylated to total Src by 176.0 6 30.0% as compared with controls (p , 0.05 by unpaired t test), which was reduced to 102.1 6 40.4% of controls when co-applied with dasatinib (Fig. 5, C and D). Because the Src-pY418 antibody also recognizes conserved corresponding phosphopeptide in other SFKs, including Fyn, and because both Src and Fyn are expressed in b-cells (47), we further tested the requirement of Src activity in leptin-induced response by expressing a dominant-negative kinase-dead Src mutant (48) (see "Experimental procedures") in INS-1 832/13 cells. For this, we monitored leptin-induced increase of K ATP channel surface expression by surface biotinylation, as described in our previous studies (7,9). As expected, leptin caused an increase in surface biotinylated SUR1, the regulatory subunit of the b-cell K ATP channel, in control cells. However, in cells transfected with the kinase-dead Src mutant leptin failed to show an increase in surface biotinylated SUR1 (Fig. 5, E and F). This result lends further support to the requisite role of Src activity in mediating the effect of leptin on K ATP channel trafficking.
Leptin signaling through Src-mediated phosphorylation of GluN2A is conserved in human b-cells We have previously shown that leptin induces hyperpolarization that depends on NMDARs in human b-cells as in INS-1 832/13 cells (19). Having found that leptin potentiates NMDARs via Src-mediated phosphorylation of GluN2A in INS-1 832/13 cells, we next tested whether the same mechanism applies to human b-cells. Cell-attached current-clamp recordings were made in human b-cells dissociated from islets Statistical analysis was conducted using Friedman's test (p = 0.0239) followed by a post hoc Dunn's multiple-comparison test with significance set to p , 0.05 (*), as compared with baseline. Inset, comparison of total charge (pA 3 ms) observed for each NMDA puff between leptin and leptin 1 AZD treatments normalized to baseline. **, p , 0.005, paired Student's t test as compared with leptin.
Leptin and GluN2A phosphorylation in b-cells of three different nondiabetic donors obtained through the Integrated Islets Distribution Program (IIDP). All donor information can be found in Table 1. Bath application of leptin alone induced a mean membrane hyperpolarization of 243.7 6 8.9 mV (n = 9 cells from three donors) (Fig. 6, A and B). The GluN2A-selective inhibitor TCN-201 largely eliminated the leptin response (DV m = 29.1 6 4.2 mV, n = 12 cells from 3 donors; p , 0.05, unpaired t test) (Fig. 6, A and B), suggesting that GluN2A-containing NMDARs underlie leptin signaling in human b-cells. We next tested whether leptin modulation of NMDARs and membrane potential in human b-cells requires Src. As shown in the example traces in Fig. 6C, the SFK inhibitor AZD0530 nearly abolished the ability of leptin to hyperpolarize human b-cells (DV m = 20.4 6 1.2 mV; n = 6 cells from two nondiabetic donors), whereas another SFK inhibitor, dasatinib, also diminished leptin-induced hyperpolarization, albeit to a lesser extent (DV m = 213.5 6 2.8 mV, n = 7 cells from three donors) compared with human b-cells treated with leptin alone (DV m = 233.5 6 6.0 mV; n = 17 cells from five nondiabetic donors; p , 0.0005 between AZD0530 and control, and p , 0.05 between dasatinib and control, Welch's t test) (Fig. 6D). Taken together, results from both INS-1 832/13 and human b-cells establish a mechanistic link between leptin receptor signaling and modulation of GluN2A-containing NMDARs via Src kinase.
Leptin signaling through NMDARs is disrupted in b-cells from db/db mice and from human type 2 diabetic donors Leptin resistance is a common pathological feature of type 2 diabetes associated with obesity (49). To test how the leptinsignaling pathway pertinent to this study is affected in diabetic b-cells, we examined b-cells from leptin-insensitive db/db mice, which harbors a mutation in the leptin receptor gene that renders ObRb signaling-defective (50), as well as from human obese type 2 diabetic donors. Cell-attached recordings of  (Fig. 7, A and B), suggesting that K ATP channels are functionally expressed in these mice. These results also provide clear evidence that leptin-induced hyperpolarization requires functional leptin receptors. Next, we tested whether direct activation of NMDARs using the receptor agonist NMDA can bypass   To test whether findings from the monogenic leptin-resistant mouse b-cells are also reproduced in human diabetic b-cells, we examined b-cells from obese (body mass index . 30) diabetic donors with presumably different genetic background through the IIDP (see Table 1 for donor information). We first tested whether these cells were resistant to leptin. Whole-cell recordings were performed to assess K ATP and Kv2.1 current density with or without leptin treatment (10 nM for 30 min) as described previously from b-cells taken from normal donors (7,9). In contrast to normal human b-cells (7,9), leptin did not increase the current density of K ATP channels (107.4 6 23.2 for control versus 75.4 6 11.7 pA/pF for leptintreated, n = 6) or Kv2.1 channels (683.3 6 146.3 for control versus 720.3 6 158.0 pA/pF for leptin-treated, n = 10) (Fig. 8A). Moreover, leptin failed to potentiate NMDAR currents in b-cells from diabetic donors (12.4 6 2.3 pA for control versus 11.3 6 1.8 pA following leptin treatment, n = 5) (Fig. 8B), in contrast to b-cells from normal donors we reported previously (19). Consistent with these results, leptin did not induce mem-brane hyperpolarization in diabetic human b-cells (DV m = 1.3 6 2.1 mV for leptin, n = 8) (Fig. 8, C and D). Conversely, application of 50 mM NMDA in the bath solution to directly activate NMDARs in b-cells from the same donors used to test leptin response did induce significant hyperpolarization (DV m = 222.0 6 8.6 mV, n = 9) (Fig. 8, C and D), indicating that NMDAR function and downstream signaling events were not compromised as was observed in b-cells from db/db mice. These results show that b-cells from a sample of obese diabetic donors were leptin-resistant but retained the ability to hyperpolarize in response to direct NMDAR activation.

Discussion
The results presented in this study elucidate the mechanism by which leptin potentiates NMDAR currents to regulate pancreatic b-cell excitability. Multiple lines of evidence support our conclusion that leptin modulation of b-cell membrane potential requires the phosphorylation of GluN2Acontaining NMDARs by Src kinase. First, application of TCN-201, a potent and highly selective inhibitor for GluN2Acontaining NMDARs, diminished leptin-induced membrane hyperpolarization in both INS-1 832/13 and human b-cells. Second, inhibition of SFKs blocked the ability of leptin to potentiate NMDAR currents and hyperpolarize b-cells. Third, leptin increased phosphorylation of Src at Tyr-418, a site that is required for its catalytic activity (45,46), and dominantnegative suppression of Src activity via a kinase-dead mutant prevented the ability of leptin to increase surface K ATP channels. Finally, mutation of known Src kinase phosphorylation sites, specifically Tyr-1292 and Tyr-1387, located in the C terminus of GluN2A prevented leptin-induced hyperpolarization. It is worth noting that Src has many downstream targets. For example, Src has been shown to regulate Kv2.1 activity (51) and actin dynamics (52,53). Therefore, results from experiments involving pharmacological or molecular manipulation of Src activity could be due to Src regulation of other targets in addition to phosphorylation of NMDARs and downstream trafficking of K ATP and Kv2.1 channels. Nonetheless, the collective evidence presented in this study together with our previously published studies (7,9,19) points to a signaling pathway in which leptin activates Src kinase to phosphorylate GluN2A of NMDARs, which leads to potentiation of NMDAR currents and increased Ca 21 influx. The increased Ca 21 influx then activates CaMKKb and AMPK, resulting in PKA-dependent actin depolymerization and trafficking of K ATP and Kv2.1 channels to the cell surface, which culminates in hyperpolarization of b-cell membrane potential and reduced insulin secretion. Given Ca 21 influx is normally associated with b-cell depolarization and insulin secretion, it is likely that the Ca 21 influx through NMDARs triggered by leptin is localized and specifically coupled to downstream events that regulate potassium channel trafficking and hyperpolarize the membrane. Importantly, we demonstrate in b-cells from mice deficient in leptin signaling and obese diabetic human donors that direct activation of NMDARs by NMDA can bypass leptin resistance and induce membrane hyperpolarization, suggesting that the signaling pathway downstream of Src kinase modulation of NMDARs remains operational and may be targeted to modulate insulin secretion.

Mechanism of NMDAR potentiation by leptin
NMDARs are allosterically modulated by a variety of endogenous extracellular ions. For example, Mg 21 can bind within the pore region of NMDAR and cause a voltage-dependent block (20). In addition, Zn 21 can promote voltage-dependent and voltage-independent inhibition of NMDAR activity (54-56). Voltage-dependent inhibition, like that observed with Mg 21 , occurs at a low-affinity GluN1 Zn 21 -binding site located within the channel pore (56), whereas voltage-independent inhibition appears to be mediated by a high-affinity Zn 21 -binding site located outside the channel pore and is associated with GluN2 subunits (57). Of the GluN2 subunits, GluN2A is 200-fold more sensitive to Zn 21 than GluN2B (58,59). The binding of Zn 21 to these subunits is thought to trap the receptor in a low open probability state, reducing their activity (54, 55). Zn 21 has been shown to be co-released with several transmitters, including glutamate, and at some glutamatergic synapses within the brain, vesicular Zn 21 is thought to diminish NMDAR activity (60). Interestingly, b-cells contain and release high levels of Zn 21 due to its role in the biosynthesis and packaging of insulin (61). We speculate that NMDARs residing on the b-cell membrane would be subjected to high concentrations of Zn 21 inhibition during bouts of insulin secretion. Of note, phosphorylation of GluN2A by Src has been shown to potentiate NMDAR currents by reducing tonic inhibition of the receptor by Zn 21 (23,25). A possible scenario is that under high glucose concentrations when b-cells are depolarized to relieve external Mg 21 block, NMDAR activity is still limited by Zn 21 co-released with insulin. However, inhibition by Zn 21 could be reduced by phosphorylating GluN2A via leptin-induced Src activation to regulate K ATP and Kv2.1 surface expression and tune insulin secretion. In this way, GluN2A phosphorylation affords a mechanism to allow modulation of b-cell response to glucose stimulation by additional inputs such as leptin. Interestingly, in cells overexpressing GluN2A phosphomutants that failed to respond to leptin, bath application of 50 mM NMDA still induced membrane hyperpolarization. This suggests that NMDAR currents activated by endogenously released ligands and potentiated by leptin represent a fraction of total NMDAR currents that were activated by 50 mM NMDA under our experimental conditions.

Mechanism of Src activation by leptin
Expression of ObRb mRNAs in b-cells has been well-documented (1,6,(62)(63)(64)(65). It is assumed that leptin binds to ObRb expressed by pancreatic b-cells to suppress glucose-stimulated insulin secretion. However, leptin has also been reported to transactivate other cytokine receptor signaling molecules, including epidermal growth factor receptor, type 1 insulin-like growth factor receptor, low-density lipoprotein receptorrelated protein, and vascular endothelial growth factor receptor (13). Our finding that db/db b-cells lack leptin-induced hyperpolarization lends strong support to the requirement of ObRb for leptin regulation of K 1 channel trafficking. The mechanism by which ObRb activation leads to Src activation in b-cells awaits further investigation. The enzymatic activity of Src is tightly regulated by tyrosine phosphorylation at Tyr-418 in the catalytic domain and Tyr-529 at its C-terminal tail. When Tyr-529 is phosphorylated, the C-terminal tail acts as an autoinhibitory peptide to block kinase activity, and its dephosphorylation relieves autoinhibition, leading to autophosphorylation of Tyr-418, which further activates the kinase (38). Activation of ObRb is known to activate several tyrosine phosphatases, such as Shp2, which could dephosphorylate pY529 to activate the kinase (13,38). Alternatively, direct recruitment of Src via the kinase's Src homology domains to phosphotyrosine in the activated leptin receptor complex or regulation by other kinases activated by leptin are also possibilities (38).
Tyrosine phosphorylation sites in GluN2A and GluN2B have been extensively investigated in recombinant expression systems and in neurons. The three GluN2A tyrosine residues examined in our study have all been shown to contribute to NMDAR current modulation (24,25,43). It is interesting that our data suggest that phosphorylation of Tyr-1292 and Y1387F but not Tyr-1325 are important for leptin response. This is in contrast to previous findings in the striatum neurons that Tyr-1325 is critical for Src-induced increase of NMDAR activity to regulate depression-related behavior (43). Very recently, Bland et al. (66) showed that leptin controls glutamatergic synaptogenesis and NMDAR trafficking via GluN2B phosphorylation by Fyn. Thus, the precise mechanisms and consequences of NMDAR modulation by leptin and SFKs are likely to be cell context-dependent.

Implications for leptin resistance and type 2 diabetes
We find that b-cells from obese type 2 diabetic human donors no longer hyperpolarize in response to leptin, as was also seen in b-cells from leptin-resistant, obese diabetic db/db mice. Examination of K ATP and Kv2.1 currents confirms that the lack of hyperpolarization is due to the lack of increase in trafficking of these channels to the cell surface, indicating that leptin signaling was disrupted. Interestingly, we found that NMDAR current density was similar in b-cells from normal and obese diabetic donors with detectable currents (12.26 6 4.12 pA/pF from normal donors, n = 16 (total of 35 cells from five donors (Table 1); 19 cells had no detectable currents) versus 11.60 6 2.17 pA/pF from obese diabetic donors, n = 20 (total of 26 cells from two donors ( Table 1); 6 cells had no detectable currents). Moreover, direct activation of NMDAR by NMDA triggered membrane hyperpolarization in both db/db b-cells and b-cells from human diabetic donors. Thus, despite leptin resistance, signaling mechanisms downstream of NMDARs (i.e. activation of CaMKKb, AMPK, and PKA-dependent actin depolymerization we reported previously (7,9,19)) remain functional to promote K ATP and Kv2.1 channel trafficking to modulate b-cell excitability. The ability of leptin to temper glucose-stimulated insulin secretion has been proposed as part of an adipoinsular feedback loop between adipocytes and b-cells to prevent excessive secretion of insulin, an adipogenic hormone, thereby limiting fat mass (67). Disruption of leptin signaling in b-cells has been shown to disrupt glucose homeostasis in some animal models (64,65), although others argue against a major role of b-cell leptin signaling in glucose homeostasis (68). Whether the controversial findings were due to different mouse strains and genetic models used (69) remains to be resolved. Nonetheless, it is tempting to speculate on the potential of exploiting the leptin signaling pathway we have identified for the prevention and treatment of obesity-associated type 2 diabetes. In obese individuals, hyperleptinemia may result in leptin resistance (49,70), which may lead to excessive insulin secretion. Hyperinsulinemia can then cause insulin resistance and hyperglycemia, further fueling obesity, leading to a loss of glucose control and eventually b-cell failure as seen in type 2 diabetes (71). Based on our finding that obese diabetic human b-cells, despite not being able to respond to leptin, retain response to NMDAR activation, stimulating NMDAR in leptin-resistant obese, prediabetic individuals may help to prevent excessive insulin secretion and thereby prevent or slow the development of type 2 diabetes (71). On the other hand, inhibition of NMDARs in individuals who have already developed type 2 diabetes may stimulate insulin secretion and alleviate hyperglycemia, as has been reported by Marquard et al. (17).
In summary, the current study identifies a novel leptin-signaling mechanism in b-cells wherein leptin activates Src kinase to phosphorylate GluN2A-containing NMDARs and potentiate NMDAR activity, thereby reducing b-cell excitability and insulin secretion. The findings build on our previously identified signaling pathway and provide a molecular explanation for the action of leptin. Moreover, they raise the therapeutic potential of targeting the pathway for the prevention and treatment of type 2 diabetes.

Dissociation of human pancreatic b-cells
Human b-cells were dissociated from human islets obtained through the Integrated Islets Distribution Program, as described previously (7,9,19). 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% BSA, pH 7.4. Dissociated cells were then plated on 0.1% gelatin-coated coverslips placed in 35-mm culture dishes (Falcon 35-3002). For electrophysiological experiments, b-cells were initially identified using the high autofluorescence signature of b-cells to 488-nm excitation, as these cells have high concentrations of unbound flavin adenine dinucleotide (72,73). Dithizone (Sigma-Aldrich) staining was then used to further confirm b-cell identity at the end of each recording (74). Donor information for specific experiments is provided in Table 1.

Dissociation of mouse pancreatic b-cells
Mice were purchased from the Jackson Laboratory (Bar Harbor, ME, USA): control (C57BL/6J) and db/db [B6.BKS(D)-Lepr db /J]. Pancreatic islets were isolated from the mice as described previously (75) and were performed in compliance with institutional guidelines and approved by the Oregon Health and Science University Animal Care and Use Committee. Following isolation, islets were cultured overnight in FBSsupplemented RPMI 1640 medium at 37°C and 5% CO 2 . Pancreatic b-cells were dissociated from islets using Spinners/ EGTA solution (116 mM NaCl, 5.37 mM KCl, 0.8 mM MgSO 4 , 26 mM NaHCO 3 , 11.67 mM NaH 2 PO 4 , 5.5 mM D-glucose, 3 mM EGTA, 1% BSA, 1% phenol red (pH 7.4)). Islets were incubated twice in Spinners/EGTA for 10 min with slight agitation every 5 min. In between incubations with Spinners/EGTA, the islets were washed with FBS-supplemented RPMI 1640 medium. Dissociated cells were plated on glass coverslips coated with 1% gelatin (Sigma-Aldrich) and cultured overnight. The identity of individual b-cells was confirmed by dithizone staining at the end of each experiment (74).
Immunocytochemistry INS-1 832/13 cells were fixed in 2% paraformaldehyde in PBS for 10 min at room temperature, permeabilized with 0.2% Triton X-100 in 1% BSA/PBS, and blocked for 60 min with 1% BSA in PBST (PBS 1 0.1% Tween 20) before being incubated overnight at 4°C with primary rabbit polyclonal antibodies directed against phospho-Src Y418 (Abcam, ab4816). Proteins were visualized using Alexa 488-conjugated secondary antibodies. Fluorescent images were acquired using a Zeiss LSM780 confocal microscope equipped with a 363 oil immersion objec-tive. Images were processed and analyzed using NIH ImageJ software (76). For surface staining of GFP-GluN2A WT and GFP-GluN2A Y1292F,Y1325F,Y1387F , cells were incubated in cold DPBS containing anti-GFP antibody (1:100; ThermoScientific, G10362) at 4°C for 30 min. Cells were then fixed as described above, and surface GFP was visualized using a Cy3-conjugated secondary antibody.

Surface biotinylation
INS-1 832/13 cells were incubated in RPMI 1640 for 1 h at 37°C prior to a 30-min treatment with vehicle or leptin (10 nM). Cells were then washed four times with cold DPBS and incubated with 1 mg/ml EZ-Link Sulfo-NHS-SS-Biotin (Pierce) in DPBS with vehicle or leptin for 30 min at 4°C. The reaction was terminated by incubating cells twice for 5 min with DPBS containing 50 mM glycine followed by two washes with cold DPBS. Cells were then lysed in 300 ml of lysis buffer as described above, and 500 mg of total lysate was incubated with 50 ml of an 50% slurry of NeutraAvidin-agarose beads (Pierce) overnight at 4°C. Biotinylated proteins were eluted with 23 protein loading buffer for 15 min at 37°C. Both eluent and input samples (50 mg of total cell lysate) were analyzed by immunoblotting using anti-SUR1 antibody (raised against a hamster SUR1 C-terminal peptide KDSVFASFVRADK) as described previously (7).

Electrophysiology
Electrophysiological recordings were conducted using an Axon 200B amplifier (Molecular Devices, Sunnyvale, CA, USA) with Clampex 9.2 (pCLAMP) software. Signals were acquired at 20 kHz and filtered at 2 kHz. Recording electrodes (tip resistances ranged between 3 and 6 megaohms) were pulled from nonheparinized Kimble glass (Thermo Fisher Scientific, Waltham, MA, USA) using a P-97 micropipette puller (Sutter Instruments). For whole-cell recording of NMDA currents, external Tyrode's solution contained 137 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl 2 , 0.5 mM MgCl 2 , 5 mM Na-HEPES, 3 mM NaHCO 3 , and 0.16 mM NaH 2 PO 4 , (pH 7.2). The external solution was supplemented with 0.1 mM glycine and 11 mM glucose, and in some experiments Mg 21 was omitted as specified in the figure legends. The internal pipette solution contained 140 mM potassium gluconate, 6 mM EGTA, 10 mM HEPES, 5 mM K 2 ATP, 1 mM CaCl 2 (pH 7.2). To induce NMDA-mediated currents, cells were held at 270 mV, and NMDA (1 mM) was puffed (3-5 p.s.i. for 0.5 s) with carbogen using 1-0.5-megaohm micropipettes connected to a Multi-function Microforge Controller DMF1000 (World Precision Instruments, Sarasota, FL, USA) equipped with a pressure regulator. In experiments measuring NMDAR current response to Mg 21 block (Fig. 1B), leptin (Figs. 3 and 8B), and leptin plus AZD (Fig. 3), 1 mM NMDA was puffed repeatedly at 1-min intervals 3-5 times to establish control baseline. Only cells that showed reproducible currents to repeated NMDA puffs during control baseline measurements were subsequently treated with 100 mM Mg 21 , 10 nM leptin, or leptin plus AZD for 5 min, after which 1 mM NMDA was again puffed at 1-min intervals 3-5 times. The currents before and after treatments were then averaged, and the averaged values are shown as individual data points in the figures.
Whole-cell K ATP and Kv2.1 current recordings (Fig. 8A) were conducted as described previously (7,9). For K ATP currents, cells were held at 270 mV, and K ATP currents were recorded at two voltage steps (250 and 290 mV) applied every 2 s. Pipette solution contained 140 mM KCl, 10 mM K-HEPES, 1 mM K-EGTA, pH 7.3. The bath solution contained 137 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl 2 , 0.5 mM MgCl 2 , 5 mM Na-HEPES, 3 mM NHCO 3 , 0.16 mM NaH 2 PO 4 , pH 7.2. Diazoxide (200 mM) was applied to the bath solution immediately after break-in to maximally stimulate K ATP channels. After the current had plateaued, 300 mM tolbutamide (a K ATP channel inhibitor) was applied, and residual currents were subtracted from maximal currents observed in diazoxide and divided by cell capacitance to obtain K ATP current density. For Kv2.1, micropipettes were filled with an internal solution containing 140 mM KCl, 1 mM CaCl 2 , 2 mM MgCl 2 , 5 mM EGTA, 5 mM ATP, 10 mM glucose, and 10 mM HEPES, pH 7.3. The bath solution contained the following: 140 mM NaCl, 5 mM KCl, 4 mM MgCl 2 , 11 mM glucose, 10 mM HEPES, pH 7.3. Ca 21 was excluded from the bath solution to eliminate calcium channel currents. A 30-ms prepulse to 210 mV was used to inactivate transient potassium channel currents and voltage-dependent Na 1 currents. The sustained current at 180 mV, after subtracting currents remaining in a 10 mM concentration of the potassium current blocker tetraethylammonium, was divided by cell capacitance for Kv2.1 current density calculation.
For cell-attached recording to monitor membrane potentials, micropipettes were filled with 140 mM NaCl. The bath solution contained 137 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl 2 , 0.5 mM MgCl 2 , 5 mM Na-HEPES, 3 mM NHCO 3 , 0.16 mM NaH 2 PO 4 , 11 mM glucose, pH 7.2. The liquid junction potential was measured to be 210 mV and corrected from the recorded V m . Seal resistances between the recording pipette and the cell membrane ranged between 4 and 11 gigaohms, and membrane potentials were monitored in current-clamp mode (I = 0). After correction for the liquid junction potential, the average starting V m was around 24 mV. Note that the V m estimated in this configuration depends on the ratio of the seal resistance and the combined patch and cell resistance, where the ratio of recorded V m and true membrane potential = (R seal /R patch 1 R cell )/(1 1 (R seal /R patch 1 R cell )) (32). As such, the recorded V m is an under-estimate of the actual membrane potential (i.e. more depolarized than the actual V m ) (31,32). Because the recorded V m under this configuration is only a proxy of the true membrane potential, we only used it to track changes in membrane potential (31,32). Seal resistance was monitored before and after the recording, and only cells that showed stable baseline membrane potential prior to leptin/drug application and that maintained good seal resistance were included for analysis. In addition, the K ATP channel opener diazoxide or inhibitor tolbutamide was used when applicable to ensure cell and patch integrity as we reported previously (19) and in the current study (see Figs. 4 and 7). For whole-cell current-clamp recordings of membrane potential (Fig. 2, C and D), pipette solution contained 130 mM potassium gluconate, 10 mM KCl, 10 mM HEPES, 6 mM EGTA, 0.1 mM CaCl 2 , 1 mM MgCl 2 , and 5 mM ATP, with or without the SF activator YEEI phosphopeptide (EPQpYEEIPIYL at 1 mM). All recordings were analyzed using Clampfit 9.2 (pCLAMP).

Statistical analysis
Results are expressed as mean 6 S.E. One-way analysis of variance with a post hoc Dunnett's test (Fig. 5E) or Friedman's test (Fig. 3B) was used for multiple comparisons where different experimental conditions were carried out side by side. For most experiments, difference between control and treated groups was tested using a paired t test or unpaired t tests (for comparable sample sizes) or Welch's t test (for significantly unequal sample sizes in Fig. 6D) as detailed in the figure legends. The level of statistical significance was set at p , 0.05.

Data availability
All data are contained in the article.
Acknowledgments-We thank Dr. Christopher Newgard for the rat insulinoma INS-1 clone 832/13 cells. Funding and additional information-This work was supported by National Institutes of Health Grants R01DK057699 and 3R01DK057699-14S1 (to S.-L. S.). P. K. was supported by National Institutes of Health Grant P51 OD011092 (to the Oregon National Primate Research Center). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflict of interest-The authors declare that they have no conflicts of interest with the contents of this article.