Antagonism of Rat β-Cell Voltage-dependent K+ Currents by Exendin 4 Requires Dual Activation of the cAMP/Protein Kinase A and Phosphatidylinositol 3-Kinase Signaling Pathways*

Antagonism of voltage-dependent K+ (Kv) currents in pancreatic β-cells may contribute to the ability of glucagon-like peptide-1 (GLP-1) to stimulate insulin secretion. The mechanism and signaling pathway regulating these currents in rat β-cells were investigated using the GLP-1 receptor agonist exendin 4. Inhibition of Kv currents resulted from a 20-mV leftward shift in the voltage dependence of steady-state inactivation. Blocking cAMP or protein kinase A (PKA) signaling (Rp-cAMP and H-89, respectively) prevented the inhibition of currents by exendin 4. However, direct activation of this pathway alone by intracellular dialysis of cAMP or the PKA catalytic subunit (cPKA) could not inhibit currents, implicating a role for alternative signaling pathways. A number of phosphorylation sites associated with phosphatidylinositol 3 (PI3)-kinase activation were up-regulated in GLP-1-treated MIN6 insulinoma cells, and the PI3 kinase inhibitor wortmannin could prevent antagonism of β-cell currents by exendin 4. Antagonists of Src family kinases (PP1) and the epidermal growth factor (EGF) receptor (AG1478) also prevented current inhibition by exendin 4, demonstrating a role for Src kinase-mediated trans-activation of the EGF tyrosine kinase receptor. Accordingly, the EGF receptor agonist betacellulin could replicate the effects of exendin 4 in the presence of elevated intracellular cAMP. Downstream, the PKCζ pseudosubstrate inhibitor could prevent current inhibition by exendin 4. Therefore, antagonism of β-cell Kv currents by GLP-1 receptor activation requires both cAMP/PKA and PI3 kinase/PKCζ signaling via trans-activation of the EGF receptor. This represents a novel dual pathway for the control of Kv currents by G protein-coupled receptors.

a novel dual-signal pathway mechanism for the regulation of Kv currents. Additionally, the present work suggests a potential mechanism contributing to the glucose dependence of the insulinotropic effect of GLP-1.
Electrophysiology-Cells were patch-clamped in the whole-cell configuration using an EPC-9 amplifier and PULSE software (HEKA Electronik, Lambrecht, Germany). Patch pipettes were prepared from 1.5-mm thin walled borosilicate glass tubes using a two-stage micropipette puller (Narishige, Tokyo, Japan) and had typical resistances of 3-6 megaohms when fire polished and filled with intracellular solution containing the following (in mM): KCl, 140; MgCl 2 ⅐6H 2 0, 1; EGTA, 1; HEPES, 10; MgATP, 5; pH 7.25 with KOH. Extracellular solutions contained the following (in mM): NaCl, 140; CaCl 2 , 2; KCl, 4; MgCl 2 ⅐6H 2 O, 1; HEPES, 10; pH 7.30 with NaOH. ␤-Cells were positively identified by the lack of an inward voltage-dependent Na ϩ current from a holding potential of Ϫ70 mV (15,16). Outward currents were elicited from a holding potential of Ϫ70 mV by a series of 500-ms depolarizing pulses in 20-mV increments to ϩ70 mV or by a single depolarizing pulse to ϩ30 mV for 500 ms or 8 s. Sustained outward current was taken as the mean current during the final 25 ms of the depolarization. The voltage dependence of steady-state inactivation was investigated by holding the cells at potentials from Ϫ80 to ϩ30 mV for 15 s followed by a 5-ms pre-pulse to Ϫ70 mV and a 500-ms depolarization to ϩ30 mV to elicit outward currents.
Electrophysiological recordings were obtained at 32-35°C and normalized to cell capacitance. The bath (ϳ0.5 ml) was perfused continuously at 1 ml/min (Minipuls peristaltic pump; Gilson Inc., Middleton, WI). Currents were filtered upon acquisition at 10 kHz, and capacitive transients were cancelled with the automatic compensation functions of the Pulse software (HEKA Electronik). Whole-cell conductance was calculated as G ϭ I/(V Ϫ V eq ), where G is whole-cell conductance, I is current, V is the test potential, and V eq is the equilibrium potential of K ϩ (Ϫ94.1 mV under the present conditions). Current-voltage relationships were compared by repeated measures analysis of variance followed by a Tukey post-test to compare currents elicited at various voltages. Voltage dependence of activation and steady-state inactivation curves were fit with a Boltzman function: where G is whole-cell conductance, V is test potential, V 50 is the voltage at which half the channels are inactivated, and s is the slope of the curve, using Prism 3.03 (Graphpad Software, San Diego, CA). Time constants of activation or inactivation were derived by fitting the waveform with either a single or double exponential function using PulseFit software (HEKA Electronik) and compared using the Student's t test (two groups).
Immunoblotting was performed as described previously (5). Twentyfive g of protein from each sample was separated on a 10% polyacrylamide gel and transferred to PVDF-Plus TM membrane (Fisher). Primary antibodies (see above) were detected with appropriate secondary antibodies (sheep anti-mouse, 1:3000 and donkey anti-rabbit, 1:7500; Amersham Biosciences) for 1 h at room temperature. Visualization was by chemiluminescence (ECL; Amersham Biosciences) and exposure to Kodak film (Eastman Kodak Co.) for 5 s to 10 min.
Rat ␤-cells were voltage-clamped in the whole-cell configuration at near-physiological temperatures (32-35°C). Outward voltage-dependent K ϩ currents were similar to those we have reported previously (4, 13, 18) (Fig. 1B) and could be blocked with the general Kv channel antagonist tetraethylammonium (15 mM) (Fig. 1B). Maximum steady-state outward currents were blocked by 40.8 Ϯ 4.1% (n ϭ 9, p Ͻ 0.001) upon perfusion of GLP-1 (10 Ϫ8 M) (Fig. 1B). To ensure that currents were stable, control currents were recorded every min for ϳ5 min prior to addition of GLP-1. Kv currents were maximally inhibited at ϳ6 min after exposure to GLP-1 (Fig. 1C). Partial washout of this effect was observed (not shown). However, because of technical difficulties in recording from cells at 32-35°C for prolonged periods (Ͼ20 min), experiments were not of sufficient duration to observe complete washout.
cAMP/PKA Signaling Is Necessary but Not Sufficient for Antagonism of Kv Currents by Exendin 4 -The importance of cAMP/PKA signaling for the insulinotropic effect of GLP-1 receptor activation is well known (9). To investigate the involvement of cAMP and PKA signaling in the inhibitory effect of exendin 4 on Kv currents, ␤-cells were pre-treated ( outward K ϩ currents in cells pre-treated with either antagonist were not different from untreated cells and could not be antagonized significantly by treatment with exendin 4 (10 Ϫ8 M) (Fig.  3A). These results demonstrate the necessity of the cAMP/PKA signaling pathway for Kv current inhibition by exendin 4.
Because activation of the cAMP/PKA pathway by GLP-1 is mediated by G-protein activation of adenylyl cyclase (9), we investigated whether general G-protein activation was sufficient to reproduce the inhibitory effect of GLP-1 receptor activation on ␤-cell Kv channels. Inclusion of the non-hydrolysable GTP analogue GMP-PNP (10 nM) in the pipette solution decreased outward K ϩ current (by 55.7 Ϯ 6.0%, n ϭ 8, p Ͻ 0.001) similar to exendin 4 (Fig. 3B). We next investigated whether direct activation of the cAMP/PKA pathway could replicate this inhibitory effect. Addition of cAMP (100 M) to the pipette solution was unable to decrease voltage-dependent outward K ϩ currents in rat ␤-cells (n ϭ 7) (Fig. 3B). Also, intracellular dialysis of the constitutively active PKA catalytic subunit (200 units/ml) had no effect on currents compared with dimethyl sulfoxide controls (n ϭ 6), which were themselves not different from untreated cells (Fig. 3B). Forskolin (5 M) and isobutylmethylxanthine (100 M) together blocked currents by 35.6 Ϯ 7.5% (n ϭ 5, p Ͻ 0.05) (Fig. 3B). However, forskolin (and its inactive analogs) directly inhibit Kv currents in insulin secreting cells (20,21). Additionally, ␤-cell Kv currents could not be antagonized by treatment with the Epac-selective cAMP analog 8-pCPT-2Ј-O-Me-cAMP (50 M, n ϭ 7) (Fig. 3B). Our inability to antagonize voltage-dependent K ϩ channels in ␤-cells by activating the cAMP/PKA pathway is in agreement with previous reports (20,21) and suggests that additional pathway(s) are also required for the inhibitory action of GLP-1.
Phospho-site Screening Identifies PI3 Kinase as an Additional Candidate Pathway-To identify candidate signaling pathways that may be involved in the GLP-1 receptor-mediated reduction of voltage-dependent K ϩ currents, MIN6 insulinoma cells were treated with GLP-1 (10 Ϫ7 M) for 10 min, and protein lysates were screened with multiple phosphosite-specific antibodies using the Kinetworks TM KPPS 1.1 screen. MIN6 insulinoma cells were used in this case, because they allow for preparation of the necessary amount of protein lysates for the phospho-screen, and in our hands they exhibit good insulin secretory responses to GLP-1 (ϳ1.5-fold). Two separate experiments were conducted, and the -fold increases in phosphorylation of 40 sites on 30 different proteins are shown in Table I.
Interestingly, a number of known PDK1 phosphorylation sites are up-regulated. These include Akt1 Ser-473 and Thr-308, PKC␣/␤ Thr-638/Thr-641, and PKC␦ Thr-505 (Table I). Because PDK1 is a direct effector of PI3 kinase signaling, these results may be taken to suggest that this pathway is activated by GLP-1. Activation of PI3 kinase by GLP-1 is supported by the work of others (22)(23)(24). Additionally, PI3 kinase signaling may be involved in activation of the MAP kinase pathway either through the ability of PI3 kinase itself to phosphorylate proteins or through the recruitment of small G-proteins (25,26). Notably the MEK1/2 Ser-221/Ser-225 site was phosphorylated, whereas the MEK3 Ser-189/Thr-197 site was paradoxically dephosphorylated. This may be significant, because MEK1 phosphorylation, through the activation of the p110␥ isoform of PI3 kinase, has been implicated in the activation of MAP kinase (26), although it is not clear whether PI3 kinase can phosphorylate MEK1 directly. Although the present phospho-screen was unable to detect phosphorylated MAP kinase, others have implicated the activation of this pathway as the mechanism for the effects of GLP-1 on ␤-growth and differentiation (22,24,27). In the experiments below, the activation of the PI3 kinase signaling pathways is confirmed and investigated further using pharmacological antagonists.
PI3 Kinase and PKC Activity Are Necessary for Regulation of Kv Channels by Exendin 4 -In support of a role for PI3 kinase signaling, the effect of exendin 4 on Kv currents was completely abolished by wortmannin (100 nM, n ϭ 6), a PI3 kinase inhibitor (Fig. 4). Confirmation of this result with the structurally unrelated PI3 kinase antagonist LY294002 was not possible, because this compound is also a potent and direct antagonist of Kv channels (28). The downstream effectors of PI3 kinase responsible for regulation of Kv currents by GLP-1 receptor activation were investigated using a number of pharmacological antagonists. Antagonism of p70 S6 kinase (rapamycin, 10 nM, n ϭ 6), MEK1 (PD98059, 20 M, n ϭ 6), and conventional PKC isoforms (calphostin C, 500 nM, n ϭ 7) failed to prevent the reduction of Kv currents in rat ␤-cells by exendin 4 (10 Ϫ8 M) (Fig. 4). Possible downstream effectors of PI3 kinase include atypical PKC isoforms, because the effect of exendin 4 could also be prevented by bisindolylmaleimide (100 nM, n ϭ 7) (Fig. 4), which antagonizes PKC at the ATP binding site (29,30). However, it is important to note that this compound alone was capable of inhibiting ␤-cell Kv currents (Fig. 4). We investigated a role for PKC, because this isoform is known to be activated by GLP-1 (22) and associate with Kv channel regulatory subunits (31), and phosphorylation at a conserved PKC site is known to reduce current through cloned Kv channels (32,33). A role for the atypical PKC isoform is supported as intracellular dialysis of the PKC pseudosubstrate inhibitor (25 M), which had no effect alone, also prevented Kv current reduction by exendin 4 (Fig. 4).

Transactivation of PI3 Kinase via the EGF Receptor Is Necessary but Not Sufficient for Antagonism of Kv Channels by
Exendin 4 -The p110␥ isoform of PI3 kinase is the only isoform known to be directly activated by G-protein-coupled receptors, and this occurs through an interaction with G␤␥ (34). However, exendin 4 could inhibit Kv currents equally in ␤-cells from p110␥ ϩ/Ϫ (47.1 Ϯ 4.9%, n ϭ 4, p Ͻ 0.001) and Ϫ/Ϫ (53.7 Ϯ 5.9%, n ϭ 4, p Ͻ 0.001) mice (Fig. 5A). To investigate whether exendin 4 activates PI3 kinase through the direct activation of a receptor other than the GLP-1 receptor, we studied ␤-cells from GLP-1 receptor Ϫ/Ϫ mice. Exendin 4 had no effect on Kv currents in ␤-cells lacking the GLP-1 receptor, even in the presence of 100 M cAMP (Fig. 5A). Because one recent study (35) demonstrates that a Src kinase-mediated trans-activation of EGF receptors and subsequent activation of PI3 kinase is involved in the effects of GLP-1 on ␤-cell proliferation, we examined whether this was the mechanism of PI3 kinase activation in the present study. Both the EGF receptor tyrosine kinase antagonist AG1428 (250 nM, n ϭ 6) and the Src kinase antagonist PP1 (10 M, n ϭ 7) prevented exendin 4-mediated reductions in ␤-cell Kv currents (Fig. 5B). We detected EGF receptor (erbB   insulinoma cells (not shown), consistent with previous reports (35,36). We further examined whether activation of PI3 kinase alone could replicate the effect of exendin 4 on ␤-cell Kv currents. The EGF receptor agonist betacellulin (5 ng/ml), which activates PI3 kinase in INS-1 ␤-cells (35), reduced ␤-cell Kv currents somewhat (by 20.2 Ϯ 4.5, n ϭ 5, p Ͻ 0.05) (Fig. 6A) but could not completely replicate the effect of exendin 4. Activation of PI3 kinase with insulin (200 nM) was without effect (n ϭ 5) (Fig.  6A). Also, ␤-cell Kv currents were not antagonized by activation of conventional PKC isoforms (200 nM TPA, n ϭ 6) (Fig. 6A). These results suggest that Src kinase-mediated trans-activation of the EGF receptor and subsequent activation of PI3 kinase/PKC signaling is required, but not sufficient, for antagonism of ␤-cell Kv currents by exendin 4.
Betacellulin Can Fully Replicate the Effect of Exendin 4 on ␤-cell Kv Currents in the Presence of Increased cAMP-Because both cAMP/PKA signaling and EGF receptor activation are necessary, but not each sufficient in themselves for Kv current regulation by GLP-1 receptor activation in ␤-cells, we investigated the effect of activating both signaling pathways in parallel. To activate PKA, 100 M cAMP was included in the patch pipette. Under these conditions, insulin (200 nM) now tended to antagonize ␤-cell Kv currents (by 26.8 Ϯ 4.4%, n ϭ 5) although this did not reach statistical significance (Fig. 6A). However, betacellulin (5 ng/ml) was now able to antagonize Kv currents by 40.5 Ϯ 3.7% (n ϭ 6, p Ͻ 0.001) (Fig. 6, A and B). Betacellulin and cAMP together also caused a hyperpolarizing shift in the voltage dependence of steady-state inactivation (V 50 ϭ Ϫ62.0 Ϯ 2.0 mV, n ϭ 6, p Ͻ 0.001), entirely replicating the effect of exendin 4 (Fig. 6C). Betacellulin alone resulted in an intermediate shift (V 50 ϭ Ϫ55.7 Ϯ 2.3 mV, n ϭ 5, p Ͻ 0.05) (Fig. 6C). TPA (200 nM, n ϭ 6) did not reduce ␤-cell Kv currents in the presence of increased cAMP, which is consistent with a role for the atypical PKC isoform as opposed to conventional PKC isoforms activated by phorbol esters. These and the above results suggest that the GLP-1 receptor antagonizes ␤-cell Kv currents by activating both the cAMP/PKA and EGF receptor/ PI3 kinase signaling pathways. DISCUSSION Regulation of voltage-dependent K ϩ channel activity in pancreatic ␤-cells by GLP-1 may contribute significantly to the to the excitatory and insulinotropic effects of this incretin hormone (13). Importantly, because antagonism of ␤-cell Kv current stimulates insulin secretion in a glucose-dependent manner (4 -7), block of these channels may contribute to the well known glucose-dependent effects of GLP-1. Recently, alternative mechanisms for the glucose dependence of GLP-1 have also been proposed, including the ADP-dependent inhibition of K ATP channels by PKA (37) and the sensitization of Ca 2ϩinduced Ca 2ϩ release by activation of the guanine-nucleotide exchange factor Epac2 (38). However, the exact mechanism of glucose dependence remains to be established and may result from a complex interaction of the many known targets of GLP-1 signaling (9).
Rat ␤-cell Kv currents were antagonized by both GLP-1 and the GLP-1 receptor agonist exendin 4, confirming our previous results (13). Indeed, immunoprecipitation of a tagged GLP-1 receptor over-expressed in HIT-T15 cells was able to pull down the Kv2.1 channel ␣-subunit, demonstrating an association that may be direct or mediated by intermediate protein-protein interactions. GLP-1 blocked currents by ϳ6 min, whereas washout of the effect occurs over a longer time period (Ͼ20 min). These relatively slow time courses are consistent with the complex nature of the signaling pathway involved. Current block resulted from a leftward shift in the voltage dependence of steady-state inactivation. Functionally this means that after GLP-1 treatment a greater number of Kv channels will already be inactivated at a given membrane potential and therefore unavailable for action potential repolarization. This is expected to prolong the action potential, leading to greater influx of Ca 2ϩ and a greater insulin secretory response (13).
The G-protein-coupled GLP-1 receptor regulates the majority of its known targets through the classic cAMP/PKA signaling pathway (9). ␤-Adrenergic stimulation reduces Kv currents in lymphocytes (39) and enhances Kv currents in cardiac myocytes (40), and in both these tissues cAMP/PKA signaling has been implicated (41,42). Antagonism of rat ␤-cell Kv currents by exendin 4 is dependent on the cAMP/PKA signaling pathway as demonstrated by the ability of Rp-cAMPs and H-89 to block this effect (Fig. 3A). We were unable to replicate the effect of exendin 4 with activators of this pathway including cAMP and the constitutively active PKA catalytic subunit (Fig. 3B), which is consistent with previous reports in mouse ␤-cells (20) and the INS-1 insulinoma cell line (21). We were also not able to block Kv currents with the Epac-selective cAMP analog. The small but significant effect of forskolin and isobutylmethylxanthine cannot be taken as evidence that activation of cAMP signaling alone can block currents, because forskolin (and inactive analogs) is known to block Kv currents directly (20,21). It therefore seems clear from these results that GLP-1 receptor-mediated block of Kv currents requires an additional signaling pathway. Recent work has identified additional pathways activated by the GLP-1 receptor, including the PKA-independent activation of Epac2 (38) and EGF receptor trans-activation and subsequent activation of PI3 kinase (35). Phospho-site screening of MIN6 cells treated acutely with GLP-1 indicates that the PI3 kinase signaling pathway is activated. because a number of phosphorylation sites downstream of the PI3 kinase-activated PDK1 were up-regulated. This is not surprising, because signaling through the GLP-1 receptor results in phosphorylation of Akt (43) and activation of PKC (22) in INS-1 insulinoma cells. Additionally, the MEK1/2 Ser-221/Ser-225 phosphorylation site was up-regulated, and this may also be downstream of PI3 kinase (26). Our inability to detect up-regulation of ERK phosphorylation at 10 min following GLP-1 treatment may be expected, because ERK phosphorylation in response to GLP-1 is transient (detectable at 5 but not 15 min) in the absence of high glucose concentrations (44). These results suggest that GLP-1 treatment causes the phosphorylation of key activation sites of numerous downstream targets of PI3 kinase.
Functionally, activation of the PI3 kinase pathway was determined to be necessary for the ability of exendin 4 to antagonize ␤-cell Kv currents, because the effect could be completely abolished by the PI3 kinase inhibitor wortmannin (Fig. 4). Direct activation of the G-protein-regulated p110␥ isoform of PI3 kinase is not involved, because exendin 4 could still antagonize currents in ␤-cells from mice lacking p110␥. PI3 kinase was also not activated by a direct effect of exendin 4 on an alternate receptor, because currents in ␤-cells lacking the GLP-1 receptor were not blocked by exendin 4, even when the cAMP/PKA pathway was activated. A recent study (35) dem-onstrates that, in the INS-1 insulinoma cell line, GLP-1 receptor activation leads to PI3 kinase activation through a mechanism involving Src kinase-mediated trans-activation of the EGF receptor. The involvement of this pathway in exendin 4-mediated Kv current block was demonstrated using antagonists of Src family protein tyrosine kinases and EGF receptor tyrosine kinase activity, which were each able to prevent current inhibition by exendin 4. Although the phospho-site screen demonstrates a down-regulation of Src Tyr-418 phosphorylation, an effect expected to reduce activity, it should be noted that the antibody used also recognizes conserved sites on related kinases (45), and the Src family tyrosine kinase inhibitor used here (PP1) is non-selective among different Src family members (46).
Because activation of either cAMP/PKA signaling or PI3 kinase signaling (with insulin or betacellulin) alone was insufficient to replicate the effect of exendin 4 on ␤-cell Kv currents, we examined the effect of concomitant activation of both pathways. In the presence of increased intracellular cAMP, betacellulin could completely replicate the inhibitory action of exendin 4. TPA was still without effect, confirming no role for the typical PKC isoforms. Importantly, the mechanism of current block by exendin 4, an ϳ20-mV hyperpolarizing shift in the steady-state voltage dependence of inactivation, was completely replicated by treatment with cAMP and betacellulin. Despite the fact that both insulin and EGF receptor signaling activate PI3 kinase, we did not observe a significant reduction in Kv current upon treatment with insulin, even in the presence of high intracellular cAMP. This functional separation of the two signaling pathways is consistent with the inability of EGF receptor ligands to replicate the effect of insulin on glucose uptake in adipocytes (47,48) and may result from the differential activation of PI3 kinase isoforms (49) or the subcellular compartmentalization of signaling molecules (50,51). These results provide further evidence that both cAMP/PKA and EGF receptor-mediated PI3 kinase signaling are necessary for Kv current block in pancreatic ␤-cells.
Downstream of PI3 kinase, the effect of exendin 4 on ␤-cell Kv currents does not involve MAP kinase, P70 S6 kinase, or conventional isoforms of PKC, because inhibitors of these did not prevent antagonism of currents. Bisindolylmaleimide, which can inhibit both conventional and atypical PKC isoforms at the ATP binding site (29), was able to prevent the effect of exendin 4 on ␤-cell Kv currents. It should be noted that although antagonism of currents by exendin 4 could be prevented by bisindolylmaleimide, this PKC inhibitor alone caused a 40% reduction in Kv currents. A role for the atypical PKC isoform is supported, because the PKC pseudosubstrate, which, in contrast to bisindolylmaleimide, did not affect control currents, prevented antagonism of currents by exendin 4 (Fig. 4). The involvement of PKC in mediating current reduction is also supported by studies demonstrating that this isoform is known to associate with Kv channel regulatory subunits (31), and phosphorylation at a conserved PKC site is known to reduce current through cloned Kv channels (32,33).
Here we investigated the mechanism and signal transduction pathways involved in Kv current block by the GLP-1 receptor agonist exendin 4 in primary rat ␤-cells. Kv currents were antagonized as the result of an ϳ20-mV hyperpolarizing shift in the voltage dependence of inactivation, whereas voltagedependent activation and current kinetics were unaffected. The classic cAMP/PKA signaling pathway was necessary for this effect, as was the more recently recognized ability of the GLP-1 receptor to trans-activate the EGF receptor and subsequent PI3 kinase signaling. Importantly, activation of both these pathways together replicated the effect of exendin 4 on Kv currents. Antagonism of ␤-cell Kv currents by GLP-1 receptor activation may be an important component to the glucose-dependent insulinotropic effect of GLP-1, because antagonists of Kv currents are also glucose-dependent secretagogues.