Both Gi and Go Heterotrimeric G Proteins Are Required to Exert the Full Effect of Norepinephrine on the β-Cell KATP Channel*

The effects of norepinephrine (NE), an inhibitor of insulin secretion, were examined on membrane potential and the ATP-sensitive K+ channel (KATP) in INS 832/13 cells. Membrane potential was monitored under the whole cell current clamp mode. NE hyperpolarized the cell membrane, an effect that was abolished by tolbutamide. The effect of NE on KATP channels was investigated in parallel using outside-out single channel recording. This revealed that NE enhanced the open activities of the KATP channels ∼2-fold without changing the single channel conductance, demonstrating that NE-induced hyperpolarization was mediated by activation of the KATP channels. The NE effect was abolished in cells preincubated with pertussis toxin, indicating coupling to heterotrimeric Gi/Go proteins. To identify the G proteins involved, antisera raised against α and β subunits (anti-Gαcommon, anti-Gβ, anti-Gαi1/2/3, and anti-Gαo) were used. Anti-Gαcommon totally blocked the effects of NE on membrane potential and KATP channels. Individually, anti-Gαi1/2/3 and anti-Gαo only partially inhibited the action of NE on KATP channels. However, the combination of both completely eliminated the action. Antibodies against Gβ had no effects. To confirm these results and to further identify the G protein subunits involved, the blocking effects of peptides containing the sequence of 11 amino acids at the C termini of the α subunits were used. The data obtained were similar to those derived from the antibody work with the additional information that Gαi3 and Gαo1 were not involved. In conclusion, both Gi and Go proteins are required for the full effect of norepinephrine to activate the KATP channel.

Insulin secretion is inhibited by ␣ 2 -adrenergic receptor activation (1,2). The ␣ 2 -adrenergic receptors at the plasma membrane are linked to pertussis toxin (PTX) 2 -sensitive G i and G o proteins. Upon interacting with the ␣ 2 -adrenergic receptors, catecholamines (i.e. epinephrine and norepinephrine (NE)), two important physiological inhibitors of insulin secretion, activate G i and/or G o proteins and inhibit the exocytosis of insulin-containing granules (2). Four mechanisms have been suggested to account for their inhibitory effect: 1) activation of ATP-sensitive K ϩ channels (K ATP ) and repolarization of the ␤-cells; 2) inhibition of L-type Ca 2ϩ channels; 3) decreased activity of adenylyl cyclase; and 4) inhibition of exocytosis at a "distal" site in stimulus-secretion coupling, which is beyond the elevation of intracellular Ca 2ϩ and beyond the potentiating actions of cAMP and diacyglycerol (2). Although it is generally accepted that the "distal" inhibitory effect is the most dominant, the relative contribution of each of these four mechanisms to the overall inhibition of insulin release is not clear.
Considering the effects of catecholamines on the membrane potential in the insulin-secreting cells, it is clear that catecholamines can repolarize the cell membrane in a PTX-sensitive manner. However, the explanations for this phenomenon are mixed. Abel et al. (3) reported that adrenaline-induced hyperpolarization was due to activation of K ATP channels but that the effect was insufficient to cause the sustained inhibition of insulin secretion in INS-1 cells associated with it (i.e. the distal effect is dominant). Schermerhorn and Sharp (4) also found that NE acted on the K ATP channel and produced different effects on [Ca 2ϩ ] i in oscillating and nonoscillating HIT-T15 cells. However, Rorsman et al. (5) proposed that epinephrine suppressed mouse ␤-cell electrical activity by a G protein-dependent mechanism, which culminated in the activation of a sulfonylurea-insensitive low conductance K ϩ channel distinct from the K ATP channel. Further evidence for the existence of this G protein-regulated low conductance K ϩ channel comes from studies with SUR1 knock-out mice that do not have functional K ATP channels. Islet cells from these mice are hyperpolarized by epinephrine in a PTX-sensitive manner, and the hyperpolarization is due to activation of a low conductance K ϩ channel (6). The expression and catecholamine control of the K ATP channel and the low conductance channel in ␤-cells of different origins and conditions (e.g. knockouts) remain to be defined. However, although it is established that the action of catecholamines is coupled to the activation of G i and G o proteins, information about the exact identities of these G proteins and the corresponding subunits is not available. Therefore, in this study, we investigated the effects of NE on the membrane potential and on the K ATP channels in INS 832/13 cells, and we identified the G proteins involved. For this, we used a combination of whole cell configuration and outside-out single chan-nel recording to monitor the membrane potential and single K ATP currents in parallel. In order to clarify which G proteins and subunits are responsible for the actions of NE, we used antisera raised against various G protein ␣ and ␤ subunits, anti-G␣ common , anti-G␤, anti-G␣ i1/2/3 , and anti-G␣ o . We also used small peptides specific for the C termini of the ␣ subunits that specifically block G protein interactions with receptors. The antisera or peptides were diffused into the cells separately and in combination, and their effects on the INS 832/13 cells were determined. NE hyperpolarized the cell membrane via enhancing the open activities of the K ATP channels in an ATP-dependent manner. Although the action of NE was mediated by G␣ i and G␣ o subunits, the effects of the two types of subunits were not redundant. Applying one of their corresponding antibodies or peptides only partially inhibited the action of NE. The effect of NE was fully blocked only by the combination of antibodies against G␣ i1/2/3 and G␣ o and by the blocking peptides for G␣ i1/2 and G␣ o2 but not the peptides for G i3 and G␣ o1. The data show that activation of K ATP channels by NE contributes to the NEinduced inhibition of insulin release in the INS 832/13 cells and that the action requires both G␣ i1/2 and G␣ o2 proteins.

EXPERIMENTAL PROCEDURES
Cell Culture-INS 832/13 cells (a kind gift by Dr. C. B. Newgard) were cultured in complete RPMI 1640 medium supplemented with 10% fetal bovine serum, 100 g/ml streptomycin, and 100 units/ml penicillin at 37°C in a 95% air, 5% CO 2 atmosphere. The cells (passage numbers 60 -66) were divided once a week by treatment with trypsin, and the medium was changed twice between divisions. The measurements were performed 1-2 days after cell division.
Electrophysiology and Solutions-Both whole cell current clamp mode and outside-out single channel recording were applied. Whole cell recordings were performed using fire-polished electrodes, pulled from borosilicate glass and showing an open resistance of 2-5 megaohms. Signals were amplified and acquired by a PULSE 8.75-controlled EPC-10 amplifier (HEKA Electronics). Data were filtered through a four-pole low pass Bessel filter at 2 kHz, and the sampling rate was 10 kHz. For single channel recording, data were further digitally filtered at 500 Hz. Capacitative transients and series resistance were compensated, using the circuitry incorporated in the amplifier. The cells were continuously superfused with standard extracellular solution (solution A) containing 140 mM NaCl, 5.6 mM KCl, 1.2 mM MgCl 2 , 2.6 mM CaCl 2 , and 10 mM HEPES-NaOH (pH 7.4) and glucose with adjusted concentrations. For the perforated whole cell configuration, the pipette solution (solution B) was composed of 10 mM KCl, 76 mM K 2 SO 4 , 10 mM NaCl, 1 mM MgCl 2 , 10 mM HEPES-KOH (pH 7.3) and amphotericin B (200 g/ml). For the standard whole cell configuration, the patch pipette was filled with solution C containing 145 mM potassium glutamate, 8 mM NaCl, 1.0 mM MgCl 2 , 2 mM ATP-Mg, 0.5 mM GTP, 0.3 mM cAMP, and 10 mM HEPES-KOH (pH 7.3). The membrane potential was measured under perforated or standard whole cell configurations, and during the recording periods the cells were bathed in solution A supplemented with 16.7 mM glucose. The outside-out patch was established after the formation of standard whole cell configuration and was continuously bathed in solution A supplemented with 5 mM tetraethylammonium and 2.8 mM glucose. The pipette solution specific for the single channel recording (solution D) was composed of 140 mM KCl, 8 mM NaCl, 1.0 mM MgCl 2 , 0.1 mM ATP-Mg, 0.5 mM GTP, 0.3 mM cAMP, 5 mM EGTA, and 10 mM HEPES-KOH (pH 7.3). NE and tolbutamide were freshly prepared in stock solutions before experiments and were added into solution A at final concentrations of 5 and 300 M, respectively. All experiments were performed at room temperature (20 -23°C). The antibodies used in this study were all purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Antibodies were added to the pipette solution with adjusted concentrations and were applied intracellularly by dialysis. In order to reach equilibrium of the antibodies between the pipette tip and the cytosol, monitoring the membrane potential as well as establishing the outside-out membrane patch were performed 2-5 min after the cell membrane was ruptured. Experiments using the peptides analogous to the C termini of G␣ i1/2 , G␣ i3 , G␣ o1 , G␣ o2 , and a randomized peptide were performed exactly as those using the antibodies.
The analysis program of single channel recording was written in our laboratory based on the software of Igor Pro version 5.03 (WaveMetrics, Inc.). The maximum number of functional channels (N) in the patch was estimated by observing the number of peaks detected on the histogram distribution of amplitude. As an index of channel activity, NP o (the maximum number of channels in the patch times the open probability) was calculated as follows, Statistics-Data are presented as means Ϯ S.E. Significance tests were performed using Student's test.

Effects of NE on the Membrane Potential of INS 832/13 Cells-
The perforated whole cell configuration was applied to investigate the effect of NE on the membrane potential in intact INS 832/13 cells. The cells were continuously superfused with solution A containing 16.7 mM glucose, and the voltages were monitored by using pipette solution B. In Fig. 1A, the membrane potential was depolarized by high glucose to ϳϪ40 mV with continuous bursts of action potentials to ϳ20 mV, which were triggered by rapid Ca 2ϩ influx. When challenged with NE (5 M) in contact with the extracellular membrane surface, the cell was markedly hyperpolarized to ϳϪ60 mV and was redepolarized to ϳϪ40 mV upon the removal of NE, showing that the effect of NE had a rapid onset and was rapidly reversible. The apparent delay in onset of the action of NE (Fig. 1A) and tolbutamide ( Fig. 1B) was caused by the perifusion system and a lag period before NE and tolbutamide reach the cells. The sensitivity of NE-induced hyperpolarization to the classic K ATP channel inhibitor, tolbutamide, was tested. Tolbutamide (300 M) reversed the NE-induced fall in membrane potential (Fig. 1B). Similarly, when tolbutamide was added first, NE had no effect on membrane potential (not shown). NE was not capable of hyperpolarizing the cells that were precultured with PTX (150 ng/ml) for 24 h (Fig. 1C), indicating the involvement of G i and/or G o proteins in the regulation of the membrane potential by NE. Data presented in Fig. 1, A-C, respectively, are typical of three sets of experiments. As summarized in Fig. 1D, on average, 5 M NE hyperpolarized the membrane potential from Ϫ36 Ϯ 1.5 mV to Ϫ55 Ϯ 2.1 mV (n ϭ 14, p Ͻ 0.01), and tolbutamide counteracted this hyperpolarization by restoring the membrane potential from Ϫ57 Ϯ 3.2 to Ϫ31 Ϯ 1.8 mV (n ϭ 10, p Ͻ 0.01). In the cells pretreated with PTX, the membrane potentials before and after the challenge with NE were Ϫ43 Ϯ 4.0 and Ϫ45 Ϯ 3.7 mV (n ϭ 5, p Ͼ 0.05). Similarly, PTX treatment blocked the effect of NE to decrease the frequency of Ca 2ϩ -triggered action potentials. Analysis of the data in Fig. 1 showed that the frequency of the action potentials was 20.6 Ϯ 5.5/ min prior to NE treatment, 17.7 Ϯ 4.2/min during NE treatment (p Ͼ 0.05, n ϭ 5) and 16.8 Ϯ 4.3/min after the washout of NE. These data indicate that NE hyperpolarized the cell membrane via its effect on the K ATP channels. To this point, the observations were obtained in the relatively intact cell and could not elucidate whether the change of membrane potential was made by the direct interaction between the K ATP channels and the G i/o proteins or by indirect interaction with other cytosolic messengers. To address this, the same experimental procedures were repeated, except that the standard whole cell configuration was employed and the patch pipette was filled with solution C containing 2 mM ATP, 0.5 mM GTP, and 0.3 mM cAMP. As shown in Fig. 2A, the cell membrane was again reversibly hyperpolarized by 5 M NE. When ATP was omitted from the pipette solution, the cell was hyperpolarized, and, as expected, NE was not capable of inducing any further fall of membrane potential (Fig. 2B). The effects of NE on the membrane potential recorded under the standard whole cell configuration in the presence and absence of ATP i are summarized in Fig. 2C. In the presence of 2 mM ATP in the pipette solution C, the membrane potentials were Ϫ41 Ϯ 3.5 mV (control), Ϫ58 Ϯ 3.1 mV (with NE), and Ϫ42 Ϯ 4.7 mV (wash out). These values are similar to those obtained with the perforated whole cell configuration. The effect of NE on membrane potential was significantly different from both the pre-NE and post-NE values (n ϭ 5, p Ͻ 0.01). When ATP i was omitted, the membrane potentials were not significantly changed by NE (n ϭ 5, p Ͼ 0.05), remaining at ϳϪ60 mV before, during, and after NE treatment.
Effects of NE on ATP-sensitive K ϩ Channels under Single Channel Recording Conditions-The effects of NE on single K ATP channels were examined in detail. All single channel recordings presented in this paper were derived exclusively from currents passing through K ATP channels in outside-out patches and with the membrane potential clamped at 0 mV. The Ca 2ϩ -activated nonselective cation channels and the FIGURE 1. Effect of NE on the membrane potential under control and PTX-treated conditions. The representative traces shown in A-C were obtained using the perforated whole cell current clamp configuration. In A, NE (5 M) reversibly hyperpolarized the membrane. In B, the NE-induced hyperpolarization was blocked by 300 M tolbutamide, indicating that the effect of NE was on the K ATP channel. In C, NE failed to induce hyperpolarization in PTX-pretreated cells. D summarizes the effects of NE on membrane potential under the different experimental conditions imposed. The lines indicate the perfusion periods of NE and tolbutamide. **, p Ͻ 0.01; #, p Ͼ 0.05 relative to control values. Ca 2ϩ -activated K ϩ channels were eliminated by keeping the [Ca 2ϩ ] i low, Ͻ10 Ϫ9 M, and by applying 5 mM tetraethylammonium extracellularly (7). Because the number of operational K ATP channels per membrane patch is often unknown, NP o was introduced to estimate the channel activity (see "Experimental Procedures"). Fig. 3A shows the effect of NE (5 M) on single K ATP channels in an outside-out membrane patch. The pipette solution D was in direct contact with the cytosolic side of the membrane patch, and the outside of the patch was continuously superfused with solution A. The single channel currents were monitored before, during, and after NE treatment. In Fig. 3A, NE-induced enhancement of the open activity of K ATP channels was clearly visualized, and this effect was completely reversible, since the channel activity decreased to the control level upon removal of NE (Fig. 3E). The values of NP o determined under test conditions were normalized to the control value, which was displayed as 100%. As summarized in Fig. 3E, on average, 5 M NE increased the value of NP o from 100% in the control situation to 209 Ϯ 17% and returned back to 106 Ϯ 11% of the pretest value upon removal of NE (0.1 mM ATP i was present, n ϭ 12, p Ͻ 0.01). When ATP was excluded from the pipette solution, the channels exhibited robust open activity, and the channel activity could not be further increased by the application of NE (Fig. 3B), which is similar to the situation shown in Fig. 2B. The results shown in Fig. 3, A and B, are typical of two sets of experiments under different conditions of ATP i . On average, when ATP i was omitted, the mean value of NP o was 103 Ϯ 3.4% in the presence of NE (n ϭ 7, p Ͼ 0.05). When tolbutamide was present, the subsequent addition of NE had no effect on K ϩ channel activity (not shown).
These data are consistent with the observations shown in Fig. 2 that the NE-induced modification of electrical properties in INS 832/13 cells cannot be detected in the absence of ATP i . The current frequency distribution for the channels studied under control and NEtreated conditions in the presence of 0.1 mM ATP is shown in Fig. 3C and in the absence of ATP in the pipette in Fig. 3D. There is no evidence for the presence of more than one channel in the cell under either of these conditions. In order to investigate whether NE also influences the conductance of single K ATP channels, single K ATP currents were recorded in the presence and absence of 5 M NE over the range of Ϫ40 to ϩ60 mV. The mean values of single current amplitudes were plotted versus the corresponding clamped voltages (Fig. 3F). The current-voltage relationship was nearly linear at negative potentials but showed a slight inward rectification at membrane potentials higher than 10 mV, as previously described in RINm5F cells by Findlay (8). The two I-V curves obtained under control and test conditions were overlapping. The slope conductance fitted between Ϫ40 and 20 mV was ϳ20 picosiemens in control and 19.2 picosiemens in the presence of NE, indicating that application of NE did not induce any significant modification of single channel current amplitude. The single channel conductance of 20 picosiemens is similar to that of other reports for the KATP channel in clonal cells (8 -10).
Effects of Antibodies against Common G ␣ Subunits on the NE-induced Electrical Changes-To identify the G protein subunits involved in NE-mediated hyperpolarization and activation of the K ATP channels, antibodies raised against various G protein subunits were tested. Anti-G␣ common (2 g/ml) was added to the pipette solutions C and D, for the voltage measurement and outside-out single channel recordings, respectively. Once the standard whole cell configuration was established, 2-5 min was allowed for the antibodies to reach  potential. The membrane potential was Ϫ42 Ϯ 4.3 mV in control, Ϫ41 Ϯ 3.7 mV in the presence of NE, and Ϫ42 Ϯ 4.3 mV upon removal of NE (n ϭ 15, p Ͻ 0.01). In order to exclude possible nonspecific effects of anti-G␣ common , the experiments were controlled by replacing anti-G␣ common with IgG and by measuring the response of the membrane potential to NE. As shown in Fig. 4C, NE induced its anticipated effect on membrane potential, and the results were analogous to those shown in Fig. 2C. The effect of anti-G␣ common on the NE-induced activation of K ATP channels was investigated by adding anti-G␣ common (2 g/ml) to pipette solution D. 5 M NE failed to induce a visible change of channel activity (Fig. 4B). As summarized in Fig. 4D, with anti-G␣ common in the pipette solution D, the normalized NP o was almost unchanged from 100% in control to 103 Ϯ 3.3% in the presence of 5 M NE and 99 Ϯ 4.8% with the removal of NE (n ϭ 18, p Ͼ 0.05). However, the NE effects remained in a series of control experiments using IgG. NP o was increased by NE from 100% to 221 Ϯ 12% and back to the precontrol level of 98 Ϯ 2.7% when NE was washed out (n ϭ 6, p Ͻ 0.01).
Effects of Antibodies against Specific Subunits of G␣ i and G␣ o on the NE-induced Electrical Changes-From the results presented above, we predicted that at least one of these two types of subunits, G␣ i and G␣ o , should be able to mimic, to some extent, the effects of anti-G␣ common on the electrical properties in INS 832/13 cells. To test this, monoclonal antibodies raised specifically against G␣ i1/2/3 and G␣ o were applied. Adding anti-G␣ i1/2/3 or anti-G␣ o (2 g/ml for each) to the cytosolic side of the patches did not fully block NE (5 M) to increase K ATP channel activity. As can be seen in the summary in Fig. 5A, in the presence of individual anti-G␣ i1/2/3 and anti-G␣ o , NE still increased the normalized NP o values from 100% to 156 Ϯ 7.8% and to 148 Ϯ 5.1%, respectively for anti-G␣ i1/2/3 (n ϭ 17, p Ͻ 0.05) and anti-G␣ o (n ϭ 19, p Ͻ 0.01), respectively. After removal of NE, both NP o values determined in the presence of G␣ i1/2/3 and anti-G␣ o individually returned to the precontrol levels.
Because anti-G␣ i1/2/3 and anti-G␣ o only partially attenuated the effects of NE (by ϳ44 and ϳ52%, respectively), the effect of these antisera in combination was examined. In the patches treated with G␣ i1/2/3 and anti-G␣ o simultaneously, the normalized NP o , in the presence of NE, was 101 Ϯ 6.1% (n ϭ 15, p Ͻ 0.01). This represents a complete inhibition of the effects of NE on K ATP channel activity by anti-G␣ i1/2/3 and anti-G␣ o when applied together. In considering why individual anti-G␣ 1/2/3 and anti-G␣ o only partially block the action of NE, several possibilities exist. First, G␣ i1/2/3 and G␣ o are redundant, and both have the same targeting sites on the K ATP channel. In this case, the partial inhibition observed could be due to an insufficient amount of antibodies applied. Second, the effects of G␣ i1/2/3 and G␣ o on the K ATP channels might not be redundant, so that both independently impair the closure of K ATP channels by targeting different sites. In this  FEBRUARY 29, 2008 • VOLUME 283 • NUMBER 9 JOURNAL OF BIOLOGICAL CHEMISTRY 5311 case, each alone would not fully block the effect of NE. Third, the inhibitory effect observed from one of the two types of antibodies, either anti-G␣ i1/2/3 or anti-G␣ o , could be artifactual, due to cross-reaction between the two antibodies (unlikely, since both antibodies were affinity-purified). To answer the questions of redundancy and cross-reactivity, the experiments were repeated, except that the concentrations of the individual anti-G␣ i1/2/3 and anti-G␣ o were modified either by increasing 2-fold (4 g/ml) or by lowering 10-fold (0.2 g/ml). Comparing the summary data in Fig. 5B with that present in Fig. 5A, increasing the amount of antibody did not change the outcome from that seen with the initial concentration of 2 g/ml. At higher concentrations of anti-G␣ i1/2/3 or anti-G␣ o (4 g/ml for each) the NP o values in the presence of NE were 151 Ϯ 4% (n ϭ 13, p Ͻ 0.01) and 152 Ϯ 6% (n ϭ 14, p Ͻ 0.01), respectively (similar to the results in Fig. 5A). Full inhibition was only achieved by combining both antibodies (NP o ϭ 108 Ϯ 4%, n ϭ 15, p Ͻ 0.01). Thus, the effects of individual G␣ i1/2/3 and G␣ o are not redundant. The possibility of cross-reaction was minimized by decreasing the concentrations of anti-G␣ i1/2/3 and anti-G␣ o to 0.2 g/ml for each (one-tenth of the original amount of antibody). Under these conditions, when any possible cross-reaction should be much reduced or eliminated, the effects of anti-G␣ i1/2/3 and anti-G␣ o were still clearly observed (see Fig.  5C). Anti-G␣ i1/2/3 and anti-G␣ o attenuated the effects of NE on NP o by ϳ34% (NP o ϭ 164 Ϯ 4%, n ϭ 11, p Ͻ 0.01) and ϳ28% (NP o ϭ 172 Ϯ 7%, n ϭ 9, p Ͻ 0.01), respectively. When applied together, the effect of the two antibodies was greater than either one alone and resulted in 51% inhibition (NP o ϭ 149 Ϯ 5%, n ϭ 11, p Ͻ 0.01). Next, both anti-G␣ i1/2/3 (2 g/ml) and anti-G␣ o (2 g/ml) were applied simultaneously to counteract NE-induced hyperpolarization under the standard whole cell configuration. As summarized in Fig. 5D, the mean values of the membrane potentials were Ϫ43 Ϯ 3.7 mV, Ϫ45 Ϯ 4.2 mV, and Ϫ43 Ϯ 5.1 mV, before, during, and after NE treatment, respectively (n ϭ 15, p Ͼ 0.05), data that are consistent with the observations from the single channel experiments presented earlier. Control experiments (shown in Fig. 5, A-D) were performed with mouse and goat IgG, separately or combined as appropriate.

Norepinephrine Effect Requires both G i and G o Proteins
Effects of G␤ on NE-induced Electrical Activities-To evaluate the contribution of G␤ subunits to the effects of NE, polyclonal antibodies raised against G␤ subunits (anti-G␤) were added to the pipette solution C or D, with respect to voltage measurement and outside-out single channel recording. In the presence of anti-G␤, the example traces corresponding to the effects of NE on the membrane potentials and on the K ATP channels are presented in Fig. 6, A and B, respectively. In contrast to the effects of anti-G␣ common , anti-G␣ i1/2/3 , and anti-G␣ o , NE-mediated regulation of the mem- brane potential and the K ATP channels was not changed by anti-G ␤ . As summarized, in the presence of anti-G ␤, NE reduced membrane potential from Ϫ45 Ϯ 3.2 to Ϫ58 Ϯ 4.7 mV (n ϭ 9, p Ͻ 0.01) (Fig. 6C) and increased NP o from 100% to 212 Ϯ 12% (n ϭ 14, p Ͻ 0.01) (Fig. 6D), similar to the control experiments. In addition to concluding that the ␤␥ subunit has no role in the activation of the K ATP channels, these results also exclude an effect of NE on K ϩ channels of the Kir3 family in these cells, which are activated by ␤␥ (11)(12)(13)(14).
In order to check the results obtained with antibodies against G␣ i and G␣ o and in order to obtain further information on the specificity of the G i and G o proteins involved, we used a second independent approach to block the effect of the G i and G o proteins. This was to prevent the activation of the G proteins at the ␣ 2 -adrenergic receptor with peptides containing the same sequences as the last 11 amino acids of the C termini of G␣ i1/2 , G␣ i3 , G␣ o1 , and G␣ o2 (15,16), using a randomly synthesized peptide (G i R) as control. These specific peptides have been shown to effectively block G protein interaction with the receptors for thrombin and adenosine (15,16) and to block the action of adenosine on a potassium channel using a whole cell tech-nique similar to that used here (16). The results of these experiments are shown in Fig. 7. NE increased NP o from 100% to 226 Ϯ 6% (n ϭ 6, p Ͻ 0.01). Using concentrations of 20 M, peptides for G␣ i1/2 reduced the effect of NE to 144 Ϯ 3% (n ϭ 6, p Ͻ 0.01). Peptides for G␣ o2 reduced the NE effect to 143 Ϯ 6% (n ϭ 4, p Ͻ 0.05), essentially to the same extent as those for G␣ i1/2 . Importantly, peptides for G␣ i3 and G␣ o1 had no significant effects. Using concentrations of 60 M, the effects of the peptides were almost identical to those of peptides used at 20 M. Thus, a peptide concentration as low as 20 M is sufficient to produce maximal effects. Finally, different combinations of peptides were used. The combination of peptides for G␣ i3 and G␣ o1 had no significant effect. The combination of peptides for G␣ i3 and G␣ o2 reduced the effect of NE to 142 Ϯ 9% (n ϭ 8, p Ͻ 0.01), similar to the effect of G␣ o2 alone. The combination of peptides for G␣ i1/2 plus G␣ o1 reduced the NE effect to 144 Ϯ 9% (n ϭ 6, p Ͻ 0.05), similar to those for G␣ i1/2 alone. The last combination to be tested, that for the G proteins that appear to be the mediators for NE (i.e. G␣ i1/2 plus G␣ o2 ), reduced the effect of NE to 105 Ϯ 1% (n ϭ 7, p Ͻ 0.01), effectively eliminating the response to NE. These results confirm those obtained with antibodies against the ␣ subunits and the conclusion that both G i and G o proteins are required to fully activate the channel. Furthermore, they demonstrate the specificity of the G protein involvement in that G␣ i3 and G␣ o1 are without effect.
Given these data and the conclusion that both G i and G o proteins appear to be required for full activation of the K ATP channel, an alternative explanation could be that two distinct channels are responsible for the effects of NE and that G␣ i2 works on one and G␣ o2 works on the other. Reasons for thinking this are the reports that islets from SUR1 Ϫ/Ϫ mice are still responsive to epinephrine-induced hyperpolarization. This would argue that the K ATP channel is not the inward rectifier responsible for the hyperpolarization in these mouse ␤-cells  The individual peptides corresponding to the C termini of G␣ i1/2 or G␣ o2 subunits only partially inhibit the effects of NE on K ATP channels, whereas the combination of the two inhibit completely. Peptides corresponding to G␣ i3 or G␣ o1 , a combination of the two, and a randomized control peptide had no effect. Combinations of peptides for active (G␣ i1/2 or G␣ o2 ) and inactive (G␣ i3 or G␣ o1 ) ␣ subunits inhibited similarly to the effect of the active peptides alone. (17). Also, Kir3.1 is present in at least some ␤-cells; its current is pertussis toxin-and therefore G i /G o protein-sensitive and causes hyperpolarization in response to norepinephrine (18). Rorsmann (5) also proposed that epinephrine causes hyperpolarization via a G protein-sensitive, sulfonylurea-insensitive low conductance potassium channel distinct from K ATP . An obvious candidate for an additional channel would be Kir3.1, which is G protein-sensitive and has been reported present in mouse ␤-cells and clonal cell line(s) derived from mice (18). Finally, somatostatin activates two inwardly rectifying K ϩ channels in MIN-6 cells (also mouse-derived), the K ATP and a G protein-gated inwardly-rectifying K ϩ channel (19). Despite these reports, the possibility of two distinct channels in our study using a rat-derived clonal cell line seems unlikely for the following reasons. 1) Tolbutamide gives complete blockade of K ϩ channel activity. 2) The I/V relationship in Fig. 3F is typical of the K ATP channel and is not distorted as would be the case if two channels were operating simultaneously. 3) There is no evidence in the I/V relationship in Fig. 3D for a "classical" steep inwardly rectifying channel, such as Kir3.1, even when studied down to Ϫ100 mV (data not shown). 4) The frequency distribution histograms in Fig. 3, C and D, provide evidence for only one channel. 5) The G protein-gated inwardly-rectifying K ϩ channels are activated by ␤␥, and we would have detected any effect of NE on G protein-gated inwardly-rectifying K ϩ channel activity during our experiments with the anti-␤ antibody. However, in a final test for the presence of Kir3.1, an inhibitor of the channel SCH23390 was used (20). These experiments were carried out in the same series as the peptide inhibitor studies, so that NE increased NP o to 226 Ϯ 6%. In the presence of SCH23390 at the high concentration of 150 M, NE increased NP o to 204 Ϯ 7%, not significantly different from the control value (n ϭ 8, p Ͼ 0.05) and nowhere near the significant reductions to 144 and 143% seen with the individual peptides for G␣ i2 and G␣ o2 , respectively.

DISCUSSION
The effects of NE on the electrical activity of the INS-1-derived clone 832/13 have been investigated. The INS 832/13 cells exhibit markedly enhanced and stable responses to glucose and several of its known potentiators (21) and have been used extensively as a model system for mammalian ␤-cells (22)(23)(24)(25)(26). We first studied the effects of NE on the membrane potential. The standard whole cell configuration has the disadvantage of disturbing the metabolic equilibrium of the cell (27) and may attenuate or prevent glucose-evoked effects. For this reason, we started our experiments with the perforated whole cell configuration, which keeps the cells relatively intact and efficiently prevents the loss of macromolecules from the cytosol. NE-induced hyperpolarization of the cells was completely reversible, inhibited by tolbutamide, and abolished by pretreatment with PTX, indicating the involvement of G i /G o proteins. NE hyperpolarized the cells via a G i /G o protein coupling of the ␣ 2 -adrenergic receptors, as described previously in HIT-T15 cells (5). Sieg et al. (6) and Rorsman et al. (5) also reported that catecholamines hyperpolarized mouse pancreatic ␤-cells. However, they attributed the changes of membrane voltage to the activation of ATP-insensitive K ϩ channels. These results are not necessarily controversial, because the expression of K ϩ channels in clonal cells derived from the rat and hamster may well be different. In primary cells also there is evidence for differences between mouse and rat ␤-cells, and human ␤-cells may be different again. Therefore, it is important to point out that the KATP channel-dependent mechanism by which norepinephrine hyperpolarizes the INS 832/13 cell may be different from that of other clonal and primary cells, which may also differ among themselves. Nevertheless, the use of a model system, such as the 832/13 cell, to study the G protein interactions with an individual ion channel is valid.
It is generally accepted that glucose-stimulated biphasic insulin secretion involves at least two signaling pathways (26), the K ATP channel-dependent and K ATP channel-independent pathways, respectively. In the former, enhanced glucose metabolism increases the cellular ATP/ADP ratio, closes K ATP channels, and depolarizes the cell. According to our results, we assume that the K ATP channels in the study are directly or indirectly regulated by activated G i /G o proteins upon the binding of NE to the ␣ 2 -adrenergic receptors. In an indirect way, G i /G o could relieve the inhibition of K ATP channels from ATP by triggering a series of downstream reactions that either impair glucose metabolism and thus reduce the production of ATP or result in a change of cytosolic second messengers (e.g. decreased activity of adenylyl cyclase). Alternatively, it is also possible that the closure of the K ATP channels is impaired due to the direct interaction between G i /G o proteins and the K ATP channels. Because the perforated whole cell configuration applied does not provide further information about the mechanisms underlying the action of NE, the same experiments were repeated, except that the standard whole cell configuration was applied. It is assumed that intracellular glucose metabolism is impaired after the connection between cytosol and pipette solution is established by mechanically breaking the cell membrane. However, in the presence of 2 mM ATP in the pipette solution, the cell membrane was still depolarized under control conditions and hyperpolarized upon challenge with NE, indicating that the effect of NE on the membrane potential is independent of glucose metabolism. As anticipated, when ATP was omitted from the pipette solution, the membrane voltage remained at ϳϪ60 mV in the absence and presence of NE. Next, we examined the effects of NE on single K ATP channels by using the outside-out membrane patch. Single channel analysis revealed that, in the presence of ATP i , NE enhanced the open activity of K ATP channels by more than 2-fold without changing the single channel conductance. There are many examples of ion channels that are regulated by G protein coupled receptors. The signaling pathways used include second messengers, phosphorylation, and direct regulation by G proteins (28 -30). In our studies, the regulation of K ATP channels by NE is clearly a membrane-delimited phenomenon, since we show that NE can modulate the channel activity when applied to excised membrane patches in a cell-free environment. The membrane-delimited effect of NE on the K ATP channels is also consistent with the observation that NE can induce membrane hyperpolarization even under the conditions of the standard whole cell configuration. In this regard, the action of NE resembles the neurohormonal modulation of other K ϩ channels that can also proceed without the participation of cytosolic second messengers (31). Considering the rapid onset and recovery of K ATP channels from NE-induced activation, two possibilities exist. Either the G proteins act directly on the subunits of the K ATP channels, or the functional G proteins trigger some unknown factors, which are located in the cell membrane and can directly activate the K ATP channels. Closely comparing the current traces in Fig. 3, A and B, we found that NE, under the experimental conditions employed, antagonized the ATP i -mediated inhibition of K ATP channels. However, the action of NE was not sufficient to restore channel activity to the spontaneous level seen in the absence of ATP i . This could be due to an insufficient amount of NE applied, to insufficient G proteins to activate all of the channels, or to the simple inability of norepinephrine to activate the channels as completely as occurs in the absence of ATP. The first possibility is unlikely, because we used supermaximal concentrations of NE (even 10 M did not increase its effect on NP o . Specific antisera raised against various ␣ subunits of G proteins (G␣ common ) and against G␤ have been used to examine which subunits of the pertussis toxin-sensitive G i /G o proteins are involved in NE-induced electrical modulation. Anti-G␣ common was able to completely inhibit the effects of NE on both the K ATP channels and the membrane potential (Fig. 4); however, neither effect was influenced by anti-G␤ under the same experimental conditions (Fig. 6). Previous studies have shown that K ATP channels in insulin-secreting cell lines are activated by physiological inhibitors other than norepinephrine. Dunne et al. (32) found that galanin enhanced K ATP channel activity in the insulinoma cell line, RINm5F, via pertussis toxin-sensitive G proteins. de Weille et al. (33) reported that somatostatin regulated K ATP channels in RINm5F cells, and the presence of intracellular GTP was required. Although these observations clearly indicated that G proteins played a role in channel modulation, they did not provide evidence as to which G proteins and corresponding subunits were involved. This problem was approached by Ito et al. (34), who reported that G␣ activated K ATP channels and G␤ activated K Ach channels in cardiac cell membranes, and by Ribalet and Eddlestone (35), who found that ␣ subunits of G i or G o proteins stimulated the K ATP channels in HIT-T15 and RINm5F cell lines. Our findings agree with these observations but go further. In this study, the activation of K ATP channels is mediated by G␣ subunits, not by G␤. Furthermore, only when anti-G␣ i1/2 and anti-G␣ o were applied together did the antibodies achieve complete blockade of the effects of NE on K ATP channels as well as on membrane potential. Additional compelling evidence that both G i and G o are required for full activation of the channel was gained from the C-terminal peptide studies. Blocking activation of the G proteins by the norepinephrine-occupied ␣ 2 -adrenergic receptor with peptides specific for individual G protein C termini provided data nearly identical to data obtained by the antibody studies together with the additional information that G␣ i3 and G␣ o1 are not involved in activating the channel. An analogous observation that two G i proteins (G i2 and G i3 ) were required for the full inhibition of adenylyl cyclase has been made in RINm5F cells (36). In that study, it was concluded that individual G i2 and G i3 proteins only partially mediated the inhibition of adenylyl cyclase by galanin, whereas the combination of the two G pro-teins functioned additively. However, since there are at least two isoforms of adenylyl cyclase present in ␤-cells (37), an alternative explanation may be that the two enzymes are affected separately by G i2 and G i3 .
Although the current study strongly suggests that different targeting sites of G i or G o are involved in the NE-mediated activation of the K ATP channels, the localization of these targeting sites and the underlying molecular mechanisms of the actions of G i /G o proteins on these sites are not known. Most likely, the sites are located on the K ATP channels. However, we cannot exclude the possibility that one or both of the G i /G o proteins regulate the K ATP channels by affecting other unknown mediators, which are tightly embedded in the cell membrane and modify the channel activity upon the interaction of G i /G o proteins.
K ATP channels are formed from an inward rectifier potassium channel (Kir6.x) and a sulfonylurea receptor (SUR) subunit, arranged with 4:4 stoichiometry (38,39). The different isoforms of Kir6.x (Kir6.1 or Kir6.2) and SUR (SUR1, SUR2A, or SUR2B) show distinct expression patterns. For insulin-secreting cells, the K ATP channels are assembled from Kir6.2 and SUR1. Kir6.2 forms the channel pore and provides the ATPbinding site, whereas the SUR1 acts as a regulatory subunit, endowing the channel with sensitivity to high affinity sulfonylurea inhibition and activation by Mg-nucleotides and channel openers (40 -45). If we hypothesize that G i /G o proteins directly interact with K ATP channels, both subunits of Kir6.2 and SUR1 can be the candidates of the targeting sites. Cloning of the K ATP channel subunits has already shown that a mutant Kir6.2 lacking the 26 C-terminal residues forms functional channels in the absence of SUR, thus indicating that the ATP-inhibitory site may be located on Kir6.2 (46). Subsequently, by mutagenesis studies, several regions in the cytoplasmic N and C termini of Kir6.2 that profoundly affect the ATP inhibition of the K ATP channels have been located (47)(48)(49)(50). Based on these studies, our electrophysiological observations can be accounted for in at least three ways: G i /G o proteins may 1) directly interact with the ATP-binding site and decrease the channel sensitivity to ATP; 2) affect certain regions on Kir6.2 and thus impair channel closure; 3) interfere with the transduction mechanism via which ATP-binding initiates pore closure. The second transmembrane domain of Kir6.2 (TM2) has been proposed to line the intracellular mouth of the pore, since mutations within this region affect the single channel conductance, ion selectivity, and sensitivity to the blocking effects of intracellular cations, such as Mg 2ϩ . The fact that the single channel conductance of the K ATP channels was not changed by NE may hint that the targeting sites of G i /G o proteins are not located on TM2. However, this conclusion is not supported by recent mutagenesis studies showing that many of the mutations relevant to the ATP-inhibitory effect are located in the cytoplasmic end of TM2 (43,44). Additional experiments are needed to resolve this. The alternative candidate for the targeting site of G i /G o proteins is the SUR1 subunit. The sulfonylurea receptor modulates the activity of Kir6.2, and binding of sulfonylureas or channel openers to SUR results in a decrease or increase of the channel activity, respectively. It is possible that G i /G o proteins directly affect SUR1 and consequently increase the channel activity. Since the effects of NE on the K ATP channels were fully blocked by tolbutamide, the potency of the influences of G i /G o proteins on SUR1 itself or on the transduction from SUR1 to Kir6.2 must be much weaker, compared with that of tolbutamide, or the effect of the latter prevents NE from acting.
In summary, we have shown that NE stimulates the K ATP channel activity in INS 832/13 cells and hyperpolarizes the membrane potential. This modulation is PTX-sensitive and mediated by a combination of the G␣ i and G␣ o subunits, both of which are required for full activation of the channel.