Modulation of ATP-sensitive potassium channels by cGMP-dependent protein kinase in rabbit ventricular myocytes.

This investigation used a patch clamp technique to test the hypothesis that protein kinase G (PKG) contributes to the phosphorylation and activation of ATP-sensitive K(+) (K(ATP)) channels in rabbit ventricular myocytes. Nitric oxide donors and PKG activators facilitated pinacidil-induced K(ATP) channel activities in a concentration-dependent manner, and a selective PKG inhibitor abrogated these effects. In contrast, neither a selective protein kinase A (PKA) activator nor inhibitor had any effect on K(ATP) channels at concentrations up to 100 and 10 microm, respectively. Exogenous PKG, in the presence of both cGMP and ATP, increased channel activity, while the catalytic subunit of PKA had no effect. PKG activity was prevented by heat inactivation, replacing ATP with adenosine 5'-O-(thiotriphosphate) (a nonhydrolyzable analog of ATP), removing Mg(2+) from the internal solution, applying a PKG inhibitor, or by adding exogenous protein phosphatase 2A. The effects of cGMP analogs and PKG were observed under conditions in which PKA was repressed by a selective PKA inhibitor. The results suggest that K(ATP) channels are regulated by a PKG-signaling pathway that acts via PKG-dependent phosphorylation. This mechanism may, at least in part, contribute to a signaling pathway that induces ischemic preconditioning in rabbit ventricular myocytes.

The majority of the studies on the contribution of phosphorylation to K ATP channel activity to the cardioprotective effects of ischemic preconditioning have centered on the role of protein kinase C. Protein kinase C may act as a link between one or more receptor-mediated pathways and increased K ATP channel activity and thus lead to ischemic preconditioning (9).
The release of endogenous substances, such as adenosine, bradykinin, nitric oxide (NO), and prostacyclin (10 -12), has been proposed as a potential mechanism of ischemic preconditioning. These substances increase cGMP via direct stimulation of myocardial cells or via the endothelium. It was reported that the cGMP levels in preconditioned hearts are higher than in nonpreconditioned hearts (13)(14)(15). Since cGMP can induce protein phosphorylation via protein kinase G (PKG) activation, the involvement of PKG-dependent phosphorylation in ischemic preconditioning is expected. To date, however, the role of PKG in ischemic preconditioning and the possible PKG target proteins in this process are not well understood.
There is evidence that K ATP channels have potential phosphorylation sites, including serine/threonine residues (16 -19), and that the channels are activated by phosphorylation of these residues (20). Since PKG is a serine/threonine protein kinase, it is likely that PKG leads to phosphorylation of K ATP channels and ischemic preconditioning. In fact, the regulation of K ATP channels by cGMP was recently shown in follicle-enclosed oocytes (21,22), pancreatic ␤-cells (23,24), vascular smooth muscle cells (25,26), and cardiac myocytes (24,27), and it was suggested that this effect was mediated by PKG activation. However, there is no direct evidence that PKG activates K ATP channels.
In this study, we investigated the role of PKG-dependent phosphorylation in modulating K ATP channels using isolated rabbit ventricular myocytes. The experiments examined the signal transduction pathways involved in PKG-dependent phosphorylation. Our findings demonstrate that the NO/cGMP/ PKG-signaling pathway facilitates the activity of K ATP channels via phosphorylation. Our findings may be important for understanding the mechanism by which PKG acts as a link in one or more known receptor-mediated pathways to increase K ATP channel activity during ischemic preconditioning.

EXPERIMENTAL PROCEDURES
Materials-ATP and glibenclamide were added to either the extracellular or intracellular solution, following the experimental protocols described below. Pinacidil (RBI, Natick, MA) was freshly prepared before the experiments and diluted in the test solution to obtain the indicated final concentrations. S-Nitroso-N-acetylpenicillamine (SNAP) was purchased from Calbiochem. Okadaic acid (OA) was purchased from RBI, stored at a stock concentration of 100 M in ethanol at 4°C, and used at a final concentration of 5 nM. PP2A was purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). PKG was obtained from Promega (Madison, WI); protein kinase A (PKA) from BIOMOL (Plymouth Meeting, PA); and (S p )-8-Br-PET-cGMPS, (R p )-8-Br-PET-cGMPS, 8-pCPT-cGMP, (S p )-5,6-DCl-cBIMPS, and (R p )-8-pCPT-cAMPS from Biolog Life Science Institute (Bremen, Germany). All other chemicals used in this study, unless specified otherwise, were from Sigma. After the addition of drugs to the test solution, the pH was readjusted to 7.4 with KOH.
Cell Isolation-Single ventricular myocytes were isolated from rabbit hearts by an enzymatic dissociation procedure, as discussed previously (28). Briefly, in accordance with national animal care guidelines, rabbits weighing 150 -280 g were anesthetized with sodium pentobarbital (50 mg/ml, 1 ml/kg of body weight) and heparin (300 IU/ml). After adequate anesthesia was achieved, sternostomy was performed, and the heart was exposed. Artificial perfusion of the heart was established by cannulation of the aorta. The heart was then removed and placed in a Langendorff perfusion apparatus, and an enzymatic method was used to isolate single ventricular cells for electrophysiological experiments.
Electrophysiological Recording and Data Analysis-Single channel currents were measured in the cell-attached and inside-out patch configurations of the patch clamp technique (29). Channel activity was measured using a patch clamp amplifier (EPC-7, LIST, Darmstadt, Germany; Axopatch-1D, Axon Instruments, Foster City, CA). The DAD-12 superfusion system (Adams & List Associates, New York) was used to rapidly exchange (within 100 ms) the bath solution and drugs in most experiments. Experiments were done at a room temperature of 25 Ϯ 2°C. Membrane currents were filtered at 0.1-10 kHz and digitized at 0.4 -20 kHz. All current recordings were stored in digitized format on digital audiotapes using a DTR-1200 recorder (Biologic, Grenoble, France). To analyze single channel activity, the data were transferred to a computer (IBM-PC, Pentium-III 450, Pusan, Korea) with pCLAMP software (version 6.3; Axon Instruments Inc., Burlingame, CA) through an analog-to-digital converter interface (Digidata-1200, Axon Instruments Inc., Burlingame, CA). The threshold for judging the open state was set at half of the single channel amplitude (30). The open time histogram was formed from continuous recordings of more than 60 s. The open probability (P o ) was calculated using the following formula in Equation 1, where t j is the time spent at current levels corresponding to j ϭ 0, 1, 2, ...N channels in the open state; T d is the duration of the recording; and N is the number of channels active in the patch. The number of channels in a patch was estimated by dividing the maximum current observed, during an extended period at zero ATP, by the mean unitary current amplitude. In order to exclude cross-activation between PKA and PKG during measurements of the effects of PKG activation on K ATP channel activity, we performed the experiments under conditions in which PKA was inhibited by the potent and selective PKA inhibitor (R p )-8-pCPT-cAMPS, unless otherwise stated. (R p )-8-pCPT-cAMPS (10 M) has no effect on K ATP channels.
Statistical Analysis-The data were statistically analyzed using either Student's unpaired t test when two treatment groups were compared or one-way analysis of variance followed by a post hoc Student-Newman-Keuls test when all pairwise comparisons among the different treatment groups were made. Tests were considered significant when p Ͻ 0.05. All data are presented as means Ϯ S.E.

Effect of NO Donors on the K ATP Channel Activity of Rabbit
Ventricular Myocytes-K ATP channels were characterized by their conductance over a voltage range of Ϫ80 to Ϫ20 mV (77.8 Ϯ 3.5 pS), over a unitary current range (2.8 Ϯ 0.2 pA; Ϫ40 mV), and through their responses to potassium channel openers, ATP, and glibenclamide.
In order to test the role of the PKG signaling pathway in the regulation of K ATP channels in ventricular myocytes, we first investigated the effects of NO on K ATP channels in cell-attached patches. NO stimulates soluble guanylate cyclase and produces its effects by increasing intracellular cGMP concentration, possibly leading to the activation of PKG (31). We used SNP and SNAP, potent stimulators of cGMP formation, which are known NO donors (32,33). After gigaseal formation, the pipette potential was set to Ϫ40 mV, and the bath solution was switched from normal Tyrode solution to high K ϩ solution. Under these conditions, little single channel activity was recorded. Subsequent application to the bath of pinacidil (50 M) opened K ATP channels with a single channel current amplitude of 2.9 Ϯ 0.4 pA (n ϭ 7 patches). As the concentration of SNAP wasincreased,K ATP channelactivityincreasedinaconcentrationdependent fashion (Fig. 1A). The addition of glibenclamide (30 M) immediately suppressed this channel activity, confirming that the observed openings were due to K ATP channels. In is the concentration of the NO donor, K d is the concentration of the NO donor at the half-maximal activation of the channel, and n is the Hill coefficient. NO donors were found to activate K ATP channels with the following K d values: 7.9 Ϯ 1.2 M for SNAP (n ϭ 8 patches) and 66.9 Ϯ 1.3 M for SNP (n ϭ 11 patches). The preceding equations were used throughout the remainder of the experiments to determine for each compound used the concentration-response relationship for channel activation.
The effects of NO donors on K ATP channel activity were examined using inside-out and outside-out patch configurations to exclude the possibility that NO donors act on K ATP channels directly. Fig. 2A shows a representative result obtained for an inside-out patch. The application of NO donors to the bath failed to enhance the channel activity at Ϫ40 mV. The average P o values before and during the addition of NO donors were as follows: 0.16 Ϯ 0.07 and 0.17 Ϯ 0.09 for 1 mM SNP (p Ͼ 0.05, n ϭ 5 patches) and 0.18 Ϯ 0.06 for 300 M SNAP (p Ͼ 0.05, n ϭ 5 patches). The application of NO donors to the extracellular surface of the outside-out patch also failed to enhance the channel activity (Fig. 2B). The average P o values before and during the addition of NO donors were as follows: 0.11 Ϯ 0.07 and 0.13 Ϯ 0.06 (p Ͼ 0.05, n ϭ 3 patches) for 1 mM SNP (p Ͼ 0.05, n ϭ 3 patches) and 0.12 Ϯ 0.05 for 300 M SNAP (p Ͼ 0.05, n ϭ 3 patches).
In the experiments described in Fig. 3, increasing the concentration of (R p )-8-Br-PET-cGMPS, the most potent known inhibitor of PKG (34), inhibited the SNAP-induced K ATP channel activity (Fig. 3, A and C). In Fig. 3D, the channel activity for the (R p )-8-Br-PET-cGMPS concentration used was normalized using the equation: is the relative open probability (P o ), P o,max is the P o without (R p )-8-Br-PET-cGMPS, and P o,min is the P o at a given concentration of 10 M (R p )-8-Br-PET-cGMPS. The continuous line in the graph is the curve fitted to the Hill equation using the least-square method: is each concentration of (R p )-8-Br-PET-cGMPS, K d is the concentration of (R p )-8-Br-PET-cGMPS at the half-maximal inhibition of the channel, and n is the Hill coefficient. The K d value for this inhibitory effect was 0.12 Ϯ 0.01 M (n ϭ 5 patches). The preceding equations were used throughout the remainder of the experiments to determine the concentration-response relationship for channel inhibition by the compounds used. (R p )-8-pCPT-cAMPS had no effect on the channel (Fig. 3, B and C). In Fig. 4, we tested the effect of (R p )-pCPT-cGMP, another potent inhibitor of PKG, using cell-attached patches. The po-tentiating effect of SNP was suppressed by (R p )-pCPT-cGMP (100 M) in a reversible manner in all of the cells tested (n ϭ 5 patches); the average P o before and during the addition of (R p )-pCPT-cGMP was 0.24 Ϯ 0.05 and 0.09 Ϯ 0.05 (p Ͻ 0.05, n ϭ 5 patches). These results suggest that NO donors facilitate the pinacidil-induced K ATP channel activity via a cGMP/PKGdependent mechanism.
Effects of PKG Activators on K ATP Channels in Rabbit Ventricular Myocytes-The involvement of PKG in the regulation of K ATP channels was further confirmed in an experiment using a potent activator of PKG, (S p )-8-Br-PET-cGMPS (Fig. 5, A and C). It has been reported that (S p )-8-Br-PET-cGMPS is the only compound that displays significant selectivity for PKG over PKA (34). In seven patches, (S p )-8-Br-PET-cGMPS stimulated half-maximal K ATP channel activity at a concentration of 4.12 Ϯ 0.96 M (Fig. 5D). In five separate patches, the specific activator for PKA, (S p )-5,6-DCl-cBIMPS (35), had no effect on the channel (Fig. 5, B and C). In Fig. 5E, the pinacidil-induced K ATP channel activity was reversibly facilitated by another PKG activator, 8-pCPT-cGMP (36). The average P o increased 1.78 Ϯ 0.12 times following the addition of 100 M 8-pCPT-cGMP (P o ϭ 0.17 Ϯ 0.09) compared with the P o (0.09 Ϯ 0.04) recorded before the addition of 8-pCPT-cGMP (p Ͻ 0.05, n ϭ 6 patches). The pinacidil-induced single channel activity was inhibited by the subsequent application of 30 M glibenclamide (P o ϭ 0.002 Ϯ 0.001). Summarized data are shown in Fig. 3F.
Effect of PKG Activation on K ATP Channels in Excised Insideout Patches-To evaluate more directly the involvement of the cGMP/PKG-dependent mechanism in the activation of the K ATP channel, we tested various combinations of cGMP, PKG, and ATP in excised inside-out patches. cGMP alone (100 M; p Ͼ 0.05, n ϭ 4 patches), cGMP and ATP together (100 M each; p Ͼ 0.05, n ϭ 4 patches), PKG alone (5 units/l; p Ͼ 0.05, n ϭ 4 patches), and PKG and cGMP together (5 units/l and 100 M each; p Ͼ 0.05, n ϭ 4 patches) had no effect on the channel activity (data not shown). Facilitation of K ATP channel activity in excised inside-out patches was only observed when PKG was applied in the presence of ATP and cGMP together. A representative case is shown in Fig. 6A. After excision of the patch, ATP (100 M) and cGMP (100 M) inhibited spontaneous K ATP channel opening. In the continuous presence of ATP and cGMP, increasing concentrations of PKG enhanced the channel activity. The K d value for this stimulatory effect was 0.08 Ϯ 0.02 units/l (Fig. 6B, n ϭ 6 patches).
(R p )-8-Br-PET-cGMPS inhibited the PKG-induced K ATP channel activation in a concentration-dependent fashion (Fig.  6C). The K d value for this inhibitory effect was 0.09 Ϯ 0.02 M (Fig. 6D, n ϭ 5 patches). The specificity of PKG in stimulating channel activity was confirmed using heat-inactivated PKG. In the continuous presence of 100 M/l ATP at the intracellular surface, the application of heat-treated PKG (5 units/l) to the intracellular surface with cGMP (100 M) failed to enhance the channel activity (data not shown). Additionally, the removal of Mg 2ϩ from the bath solution (by the addition of 1 mM EDTA) prevented PKG activation (data not shown).
In Fig. 6E, we replaced ATP with ATP␥S, a nonhydrolyzable analog of ATP, under similar experimental conditions. K ATP channels were inhibited by ATP␥S (100 M) and cGMP (100 M), and P o was reduced from 0.317 to 0.063 (Fig. 6E). However, the subsequent addition of PKG (5 units/l) failed to enhance the channel activity (P o ϭ 0.083). Similar effects were observed in four other patches.
PKG-induced Activation of K ATP Channels Is Reversed by Protein Phosphatase-Taken together, the preceding results suggested that PKG activated K ATP channels by an apparent phosphorylation-dependent mechanism. If PKG phosphorylated K ATP channels directly, one would have expected protein phosphatase to inhibit PKG action. We tested the effect of protein phosphatase (PP2A) on the PKG-induced K ATP channel activation.  Ϫ40 mV). B, the relationship between the PKG concentration and relative channel activity from a series of experiments similar to that shown in A. The channel activity for PKG was normalized using the equation under "Results." The solid line was drawn from calculations that are described under "Results." C, the effect of the PKG inhibitor (R p )-8-Br-PET-cGMP on PKG activation-induced K ATP channel activity, demonstrating the specificity of PKG in stimulating the K ATP channel activity. PKG increased the channel activity in the presence of a combination of ATP, cGMP, and (R p )-8-pCPT-cAMPS. In the same patch, various concentrations of (R p )-8-Br-PET-cGMP were added to the bath solution. Note that (R p )-8-Br-PET-cGMP reversed the stimulatory effects of PKG on K ATP channel activity. D, the relationship between (R p )-8-Br-PET-cGMP concentration and relative channel activity from a series of experiments similar to that shown in C. The channel activity for (R p )-8-Br-PET-cGMP was normalized using the equation under "Results." The solid line was drawn from calculations that are described under "Results." E, the effect of replacing ATP with ATP␥S on the channel activity in inside-out patches. Note that PKG together with cGMP had no effect on ATP␥S-inhibited channels. Current recording was from an inside-out patch held at Ϫ40 mV. Data were sampled at 20 kHz and filtered at 1 kHz. The dashed line indicates the zero current level. opening, and P o was reduced from 0.163 to 0.012. In the continuous presence of ATP and cGMP, PKG (5 units/l) enhanced the channel activity (P o ϭ 0.122). The application of exogenous PP2A inhibited the PKG-mediated K ATP channel activity (P o ϭ 0.032). A similar decrease in channel activity with PP2A was observed in five other patches (Fig. 7B).
The effect of PP2A was in turn inhibited by OA, a potent inhibitor of type 1 and 2A protein phosphatases (Fig. 7C). OA was used at a low concentration (5 nM) to specifically block the activity of PP2A in excised inside-out patches. The application of PKG in the presence of ATP and cGMP caused an increase in the channel activity (P o ϭ 0.27). When ATP, cGMP, and PKG were then washed out and OA was applied to the patches, K ATP channel activity was maintained (P o ϭ 0.28). In the presence of OA, PP2A did not alter the channel activity (P o ϭ 0.27). Identical results were obtained in six out of seven patches examined (Fig. 7D). As shown in Fig. 7E, similar effects were observed even in the presence of a PKA inhibitor.
Effect of PKG on the Properties of K ATP Channels in Rabbit Ventricular Myocytes- Fig. 8A shows the all-points histograms of single channel current amplitude made from the same patch under control conditions and in the presence of PKG activation. The I-V relationships obtained from patches under control conditions and in the presence of PKG activation are shown in Fig.  8B. PKG activation had no effect on the I-V relationship or slope conductance value (from 78.9 Ϯ 5.3 to 78.5 Ϯ 3.9 pS, n ϭ 6 patches).
To examine the effect of PKG activation on the gating kinetics of the channels, the open time and closed time histograms were calculated at a membrane potential of Ϫ50 mV relative to the reversal potential. The open-time histogram (Fig. 8C, a), which was analyzed from the current record filtered at a cut-off frequency of 5 kHz, revealed a single exponential distribution with a time constant of 1.38 ms under control conditions. In the presence of PKG activation, the open time constant ( o ) did not differ from that seen in the absence of PKG activation ( o ϭ 1.40 ms). Burst lifetime was defined as the opening period observed in the records filtered at a cut-off frequency of 0.2 kHz. The histogram of burst duration consisted of a single exponential distribution (Fig. 8C, c). Its time constant, designed as b , was markedly prolonged by PKG activation (from 20 to 35 ms). The histogram of time closed within bursts was best fitted to a single exponential function (Fig. 8C, b). This analysis was performed discarding closing times longer than 4 ms, with filtering at a cut-off frequency of 5 kHz. The constant of the time closed within bursts was designed as c . The value of c was not changed markedly by PKG activation (from 0.30 to 0.31 ms). The closed time between bursts was analyzed by using records filtered at a cut-off frequency of 0.2 kHz (Fig. 8C,  d). The histogram was fitted using a biexponential function, with the time constants of fast ( c1 ) and slow ( c2 ) components. The value of c1 was influenced by PKG activation (from 74 to 40 ms). The value of c2 was 293 ms under control conditions. This value was markedly decreased to 105 ms by PKG activation. These findings suggest that PKG activation increases the channel activity by increasing burst duration and decreasing the interburst interval.

DISCUSSION
In cardiac muscle, K ATP channels, which open when the intracellular ATP concentration ([ATP] i ) falls below a critical level, are known to be involved in ischemic preconditioning, a mechanism that protects the heart against ischemic injury (37)(38)(39). In addition to the more obvious effect of diminishing [ATP] i on the activation of this channel, various extracellular and intracellular modulators, such as adenosine (40), protein kinase C (41), ADP (42), and lactate (43), have been proposed to stimulate K ATP channel activity and thus contribute to ischemic preconditioning. The cGMP/PKG-pathway has also been suggested to play a cardioprotective role against ischemicreperfusion injury (10,11,15), and it was proposed that the modulation of Ca 2ϩ availability (44,45), reducing myofilament sensitivity to Ca 2ϩ (46), and vasodilation with antiplatelet effects (47) were involved in cardioprotectivity. Although the cGMP/PKG-pathway and K ATP channels have both been implicated in mediating the protective effect of ischemic preconditioning, the signaling mechanism by which they are linked remains poorly understood. The objective of this study was to determine the mechanism by which PKG activates K ATP channels and thus could link these two effectors in a signaling pathway that produces cardioprotection during ischemic preconditioning.
Our findings demonstrate that NO donors and PKG activators facilitate the pinacidil-induced K ATP channel activity, and a selective PKG inhibitor prevents their effects. PKG, in the presence of both cGMP and ATP, increases the channel activity. This action of PKG is prevented by heat inactivation, replacing ATP with ATP␥S (a nonhydrolyzable analog of ATP), removing Mg 2ϩ from the internal solution, applying a PKG inhibitor, or adding exogenous PP2A. Taken together, these results imply that a series of signal transduction pathways is involved in the regulation of K ATP channels; NO donors release NO, which activates guanylate cyclase in cardiac myocytes, causing cGMP accumulation and the activation of PKG, which subsequently phosphorylates and activates the K ATP channels.
However, some previous studies reported contradictory results. Shinbo and Iijima (27) demonstrated that NO potentiated the K ATP channel activity produced by K ϩ channel openers in guinea pig ventricular myocytes. Under the same conditions, however, these investigators found that 8-Br-cGMP inhibited K ATP channels. Therefore, they suggested that the mechanism of the NO-induced potentiation of K ATP channels was independent of the cGMP/PKG-signaling pathway in the heart. In addition, Tsuura et al. (24) reported that in rat ventricular myocytes, NO donors had no effect on K ATP channels. Differences in experimental conditions and animal species between their study and ours may be responsible for the discrepancies in results.
The major variations in experimental conditions are the mode of patch recordings and the compounds used. Shinbo and Iijima (27) and Tsuura et al. (24) carried out experiments using the whole-cell and cell-attached modes of the patch-clamp method, whereas we only used the cell-attached mode. It is difficult to know how the mode of patch clamping affects the action of NO/cGMP on K ATP channels. Since our study showed that the effect of NO/cGMP on K ATP channels was not direct, but required a series of signal transductions involving PKG activation and phosphorylation, it is plausible that dialysis of the cellular components during whole-cell patch clamping attenuated the effect of NO/cGMP. Shinbo and Iijima (27) used 8-Br-cGMP (100 -500 M), claiming that it is a specific activator of PKG. However, this compound has several major limitations that might influence the way in which their experimental data should be interpreted. First, 8-Br-cGMP at 100 -500 M can also strongly activate PKA because half-maximal stimulation of PKA occurs at 2.8 -12 M 8-Br-cGMP (36). Second, 8-Br-cGMP is hydrolyzed by certain phosphodiesterases, generating the corresponding 5Ј-monophosphate analogs and, subsequently, the nucleoside analogs (36). Third, 8-Br-cGMP is as polar as cAMP (48), which implies insufficient membrane permeability. Therefore, it is unclear whether the 8-Br-cGMP-mediated effects on K ATP channels observed by Shinbo and Iijima (27) indeed occurred via PKG. In our study, to avoid misinterpretation of the experimental data, we utilized cyclic nucleotide analogs as new bio- FIG. 8. Effects of PKG activation on the properties of K ATP channels from an inside-out patch. A, all-points histograms of current amplitude made from the same patch under control conditions (left panel) and in the presence of PKG activation (right panel). *, the current level during K ATP channel opening. The patch was held at Ϫ40 mV. Note that the amplitude of the channel was not affected by PKG activation. B, the current-voltage relationship made from the same patch under control conditions (E) and in the presence of PKG activation (‚). Note that PKG did not change the conductance of the channel current. C, the effects of PKG on the distribution of the open and closed times of K ATP channels. The histograms of the open (a) and closed (b) time within bursts were analyzed from the current records at 5 kHz. The histograms of the burst (c) and interburst (d) durations were analyzed from the current records at 0.2 kHz. The membrane potential was Ϫ50 mV. Time constants of the interburst duration histogram were fitted to two exponential equations (fast and slow). The others were fitted to single exponential equations. PKG increased the burst duration and shortened interburst duration. chemical tools. These analogs are highly membrane-permeable, stable against phosphodiesterase hydrolysis, and display sufficient PKG/PKA specificity. (S p )-8-Br-PEP-cGMPS was used as a PKG activator, (R p )-8-Br-PEP-cGMPS as a PKG inhibitor, (S p )-5,6-DCl-cBIMPS as a PKA activator, and (R p )-8-pCPT-cAMPS as a PKA inhibitor. Since (S p )-8-Br-PEP-cGMPS exhibits sufficient selectivity for PKG versus PKA (34), this compound is the best choice for triggering PKG activation. (R p )-8-Br-PEP-cGMPS was found to be the most potent inhibitor of PKG and inhibited PKG more than 300 times more potently than PKA (34). (S p )-5,6-DCl-cBIMPS was reported to be the best specific activator of PKA and a 300-fold less potent activator of PKG (35). (R p )-8-pCPT-cAMPS is a better inhibitor of PKA (49). Although cGMP and cAMP analogs normally have effects on the PKG and PKA pathways, respectively, it should be recognized that high concentrations of cGMP analogs can at least partially affect PKA signaling, and the converse is also true for cAMP analogs. Therefore, we carefully constructed concentration-response curves using these analogs to determine the relative potencies of the activators and inhibitors. We found that (S p )-8-Br-PEP-cGMPS facilitated channel activity at concentrations of up to 100 M and (R p )-8-Br-PEP-cGMPS inhibited the channel activity at concentrations of up to 10 M, whereas (S p )-5,6-DCl-cBIMPS and (R p )-8-pCPT-cAMPS had no effect on the channels at concentrations of up to 100 and 10 M, respectively. Additionally, we observed the effects of NO/cGMP analogs under conditions in which PKA was inhibited by (R p )-8-pCPT-cAMPS. These results ruled out the importance of PKA in K ATP channel regulation, indicating that the observed effects of NO/cGMP analogs are truly PKG-mediated.
If this interpretation were correct, one would expect that activation of PKG would also directly increase K ATP channel activity, even in excised patches. Neither Tsuura et al. (24) nor Shinbo and Iijima (27) directly tested the effect of PKG on K ATP channels. We tested the effects of exogenous PKG using insideout patches. Since PKG and PKA share some similarities in protein substrate sequence specificity, it is possible that PKG phosphorylates PKA-selective sites. To determine which is the more potent kinase in activating K ATP channels, we carefully constructed concentration-response curves using PKG and PKA (catalytic subunit). This experiment produced the following lines of evidence clearly showing that PKG acts directly on the K ATP channels: (a) in contrast to PKG, the catalytic subunit of PKA at the same concentration ranges had no effect on the channel (data not shown); (b) the stimulating effects of PKG still occurred under conditions in which PKA was inhibited by (R p )-8-pCPT-cAMPS; and (c) K ATP channel stimulation by PKG was reversibly inhibited by (R p )-8-Br-PEP-cGMPS.
The preceding findings strongly suggest that K ATP channels are stimulated by PKG but not by PKA. If PKG phosphorylated K ATP channels directly, one would expect that protein phosphatase could inhibit PKG action. Our findings demonstrated that an endogenous membrane-associated PP2A was responsible for the reversal of PKG-mediated activation of K ATP channels and that exogenous PP2A inhibited PKG-induced K ATP channel activity. Furthermore, replacement of ATP with ATP␥S and removal of Mg 2ϩ from the internal solution prevented the effect of PKG. Together, these results show that PKG activates K ATP channels by an apparent phosphorylation-dependent mechanism, suggesting that the activity of a K ATP channel depends on its net phosphorylation state, and this in turn depends on the balance between opposing PKG and phosphatase activities. Presumably, these processes of phosphorylation and dephosphorylation provide a mechanism by which K ATP channel activity can be reversibly controlled. To our knowledge, our findings provide the first direct evidence that K ATP channels can be opened through PKG-dependent phosphorylation in rabbit ventricular myocytes.