Regulation of large calcium-activated potassium channels by protein phosphatase 2A.

Vasodilating agents induce relaxation of mesangial cells, in part through cGMP-mediated activation of large calcium-activated potassium channels (BKCa). Normally quiescent in cell-attached patches, the response of BKCa to nitric oxide, atrial natriuretic peptide, and dibutyryl cGMP (Bt2cGMP) is characterized by a biphasic increase and then decrease ("rundown") in open probability. Using the patch-clamp method in conjunction with phosphatase inhibitors, we investigated whether the run-down phase was the result of dephosphorylation by an endogenous protein phosphatase. In cell-attached patches, cantharidic acid (500 nM), okadaic acid (100 nM), and calyculin A (100 nM), nondiscriminant inhibitors of protein phosphatases 1 (PP1) and 2A (PP2A) at these concentrations, caused a significantly greater and sustained response of BKCa to Bt2cGMP. Within 2 min, the response of BKCa to the combination of cantharidic acid and Bt2cGMP was greater than the response to these agents added separately. Incubation of mesangial cells with okadaic acid for 20 min at a concentration (5 nM) specific for PP2A increased the basal open probability of BKCa and completely inhibited rundown after activation by Bt2cGMP. Incubation with calyculin A (10 nM), a more potent inhibitor of PP1, did not affect BKCa activity. In inside-out patches, Bt2cGMP plus MgATP caused a sustained activation of BKCa that was inhibited by exogenous PP2A but not PP1. It is concluded that either BKCa or a tightly associated regulator of BKCa is a common substrate for endogenous cGMP-activated protein kinase, which activates BKCa, and PP2A, which inactivates BKCa, in human mesangial cells.

acterized in mesangial cells (5,6) and vascular smooth muscle (7), are not involved in setting resting potential but respond in a negative feedback manner to agonist-induced increases in contractile tone. Agonists such as angiotensin II elevate intracellular calcium and depolarize the membrane potential, producing an activation of BK Ca . The hyperpolarizing membrane potential inhibits further entry of cell calcium by inactivating voltage-gated calcium channels. The gain in this feedback mechanism is increased by smooth muscle relaxants such as nitric oxide and atrial natriuretic peptide that, via cGMPactivated kinase, lower the voltage and calcium thresholds for activating BK Ca (8,9). However, activation of BK Ca "on cell" by vasorelaxants or Bt 2 cGMP is followed by an inactivation or run-down phase, in which BK Ca returns to base line 20 s after peak activity.
Substrate regulation by phosphorylation is a dynamic balance between the forward kinase phosphorylation and the reverse dephosphorylation by a protein phosphatase. Several studies have now shown that vasorelaxants activate BK Ca through guanylyl cyclase and cGMP-dependent protein kinase in both smooth muscle and mesangial cells (9 -11); however, the role of protein phosphatase has been addressed only very recently (9,12). Two such studies on tracheal smooth muscle and neurohypophyseal cells supported the notion that cGMP stimulated BK Ca by activating protein phosphatase 2A (12,13), which activated BK Ca by dephosphorylation. In contrast, our laboratory previously showed that the mesangial BK Ca was activated by cGMP-activated protein kinase in the presence of okadaic acid, an inhibitor of protein phosphatases 1 and 2A (9,13), suggesting that these phosphatases inactivated, rather than activated, the mesangial BK Ca .
The present studies were performed to elucidate the signaling pathways involved in the run-down phase following activation of BK Ca by cGMP-dependent protein kinase. Using the cell-attached configuration and established phosphatase inhibitors, we specifically investigated the specific endogenous protein phosphatase involved in regulating BK Ca . Using inside-out patches, we then determined the regulation of BK Ca by protein phosphatases 1 and 2A.
Patch-Clamp Methods-Mesangial cells were prepared for analysis of single BK Ca channels using standard patch-clamp techniques previously described (5,15). Current recordings were made after obtaining gigaohm seals with the patch electrode on the surface of the cell (cell attached) or after withdrawing the patch (excised, inside out). The unitary current (i), defined as zero for the closed state, was determined as the mean of the best fit Gaussian distribution of the amplitude histograms. Channels were considered in an open state when the current was Ͼ(n Ϫ 1 ⁄2)i and Ͻ(n ϩ 1 ⁄2)i, where n is the maximum number of current levels observed. The probability of a channel existing in an open state (P o ) is defined as the time spent in the open state divided by the total time of the recording. In all cases, ϪV p implies the holding potential relative to the pipette.
Experimental Design and Solutions-In all cell-attached experiments, the pipette solution contained 140 mM KCl plus 10 mM HEPES buffer, pH 7.4, and the bath solution contained 135 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , and 10 mM HEPES, pH 7.4. The free Ca 2ϩ concentration of the bath, initially 1.0 mM, was adjusted to lower concentrations by buffering with EGTA as described previously (16). In excised patches, the bathing solution was replaced with 140 mM KCl, 1 M CaCl 2 , 10 mM HEPES, pH 7.4. Okadaic acid, cantharidic acid, and calyculin A, established inhibitors of PP1 and PP2A, were used to establish the role of endogenous protein phosphatase in regulation of BK Ca . While monitoring BK Ca channels in cell-attached patches, these inhibitors were added to the bathing solution in concentrations nonspecific for inhibition of PP1 and PP2A. The holding potential (ϪV p ) was either 80 mV or 0 mV, which resulted in outward and inward currents, respectively. The effects of these inhibitors on the cGMP-dependent activation of BK Ca was determined by adding inhibitor either in the continued presence of 10 M dibutyryl cGMP (Bt 2 cGMP) or at least 30 s before the addition of Bt 2 cGMP. To determine if the endogenous protein phosphatase was either 1 or 2A, cells were incubated in either 5 nM okadaic acid (specific for PP2A) or 10 nM calyculin A for 10 -30 min before obtaining a seal. Specific inhibition of BK Ca by either exogenous PP2A or PP1 was determined in inside-out patches with 140 mM KCl in the bath and holding potentials of either 40 mV or Ϫ40 mV for outward or inward currents, respectively. BK Ca were activated either by Bt 2 cGMP plus MgATP at a holding potential of 40 mV or by Bt 2 cGMP plus MgATP plus cGMP-activated protein kinase at a holding potential of Ϫ40 mV. PP2A (0.5 unit/ml) or PP1 (1 unit/ml) was added in the continued presence of Bt 2 cGMP plus MgATP with or without cGMP-activated protein kinase. Groups were compared for statistical significance using the paired t test or the ANOVA plus the Student-Newman-Keuls test as appropriate. Cyclic GMP-activated kinase was purchased from Promega. All other chemicals used in this study were purchased from Sigma or Calbiochem.

Effects of Pharmacological Inhibitors of Protein Phosphatases on Rundown of cGMP-activated BK Ca -Experiments
were performed to determine if the run-down phase after adding Bt 2 cGMP was the result of either PP1 or PP2A. Fig. 1A shows the effects of cantharidic acid (500 nM), a nonspecific inhibitor of PP1 and PP2A, on the rundown of Bt 2 cGMP-activated BK Ca in cell-attached patches in the absence of a phosphatase inhibitor (upper tracing) and with the simultaneous addition of cantharidic acid (lower tracing). The P o of BK Ca increased from Ͻ0.01 to 0.85 after 10 s with the addition of 10 M Bt 2 cGMP and then returned to base line in the next 90 s; after 2 min the P o was Ͻ0.01. However, when Bt 2 cGMP and cantharidic acid were added together, BK Ca was activated to a P o that was sustained at 0.78 after 2 min. Note that the channel amplitude diminished after addition of cantharidic acid. A decrease in the amplitude of BK Ca is the result of a decrease in electrochemical potential due to the combination of a decrease in the intracellular potassium concentration and the hyperpolarization of the membrane potential as BK Ca is activated in the cell membrane (8, 17).  In the continued presence of cantharidic acid, Bt 2 cGMP further increased BK Ca after 2 min to a sustained P o of 0.40 and 0.035 at 80 and 0 mV, respectively. Thus, cantharidic acid eliminated the rundown in the response of BK Ca to Bt 2 cGMP. Moreover, the effects of Bt 2 cGMP and phosphatase inhibitor were potentiating and independent of the holding potential.
A summary of the separate and combined effects on BK Ca of cantharidic acid and Bt 2 cGMP are shown in Fig. 1C. Two minutes after the separate additions of Bt 2 cGMP and 500 nM cantharidic acid, the open probability of BK Ca was 0.011 Ϯ 0.006 and 0.16 Ϯ 0.05, respectively. Two minutes after the addition of Bt 2 cGMP plus cantharidic acid, the P o increased to a significantly higher value of 0.48 Ϯ 0.10. This value was more than twice the value of the sum of the separate additions, suggesting a potentiated effect of Bt 2 cGMP and cantharidic acid. Fig. 2 shows the effects of 100 nM okadaic acid and 100 nM calyculin A on cGMP-activated BK Ca in cell-attached patches. At these concentrations, the phosphatase inhibitors are nonspecific for PP2A and PP1 and are effective within 120 s. As shown in Fig. 2A (ϪV p ϭ 0 mV), BK Ca activated within 5 s from Ͻ0.001 to 0.10 in response to 10 M Bt 2 cGMP and then returned to base line after approximately 5 more seconds. In the continued presence of Bt 2 cGMP, the addition of okadaic acid reactivated BK Ca to a greater and sustained value of 0.83 after 120 s. Similar results were obtained with 100 nM calyculin A (Fig. 2B, ϪV p ϭ 80 mV). The P o of BK Ca increased from Ͻ0.001 to 0.42 within 5 s after the addition of Bt 2 cGMP and then returned to base line within the next 20 s. However, after the addition of Bt 2 cGMP plus 100 nM calyculin A, the P o increased and remained at 0.65. The increase in channel amplitude at 0 mV and the reduction in amplitude at 80 mV shows that the membrane potential is hyperpolarizing, as BK Ca is activated by either okadaic acid or calyculin A.
Pharmacological Differentiation between Endogenous PP1 and PP2A-Low concentrations of okadaic acid (5 nM) and calyculin A (10 nM) were used to determine if the endogenous phosphatase was PP1 or PP2A. Although IC 50 values for phosphatase inhibition for okadaic acid and calyculin A are dependent on the substrate and purity of inhibitor, in general okadaic acid is at least 30-fold more potent for PP2A than PP1 (18,19), and calyculin A is 30-fold more potent than okadaic acid for inhibiting PP1 (19,20). Therefore, if PP1 were the endogenous phosphatase, it would be expected that BK Ca would be activated by calyculin A more than okadaic acid. If PP2A were the endogenous phosphatase, it would be expected that BK Ca would be activated more by okadaic acid than calyculin A.
In the experiments of Fig. 3, okadaic acid (5 nM) and calyculin A (10 nM) were added to the bathing medium 20 -50 min before obtaining a cell-attached patch. As shown by the upper current tracing of Fig. 3A, in the presence of okadaic acid, BK Ca   FIG. 2. Effects of 100 nM okadaic acid and 100 nM calyculin A on Bt 2 cGMP-activated BK Ca in cell-attached patches. A (0 mV), BK Ca was activated within 5 s from Ͻ0.001 to 0.10 in response to 10 M Bt 2 cGMP (DB-cGMP) and then returned to base line after approximately 5 more seconds. In the continued presence of Bt 2 cGMP, the addition of okadaic acid reactivated BK Ca to a greater and sustained value of 0.83 after 120 s. Similar results were obtained with 100 nM calyculin A (B, 80 mV). The P o of BK Ca increased from Ͻ0.001 to 0.42 within 5 s after the addition of Bt 2 cGMP and returned to base line within the next 20 s. However, after the addition of Bt 2 cGMP plus 100 nM calyculin A, the P o increased and remained at 0.65. The increase in channel amplitude at 0 mV and the reduction in amplitude at 80 mV shows that the membrane potential is hyperpolarizing as BK Ca is activated by either okadaic acid or calyculin A. All other conditions are the same as in Fig. 1. In the control cells, the P o was increased by Bt 2 cGMP to a value of 0.19 Ϯ 0.04 after 5 s and then ran down to base line after 120 s (n ϭ 6). In the presence of 5 nM okadaic acid, the basal P o was 0.18 Ϯ 0.07 (n ϭ 4) and increased to 0.29 Ϯ 0.05 and 0.25 Ϯ 0.05 after 5 s and 120 s, respectively. After incubation with calyculin A, the P o (n ϭ 4) increased from 0.002 Ϯ 0.001 to 0.20 Ϯ 0.05 after 5 s and returned near the base line value (0.008 Ϯ 0.006) after 120 s. The asterisks denote significant (p Ͻ 0.05) increases in basal and Bt 2 cGMP (120 s) when compared with the control and calyculin A groups using the ANOVA plus the Student-Newmann-Keuls test.

FIG. 3. Results of cell-attached experiments showing typical tracings (A) and a summary (B) of the activation of BK Ca by Bt 2 cGMP after incubating cells for 20 -50 min with 5 nM okadaic acid and 10 nM calyculin
was active (P o ϭ 0.063) after obtaining the seal at 80 mV. The addition of Bt 2 cGMP increased the P o further to a sustained value of 0.32. As shown in Fig. 3B, BK Ca was relatively quiescent (P o Ͻ 0.005) after the addition of calyculin A and was activated transiently to 0.227 on the addition of Bt 2 cGMP. This rundown to base line approximately 20 s after peak activation was similar to control experiments (see Fig. 1A). These results are summarized in Fig. 3B. In control cells, the P o of BK Ca was increased by Bt 2 cGMP to a value of 0.19 Ϯ 0.04 after 5 s and ran down to base line after 120 s (n ϭ 6). After incubation with okadaic acid, the basal P o was 0.18 Ϯ 0.07 and increased to 0.29 Ϯ 0.05 and 0.25 Ϯ 0.05 after 5 and 120 s, respectively (n ϭ 4). After incubation with calyculin A (n ϭ 4), the P o increased from a basal of 0.002 Ϯ 0.001 to 0.20 Ϯ 0.05 after 5 s and returned near the base line value (0.008 Ϯ 0.006) after 120 s. Using the ANOVA plus the Student-Newman-Keuls test, the effects of okadaic acid at basal and Bt 2 cGMP (120 s) were significantly greater than the respective P o values for control and calyculin A. These results indicate that BK Ca are maintained quiescent in cell-attached patches by endogenous PP2A. Moreover, specific inhibition of PP2A (by 5 nM okadaic acid) prevents the rundown after activation of BK Ca by Bt 2 cGMP in cell-attached patches.
Effects of Exogenous PP2A and PP1-The inside-out patch configuration was used to determine the effects of exogenous PP2A and PP1 on BK Ca . Fig. 4 shows the inactivation of cGMPactivated BK Ca by PP2A. As shown in the continuous tracing of Fig. 4A, dibutyryl cyclic GMP plus MgATP activated BK Ca from 0.029 to 0.332, and the subsequent addition of PP2A inactivated BK Ca to 0.098. However, there was no effect of PP1 on cGMP-activated BK Ca (not shown). These data are summarized in the bar graph of Fig. 4B. In each experiment, after activation of BK Ca by Bt 2 cGMP plus MgATP in either the absence or presence of cGMP-activated protein kinase, PP2A (n ϭ 5) decreased the P o from 0.57 Ϯ 0.11 to 0.40 Ϯ 0.14 (p Ͻ 0.025, paired t test). However, the P o was 0.66 Ϯ 0.11 and 0.70 Ϯ 0.13 (n ϭ 3; not significant) before and after the addition of PP1. DISCUSSION This study further defined the signal transduction pathways for regulating BK Ca channels in a contractile cell. Three phosphatase inhibitors, in concentrations that inhibit both PP1 and PP2A, caused a larger and more sustained increase in open probability of BK Ca in response to Bt 2 cGMP, the second messenger mediator for relaxation by nitric oxide and atrial natriuretic peptide. Basal open probability of BK Ca was increased by okadaic acid in concentrations specific for PP2A but not by calyculin A, a more potent inhibitor of PP1, indicating that endogenous PP2A but not PP1 was maintaining BK Ca in a dephosphorylated quiescent state. It was shown that exogenous PP2A, but not PP1, applied to the cytosolic side of BK Ca in inside-out patches can specifically inhibit the activation of BK Ca by cGMP-dependent protein kinase.
Biphasic Response of BK Ca to cGMP-In both smooth muscle and mesangial cells, BK Ca are activated by cGMP-dependent protein kinase (7,9). However, after activation by Bt 2 cGMP in cell-attached patches, the open probability of the mesangial BK Ca rapidly runs down to base-line levels. The present study shows that the run-down phase is due to the presence of PP2A, which would dephosphorylate BK Ca . Phosphatase-induced channel rundown has been more commonly described for channels in excised patches. Kubokawa et al. (21) found that renal K d(ATP) channels run down in excised patches due primarily to the presence of PP2A. However, phosphatase-induced rundown is not only found in excised patches; it was also shown that okadaic acid prevents rundown of Ba 2ϩ current (whole cell) in dissociated helix neurons (22).
It is not understood why the effects of cGMP-dependent kinase are transient and ultimately overcome by a phosphatase that presumably dephosphorylates and inactivates BK Ca despite the continued presence of Bt 2 cGMP. However, several mechanisms could be involved in the temporary inhibition and then activation of a protein phosphatase to initiate the rundown phase. A similar type of biphasic activation was demonstrated for Ca 2ϩ /calmodulin-dependent protein kinase II, also a substrate for PP2A (23). An increase in intracellular Ca 2ϩ in the rat brain is accompanied by a sequential autophosphorylated increase and then decrease in phosphorylation level of Ca 2ϩ /calmodulin-dependent protein kinase II. The decrease in phosphorylation was blocked by 1 nM okadaic acid. It was suggested by these authors that an increase in intracellular calcium autophosphorylated a serine/threonine protein kinase (described by Guo et al. (24)) that would temporarily phosphorylate and inhibit PP2A. In time, PP2A would autodephosphorylate and inactivate Ca 2ϩ /calmodulin-dependent protein kinase II. A similar mechanism may be involved whereby cGMP temporarily activates an inhibitor of PP2A. Although a recent study described inhibition of PP1 by cGMP-dependent protein kinase (25), a cGMP-activated inhibitor of PP2A has not been described.
Regulation of BK Ca by Phosphatases-Although several studies have described regulation of ion-selective channels by cAMP-and cGMP-dependent protein kinases, the reversal of channel phosphorylation by phosphoprotein phosphatases has been investigated only recently (26 -31). The present study is one of a few that have now implicated PP2A as a physiological regulator of BK Ca channels (12, 13, 32). However, for cGMPactivated protein kinase-activated BK Ca , at least two previous studies using three different cell types (12, 13) have shown either that PP2A activates BK Ca or cGMP-activated protein kinase does not activate BK Ca in the presence of inhibitors of PP2A. These results contrast with our study which showed that BK Ca was inactivated by PP2A and activated by either cGMPactivated protein kinase or inhibitors of PP2A.
Our disparate results may be explained by another study by Reinhart et al. (32) who have shown that BK Ca from the brain expresses two types of channels in planar bilayers with respect to PP2A regulation. Type 1 channels are activated by cAMP-activated protein kinase and inactivated by PP2A. Type 2 BK Ca channels are inactivated by cAMP-activated protein kinase and activated by PP2A. Although this was a protein kinase A and not a cGMP-activated protein kinase-activated mechanism, it is possible that there are two types of BK Ca with respect to regulation of phosphorylation and dephosphorylation by cGMP-activated kinase. However, these type 2 BK Ca channels have not been observed in mesangial cells. 2 When the properties of the PP2Aactivated (type 2) channels in tracheal smooth muscle (12) are compared with the mesangial BK Ca of this study, we find that the reported single channel conductance (in symmetrical 140 mM KCl) of the tracheal BK Ca (257 Ϯ 13 picosiemens) is somewhat larger than the mesangial BK Ca (206 Ϯ 18 picosiemens, see Ref. 5). However, voltage-gated activation response in 1 M Ca 2ϩ is similar (see Refs. 5 and 12). It remains to be determined if these are distinct isoforms of BK Ca .
The dominant type of BK Ca in smooth muscle could be determined by the contractile response to low concentrations of okadaic acid. Since PP1 dephosphorylates myosin light chain kinase (33), higher concentrations of okadaic acid would be expected to induce contractions. If type 1 channels predominate, it would be expected that okadaic acid in concentrations of 1-10 nM would inhibit smooth muscle contraction. That range of concentrations of okadaic acid would induce smooth muscle contraction if type 2 channels predominate. In support of type 1 channels, at least two studies have shown that low concentrations of okadaic acid relax vascular smooth muscle (34,35). Although the kinase specific for activation of the mesangial BK Ca appears identical to smooth muscle, the effects of okadaic acid on mesangial contraction have not been determined.
It would not be surprising if BK Ca is regulated by different mechanisms in different cell types. Tseng-Crank et al. (36) found that HSLO, the human gene encoding the calcium-activated potassium channel, contains multiple splice variants in the brain. It was recently found that the mesangial BK Ca contains at least two of these variants of HSLO (37). It is therefore feasible that the differential expression of these variants would confer different sites for regulation by kinases and phosphatases.
In summary, we have demonstrated that the mesangial BK Ca is not regulated by an all-or-none phosphorylation or dephosphorylation mechanism but rather by a dynamic enzymatic balance between cGMP-activated protein kinase and protein phosphatase 2A, which specifically activate and inactivate BK Ca , respectively. It remains to be established, however, if there are multiple molecular variants of BK Ca that are differentially modulated by kinase/phosphatase signal transduction mechanisms.