Conditional and Unconditional Inhibition of Calcium-activated Potassium Channels by Reversible Protein Phosphorylation*

Large conductance, calcium-activated potassium channels (BKCa or maxi-K) are important determinants of membrane excitability in many cell types. We used patch clamp techniques to study the biochemical regulation of native BKCa channel proteins by endogenous Ser/Thr-directed protein kinases and phosphatases in cell-free membrane patches from rat pituitary tumor cells (GH4C1). When protein kinase activity was blocked by removing ATP, endogenous protein phosphatases slowly increased BKCa channel activity approximately 3-fold. Dephosphorylated channels could be activated fully by physiological increases in cytoplasmic calcium or membrane depolarization. In contrast, endogenous protein kinases inhibited BKCa channel activity at two functionally distinct sites. A closely associated, cAMP-dependent protein kinase rapidly reduced channel activity in a conditional manner that could be overcome completely by increasing cytoplasmic free calcium 3-fold or 20 mV further depolarization. Phosphorylation at a pharmacologically distinct site inhibited channel activity unconditionally by reducing availability to approximately half that of maximum at all physiological calcium and voltages. Conditional versus unconditional inhibition of BKCa channel activity through different protein kinases provides cells with a powerful computational mechanism for regulating membrane excitability.

Large conductance, calcium-activated potassium channels (BK Ca or maxi-K channels) are uniquely powerful determinants of electrical activity in the nervous, endocrine, and vascular systems, since they respond directly to both membrane depolarization and calcium accumulation (1). The sensitivity of BK Ca channels to such physiological stimuli can be modified by enzyme pathways mediating reversible phosphorylation of the ion channel proteins or other closely associated regulatory proteins (2). Such reversible protein phosphorylation is the primary mechanism for regulating ion channel activity on the physiological time scale of seconds and minutes (2)(3)(4). Modification of the behavior of BK Ca channels has profound effects on the frequency and duration of action potentials in excitable cells, thereby controlling the physiological function of these cells.
Electrophysiological measurements indicate that the activity of BK Ca channels in native cells may be up-or down-regulated by reversible protein phosphorylation. For example, protein kinase-induced phosphorylation enhances calcium-activated potassium currents in smooth muscle (5,6), whereas in photoreceptors (7), hippocampal neurones (8 -11), and neurendocrine cells (12), the currents are inhibited by kinase activity. BK Ca channels exist as multimeric protein complexes composed of two integral membrane subunits, the pore-forming ␣ subunit and the regulatory ␤ subunit (1). All ␣ subunits are apparently coded by the same gene, the slo gene, but alternative RNA splicing during development produces functionally distinct channel proteins in different cell types (13)(14)(15). Hence, opposite modulatory effects of protein phosphorylation are likely to reflect differential modulation of BK Ca channel subtypes derived from alternately spliced mRNA transcripts or could arise as a result of variations in the ␣ and ␤ subunit composition of the channel complexes (1,16) or more indirect signaling cascades.
We have used the patch-clamp technique to investigate the modulation of BK Ca channel behavior by reversible protein phosphorylation at the molecular level by recording channel opening behavior in cell-free patches of native membrane. Thus, we report studies of BK Ca channel modulation under controlled conditions in the absence of alternative splicing or diffusible regulatory proteins, where the open probability (P o ) of the channels was determined under a variety of experimental conditions to promote or inhibit protein phosphorylation. In mammalian endocrine and nervous systems, most examples of BK Ca channel regulation by hormones and neurotransmitters involve channel inhibition by protein phosphorylation (1, 9 -11). Similarly, the BK Ca channels in rat pituitary tumor cells (GH 4 C 1 cells) are also inhibited by phosphorylation. In these cells, maximal stimulation of either protein kinase A (PKA) or C (PKC) completely suppresses BK Ca channel activity in the physiological voltage range (17,18); conversely, hormones that inhibit secretion from GH 4 C 1 cells have been shown to stimulate BK Ca channels through protein dephosphorylation (12,19). Thus, the BK Ca channels in GH 4 C 1 cells are regulated in the same way as many other physiologically relevant examples of BK Ca channels in the brain. For this reason, we have chosen to study the effects of reversible protein phosphorylation on the calcium and voltage dependence of single native BK Ca channels from GH 4 C 1 cells. Previous studies at the single-channel level have focused on BK Ca channels that are stimulated by protein phosphorylation (5, 20 -23). In this paper, we report evidence of BK Ca channel modulation using a cell-free system that qualitatively and quantitatively reproduces the effects of hormones on intact cells. This is the first study of native BK Ca channels in native membranes that systematically varies all three of the molecular mechanisms known to regulate these channels: voltage, intracellular free calcium concentration, and protein phosphorylation. BK Ca channel modulation has not previously been examined at sufficiently high open probabilities to address the molecular mechanism of modulation across the complete range of channel activity. To measure the effects of reversible protein phosphorylation on BK Ca channel activity reproducibly over the entire physiological range, 0.01 Յ P o Յ 0.9, we have found it essential to use 1,2-bis(aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid (BAPTA) 1 and its dibromo analogue to buffer calcium on the cytoplasmic side of the membrane patches. EGTA, which saturates at submicromolar concentrations of calcium (24 -26), is an ineffective buffer of physiological calcium transients. In contrast, BAPTA and dibromo-BAPTA bind calcium rapidly and independently of pH and physiological magnesium concentrations with an affinity that allows effective buffering of free calcium at concentrations up to and above 10 M (27). With this experimental precaution, our measurements have revealed a novel property of BK Ca channel regulation by protein phosphorylation. Our findings have important implications for the control of BK Ca channel proteins and demonstrate that reversible protein phosphorylation cascades play an integral role in controlling cellular excitability through their actions on BK Ca channels.

EXPERIMENTAL PROCEDURES
Cell Culture-BK Ca channel activity was studied in GH 4 C 1 cells, an immortal cell line derived from rat anterior pituitary tumor cells (28). Cells were maintained at 37°C in sterile Ham's F-10 culture medium supplemented with fetal bovine serum (2.5%, vol), equine serum (12.5%, vol), and antibiotics (penicillin and streptomycin). Cells were plated onto plain or collagen-coated glass coverslips 1-6 days before patch clamp recording, and the culture medium was replaced every 3-4 days. Cells from culture passage 2-28 were used in these studies.
Electrophysiological Recording-The activity of single BK Ca channels was investigated using conventional patch clamp recording techniques (29). Patch pipettes were fabricated from borosilicate glass capillary tubing (7052; Garner Glass Co, Claremont, CA), the pipette shank was coated with Sylgard (Dow Corning, Midland, MI), and the tips were fire-polished just before use. These pipettes had resistances in the range of 4 -8 megaohms when filled with electrolyte (see below for composition). Single-channel activity was recorded at room temperature from cell-attached patches of membrane on intact cells and from cell-free, inside-out patches using an Axopatch 1C amplifier and TL-1 interface (Axon Instruments, Burlingame, CA). Data were recorded and analyzed using pClamp software, version 6.0.1 (Axon Instruments). Voltage-dependent channel activity was measured over the range Ϫ40 to 100 mV, with the applied voltage stepped at random. BK Ca channels were identified by their conductance, ϳ120 picosiemens in a physiological [K ϩ ] gradient. The stochastic behavior of ion channels in these patches was interpreted using standard analytical procedures. Channel open probability (P o ) was determined from continuous records of channel activity over 10, 20, or 30 s at a given potential; P o may show some variability when measured over shorter periods (30). Excised membrane patches commonly contained more than one active BK Ca channel. The number of active channels could be counted with confidence (N Յ 3) when all the channels were stimulated to open simultaneously, giving high open probabilities (P o near 1) during a 10-s period. In these patches, P o was calculated according to the relationship, where i ϭ 1 to N, N is the maximum number of channels open simultaneously at the most depolarized membrane potentials (positive to ϩ60 mV; N Յ 3), t t is the total time of the recording and t i is the cumulative time during which exactly i channels are open. When channel behavior in multichannel patches was not monitored across the complete voltage range, activity is expressed as N*P o . Data are expressed as the mean value Ϯ S.D. from multiple independent measurements (n Ն 3). Statistical analyses were made by Student's t test (2-tailed) for paired or unpaired data as appropriate; the null hypothesis was rejected when p Ͻ 0.05. Boltzmann curves were generated by best fit to complete data sets using the Origin 4.0 software package (Microcal Software Inc., Northampton, MA); P o (min) was constrained at 0, and where appropriate, P o (max) was constrained at 0.87 (mean value from experimental data).
Solutions and Drugs-For cell-attached recording, intact cells were bathed in high [K ϩ ] solution with the following composition: 140 mM KCl, 10 mM HEPES (pH 7.2 with NaOH), 1 mM MgCl 2 , 1 mM CaCl 2 , 30 mM glucose, such that E K ϭ 0 mV. Pipette-filling solution had the following composition: 135 mM NaCl, 5 mM KCl, 10 mM HEPES (pH 7.4 with NaOH), 1 mM MgCl 2 , 1 mM CaCl 2 , 0.001 mM tetrodotoxin. After seal formation (Ͼ30 gigaohms), cell-free patches were excised into a similar high [K ϩ ] solution containing the calcium buffers BAPTA or dibromo-BAPTA (1 mM). Free calcium concentration ([Ca 2ϩ ]) was adjusted in the ranges 10 Ϫ8 to 5 ϫ 10 Ϫ7 M (with BAPTA as the calcium buffer) and 10 Ϫ6 to 10 Ϫ5 M (with dibromo-BAPTA as the calcium buffer) by the addition of appropriate amounts of a stock solution of CaCl 2 (1 M), according to the relationship, Use of 1 mM BAPTA or dibromo-BAPTA allowed accurate buffering of [Ca 2ϩ ] at the intracellular face of the membrane and avoided the artifacts associated with the slow Mg 2ϩ -and pH-dependent buffering by EGTA (27). Okadaic acid was used to inhibit the activity of specific protein phosphatases (31). Consistent effects of okadaic acid were obtained by careful treatment of the inhibitor. Stock okadaic acid solutions were prepared in sterile, dry dimethyl sulfoxide (Me 2 SO) and stored in airtight containers at Ϫ20°C for up to 2 months. Stock solutions were diluted in bath solution to give a final Me 2 SO concentration of 0.1%; the activity of okadaic acid was preserved by limiting the thawing and refreezing of stock solutions and avoiding sonication of experimental solutions.

RESULTS
When GH cells are voltage-clamped through perforated patches in a physiological salt solution, BK Ca channels dominate the steady-state membrane conductance at depolarized voltages (12,32,33). In contrast, individual channels in cellattached patches on resting cells often showed surprisingly low activity: P o Ͻ 0.1 even at ϩ40 mV. However, when membrane patches were excised into an ATP-free bathing solution (Fig. 1), the mean open probability (P o ) increased slowly but significantly over 10 -15 min, reaching a new steady-state level that was 258 Ϯ 167% greater than the activity measured immediately following patch excision (n ϭ 11; p Ͻ 0.01). In many patches, an increase in the number of BK Ca channels open simultaneously (N) was also observed. The rate of increase in channel activity was not affected by varying calcium concentration the range 0.1 to 1.0 M or by varying the membrane potential of the patch from 0 to ϩ40 mV (not shown).
Dephosphorylated BK Ca Channels Respond to Voltage and Calcium in the Physiological Ranges-Once BK Ca channel activity had stabilized in MgATP-free solution, it remained stable for the duration of the recording (up to 90 min, Fig. 2A). Nevertheless, the mean open probability of the channels in the patch could be increased independently by further depolarization (V) or by increasing the free calcium concentration ([Ca 2ϩ ] i ) at the intracellular face of the membrane, up to a maximum P o ϭ 0.86 Ϯ 0.11 (n ϭ 10 complete P o V curves at 5 different [Ca 2ϩ ] in 8 patches) (Fig. 2B). These effects were reproducible and fully reversible and showed no sign of inactivation during prolonged stimulation. At calcium concentrations between 0.1 and 10.0 M, an e-fold increase in P o over the linear range from 0.1 to 0.8 was produced by increasing the voltage 14.6 Ϯ 6.6 mV (n ϭ 8 independent curves, in 6 patches). On average, increasing free calcium 10-fold at the cytoplasmic surface reduced the voltage required to produce half-maximal activity (V 1/2 ) by 55mV without changing the intrinsic voltage dependence of activity (Fig. 2C). Thus, when protein kinases are inhibited, native BK Ca channels are extremely sensitive to changes in voltage and calcium in the physiological ranges. The 1 The abbreviations used are: BAPTA, 1,2-bis(aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid; PKA and PKC, protein kinases A and C, respectively; PP1, protein phosphatase 1; PP2A, protein phosphatase 2A.
dissociation constant (K D ) for Ca 2ϩ binding was 1.8 M (calculated from the linear fit to the data in Fig. 2C (34)). Hence, although depolarization greater than any action potential (ϳϩ55 mV) would be required for half-maximal channel activity at the resting calcium level in the cell (ϳ0.1 M), halfmaximal channel activity would be sustained below the resting membrane potential of the cell (ϳϪ55 mV) as long as free Ca 2ϩ levels underneath the plasma membrane exceeded 10.0 M.
Protein Dephosphorylation Increases BK Ca Channel Activity-Previous studies show that protein dephosphorylation stimulates BK Ca channel activity in GH 4 C 1 cells (12,19); this behavior is similar to that of BK channels in other neuronal cell types (3, 9 -11). The increase in BK Ca channels activity we observed following patch excision into ATP-free solutions is consistent with this published data. Two independent pharmacological manipulations of the patches provide additional evidence that endogenous protein phosphatases are responsible for the increase in BK Ca channel activity. First, inhibition of serine/threonine-directed protein phosphatases with the structurally unrelated microbial toxins 1 M microcystin (n ϭ 3) or 1 nM okadaic acid (n ϭ 5) to selectively inhibit the PP2A family of phosphatases (35) completely prevented the increase in channel activity when they were added before patch excision (Fig. 1B). In contrast, if the inhibitors were not added until after channel activity had increased in ATP-free solution, they had no effect on channel activity (not shown), which rules out direct effects of the toxins on channel gating. Second, subsequent addition of MgATP at the intracellular face of the membrane rapidly reduced activity back to the low level recorded immediately following patch excision ( Fig. 3A; n ϭ 6). This Channels to Calcium-The activity of BK Ca channels in dephosphorylated membrane patches was reduced dramatically by subsequent protein phosphorylation (Fig. 3). The addition of 1 mM MgATP (but not ADP or GTP; see above) to the solution at the intracellular face of the excised patch caused a shift of 19 Ϯ 10 mV (n ϭ 5 patches; p Ͻ 0.05) toward more depolarizing potentials in the voltage dependence of BK Ca channel activity without changing the slope of the relationship (Fig. 3C). We interpret the reduction in activity as a decrease in channel sensitivity to calcium because the same effect was observed at two calcium concentrations (1 and 5 M) and because the intrinsic voltage-dependence of channel activity remained unchanged. The decrease in P o was achieved predominantly by a decrease in the frequency of channel opening with no significant change in the mean open time. Nevertheless, the maximum attainable channel activity was not suppressed by MgATP in excised patches; channels could be stimulated maximally by raising calcium or by depolarizing further (Fig. 3). In three patches, P o(max) was 0.76 Ϯ 0.07 following run up in the absence of ATP and 0.71 Ϯ 0.10 following the addition of MgATP (no significant difference; p Ͼ 0.1).
The inhibitory effect of MgATP on channel activity was not potentiated by subsequent addition of low concentrations (Յ10 nM) of okadaic acid or microcystin (not shown), indicating that the basal activity of PKA was substantially higher than that of the phosphatase. This conclusion is consistent with the low activity of BK Ca channels in cell-attached patches observed before excision and is supported by the observation that incubating cells in 1 nM okadaic acid and 1 mM cpt-cAMP for 10 min before patch excision also had no significant effect on channel activity in cell-attached patches (n ϭ 4). However, the open probability of the channels after patch excision from cells treated with okadaic acid could not be increased beyond P o ϭ 0.47 Ϯ 0.19 (n ϭ 10 complete P o V curves over a range of [Ca 2ϩ ] in 6 patches) at any calcium concentration (Յ50 M) or voltage (Յϩ100mV) we examined (Fig. 4). In all but one of the patches treated with okadaic acid before excision, maximal opening probability was Ͻ0.5. In contrast, in every single one of the dephosphorylated patches and those patches that were rephosphorylated after excision, maximal channel open probability could be driven to P o Ͼ 0.7. Hence, we interpret this suppressed channel activity as an effect on all of the channels rather than on a subset of the channel population. Our data indicate that in intact cells there is an additional kinase regulating the activity of BK Ca channels to produce a significantly different physiological effect on cell excitability. PKA-mediated phosphorylation reduces the sensitivity of BK Ca channels to increases in voltage and calcium, but phosphorylation by the second kinase reduces the maximum attainable channel activity by half. In other words, the inhibition by PKA is conditional, but the inhibition by the second kinase is unconditional. DISCUSSION The data reported here establish clearly that native calciumactivated potassium channels with the largest conductance (BK Ca or maxi-K) are inhibited by endogenous protein kinases and stimulated by endogenous protein phosphatases that remain closely associated with the channels in cell-free patches from an immortalized endocrine rat pituitary cell line (GH 4 C 1 ). Our results are entirely consistent with the physiological stimulation of excitability and secretion in the anterior pituitary by hypothalamic neuropeptides, which increase protein kinase activity through G s -and G q -coupled receptors (36,37). Such kinase-mediated inhibition demonstrates that BK Ca channels in GH 4 C 1 cells behave in the same way as the calcium-activated potassium currents in hippocampal neurons of the mammalian central nervous system (3, 8 -11). In contrast, most previous single-channel studies of BK Ca channel modulation have been conducted on channels that are stimulated by protein kinases and inhibited by protein phosphatases, using either native channels in nonneuronal tissues (5,21,22) or recombinant channels that are expressed heterologously in Xenopus oocytes, where the channels are much less sensitive to calcium and phosphorylation (38 -40). Nevertheless, singlechannel studies in neurons have also shown stimulation of BK Ca channel activity by protein kinases (20,23,40). Similar diversity in the response to protein kinases and phosphatases has been reported for BK Ca channels in cells from the vascular system (42)(43)(44). As yet there is only one slo gene encoding the pore-forming subunit of the BK Ca channel in any single species; hence, such diametrically opposed differences in modulation by phosphorylation must reflect alternative splicing of the gene or interaction with additional regulatory subunits or more indirect signaling cascades through which the protein kinases produce their effects.
We have used a number of criteria to identify the specific biochemical mechanism of the observed "run-up" of BK Ca channel activity in excised membrane patches, although we have not characterized as clearly all of the enzymes or their substrates that are responsible for modulating the availability of BK Ca channels for opening. This is the first study to investigate the effect of protein phosphorylation on BK Ca channel activity at maximal open probabilities. As a result, we have discovered that pharmacologically distinct protein kinases produce functionally distinct effects on channel behavior. One kinase, probably PKA, reduces stimulation of BK Ca channel activity by calcium and voltage without reducing the availability of the channels. For an enzyme, this would correspond to a reduction in K m without a change in V max . A second, unidentified kinase reduces the availability of the channels at all calcium concentrations and voltages, corresponding to a decrease in V max . The exact mechanisms by which the functional consequences of protein phosphorylation are achieved remain unclear. Both ␣ and ␤ subunits of the channel complex contain potential sites for phosphorylation by protein kinase A and protein kinase C (16). The C-terminal "tail" of the ␣ subunit has also been demonstrated to confer the calcium-sensing properties of BK Ca channels (38); thus, the addition or removal of charged phosphate groups could alter the calcium sensitivity of channel activity by influencing the "Ca 2ϩ sensor" in the protein. Alternatively, co-expression of the ␤ subunit has been shown to enhance the calcium sensitivity of BK Ca channels (13,39,45) so reversible phosphorylation could modulate channel behavior by promoting or inhibiting interactions between the ␣ and ␤ subunits in the heteromultimeric channel complex. Fig. 5 illustrates one hypothesis that integrates the results of this study with previously published results from intact GH 4 C 1 cells (12,18) and from BK Ca channels isolated from rat brain and reconstituted into lipid bilayers (41,46). When protein phosphorylation is prevented, the channels respond robustly to physiological increases in voltage and calcium. Phosphorylation at one of the modulatory sites inhibits channel activity in a conditional manner that can be overcome by increasing cytoplasmic calcium 3-fold or depolarizing the membrane by an additional 20 mV. Phosphorylation of this "conditional" site requires PKA activity and is protected by low concentrations of okadaic acid, which leads us to postulate that it is regulated by a type 2A-like protein phosphatase. In support of this proposal, it has been shown in bilayers that the inhibitory effect of PKA on brain BK Ca channels is reversed selectively by PP2A but not PP1 (41).
Our results also demonstrate a second pharmacologically and functionally distinct phosphorylation site that regulates channel availability. Phosphorylation at this site modulates BK Ca channel behavior in an unconditional manner by reduc- ing the maximum attainable open probability to P o Ͻ0.5, even with unphysiologically high calcium concentrations or depolarized membrane potentials. PKC is a powerful inhibitor of BK Ca channel activity in intact GH 4 C 1 cells (18) and in neurons (9). In our experiments, however, unconditional inhibition is lost when the membrane patch is excised from the cell, which implies either that this kinase is not functionally coupled to the channels in cell-free patches or that basal protein phosphatase activity is much greater than the activity of the kinase at this site. Preliminary experiments in which cell-free patches were treated with MgATP and a higher concentration (100 nM) of okadaic acid indicate that channel availability is suppressed under these conditions, 2 which implies a direct role for another phosphatase less sensitive to okadaic acid than PP2A. Protein phosphatase 1 (PP1) is much less sensitive to okadaic acid than PP2A (31), and PP1 selectively reverses the effects of PKC on BK Ca channels in other systems (23,46).
In summary, we have identified two functionally distinct effects of endogenous protein kinases and phosphatases on the activity of calcium-and voltage-activated potassium channels. Phosphorylation at a putative PKA/PP2A site reduces the calcium sensitivity of BK Ca channels without altering the voltage dependence of activation or suppressing maximum channel availability. In contrast, phosphorylation at a putative PKC/ PP1 site reduces the availability of the channels at all calcium concentrations and voltages in the physiological range. Such conditional and unconditional regulation by two pharmacologically distinct signaling pathways provides the nervous and neuroendocrine systems with a powerful computational process at the cellular level. Activation of G s -coupled receptors would inhibit BK Ca channel activity and, hence, stimulate cell excitability in a conditional manner, an effect that could be reversed when the intracellular calcium concentration increased or the membrane depolarized further. In contrast, activation of G qcoupled receptors would inhibit BK Ca channel activity and stimulate excitability unconditionally until the effects of PKC were reversed by protein phosphatase activity.