A Molecular Switch for Specific Stimulation of the BKCa Channel by cGMP and cAMP Kinase*

The cGMP and the cAMP pathways control smooth muscle tone by regulation of BKCa (BK) channel activity. BK channels show considerable diversity and plasticity in their regulation by cyclic nucleotide-dependent protein kinases. The underlying molecular mechanisms are unclear but may involve expression of splice variants of the BK channel α subunit. Three isoforms, BKA, BKB, and BKC, which were cloned from tracheal smooth muscle, differed only in their C terminus. When expressed in HEK293 cells, cGMP kinase (cGK) but not cAMP kinase (cAK) stimulated the activity of BKA and BKB by shifting the voltage dependence of the channel to more negative potentials. In contrast, BKC was exclusively stimulated by cAK. BKC lacks a C-terminal tandem phosphorylation motif for protein kinase C (PKC) with Ser1151 and Ser1154. Mutation of this motif in BKA switched channel regulation from cGK to cAK. Furthermore, inhibition of PKC in excised patches from cells expressing BKA abolished the stimulatory effect of cGK but allowed channel stimulation by cAK. cAK and cGK phosphorylated the channel at different sites. Thus, phosphorylation/dephosphorylation by PKC determines whether the BK channel is stimulated by cGK or cAK. The molecular mechanisms may be relevant for smooth muscle relaxation by cAMP and cGMP.

The cGMP and the cAMP pathways are major regulators of smooth muscle contractility (1). It is widely accepted that activation of large conductance Ca 2ϩ -activated K ϩ (BK) channels via cGMP kinase (cGK) 1 (2,3) and cAMP kinase (cAK) (4,5) contributes to relaxation (6). However, the responses of BK channels to hormones and drugs that activate either the cAMP or the cGMP-signaling pathway vary from tissue to tissue (7)(8)(9)(10) and even within the same tissue (11,12). Alternatively, spliced mRNA transcripts and reversible protein phosphorylation of the BK channel ␣ subunit or closely associated proteins have been shown as major mechanisms for the diversity of channel regulation by protein kinases (13). Particularly, the large C-terminal domain of the BK channel ␣ subunit has been suggested as the target for phosphorylation by several serine/ threonine-directed kinases (14). The molecular requirements at the BK channel level, however, that are necessary for modulation by specific cyclic nucleotide-dependent protein kinases have not yet been resolved. Recently, phosphorylation sites for cAK (15) and cGK (16,17) have been identified that may be important for specific BK channel activation. Since these sites are fully conserved in all mammalian alternative splice variants known so far, regulation of BK channels via both the cGK and cAK would be expected. However, in most reports on cyclic nucleotide-dependent channel stimulation, effects from either cGK (3, 16 -19) or cAK (13,15,20,21) have been described. In particular, it is not clear whether cGK and cAK can substitute for each other as channel regulators, whether they have additive effects, or whether specific requirements are needed that allow distinct regulation by either kinase. The elucidation of molecular mechanisms underlying BK channel regulation is even more complicated by the fact that BK channels apart from cGK and cAK are presumably modulated by additional protein kinases such as PKC (22) and Src (21). These protein kinases may interact with each other in modifying channel activity. Thus, BK channels seem to integrate the output from several signaling cascades.
With respect to channel regulation by protein kinases, it has been an important issue in recent years to identify the underlying molecular mechanisms (13, 15-19, 21, 23). In this study, we first analyzed whether distinct mRNA transcripts in tracheal smooth muscle represent the molecular basis for differential regulation of BK channel by cAK and cGK. We identified three C-terminal isoforms of the BK channel ␣ subunit that are presumably generated by alternative splicing at cryptic intron sites within a C-terminal exon. All of the isoforms showed similar single channel conductance and Ca 2ϩ sensitivity. The presence of the C-terminal tandem phosphorylation site for PKC in two isoforms allowed stimulation of channel activity by cGK, suggesting the involvement of PKC in cGK-dependent channel regulation. However, the absence of this motif in another splice variant resulted in a loss of stimulation via cGK but led to activation by cAK. The hypothesis of phosphorylation by PKC as a prerequisite for the stimulatory effect of cGK was proved by mutation of respective serines and the use of specific PKC inhibitors. The results suggest that constitutive phosphorylation by PKC at the C terminus allows conditional activation by cGK. In turn, stimulation by cAK occurs at the dephosphorylated channel and at splice variants lacking the C-terminal PKC motif. The mechanisms identified here may contribute to the relaxation induced by NO and ␤ 2 -adrenoreceptor agonists in smooth muscle.

MATERIALS AND METHODS
Cloning-BK channel ␣ subunits were cloned from a bovine tracheal smooth muscle oligo dT-primed cDNA library. About 800,000 clones of the library were screened with the EcoRV/HaeII fragment corresponding to the region between the S0 and S6 segment. The probe (nucleotides 400 -989) was isolated from the human BK channel ␣ subunit cDNA, kindly provided by Dr. L. Toro (GenBank TM U11058) (24). Isolated full-length clones were sequenced and constructed into the expression vector pcDNA3 (Invitrogen), which directed expression of BK channel ␣ subunits in HEK293 cells starting translation from the M1 methionine (24). The GenBank TM accession numbers for the BK channel isoforms are AY033472 (BK A ), AY033473 (BK B ), AY033474 (BK C ).
Site-directed Mutagenesis-Mutants were generated by extended overlap polymerase chain reaction using the primer pairs as indicated. The resulting amplicons were digested with the given restriction enzymes and cloned into pcDNA3 containing the respective isoform of the BK channel. All mutations were confirmed by sequence analysis. Primer sequences are as follows (mutated bases in bold letters): BK A S1151A, M1/SP6, F8/M2, template BK A , ApaI; BK A S1154A, M3/SP6, F8/M4, template BK A, ApaI; BK A S1151A/S1154A, M5/SP6, F8/M6, template BK A, ApaI; BK A S1134A, M7/SP6, F8/M8, template BK A, ApaI; BK C S1134A, M7/SP6, F8/M8, template BK C, ApaI; BK A S1134A/ S1151A/S1154A, M7/SP6, F8/M8, template BK A S1151A/S1154A, ApaI; BK A S922A/S1151A/S1154A, M9/R12, F6/M10, template BK A S1151A/ S1154A, AflII and SacII. Functional Expression in HEK293 Cells and Electrophysiological Measurements-HEK293 cells were transfected with the channel cDNA plasmids by calcium phosphate precipitation for 18 h at 35°C and 3% CO 2 . The transfection efficiency varied between 40 and 70%. Macroscopic currents were recorded 48 h after transfection from inside-out patches excised from HEK293 cells by using a List EPC 7 amplifier connected via a 16-bit analog-to-digital interface to a Pentium IBM clone computer (25). Signals were filtered at 1 kHz by a 10-pole Bessel filter and sampled at 3 kHz. Data acquisition and analysis was performed with an ISO-3 patch-clamp program (MFK, Niedernhausen, Germany). The patch pipettes were pulled from borosilicate glass (fiberfilled) with a resistance of 3-4 megaohms and were filled with physiological saline solution containing 127 mM NaCl, 5.9 mM KCl, 2.4 mM CaCl 2 , 1.2 mM MgCl 2 , 11 mM glucose, and 10 mM HEPES adjusted to pH 7.4 with NaOH. The bath solution (cytosolic surface of the patch) contained 134 mM KCl, 6 mM NaCl, 1.2 mM MgCl 2 , 5 mM EGTA, 11 mM glucose, 3 mM dipotassium ATP, and 10 mM HEPES (pH 7.4). Depending on the experiment, the free Ca 2ϩ concentration of the bath solution was changed from 0.3-10 M by changing the Ca 2ϩ concentration in the corresponding solution. The appropriate amounts of CaCl 2 were added, and the pH was adjusted according to a computer program on the basis of the binding constants of Fabiato and checked by fura 2 fluorescence (for details see Ref. 19). Experiments with iberiotoxin (Latoxan, Rosans, France) were carried out on outside-out patches, and bath and pipette solution were opposite to those used in inside-out experiments. A multibarreled perfusion pipette placed 200 m away from the patch was used to switch the superfusion solution. Electrophysiological experiments were performed at 20 -22°C, the holding potential was Ϫ20 mV.
Conductance (G) values were calculated as G ϭ I/V m . Usually, each experiment was normalized by dividing G values by G max , where G max was defined as the largest G value obtained in each experiment. G/G max voltage curves were fitted with a Boltzmann equation of the form G/G max ϭ (1 ϩ exp((V 1/2 Ϫ V m )/k]) Ϫ1 with the use of Origin (version 6; Microcal Software, Northampton, MA). The data are expressed as means Ϯ S.E.
Protein Kinases and PKC Inhibitors-Pure cGK I␣ (26) or the catalytic subunit of cAMP kinase (19) were prepared as described. The cGK and cAK were applied to the membrane patches at a concentration of 300 nM. The cGK was consistently used in the presence of 10 M cGMP. The PKC inhibitor peptide PKC 19 -31 was from Calbiochem and was applied at a concentration of 5 M. The non-peptide PKC inhibitor Ro31-8220 was obtained from Biomol (Hamburg, Germany) and was used at a concentration of 1 M.

BK Channel Isoforms Cloned from Tracheal Smooth Muscle-
The BK channels in native tracheal smooth muscle cells show diversity in modulation by cGK and cAK (4,19,27). To analyze whether additional yet unidentified mRNA transcripts code for BK channels that might be specifically regulated by either cGK or cAK, we cloned BK channel ␣ subunits from smooth muscle tissue of bovine trachea. In total, nine fulllength clones containing the M1 methionine (24) and poly(A) ϩ tails of different length were isolated. In general, the bovine BK channel primary structure is identical with that from human (24,28) with the exception of repetitive amino acid sequences at the N terminus varying in length (Fig. 1A) and amino acids at positions Thr 139 , Ser 710 , Ser 711 of the bovine sequence that are changed to Ala, Asn, Thr in the human sequence, respectively. The nine clones obtained by cloning from tracheal smooth muscle consisted of three populations differing only at their C terminus ( Fig. 1). Four clones of 6.1-6.6 kb were grouped into class BK A . The BK A isoform has been identified in a variety of species including human (24,28). One clone of 4.4 kb differed from the BK A clones in amino acids 1160 -1166 at the C terminus and was classified as BK B . This clone contained a 3Ј-untranslated region of 0.6 kb that was different from the 3Ј-untranslated region of 3 kb from BK A . The BK B -specific C terminus was recently also identified in rat (GenBank TM U40603) and chicken (29). Another four clones (BK C ) of 3.3 kb were lacking the last 26 amino acids at the C terminus of BK A . The C-terminal sequence 1137 HSIP 1140 , which is common to all isoforms, is followed by the sequence 1141 PQT 1143 in BK C before the stop codon and the poly(A) ϩ tail. BK C does not contain a 3Ј-untranslated region, and the poly(A) tail directly follows the stop codon. Since the analysis of the genomic structure of the BK channel in rat (30) revealed a single exon encoding the C-terminal part of the BK A isoform (amino acids 1085-1166), the BK B and BK C isoforms may be generated by alternative splicing at cryptic intron splice sites within this exon. In fact, inspection of the nucleotide sequence at the junction sites reveals the presence of consensus motifs of intron 5Ј donor splice sites (31) in BK A , namely GT at the junction of BK A to BK B (site 2 in Fig. 1B) and AT at the junction of BK A to BK C (site 1 in Fig.  1B). The consensus motifs GT and AT are conserved at the respective splice site in mammalian BK channel nucleotide sequences (Fig. 1B). Apart from BK B , additional C-terminal isoforms have been identified from others that probably result from splicing at site 2 including Rat1BK, Rat2BK, and MouseBK (for details see the legend of Fig. 1B). Thus, the GT and AT consensus motifs at sites 1 and 2 may serve as cryptic 5Ј donor splice sites within the C-terminal exon.
Basic Properties of BK Ca Channels Expressed in HEK Cells-Membrane currents of nontransfected native HEK293 cells were measured in inside-out patches. Depolarizing steps from a holding potential of Ϫ20 mV to positive potentials resulted in a small outward current that showed no inactivation and exhibited no Ca 2ϩ sensitivity. The steady-state current at ϩ80 mV was 27.1 Ϯ 2.5 pA (n ϭ 30). The amplitude of outward currents in HEK293 cells expressing BK A was considerably larger than in native HEK cells. Mean current amplitude obtained under steady-state conditions at ϩ80 mV in cells expressing BK A was 1851.3 Ϯ 176.0 pA (n ϭ 34). Typical examples of outward currents obtained from a native and a transfected cell are shown in Fig. 2, A and B. The results demonstrate that the outward current of transfected cells was contaminated by an endogenous current component by Ͻ2%. In some inside-out patches with low channel density, single channel conductance was determined in the presence of symmetrically high potassium of 140 mM. Channel openings could be detected at membrane potentials from Ϫ60 mV to ϩ80 mV ( Fig. 2C shows an example from a cell transfected with BK A ). The current-voltage relations of the three C-terminal splice variants were linear in the voltage range between Ϫ60 and ϩ60 mV and showed reversal potentials close to 0 mV (Fig. 2D). The mean unitary conductances were 276 Ϯ 8.1 picosiemens (BK A ; n ϭ 6), 260 Ϯ 12.3 picosiemens (BK B ; n ϭ 6), and 272 Ϯ 16.3 picosiemens (BK C ; n ϭ 7). To further determine the basic channel properties of the BK Ca channel isoforms, Ca 2ϩ dependence was investi-gated in inside-out patches. When the Ca 2ϩ concentration was increased from 0.3 to 10 M, a shift to the left of the normalized conductance-voltage relations by Ϸ100 mV was observed (Fig.  3A). In eight patches from clones BK A and BK B and in 10 patches from clone BK C , the mean half-maximal activating voltage (V 1/2 ) was 100 Ϯ 5, 95 Ϯ 5, and 113 Ϯ 9 mV in 0. The amino acid sequence of BK B and BK C different from that in BK A are underlined and given in bold letters. Amino acid numbers are given at the right margin. The phosphorylation site for cGK, Ser 1134 , (16,17), and the C-terminal consensus sites Ser 1151 /Ser 1154 for phosphorylation by PKC are indicated in bold italics. B, alignment of C-terminal nucleotide sequences of BK A , BK B , BK C (BovBK) with BK channel isoforms from different species, human BK (GenBank TM U11058), rat1 BK (GenBank TM AF135265), mouse BK (GenBank TM AF156674), and rat2 BK (GenBank TM U40603). The indicated nucleotide sequence of bovine BK A and human BK is part of a single exon encoding the C-terminal amino acids 1085-1166 (numbering of BK A ) (30). Splicing within this exon at the putative cryptic intron site 1 starting with the consensus nucleotides GT or AT (underlined) may result in the BK C -specific C-terminal nucleotide sequence (indicated in bold letters). Splicing at the putative cryptic intron site 2 starting with the consensus site GT (underlined) may result in the BK B -specific C-terminal nucleotide sequence (indicated in bold letters). The BK B -specific C-terminal sequence is also found in the rat2 BK. Further isoforms with specific C termini (indicated in bold letters) such as the indicated rat1 BK and mouse BK are probably also generated by splicing at site 2. Triplets represent the reading frame. Numbers of nucleotides are given at the left and right margin of the alignments.  (Fig. 3B). The difference between the V 1/2 values was not statistically significant (p Ͼ 0.1). To determine the sensitivity of the BK channel isoforms to the specific peptide blocker iberiotoxin, outside-out patches in the presence of 1 M [Ca 2ϩ ] i were exposed to increasing concentrations of the toxin (Fig. 3C). Mean half-maximal inhibitory concentrations of iberiotoxin were 5.8 Ϯ 0.7 nM (BK A ; n ϭ 5), 5.7 Ϯ 0.7 nM (BK B ; n ϭ 7), and 8.3 Ϯ 2.0 nM (BK C ; n ϭ 7).

Regulation of C-terminal Splice Variants by cGK and cAK-
The activity of BK channels is modulated in native cells by cAK and cGK. To investigate whether the C-terminal splice variants expressed in HEK293 cells are also regulated by the kinases, either the catalytic subunit of cAK or cGK plus cGMP were applied to the cytosolic face of excised inside-out membrane patches (Fig. 4). In the presence of cGK (upper panels), a 26and 31-mV shift to the left of the normalized conductancevoltage relation was observed in splice variants BK A (V 1/2 before, 57.9 Ϯ 5.9 mV; V 1/2 after, cGK 32.2 Ϯ 3.5 mV; n ϭ 7) and BK B (V 1/2 before, 54.8 Ϯ 6.1 mV; V 1/2 after, cGK 24.1 Ϯ 2.8 mV; n ϭ 8). Membrane conductance of splice variant BK C , however, did not respond significantly to cGK, V 1/2 was 54.5 Ϯ 5.8 mV before and 47.5 Ϯ 5.1 mV after application of the kinase. Opposite to cGK, cAK (lower panels) had no significant influence on BK A (V 1/2 before, 46.2 Ϯ 5.3 mV; V 1/2 after, cAK 43.7 Ϯ 4.8 mV; n ϭ 8) and BK B (V 1/2 before, 51.0 Ϯ 5.6 mV; V 1/2 after, cAK 49.4 Ϯ 5.1 mV; n ϭ 8) but induced a 35-mV shift to the left of the normalized conductance-voltage relation obtained from BK C (V 1/2 before, 62.3 Ϯ 6.9 mV; V 1/2 after, cAK 27.8 Ϯ 3.1 mV; n ϭ 6). The results of Fig. 4 suggest that the C-terminal amino acids determine whether the channel is responsive to cGK or cAK.
Effects of cGK and cAK after Mutation of the Putative PKC Sites Ser 1151 and Ser 1154 -Inspection of the C terminus revealed the tandem motif RXKS 1151 RXS 1154 RXK (32) for phosphorylation by PKC, which was present in BK A and BK B but not in isoform BK C (Fig. 1A). To study the possible influence of this site on the channel regulation by cGK and cAK, a mutant BK A was created in which Ser 1151 and Ser 1154 were replaced by alanine (S1151A/S1154A). Inside-out patches excised from HEK293 cells expressing the double mutant construct proved insensitive to stimulation by cGK but became responsive to cAK, similar as the BK C isoform with its truncated C terminus (Fig. 5A). V 1/2 obtained from conductance-voltage relations was 53.6 Ϯ 6.2 mV before (control) and 49.7 Ϯ 5.4 mV after the application of cGK (n ϭ 8; not significant). In contrast, cAK shifted V 1/2 by 31 mV from 40.6 Ϯ 4.3 to 9.7 Ϯ 1.3 mV (n ϭ 12). These results indicate that the intact putative tandem PKC motif is a prerequisite for the channel to become regulated by cGK. To evaluate whether cAK stimulated the activity of the mutant BK A isoform by phosphorylation of Ser 1134 , the site phosphorylated by cGK (16,17), this serine was changed to alanine. As expected, the single mutation S1134A in BK A resulted in loss of stimulation of channel activity by cGK (Fig.  5B). The inability of cAK to activate wild-type BK A channels was unaffected by mutation S1134A (Fig. 5B), demonstrating that the mutation of the tandem PKC site in BK A was responsible for the specific activation by cAK. Mean V 1/2 values in Fig.  5B were 58.9 Ϯ 6.1 mV (control) and 60 Ϯ 6.3 mV (cGK; n ϭ 6) and 48.9 Ϯ 5.3 mV (control) and 51.0 Ϯ 5.5 mV (cAK: n ϭ 8). When the double mutation in BK A S1151A/S1154A was combined with the mutation S1134A, a channel was created that

FIG. 4. Regulation of BK channel isoforms expressed in HEK293 cells by cGK and cAK.
Normalized conductance-voltage relations from inside-out patches in the absence and presence of cGK (upper panels) and cAK (lower panels). cGK shifts the curve to the left in patches expressing BK A (n ϭ 7) and BK B (n ϭ 8) but not in those expressing BK C (n ϭ 8). Cyclic AK has no effect on patches expressing BK A (n ϭ 8) and BK B (n ϭ 8) but induced a left-shift of the curve obtained with BK C (n ϭ 6). Each patch represents a different cell. Currents were evoked by applying a 200-ms depolarizing pulse every 5 s in 20-mV increments from a holding potential of Ϫ20 mV to Ϯ 120 mV. The data were fitted by a Boltzmann equation. The insets show representative BK current recordings elicited by 200 ms pulses from Ϫ20 to ϩ40 mV. Closed symbols represent currents in the presence of either cGK or cAK; [Ca 2ϩ ] in the bath solution was 1 M. Cells from the same transfection were used for testing BK channel conductance in the absence and presence of the respective kinase as V 1/2 values slightly varied between different rounds of transfection.
failed to respond to cGK (not shown) but was activated by cAK, indicating that cAK phosphorylates a different site of the BK channel than cGK (Fig. 5C). Cyclic AK produced a shift to the left by 30 mV; the mean V 1/2 values were 55.8 Ϯ 5.9 mV before and 25.4 Ϯ 3.2 mV (n ϭ 10) in the presence of cAK. Similarly, mutation of the cGK site S1134A in splice variant BK C had no influence on the stimulatory action of cAK; V 1/2 was shifted to the left by 34 mV from 45.4 Ϯ 5.2 mV (control) to 1.7 Ϯ 2.2 mV in the presence of cAK (n ϭ 6; data not shown). Since phosphorylation of Ser 869 of the human BK channel has been suggested to account for the stimulatory effect of cAK (15); the corresponding Ser 922 of the bovine BK A channel was mutated to alanine in addition to the S1151A/S1154A exchange. The cAKmediated increase in BK channel activity was completely eliminated in patches from HEK cells expressing this S922A mutant (Fig. 5C). The mean V 1/2 values were 52.0 Ϯ 5.4 mV before and 51.0 Ϯ 5.5 mV (n ϭ 13) in the presence of cAK. This finding confirms that cAK and cGK stimulate BK channel activity by phosphorylating different sites of the BK channel ␣ subunit, i.e. Ser 922 and Ser 1134 , respectively.
In summary, these results suggest that an intrinsic PKC activity of HEK293 cells phosphorylates the BK A and BK B isoforms and consequently allows channel stimulation by cGK. The absence of the PKC consensus motif in the BK C isoform leads to exclusive channel stimulation by cAK. Further in depth analysis revealed that phosphorylation of both serines were involved in the activation switch, because mutation of either Ser 1151 or Ser 1154 alone led to an intermediate phenotype with respect to channel activation by cGK and cAK (data not shown). Inhibition of PKC Switches BK Channel Activation from cGK to cAK-To provide additional evidence for the involvement of PKC in the regulation of the BK channel, inside-out patches expressing the BK A isoform were superfused with the inhibitory peptide of PKC (PKC 19 -31 ) or the PKC inhibitor Ro31-8220. To allow complete dephosphorylation of the PKC sites, inside-out patches were first superfused for 5 min with ATPfree solution to which either PKC 19 -31 (Fig. 6A) or Ro31-8220 (Fig. 6B) had been added. This solution was then changed to a solution containing ATP, PKC 19 -31 , or Ro31-8220 and either cGK or cAK. Similarly to the results obtained with the PKC mutation S1151A/S1154A, PKC 19 -31 , and Ro31-8220 prevented the stimulatory action of cGK on BK A and switched the responsiveness of the channel toward cAK. V 1/2 values obtained from conductance-voltage relations in the presence of PKC 19 -31 (Fig. 6A) were not significantly changed by cGK (before, 55.0 Ϯ 5.9 mV; after cGK, 49.3 Ϯ 5.7 mV; n ϭ 7), whereas a shift to the left by 28 mV was produced by cAK (before 48.8 Ϯ 5.3 mV; after cAK 20.3 Ϯ 2.5 mV; n ϭ 7). Similar results were obtained with Ro31-8220 (Fig. 6B); the mean V 1/2 values before (57.4 Ϯ 6.3 mV) and after cGK (52.5 Ϯ 6.1 mV; n ϭ 8) were not signifi- FIG. 5. Activation of the BK A isoform by cGK requires the putative PKC phosphorylation sites Ser 1151 and Ser 1154 . Normalized conductance-voltage relations from inside-out patches excised from HEK293 cells expressing BK A . A, mutation of the tandem PKC site, S1151A/S1154A, switches channel activation from cGK (n ϭ 8) to cAK (n ϭ 12). B, mutation of the putative cGK site, S1134A, abolishes the cGK effect (n ϭ 6), whereas cAK remains ineffective (n ϭ 8). C, left, combined mutations S1151A/S1154A and S1134A do not affect the shift to the left of the curve by cAK (n ϭ 10). C, right, combined mutations S1151A/S1154A and S922A prevent channel stimulation by cAK. A-C, each patch was from a different cell. Currents were evoked by applying a 200-ms depolarizing pulse every 5 s in 20-mV increments from a holding potential of Ϫ20 mV to Ϯ 120 mV. The insets show representative BK current recordings elicited by 200-ms pulses from Ϫ20 to ϩ40 mV. Closed symbols represent currents in the presence of cGK or cAK; [Ca 2ϩ ] in the bath solution was 1 M. ctr, control. cantly different, whereas cAK produced a shift to the left by 31 mV (before 54.5 Ϯ 5.7 mV; after cAK 23.0 Ϯ 2.9 mV; n ϭ 8). The results demonstrate that inhibition of PKC, apparently associated with the BK channel in the excised patch, switches stimulation of channel activity from cGK to cAK. DISCUSSION The Ca 2ϩ -activated K ϩ channel of the BK type is subject to regulation by both the cAMP and the cGMP signaling pathway. Its contribution to smooth muscle relaxation has recently been demonstrated by gene deletion of the Ca 2ϩ sensitivity-enhancing ␤ 1 subunit that resulted in hypertension of gene-targeted mice (33,34). Moreover, blockade of BK channels can interfere with the relaxation-promoting effect from NO/cGMP (35,36). Stimulation of BK channels by cyclic nucleotide-dependent protein kinases was attributed to the phosphorylation of the ␣ subunit at sites specific for either cAK (15,23) or cGK (16,17). Although these reports suggested the existence of specific phosphorylation sites that are conserved in mammalian BK channels, they do not explain the diversity of channel regulation by cAK (4,5,37,38) and cGK (2,18,19) found in native cells. Further requirements are apparently needed to allow specific channel stimulation by either of these kinases. The objective of this study was therefore to address the question of whether alternative exon splicing of the ␣ subunit and/or additional determinants are responsible for specific enhancement of channel activity by cAK and cGK. We identified three full-length splice variants in tracheal smooth muscle differing only at their C terminus. In contrast to BK channel isoforms identified recently (39), the single channel conductance, Ca 2ϩ , voltage, and iberiotoxin sensitivity of the isoforms BK A , BK B , and BK C , were nearly identical (Figs. 2 and 3). The unexpected finding that BK A and BK B were exclusively stimulated by cGK, whereas BK C was stimulated only by cAK, prompted us to search for the determining factor. We identified a tandem PKC motif close to the C terminus that is present in BK A and BK B but not in BK C . The residues Ser 1151 and Ser 1154 of this motif have an optimal consensus sequence for PKC-dependent phosphorylation (32). Our results strongly suggest that these sites are an effective PKC substrate for the following reasons. (i) PKC inhibition by two different inhibitors switched the specificity of channel stimulation from cGK toward cAK (Fig. 6). Since PKC was inhibited in the inside-out configuration, we suppose that the PKC remains functionally associated with the channel in the detached patch, similar to that suggested re-cently (40). (ii) Cyclic GK fails to stimulate channel activity when PKC-dependent phosphorylation of the C terminus is prevented due to mutation of both Ser 1151 and Ser 1154 (Fig. 5). Mutation of either Ser 1151 or Ser 1154 alone yielded only an intermediate phenotype with respect to channel stimulation by cAK and cGK.
The channel activity in inside-out patches from HEK293 cells expressing BK A or BK B is per se stimulated by cGK (Fig.  4). This suggests that phosphorylation of the tandem motif by PKC occurs constitutively. A constitutive or unconditional phosphorylation possibly mediated by PKC has recently been suggested for BK channels in GH 4 C 1 cells (41). Due to the phosphorylation at the tandem motif, the switch of channel regulation from cGK to cAK in the presence of PKC inhibitors may involve the activity of a phosphatase that dephosphorylates the channel at the PKC motif before the channel becomes responsive to cAK (Fig. 6). Regulation of BK channels by a closely associated phosphatase has been shown in neuronal (40) and smooth muscle cells (19,37). The present study also clearly demonstrates that cGK and cAK phosphorylate different sites of the channel. In agreement with Alioua et al. (16) and Fukao et al. (17), Ser 1134 was identified as target for cGK-dependent phosphorylation. In contrast to cGK, cAK does not phosphorylate Ser 1134 . Cyclic AK enhances channel activity by phosphorylating Ser 922 , a result that confirms recent findings (13,15). The phosphorylation of an additional cAK site introduced by a facultative STREX exon was shown to inhibit BK channel activity (13). Hence, the lack of cAK to stimulate BK A and BK B channels that do not contain the STREX exon could be due to an inhibitory effect mediated by phosphorylation of the tandem PKC site.
Cyclic AK was demonstrated to bind to the large C-terminal part of the BK channel from Drosophila (21) and to coimmunoprecipitate with a mouse brain BK channel expressed in HEK293 cells (13). We speculated, therefore, whether phosphorylation of the tandem PKC site is necessary for binding of cGK to the channel. Since BK C with its truncated C terminus is unresponsive to cGK, we tested a phosphopeptide comprising the last 26 amino acids of the BK A channel including the phosphorylated Ser 1151 and Ser 1154 for competition with cGK. This phosphopeptide was not able to attenuate the stimulatory effect of cGK, 2 7. Scheme illustrating the role of BK channel isoforms in cAMP-and cGMP-mediated smooth muscle relaxation. The observation that PKC inhibits BK channels in smooth muscle (44) suggests that phosphorylation of the BK channel by PKC supports the contractile response of hormones coupling to the diacylglycerol (AC)/PKC and inositol 1,4,5trisphosphate (IP 3 ) pathway. In turn, phosphorylation of the tandem motif by PKC promotes the counter-regulatory effect of the NO/cGMP/cGK cascade, since phosphorylation by PKC is a prerequisite for cGK to become an effective activator of BK A and BK B channel isoforms. Dephosphorylation of the tandem PKC motif mediated by a phosphatase or expression of isoforms like BK C lacking the C-terminal tandem PKC site allow BK channel activation by cAK. acts with the phosphorylated C-terminal portion of the channel. Contrary to our findings with the BK A and BK B channel from bovine tracheal smooth muscle, a specific BK channel isoform from mouse brain is regulated by cAK when expressed in HEK293 cells (13). Comparison of the respective primary structures reveals that this BK channel from mouse brain (14) has an extended C terminus starting downstream of the PKC tandem motif, suggesting that an altered C terminus may affect phosphorylation by PKC. Preferential activation of BK channels by cAK may be achieved by specific splice variants like the BK C isoform, lacking the C-terminal tail including the tandem motif for PKC (Fig. 7). Previous studies confirmed the presence of mRNA transcripts in the range of 4 kb in smooth muscle tissue (28). These mRNA species may represent BK channels similar to the BK B and BK C isoform cloned from tracheal smooth muscle in this study. Intriguingly, the BK B and BK C isoforms result from splicing at cryptic intron sites within the C-terminal exon. This processing of the pre-mRNA might occur in a tissue-specific manner and further adds complexity to the functional role of splice variants.
In smooth muscle, the BK channel is a multimeric macrocomplex consisting of four ␣ and four ␤ 1 subunits (42). The co-expression of the ␤ 1 subunit may affect the molecular mechanisms of channel regulation identified here. However, a previous report studying channel regulation by cAK clearly demonstrated that enhancement of channel activity was not altered when the ␤ subunit was co-expressed (15). In addition to auxiliary subunits, the interaction of protein kinases with BK channels may be influenced by the host cell. When choosing Chinese hamster ovary cells as hosts, both cAK and cGK failed to stimulate BK channels (43). In the latter report, the transfected human BK channel starts translation from methionine M 3 (24). The difference in the present study cannot be explained by the start methionine, since the human BK channel with M 3 as translational start was similarly activated by cGK when expressed in HEK293 cells as the bovine BK A isoform with M 1 as the translational start. 2 This finding suggests that a specific membrane environment and/or putative scaffold and signaling proteins of host cells are needed for the functional interaction of kinases and channel isoforms. In conclusion, the results of this study establish a complex link between PKC and the ability of cGK and cAK to regulate BK channels that contain typical C termini like BK A and BK B . We show for the first time that phosphorylation by PKC controls the activation of the channel by cGK. This demonstration is a further step in our understanding of the diversity and complexity of BK channel regulation. The findings may be particularly significant for mechanisms of smooth muscle relaxation and may explain specific channel regulation in native cells by either cAMP or cGMP pathways.