A Catalytically Inactive Mutant of Type I cGMP-dependent Protein Kinase Prevents Enhancement of Large Conductance, Calcium-sensitive K+ Channels by Sodium Nitroprusside and cGMP*

The activation of large conductance, calcium-sensitive K+ (BKCa) channels by the nitric oxide (NO)/cyclic GMP (cGMP) signaling pathway appears to be an important cellular mechanism contributing to the relaxation of smooth muscle. In HEK 293 cells transiently transfected with BKCa channels, we observed that the NO donor sodium nitroprusside and the membrane-permeable analog of cGMP, dibutyryl cGMP, were both able to enhance BKCa channel activity 4–5-fold in cell-attached membrane patches. This enhancement correlated with an endogenous cGMP-dependent protein kinase activity and the presence of the α isoform of type I cGMP-dependent protein kinase (cGKI). We observed that co-transfection of cells with BKCa channels and a catalytically inactive (“dead”) mutant of human cGKIα prevented enhancement of BKCa channel in response to either sodium nitroprusside or dibutyryl cGMP in a dominant negative fashion. In contrast, expression of wild-type cGKIα supported enhancement of channel activity by these two agents. Importantly, both endogenous and expressed forms of cGKIα were found to associate with BKCa channel protein, as demonstrated by a reciprocal co-immunoprecipitation strategy. In vitro, cGKIα was able to directly phosphorylate immunoprecipitated BKCa channels, suggesting that cGKIα-dependent phosphorylation of BKCa channels in situ may be responsible for the observed enhancement of channel activity. In summary, our data demonstrate that cGKIα alone is sufficient to promote the enhancement of BKCa channels in situ after activation of the NO/cGMP signaling pathway.

The elevation of intracellular cGMP 1 in response to endothelium-derived nitric oxide (NO) or clinically prescribed nitrovasodilators, such as nitroglycerin and sodium nitroprusside, is known to play an important role in the hypotensive actions of these agents (1,2). Similarly, elevation of cGMP by the phosphodiesterase inhibitor sildenafil (Viagra) (3) appears to underlie the smooth muscle-relaxing and anti-impotence effects of this drug. Although the exact mechanism(s) by which elevated cGMP causes smooth muscle relaxation has not been clearly defined, cGMP is known to influence a number of cellular processes (4), such as the levels of cytosolic free calcium, myosin light chain dephosphorylation (5), and the activity of voltagedependent, L-type calcium channels (6).
In both vascular and nonvascular smooth muscle, activation of large conductance, calcium-sensitive K ϩ channels (maxi-K or BK Ca channels) is reported to occur in response to endogenous NO or exogenous NO donors (7)(8)(9)(10)(11)(12). In many cases, addition of exogenous cGMP appears to mimic the effects of NO and NO donors on BK Ca channel activation (8,10,(12)(13)(14)(15), suggesting that cGMP acts downstream of NO. Physiologically, BK Ca channels appear to be important cellular effectors for the vasodilatory actions of the NO/cGMP signaling pathway because blockade of BK Ca channels can interfere with the relaxationpromoting effects of NO (16 -18).
A major intracellular target for cGMP in smooth muscle is the type I cGMP-dependent protein kinase (cGKI), a serine/ threonine protein kinase that is widely expressed in mammalian tissues (19,20). This kinase is encoded by a single gene, which gives rise to two alternatively spliced isoforms, ␣ and ␤, differing only in their N-terminal domains (36% of the first 103 amino acids of the ␤ isoform are identical to those of the ␣ isoform) (19,21). Both the ␣ and ␤ isoforms of the type I cGMP-dependent protein kinase functionally exist as homodimers (i.e. ␣/␣ and ␤/␤), in which each subunit contains a catalytic domain, 2 cGMP-binding sites, and an N-terminal dimerization region (4,19). Smooth muscle expresses both isoforms (22), although the biological roles of each are not well understood.
To examine the role played by cGKI␣ in the activation of cellular BK Ca channels by the NO/cGMP signaling cascade, we created a catalytically inactive or "dead" mutant of cGKI␣ that could be co-expressed with murine BK Ca channels. Utilizing a dominant negative suppression strategy (23,24), we observed that dead cGKI␣ prevented activation of BK Ca channels in cell-attached patches of HEK 293 cells in response to the nitrovasodilator sodium nitroprusside (SNP) or dibutyryl cGMP, a membrane-permeable analog of cGMP. Using a reciprocal co-immunoprecipitation strategy, we found that the endogenous and expressed forms of cGKI␣ were able to associate with BK Ca channel protein. We also observed that cGKI␣ is able to directly phosphorylate BK Ca channels in vitro, supporting the hypothesis that a similar event may be responsible for the enhancement of channel activity by cGMP in situ. Taken to-gether with previous observations, our results strongly suggest that the ␣ isoform of type I cGMP-dependent protein kinase represents a major downstream effector in the activation of BK Ca channels by the NO/cGMP signaling pathway in situ.

MATERIALS AND METHODS
Rabbit polyclonal antibodies against the mouse BK Ca channel ␣ subunit and the human type I cGMP-dependent protein kinase were obtained from Chemicon International and Calbiochem, respectively. A horseradish peroxidase-linked, mouse anti-rabbit IgG monoclonal antibody (clone RG-96) was purchased from Sigma Chemical Co. The purified, recombinant cGMP-dependent protein kinase I␣ enzyme was purchased from Calbiochem. The cGMP-dependent protein kinase-selective peptide substrate RKRSRAE was obtained from Peninsula Laboratories. The Lowry protein assay kit (detergent compatible) was purchased from Bio-Rad Laboratories.
Construction and Transfection of cDNA Plasmids-The cDNAs encoding the mouse BK Ca channel (mSlo) ␣ subunit (25), the wild-type green fluorescent protein (26), and the human cGKI␣ (27) were subcloned into the polylinker region of the SV40 promoter-based mammalian expression plasmid SR␣ using standard techniques. Site-directed mutagenesis was carried out using the Transformer mutagenesis kit (CLONTECH), which is based upon the unique restriction site elimination strategy (28). The enzymatically dead form of cGKI␣ was prepared by a Lys to Met substitution at amino acid position 393 within the kinase's catalytic domain.
Transient transfection of HEK 293 cells (50 -80% confluence) was carried out in 35-mm tissue culture dishes using the lipofection technique. Briefly, 6 -8 l of LipofectAMINE (Life Technologies, Inc.) was mixed together with ϳ1.5 g of plasmid cDNA in 1 ml of serum-free culture medium (Dulbecco's modified Eagle's medium supplemented with L-glutamine and 4.5 g/liter D-glucose) and placed on cells for 4 -6 h in a humidified incubator containing 5% CO 2 at 37°C. DNA-containing medium was then aspirated and replaced with serum-containing medium. The following day, cells were detached from the dish by treatment with 0.05% (w/v) trypsin/0.5 mM EDTA and replated onto sterile glass coverslips. Electrophysiological recordings were performed on days 3-5 after transfection (day 1). For biochemical studies, cells detached from 35-mm dishes were replated onto 100-mm dishes to prevent overgrowth. These cells were then harvested on days 3-4 after transfection.
Electrophysiology-Macroscopic currents were recorded at 35 Ϯ 0.5°C from cell-attached membrane patches of HEK 293 cells using an Axopatch 200B patch clamp amplifier and pClamp 6.03 software. BK Ca channel currents were activated by voltage clamp pulses delivered from a holding potential of 0 mV to membrane potentials ranging from Ϫ90 to ϩ150 mV; tail currents were recorded at Ϫ80 mV. Current traces were filtered at 2-5 kHz (4-pole Bessel filter) and acquired on a Dell Pentium II-based computer at a sampling frequency of 8 -10 kHz using a Digidata 1200 analog/digital interface. Recording micropipettes were pulled from thin-walled borosilicate glass capillaries (1.2 mm inner diameter; 1.5 mm outer diameter; WPI, Sarasota, FL) using a Sutter P-89 horizontal electrode puller. Micropipettes were filled with a solution containing 5 mM KCl, 140 mM KOH, 1 mM MgCl 2 , 1 mM CaCl 2 , and 10 mM HEPES (pH adjusted to 7.3 with methanesulfonic acid) and had tip resistances of 2-3.5 megaohms. The bath solution contained 5 mM KCl, 140 mM KOH, 1 mM MgCl 2 , 1 mM CaCl 2 , and10 mM HEPES (pH adjusted to 7.3 with methanesulfonic acid). The recording chamber (ϳ0.3 ml volume) was perfused by gravity flow at a constant rate of 1-1.5 ml/min, using a set of manually controlled solenoid valves to switch between solutions. Reagents were added directly to the solution reservoir tubes at the concentrations indicated.
Transfected HEK 293 cells seeded on coverslips were placed in a temperature-controlled recording chamber on the stage of a Nikon Eclipse TE300 inverted microscope. Individual cells expressing BK Ca channels were then identified visually by co-expression of the marker protein green fluorescent protein under epifluorescence using 480 nm excitation and 510 nm emission filters.
After measurements of protein concentration were carried out using a modified Lowry procedure (29), lysates were mixed with Laemmli sample buffer containing 1% (v/v) ␤-mercaptoethanol and incubated for 20 -30 min at 70°C, and the proteins were then separated by SDSpolyacrylamide gel electrophoresis (30). The resolved proteins were electrotransferred to nitrocellulose membrane at 4°C in a buffer containing 25 mM Tris, 192 mM glycine, 0.1% (w/v) SDS, and 20% (v/v) methanol for either ϳ2 h at 80 -90 V or overnight at 35 V (31). Membranes were first dried in a fume hood to fix proteins and then rinsed briefly in a buffer containing 20 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.1% (v/v) Tween 20 (TTBS). Membranes were incubated at room temperature for 20 -30 min in TTBS containing 10% (w/v) skim milk powder to block nonspecific binding of antibodies and then rinsed three times for 5 min each time in TTBS. Incubation of membranes with primary antibodies was carried out in TTBS containing 1% (w/v) skim milk powder for 1-2 h at room temperature, followed by three to five 10-min washes with TTBS alone. Membranes were then incubated for ϳ1 h at room temperature with a horseradish peroxidase-linked, mouse anti-rabbit secondary antibody also diluted in TTBS/1% (w/v) skim milk powder, followed by three to five 5-min washes with TTBS. After the final wash, blots were developed immediately by applying the Super-Signal chemiluminescence reagent (Pierce Chemical Co.) for ϳ2 min and then exposing the blots to x-ray film (Hyperfilm; Amersham Pharmacia Biotech).
In Vitro cGMP-dependent Protein Kinase Assay-This assay was performed as described previously by Wolfe et al. (32), with minor modifications. Transiently transfected HEK 293 cells growing on a 100-mm culture dish (50 -80% confluence) were harvested in 1 ml of lysis buffer, sonicated for 5-10 s on ice, and then centrifuged for 10 min at 4°C at 15,000 rpm using a microcentrifuge. The supernatants were removed and kept on ice.
The assay of cGMP-dependent protein kinase activity was performed at 30°C in a final reaction volume of 60 l containing (final concentrations) 20 mM Tris-HCl, pH 7.5, 20 mM magnesium acetate, 2 mM DTT, 20 M cGMP, 0.1 mM isobutylmethylxanthine, 160 M synthetic peptide substrate (RKRSRAE) (33), and 200 M Na 2 -ATP (radiospecific activity, 150 -300 cpm/pmol). After the addition of 20 -50 g of soluble cell lysate to start the reaction, incubation was carried out for 10 min and stopped by spotting a 30-l aliquot of the reaction mixture onto a P-81 phosphocellulose disc (Whatman). Discs were then placed immediately in 0.5% (v/v) phosphoric acid, washed for 3 ϫ 5 min in 500 ml of 0.5% phosphoric acid, rinsed briefly with acetone, and dried. Radioactivity bound to the paper discs was quantified by Cerenkov counting using a Beckman LS6000 liquid scintillation counter.
In Vitro Phosphorylation Assay-The BK Ca ␣ subunit was immunoprecipitated from transiently transfected HEK 293 cells as follows: on day 3-4 after transfection, cells growing on a 100-mm dish were harvested in 1 ml of lysis buffer, sonicated briefly, and then centrifuged for 10 min at 4°C at 15,000 rpm in a microcentrifuge. The supernatant was diluted to 0.4 -0.5 mg protein/ml, and a 1.4-ml aliquot of the diluted lysate was transferred to a clean microcentrifuge tube. Bovine serum albumin was then added to a final concentration of 1 mg/ml. This sample was pre-cleared by the addition of 30 l of a 50% slurry (v/v) of rehydrated protein A-Sepharose beads (Amersham Pharmacia Biotech), followed by rotation at 4°C for 1 h. Samples were centrifuged for 5 min at 10,000 rpm to pellet the beads, and the soluble material was transferred to a clean microcentrifuge tube. Pre-cleared supernatants were then incubated for 4 -16 h at 4°C with ϳ1.5 g of anti-BK Ca channel antibody, followed by further incubation for 2 h with 30 l of protein A-Sepharose beads (50% slurry). The beads were pelleted by centrifugation at 4°C for 5 min at 3,000 rpm and then washed twice by resuspension in 1 ml of wash buffer containing 20 mM Tris-HCl, pH 7.4, 140 mM NaCl, 5 mM KCl, 1 mM DTT, 1 mM EDTA, 0.2 mM EGTA, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 0.1% (v/v) Triton X-100, and 5 g/ml each of aprotinin, leupeptin, and pepstatin A, followed by a final wash in the same buffer minus Triton X-100. The beads were then resuspended in 20 -30 l of a buffer containing 20 mM Tris HCl, pH 7.5, and 2 mM DTT; 15-l aliquots were used directly in the assay.
The phosphorylation reaction was carried out for 20 min at 30°C in a final volume of 40 l containing (final concentrations) 20 mM Tris-HCl, pH 7.5, 20 mM magnesium acetate, 2 mM DTT, 0.1 mM isobutylmethylxanthine, 20 M cGMP, 5,000 units of recombinant bovine type I␣ cGMP-dependent protein kinase, and 15 l of resuspended protein A-Sepharose beads containing immunoprecipitated BK Ca ␣ subunit. The reaction was started by the addition of [␥-32 P]ATP (20 M; final concentration, 20,000 -30,000 cpm/pmol) and stopped by addition of concentrated (4ϫ) Laemmli sample buffer directly to the assay tubes.
Reaction mixtures were then resolved by SDS-polyacrylamide gel electrophoresis. Gels were stained and destained and then directly exposed to x-ray film in a Kodak BioMax cassette at room temperature for 12-24 h.
Co-immunoprecipitation of cGKI␣ and BK Ca Channels-After transient expression of cGKIa and BK Ca channel cDNAs, either alone or together, the immunoprecipitation protocol was carried out as described above, with minor modifications. Upon addition of 0.5-1.0 ml of lysis buffer, cells were kept on ice for 20 -30 min and then disrupted by 10 -12 up and down strokes of a disposable plastic pestle designed to fit 1.5-ml Eppendorf centrifuge tubes. For immunoprecipitation of cGKI␣, ϳ4 g of rabbit polyclonal anti-cGKI antibody (Calbiochem) was added to each pre-cleared cellular lysate. Subsequent steps in the protocol were unchanged.
Statistical Analysis-A one-way analysis of variance was used to determine statistical significance among groups of values; mean values were considered to be significantly different at a level of p Ͻ 0.05.

RESULTS
Characterization of Transfected Type I cGMP-dependent Protein Kinase-To examine the functional importance of cGKI␣ in the regulation of BK Ca channels by cGMP and the nitrovasodilator compound SNP, we hypothesized that a catalytically inactive mutant of cGKI␣ could be used in a dominant negative strategy (23,24) to suppress or interfere with the actions of endogenous cGKI␣ in an intact cell. Earlier reports have shown that native HEK 293 cells contain an endogenous cGMP-dependent protein kinase activity (15,34,35); our new data, described below, strongly suggest that this activity can be accounted for by the presence of the ␣ isoform of type I cGMPdependent protein kinase in these cells.
To prepare a catalytically inactive mutant of human cGKI␣ (27), we replaced a critical lysine residue (Lys 393 ) in the ATPbinding motif of the enzyme's catalytic domain with methionine (36,37). The effectiveness of this substitution was verified using two complementary biochemical approaches. First, an in vitro kinase assay was performed using soluble cell lysates from HEK 293 cells transiently transfected with cDNA constructs encoding a BK Ca channel ␣ subunit together with either the wild-type or mutant form of cGKI␣. Fig. 1 shows that the detergent-solubilized lysate from cells transfected with the BK Ca channel cDNA alone displayed modest cGMP-dependent protein kinase activity, as quantified by measuring phosphorylation of a cGKI-selective peptide substrate (33).
In the presence of cGMP, substrate phosphorylation was increased ϳ2-fold above background levels observed in the absence of cGMP, and this activity was comparable to that present in mock-transfected cells. However, cGMP-dependent substrate phosphorylation was strongly enhanced (ϳ8-fold) in the lysate from cells co-transfected with BK Ca channels and wild-type cGKI␣, demonstrating expression of intact, biochemically active protein kinase. In contrast to this situation, cells transiently transfected with BK Ca channels and the catalytically inactive or dead form of cGKI␣ displayed a level of cGMPdependent kinase activity that was very similar to that of cells transfected with BK Ca channels alone. This finding is thus consistent with results using other protein kinases (36,38), in which mutation of this invariant lysine residue within the catalytic domain is sufficient to block expression of enzymatic activity. It is interesting that co-expression of dead cGKI␣ did not lower the total observable cGMP-dependent protein kinase activity measured in vitro. This observation may be due to: (a) transfection efficiency of HEK 293 cells with the Lipo-fectAMINE reagent (Life Technologies, Inc.) ranges from 20 -30% in our hands; therefore, endogenous cGMP-dependent activity in the majority of cells remains unaffected, and (b) the assay mixture for cGMP-dependent protein kinase activity contains saturating concentrations of Mg-ATP, cGMP, and peptide substrate, therefore the presence of inactive kinase molecules would not be expected to interfere with the activity of native kinase by depletion of essential reagents in vitro.
Co-expression of BK Ca Channels and cGKI␣- Fig. 2A shows a Western blot of the detergent-soluble cellular lysates used for the in vitro kinase assays shown in Fig. 1.
Using an anti-cGKI antibody, we observed modest expression of an endogenous immunoreactive protein with a molecular mass of ϳ75 kDa, corresponding to the ␣ isoform of cGKI (22). No expression of a 78 -80-kDa ␤ isoform was detected in our HEK 293 cells, although both ␣ and ␤ isoforms were readily observed under the same Western blotting conditions in isolated smooth muscle myocytes from rabbit aorta (data not shown). In contrast, HEK 293 cells co-transfected with cDNAs encoding either the wild-type or mutant form of cGKI␣ showed very strong expression of a similar 75-kDa band, consistent with the presence of exogenous cGKI␣ protein. The lower immunoreactive bands with a molecular mass of ϳ60 kDa most likely represent proteolytic fragments of the full-length wildtype and mutant cGKI␣ (19). Taken together, the results shown in Figs. 1 and 2A demonstrate that both the wild-type and mutant forms of recombinant human cGKI␣ can be strongly expressed in HEK 293 cells, with only the wild-type enzyme displaying significant cGMP-dependent kinase activity. We then examined whether co-transfection of cGKI␣ influenced the expression pattern of BK Ca channels themselves. Fig.  2B shows a Western blot of the same cellular extracts used in Fig. 2A, which was probed with an antibody against the BK Ca channel ␣ subunit. A single immunoreactive band of ϳ125 kDa was observed, the level of which was comparable under all three transfection conditions. Similar results were obtained in two additional experiments. Taken together, these observations demonstrate that co-transfection of HEK 293 cells with either wild-type or mutant cGKI␣ cDNA leads to the expected expression of both protein and kinase activity, without altering the expression pattern of BK Ca channel protein.
Effects of the NO/cGMP Signaling Pathway on BK Ca Channel Activity-The functional importance of cGKI␣ in the regulation of BKCa channels was examined by using patch clamp techniques to record BK Ca channel activity in cell-attached membrane patches of transfected HEK 293 cells in the absence and presence of 100 M SNP. Fig. 3A shows BK Ca channel activity before and during bath application of 100 M SNP.
In the absence of SNP, a modest level of BK Ca channel activity was observed, the magnitude of which varied from cell to cell. However, after ϳ4 min of exposure to SNP, we observed a large increase in the amplitude of BK Ca channel macroscopic current; this increase typically peaked during 2-6 min of exposure and was reversible over several minutes upon washout of SNP from the bath (Fig. 3C). SNP produced an average increase of ϳ4-fold in current magnitude compared with control, as quantified in Fig. 3C. This observation is thus consistent with recent findings reported by Fukao et al. (15) using whole cell voltage clamp methodologies that SNP could augment the activity of canine colonic BK Ca channels expressed in HEK 293 cells. Using our dominant negative strategy, we observed that in cells co-transfected with cDNAs encoding the BK Ca channel and the catalytically dead form of cGKI␣, the effect of bath-applied SNP on BK Ca channel activity was significantly blunted compared with cells expressing BK Ca channels alone (Fig. 3, B and C). Interestingly, in cells co-transfected with BK Ca channels and wild-type cGKI␣, bath application of SNP also produced a large enhancement of BK Ca channel activity, although the effect was not significantly higher than that observed with cells expressing BK Ca channels alone (Fig. 3C). This result suggests that the bath concentration of SNP and the amount of endogenous cGMP-dependent protein kinase in HEK 293 cells are sufficiently high to produce maximal stimulation of BK Ca channel activity; further expression of wild-type cGKI␣ via transfection does not appear to augment this already maximal response.
Although SNP is known to elevate intracellular cGMP in several cell types via the NO-dependent activation of guanylyl cyclase (39), it is also capable of generating chemical products (i.e. peroxynitrites, S-nitrosothiols, and ferrocyanates) that may directly influence BK Ca channel activity (40,41). To address whether SNP may initiate other cellular mechanisms not dependent upon activation of cGKI, we examined the effect of dibutyryl cGMP, a membrane-permeable form of cGMP, on BK Ca channels expressed alone or in the presence of co-transfected wild-type cGKI␣ or catalytically dead cGKI␣. Fig. 4A shows cellattached patch clamp recordings of BK Ca channels under control conditions and after activation by dibutyryl cGMP.
Dibutyryl cGMP (1 mM) was back-filled in the recording pipette and allowed to diffuse to the membrane patch at the pipette tip after formation of a high resistance, gigaohm seal. Over a 10-min recording period, we observed that dibutyryl cGMP (db-cGMP) produced an increase (4 -5-fold above control) in BK Ca channel activity that was qualitatively similar to that seen with exposure to SNP (compare Fig. 4, A and C with Fig.  3, A and C). In the absence of db-cGMP, no significant change

FIG. 2. Protein expression levels of wild-type and catalytically dead cGKI␣ in transiently transfected HEK 293 cells. A shows a
Western blot of equal amounts of total cellular lysates (ϳ75 g/lane) from HEK 293 cells transfected with the BK Ca channel ␣ subunit alone (BK alone) or together with either wild-type cGKI␣ (BKϩ cGKI␣) or catalytically inactive cGKI␣ (BK ϩ dead cGKI␣). An antibody specifically recognizing the type I cGMP-dependent protein kinase detected a ϳ75-kDa band in each lane; however, the intensity of this band was much greater in cells transfected with cGKI␣ cDNA. The lower band at ϳ60 kDa most likely represents a proteolytic fragment of full-length cGKI␣ (19). After detection of cGKI␣, the blot was stripped and reprobed with an antibody recognizing the BK Ca channel ␣ subunit. A single band of ϳ125 kDa was detected in each lysate (B). The electrophoretic mobility of molecular mass markers (in kDa) is indicated to the right of each panel.

FIG. 3. Effect of transiently expressed wild-type and catalytically dead cGKI␣ on the enhancement of BK Ca channel activity by sodium nitroprusside.
A and B show cell-attached patch clamp recordings of macroscopic BK Ca channel currents from HEK 293 cells transiently transfected with BK Ca channel ␣ subunit alone (A) or together with catalytically inactive (dead) cGKI␣ (B). BK Ca channel currents in cell-attached membrane patches were evoked by voltage clamp steps from Ϫ90 to ϩ150 mV, in 10-mV increments; the membrane patch was held at 0 mV (refer to the inset). The top set of traces in each panel was recorded shortly after the formation of a gigaohm seal (designated as 0 min). Using continuous bath superfusion, cells were then exposed to 100 M SNP, and current families were recorded every 2 min for up to 6 -7 min after the start of exposure (the bottom set of traces in each panel). The fold change in current amplitude at ϩ100 mV in the absence or presence of SNP exposure versus initial control level is plotted in C for cells transfected with BK Ca channels alone or together with either catalytically inactive (dead) cGKI␣ or wild-type cGKI␣. The reversibility of stimulated current amplitude in cells expressing BK Ca channels alone after SNP wash-out (SNP W/O) for 6 -8 min is also indicated. Current amplitudes were quantified by measuring the average steadystate current over the last 5 ms of the depolarizing pulse. Data are presented as the means Ϯ S.E. (n ϭ 4 -6 cells in each group). An asterisk indicates that these values are significantly different from the control value in the absence of SNP, p Ͻ 0.05. in BK Ca channel activity was observed over the same time period (Fig. 4C). These findings are thus in agreement with those recently reported by Alioua et al. (14) for the activation of human BK Ca channels expressed in Xenopus oocytes by db-cGMP. Most interestingly, however, in cells co-transfected with BK Ca channels and dead cGKI␣, we observed that db-cGMP was no longer effective in enhancing channel activity when assayed over a 10-min recording period (Fig. 4, B and C). This result is thus similar to that observed with exposure of cotransfected cells to SNP (see Fig. 3B) and suggests that both SNP and db-cGMP act via a type I cGMP-dependent protein kinase to enhance BK Ca channel activity. Under the condition of BK Ca channels co-transfected with wild-type cGKI␣, db-cGMP was observed to increase current magnitude ϳ5-fold above control levels (see Fig. 4C). However, this increase was not significantly greater than that observed for BK Ca channels expressed alone, as was the case for stimulation by SNP. This result further supports the likelihood that under the conditions of our experiment, the level of endogenous cGKI␣ is sufficient to produce maximal enhancement of BK Ca channel activity.
Direct Phosphorylation of BK Ca Channels by Purified cGKI␣-Whereas it has been generally hypothesized that the regulation of BK Ca channels by cGMP may involve a phosphorylation event, only recently have results appeared in the literature directly supporting such a mechanism (9,14,15,42,43). To examine whether BK Ca channels expressed in HEK 293 cells may undergo direct phosphorylation in the presence of cGKI␣, we isolated expressed BK Ca channels by immunoprecipitation and then incubated these purified channels in either the absence or presence of purified cGKI␣. Fig. 5A shows an autoradiogram of an in vitro phosphorylation assay using immunoprecipitates from cells transfected with BK Ca channel cDNA or cells transfected with empty vector alone (mock-transfected cells).
In the lane containing immunoprecipitated BK Ca channels plus purified cGKI␣, a major phosphorylated band of ϳ125 kDa was observed, most likely corresponding to the BK Ca channel ␣ subunit. The second major phosphoprotein observed in the same lane, migrating at ϳ75 kDa, corresponds to the purified cGKI␣ after autophosphorylation (4). As shown in Fig. 5A, addition of purified cGKI␣ to the immunoprecipitate from mock-transfected cells produced only a single phosphoprotein of ϳ75 kDa, again corresponding to the autophosphorylated cGKI␣ enzyme. The absence of this 75-kDa band from both reactions lacking addition of the purified cGKI␣ further supports this conclusion. To demonstrate the presence of BK Ca channel protein in our reactions, we performed a Western blot on the total cellular lysates and immunoprecipitates from both sets of transfected cells. Fig. 5B shows that the BK Ca channel ␣ subunit is strongly expressed in the total cellular lysate of positively transfected cells, with significant recovery in the immunoprecipitate. However, a similar immunoreactive band was not detected in either the lysate or immunoprecipitate of mock-transfected cells, consistent with the observed lack of a phosphoprotein of ϳ125 kDa in the autoradiogram of Fig.5A.
Interaction of cGKI␣ with BK Ca Channels in Situ-To address the potential mechanism by which expression of the catalytically inactive form of cGKI␣ interfered with stimulation of BK Ca channel activity by SNP and db-cGMP, BK Ca channels and cGKI␣ (wild-type or catalytically inactive forms) were transiently expressed either alone or together in HEK 293 cells. BK Ca channels were then directly immunoprecipitated, and the immunoprecipitates were probed by Western blot for the presence of associated cGKI␣.
Using this co-immunoprecipitation strategy, we found a small amount of expressed cGKI␣ associated with isolated BK Ca channels, and we found that brief stimulation by dibutyryl cGMP modestly enhanced this interaction (Fig. 6A). Under these same conditions, we further observed that the catalytically inactive form of cGKI␣ associated with immunoprecipitated BK Ca channels to a much greater extent compared with the expressed wild-type kinase. However, in cells transfected with BK Ca channels alone, we were unable to detect the presence of endogenous cGKI␣ in BK Ca channel immunoprecipitates. This inability to capture such a steady-state interaction may reflect the combination of relatively low amounts of endogenous kinase present in these cells compared with BK Ca channels and the transient nature of interaction of cGKI␣ with the BK Ca channel substrate. Similarly, in their recent study, Wang et al. (44) could not detect either endogenous cAMP-dependent protein kinase or cSrc tyrosine kinase in immunoprecipitates of recombinant Drosophila Slo channels transiently expressed in HEK 293 cells. However, when either kinase was co-expressed along with the channel, channel-kinase interactions could be observed using such a coimmunoprecipitation strategy. To carry out stimulation of individual cells by membrane-permeable cGMP, the extreme tip of the recording micropipette was first filled with normal pipette solution (see "Materials and Methods"), and then the majority of the pipette shaft was back-filled with the same solution containing 1 mM db-cGMP. In this strategy, dibutyryl cGMP was allowed to diffuse to the pipette tip and then across the cell membrane over the course of several minutes. BK Ca channel currents in cell-attached membrane patches were evoked by voltage clamp steps from Ϫ90 to ϩ150 mV, in 10-mV increments; steps were delivered from a holding potential of 0 mV (refer to the inset).
The top set of traces in each panel was recorded shortly after formation of a gigaohm seal (designated as 0 min). Cells were then continuously superfused under constant bath flow, and current families were recorded at 2-min intervals for up to 10 min after the initial seal formation (the bottom set of traces in each panel). The fold change in current amplitude at ϩ100 mV (average steady-state current over the last 5 ms of the pulse) in the absence or presence of db-cGMP exposure at 10 min versus initial control level (0 min) is plotted in C for cells transfected with BK channels alone or together with either catalytically inactive (dead) cGKI␣ or wild-type cGKI␣. Data are presented as the means Ϯ S.E. (n ϭ 3-6 cells in each group). The asterisk indicates that these values are significantly different from control (BK alone), p Ͻ 0.05. When BK Ca channel immunoprecipitates were reprobed for the presence of channel protein, we observed that similar amounts were recovered from cells transfected under each condition (Fig. 6B). This finding indicates that unequal immunoprecipitation of BK Ca channel protein cannot account for the difference observed in the levels of co-immunoprecipitated cGKI␣ shown in Fig. 6A. To further verify this result, we then probed equal amounts of the starting whole cell lysates for the expression of type I cGMP-dependent protein kinase to ensure that differences in the level of cGKI␣ expression between conditions did not account for the differential co-immunoprecipitation of cGKI␣ shown in Fig. 6A. Our observation that expression of the transiently expressed, wild-type cGKI␣ was greater than that of the catalytically inactive form of the kinase, which was greater than that of endogenous cGKI␣ (see Fig. 6C), indicates that gross differences in cGKI␣ expression can not account for the differential co-immunoprecipitation of wildtype and dead cGKI␣ with BK Ca channels presented in Fig. 6A. In Fig. 6C, the amount of endogenous type I cGMP-dependent protein kinase was below detection in cells transfected with BK Ca channels alone, likely due to the low amount of whole cell lysate loaded per lane (i.e. ϳ15 g). By restricting the amount of protein loaded per lane in this particular experiment, we were able to achieve a better comparison of the expression levels between the wild-type and catalytically inactive forms of cGKI␣ in the whole cell lysates shown in Fig. 6A.
This important association between BK Ca channels and cGKI␣ was further examined by performing reciprocal co-immunoprecipitation, in which we probed anti-cGKI immunoprecipitates for the presence of co-associated BK Ca channel protein. As expected, we observed that BK Ca channels co-immunoprecipitated with co-expressed wild-type or catalytically inactive cGKI␣ (Fig. 6D), although there was not the same marked difference as seen in Fig. 6A. Importantly, we also observed that the endogenous form of cGKI␣ in HEK 293 cells is able to co-associate with expressed BK Ca channels, as demonstrated by the presence of these two proteins in the same anti-cGKI␣ immunoprecipitates. When these immunoprecipitates were reprobed for the presence of cGKI␣ protein, we observed similar levels of either expressed wild-type or inactive cGKI␣, along with modest amounts of the endogenous form of cGKI␣ (Fig. 6E). A Western blot of the initial whole cell lysates probed for the BK Ca channel ␣ subunit demonstrates similar expression of BK Ca channel protein in the four groups of transfected cells (Fig. 6F), thus confirming equal starting conditions for the immunoprecipitation results shown in Fig. 6D.
Finally, we believe that the difference in co-immunoprecipitation data shown in Fig. 6, A and D, may reflect the relative expression of cGKI␣ versus BK Ca channels. If we consider that cGKI␣ is expressed to a greater level than BK Ca channel protein under the conditions of our transient co-transfection, then we would anticipate that there is a greater likelihood to observe cGKI␣ co-associated with BK Ca channel immunoprecipitates because the kinase molecules are present in excess quantity. However, for cGKI␣ immunoprecipitation, the majority of either wild-type or inactive kinase molecules would not be associated with BK Ca channel protein, which decreases the probability that immunoprecipitated cGKI␣ will have a channel molecule bound to it. This situation leads to a low recovery of kinase co-associated with the channel, which effectively dilutes any observable differences in the detected co-associations. This idea is supported by our observation that expressed BK Ca channels can be detected in immunoprecipitates of endogenous cGKI (Fig. 6D), which is present at much lower levels than the expressed forms of the kinase. This low expression ratio be-tween endogenous kinase and expressed channel thereby increases the likelihood that an immunoprecipitated kinase molecule will be co-associated with a channel protein. Therefore, immunoprecipitation of the lesser of these two proteins in-FIG. 6. Interaction of cGMP-dependent protein kinase with BK Ca channels in situ. BK Ca channels or the ␣ isoform of human cGKI␣ (wild-type or catalytically inactive forms) were transiently expressed either alone or together in HEK 293 cells; transfection conditions for each lane in A-C and D-F are indicated at the top of A and D, respectively. Intact cells were stimulated for 3 min at 30°C in either the absence (Ϫ) or presence (ϩ) of 1 mM db-cGMP, as indicated above each lane in A and D. After stimulation, medium was aspirated, and cells were lysed immediately and subjected to immunoprecipitation using either an anti-BK Ca channel ␣ subunit antibody (A and B) or an anti-GKI antibody (D and E). A and E show a Western blot of the isolated immunoprecipitates probed with an antibody versus cGKI␣. These blots were then stripped and reprobed using an anti-BK Ca channel ␣ subunit antibody (B and D). C and F show Western blots of the initial cell lysates (ϳ15 g protein loaded/lane in C, ϳ45 g protein loaded/lane in F) from each of the transfection conditions, which were probed with an antibody versus cGKI␣ (C) or the BK Ca channel ␣ subunit (F). The prominent bands of ϳ55 kDa detected in A, B, D, and E represent the heavy chains of the rabbit polyclonal anti-BK Ca channel and anti-cGKI antibodies used for immunoprecipitation. The electrophoretic mobility of molecular mass standards is shown to the right of each panel.
creases the likelihood of detecting an interaction with the more abundant partner. DISCUSSION In this study, we have examined the importance of the cGKI in the regulation of a BK Ca channel by the NO/cGMP signaling pathway in intact cells. To do so, we created a catalytically inactive mutant of the ␣ isoform of cGKI (see "Materials and Methods") that could be transfected and expressed in mammalian cells. We anticipated that this mutant would selectively target and interfere with the function of endogenous cGKI␣ in a dominant negative fashion (23,24), which is not possible with the nonselective cGMP-dependent protein kinase inhibitor KT5823 (15) or the disruption of the cGKI gene (45), leading to loss of both ␣ and ␤ isoforms. The results shown in Figs. 1 and 2 demonstrate that HEK 293 cells transiently transfected with BK Ca channel cDNA expressed a measurable level of endogenous cGMP-dependent protein kinase activity that correlated with the presence of the ϳ75-kDa ␣ isoform of cGKI, as detected by Western blotting. Transient expression of wild-type cGKI␣ produced a large increase in cGMP-dependent protein kinase activity, as measured in total cell lysates, which correlated with a large increase in immunoreactive cGKI␣. In contrast, transient expression of the catalytically inactive or dead mutant of cGKI␣ produced no change in measurable cGMP-dependent protein kinase activity compared with control (BK Ca channel alone) but led to a similar large increase in the expression of immunoreactive type I cGMP-dependent protein kinase, indicating the presence of transfected cGKI␣ protein.
Having established both the activity and expression of endogenous cGMP-dependent protein kinase in our HEK 293 cells, we examined whether stimulation of this intrinsic pathway by the NO donor SNP or membrane-permeable db-cGMP could result in altered BK Ca channel activity. Using the cellattached recording mode of the patch clamp technique (to keep the intracellular milieu intact), we observed that exposure of cells to either SNP (Fig. 3A) or db-cGMP (Fig. 4A) significantly increased the magnitude of macroscopic BK Ca channel currents, in agreement with recent observations of other investigators (14,15). Taken together with our biochemical data above, these electrophysiological data would be consistent with a role for endogenous cGMP-dependent protein kinase in the augmentation of BK Ca channel activity by SNP and db-cGMP in these cells. Given that the primary function of protein kinases is to phosphorylate selected substrates, we anticipated that BK Ca channels would undergo serine/threonine phosphorylation in the presence of cGKI␣. The results shown in Fig. 5 clearly demonstrate that purified cGKI␣ can readily phosphorylate immunoprecipitated BK Ca channels in vitro, in agreement with the findings of others (14,42). Such direct phosphorylation of BK Ca channels by cGKI␣ in situ would thus serve as the basis for the enhancement of BK Ca channel activity observed electrophysiologically after the addition of cGMP, Mg-ATP, and cGKI␣ to excised membrane patches (9,15,43,46) and would also explain how enhancement could be maintained after excision of inside-out membrane patches from stimulated cells (14). 2 If cGKI␣ is indeed a critical component in the regulation of BK Ca channel activity by the NO/cGMP signaling pathway, then we would predict that expression of the catalytically inactive cGKI␣ mutant described above would selectively prevent such augmentation in response to SNP or db-cGMP. As shown in Figs. 3B and 4B, co-expression of BK Ca channels with dead cGKI␣ does in fact preclude augmentation of channel activity by either SNP or db-cGMP when compared with BK Ca channels expressed alone. This observation thus suggests that the ␣ isoform of cGKI alone is sufficient to support the regulation of BK Ca channel activity by the NO/cGMP signaling pathway in cells expressing a type I cGMP-dependent protein kinase. Based on these novel results, along with the previous observations of others (9,14,15,43), we conclude that cGKI acts directly on BK Ca channels in situ, resulting in enhanced channel activity. Our conclusion thus agrees with that of a recent study by Sausbier et al. (47), who used the cGKI-deficient mouse to demonstrate the important role of cGKI in both the activation of BK Ca channels and nitric oxide/cGMP-dependent vasodilation. Recent studies from Han et al. (48) and White et al. (49) further demonstrate that cGKI is the primary protein kinase involved in the activation of BK Ca channels in coronary smooth muscle myocytes by vasodilatory, cAMP-elevating agents such as dopamine or forskolin.
Several groups have already reported that protein kinases may physically associate with membrane ion channels (50 -54), presumably as part of a phosphorylation-dependent regulatory mechanism. Our findings (see Fig. 6, A and D) that cGKI␣ can associate with mammalian BK Ca channels are analogous to recent results showing that endogenous cAMP-dependent protein kinase and Src tyrosine kinase can independently associate with native BK Ca channels immunoprecipitated from Drosophila head (44). In the context of phosphorylation-dependent regulation, the stronger association observed for catalytically inactive cGKI␣ with BK Ca channels compared with wild-type kinase may serve to explain how the dead kinase acts in a dominant negative fashion to suppress enhancement of channel activity via the NO/cGMP signaling pathway. This stronger interaction of dead kinase with the channel could thus account for the observed suppression of channel activity in situ by (a) binding and depletion of the channel as a phosphorylation substrate and/or (b) displacement of active cGKI␣ from anchoring proteins that localize the kinase near the BK Ca channel complex (23,24).
In summary, the findings of our study using a dominant negative suppression strategy implicate an important role for cGKI␣ in the enhancement of BK Ca channel activity by the NO/cGMP signaling pathway in intact cells. These results are further consistent with the observed phenotype of cGKI-deficient knockout mice, which display impaired endothelium and NO-dependent relaxation of smooth muscle, resulting in vascular and intestinal dysfunction (45,47). The question of whether NO and NO donors may also be able to directly activate BK Ca channels is not supported by our data and remains controversial (55)(56)(57)(58); it is possible that such a phenomenon may depend upon the specific preparation in use, along with the types and concentrations of agents under study.