A Short Motif in Kir6.1 Consisting of Four Phosphorylation Repeats Underlies the Vascular KATP Channel Inhibition by Protein Kinase C*

Vascular ATP-sensitive K+ channels are inhibited by multiple vasoconstricting hormones via the protein kinase C (PKC) pathway. However, the molecular substrates for PKC phosphorylation remain unknown. To identify the PKC sites, Kir6.1/SUR2B and Kir6.2/SUR2B were expressed in HEK293 cells. Following channel activation by pinacidil, the catalytic fragment of PKC inhibited the Kir6.1/SUR2B currents but not the Kir6.2/SUR2B currents. Phorbol 12-myristate 13-acetate (a PKC activator) had similar effects. Using Kir6.1-Kir6.2 chimeras, two critical protein domains for the PKC-dependent channel inhibition were identified. The proximal N terminus of Kir6.1 was necessary for channel inhibition. Because there was no PKC phosphorylation site in the N-terminal region, our results suggest its potential involvement in channel gating. The distal C terminus of Kir6.1 was crucial where there are several consensus PKC sites. Mutation of Ser-354, Ser-379, Ser-385, Ser-391, or Ser-397 to nonphosphorylatable alanine reduced PKC inhibition moderately but significantly. Combined mutations of these residues had greater effects. The channel inhibition was almost completely abolished when 5 of them were jointly mutated. In vitro phosphorylation assay showed that 4 of the serine residues were necessary for the PKC-dependent 32P incorporation into the distal C-terminal peptides. Thus, a motif containing four phosphorylation repeats is identified in the Kir6.1 subunit underlying the PKC-dependent inhibition of the Kir6.1/SUR2B channel. The presence of the phosphorylation motif in Kir6.1, but not in its close relative Kir6.2, suggests that the vascular KATP channel may have undergone evolutionary optimization, allowing it to be regulated by a variety of vasoconstricting hormones and neurotransmitters.

Vascular ATP-sensitive K ؉ channels are inhibited by multiple vasoconstricting hormones via the protein kinase C (PKC) pathway. However, the molecular substrates for PKC phosphorylation remain unknown. To identify the PKC sites, Kir6.1/SUR2B and Kir6.2/SUR2B were expressed in HEK293 cells. Following channel activation by pinacidil, the catalytic fragment of PKC inhibited the Kir6.1/SUR2B currents but not the Kir6.2/SUR2B currents. Phorbol 12-myristate 13-acetate (a PKC activator) had similar effects. Using Kir6.1-Kir6.2 chimeras, two critical protein domains for the PKC-dependent channel inhibition were identified. The proximal N terminus of Kir6.1 was necessary for channel inhibition. Because there was no PKC phosphorylation site in the N-terminal region, our results suggest its potential involvement in channel gating. The distal C terminus of Kir6.1 was crucial where there are several consensus PKC sites. Mutation of Ser-354, Ser-379, Ser-385, Ser-391, or Ser-397 to nonphosphorylatable alanine reduced PKC inhibition moderately but significantly. Combined mutations of these residues had greater effects. The channel inhibition was almost completely abolished when 5 of them were jointly mutated. In vitro phosphorylation assay showed that 4 of the serine residues were necessary for the PKC-dependent 32 P incorporation into the distal C-terminal peptides. Thus, a motif containing four phosphorylation repeats is identified in the Kir6.1 subunit underlying the PKC-dependent inhibition of the Kir6.1/ SUR2B channel. The presence of the phosphorylation motif in Kir6.1, but not in its close relative Kir6.2, suggests that the vascular K ATP channel may have undergone evolutionary optimization, allowing it to be regulated by a variety of vasoconstricting hormones and neurotransmitters.
ATP-sensitive K ϩ (K ATP ) channels play an important role in vascular tone regulations (1)(2)(3). Such a function attributes to channel regulation by a variety of vasodilating and vasoconstricting hormones and neurotransmitters (3)(4)(5)(6)(7). Therefore, the understanding of the molecular basis for channel regulation has an impact on the design of therapeutic modalities by targeting at specific molecular substrates of the channel. It is known that the major isoform of K ATP channels in vascular smooth muscles is composed of Kir6.1 and SUR2B (8 -12). Genetic disruption of Kir6.1 or SUR2 indeed results in a phenotype of Prinzmetal angina with a high rate of sudden death (13,14), consistent with the importance of the channel in vascular regulations.
Experimental evidence suggests that vasoconstrictors act on the vascular K ATP channel through the PKC 2 signaling system. Our previous studies have shown that the Kir6.1/SUR2B channel and its counterpart in vascular smooth muscle cells are inhibited by vasopressin and that channel inhibition can be abolished by specific PKC blockers (6). Similar observations have been made by other groups with endothelin (15), muscarinic M3 receptor agonist (16), and angiotensin II (17). The effect of angiotensin II on the vascular K ATP channel requires translocation of PKC⑀ to plasma membranes (18,19). By comparing the effects of the acetylcholine M3 receptor on Kir6.1/ SUR2B and Kir6.2/SUR2B channels, Quinn et al. (16) suggest that Kir6.1/SUR2B channel inhibition is mediated via a direct effect of PKC rather than a change in phosphatidyl 4,5-bisphosphate concentrations. They also show evidence for Kir6.1 phosphorylation using in vitro biochemical assay (16). Consistently, purified PKC inhibits the cloned Kir6.1/SUR2B and vascular smooth muscle endogenous K ATP channels in inside-out patches where cytosolic soluble components are absent (4,20). Although these previous studies have significantly improved our understanding of vascular K ATP channel regulation by vasoconstrictors, the molecular substrate of PKC remains unknown. Therefore, we performed these studies to identify the critical protein domain and amino acid residues for PKC phosphorylation.

MATERIALS AND METHODS
Molecular Biology-Rat Kir6.1 (GenBank TM accession number D42145), mouse Kir6.2 (D50581), and mouse SUR2B (D86038) cDNAs were used in the present study. The cDNAs were cloned in the eukaryotic expression vector pcDNA3.1 and used for mammalian cell expression (5,6). Kir6.1-Kir6.2 chimeras were produced by overlap extension using PCR (Pfu DNA polymerase; Stratagene, La Jolla, CA) (21). Site-specific mutations were made using a site-directed mutagenesis kit (Stratagene). The orientation of the constructs and the correct muta-* This work was supported by National Institutes of Health Grant HL067890.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  tions were confirmed with DNA sequencing. The constructs were expressed in HEK293 cells (American Type Culture Collection, Manassas, VA) as detailed in our previous reports (5,6). Patch clamp experiments were performed at room temperature as described previously (5,6). In brief, the bath solution for whole-cell recording contained (in mmol/L) 10 KCl, 135 potassium gluconate, 5 EGTA, 5 glucose, and 10 HEPES (pH 7.4). The pipette solution contained (in mmol/L) 10 KCl, 133 potassium gluconate, 5 EGTA, 5 glucose, 1 K 2 ATP, 0.5 NaADP, and 10 HEPES (pH 7.4), in which the free Mg 2ϩ concentration was adjusted to 1 mmol/L using MgCl 2 . PMA and other chemicals were purchased from Sigma if not otherwise stated. Chemicals were prepared in high concentration stock solution in doubledistilled H 2 O or Me 2 SO and were diluted in bath solution to experimental concentrations immediately before usage. Glibenclamide, pinacidil, and PMA were applied to cells using a perfusion system. PKC inhibitory peptide 19 -31 (PKCi) (10 mol/L; Calbiochem) was applied to the pipette solution. To avoid ATP degradation, all ATP-containing solutions were made immediately before experiments and used for no longer than 4 h. Inside-out patches were performed with symmetric high K ϩ in bath solution and pipette solution (in mmol/L): 10 KCl, 135 potassium gluconate, 5 EGTA, 5 glucose, and 10 HEPES (pH 7.4), with [Mg 2ϩ ] adjusted to 1 mmol/L using MgCl 2 . After a giga-seal was formed, the patch was excised, and the intracellular side was exposed to bath solution. Constant single voltages of Ϫ60 mV were applied to the patch from a holding potential of 0 mV. PMA inhibited the WT and some mutant channels gradually with a plateau reached in 4 -5 min of exposure. Therefore, the effect of PMA on the current amplitude was measured at 5 min and expressed as % current inhibition, i.e. the plateau level of current inhibition during PMA exposure divided by the difference of the maximum channel inhibition by glibenclamide from the maximum activation by pinacidil.
Western Blotting-MBP-Kir6.1C constructs with or without mutants were generated with the distal C-terminal segment (residues 346 -420) of Kir6.1 fused to the C terminus of MBP using the same methods detailed in our previous report (5). The constructs were then transformed into protease-deficient Escherichia coli BL21, in which MBP fusion peptides were induced with 0.3 mmol/L isopropyl thiogalactoside for 2 h. The MBP fusion peptides were purified using amylose resin according to the procedure of the provider (New England Biolabs). The peptides were run on 10% SDS-polyacrylamide gel and transferred to a polyvinylidene difluoride membrane (Bio-Rad). Western blotting was performed using rabbit primary antibody against the Kir6.1 C terminus (KRNSMRRNNSMRRSN, corresponding to amino acid residues 382-396 of rat Kir6.1; 1:2000; Sigma) and secondary goat anti-rabbit IgG conjugated with alkaline phosphatase. The protein bands were depicted with bromochloroindolyl phosphate and nitro blue tetrazolium.
In Vitro Phosphorylation-The fusion peptides were treated with the catalytic fragment of PKC (cPKC) (BIOMOL, Plymouth, PA) under the following condition: ϳ5 g of fusion peptides in 5 l of modified elution buffer (in mmol/L) (200 NaCl, 30 Tris-HCl, 6 EDTA, 10 maltose, pH 7.4), 5 l of 5ϫ reaction buffer (125 Tris-HCl, 0.1 EGTA, pH 7.5), 5 l of MgATP solu-tion (20 MOPS, 25 ␤-glycerophosphate, 5 EGTA, 1 Na 3 VO4, 1 dithiothreitol, 75 MgCl 2 , 0.5 ATP, pH 7.2), 10 ng of cPKC in 10 l of H 2 O, and 1 l of 5 Ci/l [␥-32 P]ATP (PerkinElmer Life Sciences). After 60 min of reaction, 5 l of 5ϫ protein loading buffer were added to each sample to terminate the reaction. The samples were subjected to electrophoresis in 10% SDSpolyacrylamide gel. The gel was stained with Coomassie Blue and photographed. The gel was then fixed and dried. Autoradiography was performed using a Fuji BAS 2500 imaging plate. The in vitro phosphorylation experiment was repeated twice.
Data are presented as the mean Ϯ S.E. Differences in means were tested by an analysis of variance or Student's t test and were accepted as significant if p Յ 0.05.
In inside-out patches, channel activity was low in the absence of nucleotides. The Kir6.1/SUR2B channel was activated in the presence of MgADP/ATP, which was further augmented with pinacidil application ( Fig. 2A). When cPKC was applied to the internal patch membranes, the Kir6.1/SUR2B currents were markedly inhibited (61.1 Ϯ 3.8%, n ϭ 6) (Fig. 2, A and C). Thus, these results indicate that the Kir6.1/SUR2B channel is inhibited by PKC independently of cytosolic soluble components, consistent with previous reports (20). In contrast, PMA and cPKC had no effect on the Kir6.2/SUR2B channel in whole-cell recordings ( Fig. 1, D-F) or in inside-out patches (Fig. 2, B and C), suggesting that the SUR2B subunit is not critical.
Critical Protein Domains for PKC-dependent Channel Inhibition-The differential PKC sensitivity of Kir6.1/SUR2B compared with Kir6.2/SUR2B suggests that critical protein domains for PKC regulation are located on the Kir subunit. Therefore, we constructed Kir6.1-Kir6.2 chimeras and expressed them with SUR2B in HEK293 cells. Kir6.1 and Kir6.2 were divided into three regions: the N terminus, the core sequence containing two transmembrane domains and the pore loop, and the C terminus (Fig. 3). Six chimeras were thus obtained, all of which showed functional channels and were sensitive to pinacidil and glibenclamide. When the N terminus of Kir6.1 was replaced with that of Kir6.2 (named 211), channel inhibition by PMA was greatly diminished (Fig. 3, A and D). A similar result was observed in the 112 channel, whose C terminus was from Kir6.2 (Fig. 3, B and D). Consistent with this, replacement of either the N or C terminus of Kir6.2 (122 and 221, respectively) did not confer PMA sensitivity to the chimeric channels (Fig. 3D). When both the N and C termini of Kir6.1 were constructed, the 121 channel responded to PMA in the same way as Kir6.1 (Fig. 3, C and D). Therefore, both the N and C termini are required for PKC-dependent channel inhibition.
We further divided the N terminus into two segments and the C terminus into three. These chimeras showed similar responses to pinacidil and glibenclamide as the WT channels. When the distal N terminus was replaced, the Kir6.1-21_111 channel responded to PMA to the same extent as Kir6.1 (71.1 Ϯ 4.2%, n ϭ 5) (Fig. 4A). However, replacement of the proximal N terminus significantly decreased PMA effects, although such an effect was somehow incomplete (36.7 Ϯ 7.8%, n ϭ 5) (Fig. 4B). A serine residue (Ser-40) was found in the proximal N terminus of Kir6.1 but not in Kir6.2. When Ser-40 was mutated to the corresponding residue (Lys-39) in Kir6.2, the Kir6.1-S40K channel was still inhibited in the same way as the WT Kir6.1 channel (Fig. 5E). Because no PKC site was found in the N terminus, how this protein domain is involved in the PKC action remains to be understood. When the proximal C terminus was replaced, Kir6.1-11_211 responded to PMA to the same degree as the WT channel (Fig. 4C). With the middle or distal segment swapped, the Kir6.1-11_121 and Kir6.1-11_112 channel responses to PMA were significantly diminished (Fig. 4, D and  E). PMA inhibition was further decreased in a chimera in which both middle and distal segments were from Kir6.2 (Kir6.1-11_122) (Fig. 4F).
Phosphorylation of a Stretch of Serine Repeats in the C Terminus by PKC-There are multiple consensus PKC sequences in the distal C terminus from residue 346 to the end (residue 424). Particularly remarkable are the SXRRXN/ SXRKXN repeats at residues 379, 385, 391, and 397. We thus performed alanine mutational screening. Mutation of any individual residues produced moderate but significant reduction of the PKCdependent inhibition (Fig. 5, B and  E). Mutation of Ser-354 in the same protein domain showed a similar effect (Fig. 5E). A combined mutation of 3 or 4 (Ser-379, Ser-385, Ser-391, Ser-397) had greater effects (Kir6.1-3A and Kir6.1-4A) (Fig. 5, C and E). When all of these 5 serine residues were jointly mutated (Kir6.1-5A), the PKC-dependent channel inhibition was almost completely eliminated (13.3 Ϯ 4.5%, n ϭ 5; note that there were ϳ10% desensitizations contributing to the remaining effect) (Fig. 5, D and E).
To confirm that these serine residues were indeed phosphorylated, we generated several C-terminal peptides of Kir6.1 (residues 346 -424) with and without mutations of the serine residues. After extraction and purification, the peptides were subjected to SDS-PAGE separation, Western blotting, and in vitro phosphorylation. In the Coomassie Blue stain and Western blot, a 52-kDaband was detected, which was consistent with the peptide size (Fig. 6B), whereas there was another 49-kDa band indicating a degraded protein fragment. An in vitro phosphorylation experiment showed strong 32 P incorporation in both bands of the WT peptide (Fig. 6C), consistent with our functional assays. The strength of 32 P incorporation was substantially reduced in the Kir6.1-3A peptide. The Kir6.1-4A and Kir6.1-5A peptides were only weakly phosphorylated. Comparing the phosphorylation patterns, we found that Ser-354 was not phosphorylated (see Fig. 6 legend). The 32 P incorporation in the Kir6.1-4A and Kir6.1-5A peptides was almost completely eliminated in comparison with the WT, consistent with our mutational analysis in this region (S403A, S404A, and T414A) (Fig. 5E), suggesting that residues other than Ser-379, Ser-385, Ser-391, and Ser-397 have very little effect, if any, on PKC-dependent channel inhibition.

DISCUSSION
The present studies show evidence for the molecular basis underlying Kir6.1/SUR2B channel inhibition by PKC. The channel inhibition relies on a short motif consisting of four phosphorylation repeats in the Kir6.1 subunit. Graded channel inhibitions are produced by phosphorylation of different numbers of these serine residues. The presence of the phosphorylation motif in Kir6.1, but not in its close relative Kir6.2, suggests that the vascular K ATP channel may have undergone evolutionary optimization allowing it to be regulated by a variety of vasoconstrictors.
PKC acts on several isoforms of K ATP channels. Previous studies indicate that PKC activation leads to inhibition of the vascular Kir6.1/SUR2B channel, and this effect is likely mediated by direct phosphorylation of the channel protein (16,20). PKC activates the pancreatic isoform (Kir6.2/SUR1) and striated muscular isoform (Kir6.2/SUR2A) of K ATP channels (22).
Expressed with SUR2B, the Kir6.2 channel is activated by PKC or shows no response, depending on recording conditions (16,20). A critical PKC phosphorylation site (Thr-180) has been found in Kir6.2 mediating the fast channel activation (22). However, it is unclear whether a corresponding site in Kir6.1 (Thr-190) plays a similar role, as mutation of the residue leads to a nonfunctional channel (5,20). PKC also causes Kir6.2/ SUR1 and Kir6.2/SUR2A channel internalization (23). Because the effect of PKC on Kir6.1/SUR2B (inhibition) is clearly in contrast to that on Kir6.2-containing channels (activation or no effect), the Kir subunit is likely to be targeted by PKC regulation. The differential responses of Kir6. 1    Kir6.1 N terminus, core, and C terminus refer to residues 1-71, 72-186, and 187-424, respectively. Kir6.2 N terminus, core, and C terminus refer to residues 1-70, 71-176, and 177-390, respectively. and C termini are important for the sensitivities (24 -27). A notable difference between Kir6.1 and Kir6.2 channels is that in the absence of nucleotides the Kir6.2 channels are spontaneously open, whereas the Kir6.1 channels are completely closed (28), a phenomenon that is also explained as fast rundown of the Kir6.1 channels (29). Using Kir6.1-Kir6.2 chimeras, Kondo et al. (28) found that both the N and C termini are important for this difference, whereas the N terminus seems to be more critical; 5 residues in the proximal N terminus are found to play a major role in the difference. The Kir6.2 channels are also activated by intracellular H ϩ (30). The pH-dependent channel gating requires both N and C termini, although there is only one protonation site (His-175) located in the C terminus (21,31). The importance of the proximal N terminus in channel gating has been studied with different ligands, indicating that this region is indeed involved in Kir6.2 channel gating rather than ligand binding (32,33). In addition to Kir6.2, the Kir2.3 channel is gated by acidic pH in an N terminus-dependent manner, and there is no protonation site in the N terminus (34). The requirement of the N terminus for PKC-dependent channel inhibition found in our current studies is thus consistent with these previous reports, suggesting that this protein domain is likely to be involved in interactions with other protein domain(s) in channel gating.
The distal C terminus (residues 346 -424) of Kir6.1 is critical for PKC-dependent channel inhibition. This protein domain is a direct target of PKC, as multiple PKC phosphorylation sites are found in this narrow region. Our results indicate that there are at least four phosphorylation sites in this region, i.e. residues 379, 385, 391, and 397. Our patch clamp study suggests that Ser-354 is another, a result that is not supported by our in vitro phosphorylation assay. Therefore it is unclear whether Ser-354 can be phosphorylated in vivo. Of them, Ser-397 in human Kir6.1 has been previously reported to be phosphorylated by an unknown kinase in a global phosphorylation screening study (35). None of the rest has been studied previously. Interestingly, the effects of phosphorylation of these residues are additive or cumulative. Mutation of each individual residue reduces PKC effects by ϳ20%. Combined mutations of multiple residues have greater effects. Channel inhibition is almost completely abolished with mutations of all five (including Ser-354). Multiple phosphorylation sites often result in sequential phosphorylation, which does not seem to be the case in Kir6.1/SUR2B, as our results suggest that none of the residues shows a dominant effect over others. Of the five putative phosphorylation sites, 4 serine residues appear in a clear pattern of repeats (SXRRXN/SXRKXN). To our knowledge, this is the first demonstration regarding such a phosphorylation motif in ion channels.
The distal C termini of inward rectifier K ϩ channels play a critical role in channel trafficking and thus control the number of functional channels on the cell membrane (36 -39). There is an endoplasmic reticulum retention sequence (RKR, residues 381-383 in Kir6.1 and 369 -371 in Kir6.2) (40), which warrants the Kir6 subunit targeting the membrane together with the SUR subunit. A dileucine motif (Leu-355/Leu-356) in Kir6.2 is involved in channel endocytosis (23). Our results indicate that the distal C terminus of Kir6.1 is also involved in channel activity control (gating). In the narrow C-terminal region where the last four PKC sites are found, there are 9 alkaline residues, including the RKR endoplasmic reticulum retention signal, making this region extremely positively charged. The positive charges appear to favor the open state in the presence of pinacidil or other channel openers. These charges are largely neutralized following phosphorylation of the serine repeats, which may attenuate the channel opening.
It is noteworthy that although we used PMA and pinacidil to optimize the experimental condition, similar PKC-dependent channel inhibition has been shown to play a role in the channel modulation by vasopressin in our previous studies (6). Consistently, endothelin (15), muscarinic M3 receptor agonist (16), and angiotensin II (17) also inhibit the vascular K ATP channels in a PKC-dependent manner. Because some of these previous studies were performed in mesenteric arteries and dissociated vascular smooth muscles cells, the channel modulation mechanism demonstrated in the present study is of physiological significance.
Why does the PKC phosphorylation motif exist in Kir6.1 but not in Kir6.2? The phosphorylation motif renders 16 phosphorylation sites in a K ATP channel with four Kir6.1 subunits. This, as well as the cumulative nature of each phosphorylation, may allow the channel to be elaborately modulated according to the levels of PKC activation, which is consistent with the functional needs of the vascular K ATP channel targeted by a variety of vasoconstricting hormones and neurotransmitters. The presence of the phosphorylation motif in Kir6.1, but not in its close  1-3A must occur at Ser-379. That the 49-kDa band of Kir6.1-4A was not phosphorylated also suggested that Ser-354 was not phosphorylated. In addition, phosphorylation on the 52-kDa band of Kir6.1-4A and Kir6.1-5A suggests that there are unidentified phosphorylation site(s) located C-terminal to Ser-397, but they were not functionally important according to mutational analysis in Fig. 5E. relative Kir6.2, suggests that the vascular K ATP channel may have undergone evolutionary optimization serving for the vascular regulation under various physiological and pathophysiological conditions.