Characteristic Interactions with Phosphatidylinositol 4,5-Bisphosphate Determine Regulation of Kir Channels by Diverse Modulators*

The activity of specific inwardly rectifying potassium (Kir) channels is regulated by any of a number of different modulators, such as protein kinase C, Gq -coupled receptor stimulation, pH, intracellular Mg2+ or the βγ-subunits of G proteins. Phosphatidylinositol 4,5-bisphosphate (PIP2) is an essential factor for maintenance of the activity of all Kir channels. Here, we demonstrate that the strength of channel-PIP2 interactions determines the sensitivity of Kir channels to regulation by the various modulators. Furthermore, our results suggest that differences among Kir channels in their specific regulation by a given modulator may reflect differences in their apparent affinity of interactions with PIP2.

Inwardly rectifying potassium (Kir) 1 channels comprise a superfamily composed of seven subfamilies (Kir 1-7) containing at least 15 members in mammals. Kir channels are expressed in many tissues and serve important roles in cellular physiology such as cell excitability, K ϩ homeostasis, and insulin secretion. The activity of Kir channels is regulated by many intracellular factors and second messengers (1). Examples of the physiological importance of Kir regulation include: the ATP dependence of Kir6.X (K ATP ) channels in the control of insulin secretion (2,3) and the determination of myocardial resistance to hypoxia (4,5); the regulation of Kir3.X (GIRK) K ϩ channels by G-proteins that accounts for the vagal control of heart rate (e.g. Ref. 6); the regulation of Kir1.X (ROMK) channels by K ϩ and pH, which controls K ϩ secretion in the kidney (7), and the regulation of Kir4.1 channels in glial cells (8) and the inner ear (9), controlling K ϩ homeostasis and excitability.
A common feature of Kir channels that has emerged recently is that they all require the membrane phospholipid, PIP 2 to maintain their activity (6, 10 -13). PIP 2 , although only a minor component of the plasma membrane phospholipids, plays an important role in cell signaling; PIP 2 serves as the substrate for PLC (phospholipase C)-mediated hydrolysis into the two ubiquitous second messengers diacylglycerol and inositol triphosphate. Recent studies have suggested that PIP 2 directly mediates receptor-induced modulation of Kir channel activity (14,15). Defects in channel-PIP 2 interactions appear to underlie Andersen's and Bartter's syndrome (13).
The molecular mechanism by which Kir channels are regulated as a consequence of stimulation of different PLC isoforms, such as PLC␤ coupled to receptor signaling through G q -or PLC␥-coupled to tyrosine kinase receptors, remains unsettled. Kir2.3, Kir3.X, and Kir 6.X channels have been shown to be inhibited through activation of PKC (16 -20), directly by hydrolysis of PIP 2 (14,15,(21)(22)(23), or even directly by Mg 2ϩ possibly acting as a second messenger (24). pH regulation of Kir channels has been extensively studied. It is generally believed that intracellular pH inhibits Kir channels through protonation of certain amino acid residues within the cytosolic channel domains (25)(26)(27). However, a mechanism other than mere protonation is needed to explain pH inhibition of Kir channels (26,28). Recent studies indicate that PIP 2 also plays an important role in the pH-induced inhibition of Kir channels (29 -31).
In the present study, we aimed to explore the hypothesis that channel-PIP 2 interactions determine modulation of Kir channel activity by multiple regulatory factors. We predominantly used two members of the Kir2 family, Kir2.1 (IRK1) and Kir2.3 (IRK3), and their mutants to show that the strength of channel-PIP 2 interactions determines the degree of inhibition of these Kir2 channels by receptor (M1 and EGFR) stimulation, PKC activation (via PMA treatment), intracellular Mg 2ϩ and pH. Our results strongly suggest that the apparent affinity of channel-PIP 2 interactions is a key element in determining modulation of Kir channels.

EXPERIMENTAL PROCEDURES
Molecular Biology-All cDNA constructs were subcloned into the pGEMHE plasmid vector and used as described. Point mutants were produced by Pfu mutagenesis with a QuikChange kit (Stratagene, La Jolla, CA). Sequences were confirmed by DNA sequencing. Recombinant Kir channels, M1 and EGF receptors were expressed in Xenopus laevis oocytes as previously described (35). cRNA was produced with T7 RNA polymerase using a kit (Ambion, Austin, TX). cRNA of the various Kir channels and their mutants and of receptors was injected in the range of 0.5-10 ng/oocyte depending on the functional expression level of the given construct. * This work was supported in part by National Institutes of Health Grant HL59949 (to D. E. L.). 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.
Electrophysiology-Recordings in X. laevis oocytes were performed 2-4 days following cRNA injection.
Whole oocyte currents were measured by conventional two-microelectrode voltage clamp with a GeneClamp 500 amplifier (Axon Instruments). Electrodes were filled with 3 M KCl dissolved in 1% agarose to prevent the leakage of KCl into the oocytes. The electrodes had a resistance less than 1 M⍀. Oocytes were constantly perfused with a high potassium solution (ND96K) containing (in mM) 96 KCl, 1 NaCl, 1.8 CaCl 2 , 1 MgCl 2 , 5 HEPES (pH 7.4). A low potassium solution (ND96) was used in some experiments to inhibit most of the Kir currents at Ϫ80 mV. ND96 contained (in mM) 96 NaCl, 1 KCl, 1.8 CaCl 2 , 1 MgCl 2 , 5 HEPES (pH 7.4). Current amplitudes were measured at ϩ80 mV and Ϫ80 mV. Data acquisition and analysis were carried out using pClamp8 (Axon Instruments) and Origin 6 (Microcal) software.
Chemicals-PIP 2 was purchased from Roche Applied Science and Roche Diagnostics; DiC 8 PIP 2 was purchased from Echelon. PIP 2 antibody was purchased from Assay Technologies (Ann Arbor, MI). PIP 2 and PIP 2 antibody were prepared as described previously (12,13). All other chemicals were purchased from Sigma.
Error bars in the figures represent S.E. Each experiment shown or described was performed on 3-5 oocytes of the same batch. A minimum of 2-3 batches of oocytes was tested for each experiment shown. The unpaired t test was used to assess statistical significance in all cases where indicated.  (24), and pH (28), even though they belong to the same subfamily of Kir channels.

Activation of M1 and EGF Receptors Inhibits
When expressed in Xenopus oocytes, both Kir2.1 and Kir2.3 channels gave substantial inwardly rectifying currents. Fig. 1, A and B, shows the current traces recorded by two-electrode voltage clamp at ϩ80 (above the dotted zero current line) and Ϫ80 mV (below the dotted zero current line) in a high extracellular K ϩ bath solution (ND96K). When M1 muscarinic receptors were coexpressed with these Kir channels and were stimulated by acetylcholine (ACh, 5 M), Kir2.3 currents were consistently inhibited, whereas Kir2.1 currents were not (Fig.  1, A-C). Transient Ca 2ϩ -activated Cl Ϫ currents seen at ϩ80 mV were indicative of PIP 2 hydrolysis by PLC, which is known to be coupled to M1 receptor signaling (e.g. Ref. 14). Full recovery of Kir2.3 currents from inhibition could be obtained following long term washout of ACh (data not shown).
We investigated whether PLC-mediated hydrolysis of PIP 2 through a G protein-independent pathway would also result in Kir2.3 inhibition. In addition, we tested whether G protein-dependent and -independent pathways of PIP 2 hydrolysis were occlusive of each other in inhibiting K ϩ currents. The epidermal growth factor receptor (EGFR) hydrolyzes PIP 2 via direct coupling to PLC␥. Kir2.3 channel, M1, and EGF receptors were coexpressed in oocytes and ACh (5 M) and EGF (100 ng/ml) were applied through the bath solution. EGF induced a similar inhibition of Kir2.3 currents as ACh, and the inhibition of Kir2.3 by ACh or EGF was occlusive for the other agent ( Fig. 1, D-F). EGF has been shown not to induce a significant inhibition of Kir2.1 currents (14). Thus, two independent pathways that hydrolyze PIP 2 showed similar effects on Kir2.3 and Kir2.1 channels. Moreover, each pathway occluded the effect of the other in inhibiting Kir2.3 currents.
ACh-and EGF-induced Inhibition of Kir Currents Shows a Strong Correlation to the Apparent Affinity of Interaction with PIP 2 -We have previously identified mutations that alter channel-PIP 2 interactions (e.g. Ref. 12 and 13). We first used PIP 2 antibody to functionally assess channel-PIP 2 interactions (10,12). Fig. 2A shows the current traces recorded from inside-out macropatches held at Ϫ80 mV. PIP 2 antibody inhibited Kir2.3 significantly faster than Kir2.1 currents, suggesting that Kir2.1 channels interact tighter with PIP 2 than Kir2.3 channels do. We previously identified an amino acid difference between Kir3.4* and Kir2.1 channels, which is critical for interactions with PIP 2 (12). Mutation of isoleucine in Kir3.4*(Ile 229 ) to the corresponding Kir2.1 leucine (Leu 222 ) strengthened interactions of Kir3.4* with PIP 2 . Kir2.3 channels also have an isoleucine (Ile 213 ) in the position corresponding to GIRK4 (Ile 229 ). We made the point mutation of Kir2.3 (I213L) and tested the kinetics of inhibition by PIP 2 antibody. As shown in Fig. 2A, Kir2.3 (I213L) significantly slowed the time course of inhibition by PIP 2 antibody. In addition, the point mutant Kir2.1(R312Q) was inhibited by PIP 2 antibody with much faster kinetics than control ( Fig.  2A), consistent with previous results (13). Summary data in Fig. 2B show the time needed for PIP 2 antibody to inhibit half the channel currents.
Thus, Kir2.3(I213L) behaved similarly to Kir3.4(I229L) in showing slower kinetics of PIP 2 antibody inhibition of these mutant currents compared with control. In fact, there is good correlation of the residue type at this position ( Fig. 2C, highlighted) with the strength of interaction of PIP 2 with Kir channels: channels possessing a Leu residue generally show stronger interactions with PIP 2 than channels possessing an Ile residue at this position (see Fig. 4C and "Discussion"). We proceeded to test whether this mutation caused an increase in the apparent affinity of Kir2.3 to PIP 2 . We used the 8-carbon long acyl chains on PIP 2 (DiC 8 ), a synthetic water soluble analog of PIP 2 , to construct dose-response curves of PIP 2 on channel activity. Shifts in the PIP 2 dose-response curves for Kir currents with specific mutations affecting channel-PIP 2 interactions have previously suggested changes in the channel-PIP 2 apparent affinity (13). Consistent with the PIP 2 antibody (Fig. 2D), the relative apparent affinities of these channels for PIP 2 were in the same order (Kir2.1: . Given these results we will use interchangeably the terms "strength" of channel-PIP 2 interactions or "apparent affinity" of a channel to PIP 2 . We next proceeded to test whether these mutations that altered channel-PIP 2 interaction also affected modulation of Kir2.3 and Kir 2.1 currents by M1 and EGFR stimulation. Fig.  2, E and F, shows that Kir2.3(I213L) was inhibited significantly less than Kir2.3 by either M1 or EGFR stimulation, whereas Kir2.1(R312Q) showed greater inhibition than Kir2.1 by M1 and EGFR stimulation. Therefore, we conclude that the extent of agonist-induced current inhibition of these Kir channels is controlled by the apparent affinity of interactions between the channels and PIP 2 .

PMA-and ACh-induced Inhibitions of Kir Currents Show a
Strong Correlation with the Apparent Affinity of a Channel to PIP 2 -Previous studies have shown that Kir2.3 channels are inhibited by PMA through PKC-mediated phosphorylation (32,33). In contrast, Kir2.1 channels are generally not modulated by PKC activation. We performed similar experiments to those described above using PMA-modulated Kir channels expressed in oocytes. When PMA (100 nM) was applied to oocytes expressing Kir2.3 channels, currents were inhibited irreversibly; the inhibition started with a delay of ϳ2 min and reached a plateau 4 -5 min following PMA application (Fig. 3A); Kir2.3 currents were inhibited to about 50% of their original amplitude (Fig.  3E) and no further inhibition was observed after PMA concentration was increased from 100 nM to 1 M; the inactive form of PMA, 4-␣-PMA, had no effect on Kir2.3 currents (data not shown); a specific PKC blocker, bisindoylmaleimide, prevented PMA-induced inhibition of Kir2.3 currents (Fig. 3E). These results suggest that PMA inhibited Kir2.3 currents through a PKC-mediated mechanism. In contrast, PMA did not inhibit significantly Kir2.1 currents (Fig. 3, B and E). Kir2.3(I213L), which increased the apparent affinity of interactions with PIP 2 , completely abolished PMA-induced inhibition (Fig. 3, C and E). In contrast, Kir2.1(R312Q), which showed decreased affinity of interactions with PIP 2 , did exhibit inhibition by PMA (Fig. 3, D and E). Thus, apparent affinity of channel-PIP 2 interactions determines modulation of these Kir channels by PMA.
To further examine the generality of our conclusion, we tested for ACh-and PMA-induced inhibition in the presence and absence of G␤␥. Activation of Kir3.4* channels by G␤␥ (10,12) or intracellular Na ϩ (6, 12, 34) constitutes a physiological way by which channel-PIP 2 interactions can be strengthened. Fig. 3 also shows that basal Kir3.4 currents could be inhibited by either ACh or PMA treatment (Fig. 3F). Coexpression of Kir3.4* channels with G␤␥ greatly reduced the ACh-or PMAinduced current inhibition (Fig. 3G). These results further support the conclusion that the strength of channel-PIP 2 interactions determines modulation of Kir channels by ACh and PMA. We next examined the ACh- (Fig. 4A) and PMA-induced (Fig.  4B) inhibition of most Kir channel family members. We correlated these effects to the strength of interaction with PIP 2 exhibited by each of these channels (Fig. 4C) (35,36). Heteromeric expression often also enhances channel targeting to the cell surface and yields larger currents than the corresponding homomeric ones (e.g. Kir3.1/3.4 and Kir4.1/5.1). We studied 11 homomeric and two heteromeric channels (Fig. 4). Based on our results, we ranked strength of channel-PIP 2 interactions relative to the strongest interacting channel Kir4.1. This analysis indicated the following rank order among Kir channel subfamilies (from the strongest to the weakest 4C). Both the ACh-and PMA-induced effects showed an almost perfect inverse correlation to the strength of channel-PIP 2 interactions (Fig. 4, A and B). The only exception seemed to be Kir2.2, which displayed relatively strong channel-PIP 2 interactions with disproportionately large ACh and PMA current inhibition. These results suggest that the strength of channel-PIP 2 interactions determines modulation of Kir channels through agonist-and PMA-induced effects.
Enhanced Channel-PIP 2 Interaction and PIP 2 Prevent Intracellular Mg 2ϩ Inhibition of Kir2.3 Channels-Intracellular Mg 2ϩ was shown previously to inhibit Kir2.3 channel currents, and was suggested to act as a second messenger accounting directly for the M1-induced inhibition of Kir2.3 channels (24).
In the present study we tested whether the Mg 2ϩ -induced inhibition of Kir2.3 currents was PIP 2 -dependent. Data shown in Fig. 5 were obtained from inside-out macropatches. Kir2.3 as well as Kir2.1 channels generally gave currents in the nanoampere range when recorded in the cell-attached mode at Ϫ80 mV (e.g. Fig. 5A). When an inside-out patch was excised into a Mg 2ϩ -free high K ϩ solution (ND96K), Kir2.3 currents ran up first, reaching a plateau that was maintained for a short time before starting to run down. Kir2.3 currents were more stable in a FVPP solution, which inhibits phosphatases and thus prevents breakdown of PIP 2 (10). Under such conditions, 1 mM Mg 2ϩ significantly accelerated rundown of Kir2.3 currents when applied to the inner surface of inside-out patches. Only a small portion of the rundown currents could be recovered after Mg 2ϩ was washed away (Fig. 5A). However, the time course of Kir2.3 current rundown was variable from patch to patch, possibly reflecting variable phosphatase activities. PIP 2 (10 M), applied to the inner surface of patches, could activate Kir2.3 currents after they completely ran down; normally long PIP 2 applications were required to reactivate Kir2.3 channels, possibly because of the slow diffusion of exogenous PIP 2 into the membranes and binding to the channels. Once Kir2.3 channels were reactivated by PIP 2 , Mg 2ϩ could not induce current rundown any more. Instead, Mg 2ϩ application following reactivation by PIP 2 induced a nonspecific reversible inhibition. Kir2.3(I213L) currents were less sensitive to Mg 2ϩ inhibition. rents after they were reactivated by PIP 2 . In previous work (24), Kir2.1 channels were assessed to be Mg 2ϩ -insensitive. In our experiments, we found Mg 2ϩ able to induce Kir2.1 current rundown, albeit with much slower kinetics (Fig. 5C). PIP 2 reactivated Kir2.1 currents following run-down with fast kinetics, and Mg 2ϩ induced a small, reversible inhibition of the currents that was comparable with the inhibition of Kir2.3 or Kir2.3(I213L) currents following PIP 2 reactivation. Fig. 5D shows that the Mg 2ϩ -induced rundown was much faster for Kir2.1(R312Q) than for Kir2.1 wild type. After rundown, PIP 2 reactivated Kir2.1(R312Q) slower than Kir2.1 currents, suggesting a weaker binding affinity for Kir2.1(R312Q) than for Kir2.1. Once again, Mg 2ϩ induced only a PIP 2 -independent inhibition following reactivation by PIP 2 . Fig. 5E shows summary data of the inhibition of Kir2.3, Kir2.3(I213L), Kir2.1, and Kir2.1(R312Q) by Mg 2ϩ before and after application of PIP 2 . The data are quantified 1 min following application of Mg 2ϩ .
We next proceeded to use an ATP-containing solution to test if ATP, in the presence of which endogenous synthesis of PIP 2 is possible, would reduce Mg 2ϩ -induced inhibition of the Kir currents in a manner similar to that obtained following PIP 2 application. Inside-out macropatches expressing Kir2.3, Kir2.3(I213L), Kir2.1, and Kir2.1(R312Q) channels were excised into 1 mM MgATP solution and 1 mM Mg 2ϩ was applied to the patches for a short time (1 min, Fig. 6, A-D). In the presence of MgATP, Mg 2ϩ had a small inhibitory effect on Kir2.3(I213L), Kir2.1, and Kir2.1(R312Q) currents (Fig. 6,  B-D), similar to the inhibition for these channels shown in Fig.  5

Enhanced Channel-PIP 2 Interactions and Presence of PIP 2 Prevent Intracellular Proton Inhibition of Kir Channels-
Kir2.3 channels have been shown to be regulated by intracellular pH, and multiple histidines in the C terminus of the channel have been suggested to serve as the proton action sites (28). We investigated whether pH regulation of Kir2.3 channels is PIP 2 -dependent. Both whole cell and inside-out macropatch recordings were used in the experiments. In whole cell recordings, we used azide (sodium salt) and KHCO 3 to lower intracellular pH. Azide is a metabolic inhibitor and has been used to activate K ATP channels expressed in Xenopus oocytes by lowering internal ATP concentration (37). When oocytes expressing Kir2.3 channels were treated with 3 mM azide, a fast and reversible inhibition of Kir2.3 currents was seen (Fig. 7A). We considered this inhibition to be the result of lowered intracellular pH, because: 1) metabolic inhibitors are known to decrease intracellular pH (38); 2) injection of 20 mM HEPES into the oocytes largely prevented the inhibition (data not shown); 3) Kir2.3 currents were directly inhibited by intracellular acid- ification (see below); and 4) Kir3.4 channels that were activated by intracellular acidification were also activated by azide. 2 Recently, it was reported that metabolic inhibitors including azide inhibited Kir2 (2.2 and 2.3) channels through a pHmediated process (39). We used azide mainly because of its pathophysiological relevance to the conditions like ischemia and hypoxia. KHCO 3 has also been used previously as a modulator of intracellular pH (25,40). As it can be seen in Fig. 7A, when KCl in the bath solution was replaced by KHCO 3 (pH in the solution was adjusted to 7.4), a larger inhibition for Kir2.3 currents was seen than for Kir2.3(I213L); the inhibition was fast and reversible. A low-potassium solution (ND96) was applied at the end of the recording to set the baseline for current measurements. Whereas KHCO 3 induced smaller inhibition of Kir2.3(I213L) currents, azide did not inhibit Kir2.3(I213L) cur-rents at all (Fig. 7B). Azide failed to inhibit Kir2.1 currents or Kir2.1(R312Q) currents; the intracellular pH change induced by azide may be too small to distinguish the differential sensitivity of Kir2.1 and Kir2.1(R312Q) channels to pH (see Fig. 8E, pH 6.8); KHCO 3 induced a larger inhibition on Kir2.3(I213L) currents than on Kir2.1 currents (Fig. 7, C and D). Summary data are shown in Fig. 7E.
We next tested pH sensitivity of Kir2.3, Kir2.3(I213L), Kir2.1, and Kir2.1(R312Q) channel currents in inside-out macropatches. Fig. 8 shows that Kir currents were inhibited by decreasing pH with varying sensitivities, which again correlated well with their strength of interaction with PIP 2 . Fig. 8E shows dose-response relationship curves between Kir channel currents and intracellular pH, which were fitted by a Hill equation . Kir2.3, Kir2.3(I213L), Kir2.1, and Kir2.1(R312Q) were inhibited by decreasing the pH with pK (i.e. the pH value for half-maximal current inhibition) of 6.77, 6.27, 4.93, and 2 X. Du, H. Zhang, and D. E. Logothetis, unpublished data. 5.72, respectively. Hill coefficients were 2.4, 1.9, 1.9, and 2.2 for Kir2.3, Kir2.3(I213L), Kir2.1, and Kir2.1(R312Q) currents, respectively. We then tested if Kir 2.3 currents would become less sensitive to proton inhibition following activation by PIP 2 . Fig.  8F shows such an experiment. In this recording, Kir2.3 currents were inhibited when the bath pH was changed briefly from 7.4 to 6.5; the proportion of the current inhibited at pH 6.5 was increased during current rundown (e.g. the second application of pH 6.5 solution reduced currents to a greater extent than the first application), indicating a dependence on membrane PIP 2 that was presumably lost gradually after patch excision. PIP 2 reactivated the currents that became less and less sensitive to the pH 6.5-induced inhibition during the slow reactivation course. In four patches, pH 6.5 inhibited 22 Ϯ 7.4% of Kir2.3 currents in the presence of PIP 2 , which is significantly (p Ͻ 0.01) less than 79 Ϯ 11% inhibition of the currents in the absence of PIP 2 . These results strongly suggested that internal pH regulated the Kir channels in a PIP 2 -dependent manner. DISCUSSION In the present study, we show that the strength of channel-PIP 2 interactions controls the sensitivity of Kir channels to regulation by receptor stimulation, PKC, intracellular Mg 2ϩ , and pH. Our results suggest that differences in the regulation of Kir channels (e.g. phosphorylation, Mg 2ϩ , and pH binding sites) may reflect differences in the strength of channel-PIP 2 interactions.
Kir Channel-PIP 2 Interaction and Kir Channel Modulation by Receptor or PMA Activation-Our data clearly show that the affinity of channel-PIP 2 interactions controls inhibition of Kir2.1 and Kir2.3 channels by receptor stimulation and PMA. We found that Kir2.3 channels were otherwise inhibited by agonist and PMA stimulation, and became insensitive to these modulators when the affinity of their interaction with PIP 2 was increased by the single point mutation Kir2.3 (I213L); on the other hand, Kir2.1 channels, which are normally not inhibited by agonist and PMA, became sensitive to inhibition when the affinity of their interaction with PIP 2 was decreased by the single point mutation Kir2.1(R312Q). Kir2.3(I213L) is homologous to Kir3.4(I229L), a mutant we identified in previous work and showed that it enhanced the affinity of Kir3.4 channel interactions with PIP 2 (12). Fig. 2C Fig. 4C), and are modulated by receptor or PKC activation (18,20,24,(41)(42)(43) (Fig. 4, A and B); Kir channels (Kir1.1, Kir2.1, Kir 2.4, Kir4.1, and Kir4.2) that have a leucine in this position generally show high affinity interactions with PIP 2 (10, 12) (Fig. 4C), and are not modulated by agonist or PKC activation (24) (Fig. 4, A and B). It is likely that an isoleucine or leucine at this position plays an important role in affecting Kir channel interactions with PIP 2 in all these Kir channels.
One exception to this correlation seems to be Kir7.1 channel, which although has a leucine at this position shows weak affinity interactions with PIP 2 (Fig. 4C). Accumulating evidence supports the hypothesis that channel-PIP 2 interaction are electrostatic in nature, where the negatively charged phosphate groups of the phospholipid interact with positively charged amino acids in the Kir protein (12,13,44,45). The effects of isoleucine or leucine residues ought to be indirect, as these residues are not charged. It is likely that these residues exert their effects by influencing the way neighboring residues interact with PIP 2 . Five positions upstream of the isoleucine or leucine residues there is a positively charged arginine (at the beginning of the sequence shown in Fig. 2B) that is conserved in all Kir channels. Our previous work demonstrated that this arginine is a crucial residue in the interactions of Kir channels with PIP 2 (12,13). When we map these two residues (arginine and isoleucine or leucine) to the structure of a recently crystallized C terminus of the Kir3.1 channel (46), they are co-localized to positions close to the inner membrane (not shown). A leucine could be stabilizing the interaction of the arginine with PIP 2 , whereas an isoleucine could be destabilizing this interaction. Arginine 312 in Kir2.1 is another important PIP 2 -interacting residue previously identified (13). We chose to mutate this residue because even though Kir2.1(R312Q) had a weaker interaction with PIP 2 , it still gave measurable currents to study, whereas mutation of other PIP 2 -interacting residues (e.g. R218Q) often resulted in channels with currents too small to be reliably measured in most batches of oocytes. The mechanism underlying inhibition of Kir channels by receptor (G q -coupled or growth factor) stimulation is still under debate. Cell signaling through receptor activation involves PIP 2 hydrolysis and subsequent activation of its downstream product PKC. Direct PIP 2 hydrolysis has been implicated in Kir channel inhibition (14,15,(21)(22)(23). However, a PKC-mediated mechanism has also been suggested (16 -19), because receptormediated inhibition can be prevented by PKC inhibitors and mimicked by PMA that is known to activate PKC. A few attempts have been made to identify potential phosphorylation sites of PKC on Kir channels. Whereas most studies have failed to identify any PKC phosphorylation sites (24, 32, 42), one study has reported that a single residue, threonine at the N terminus of Kir2.3 (Thr 53 ) is responsible for PMA-mediated inhibition; mutation of Kir 2.3(Thr 53 ) to the corresponding residue of Kir2.1 (Kir 2.3(T53I) completely eliminated the response of the channel to PMA; the reverse mutation conferred Kir2.1(I79T) with sensitivity to PMA almost identical to Kir2.3. This threonine is not a known PKC phosphorylation site and it was not clear how it affected the response of the channel to PMA (33). One possibility is that Thr 53 may behave like Ile 213 that we identified in this study, namely to affect channel-PIP 2 interactions. PIP 2 -interacting sites have been identified in the N terminus of Kir channels (13) where Thr 53 resides. In light of all the findings so far, it seems unlikely that specific PKC phosphorylation site(s) exist only in channels sensitive to PKC modulation; it is more likely that common PKC site(s) exist in most Kir channels, and only channels that have a weaker interaction with PIP 2 would allow PKC-mediated phosphorylation to take place or to be manifested. The assumption here is that the PMA effect seen on Kir channels is indeed through PKC activation (see Ref. 47), and PKC acts on the channel directly, not through some accessory molecules (48). As for the mechanism of receptor-mediated inhibition of Kir channels, our data suggest the following scenario: receptor activation could first hydrolyze PIP 2 reducing the apparent affinity of the channel to PIP 2 ; however, for those channels that have a high apparent affinity to PIP 2 , the remaining interaction is still strong enough to prevent PKC-mediated inhibition; for those channels that have a lower apparent affinity to PIP 2 , the remaining interactions are weakened to a point that PKC action is permitted.
Mg 2ϩ Inhibition of Kir Currents Depends on PIP 2 -It was suggested that receptor (M1)-induced inhibition of Kir2.3 channels proceeded through intracellular Mg 2ϩ acting as a direct second messenger (24). This conclusion was based mainly on the fact that M1 inhibited Kir2.3 currents but not Kir2.1 currents, an effect that could be mimicked by intracellular Mg 2ϩ . Partial exchange of the N and C termini of Kir2.1 channels with those from Kir2.3 conferred the chimeric channel with sensitivity to M1 and Mg 2ϩ inhibition. These results are consistent with our present data. However, we argue that the different sensitivity of Kir2.1 and Kir2.3 channels to intracellular Mg 2ϩ is a reflection of their different interaction properties with PIP 2 . In our experiments, Mg 2ϩ inhibition of Kir channels exhibited two distinct characteristics. First, in the absence of PIP 2 , Mg 2ϩ normally inhibited Kir channels irreversibly. Second, the only difference between Kir2.1 and Kir2.3 channels was that Mg 2ϩ took a longer time to inhibit Kir2.1 currents (Fig. 5, A and C). Increasing the strength of PIP 2 interactions, as with Kir2.3(I213L), slowed the Mg 2ϩ -induced inhibition of K ϩ currents, whereas decreasing the strength of these interactions, as with Kir2.1(R312Q), accelerated the kinetics of block (Fig. 5, A-E). In the presence of PIP 2 , Mg 2ϩ inhibited mutant and wild-type Kir2.1 and Kir2.3 currents reversibly and with a similar potency (Fig. 5, A-D). These data are consistent with the notion that Mg 2ϩ , a known phosphatase activator, induced Kir channel inhibition through enhancing PIP 2 hydrolysis and this effect could be compensated by exogenously applied PIP 2 ; Mg 2ϩ , also a known Kir channel pore blocker (e.g. Ref. 49), rapidly and reversibly blocked Kir channels in the presence of PIP 2 . We also attempted to use a MgATP-containing solution to mimic the physiological conditions, where PIP 2 would be synthesized endogenously. However, MgATP was apparently not sufficient to prevent Kir2.3 currents from continuously running down (Fig. 6A), which made assessment of the Mg 2ϩ action difficult. Nevertheless, it is apparent in the records that Mg 2ϩ induced a smaller inhibition of Kir currents in the presence of MgATP (Fig. 5). Although the possibility that Mg 2ϩ plays a direct role in the M1-induced inhibition of Kir2.3 channels cannot completely be excluded, it is likely that the predominant effect of Mg 2ϩ is mediated by enhancing hydrolysis of PIP 2 .
pH Regulation of Kir Channels Depends on PIP 2 -Our data indicate that intracellular acidic pH inhibits Kir channels in a PIP 2 -dependent manner. In both whole cell and inside-out patch recordings, inhibition of Kir2.3, Kir2.3(I213L) Kir2.1, and Kir2.1(R312Q) currents correlated well with the strength of channel-PIP 2 interactions. Furthermore, in the presence of PIP 2 , Kir2.3 currents became less sensitive to pH inhibition (Fig. 8E). A number of Kir channels have been shown to be regulated by intracellular pH; the Kir channels that are inhib- The dotted line indicates zero current level. E, the current-pH relationship was fitted by a Hill function, I ϭ 1/(1 ϩ (pK/pH) n , where pK is the pH at which the current is 50% inhibited and n is the Hill coefficient. F, pH 6.5 inhibition of Kir2.3 currents followed activation by PIP 2 . pH 6.5 solution was applied multiple times as indicated. The dotted line indicates zero current level. ited by intracellular pH include Kir1.1 channels (25,26,31), Kir2.3 channels (28,39,40), Kir4.1 and Kir5.1 channels (30); Kir 6.1 and Kir6.2 channels, on the other hand, are activated by intracellular proton in a certain concentration range (50). Kir1.1 is the best studied channel with regard to understanding the mechanism of pH inhibition. A lysine (Lys 80 ) was identified in the N terminus of the Kir1.1 channel that serves as a pH sensor (25). However, a lysine residue has a nominal pK a of 10.5, whereas Kir1.1 is inhibited by intracellular pH with a pK a of 6.8. An "Arg-Lys-Arg triad" hypothesis has been proposed to explain this three-order of magnitude discrepancy (26): two arginines in the N (Arg 41 ) and C (Arg 311 ) terminus of Kir1.1 channels form the "triad" with Lys 80 that shifts the pK a of the lysine into the physiological range via electrostatic interactions. It is interesting to note that Arg 311 in Kir1.1 channels is homologous to Arg 312 in Kir2.1, a PIP 2 interaction site (13) (Fig. 3A). More recent studies suggest that PIP 2 plays an important role in the pH inhibition of Kir1.1 channels (29,31); it was reported that the absence of a N-terminal PIP 2 interaction site was a prerequisite for pH inhibition of Kir1.1 channels in the physiological range; introduction of an arginine into this site increased affinity of Kir1.1 channels to PIP 2 and shifted the pK a of pH inhibition of Kir1.1 channels to a much lower value (31). A PIP 2 dependence was also observed for pH inhibition of Kir4.1-Kir5.1 channel currents (30). Thus, our data combined with the above reports suggest that channel-PIP 2 interactions may act like a switch that controls pH inhibition of Kir channels.
The current study presents a general mechanism underlying Kir channel regulation. PIP 2 , a membrane phospholipid that is known to be important in maintaining Kir channel activity, may act as the final checkpoint for regulation of Kir channels by other modulators, which may have action sites on Kir channels common to all or specific to one Kir channel. On the other hand, although it is less likely, the Kir channel modulators we discussed here may also influence Kir channel-PIP 2 interactions and thus regulate Kir channel activities. Two recently published papers show that PKC activation inhibits Kir channel activity in association with a reduction in the membrane PIP 2 level (51, 52), through a yet unknown mechanism. In either case, channel-PIP 2 interactions play a critical role in the modulation of Kir channel activity.