Diverse Kir modulators act in close proximity to residues implicated in phosphoinositide binding


  • This report was presented at The Journal of Physiology Symposium on Regulation of ion channels and transporters by phosphatidylinositol 4,5-bisphosphate (PIP2), Baltimore, MD, USA, 2 March 2007. It was commissioned by the Editorial Board and reflects the views of the author.

Corresponding author D. E. Logothetis: Department of Structural and Chemical Biology, Mount Sinai School of Medicine, New York, NY 10029, USA. Email: diomedes.logothetis@mssm.edu


Inwardly rectifying potassium (Kir) channels were the first shown to be directly activated by phosphoinositides in general and phosphatidylinositol bisphosphate (PIP2) in particular. Atomic resolution structures have been determined for several mammalian and bacterial Kir channels. Basic residues, identified through mutagenesis studies to contribute to the sensitivity of the channel to PIP2, have been mapped onto the three dimensional channel structure and their localization has given rise to a plausible model that can explain channel activation by PIP2. Moreover, mapping onto the three-dimensional channel structure sites involved in the modulation of Kir channel activity by a diverse group of regulatory molecules, revealed a striking proximity to residues implicated in phosphoinositide binding. These observations support the hypothesis that the observed dependence of diverse modulators on channel–PIP2 interactions stems from their localization within distances that can affect PIP2-interacting residues.

In the past decade work from many laboratories has established that phosphoinositides are critical determinants of ion channel activity. Following the pioneering work of Hilgemann & Ball (1996) as well as of Fan & Makielski (1997), numerous reports have implicated phosphoinositides, such as phosphatidylinositol-4,5-bisphosphate (PI-4,5-P2 or PIP2), in the direct control of the activity of just about every type of ion channel (for reviews see Hilgemann et al. 2001; Suh & Hille, 2005). Among the first ion channels whose activity was shown to depend on PIP2 were the inwardly rectifying potassium (Kir) channels (Hilgemann & Ball, 1996; Fan & Makielski, 1997; Huang et al. 1998; Sui et al. 1998; for reviews see Sui et al. 1999; Mark & Herlitze, 2000; Ruppersberg, 2000; Rohacs et al. 2002; Stanfield et al. 2002; Takano & Kuratomi, 2003; Xie et al. 2006).

Kir channels and phosphoinositides

Kir channels belong to a class of K+ channels that regulate such important functions as membrane excitability, heart rate, vascular tone, insulin release and salt flow across epithelia. The K+ flux is greater in the inward compared with the outward direction, a process that has been referred to as anomalous or inward rectification (Katz, 1949). Mg2+ ions and polyamines have been identified to block conduction upon depolarization and thus to reduce outward current (Matsuda et al. 1987; Vandenberg, 1987; Fakler et al. 1994; Lopatin et al. 1994). This mechanism enables cells to conserve intracellular potassium and facilitate entry of potassium ions into the cell. Fifteen Kir channels have been identified and classified in seven subfamilies (Kir1–7) (reviewed by Reimann & Ashcroft, 1999). Kir channels are tetramers of individual subunits, whose secondary structure is characterized by two transmembrane-spanning segments (TM1 and TM2) connected by a pore (P) region and flanked by amino (N-) and carboxy (C-) terminal cytoplasmic domains (e.g. Ho et al. 1993; Dascal et al. 1993; Kubo et al. 1993). Recently, the complete atomic resolution structures of two bacterial Kir channels (KirBac1.1 and KirBac3.1) have been determined by X-ray crystallography (Kuo et al. 2003; J. M. Gulbis, A. Kuo, B. Smith, D. A. Doyle, A. Edwards, C. Arrowsmith et. al., unpublished observations). In addition, 3-dimensional (3-D) X-ray structures of the cytosolic domains of three mammalian Kir channels (Kir2.1, Kir3.1 and Kir3.2) have been solved (Nishida & MacKinnon, 2002; Pegan et al. 2005; Inanobe et al. 2007). All Kir channels have been shown to exhibit distinct affinities and stereospecificities to phosphoinositides (Rohacs et al. 2003; Du et al. 2004). Moreover, it has been demonstrated that regulation of Kir channels by diverse modulators (e.g. phosphorylation, pH, interactions with signalling molecules such as G protein subunits, etc.) shows great dependence on the strength of channel–PIP2 interactions, as if the effects of these modulators are manifested by adjusting channel–PIP2 interactions (Du et al. 2004). Thus, Kir channels that show high affinity for PIP2 are hardly modulated (e.g. Kir2.1), while Kir channels that exhibit moderate (e.g. Kir2.3) to low (e.g. Kir3) affinity for PIP2 are quite sensitive to modulatory inputs. How is it that such diverse modulatory action can all depend on channel–PIP2 interactions? The recent structural advances allow us to localize in the channel three dimensional structure the precise position of residues whose mutation alters the sensitivity of Kir channels to PIP2 as well as the sites utilized by modulatory inputs. Are the channel sites modified by phosphorylation or protons proximal to where the channels interact with PIP2, so that such modifications can alter channel–PIP2 interactions? This review aims to examine the plausibility of such a hypothesis by mapping sites where modulators act relative to sites that affect channel–PIP2 interactions.

Systematic mutagenesis analysis on Kir2.1 (Lopes et al. 2002) and Kir6.2 (Shyng et al. 2000) have identified a number of basic residues, mutation of which alters Kir channel sensitivity to PIP2. In Kir2.1, for example, positioning of these residues (five arginines, four lysines and a histidine) onto the 3-D structure of this channel (Pegan et al. 2005) identifies a pocket where PIP2 could bind. Many of the identified basic residues are conserved among different Kir channels and seem to exert similar effects on PIP2 sensitivity (e.g. Kir2.1 is similar to Kir1.1, see Lopes et al. 2002).

In order to compare positions of sites modified by diverse modulators relative to those of PIP2-sensitive residues, we have mapped the conserved PIP2-sensitive residues we identified in Kir2.1 (Pegan et al. 2005) to the 3-D structure of the highly modulated Kir3.1 channel (Nishida & MacKinnon, 2002; Pegan et al. 2005) shown in Fig. 1. The putative PIP2-binding pocket is formed on one hand by the N-terminus of Kir3.1 and on the other hand by the C-terminus of the channel, which in the 3-D structure comes close to the membrane. The N-terminus is contiguous to a short amphipathic α-helix, the slide helix, which on its hydrophobic surface interacts with the plasma membrane. Although it is not clear yet which of these identified residues are engaged in direct interactions with PIP2, their congregation into a putative pocket of interaction of Kir3.1 with PIP2 is consistent with the picture that emerges from examination of 25 known complexes of proteins with phosphoinositides that show a pocket or a groove in the protein that binds the phosphoinositide (Rosenhouse-Dantsker & Logothetis, 2007).

Figure 1.

Residues affecting PIP2 sensitivity form a pocket that involves both the N- and C-termini of Kir channels
A, the known cytosolic structure of Kir3.1 with a modelled transmembrane domain based on KirBac3.1. The cyan-highlighted residues are conserved basic amino acids sensitive to PIP2 as suggested by functional evidence tested on Kir2.1. B, surface Kir3.1 model with PIP2 sites shown in cyan. Top view of the four (1–4) subunits of the Kir3.1 channel that face the membrane, without the transmembrane domain. The channel is represented as a transparent surface model to emphasize the important surface-accessible residues.

These results and considerations have given rise to a model of gating that places channel–PIP2 interactions as a final step in the transitions between the closed and the open states (Logothetis et al. 2007). If the hypothesis raised in the present review is correct and modulators adjust channel activity by acting at positions that can affect channel–PIP2 interactions, then we will have realized a structural framework within which to seek experimental evidence and test the hypothesis.

Xie et al. (2006) have recently published a detailed review of studies reporting modification of channel–PIP2 interactions by a number of diverse modulators of Kir channels. The present manuscript examines specific examples localizing in 3-D space sites where such diverse modulators act relative to the putative channel–PIP2 interactions.

Cytoplasmic regulatory factors modulate Kir channels by affecting channel–PIP2 interactions

Du et al. (2004) tested the hypothesis that the affinity of channel–PIP2 interactions determines the regulatory outcome of diverse modulators on Kir channel activity. Below, we briefly list specific examples of Kir channel modulation that exhibit dependence on channel–PIP2 interactions. A more detailed discussion of these and other examples can be found in the review by Xie et al. (2006).

In Kir1 channels, studies in Chow-Long Huang's laboratory examined the protein kinase A (PKA) up-regulation of Kir1.1 channel activity and showed that PKA-dependent phosphorylation augmented PIP2 effects in these channels (Liou et al. 1999). Mutation of two PKA phosphorylation sites (S219A and S313A) reduced PIP2 affinity for Kir1.1 channels, mimicking the dephosphorylated state. Other studies in Chow-Long Huang's group have examined the inhibitory effects of PMA on Kir1 channel activity, showing that mutations that decrease channel–PIP2 interactions enhance PMA inhibition (Zeng et al. 2003).

Similar results were obtained for Kir2 channels as well (Du et al. 2004). Du and colleagues utilized predominantly two members of the Kir2 subfamily, Kir2.1 that interacts strongly with PIP2 and Kir2.3 that shows an apparent affinity for PIP2 that is lower than Kir2.1. Additionally, they used two point mutants, Kir2.1(R312Q) that weakened and Kir2.3(I213L) that strengthened channel–PIP2 interactions, and compared them to the corresponding wild-type controls. A robust inhibitory dependence of these Kir2 currents on channel–PIP2 interactions was shown for a number of diverse modulators: (a) by PLCβ (through stimulation of the muscarinic type 1 receptor – M1); (b) by PLCγ (through stimulation of the epidermal growth factor receptor – EGFR); (c) by direct activation of PKC (through stimulation with the phorbol ester, phorbol myristate acetate – PMA); (d) by lipid phosphatases (stimulated by increased levels of intracellular Mg2+); or (e) by protons (applied through changes in intracellular pH). The inhibitory effect correlated inversely with the apparent affinity of the channels for PIP2. Thus, the stronger the channel–PIP2 interactions the less the inhibitory effect.

Although homomeric Kir4.1 is inhibited by severe intracellular acidification, coexpression of Kir4.1 with the inactive Kir5.1 greatly enhanced channel sensitivities to CO2 and pH inhibition. PIP2 modulated Kir4.1/5.1 currents by enhancing the channel open probability and reducing the channel sensitivity to intracellular protons (Yang et al. 2000; Pessia et al. 2001). In contrast, PIP2 did not exert these effects in homomeric Kir4.1 channels that display the strongest interactions with PIP2 compared with all other Kir channels (Du et al. 2004). A more detailed discussion of pH sensitivity in other Kir members (Kir1.1 and Kir2.3) is presented below, with particular emphasis on the proton modulation of Kir1.1.

PIP2 has been shown to dramatically decrease ATP sensitivity of KATP channels, which are obligate heteromers of Kir6 subunits with distinct isoforms of the sulphonylurea receptor – SUR (Baukrowitz et al. 1998; Shyng & Nichols, 1998). It has not been settled yet whether the PIP2 and ATP binding sites share common structural elements or they communicate allosterically (Xie et al. 2006). In addition, PIP2 attenuated inhibition of KATP channels by glibenclamide, a sulphonylurea that specifically blocks these channels. Finally, KATP sensitivity to PIP2 was shown to be modulated by the specific SUR subunit (SUR1 or SUR2) associating with Kir6.2 (Song & Ashcroft, 2001).

The recently determined atomic resolution structures of Kir channels offer an opportunity to map the sites that mediate the effects of diverse modulators and to examine their relative proximity to sites implicated in channel–PIP2 interactions. At a first approximation this exercise could provide a structural framework that accounts for the dependence of modulators on PIP2.

Regulatory sites of diverse modulators of Kir3 channel activity

We have chosen Kir3.1 as our Kir example onto which to map sites identified in the literature that are involved in mediating the effects of diverse modulators for a number of reasons: (a) its 3-D structure has been determined by X-ray crystallography (Nishida & MacKinnon, 2002; Pegan et al. 2005); (b) it is weakly interacting with PIP2 and thus it is amenable to a high level of modulation; (c) diverse modulators can influence its activity, such as the βγ subunits of G proteins (Gβγ), intracellular Na+, protein phosphorylation by PKA, PKC and tyrosine kinases, hydrolysis of PIP2 by PLC stimulation (both PLCβ and PLCγ), arachidonic acid and its metabolites, etc.

Gβγ modulation of Kir3 currents There are several excellent reviews summarizing results from Kir3 studies (Yamada et al. 1998; Mark & Herlitze, 2000; Stanfield et al. 2002; Bichet et al. 2003; Takano & Kuratomi, 2003; Xie et al. 2006).

At present, four mammalian members in the G protein gated K+ channel subfamily of inward rectifiers (Kir3.1–3.4) have been cloned, two of which, Kir3.1 and Kir3.4, are found in heart (Kubo et al. 1993; Dascal et al. 1993; Krapivinsky et al. 1995; Chan et al. 1996a). Kir3.4 has been shown to form heteromeric complexes with Kir3.1. Heterologous coexpression of the two subunits in Xenopus oocytes or mammalian cell lines produced heteromeric channels giving rise to large macroscopic currents with single-channel conductance and open-time kinetics indistinguishable from those of atrial ACh-activated potassium (KACh) channels (Krapivinsky et al. 1995; Chan et al. 1996a). The KACh channel has been shown since the late 1950s to underlie the heart slowing upon vagal action (Trautwein & Dudel, 1958). GTP-binding (G) proteins were later shown to couple the muscarinic receptor to KACh (Pfaffinger et al. 1985; Breitwieser & Szabo, 1985; Kurachi et al. 1986). Moreover, the Gβγ subunits directly activated KACh, making this channel the first example of an effector protein sensitive to the βγ subunits of G proteins (Logothetis et al. 1987). Expression of homomeric Kir3 subunits gave either no currents or currents that are significantly smaller than heteromers of Kir3.1 with any of the other subunits (Chan et al. 1996a). Efforts to elucidate the structural determinants responsible for the large heteromeric currents identified a P-region residue of Kir3.1 that enhanced the activity and altered the gating kinetics of other members of the Kir3 channel subfamily (Chan et al. 1996b). Single point mutants at this critical position could yield homotetrameric active channels (Kir3.4(S143T) or Kir3.4* and Kir3.1(F137S) or Kir3.1*) (Vivaudou et al. 1997). Gβγ subunit sensitivity of Kir3.1/3.4 has been reported previously (Reuveny et al. 1994; Krapivinsky et al. 1995; Duprat et al. 1995). Gβγ subunits activate not only native Kir3 heteromultimers (Logothetis et al. 1987; Krapivinsky et al. 1995), but also recombinant hetero- or homomultimeric Kir3 channels (e.g. Reuveny et al. 1994; Chan et al. 1996a). There is no qualitative difference in the Gβγ sensitivity of P-region homomultimeric mutants versus heteromultimeric channels (Vivaudou et al. 1997). These studies have produced active homomeric Kir3 subunits that have simplified structure–function studies and have enabled investigators to assess the relative contribution of each subunit to effects on heteromeric channel assemblies (e.g. He et al. 1999, 2002).

Sui et al. (1996, 1998) showed that in order to sustain Kir3.1/3.4 activity, hydrolysable forms of intracellular ATP were required to keep intact PIP2 levels in the plasma membrane. Although PIP2 alone was not sufficient to activate Kir3 channels due to their weak interactions, in the presence of Gβγ or intracellular Na+ (see below), PIP2 could gate the channel. In fact, a mutation that strengthened the Kir3.4*–PIP2 interactions (I229L) removed the need for the presence of Gβγ or intracellular Na+ during PIP2-induced activation. Huang et al. (1998) showed that in the presence of Gβγ, Kir3.1/3.4 channel–PIP2 interactions were stabilized. Figure 2A shows an experiment performed by Zhang et al. (1999), where the kinetics of block by PIP2 antibody served as an indication of the strength of channel–PIP2 interactions (slow kinetics indicating higher strength of interactions than fast kinetics indicating lower strength of interactions). In this experiment, DTT reduced the antibody disulphide bond and reversed its effect allowing for a comparison of the kinetics of block before and after treatment of the inside-out patch with Gβγ. The kinetics of inhibition were significantly slower following exposure to Gβγ than before, providing further evidence that Gβγ stabilized channel–PIP2 interactions.

Figure 2.

Gβγ induces stabilization of Kir3 channel-PIP2 interactions (A) and the localization of Gβγ-interacting sites relative to PIP2-sensitive residues (B, C)
A, PIP2 antibody rapidly blocked Kir3.4* currents. Perfusion with DTT reversed this inhibition, presumably by reducing the disulphide bond of the antibody. Inside-out macropatches were bathed in a solution containing flouride; vanadate, (pyrophospate). The summary data (n= 3) depict T50 (Time required for half maximal current inhibition by the PIP2 antibody) values for the PIP2 antibody-mediated block of Kir3.4* in the presence and absence of Gβγ. Following activation by Gβγ of the channels in the same patch, PIP2 antibody blocked the K+ current with slower kinetics than before Gβγ treatment (P < 0.01). B, the known cytosolic structure of Kir3.1 with a modelled transmembrane domain based on KirBac3.1 showing putative Gβγ-interacting residues highlighted in red, in addition to the cyan-highlighted residues that affect channel–PIP2 interactions. C, surface Kir3.1 model with PIP2 and Gβγ sites shown from the same orientation as described in Fig. 1C. The colours match the colours used in Fig. 2B.

Gβγ binds directly to a region of the C terminus (residues 273–462) and the N terminus (1–85) of Kir3.1 (Huang et al. 1995, 1997). In Kir3.4 the Gβγ binding regions were narrowed down further between residues 253–348 in the C-terminus and 41–92 in the N-terminus. Specifying exact binding sites for Gβγ has proven difficult and multiple sites in the intracellular domain have been implicated in Gβγ binding (Kunkel & Peralta, 1995; Krapivinsky et al. 1995; Huang et al. 1997; He et al. 1999, 2002; Ivanina et al. 2003). Three key residues have been identified that greatly affect Gβγ binding and transduction of its functional effects on Kir3.1* and Kir3.4* channels (He et al. 1999, 2002). Two residues that are involved in agonist-independent effects of Gβγ are: H64 and L268 in Kir3.4* and the corresponding H57 and L262 in Kir3.1*. One residue involved in agonist-dependent effects of Gβγ is L339 in Kir3.4 and the corresponding L333 in Kir3.1. These three residues have been mapped onto the Kir3.1 structure and are shown in red relative to the putative channel–PIP2 interacting residues, shown in cyan (Fig. 2B and C). L333 seems to be twice as far from the closest putative PIP2-interacting residue compared with H57 or L262. If L333 serves to stabilize channel PIP2 interactions it is possible that it accomplishes that either allosterically or through a large conformational movement that brings it within an interaction distance from a PIP2-interacting residue. The precise contribution of each of these critical Gβγ-interacting residues to the stabilization of interactions between the channel and PIP2 has not been determined yet.

Na+ modulation of Kir3 currents Next we will consider evidence that intracellular Na+ also enhances Kir currents by binding to a site close to the putative PIP2 pocket. A G protein-independent mechanism of activating recombinant Kir3 or native KACh channels was attributed to effects of intracellular Na+ (Sui et al. 1996). Hydrolysis of MgATP resulted in PIP2 formation by lipid kinases, which not only enabled Gβγ but also Na+ gating (Sui et al. 1998). Na+, which has been shown to activate both Kir3.2 (Ho & Murrell-Lagnado, 1999a,b) and Kir3.4 (Sui et al. 1996, 1998; Zhang et al. 1999), binds to an aspartate in close proximity to a PIP2-binding site, indicating that shielding of the negatively charged Asp by Na+ might increase the affinity of the channel for PIP2 (Logothetis & Zhang, 1999). Figure 3A shows results from Zhang et al. (1999) during rundown upon inside-out macropatch excision, Na+ pulses increase activity stabilizing Kir3.4* channel–PIP2 interactions but as PIP2 is depleted (by lipid phosphatases present in the patch) the effectiveness of Na+ decreases drastically. As can be seen with the Kir3.4* (D223N) by removing the negative Asp residue we mimic the Na+ effect. Kir3.1 possesses an Asn residue rather than an Asp residue at the corresponding position 217 (N217). In silico mutation of N217 to an Asp residue and minimization of the structure revealed that N217D formed a hydrogen bond with R219, a residue shown to affect PIP2 sensitivity. Our recent work, in which one Na+ ion replaced a water molecule in the immediate vicinity of N217D, has revealed following minimization of the structure the complete Na+ coordination site in the 3-D structure of Kir3.1 (A Rosenhouse-Dantsker, Q. Zhao, J. L. Sui & D. E. Logothetis, unpublished observations). Simulations with the Kir3.1(N217D) have given credit to the earlier hypothesis (Logothetis & Zhang, 1999) that Na+ engages N217D preventing it from hydrogen bonding with the neighbouring R219 residue that is involved in PIP2 sensitivity.

Figure 3.

Na+-induced stabilization of Kir3 channel-PIP3 interactions (A) and the localization of a Na+-interacting Asp residue relative to PIP2–sensitive residues (B, C).
A, Inside-out macropatches excised in a Na+-free Mg2+-containing solution were subjected to multiple, brief exposures to Na+ ions, which stimulated channel activity, the amplitude of which declined as a function of the time following excision. The dashed lines represent fits to the peak currents elicited by the brief applications of Na+ (denoted by asterisks). Kir3.4*(D223N) on the other hand failed to produce Na+-mediated stimulation of K+ currents. Following patch excision, Na+ application during the slow course of rundown not only failed to stimulate activity, but it also caused a small current inhibition (denoted by asterisks). The current rundown kinetics of the Kir3.4*(D223N) mutant were much slower than those of Kir3.4*. Moreover, no significant difference in rundown kinetics was seen in the presence (dashed line is the fit to the current troughs elicited by the brief exposures to Na+) or absence of Na+ ions. The summary data (n= 4) show the effect of Na+ ions on the current rundown kinetics for Kir3.4* (P < 0.005) and Kir3.4*(D223N). B, the known cytosolic structure of Kir3.1 with a modelled transmembrane domain based on KirBac3.1 showing the putative Na+ interacting residue, N217, highlighted in blue, in addition to the cyan-highlighted residues that affect channel–PIP2 interactions. N217 is located at the equivalent position to the position of D223 in Kir3.4*. C, surface Kir3.1 model with PIP2 and N217, the Na+-interacting residue, shown from the same orientation as described in Fig. 1C. The colours match the colours used in Fig. 3B.

The position of Kir3.1(N217D) is shown relative to the putative PIP2-interacting residues and R219 in particular (Fig. 3B and C). The effect of intracellular Na+ serves as the best example thus far of how a modulator can alter the immediate environment of critical residues that confer sensitivity to PIP2.

Kir6.2 channels which display weak interactions with PIP2 also exhibit low stereospecific interactions with phosphoinositides, showing similar activation by PI(4,5)P2, PI(3,4)P2, PI(3,4,5)P3 and even by oleoyl-coA, a negatively charged lipid (Rohacs et al. 2003). Kir2.1 channels that display strong interactions with PIP2, exhibit highly stereospecific interactions with PI(4,5)P2 and fail to be activated by the other phosphoinositides. Oleoyl-coA inhibits Kir2.1 currents in a manner that is competitive with PIP2. Specific mutations of Kir2.1 decreased its stereospecificity and enabled oleoyl-coA to activate it rather than inhibit it. These experiments have suggested that channels that show low specificity of interactions with phosphoinositides can be activated by other anionic lipids as well. Arachidonic acid and some of its leukotriene derivatives have also been shown to activate KACh in atrial cells in a PTX-independent way (Kurachi et al. 1989; Kim et al. 1989). Another study reported that homomeric Kir3.1* or Kir3.4* currents were inhibited (rather than activated) by arachidonic acid and that inhibition depended on the critical Asp that is involved in coordinating Na+ (Rogalski & Chavkin, 2001). Since Kir3 channels show low specificity for phosphoinositides (Rohacs et al. 2003), these results are consistent with the notion that other lipids may also regulate their activity. Moreover, these results suggest that heteromeric Kir3 channels may respond differently to arachidonic acid and its metabolites than homomeric ones. It will be interesting to pursue further the details of the mechanism of action of arachidonic acid at the reported site that is shared by Na+ ions.

It has been previously shown that Kir3 channel activity is modulated by an interplay between Gβγ, PIP2, Na+ and Mg2+ (Petit-Jacques et al. 1999). This study explored the combined effects of cytoplasmic regulators and found that they can act in a synergistic manner. In the physiological setting, several cytoplasmic regulators modulate channel activity at the same time. The sites at which Mg2+ acts remain unidentified. In several cases the interplay among different modulators could be based on their binding to neighbouring sites and affecting channel–PIP2 interactions in a coordinated manner.

Modulation of Kir3 channels by phosphorylation Several studies have provided evidence for putative phosphorylation of specific residues by either protein kinase C or A or by tyrosine kinases. These residues in Kir3.1 are: Y12 (not in the structure), Y67, S185, S221 and S315.

PKC effects Hill and Peralta reported that stimulation of the M1 receptor suppressed both basal and dopamine 2 receptor-activated Kir3.1/4 currents (Hill & Peralta, 2001). Overexpression of Gβγ subunits attenuated this effect. This Gq signal required the use of second messenger molecules and pharmacological evidence implicated a role for PKC and Ca2+ responses as M1 receptor-mediated inhibition of GIRK channels was mimicked by PMA and a Ca2+ ionophore. Analysis of a series of mutant and chimeric channels suggested that the mutant Kir3.4 subunit was still capable of responding to Gq signals. The authors concluded that the inhibition did not occur via phosphorylation of a canonical PKC site on the channel itself. Curiously, the authors tested the mutants under Gβγ overexpression conditions which as they had already concluded themselves attenuate the inhibition. Recent work in our lab (Keselman et al. 2007) has addressed the mechanism of Gq-mediated inhibition of Kir3 currents showing that both direct PIP2 hydrolysis as well as PKC activation contribute to the inhibition. Thus, the question of whether the mutated sites contribute to the PKC effects remains open. Our work has shown that PKC treatment decreases the affinity of the channel for PIP2.

In order to identify the PKC phosphorylation sites Mao and colleagues (Mao et al. 2004) performed systematic mutagenesis analysis on Kir3.4 and Kir3.1 subunits expressed in Xenopus oocytes. The data suggested that the heteromeric Kir3.1/4 channels were inhibited by PMA through reduction of the single channel open-state probability. Direct application of the catalytic subunit of PKC to excised patches had a similar inhibitory effect. This inhibition was mostly (∼80%) reduced by mutation of Ser-185 in Kir3.1* or the corresponding Ser-191 in Kir3.4*. The PKC-dependent phosphorylation seems to mediate the channel inhibition by the excitatory neurotransmitter substance P(SP), as specific PKC inhibitors and mutation of these PKC phosphorylation sites abolished the SP-induced inhibition of Kir3.1/4 channels.

TK effects Brain-derived neurotrophic factor (BDNF) acting through TrkB receptors strongly inhibited basal Kir3 currents in a subunit-dependent manner (Rogalski et al. 2000). Functional homomers of Kir3.1 or Kir3.4 but not Kir3.2 were inhibited by BDNF in a genistein-dependent manner. Mutations of either Y12 or Y67 in Kir3.1 or Y32 or Y53 in GIRK4 blocked significantly the BDNF inhibition. The insensitive Kir3.2 was made sensitive to BDNF by adding D41Y and P32K to generate a phosphorylation site motif analogous to the Kir3.4 one.

PKA effects First Mullner and colleagues (Mullner et al. 2000) reported that treatment of atrial cardiomyocytes with isoproterenol (isoprenaline; Iso) resulted in a distinct slow component of activation and a current increase both of which could be abolished by preincubation in 50 μm H89. In Xenopus oocytes Iso facilitated Kir3.1/Kir3.4 currents in a way similar to atrial cells. Cytosolic injection of cAMP but not Rp-cAMPS mimicked the β2-adrenergic effect. Overexpression of Gαs increased basal and agonist-induced currents and this effect was inhibited by H89.

Medina and colleagues (Medina et al. 2000) suggested that Kir3.1 but not Kir3.4 was phosphorylated when heterologously expressed. Protein Phosphatase 2A (PP2A) dephosphorylation of a protein in the excised patch abrogated channel activation by Gβγ. Deletion mutagenesis suggested that the C-terminal region between positions 373 and 419 in Kir3.1 but a mutant of all seven Ser/Thr residues between these two positions did not eliminate Kir3.1 subunit phosphorylation. Our own work has provided functional evidence that PKA treatment increases the apparent affinity for PIP2 and has identified S221 and S315 in Kir3.1 as putative phosphorylation residues (Lopes et al. 2007). These two residues correspond to the Kir1.1 residues S219 and S313, which as was discussed earlier, when mutated to Ala, they reduced the affinity of the channel for PIP2. The four identified putative phosphorylation residues, S185 (PKC), Y67 (TK), and the pair of S221/S315 (PKA), were mapped onto the Kir3.1 structure (Fig. 4A and B). As can be seen in the figure, these putative phosphorylation sites are localized next to residues implicated to mediate channel–PIP2 interactions.

Figure 4.

Localization of phosphorylation sites of the Kir 3.1 channel relative to tis PIP2–sensitive residues
A, the known cytosolic structure of Kir3.1 with a modelled transmembrane domain showing the phosphorylation sites suggested by functional evidence highlighted in orange, in addition to the cyan-highlighted residues that affect channel–PIP2 interactions. B, surface Kir3.1 model with PIP2 and the putative phosphorylation sites, shown from the same orientation as described in Fig. 1C. The colours match the colours used in Fig. 4A.

Localization of sites affecting pH sensitivity in a Kir1.1 model

The 3-D structure of Kir3.1 shown in Figs 1–4 incorporates the actual cytosolic domain structure determined by X-ray crystallography (Nishida & MacKinnon, 2002; Pegan et al. 2005) and a homology model of the transmembrane domains based on the bacterial KirBac3.1 channel (J. M. Gulbis, A. Kuo, B. Smith, D. A. Doyle, A. Edwards, C. Arrowsmith et. al., unpublished observations). Although atomic resolution structures of other Kir channels that are highly modulated are not yet available, the current structures allow us to generate homology models based on the mammalian cytosolic structures of Kir3.1, Kir3.2 or Kir2.1 and the bacterial transmembrane domains of KirBac1.1 or KirBac3.1. These homology models can be built as previously described (Lopes et al 2007).

Several Kir members (e.g. Kir 1.1, Kir2.3 and Kir4.1/5.1) have been shown to be regulated by intracellular pH (Giebisch, 1998; Qu et al. 2000; Yang et al. 2000; Pessia et al. 2001; Du et al. 2004; Hebert et al. 2005). Leung et al. (2000) suggested that PIP2 binding in Kir1.1 altered the pKa for pH gating by showing that mutations that decrease affinity for PIP2 caused an alkaline shift in pH sensitivity. Intracellular acidification reversibly reduced the open probability (Po) of Kir1.1 in the physiological range, a characteristic that makes this channel a key player in K+ homeostasis during metabolic acidosis in the kidney (Giebisch, 1998; Hebert et al. 2005). As mentioned earlier, Du et al. (2004) suggested that the pH sensitivity of Kir2.3 compared with Kir2.1 depended greatly on the strength of their respective interactions with PIP2; they showed that mutations that decreased affinity of the channel for PIP2 caused an alkaline shift, whereas mutations that increased affinity of the channel for PIP2 caused an acidic shift in pH sensitivity. Both studies in Kir1.1 (Leung et al. 2000) and Kir2.3 (Du et al. 2004) pointed to the same conclusion that the stronger the channel interactions with PIP2 the less proton-sensitive were the Kir channels tested. Rapedius et al. (2006) reached a different conclusion examining the R311W mutation. Mutations of R311 have previously been shown to affect the PIP2 sensitivity of Kir1.1 and Kir2.1 (R312) (Lopes et al. 2002). The Kir1.1(R311W) caused a 2.7 pH unit shift (from 6.5 to 9.2), consistent with the conclusions drawn by the studies of Leung et al. (2000) and Du et al. (2004) that weakening channel–PIP2 interactions increases proton sensitivity. Yet, Rapedius and colleagues concluded that changes in PIP2 affinity did not underlie the alkalinization shift. They reached this conclusion by testing for changes in pH inhibition of the wild-type and the Kir1.1(R311W) mutant under high PIP2 conditions and finding none. Since these experiments were not shown in any detail it is difficult to evaluate the effects of changes in pH sensitivity due to high PIP2 levels. It is also unclear whether the starting levels of PIP2 were low enough to be able to see an effect when inside-out macropatches were exposed to high PIP2 concentrations. In Kir4.1/5.1 heteromers, for example, where it was concluded that PIP2 controls proton sensitivity (Yang et al. 2000), when a similar experiment was performed, 10 μm PIP2 application in inside-out macropatches caused a shift of only 0.22 pH units, decreasing sensitivity to protons. Thus, given these considerations, the decrease in Kir1.1 affinity for PIP2 has not been ruled out as the underlying cause for the alkalinization shift of R311W.

Since pH sensitivity has been extensively studied in Kir1.1, we proceeded to map residues of Kir1.1 that have been implicated in pH sensitivity onto a 3-D modelled structure of this channel. We generated a homology model of Kir1.1 using the Kir2.1 cytosolic domain and the KirBac3.1 transmembrane domains as templates, as previously described (Lopes et al. 2007). Figure 5A and B shows a number of residues that have been identified to affect pH sensitivity (Dahlmann et al. 2004; Rapedius et al. 2006). We have excluded Kir1.1 (K80 and A177), which recently were shown not to comprise the pH sensor but to indirectly control the channel gate and therefore the expression of pH effects (Rapedius et al. 2006). The five residues shown could be grouped into two types: (a) those that cluster around the helix bundle crossing (I178, L179, I182) and could be affecting pH sensitivity by affecting the channel gate and (b) those that are positioned proximally to residues involved in PIP2 sensitivity (E302 is proximal to R184, R186 and R311; E318 is proximal to K188) and could be affecting pH sensitivity by altering channel PIP2 interactions. This analysis points out the potential caveats of identifying molecular determinants of a particular modulator based on function alone. Although there is no question that all the identified residues can affect pH sensitivity, their contribution could be indirect, reflecting the dependence of the coupling mechanism of the unidentified pH sensor to the gate. Thus, it appears that additional work will be needed to identify the pH sensor, its relationship to PIP2 and to the gate.

Figure 5.

Localization of pH–sensitive residues relative to Kir1.1 PIP2–sensitive residues.
A, an homology model of Kir1.1 based on the cytosolic structure of Kir2.1 and the transmembrane domain of KirBac3.1. Sites highlighted in green show residues, mutation of which affects pH sensitivity, in addition to the cyan-highlighted residues that affect channel–PIP2 interactions as determined experimentally (Lopes et al. 2002). B, surface Kir1.1 model with PIP2 and the sites that affect pH sensitivity, shown from the same orientation as described in Fig. 1C. The colours match the colours used in Fig. 5A.

Kir channel–PIP2 interactions and human disease

Molecular defects in Kir channels can result in human disease (Abraham et al. 1999). Bartter's syndrome is an autosomal recessive disease characterized by hypokalaemia, salt wasting, metabolic alkalosis, hypecalciuria, hyperreninism and hyperaldosteronism. In a subset of patients, mutations in Kir1.1 channels have been implicated in the disease (Simon et al. 1996; Derst et al. 1997; Schwalbe et al. 1998). Andersen's syndrome is an autosomal dominant trait characterized by periodic paralysis, cardiac arrhythmias and dysmorphic features. Mutations in Kir2.1 underlie the developmental and episodic electrical phenotypes of the disease (Plaster et al. 2001). Mutations that cause disease have been shown to affect channel–PIP2 interactions in both of these syndromes (Lopes et al. 2002; Donaldson et al. 2003; Davies et al. 2005). The central role that PIP2 is playing in controlling the activity of so many ion channels begs the question of why additional PIP2-dependent conditions have not yet been described. We believe this to be a matter of time from the point investigators begin to consider associations of defects in channel–PIP2 interactions with human diseases.

Future work

It is clear that understanding the details of channel–PIP2 interactions in molecular detail lies at the heart of the subject of this review. Structural and computational studies will be needed to elucidate the molecular interactions between phosphoinositides and ion channels and to explain the experimentally obtained profile of their affinities and specificities. The coupling mechanism between the channel-PIP2 binding site and the affected channel gate(s) also remains to be elucidated. Moreover, a molecular appreciation of channel–PIP2 interactions will allow us to see how modulators exert their effects by congregating around residues that interact with PIP2 to alter specific interactions. A molecular picture of such interactions will elucidate how the Gβγ subunits stabilize channel–PIP2 interactions, how distinct kinases can activate (PKA) or inhibit (PKC or TK) Kir3 activity, and how protons exert their effects on ion channel activity. Such future work will provide a framework under which the effects of additional diverse modulators of Kir activity that show dependence on channel–PIP2 interactions can be understood mechanistically.



We would like to thank past and present members of the Logothetis lab. Without their dedication and enthusiasm in studying phosphoinositide control of Kir channel activity, this manuscript would not have been possible. This work was supported by NIH grant HL59949 to D.E.L.