Synergistic Activation of G Protein–Gated Inwardly Rectifying Potassium Channels by the βγ Subunits of G Proteins and Na⁺ and Mg²⁺ Ions

Abstract

Native and recombinant G protein–gated inwardly rectifying potassium (GIRK) channels are directly activated by the βγ subunits of GTP-binding (G) proteins. The presence of phosphatidylinositol-bis-phosphate (PIP₂) is required for G protein activation. Formation (via hydrolysis of ATP) of endogenous PIP₂ or application of exogenous PIP₂ increases the mean open time of GIRK channels and sensitizes them to gating by internal Na⁺ ions. In the present study, we show that the activity of ATP- or PIP₂-modified channels could also be stimulated by intracellular Mg²⁺ ions. In addition, Mg²⁺ ions reduced the single-channel conductance of GIRK channels, independently of their gating ability. Both Na⁺ and Mg²⁺ ions exert their gating effects independently of each other or of the activation by the G_βγ subunits. At high levels of PIP₂, synergistic interactions among Na⁺, Mg²⁺, and G_βγ subunits resulted in severalfold stimulated levels of channel activity. Changes in ionic concentrations and/or G protein subunits in the local environment of these K⁺ channels could provide a rapid amplification mechanism for generation of graded activity, thereby adjusting the level of excitability of the cells.

INTRODUCTION

In atrial tissue, acetylcholine released by the vagus nerve binds to muscarinic type 2 receptors, activates K_ACh channels via pertussis toxin–sensitive G proteins, and slows the heart rate. Upon activation, the heterotrimeric G protein dissociates, allowing the G_βγ subunits to directly activate the K_ACh channel (Logothetis et al. 1987; Krapivinsky et al. 1995b). K_ACh has been shown to be composed of two types of G protein–gated inwardly rectifying potassiumb channels (GIRK1 and GIRK4),¹ associated in a heterotetrameric complex (Krapivinsky et al. 1995a; Silverman et al. 1996; Corey et al. 1998). Recombinant (GIRK) channels expressed in oocytes are also directly activated by G protein βγ subunits (Reuveny et al. 1994). In addition, GIRK channels appear to be activated independently of G proteins. In the absence of agonist, ATP hydrolysis leads to an increase in the mean open time and sensitizes channels to gating by Na⁺ ions (Sui et al. 1996). Recently, it was shown that the ATP modification of GIRK channels is mediated via phosphatidylinositol phosphates such as phosphatidylinositol-bis-phosphate (PIP₂) (Huang et al. 1998; Sui et al. 1998). PIP₂ has been implicated in the regulation of the sodium–calcium exchanger (Hilgemann and Ball 1996), the K_ATP channel (Hilgemann and Ball 1996; Fan and Makielski 1997; Baukrowitz et al. 1998; Shyng and Nichols 1998), the inwardly rectifying ROMK1 and IRK1 channels (Huang et al. 1998) and other Na⁺-gated nonselective cation channels (Zhainazarov and Ache 1999). Moreover, PIP₂ appears to be essential for GIRK channel activation by the G protein βγ subunits (Sui et al. 1998).

Here, using both native and recombinant GIRK channels, we show that Na⁺ as well as Mg²⁺ ions gate the ATP- or PIP₂-modified channels. While the two ions seem to exert their effects at distinct sites on the channel protein, they showed synergistic effects on gating. In the presence of exogenous PIP₂, G_βγ and Na⁺ and Mg²⁺ ions showed great synergism in activating the channel. However, in the absence of exogenous PIP₂, preactivation by G protein βγ subunits sensitized the channel to gating by Na⁺ but not Mg²⁺ ions. These data suggest that the synergism between Mg²⁺ and G_βγ subunits in gating GIRK channels shows a much greater dependence on PIP₂ levels than the synergism between Na⁺ and G_βγ. The synergism among ions and G_βγ proteins in the gating of GIRK channels implies that variations of the concentrations of these molecules in the local environment of these channels could play an important role in the “fine tuning” of their activity.

MATERIALS AND METHODS

Expression of Recombinant Channels in Xenopus Oocytes

Recombinant channel subunits (GIRK1, GenBank accession No. U39196; GIRK4, GenBank accession No. U39195) were expressed in Xenopus oocytes as described previously (Chan et al. 1996). Channel subunit coexpression was accomplished by coinjection of equal amounts of each cRNA (∼4 ng). The human muscarinic receptor type 2 was coexpressed with the channel subunits (∼1.5 ng injected per oocyte). The β-adrenergic receptor kinase (βARK)–PH construct, altered to incorporate the 15 NH₂-terminal residues of Src for membrane targeting, was generously provided by Dr. E. Reuveny (Weizmann Institute of Science, Rehovot, Israel). cRNA concentrations were estimated from two successive dilutions that were electrophoresed on formaldehyde gels in parallel and compared with known concentrations of a RNA marker (GIBCO BRL). Oocytes were isolated and microinjected as described previously (Logothetis et al. 1992). The oocytes were maintained at 18°C, and electrophysiological recordings were performed 2–6 d after injection at room temperature (20–22°C).

Preparation of Chicken Atrial Myocytes

The procedure used for isolating cardiac myocytes from chicken embryos has been described previously (Sui et al. 1996). In brief, atrial tissue was selected using chicken embryos from eggs incubated 14–18 d. Atrial tissue was incubated for 20–30 min at 37°C in 5 ml of Mg²⁺- and Ca²⁺-free PBS supplemented with 1–2% trypsin/EDTA solution (10×, GIBCO BRL). Isolated myocytes were collected by triturating the digested tissue in 5 ml of trypsin-free solution and stored in a high potassium (KB) solution (Isenberg and Klöckner 1982) at 4°C for up to 36 h. The cells were allowed to settle on poly-lysine–coated coverslips in the recording chamber before experiments.

Reagents

General chemical reagents, including GTP and ATP, were purchased from Sigma Chemical Co. PIP₂ (Boehringer Mannheim) was sonicated on ice for 30 min before application. Purified recombinant G protein subunits dimer β₁γ₇ was kindly provided by Dr. J. Garrison (University of Virginia, Charlottesville, VA). The stock of β₁γ₇ (0.86 μg/μl) was dissolved in 20 mM HEPES, 1 mM EDTA, 200 mM NaCl, 0.6% CHAPS, 50 mM MgCl₂, 10 mM NaF, 30 μM AlCL₃, 3 mM dithiothreitol (DTT), 3 μM GDP, pH 8.0. The final concentration was 20 nM in a solution containing 0.012% CHAPS, and 20 μM DTT. QEHA peptide (Chen et al. 1995) was kindly provided by Dr. R. Iyengar (Mount Sinai School of Medicine) and was used at a final concentration of 50 μM.

Single-Channel Recording and Analysis

Single-channel activity was recorded in the cell-attached or inside-out patch configurations (Hamill et al. 1981) using an Axopatch 200B amplifier (Axon Instruments). All pipettes used in the experiments were pulled using the WPI-K borosilicate glass (World Precision Instruments) and gave resistances of 2–8 MΩ. All experiments were conducted at room temperature (20–22°C). Single-channel recordings were performed at a membrane potential of −80 mV with acetylcholine (ACh, 5 μM) in the pipette, unless otherwise indicated. Single-channel currents were filtered at 1–2 kHz, sampled at 5–10 MHz, and stored directly into the computer's hard disk through the DIGIDATA 1200 interface (Axon Instruments). PCLAMP (v. 6.03; Axon Instruments) was used for data acquisition.

To remove the vitelline membrane, Xenopus oocytes were placed in a hypertonic solution (Stühmer 1992) for 5 min. Shrunk oocytes were transferred into a V-shaped recording chamber and the vitelline membrane was partially removed, exposing just enough plasma membrane for access with a patch pipette (Sui et al. 1996). This procedure increased the success rate of forming gigaseals.

The pipette solution contained 96 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, and 10 mM HEPES, pH 7.35. The bath solution contained 96 mM KCl, 5 mM EGTA, and 10 mM HEPES, pH 7.35. When high concentrations of Mg²⁺ ions (>5 mM) were used in the bath solution, the KCl concentration was reduced accordingly to maintain osmolarity. Gadolinium chloride at 100 μM was routinely added to the pipette solution to suppress native stretch channel activity in the oocyte membrane. For chick atrial cells, the experimental solutions were the same as those used with oocyte recordings, except that the KCl concentration was 140 mM without gadolinium chloride.

Free Mg²⁺ and ATP concentrations were estimated as described previously (Vivaudou et al. 1991). Single-channel records were analyzed using PCLAMP software, complemented with our own analysis routine, as described previously (Sui et al. 1996). Parameters used for single-channel analysis include activity of all channels in the patch (or the total open probability, NP _o), the total frequency of opening (NF _o), and the mean open time (MT _o), and averages over 5-s bins are displayed.

In experiments shown in Fig. 7, where exogenous PIP₂ was applied throughout the experiment (i.e., Fig. 7A and Fig. C), occasional applications of the same ion as a function of time in the experiment were used as control to ascertain that the synergistic effects described were not due to a time-dependent accumulation of PIP₂ in the membrane patch. Similar precautions were taken in the experiments shown in Fig. 4. Experiments used to generate the data shown in these two figures were never longer than 14 min (usually 10–13 min). Na⁺ and/or Mg²⁺ ions were applied for 30 s.

Figure 7.

Synergistic effects of G_βγ, Na⁺, and Mg²⁺ ions in activating GIRK channels. (A) The mean NP _o for seven patches are plotted for different conditions. The data were obtained from inside-out patches excised from oocytes expressing the recombinant channel GIRK1/GIRK4. The membrane was held at −80 mV and 5 μM acetylcholine was present in the pipette. PIP₂ was 2.5 μM, Mg²⁺ was 10 mM, and Na⁺ was 10 mM. SEM are indicated by the vertical bars. The mean NP _o for the channel activity was 0.011 ± 0.003 in control conditions, and 0.01 ± 0.006 during the application of 2.5 μM PIP₂. When 10 mM Mg²⁺ ions were applied with PIP₂, the mean NP _o was 0.43 ± 0.14. 10 mM Na⁺ ions gave a mean NP _o of 0.40 ± 0.12. When applied together, in the presence of PIP₂, Mg²⁺ and Na⁺ ions (10 mM each) yielded a mean NP _o of 2.08 ± 0.52. (B) Mean NP _o plots for six inside-out patches from oocytes expressing GIRK1/GIRK4. V_m was −80 mV. 5 μM acetylcholine was in the pipette. Mg²⁺ was 10 mM, Na⁺ was 10 mM, and GTPγS was 10 μM. The columns GTPγS and GTPγS+x depict the channel activity measured after the GTP analogue was washed out and substance(s) x were added. SEM are indicated by vertical bars. The mean NP _o of the channel was 0.023 ± 0.012 in control conditions and 0.035 ± 0.02 in the presence of 10 mM Mg²⁺ ions. 10 μM GTPγS gave a mean NP _o of 0.12 ± 0.045. After GTPγS washout, Mg²⁺ ions gave a mean NP _o of 0.12 ± 0.03. 10 mM Na⁺ ions gave a mean NP _o of 0.23 ± 0.06. Coapplication of Mg²⁺ and Na⁺ ions resulted in a mean NP _o of 0.26 ± 0.08. (C) Mean NP _o plots for six patches. The inside-out patches were excised from oocytes expressing GIRK1/GIRK4. V_m was −80 mV. 5 μM ACh present in the pipette. PIP₂ was 2.5 μM, Mg²⁺ was 10 mM, Na⁺ was 10 mM, and GTPγS was 10 μM. PIP₂+GTPγS refers to the channel activity (at steady state) during the application of the GTP analogue. PIP₂+GTPγS+x columns depict the channel activity measured after the washout of the GTP analogue and addition of substance(s) x. In absence of 10 mM Mg²⁺, all solutions contained 50 μM Mg²⁺. This low concentration of Mg²⁺ was necessary to render GTPγS effective. Vertical bars represent SEM. The mean NP _o for the channel activity was 0.004 ± 0.002 in control conditions and 0.0007 ± 0.0002 in the presence of 2.5 μM PIP₂. When 10 mM Mg²⁺ ions were applied to the patches in the presence of PIP₂, a mean NP _o of 0.034 ± 0.013 was obtained. Although this activity appeared small, it was significantly higher than that in PIP₂ alone (P < 0.005, paired t test, log scale). 10 mM each of Mg²⁺ and Na⁺ ions in combination yielded a mean NP _o of 0.2 ± 0.09. GTPγS gave a mean NP _o of 0.058 ± 0.03. Again, although this activity appeared relatively small, it was significantly higher than that in PIP₂ alone (P < 0.005, paired t test, log scale). When 10 mM Mg²⁺ ions were applied to the patches after the GTPγS treatment a mean NP _o of 0.42 ± 0.19 was obtained. Mg²⁺ and Na⁺ ions applied together resulted in NP _o of 1.6 ± 0.49.

Figure 4.

Concentration dependence of GIRK channel activation on internal Mg²⁺ ions. Normalized activity (NP _o) of GIRK1/GIRK4 channels is plotted for different Mg²⁺ concentrations. Inside-out patches were exposed to 2.5 μM PIP₂ before and during application of Mg²⁺. Responses were expressed in fold increase above the activity in the presence of PIP₂ and were normalized to those recorded in the presence of 1 mM Mg²⁺. Vertical bars represent SEM. The responses obtained at the low concentrations of Mg²⁺ tested (<1 mM) were significantly higher than those with PIP₂ alone (P < 0.05, paired t test). The holding potential was at −80 mV. 5 μM ACh was present in the pipette. *Significant differences from 1 mM Mg²⁺ (P < 0.01; paired t test).

RESULTS

MgATP/Na Activation of the K_ACh Channel Can Proceed Independently of the Involvement of G Proteins

It has been shown previously that Na⁺ ions can stimulate K_ACh activity in an ATP-dependent manner in the absence of agonist and internal GTP (Sui et al. 1996). The ATP-dependent modification of the channel is thought to work via the production of membrane phosphoinositide phosphates (e.g., PIP₂), which interact directly with members of the inwardly rectifying K⁺ channel family (Huang et al. 1998; Sui et al. 1998). PIP₂ also appears to be essential for G protein regulation of the K_ACh channel (Sui et al. 1998).

To further test for a dependence of the MgATP/Na⁺ activation on G protein gating of K_Ach, we designed experiments where G protein–dependent activation of the channel was impaired. As shown in Fig. 1 A, the K_ACh channel in an inside-out atrial myocyte patch was activated persistently by 10 μM GTPγS, a nonhydrolyzable analogue of GTP. Activation of the channel by GTPγS was blocked upon perfusion of the QEHA peptide. QEHA is a 27 amino acid long peptide derived from the COOH terminus of the G_βγ-sensitive adenylate cyclase 2 isoform. It has been shown to block G_βγ activation of several different effectors, including the K_ACh channel (Chen et al. 1995). QEHA (50 μM) application abolished the GTPγS activation of K_ACh in <2 min (n = 3). After washout, the channel activity remained very low, suggesting the persistence of the QEHA-blocking effect. However, under these conditions, the K_ACh channel could be activated by MgATP/Na⁺ (5/20 mM). QEHA coapplication with MgATP/Na⁺ failed to block channel activation, whereas QEHA did block GTPγS-induced activation in the same oocyte patches (n = 3) (data not shown).

Figure 1.

Impairment of G protein signaling does not affect the activation of K_ACh by MgATP and Na⁺. (A) Single channel activity (top; NP _o, bin = 5 s) plotted as a function of time. The data were obtained from an inside-out patch excised from an atrial cell. The K_ACh channel was stimulated by maintaining the membrane at −80 mV and by the presence of 5 μM acetylcholine in the pipette. 10 μM GTPγS, 50 μM QEHA, and 5/20 mM MgATP/Na⁺ were applied for the duration indicated by the bars. Sample single-channel currents in each condition at the time marked by the arrows are shown under the plot (bottom). (B) NP _o plot of K_ACh channel activity (top, bin = 5 s) in an inside-out patch from an oocyte expressing the human GIRK1/GIRK4 and the construct βARK-PH. The membrane was clamped at −80 mV and 5 μM acetylcholine was present in the pipette. Application of 10 μM GTPγS and 5/20 mM MgATP/Na⁺ are illustrated by the bars. Labeled arrows correspond to the sample single-channel currents shown under the plot (bottom).

Another way we impaired the G protein regulation of the GIRK channels was by coexpressing them in oocytes with a βγ-binding protein. We used the PH domain of βARK (βARK-PH), which specifically binds the βγ subunits of G proteins, and thus acts as a “βγ sink” (Koch et al. 1993; He et al. 1999). In oocytes coexpressing the recombinant channels GIRK1/GIRK4 and the construct βARK-PH, 10 μM GTPγS did not induce channel activity. This suggests that the βARK-PH protein bound the oocyte endogenous G proteins, such that no βγ subunits were available for channel activation (Fig. 1 B). However, in the same patches, MgATP/Na⁺ (5/20 mM) caused a >30-fold increase in channel activity. Summary data revealed that channel activities (NP _o) before, during, and after GTPγS application were similar, 0.0070 ± 0.0039, 0.0077 ± 0.0037, and 0.0097 ± 0.0045, respectively (mean ± SEM, n = 4). During application of MgATP/Na⁺, NP _o was 0.313 ± 0.221 (n = 4).

In control experiments using inside-out patches from oocytes of the same batch that coexpressed the recombinant channels GIRK1/GIRK4 alone, GTPγS caused great channel activation (n = 3, data not shown). Similar results were obtained in experiments in which we applied Na⁺ ions with PIP₂ rather than MgATP (n = 4, data not shown).

These results suggest that even when G protein regulation is impaired, Na⁺ ions are still able to activate the channel. Thus, Na⁺ gating of the channel can indeed proceed independently of G_βγ gating.

G_βγ Subunits Sensitize GIRK Channels to Gating by Na⁺ Ions

Na⁺ ions can gate GIRK channels when membrane PIP₂ levels are maintained (i.e., via hydrolysis of ATP). We next tested under conditions that did not maintain PIP₂ at a constant high level whether Na⁺ ions could gate these channels after G_βγ activation.

Fig. 2A and Fig. B, show representative and summary data from experiments where Na⁺ ions gated GIRK1/GIRK4 channels after activation by G proteins. Inside-out patches from oocytes expressing these channels showed no channel activity upon application of 20 mM Na⁺. This result suggested a low presence of PIP₂ in the membrane. However, this PIP₂ concentration was sufficient to allow persistent channel activation by a brief exposure to 10 μM GTPγS. Reapplication of Na⁺ ions produced a more than fourfold increase in the channel activity above the level obtained with GTPγS. It should be noted that the effect of Na⁺ ions on the basal channel activity was variable from patch to patch, presumably reflecting different levels of endogenous PIP₂ at the time of Na⁺ application.

Figure 2.

Na⁺ ions gate GIRK channels after activation by G protein βγ subunits. (A) Single-channel activity (NP _o, bin = 5 s) plotted as a function of time. The data were obtained from an inside-out patch excised from an oocyte expressing the recombinant channel GIRK1/GIRK4. 20 mM Na⁺ and 10 μM GTPγS were applied as indicated by the bars. The membrane was clamped at −80 mV and 5 μM acetylcholine was in the pipette solution. (B) The mean NP _o for seven patches is plotted for different conditions. Steady state channel activity after activation by GTPγS was taken as reference (GTPγS) and NP _o were normalized to it. Na⁺ concentration was 20 mM and GTPγS was 10 μM. GTPγS+Na⁺ corresponds to the application of 20 mM Na⁺ after the washout of the GTP analogue. SEM are indicated by the vertical bars. The normalized mean NP _o was 0.057 ± 0.018 (mean ± SEM) in control solution, 0.277 ± 0.095 in the presence of 20 mM Na⁺ ions, 1 after the application of 10 μM GTPγS, and 4.21 ± 0.59 in the presence of 20 mM Na⁺ ions after channel activation by GTPγS. (C) NP _o vs. time plot for the channel activity recorded in an inside-out patch from an oocyte expressing GIRK1/GIRK4. 20 mM Na⁺ and 20 nM β₁γ₇ purified subunits were applied via the bath as indicated by the bars. V_m = −80 mV. 5 μM acetylcholine was present in the pipette. (D) The mean NP _o for nine patches are plotted for different conditions. Steady state channel activity after β₁γ₇ activation (after β₁γ₇ washout) was taken as reference and NP _o was normalized to it. Na⁺ concentration was 20 mM and β₁γ₇ was 20 nM. βγ+Na⁺ refers to the application of 20 mM Na⁺ after the washout of βγ. The vertical bars represent SEM. The normalized mean NP _o was 0.084 ± 0.039 (mean ± SEM) in control solution, 0.206 ± 0.12 in the presence of 20 mM Na⁺ ions, 1 after the application of 20 nM β₁γ₇, and 3.1 ± 0.84 in the presence of 20 mM Na⁺ ions after activation of the channel by the G protein subunits.

Na⁺ ions also gated GIRK channels after stimulation of activity by purified G_βγ subunits. In Fig. 2C and Fig. D, Na⁺ ions (20 mM) applied on an inside-out patch did not affect significantly the basal activity of the channel. After washout of the Na⁺ ions, recombinant G_β1γ7 was applied on the patch at a concentration of 20 nM, causing a slow channel activation. After washout of G_β1γ7, and as activity stabilized, a second application of Na⁺ ions produced a more than threefold increase in channel activity, above the level obtained with β₁γ₇. Combined together, these data suggested that the G protein βγ subunits sensitized GIRK channels to gating by Na⁺ ions. It has been shown that the mean open time (MT _o) increased in the presence of PIP₂ that is generated by hydrolysis of ATP or exogenous application (Sui et al. 1996, Sui et al. 1998). In the present experiments, no change in the channel MT _o was observed in the different solutions perfusing the patches (data not shown). This suggests that the levels of PIP₂ in the membrane were not altered, and thus could not account for the G_βγ-dependent gating effects of Na⁺ ions.

It has been shown that Li⁺ ions stimulate GIRK channels modified by ATP to ∼10% the activity level achieved by comparable Na⁺ ion concentrations (Sui et al. 1996). However, Li⁺ ions were unable to increase the activity of the channel after activation by GTPγS. In three patches, the mean NP _o of the GIRK channel was 0.028 ± 0.013 in control conditions, 0.127 ± 0.035 after application of 10 μM of GTPγS, 0.578 ± 0.15 in the presence of 20 mM Na⁺ ions, and 0.099 ± 0.045 in the presence of 20 mM Li⁺ ions (data not shown). When applied together, Li⁺ ions were also unable to affect the gating of the GIRK channel by Na⁺ ions. This suggests that the gating effect of Na⁺ ions on the GIRK channel activated by G protein βγ subunits is specific to Na⁺ ions.

Mg²⁺ Ions Gate GIRK Channels After Channel Modification by ATP or PIP₂

In certain experiments, 5 mM MgATP increased the activity of the GIRK channels in the absence of Na⁺ ions (e.g., Sui et al. 1996). 5 mM MgATP in the solution corresponds to a free Mg²⁺ ion concentration of ∼2.1 mM (Vivaudou et al. 1991). This observation prompted us to test whether Mg²⁺ ions alone were able to gate the channel that had been modified by ATP. In Fig. 3, in an inside-out patch from an oocyte coexpressing the channel subunits GIRK1/GIRK4, Mg²⁺ ions (10 mM) had no significant effects on channel activity in the absence of ATP. After washout of Mg²⁺, the channel was activated by the combination of MgATP (2.5 mM; corresponding to ∼1.1 mM free Mg²⁺) and Na⁺ ions (20 mM). MgATP application was maintained and, upon withdrawal of Na⁺ ions, channel activity became comparable to basal levels. Application of Mg²⁺ ions (10 mM), in the continuous presence of MgATP (2.5 mM), increased channel activity to levels similar to those obtained with Na⁺ ions (as confirmed by sequential application of 10 mM of each of the ions at the end of the experiment). Withdrawal of Mg²⁺ ions caused channel activity to return to basal levels (n = 3). The MT _o of the channel activity was increased from ∼1 to ∼2 ms by the application of MgATP, but was not further modified during the gating by Mg²⁺ or Na⁺ ions.

Figure 3.

Mg²⁺ ions gate the ATP-modified GIRK channels. NP _o, NF _o, and MT _o plots (bin = 5 s) of GIRK channel activity in an inside-out patch excised from an oocyte expressing GIRK1/GIRK4. 10 mM Mg²⁺, 2.5 mM MgATP, and 20 mM Na⁺ were applied via the bath as indicated by the bars. The membrane was clamped at −80 mV and 5 μM acetylcholine was present in the pipette. NP _o, NF _o, and MT _o in 10 mM Mg²⁺ were calculated such that the amplitude levels were set to match the reduced amplitude of the channel openings in 10 mM Mg²⁺.

Using PIP₂, we could test the ability of different Mg²⁺ concentrations to activate the GIRK channels. Fig. 4 represents normalized activity of GIRK channels for different concentrations of Mg²⁺ ions. The NP _o for each concentration was calculated in reference to the NP _o measured at 1 mM Mg²⁺. Mg²⁺ ions could activate the GIRK channels at concentrations as low as 100–300 μM. Maximal activity could be obtained at a concentration of ∼7 mM Mg²⁺. At higher concentrations (e.g., 20 mM), Mg²⁺ ions resulted in a decrease of channel activity relative to lower concentrations (e.g., 7 mM). It has been shown that, at high concentrations, divalent cations can trigger aggregation of PIP₂ molecules (Flanagan et al. 1997), a result that could account for the effects of high Mg²⁺ concentrations on channel activity.

In another set of experiments, we showed that Mg²⁺, like Na⁺ gating, can occur independently of G proteins. Patches excised from oocytes coexpressing the βARK-PH domain and GIRK channels were exposed to PIP₂ (2.5 μM) and subsequently to Mg²⁺ ions. In these patches, GTPγS (10 μM) was unable to activate the GIRK channels, giving a NP _o of 0.08 ± 0.03, identical to the NP _o measured in PIP₂ (0.078 ± 0.02). Mg²⁺ ions (1 mM) could increase the channel activity approximately sixfold (n = 4, data not shown) above the activity measured in PIP₂, showing that Mg²⁺ gating could proceed independently of G_βγ gating.

These results suggest that when modified by ATP or PIP₂, GIRK channels become sensitive to either Na⁺ or Mg²⁺ ions.

Mg²⁺ Ion Gating Occurs at a Site Distinct from that of Na⁺ Action

Recent work has identified an aspartate amino acid residue as the site of action of Na⁺ ions on GIRK channels, GIRK2 (D228) and GIRK4 (D223) (Ho and Murrell-Lagnado 1999; Zhang et al. 1999). Moreover, it was shown that Na⁺ sensitivity lies entirely with the heteromeric partners of GIRK1, as this channel possesses an asparagine instead of an aspartate residue at the equivalent position. We used the point mutant GIRK4(S143T) (referred to as GIRK4*) that allows for high levels of activity of homotetrameric GIRK4 channels (Vivaudou et al. 1991) to test for Na⁺ and Mg²⁺ sensitivity. GIRK4* channel activity shows high sensitivity to both Na⁺ and Mg²⁺. Fig. 5 shows that indeed GIRK4*(D223N) loses its sensitivity to Na⁺ ions (20 mM). However Mg²⁺ ion (1 mM) sensitivity was intact (Fig. 5 A). Summary data are shown in Fig. 5 B. These data indicate that Na⁺ and Mg²⁺ ions act at distinct sites to activate GIRK channels.

Figure 5.

Mg²⁺ ions act at a site distinct of that used by Na⁺ ions. (A) Single channel activity (NP _o, bin = 5 s) plotted as a function of time. An inside-out patch from an oocyte expressing GIRK4*(D223N) (see text) and human muscarinic receptor type 2 receptor was exposed to 2.5 μM PIP₂ and 20 mM Na⁺ or 1 mM Mg²⁺ ions, and channel activity was recorded. The membrane potential was kept at −80 mV. The pipette solution contained 5 μM ACh. (B) Summary data plotting mean NP _o of four experiments such as that shown in A. The mean NP _o values were 0.14 ± 0.1 (mean ± SEM) in control conditions, 0.21 ± 0.06 in the presence of PIP₂, 0.21 ± 0.08 in the presence of PIP₂ and Na⁺ ions, and 2.97 ± 0.31 in the presence of PIP₂ and Mg²⁺ ions.

Mg²⁺ Ions Reduce the Conductance of the GIRK Channels

We observed, particularly at high concentrations (>5 mM), that internal Mg²⁺ ions reduced the amplitude of single GIRK channel currents. In Fig. 6 A, the activity of the coexpressed channel subunits GIRK1/GIRK4 from an inside-out patch was recorded at −120 mV. After activation by 10 μM GTPγS, channel activity was recorded in a solution containing 1 mM Mg²⁺ ions, showing an approximate amplitude of −3.2 pA. When the solution applied to the patch was switched to one containing 20 mM Mg²⁺ ions, the amplitude of the single openings was rapidly reduced to a lower value, approximately −2.5 pA (n = 5). In Fig. 6 B, the activity of native K_ACh channels in an inside-out patch from an atrial cell was recorded at −90 mV. After exposure to 5 μM PIP₂, the patch was perfused with a solution containing 20 mM Mg²⁺ ions, giving an amplitude of approximately −2.2 pA. When the solution applied to the patch was switched to one containing 20 mM Na⁺ and 1 mM Mg²⁺ ions, the channel amplitude immediately increased to a value of approximately −3.5 pA. This amplitude was also obtained in control conditions, where 1 mM Mg²⁺ ions were present (n = 5). The reduction in the single-channel amplitude was observed at various voltages. Since it was present at negative potentials (i.e., −80, −90, and −120 mV) where no rectification occurs, it is likely to proceed by a mechanism distinct from that of the rectification phenomenon. Mg²⁺ ions at high concentrations also decreased the amplitude of GIRK single channels when applied together with Na⁺ ions (data not shown). Thus, regardless of their ability to gate GIRK channels (see Fig. 3 and Fig. 7), Mg²⁺ ions at high concentrations (>5 mM) show a clear inhibition on single-channel current amplitudes. These data suggest that the inhibitory effect of Mg²⁺ ions on the single-channel amplitude was not dependent on their ability to gate the channel.

Figure 6.

High concentrations of Mg²⁺ ions reduce the GIRK single-channel currents. (A) Single-channel records of GIRK channels in an inside-out patch excised from an oocyte expressing GIRK1/GIRK4. The membrane was clamped at −120 mV and 5 μM acetylcholine was present in the pipette. The channel was preactivated by 10 μM GTPγS and the current traces shown were recorded after washout of the GTP analogue. The switch between bath solutions containing 1 and 20 mM Mg²⁺ is visualized by the arrow (and by the corresponding electrical artifact on the record) on top of the second current trace. Associated all-point histogram plots indicate the amplitudes resulting from the various activity levels ranging from closed to multiple open levels and are shown for the first (1 mM Mg²⁺) and third (20 mM Mg²⁺) current traces. Data points are shown on a logarithmic scale ranging from 5 to 50,000. (B) K_ACh channel activity in an inside-out patch from a cardiac cell. The membrane was held at −90 mV and the pipette contained 5 μM acetylcholine. The patch was preincubated with 5 μM PIP₂ and the current traces shown were recorded after the washout of PIP₂. The arrow on top of the second current trace visualizes the switch between bath solutions containing 20 mM Mg²⁺ and 20 mM Na⁺ + 1 mM Mg²⁺. Associated all-point histogram plots are shown for the first (20 mM Mg²⁺) and third (20 mM Na⁺ + 1 mM Mg²⁺) current traces. Data points are shown on a logarithmic scale ranging from 20 to 50,000.

Synergistic Interactions among Ions and G Protein Subunits in Gating GIRK Channels

ATP modification of GIRK channels (native or recombinant) is likely to proceed through changes in the level of membrane PIP₂ in the local environment of the channel (Huang et al. 1998; Sui et al. 1998). In Fig. 7 A, the activity of recombinant GIRK1/GIRK4 channels was not increased by the application of 2.5 μM PIP₂ alone. Mg²⁺ ions (10 mM), applied with PIP₂, stimulated activity >40-fold. As shown previously (Sui et al. 1998), Na⁺ ions (10 mM) were able to gate the channel in the presence of PIP₂, resulting in activity equivalent to that obtained with Mg²⁺ ions. When applied together, in the presence of PIP₂, Mg²⁺ and Na⁺ ions (10 mM each) showed a synergistic effect stimulating channel activity >200-fold.

We next tested whether Mg²⁺ ions, like Na⁺ ions (Fig. 2), can further gate GIRK channels after channel activation by GTPγS, under conditions that do not maintain PIP₂ at constant high levels. Fig. 7 B shows that the basal activity of GIRK1/GIRK4 recombinant channels was not affected by Mg²⁺ ions. 10 μM GTPγS increased the activity of the channel approximately fivefold above basal levels. After GTPγS washout, the channel activity was stable and, when applied to the patches, Mg²⁺ ions were unable to increase channel activity further. In contrast, Na⁺ ions (10 mM) increased activity by another twofold above the GTPγS effect. When Mg²⁺ ions were applied together with Na⁺ ions, no further increase in channel activity above the levels obtained with Na⁺ ions was seen. Thus, G protein activation sensitized the GIRK channels to gating by Na⁺ ions, but not Mg²⁺ ions.

In Fig. 7 C, we show the effects of Mg²⁺ and Na⁺ ions after stimulation of the channel by GTPγS under conditions that kept PIP₂ at a constant high level. As shown earlier, in the absence of Mg²⁺ and Na⁺ ions, PIP₂ was not able to increase the basal activity of the GIRK channels. When Mg²⁺ ions (10 mM) were applied to the patches in the presence of PIP₂, a greater than eightfold increase over control or PIP₂ activity levels occurred. Mg²⁺ and Na⁺ ions (each 10 mM) in combination could raise channel activity by 50-fold over control levels. We then applied GTPγS and studied the effects of ions on G protein–stimulated channel activity in the continuous presence of PIP₂. GTPγS was able to activate the channel >14-fold above control basal levels. After washout of GTPγS, channel activity was stable. When Mg²⁺ ions (10 mM) were applied to the patches after the GTPγS treatment in the continuous presence of PIP₂, they could enhance channel activity to levels >100-fold higher than those obtained under control conditions. Thus, in the continuous presence of PIP₂, this high level of activity was greater than that obtained with Mg²⁺ or GTPγS alone or their sum, suggesting synergistic interactions among the three molecules. Finally, when Mg²⁺ and Na⁺ ions were applied together, the channel total activity was increased 400-fold compared with control.

Similar data were obtained when the G protein β₁γ₇ subunits rather than GTPγS were used. In three cells, the total channel activity measured as the mean NP _o was 0.027 ± 0.023 in control conditions, 0.022 ± 0.02 in the presence of 2.5 μM PIP₂, 0.12 ± 0.09 in the presence of PIP₂ and 10 mM Mg²⁺ ions, and 1 ± 0.55 in the presence of PIP₂ and Mg²⁺ and Na⁺ ions. When 20 nM β₁γ₇ was applied in the presence of PIP₂, it gave a steady state activity of the channel corresponding to a mean NP _o of 0.25 ± 0.12. In the continuous presence of PIP₂ and after stimulation of the channel by βγ subunits, the mean NP _o was 1.23 ± 0.23 in the presence of Mg²⁺ ions and 2.61 ± 0.24 in the presence of Mg²⁺ and Na⁺ ions. It should be noted that the differences in channel activity (mean NP _o) for the same condition applied to the patches (for example PIP₂ + Mg²⁺ in Fig. 7A and Fig. C) may be related to differences in the level of channel expression between different batches of oocytes. Taken together, these data make four points. (a) Mg²⁺ ions can gate the channel after modification by PIP₂. At a concentration of 10 mM, their gating potency is comparable with that of 10 mM Na⁺ ions. (b) When applied together, Mg²⁺ and Na⁺ ions show synergistic effects, resulting in levels of activity higher than those induced by each of the ions separately or their summed responses. (c) After activation by GTPγS (in the absence of exogenous PIP₂ or MgATP), the GIRK channels are not further gated by Mg²⁺ ions, suggesting an important difference between Mg²⁺ and Na⁺ ions in gating these channels. (d) After channel modification by the combination of PIP₂ and GTPγS, Mg²⁺ ions do stimulate the GIRK channel activity to higher levels than those obtained with PIP₂ alone, suggesting that PIP₂ renders the βγ-activated channels sensitive to gating by Mg²⁺ ions.

DISCUSSION

In the present study, we have shown that Mg²⁺ ions at physiological concentrations are additional activators of G protein–gated potassium channels. These K⁺ channels can be activated independently either by the βγ subunits of GTP-binding proteins (Logothetis et al. 1987) or by intracellular ions, such as Na⁺ (Sui et al. 1996) or Mg²⁺ ions. Activation by either G protein subunits or ions shows an absolute dependence on the presence of PIP₂ (Sui et al. 1998). Specific combinations of these molecules show synergism and suggest differential dependence on the level of PIP₂ for channel activation. This complex dependence of K⁺ channel activity on G proteins, Mg²⁺, Na⁺ ions, and PIP₂ could serve to “fine tune” channel activity during physiological and pathophysiological conditions, where changes in the relative concentrations of these molecules might occur.

GIRK Channel Activation by Na⁺ Ions Can Be Independent of G Protein Subunit Involvement

Previous results from our laboratory showed that intracellular solution containing MgATP/Na⁺ was able to stimulate K⁺ channel activity in the absence of acetylcholine in the pipette, suggesting a G protein–independent mechanism of activation (Sui et al. 1996). Subsequently, it was further demonstrated that the ATP dependence of G protein–sensitive K⁺ channels, as well as of other inwardly rectifying channels, involved phosphoinositide formation, particularly PIP₂ (Huang et al. 1998; Sui et al. 1998). In addition, it was reported that G protein activation of the K⁺ channel showed an absolute dependence on PIP₂ (Sui et al. 1998). In the present study, we show that impairment of G protein subunit activation of the channel (by binding and competing away G_βγ from the channel with either QEHA perfusion or βARK-PH coexpression) did not prevent the MgATP/Na⁺ stimulation of activity (Fig. 1). Thus, we have provided further evidence that Na⁺ ion gating of the channel modified by ATP (or PIP₂) can be independent of G protein subunit activation.

Mg²⁺ Gating of the G Protein-gated K⁺ Channel

Mg²⁺ ions have been shown to play an essential role in the rectification properties of inwardly rectifying K⁺ channels. Unitary current–voltage relations for G protein–sensitive K⁺ channels become ohmic if the internal face of the patch is exposed to Mg²⁺-free solutions. Inward rectification is restored when Mg²⁺ is reintroduced in the bathing solutions (Matsuda 1991; Kurachi et al. 1992; Nichols and Lopatin 1997).

Mg²⁺ ions are involved in many other reactions as essential cofactors. Kurachi et al. 1986 showed that G protein activation of the native G protein–sensitive K⁺ channel was absolutely dependent on Mg²⁺, possibly due to the requirement of Mg²⁺ for the binding of GTP to the G_α subunit (also see Logothetis et al. 1987). More recently, it has been appreciated that Mg²⁺-dependent processes of ATP hydrolysis (likely to be involved in phosphorylation–dephosphorylation of phosphoinositides) regulate channel activity (Sui et al. 1996, Sui et al. 1998; Huang et al. 1998).

Our present data show that Mg²⁺ ions, in addition to their involvement in the processes mentioned above, are able to activate the ATP- or PIP₂-modified G protein–sensitive channel (Fig. 3 and Fig. 7). In the presence of PIP₂, similar concentrations of Mg²⁺ and Na⁺ ions yielded comparable levels of channel activity, suggesting equivalent gating abilities for both ions. Since PIP₂ mimics the MgATP effects on the channel, we have been able to study directly Mg²⁺ gating effects.

Mutation of the amino acid responsible for Na⁺-ion activation of GIRK channels did not interfere with Mg²⁺-ion activation. This result strongly suggests that Mg²⁺ and Na⁺ ions act on distinct sites to gate the channel.

Our data also show that Mg²⁺ ions reduced the conductance of the G protein–gated channels in a manner independent of their stimulatory effect on gating. Since this inhibitory effect of Mg²⁺ ions on conductance was present at negative potentials (−120, −90, and −80 mV), where no rectification is occurring (Kurachi et al. 1992), it is unlikely that the two processes proceed through a single mechanism. This effect of partial block on channel conductance suggests that Mg²⁺ ions act at a site located very near the pore. Chuang et al. 1997 described a chronic inhibition of the IRK3 inward rectifier channel by internal Mg²⁺ ions, which is independent of the rectification process and is voltage independent. However, the on and off rates of this inhibition were slow (in the minute range) and no reduction of the single-channel conductance was reported. Under our conditions, the blocking effect of Mg²⁺ ions occurred much more rapidly, in the range of seconds (Fig. 6).

Synergism Among G Proteins and Ions in the Gating of GIRK Channels

At higher PIP₂ concentrations, the combination of Na⁺ and Mg²⁺ ions resulted in a stimulation of channel activity that was greater than the sum of their individual effects, suggesting synergistic interactions of these ions on channel gating (Fig. 7 A).

Na⁺ ions gate the K⁺ channel in the presence of hydrolyzable ATP or PIP₂ (Sui et al. 1996, Sui et al. 1998; Fig. 3 and Fig. 7 A). In the present study, we show that (shortly after patch excision in solutions that do not replenish or supply PIP₂) application of G_βγ subunits, but not of Na⁺ or Mg²⁺ ions, results in stimulation of channel activity (Fig. 2 and Fig. 7 B). Our previous study (Sui et al. 1998) showed that in the absence of PIP₂ in the membrane (e.g., by its complete hydrolysis by exogenous PLCβ₂) no gating molecule (e.g., G_βγ or Na⁺) could activate the channel. In the present experiment under conditions that we do not expect to have depleted PIP₂, G_βγ subunits caused a much greater stimulation of activity than Na⁺ or Mg²⁺ ions. This result suggests that the dependence on PIP₂ for channel gating is greater for Mg²⁺ and Na⁺ than for G_βγ. Under these conditions, we find that Na⁺ ions do stimulate channel activity after preactivation by GTPγS or by purified G_βγ subunits (Fig. 2 and Fig. 7 B). This result suggests that G_βγ activation sensitizes the K⁺ channel gating to Na⁺ ions. Moreover, in such experiments, G_βγ subunits and Na⁺ ions act synergistically in gating the channel. Interestingly, Mg²⁺ ions were unable to gate the channel after channel preactivation by GTPγS. These data underscore an interesting difference in the gating of this channel by ions, namely at low PIP₂ levels G_βγ subunits synergize with Na⁺ but not Mg²⁺ ions to gate the channel.

This difference of the two ions on channel gating is lost at higher PIP₂ concentrations (Fig. 7 C). In such experiments, not only were the synergistic effects of the ions shown in Fig. 7 A reproduced, but also Mg²⁺ as well as Na⁺ ions cooperated with G_βγ. When applied in combination, all three gating particles showed synergistic effects (Fig. 7 C).

We have previously shown that block of the Na⁺/K⁺ pump activates K_ACh in atrial myocytes with kinetics similar to those seen for Na⁺ accumulation resulting from the block of the pump (Sui et al. 1996). Thus, it is likely that the effects of cardiac glycosides on cardiac rhythm involve the Na⁺-sensitive K_ACh channels. Under physiological conditions, local variations of [Na⁺]_i (e.g., during an action potential) and possibly [Mg²⁺]_i could provide a sensitive and fast control of the GIRK channel gating and activity. The synergism among ions and G_βγ subunits implies that variations in the local levels of these molecules could have a profound impact on the dynamic range of GIRK channel activity under normal or pathophysiologic states.

A Gating Model for GIRK Channels

Channel binding sites for PIP₂, G_βγ, and Na²⁺ ions have been identified (Huang et al. 1995; Huang et al. 1997; Kunkel and Peralta 1995; He et al. 1999). We postulate that additional distinct sites exist to completely account for the effects of gating molecules on channel activity. Fig. 8 shows the closed channel state C₀ in the absence of PIP₂. GIRK channels interact weakly with PIP₂, and as a result PIP₂ does not directly activate these channels (closed state C₁). In the absence of PIP₂, gating molecules such as G_βγ, Na⁺, or Mg²⁺ are unable to activate the channel (closed state C₂). However, in the presence of PIP₂, any of the gating molecules can cause channel activation.

Figure 8.

A model depicting gating of GIRK channels by the combination of PIP₂ with gating molecules G_βγ and/or Na⁺ and Mg²⁺ ions. (C_o) Channel closed state, in the absence of PIP₂ in the plasma membrane and gating molecules. (C₁) Channel closed state, in the presence of PIP₂ in the membrane GIRK channels experience weak interactions that in the absence of gating molecules are not of sufficient strength to gate the channel. (C₂) Channel closed state, gating molecules can interact with the channel at distinct sites but in the absence of PIP₂ they fail to gate the channel. (O) Channel open state, gating molecules in the presence of membrane PIP₂ can activate the channel and show synergism.

We envision two possible mechanisms for the synergistic action of gating molecules to activate the channel. Ions and G_βγ subunits maybe exerting their combined effects by synergistic interactions of channel sites with PIP₂. Published reports have already suggested a stronger interaction of channel with PIP₂ in the presence of either G_βγ subunits or Na⁺ ions (Huang et al. 1998; Zhang et al. 1999) Alternatively, the gating molecules could be inducing conformational changes, affecting gating directly, independently of PIP₂ interactions. Although PIP₂ is absolutely required for gating molecules to be effective, we have seen that at low PIP₂ concentrations G_βγ, unlike Na⁺ or Mg²⁺ ions, can still gate the channels. This result suggests a stronger influence of G_βγ than of Na⁺ or Mg²⁺ ions on channel gating, possibly proceeding in a PIP₂-independent manner. Further work will be required to distinguish between these possibilities.