Molecular mechanism for sodium-dependent activation of G protein-gated K+ channels


  • Ivan H. M. Ho,

    1. Department of Pharmacology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QJ, UK
    Search for more papers by this author
  • R. D. Murrell-Lagnado

    Corresponding author
    1. Department of Pharmacology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QJ, UK
    • Corresponding author
      R. D. Murrell-Lagnado: Department of Pharmacology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QJ, UK. Email:

    Search for more papers by this author


  • 1G protein-gated inwardly rectifying K+ (GIRK) channels are activated independently by Gβγ and internal Na+ via mechanisms requiring phosphatidylinositol phosphates. An aspartate (Asp) at position 226 in GIRK2 is crucial for Na+-dependent activation of GIRK1-GIRK2 heteromeric channels. We expressed wild-type and mutant GIRK1-GIRK2 channels in Xenopus oocytes and tested the effects of Na+ and neutralizing Asp226 on the functional interactions of the channels with phosphatidylinositol 4,5-bisphosphate (PIP2).
  • 2The rate of inhibition of GIRK1-GIRK2 currents by application of anti-PIP2 antibody to inside-out membrane patches was slowed > 2-fold by the D226N mutation in GIRK2 and by increasing internal [Na+]. The reverse mutation in GIRK1 (N217D) increased the rate of inhibition.
  • 3The dose-response relationship for activation by purified PIP2 was shifted to lower concentrations in the presence of 20 mM Na+.
  • 4Three synthetic isoforms of PIP2, PI(4,5)P2, PI(3,4)P2 and PI(3,5)P2, activated GIRK channels with similar potencies.
  • 5We conclude that Na+ directly interacts with Asp226 of GIRK2 to reduce the negative electrostatic potential and promote the functional interaction of the channels with PIP2.

GIRK channels in the brain couple to many different neurotransmitter receptors to regulate neurone excitability (Andrade et al. 1986; Trussell & Jackson, 1987; Penington et al. 1993). Activation of the seven transmembrane receptors catalyses the release of Gβγ subunits from Gαi/o subunits, and Gβγ directly interacts with GIRK channels to increase their open probability (Logothetis et al. 1987; Wickman et al. 1994). Four mammalian GIRK subunits have been cloned and they form one of at least seven distinct subfamilies of inwardly rectifying K+ channels (Kir). Although the different Kir subtypes are regulated differently by a variety of intracellular factors, recent reports indicate that many of these channels require membrane phosphatidylinositol phosphates for activity (Hilgemann & Ball, 1996; Fan & Makielski, 1997; Baukrowitz et al. 1998; Huang et al. 1998; Shyng & Nichols, 1998; Sui et al. 1998). In excised membrane patches, Kir channels run down and can be reactivated by applying either MgATP or phosphatidylinositol 4,5-bisphosphate (PIP2). PIP2 has been shown to bind to the proximal C-terminal region of Kir subunits and PIP2 antibodies inhibit channel activity at a rate that correlates with the binding affinity (Huang et al. 1998; Liou et al. 1999).

The binding of the anionic PIP2 to Kir channels appears to involve electrostatic interactions. A negatively charged head group on the phospholipid is required for activation of ATP-sensitive K+ (KATP) channels, and PIP2 is more effective than PIP (Hilgemann & Ball, 1996; Fan & Makielski, 1997; Baukrowitz et al. 1998; Shyng & Nichols, 1998). Screening negative charges with polycations inhibits stimulation by PIP2, as does neutralizing a highly conserved arginine within the proximal C-terminal region of ROMK1 and Kir6.2 (Fan & Makielske, 1997; Huang et al. 1998; Shyng & Nichols, 1998). In the absence of Gβγ, the interaction between PIP2 and GIRK channels is of lower affinity than the interaction between PIP2 and other Kir channels. The activation of GIRK channels by Gβγ appears to involve a substantial increase in the PIP2-channel interactions (Huang et al. 1998; Sui et al. 1998). The conformational rearrangements within GIRK channels that are induced by Gβγ binding are not known, although a cysteine lying in close proximity to the putative PIP2 binding region of GIRK4 has been shown to be important for Gβγ-dependent activation of GIRK1-GIRK4 heteromeric channels (Krapivinsky et al. 1998).

GIRK channels are activated by a rise in intracellular [Na+] via a mechanism that is independent of G proteins but does require either MgATP or PIP2 (Sui et al. 1996). We recently identified an amino acid within the GIRK2 subunit, Asp226, that is crucial for Na+-dependent activation of both GIRK2 homomeric channels and GIRK1-GIRK2 heteromeric channels (Ho & Murrell-Lagnado, 1999). GIRK4 also has an aspartate at the equivalent position whereas GIRK1 has an asparagine (Asn217). Interestingly, this amino acid is close to the putative PIP2 binding region.

We previously hypothesized that Na+-dependent activation of GIRK1-GIRK2 channels might involve the direct interaction of Na+ with Asp226, resulting in a decrease in the negative charge in the vicinity of the PIP2 binding site and a corresponding increase in PIP2 binding affinity (Ho & Murrell-Lagnado, 1999). Arguing against this being the sole mechanism for Na+-dependent activation is the observation that even after prolonged application of a high concentration (50 μM) of PIP2 to GIRK channels in excised membrane patches, subsequent application of Na+ could further enhance channel activity (Ho & Murrell-Lagnado, 1999). In this study, we used a PIP2 antibody to directly test the effect of increasing [Na+] and mutating Asp226 to asparagine on the functional interactions of PIP2 with GIRK channels. We show that the two manipulations slowed the inhibition of GIRK currents by the PIP2 antibody to a similar extent. Thus Na+ appears to effectively neutralize the negative charge on Asp226, thereby promoting the interaction with PIP2. This is likely to be at least part of the mechanism by which the G protein-independent activity of the channel is increased.


Molecular biology

Rat GIRK1 and GIRK2 cDNAs were subcloned into the pBG7.2 vector (gift from R. W. Aldrich). The human m2 muscarinic receptor was inserted into pBluescript KS(II)+. The details of construction of N217D in GIRK1 and D226N in GIRK2 are as described previously in Ho & Murrell-Lagnado (1999).


Recombinant channel subunits and the human m2 muscarinic receptor were co-expressed in Xenopus oocytes as described previously (Ho & Murrell-Lagnado, 1999). Briefly, stage V/VI oocytes were obtained by ovariectomy from Xenopus laevis frogs anaesthetized with 0.3 % (w/v) 3-amino benzoic acid (Sigma). Following the first ovariectomy the frog was allowed to recover from anaesthesia before being returned to the maintenance tank. Prior to a second ovariectomy, the frog was terminally anaesthetized in 0.6 % (w/v) anaesthetic for 1 h. It was subsequently killed by opening the chest cavity and removing the heart. Oocytes were dissociated from connective tissue using 0.3 % collagenase IA (Sigma) in calcium-free OR2 solution containing (mM): 82.5 NaCl, 2.5 KCl, 1 MgCl2 and 5 Hepes (pH 7.6). In vitro transcription of the linearized cDNAs was as described previously in Stevens et al. (1997). A similar total amount of channel subunit cRNAs was injected into each oocyte together with m2 receptor cRNA. Two-electrode voltage-clamp recordings were performed 3–5 days after microinjection. Oocytes were continually perfused with standard recording solution: (1) ND96 which contained (mM): 96 NaCl, 2 KCl, 1.8 CaCl2, 1 MgCl2 and 5 Hepes (pH 7.6 with NaOH), or (2) high K+ (hK) solution which contained (mM): 90 KCl, 1 MgCl2, 1 CaCl2 and 1 Hepes (pH 7.4 with KOH) as indicated. For patch-clamp experiments, extracellular (pipette) solution contained (mM): 96 KCl, 1 MgCl2, 1.8 CaCl2 and 10 Hepes, pH 7.2 with KOH. The intracellular (bath) solution contained (mM): 96 KCl, 10 EGTA and 20 Hepes, pH 7.2 with KOH. For antibody experiments, 0.5 mM MgCl2, 0.2 mM NaF and 0.1 mM Na3VO4 were added to the bath solution (FV solution). After addition of 5 mM MgATP to the FV solution, the free Mg2+ concentration was calculated to be ∼2 mM. NaCl (20 or 60 mM) was added to the bath solution without compensation for changes in osmolarity and ionic strength. Solutions containing MgATP were prepared freshly each day and the pH was readjusted to 7.2 after addition of MgATP. Solutions containing purified PIP2 (a gift from R. F. Irvine, Department of Pharmacology, University of Cambridge) or synthetic PIP2 analogues (Echelon Research Laboratories, Salt Lake City, UT, USA) were prepared freshly each day by sonication (Ho & Murrell-Lagnado, 1999). Reconstituted monoclonal PIP2 antibody was from PE Biosystems (UK) and was diluted 40-fold into the bath solution. Patch pipettes were coated with wax and had resistances of between 0.3 and 1.0 MΩ. Currents were recorded at −80 mV, sampled at 10 kHz and filtered at 2 kHz. Ten second ramp tests from −140 to +80 mV were performed throughout the recording to check that there were no non-inwardly rectifying channels present in the patch. Data were acquired using Pulse acquisition software (version 8.11; HEKA Electronics).

Data analysis

Exponential curves were fitted to the data with the Levenberg-Marqurdt least-squares algorithm, using Igor Pro software (version 3; Wavemetrics, Lake Oswego, OR, USA). To measure channel activity in multichannel patches, the mean current of each continuous recording segment was calculated by first subtracting the baseline current and then summing the amplitudes of all of the sample points and dividing by the number of sample points within the continuous recording. Statistical analyses were performed with Student's unpaired t test using InStat software (version 2.01; GraphPad, San Diego, CA, USA). Data were assumed to be normally distributed. Averaged data are expressed as means ±s.e.m.


Inhibitory effect of Asp226 in GIRK2 on GIRK1-GIRK2 heteromeric channels

Huang et al. (1998) proposed that the low basal activity of GIRK channels arises from their low binding affinity for PIP2 in the absence of Gβγ. If the negatively charged Asp226 in GIRK2 inhibits the interaction of the channels with PIP2 by direct electrostatic interactions, then we would predict that neutralizing this charge by substituting asparagine for aspartate would increase PIP2-channel interactions and hence increase basal activity. Correspondingly, we would also predict that substituting aspartate for Asn217 in GIRK1 might further reduce basal channel activity as compared with the wild-type GIRK1-GIRK2 channel. To test this hypothesis we expressed GIRK1-GIRK2D226N, GIRK1N217D-GIRK2 and the wild-type GIRK1-GIRK2 channels in Xenopus oocytes, together with the m2 muscarinic receptor, and compared the whole-cell basal and agonist-stimulated currents. Figure 1A shows currents recorded at −80 mV, 3 days after injection of the cRNAs indicated. For GIRK1N217D-GIRK2, the high K+-induced Ba2+-sensitive (basal) currents were extremely small, whereas the GIRK1-GIRK2D226N basal currents were ∼14-fold larger and of similar magnitude to those recorded from oocytes expressing the wild-type channel (Fig. 1B). To obtain an estimate of relative basal activity, independent of the level of channel expression, the basal current was expressed as a fraction of the total current measured in the presence of 3 μM carbachol (Fig. 1C). For the GIRK1-GIRK2D226N channel, basal activity was ∼40 % of agonist-stimulated activity; for the wild-type channel it was ∼22 % and for the GIRK1N217D-GIRK2 channel it was only ∼6 % of agonist-stimulated activity. Thus aspartate at the 226/217 position does appear to have an inhibitory effect on basal GIRK channel activity, whereas asparagine at this position enhances the ratio of basal to agonist-stimulated activity.

Figure 1.

The effect of negative charges at position 226/217 on the basal activity of the wild-type GIRK1-GIRK2 (G1/G2) and the mutant GIRK1-GIRK2D226N (4N) and GIRK1N217D-GIRK2 (4D) heteromeric channels

A, continuous currents at −80 mV, recorded by two-electrode voltage clamp from Xenopus oocytes expressing the indicated channels and the m2 muscarinic receptor. Solution changes are indicated by the bars above the current traces. CCh (carbachol), 3 μM; Ba2+, 1 mM. Dotted line indicates zero current level. B, Ba2+-sensitive basal and 3 μM CCh-induced current amplitudes at −80 mV (n= 6 for each group). C, basal currents expressed as a fraction of the total current in the presence of 3 μM CCh minus the Ba2+-insensitive current, for the wild-type and mutant channels. * Values significantly different from each other, P < 0.001 (n= 6 for each group).

Promoting the functional interactions of PIP2 with GIRK1-GIRK2 heteromeric channels by internal Na+

To compare the effective interactions of PIP2 with the wild-type and mutant channels, we measured the time course of inhibition of the channels by a PIP2-specific monoclonal antibody (Fig. 2). In inside-out patches, the channel activity was maintained at a relatively high level in the FV solution containing 5 mM MgATP. The 40-fold diluted anti-PIP2 antibody was applied in the continuous presence of MgATP, and completely inhibited the channel activity for both wild-type and mutant channels. In the absence of internal Na+, the time course of inhibition by the antibody was 2.4 ± 0.7 s (n= 3) for GIRK1N217D-GIRK2, 3.3 ± 0.5 s (n= 4) for GIRK1-GIRK2, and 6.8 ± 1.1 s (n= 3) for GIRK1-GIRK2D226N (Fig. 2). Thus, with Asp226 neutralized, the rate of inhibition by the antibody was significantly slower than for the wild-type channel (P < 0.05). Increasing internal [Na+] (60 mM) significantly slowed the rate of inhibition of the wild-type channel. The time constant for the exponential fit was 6.9 ± 0.8 s, (n= 3), which is very similar to the value for the GIRK1-GIRK2D226N mutant channel. In contrast, simply increasing the osmolarity of the bath solution by application of 120 mM sucrose had no significant effect on the time course of inhibition of GIRK1-GIRK2 currents by the antibody (τ= 3.86 ± 0.12 s, n= 3, P > 0.05). Increasing internal [Na+] blocks GIRK1-GIRK2D226N channel currents (Ho & Murrell-Lagnado, 1999) and so we did not test the effects of 60 mM Na+ on the rate at which the antibody inhibits the mutant channel currents. The results suggest that intracellular Na+ enhances the functional interactions of PIP2 with GIRK1-GIRK2 channels by effectively neutralizing the negative charge of Asp226.

Figure 2.

Inhibitory effect of anti-PIP2 antibody on the wild-type GIRK1-GIRK2 and the mutant GIRK1-GIRK2D226N and GIRK1N217D-GIRK2 heteromeric channels

A, the anti-PIP2 antibody was applied to inside-out patches in the continuous presence of 5 mM MgATP in FV solution with or without 60 mM Na+ as indicated. The time course of inhibition was fitted with single exponential curves. B, currents were normalized to the mean amplitude prior to antibody application, and overlayed. For clarity, the fitted curves for both GIRK1-GIRK2D226N and GIRK1N217D-GIRK2 channels are plotted, and the wild-type GIRK1-GIRK2 currents were smoothed. C, mean values of the time constants for the monoexponential fits. *P < 0.05 (n= 3–4 for each group).

Activation of GIRK1-GIRK2 channels by PIP2 analogues in the absence and presence of internal Na+

Further evidence for the enhancement of PIP2-channel interactions by Na+ is provided by the dose-response relationship for activation of GIRK1-GIRK2 channels by purified PIP2 liposomes in the absence and presence of 20 mM internal Na+ (Fig. 3). Channel activity ran down rapidly following patch excision into a solution lacking MgATP (Fig. 3A). Na+ was applied in the presence of 2 mM Mg2+ because Mg2+ is reported to act synergistically with Na+ to activate GIRK channels (Petit-Jacques et al. 1999). Previous patch-clamp experiments included ∼2 mM free Mg2+ in the bath solution. In the absence of ATP and PIP2, 20 mM Na+ had little effect on channel activity. After washing out Na+ and Mg2+, 1 μM PIP2 was applied for 3 min and it produced a very small but significant increase in the mean current, whereas subsequent application of Na+ and Mg2+ produced a much greater increase in current. We assume when applying PIP2 to the bath that it is the concentration within the membrane that determines the level of channel activity and therefore the dose-response relationship that we measured should be interpreted with caution. However, the response to Na+ plus Mg2+ clearly increased following application of 10 μM PIP2 as compared with 1 μM PIP2, whereas 50 μM PIP2 had no additional effect. In contrast, in the absence of Na+ and Mg2+, 50 μM PIP2 had a much greater effect than 10 μM PIP2, suggesting that the dose-response relationship is shifted towards higher concentrations (Fig. 3B). We were reluctant to use concentrations of PIP2 greater than 50 μM because of the possibility of multilayer liposomes at very high concentrations.

Figure 3.

Dose-response relationship for activation of GIRK1-GIRK2 channels by purified PIP2 liposomes in the absence and presence of internal Na+

A, GIRK1-GIRK2 channel activity ran down following patch excision into MgATP-free bath solution. C/A is cell-attached and I/O is inside-out configuration. Bars indicate the periods of application of the different reagents. Purified PIP2 was applied for 3 min each time and the mean current was measured for the final minute of this period. B, dose-response relationship for PIP2 in the absence and presence of Na+ and Mg2+. Currents were normalized to the cell-attached level. Asterisks indicate a significant difference from the current in the absence of PIP2, Na+ and Mg2+ (*P < 0.01 and **P < 0.001; n= 3–15 for each group).

PI(4,5)P2 is the predominant naturally occurring analogue of PIP2 and it is therefore likely to play an important physiological role in the regulation of Kir channels. However, there are two other analogues that exist, PI(3,4)P2 and PI(3,5)P2, and these may also play an important role, particularly if they have a higher potency than the PI(4,5)P2 analogue and/or a tendency to colocalize with GIRK channels. We tested whether or not Na+ acts synergistically with each analogue to activate GIRK1-GIRK2 channels (Fig. 4). There was no significant effect with 1 μM of the synthetic PIP2 analogues in the absence of Na+ and Mg2+, but there was a significant, albeit small, increase in the mean current upon application of 10 μM of the analogues for 3 min. For all three analogues, subsequent application of 20 mM Na+ and 2 mM Mg2+ produced a dramatic increase in the activity of the GIRK channels. The results suggest that all three PIP2 analogues interact with GIRK channels in a similar way to activate the channel.

Figure 4.

Activation of GIRK1-GIRK2 channels by synthetic PIP2 analogues

The experimental protocol was the same as that shown in Fig. 3A. Currents were normalized to the cell-attached level. Asterisks indicate a significant difference from currents in the absence of PIP2, Na+ and Mg2+ (*P < 0.05, **P < 0.01 and ***P < 0.001; n= 3–5 for each group).


We have previously shown that the GIRK2D226N mutation abolished Na+-dependent activation of GIRK2 homomeric and GIRK1-GIRK2 heteromeric channels but did not prevent the activation of these channels by agonist stimulation of coexpressed m2 muscarinic receptors (Ho & Murrell-Lagnado, 1999). Here we show that this mutation promotes the functional interaction between PIP2 and GIRK1-GIRK2 channels and that raising internal [Na+] has a very similar effect on promoting the PIP2-channel interaction to that of the D226N mutation. The simplest interpretation of these results is that Na+ directly interacts with Asp226 to reduce the negative electrostatic potential in the vicinity of the PIP2 binding site. The increase in the PIP2-channel interaction is likely to be at least part of the mechanism by which Na+ promotes GIRK channel activity. The highest concentration of PIP2 that we applied to inside-out patches did not preclude further activation of the channel by Na+, suggesting either that the membrane concentration of PIP2 around the channel did not achieve sufficiently high levels to saturate the low affinity binding sites on the channel, or that Na+ promotes channel activity by an additional mechanism.

The effect of Na+ on promoting PIP2-channel interactions was less than that reported for Gβγ binding. We observed a 2- to 3-fold slowing in the rate of current inhibition by the PIP2 antibody whereas Huang et al. (1998) reported a 6- to 15-fold slowing in the presence of Gβγ. Two amino acids within the GIRK C-terminus have been identified as important for Gβγ-mediated activation; a leucine in the middle of the GIRK1 and GIRK4 C-termini (Leu333 and Leu339, respectively) (He et al. 1999) and a cysteine in the proximal C-terminus of GIRK4 (Cys216) (Krapivinsky et al. 1998). Mutations which affect PIP2-Kir channel interactions also extend over the proximal and middle C-terminal regions. Arginines lying adjacent to the second transmembrane segment (TM2) of Kir6.2 (R176, R177) and ROMK1 (R188) are involved in electrostatic interactions with PIP2 (Fan & Makielski, 1997; Huang et al. 1998; Shyng & Nichols, 1998; Liou et al. 1999). Serines at positions 219 and 313 in ROMK1 are involved in a protein kinase A-mediated increase in PIP2 affinity (Liou et al. 1999). Two arginines at positions 218 and 228 in IRK1 were recently identified as critical for PIP2-channel interactions (Zhang et al. 1999). These are conserved in GIRK2 and GIRK4 and are adjacent to the Na+-sensing aspartate. Between these two arginines, mutating isoleucine at position 229 in GIRK4 altered PIP2-GIRK4 interactions (Zhang et al. 1999). Neutralizing the aspartate at position 223 also slowed channel rundown suggesting an increase in the PIP2-channel interaction. Thus several charged and uncharged amino acids within the C-terminus appear to be important in determining the functional interaction between inward rectifier channels and PIP2.

All three PIP2 isoforms that have been identified in vivo appear to be similarly effective at activating GIRK1-GIRK2 channels. Given the greater abundance of PI(4,5)P2, it seems likely that this isoform plays the most important role in regulating GIRK channel activity. PIP2 has multiple functional roles within the cells, such as the regulation of the actin cytoskeleton, the regulation of intracellular vesicle trafficking, and the involvement of Ca2+-sensitive exocytosis (see Hinchiffe et al. 1998 for review), as well as being involved in the regulation of membrane ion channels and transporters (Chu & Stefani, 1991; Hilgemann & Ball, 1996; Fan & Makielski, 1997; Baukrowitz et al. 1998; Huang et al. 1998; Lupu et al. 1998; Shyng & Nichols, 1998; Zhainazarov & Barry, 1999). It also serves as the precursor of important signalling molecules such as inositol 1,4,5-trisphosphate (IP3), diacylglycerol (DAG) and PI(3,4,5)P3 (Berridge, 1993; Zhang & Majerus, 1998). The role(s) of the less abundant isoforms remains unclear. Dynamic changes in the level of phosphatidylinositol phosphates by cellular signalling might contribute a ‘control point’ in the regulation of ion channels. Changing the affective interaction of PIP2 with Kir channels appears to be an important mechanism by which intracellular factors regulate their activity.


We thank Drs L. Y. Jan and J. P. Adelman for GIRK1 and GIRK2 cDNAs, respectively. We also thank Professor R. F. Irvine for purified PIP2. This work was supported by the Wellcome Trust, the Biotechnology and Biological Sciences Research Council and the Cambridge Commonwealth Trust (I.H.M.H.).