Divergent regulation of GIRK1 and GIRK2 subunits of the neuronal G protein gated K⁺ channel by Gα_i^GDP and Gβγ

Abstract

G protein activated K⁺ channels (GIRK, Kir3) are switched on by direct binding of Gβγ following activation of G_i/o proteins via G protein-coupled receptors (GPCRs). Although Gα_i subunits do not activate GIRKs, they interact with the channels and regulate the gating pattern of the neuronal heterotetrameric GIRK1/2 channel (composed of GIRK1 and GIRK2 subunits) expressed in Xenopus oocytes. Coexpressed Gα_i3 decreases the basal activity (I_basal) and increases the extent of activation by purified or coexpressed Gβγ. Here we show that this regulation is exerted by the ‘inactive’ GDP-bound Gα_i3^GDP and involves the formation of Gα_i3βγ heterotrimers, by a mechanism distinct from mere sequestration of Gβγ‘away’ from the channel. The regulation of basal and Gβγ-evoked current was produced by the ‘constitutively inactive’ mutant of Gα_i3, Gα_i3G203A, which strongly binds Gβγ, but not by the ‘constitutively active’ mutant, Gα_i3Q204L, or by Gβγ-scavenging proteins. Furthermore, regulation by Gα_i3G203A was unique to the GIRK1 subunit; it was not observed in homomeric GIRK2 channels. In vitro protein interaction experiments showed that purified Gβγ enhanced the binding of Gα_i3^GDP to the cytosolic domain of GIRK1, but not GIRK2. Homomeric GIRK2 channels behaved as a ‘classical’ Gβγ effector, showing low I_basal and strong Gβγ-dependent activation. Expression of Gα_i3G203A did not affect either I_basal or Gβγ-induced activation. In contrast, homomeric GIRK1* (a pore mutant able to form functional homomeric channels) exhibited large I_basal and was poorly activated by Gβγ. Expression of Gα_i3^GDP reduced I_basal and restored the ability of Gβγ to activate GIRK1*, like in GIRK1/2. Transferring the unique distal segment of the C terminus of GIRK1 to GIRK2 rendered the latter functionally similar to GIRK1*. These results demonstrate that GIRK1 containing channels are regulated by both Gα_i3^GDP and Gβγ, while GIRK2 is a Gβγ-effector insensitive to Gα_i3^GDP.

G protein activated K⁺ channels (GIRK, Kir3) mediate postsynaptic inhibitory effects of many neurotransmitters in the heart and brain. GIRK channels are switched on by direct binding of Gβγ following activation of G_i/o proteins via numerous G protein coupled receptors (GPCRs) (Logothetis et al. 1987; Wickman & Clapham, 1995; Dascal, 2001). Although GIRK is not considered an effector for Gα subunits, fast and specific channel activation was attributed to formation of GPCR–Gα_iβγ–GIRK signalling complexes (Huang et al. 1995; Slesinger et al. 1995; Leaney et al. 2000; Peleg et al. 2002; Ivanina et al. 2004). GIRK subunits interact with Gβγ subunits in vitro (Huang et al. 1995; Kunkel & Peralta, 1995; Huang et al. 1997; Ivanina et al. 2003; Finley et al. 2004) and in living cells before and after agonist activation (Rebois et al. 2006; Riven et al. 2006). Cytosolic GIRK segments also bind Gα_i^GDP and Gα_i^GTPγSin vitro, and GIRK channels immunoprecipitate with Gα subunits in native brain membranes (Huang et al. 1995; Ivanina et al. 2004; Clancy et al. 2005), but interaction of Gα with GIRK in vivo is under debate (Fowler et al. 2006; Rebois et al. 2006).

The main neuronal GIRK channel is a GIRK1/2 heterotetramer, composed of GIRK1 and GIRK2 subunits. These subunits can also co-assemble with GIRK3 to form GIRK1/3 and GIRK2/3. GIRK1/2 channels are abundant in the hippocampus and the cerebellum. GIRK2 can also form functional homomeric channels, which form the majority of GIRKs in the substantia nigra (Slesinger et al. 1996; Inanobe et al. 1999; Saenz del Burgo et al. 2008). In hippocampal and locus coeruleus neurons, GIRKs possess a substantial basal activity (I_basal) (Luscher et al. 1997; Torrecilla et al. 2002; Chen & Johnston, 2005; Wiser et al. 2006). The mechanisms of regulation of I_basal and its relation to the neurotransmitter-evoked activity (I_evoked) are poorly understood. We have previously presented evidence that in expression systems, Gα_i regulates GIRK gating, keeping I_basal low and preparing the channel for activation by ‘free’ Gβγ (Peleg et al. 2002; Rishal et al. 2005; Rubinstein et al. 2007). We proposed that this action was carried out by Gα^GDP (classically considered as the inactive form of Gα), possibly as a heterotrimer with Gβγ. However, the involvement of ‘active’ Gα^GTP could not be ruled out.

Divergent subunit compositions of GIRK channels have not been systematically studied. However, subunit content appears to play a role in coupling of GIRKs to GABA_B receptors (Cruz et al. 2004; Fowler et al. 2006) and GIRK1/3 was shown to have higher affinity to Gβγ compared to GIRK2/3 (Jelacic et al. 2000). When expressed as homomers, GIRK1 (GIRK1_F137S, a functional pore mutant, denoted GIRK1*) displayed low conductance compared to GIRK1/2 or GIRK1/4, and long bursts of activity (Chan et al. 1996; Vivaudou et al. 1997), whereas GIRK2 showed higher conductance than GIRK1* but low open probability with brief, flickery openings (Yi et al. 2001). In vitro studies suggested a smaller Gβγ and Gα binding surface in N and C termini of GIRK2 compared with GIRK1 (Ivanina et al. 2003; Ivanina et al. 2004). Yet, no definite differences in Gβγ or Gα_i regulation of GIRK1 and GIRK2 have been reported so far.

We explored the mechanisms of regulation of GIRK channels (heteromeric GIRK1/2 channel as well as homomeric GIRK1 and GIRK2 channels) by Gα_i and Gβγ using in vitro protein interaction assay and functional assays in Xenopus oocytes. To definitely distinguish between the effects of ‘inactive’ Gα^GDP and ‘active’ Gα^GTP, we utilized ‘constitutively inactive’ and ‘constitutively active’ Gα_i3 mutants. We demonstrate a profound regulation of the heteromeric GIRK1/2 and the homomeric GIRK1* channels by Gα_i3^GDP, probably in the context of the Gα_iβγ heterotrimer. This regulation does not occur in homomeric GIRK2 channels, leading to different regulation of GIRK1 and GIRK2 by Gβγ, and to an asymmetric regulation by Gα_iβγ heterotrimer.

Methods

DNA constructs and mRNA

The cDNA constructs were described in previous publications (Peleg et al. 2002; Rishal et al. 2005). cDNA constructs were inserted into pGEX (for production of glutathione-S-transferase (GST) fusion proteins in E. coli) or into high-expression oocyte vectors containing 5′ and 3′ untranslated sequences of Xenopusβ-globin: pGEMHE, pGEMHJ, or pBS-MXT. Rat GIRK1 (U01071), mouse GIRK2a (U11859), and human Gα_i3 (J03198) were used for oocytes expression and for deriving mutants, except human GIRK1* (U39195).

Gα_i3G203A and Gα_i3Q204L were produced using standard PCR-based methods and fully sequenced. The HA-(hemagglutinin)-tagged GIRK2 (Chen et al. 2002; Clancy et al. 2005) was subcloned into pBS-MXT. Yellow fluorescent protein (YFP) was fused in-frame before the N-terminus of GIRK1* or GIRK2, or after the C-terminus of GIRK2, essentially as described by Riven et al. (2003) and Fowler et al. (2006). G1NC and G2NC constructs were made by PCR; the transmembrane segments (GIRK1 a.a 85–184; GIRK2 a.a 96–193) were deleted, and N-terminus and C-terminus (NT and CT, respectively) were connected by a linker encoding amino acids QSTASQST in G1NC and KL in G2NC. The G2_CTG1 chimera (GIRK2_{HA(a.a 1-381)}GIRK1_(a.a371-501)) was constructed using PCR followed by blunt end ligation. All PCR products were fully sequenced. RNAs were synthesized in vitro as described (Kanevsky & Dascal, 2006) and injected into oocytes at 0.2–20 ng per oocyte. To prevent the formation of GIRK1*/5 heterotetramers, we injected antisense targeted against the oocytes’ endogenous GIRK5 subunits (Hedin et al. 1996) when studying GIRK1*.

Xenopus oocytes preparation and electrophysiology

Experiments were approved by the Tel Aviv University Institutional Animal Care and Use Committee (permit no. 11-05-064). Briefly, female frogs, maintained at 20 ± 2°C on an 11 h light–13 h dark cycle, were anaesthetized in a 0.15% solution of procaine methanesulphonate (MS222), and portions of ovary were removed through a small incision on the abdomen. The incision was sutured, and the animal was returned to a separate tank until it had fully recovered from the anaesthesia, and afterwards was returned to a large tank, together with the other postoperational animals. The animals did not show any signs of postoperational distress and were allowed to recover for at least 8 weeks until the next surgery. Following the final collection of oocytes, the frogs were killed by decapitation and double pithing while under anaesthesia. Oocytes were defolliculated by collagenase, injected with RNA (Peleg et al. 2002) and incubated for 3 days (whole cell studies) or 2–4 days (patch clamp) at 20–22°C in ND96 solution (low K⁺) containing, in mm: 96 NaCl, 2 KCl, 1 MgCl₂, 1 CaCl₂, 5 Hepes, pH 7.5, supplemented with 2.5 mm sodium pyruvate, 100 μg ml⁻¹ streptomycin and 62.75 μg ml⁻¹ penicillin (or 50 μg ml⁻¹ gentamycine). Whole-cell GIRK currents were measured using standard two-electrode voltage clamp procedures (Rubinstein et al. 2007) at 20–22°C, in high K⁺ solutions (24 mm K⁺ for GIRK1*, and 24, 48 or 96 mm for GIRK2, as indicated). K⁺ solutions at 24 and 48 mm were obtained by mixing ND96 with a 96 mm K⁺ solution containing, in mm: 96 KCl, 2 NaCl, 1 CaCl₂, 1 MgCl₂, 5 Hepes, pH 7.5. Acetylcholine (ACh) was used at 10 μm. Whole cell currents produced by the expression of untagged or YFP-tagged homomeric GIRK1* were similar in amplitude and in regulation by Gβγ, and the data were pooled. The same was observed with untagged, HA- or YFP-tagged GIRK2 channels.

Patch clamp experiments were performed as described (Peleg et al. 2002). Data acquisition and analysis were done using pCLAMP software (Molecular Devices, Sunnyvale, CA, USA) as described (Peleg et al. 2002). Currents were recorded at a holding potential of −80 mV, sampled at 10 kHz and filtered at 2 kHz. Patch pipettes had resistances of 1.2–2.5 MΩ. After seal formation, the patches were excised and exposed to air to prevent the formation of closed membrane vesicles at the tip. Stock solution (10 μm) of the purified Gβ₁γ₂ protein was diluted into 50 μl of the bath solution (final concentration of 20–40 nm as indicated), added to the 500 μl solution in the bath, and stirred. For patch recordings of heteromeric GIRK1/2 channels, the pipette solution contained, in mm: 144 KCl, 2 NaCl, 1 MgCl₂, 1 CaCl₂, 1 GdCl₃, 10 Hepes/KOH, pH 7.5. Bath solution contained, in mm: 130 KCl, 2 MgCl₂, 1 EGTA, 2 Mg-ATP, and 10 Hepes/KOH, pH 7.5. Measurements of homomeric channels were made with a pipette solution that contained, in mm: 146 KCl, 2 NaCl, 1 MgCl₂, 1 CaCl₂, 1 GdCl₃, 10 Hepes/KOH, pH 7.5. Bath solution contained, in mm: 146 KCl, 6 NaCl, 2 MgCl₂, 1 EGTA, 2 Mg-ATP, 10 Hepes/KOH, pH 7.5. (The presence of NaCl was found essential to prevent strong rundown of GIRK1* channels in excised patches.) GdCl₃ completely inhibited the stretch-activated channels. For the comparison of channel activity in excised patches vs. cell attached patches, currents were also corrected for changes in the single channel amplitude, i. For GIRK1/2, i was taken as 2.4 pA in cell-attached and 2.2 nA in excised patches (authors’ unpublished observations). For GIRK1*, i was 1.24 ± 0.03 pA in cell-attached mode and 1.06 ± 0.03 pA in excised mode. For GIRK2_HA, i was 1.4 ± 0.07 pA and 1.03 ± 0.08 pA in cell attached and excised mode, respectively. For GIRK2, i was 1.74 ± 0.05 pA and 1.49 ± 0.05 pA in cell attached and excised mode, respectively.

HEK293 transfection and electrophysiology

HEK293 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 2 mm glutamine, 10% fetal calf serum, 100 units ml⁻¹ penicillin-G sodium and 100 μg ml⁻¹ streptomycin sulfate in an atmosphere of 95% air–5% CO₂ at 37°C. Cells were transfected using the Fugene^R6 Transfection Reagent (Roche Applied Science, USA), in a 24-well dish, with cDNAs of GIRK1 and GIRK2 (0.2–0.6 μg), m2R (0.5 μg), and (when designed) of Gα_i3 (0.15–0.3 μg), m-phosducin (0.2 μg), Gβ₁ and Gγ₂ (0.2 μg each). Total DNA concentration was adjusted to 2.1 μg per 1.5 cm well by adding empty pcDNA3 vector DNA. The CD8 reporter gene system was used to visualize transfected cells (0.5 μg DNA per well). Beads coated with anti-CD8-antibodies were purchased from Invitrogen.

Whole cell recordings were performed at 21–23°C. Patch pipette solution contained, in mm: 130 KCl, 1 MgCl₂, 5 EGTA, 3 MgATP, 10 Hepes. Low-K bath solution contained, in mm: 140 NaCl, 4 KCl, 1.8 CaCl₂, 1.2 MgCl₂, 11 glucose, 2 CdCl₂, 5.5 Hepes. High-K bath solution contained 90 mm KCl and 54 mm NaCl; the rest was as in low-K solution. The osmolarity of intracellular and extracellular solutions was adjusted to 290 and 310 mosmol l^–1 respectively, with sucrose; pH was 7.4–7.6. Patch clamp was done using an Axopatch 200B (Molecular Devices). Data acquisition and analysis were done using pCLAMP software (Molecular Devices).

Biochemistry

GST-fused Gα_i3 (GST-Gα_i3) was purified as described (Rishal et al. 2003). [³⁵S]Methionine-labelled G1NC (G1N_1-84C_183-501) and G2NC (G2N_1-95C_194-414) proteins were synthesized in rabbit reticulocyte lysate (Promega Corp., Madison, WI, USA). For pull down experiments, 3 μg of GST-Gα_i3 was incubated at 30°C with 30 μm of GDP for 30 min in 50 μl of a high-K⁺ binding buffer (Rishal et al. 2003) containing 0.01% Lubrol. Purified recombinant Gβ₁γ₂ (3 μg, wild-type or His₆-tagged) or vehicle was added for another 30 min. Then, 5 μl of reticulocyte lysate containing [³⁵S]methionine-labelled G1NC or G2NC was added, the total reaction volume was brought to 300 μl, 5 μl was removed and used to measure the loaded protein (‘input’), and the incubation continued for 1 h at room temperature. Binding to glutathione-Sepharose beads and elution with 15 mm glutathione were done as described (Rishal et al. 2003). The eluted proteins were separated on 12% polyacrylamide-SDS gels. The radioactive signals from protein bands of the gels were imaged using PhosphorImager and the software ImageQuaNT (Molecular Dynamics). Western blots were performed using standard procedures, using Gβ antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, c.a., USA) and ECL reagents (Pierce Biotechnology, Inc., Rockford, IL, USA).

Imaging and immunocytochemistry

Imaging of proteins in the plasma membrane (PM) was performed either in giant PM patches or in whole oocytes. Imaging of homomeric GIRK1* in PM was performed using N or C terminally YFP-labelled channels in whole oocytes (Fig. 5A and Supplemental Fig. 2A), or immunolabelling of untagged channels in giant excised PM patches using GIRK1 antibody (Fig. 6A). GIRK2 expression was monitored in intact oocytes with either C terminally labelled YFP channel (Supplemental Fig. 2A), or GIRK2_HA channels with an extracellular HA tag (Figs 5A and 8A and C) or giant PM patches using GIRK2 antibody (Fig. 8A). Immunocytochemistry in giant PM patches was done essentially as described (Singer-Lahat et al. 2000; Peleg et al. 2002). In brief, the vitelline membrane was peeled off and the oocytes were placed on plastic coverslips (Thermanox plastic coverslip, Nunc, Naperville, IL, USA). After sticking to the coverslip, the oocyte was removed by washing with a strong jet of solution. Pieces of membrane attached to the coverslip, with their cytosolic leaflet surface exposed to the external solution, were washed until the membrane patch became transparent. Following fixation for 10 min in 1% formaldehyde, the membranes were washed three times with 5% skim milk dissolved in phosphate-buffered saline (PBS). Blocking of non-specific binding sites was done with donkey immunoglobulin G (IgG, whole molecule, 1/400, Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) for 30 min. Each coverslip was incubated for 1 h with antibodies against GIRK1 or GIRK2 (Alomone Labs, Jerusalem, Israel). Residual antibody was washed out with 5% skim milk 3 times, 5 min each. This was followed by a 30 min incubation with secondary antibody (Cy3 donkey anti-rabbit IgG, 1 : 400, Jackson ImmunoResearch Laboratories). Free secondary antibody was then washed out with PBS 3 times, 5 min each in darkness and the coverslips were mounted on a glass slide. The fluorescent labelling was examined by a confocal laser scanning microscope (LSM 510 Meta, Zeiss, Germany) with 20× or 5× objective lens. Cy3 was excited at 514 nm and the emitted light was collected between 540 and 615 nm using the spectral mode of the Zeiss 510 Meta (beam splitter HFT 405/514/633).

Figure 5. The basal activity is Gβγ dependent in GIRK1*, but Gβγ independent in GIRK2.

A, confocal images of whole oocytes expressing YFP-labelled GIRK1* (visualized using YFP) or GIRK2_HA (visualized using anti-HA antibody). B, the effect coexpression of Gβγ scavengers, m-phosducin (10 ng) and m-cβARK (5 ng), on I_basal of GIRK1* (grey) and GIRK2 (dark grey). Currents were corrected to changes in PM level of the channel (shown in Supplemental Fig. 3). n= 16–20. ***P < 0.001.

Figure 6. Gα_i3 regulates GIRK1* activation in whole oocytes.

A, examples of basal and ACh-evoked currents (continuous lines) of GIRK1* (10–15 ng), with and without coexpression of Gα (2.5 ng) and Gβγ (5 and 1 ng RNA, respectively). The images beside the current traces show the PM expression of GIRK1*, monitored in giant excised PM patches using anti-GIRK1 antibody and a secondary, fluorescently labelled antibody. The images focus on the edges of the PM patches to allow comparison with the background. The image area is 55 × 55 μm. B, summary of GIRK1* expression levels under different conditions. n= 14–19. C, summary of I_basal and I_ACh with or without coexpression of Gα_i3-wt (2.5 ng RNA per oocyte) (a), and of R_a measured in the same oocytes (b). n= 22–30. D, the effect of coexpression of Gα_i3 (wt, GA or QL, 2.5 ng each) on GIRK1*'s I_basal and I_βγ (a), and the summary of measurements of R_βγ in this series of experiments (b). n= 17–36. Statistical comparisons of multiple groups were done using one-way ANOVA followed by Dunnett's test against the control group, GIRK1* alone.*P < 0.05; **P < 0.01; ***P < 0.001.

Figure 8. GIRK2 is not regulated by Gα_i3 in whole oocytes.

Aa, changes in the amount of GIRK2 channels in PM caused by coexpression of Gα_i3 (2.5 ng RNA) and Gβγ (5 and 1 ng RNA, respectively). Data were obtained by measurements in giant PM patches using anti-GIRK2 antibody and a secondary, fluorescently labelled antibody, or in whole oocytes using external HA tag, in the same experiment. b, summary of GIRK2 expression and comparison of the two imaging methods. Open bars show the amount of GIRK2 and GIRK2_HA channels, assessed in giant PM patches using the anti-GIRK2 antibody. Grey bars show the PM expression of GIRK2_HA measured using the external HA tag. With both methods, the expression level in different groups was normalized to the control group of oocytes expressing the channel alone. Note that both methods provide very similar assessment of the relative effects of expression of Gα_i3GA (reduction in PM expression levels) and of Gβγ (no change or recovery of expression). n= 4–11. B, coexpression of Gα_i3 (wt, GA or QL, 2.5 ng each) did not significantly affect R_βγ in a series of experiments where the measured currents were corrected to the PM expression of GIRK2 or GIRK2_HA measured in the same experiments. The full details of PM expression, I_basal and I_βγ are presented in Supplemental Fig. 4A. n= 13–28. C, summary of a separate experiment in which GIRK2_HA PM expression was titrated, by injected different amounts of RNA as indicated on the images, to produce equal channel expression in the presence of coexpressed Gα_i3GA (2.5 ng) with or without Gβγ (5 and 1 ng RNA, respectively). a, the confocal images of whole oocytes obtained with an anti-HA antibody. Examples of currents in representative oocytes are shown to the left of the confocal images. See Supplemental Fig. 4B for further details. b and c, the effect of coexpression of Gα_i3GA on GIRK2's I_basal and I_βγ (b) and the summary of measurements of R_βγ (c) in groups of oocytes with equal PM channel expression. n= 8 oocytes in each group. D, examples of GIRK2 currents (a) and summaries of I_basal and I_ACh (b) and R_a (c) with or without coexpression of Gα_i3-wt (2.5 ng) in four experiments where the amounts of GIRK2 RNA were not titrated, channel expression in the PM was monitored, and current measurements were corrected for PM level changes. n= 13–28. ***P < 0.001.

Imaging of whole oocytes was done using external HA tag (GIRK2_HA, G2_CTG1) (Kanevsky & Dascal, 2006) or YFP (YFP labelled GIRK1* or GIRK2). To visualize the HA tag, whole oocytes expressing GIRK2_HA or G2_CTG1 were fixated in 4% formaldehyde (from 37% stock) in Ca²⁺-free ND96 solution for 30 min. Blocking of non-specific binding sites was done by 5% skim milk for 1 h in Ca²⁺-free ND96. Then the oocytes were incubated for 1 h with the mouse monoclonal anti-HA antibody (Santa Cruz Biotechnology), diluted 1 : 400 in 2.5% skim milk. Residual antibody was washed out with 2.5% skim milk 3 times, 5 min each. This was followed by 1 h incubation with the secondary antibody (Alexa Fluor 488 conjugated, 1 : 400, Molecular Probes/Invitrogen) in dark. Free secondary antibody was then washed out with Ca²⁺-free ND96. Oocytes were placed in a chamber with a transparent bottom, and fluorescence imaging was performed with LSM 510 (×20 objective, zoom = 2, pinhole 3 Airy units). Alexa was excited at 488 nm, and the emitted light was collected in the wavelength interval of 508–615 nm in spectral mode (with HFT 405/488 beam splitter).

Imaging of YFP labelled GIRK1* and GIRK2 channels was performed with LSM 510 (×20 objective, zoom = 2, pinhole 3 Airy units). YFP was excited at 514 nm, and the emitted light was collected in the wavelength interval of 524–609 nm in spectral mode (with HFT 405/514/633 beam splitter).

All images were obtained from optical slices from the animal hemisphere close to the oocyte's equator. Quantification of all the images was done using Zeiss LSM software. The intensity of fluorescence in the PM was measured by averaging the signal obtained from three standard regions of interest. Net fluorescence intensity per unit area was obtained by subtracting the background signal measured in uninjected oocytes. In all confocal imaging procedures, care was taken to completely avoid saturation of the signal. In each experiment, all oocytes from the different groups were studied using constant LSM settings.

Data analysis, presentation and statistics

Relative activation by agonist, R_a, was calculated in each cell as I_total/I_basal, where I_total=I_basal+I_ACh. Thus, R_a=I_total/I_basal= (I_ACh+I_basal)/I_basal= (I_ACh/I_basal) + 1. Similarly, the extent of Gβγ activation, R_βγ, was defined as R_βγ=I_βγ/I_basal, where I_βγ is the total Gβγ-dependent current. The better the activation by Gβγ, the greater is R_βγ; R_βγ= 1 when there is no effect of Gβγ. In patch clamp experiments, I_basal was measured in a cell-attached mode before excision, and I_βγ in the same patch after excision and addition of purified Gβγ (Figs 3, 7 and 9). In whole cells, I_βγ is the total GIRK current in a cell expressing Gβγ, and I_basal is the average GIRK current in oocytes of the same batch expressing GIRK alone (Rubinstein et al. 2007). These definitions are essentially the same as the ones used previously (R_a=I_ACh/I_basal; R_βγ=I_βγ/I_basal− 1) (Peleg et al. 2002; Rubinstein et al. 2007) except for the addition of a constant number, 1. The change in definition was motivated by the desire to avoid division by very small numbers during normalization to (division by) R_a or R_βγ of the ‘GIRK only’ groups. Indeed, R_a and R_βγ calculated according to the previous definition (Rubinstein et al. 2007) were often close to zero in these cells at high expression levels of GIRK1/2 or GIRK1*, when the activation by ACh or by Gβγ was weak.

Figure 3. Gα_i3GA improves GIRK1/2 activation by added Gβγ protein in excised plasma membrane patches.

A, examples of patch clamp records in oocytes expressing either GIRK1/2 alone (a) or with Gα_i3GA (b). RNAs at 1–2.5 ng per oocyte of GIRK1/2, Gα_i3GA and Gα_i3QL were injected. Approximate amplitudes of I_basal and I_βγ are indicated for illustration in a; the exact values were calculated from all-points histograms (Yakubovich et al. 2009). B, summary of patch clamp experiments. In view of large batch-to-batch variability, all data (currents, R_βγ) were normalized to those of control group (GIRK1/2 alone) recorded on the same day. Number of measurements is shown above the bars. *P < 0.05; **P < 0.01.

Figure 7. Gα_i3GA regulates GIRK1* in excised plasma membrane patches.

GIRK1* (10–17 ng) was expressed with or without Gα_i3GA (2.5 ng). A, examples of patch clamp records in oocytes expressing GIRK1* alone (a) or GIRK1* with Gα_i3GA (b), with zooms on segments of the c.a. records below the main traces. B, correlation between I_basal (measured in the cell attached mode) and R_βγ with (right) or without (left) Gα_i3GA coexpression. The data were analysed using Spearman's correlation algorithm. Correlation coefficient (r), P value and n are shown in the boxes. C, Summary of patch clamp experiments. In view of large batch-to-batch variability, all data (currents, R_βγ) were normalized to those of control group (GIRK1* alone) recorded on the same day. The number of measurements is shown on top of the I_basal bars (left graph). *P < 0.05, **P < 0.01.

Figure 9. Gα_i3GA does not regulate GIRK2 in excised plasma membrane patches.

A, examples of patch clamp records in oocytes expressing GIRK2_HA alone (10 ng) (a) or GIRK2_HA+ Gα_i3GA (Gα_i3GA was expressed at 2 ng) (b). B, correlation between I_basal (measured in the cell attached mode) and R_βγ with (right) or without (left) Gα_i3GA coexpression. The data were analysed using Spearman's correlation algorithm. Correlation coefficient (r), P value and n are shown in the boxes. C, summary of patch clamp experiments. In view of large batch-to-batch variability, all data (currents, R_βγ) were normalized to those of control group (GIRK2_HA alone) recorded on the same day. The number of measurements is shown above the bars.

Results are shown as means ±s.e.m. When summarizing several experiments, in order to minimize batch-to-batch variations and to enhance the accuracy of the statistical analysis, whole-cell GIRK currents were normalized, in each oocyte, to the average GIRK current of the control group (GIRK alone) of the same experiment (Sharon et al. 1997). Two group comparisons were done using Student's two-tailed t-test. Multiple group comparison was done using one-way analysis of variance (ANOVA) followed by Tukey's, Student–Neumann–Keuls or Dunnet's test. Correlations between two parameters were examined using Spearman's test.

Results

Gα_i3 regulates GIRK1/2 gating in HEK 293 cells, like in oocytes

We have previously observed that the I_basal of GIRK1/2 increased disproportionally as the channel expression level was increased, whereas activation by agonist or Gβγ became weaker; there is a negative correlation between I_basal and the extent of activation by agonist (R_a) or by Gβγ (R_βγ). This abnormal gating was corrected by coexpressed Gα_i3 which reduced I_basal, increased R_a and R_βγ, but did not reduce the total Gβγ-dependent current (basal + evoked), I_total, as if the presence of Gα_i3 prepares (‘primes’) the channel for activation by Gβγ (Peleg et al. 2002; Rubinstein et al. 2007). The enhancement of agonist-evoked current by coexpressed Gα_i appeared self-evident (Vivaudou et al. 1997; He et al. 1999) (free Gβγ is sequestered by Gα_i^GDP but then released after activation of GPCR). However, the enhancement of activation induced by added or coexpressed Gβγ could not be fully explained by a mechanism in which Gα reduces I_basal by ‘sequestering Gβγ away’ from GIRK (Rusinova et al. 2007), suggesting a more complex mechanism, where the GIRK channel may be regulated by Gα_i3^GDP (Peleg et al. 2002).

The aforementioned hallmarks of Gα_i regulation have been observed so far only in Xenopus oocytes, which are believed to have relatively high levels of Gβγ that help to maintain the meiotic arrest (Sheng et al. 2001; Evaul et al. 2007). Therefore, it was important to show that regulation by Gα_i does not arise from some unusual properties of Xenopus oocytes. To this end, we transiently transfected HEK 293 cells with GIRK1, GIRK2 and the muscarinic 2 receptor (m2R). Evoked currents were elicited by acetylcholine (ACh) or by coexpression of Gβγ. In general, R_a and R_βγ were somewhat higher in HEK 293 cells than in oocytes, corroborating higher basal Gβγ levels in the oocytes. Despite this quantitative difference, the fundamental hallmarks of Gα_i3 regulation were as in Xenopus oocytes (Fig. 1). There was a strong negative correlation between I_basal (taken as the indicator of expression level) and R_a or R_βγ; expression of Gα_i3 restored the low I_basal and high activation by agonist and Gβγ, whereas the membrane-attached Gβγ scavenger protein m-phosducin (Rishal et al. 2005; Rubinstein et al. 2007) diminished both I_basal and the evoked currents. For further functional assays we used Xenopus oocytes, where protein expression levels can be accurately controlled and monitored.

Figure 1. Regulation of GIRK1/2 channels expressed in HEK293 cells.

A, HEK 293 cells were transiently transfected with GIRK1, GIRK2 and the muscarinic 2 receptor (m2R). GIRK1/2 currents were studied using whole-cell voltage clamp at −80 mV. Switching from a physiological low-K⁺ solution to a high-K⁺ solution (90 mm K⁺) revealed GIRK1/2 I_basal. Addition of the agonist acetylcholine (ACh; 10 μm) evoked I_ACh. Ba²⁺ at 5 mm was then added to inhibit GIRK currents, to calculate the net I_basal and I_ACh. B, R_a is negatively correlated with I_basal (to account for differences in HEK cell size, we used current densities (in pA pF⁻¹) in all calculations rather than net currents). Correlation coefficient (r), statistical significance (P), and number of measurements (n) are indicated. R_a in HEK cells was usually greater than in Xenopus oocytes, up to 18 (in the oocytes it is usually below 5), corroborating higher basal Gβγ levels in the oocytes. C, coexpression of Gα_i3 and m-phosducin reduced I_basal similarly, but only Gα_i3 enhanced the relative activation by ACh, R_a. D, coexpressed Gα_i3, but not m-phosducin, improves activation of GIRK1/2 by expressed Gβγ. n= 3–7 cells in each group. The extent of direct activation by Gβγ (R_βγ), calculated here as the ratio of average currents in cells expressing Gβγ to those without Gβγ, was 5.4 for GIRK alone, 26 in the presence of Gα_i3, and 0.68 in the presence of m-phosducin. *P < 0.05; **P < 0.01 compared to control by ANOVA.

‘Constitutively inactive’ mutant of Gα_i3, Gα_i3GA, regulates the basal activity and Gβγ activation of heteromeric GIRK1/2

We envisaged that coexpressed Gα_i3 regulated GIRK1/2 in its GDP-bound form, but the involvement of Gα_i^GTP could not be excluded (Peleg et al. 2002; Rubinstein et al. 2007). In order to decisively distinguish between effects of Gα_i^GDP and Gα_i^GTP, we used two mutants of Gα. One is the ‘constitutively active’ Gα_i3Q204L (Gα_i3QL), which poorly hydrolyses GTP, has a reduced affinity for Gβγ and regulates effectors of Gα_i^GTP in a GPCR-independent manner (Masters et al. 1989). The second one is the ‘constitutively inactive’ Gα_i3G203A (Gα_i3GA), which strongly binds Gβγ (Ogier-Denis et al. 1996). GαGA mutants can bind GTP but do not regulate known effectors of Gα^GTP (Lee et al. 1992).

Upon injection of a standard dose of 1–2 ng RNA per oocyte, Gα_i3-wild-type (wt) and its two mutants were expressed in the PM at similar levels, more than 4-fold over the endogenous Gα_i/o, with no effect on GIRK1/2 expression level (data not shown). As described in other cells (Fishburn et al. 1999; Evanko et al. 2000), Gα and Gβγ mutually affected each other's PM levels. Expression of Gβγ further increased the PM levels of expressed Gα_i3, Gα_i3GA and Gα_i3QL by ∼ 35%, and expression of Gα_i3 or Gα_i3GA increased the PM level of coexpressed Gβγ by about 50%, whereas Gα_i3QL was without effect (data not shown).

Since both Gα mutants disrupt the G protein cycle, we activated GIRK by coexpressed Gβγ (in whole oocytes; see Reuveny et al. 1994; Rubinstein et al. 2007) or by purified Gβγ (in excised patches), thus bypassing the GPCR. Here, the channels are activated by the externally added Gβγ, without the Gα_iβγ heterotrimer dissociation and without formation of Gα_i^GTP. As controls for Gα_i3 mutants, we expressed two genetically engineered Gβγ scavengers that accumulate at the PM because they are N-terminally myristoylated, like Gα_i, and act as ‘sinks’ for Gβγ: m-phosducin and m-cβARK (a modified C-terminal part of β-adrenergic receptor kinase 1) (Rishal et al. 2005). We found that m-phosducin moderately reduced the total amount of detectable coexpressed Gβγ in the PM, strengthening the notion of its being a strong Gβγ sink, whereas cβARK had no effect (data not shown).

With the high GIRK1/2 levels used, I_basal was large (14 ± 2 μA, n= 31) and the expression of Gβγ produced little additional activation (examples in Fig. 2A, summarized in Fig. 2B); the average relative activation by Gβγ (R_βγ) in this series of experiments was 1.4 ± 0.1 (Fig. 2C). Gα_i3GA, m-phosducin and m-cβARK reduced I_basal, whereas Gα_i3QL did not alter I_basal (Fig. 2B). We titrated the RNA dosage of Gα_i3GA, phosducin and m-cβARK to attain a comparable 72–82% reduction in I_basal (at RNA doses of 2, 5 and 10 ng RNA per oocyte, respectively; data not shown) and used these doses in experiments of Fig. 2. Despite the similar decrease in I_basal, the three proteins had disparate effects on GIRK1/2's activation by Gβγ. Gα_i3GA did not reduce the total Gβγ-dependent current, I_βγ, and significantly enhanced R_βγ (by 6.15 ± 0.83-fold compared to control; n= 25). In contrast, c-βARK reduced I_βγ by ∼75%, almost completely preventing channel activation by the coexpressed Gβγ (Fig. 2B). Accordingly, in the presence of m-cβARK, R_βγ was the lowest among all conditions, only 1.21 ± 0.1 (compare with 8.6 ± 1.2 with Gα_i3GA; Fig. 2C). m-Phosducin slightly increased R_βγ, but significantly less than Gα_i3GA. Furthermore, m-phosducin also significantly reduced I_βγ, in contrast to Gα_i3GA (Fig. 2B and C). The disparate effects of Gα_i3GA and the two Gβγ scavengers confirm that the effect of Gα_i3GA is distinct from simple Gβγ sequestration away from GIRK.

Figure 2. Gα_i3GA improves GIRK1/2 activation by coexpressed Gβγ in whole oocytes.

A, examples of GIRK1/2 currents in four individual oocytes of one batch expressing GIRK1/2 with or without Gβγ (left panel), or GIRK1/2 with 2 ng Gα_i3GA, with or without Gβγ (right panel). Five nanograms of RNA of Gβ₁ and 1 ng RNA of Gγ₂ were injected. Zero current is indicated by dashed lines (note that an outward I_basal is present in the low, 2 mm K⁺ solution). B, the effects of coexpression of Gα_i3GA or Gα_i3QL (2 ng RNA), m-phosducin (5–10 ng) and m-cβARK (5 ng) on basal and Gβγ-induced whole-cell currents. (The data with 5 and 10 ng m-phosducin were pooled as they produced quite similar effects.) **P < 0.01 compared to control I_basal; #P < 0.05 or ##P < 0.01 compared with control I_βγ. C, R_βγ of the experiments summarized in B. *P < 0.05 and **P < 0.01 against control or between indicated groups. n= 8–31.

We next studied GIRK1/2 regulation by Gα mutants in excised patches, which has the advantage of strictly controlling the concentration of added Gβγ and measuring I_basal and I_βγ in the same oocyte. GIRK1/2 was expressed alone or with Gα_i3GA or Gα_i3QL. After 1–3 min of cell-attached (c.a.) recording of I_basal, patches were excised. The activity declined, reaching a new steady level within 0.5–2 min (Peleg et al. 2002). Approximately 3 min after excision, purified Gβ₁γ₂ (20 nm) was added to the bath solution to activate the channel (Fig. 3A). Coexpressed Gα_i3GA strongly reduced I_basal but significantly enhanced R_βγ compared to control group (Fig. 3B). Gα_i3QL caused a mild reduction in I_basal (P > 0.05) and no significant changes in R_βγ. I_βγ was unchanged by either Gα_i3GA or Gα_i3QL. This result (and also the similar result with GIRK1*, Fig. 7) rules out the possibility that the enhanced activation observed with Gα_i3GA coexpression was due to Gα-dependent changes in the level of Gβγ, as could potentially happen with coexpressed proteins in whole cell experiments. The effect of Gα_i3GA was in sharp contrast to that of m-cβARK, which greatly reduced Gβγ-induced activation of GIRK1/2 in excised patches (Peleg et al. 2002). Thus, in excised patches Gα_i3GA, but not Gα_i3QL, regulates the channel basal activity, and improves the relative activation by added Gβγ.

To summarize, Gα_i3GA has a unique effect on GIRK1/2 gating: similarly to Gα_i3-wt, it reduces I_basal, enhances R_βγ, and does not reduce the total I_βγ. Neither the two Gβγ scavengers tested nor Gα_i3QL can reproduce these effects. We conclude that the regulation of I_basal and of the Gβγ-evoked activation of GIRK1/2 is produced by the GDP-bound form of Gα_i3.

Gβγ enhances the interaction of Gα_i3^GDP with GIRK1 but not GIRK2

We have examined the interaction of in vitro translated cytosolic domain of GIRK1 (a tandem of full-length N- and C-termini, G1NC, which lacks the transmembrane domain; see Fig. 4A) with Gα_i3 in the presence of purified Gβγ. Interestingly, the binding of Gα_i3^GDP to G1NC was enhanced in the presence of Gβγ (Fig. 4C and unpublished observations). To examine whether this phenomenon also takes place in GIRK2, we constructed a similar tandem, G2NC, encoding the full-length cytosolic domain of GIRK2 (Fig. 4A). The in vitro translated, ³⁵S-labelled G1NC and G2NC gave a single protein band each on SDS-polyacrylamide gels (Fig. 4B and C). Furthermore, both G1NC and G2NC bound Gβγ similarly (Fig. 4B). These results suggest that in vitro synthesized G1NC and G2NC are stable, well folded proteins.

Figure 4. Biochemical and functional differences between GIRK1 and GIRK2.

A, schematic representation of the G1NC and G2NC tandems. The transmembrane portion of the channel was replaced by a short linker (shown by a loop connecting the cylinders). B, His-Gβγ binds the ³⁵S-labelled, in vitro synthesized G1NC and G2NC. Pull down experiments were performed using Ni²⁺ affinity beads. Autoradiograms of loaded (upper panel, ‘input’; 1/60 of total loaded protein) and bound ³⁵S-labelled G1NC and G2NC proteins (lower panel, ‘binding’) are shown. C, GST-Gα_i3 binds G1NC and G2NC. The binding of G1NC, but not G2NC, is enhanced by added purified Gβ₁γ₂. Pull down experiments were performed using glutathion sepharose affinity beads. These results are representative of 3 independent experiments. D, examples of basal and ACh-evoked currents in two representative oocytes expressing homomeric GIRK1* and GIRK2. Arrowheads indicate the exchange of the external solution from low, 2 mm K⁺ to high, 24 mm K⁺. The addition of 10 μm ACh is indicated by the horizontal bar. E and F, examples of currents from four different oocytes expressing homomeric GIRK1* (E, grey), GIRK2 (F, grey) or homomeric channels with coexpression of Gβγ (black traces in E and F). Holding potential was set to −80 mV. Solutions were switches as indicated. Summaries of the basal, ACh-evoked and Gβγ-evoked currents are depicted to the right of the traces in E and F. GIRK1* is depicted in light grey (E) and GIRK2 in dark grey (F). In each bar chart, the bottom bar shows I_basal, the top open bar shows I_ACh, and the total height of each bar represents I_total. G1* stands for GIRK1*, G2 for GIRK2. Diagonal strips indicate coexpression of Gβγ. n= 20–40. G, R_βγ, the extent of activation by coexpressed Gβγ. H, titrated expression of Gβγ revealed dose-dependent activation of GIRK2 (one experiment, n= 5–6) but no activation of GIRK1* (two experiments, n= 5–18). The dose of Gγ was 0.5 of that of Gβ. ***P < 0.001.

Next, binding of the in vitro translated G1NC and G2NC to a GST-fused Gα_i3 (GST-Gα_i3) (Rishal et al. 2003) was examined in the presence of GDP, and in the presence or absence of purified Gβ₁γ₂. In the absence of Gβγ, both G1NC and G2NC bound Gα_i3 (Fig. 4C). In the presence of Gβγ, the binding of G1NC to GST-Gα_i3 was enhanced ∼6-fold. In contrast, G2NC binding to GST-Gα_i3 was unaffected by the addition of Gβγ (Fig. 4C). These data suggest the formation of a strong complex between Gα_i3βγ heterotrimers and GIRK1, but not GIRK2.

Functional differences between homomeric GIRK1 and GIRK2

The asymmetric interaction of GIRK1 and GIRK2 with Gα_i3βγ suggested a different regulation of function of each subunit within the heteromeric channel. We therefore sought to investigate these differences using homomeric GIRK channels expressed in Xenopus oocytes. GIRK2 subunits form functional homomers, whereas GIRK1 does not. Thus, we used the GIRK1_F137S (GIRK1*), a pore mutant of GIRK1, able to form functional channels in the plasma membrane (PM) (Vivaudou et al. 1997). The presence of a single mutation in the pore is not likely to affect the regulation of the channel by Gα and Gβγ, which bind to the cytosolic segments. In support, heterotetrameric GIRK1*/GIRK2 functioned similarly to wild-type GIRK1/2, showing an excessive I_basal and reduced activation by agonist and Gβγ at high expression levels, as well as the typical regulation by Gα_i3GA (data not shown).

Homomeric GIRK channels were activated by ACh (via coexpressed m2R), by coexpression of Gβγ, or by the addition of purified Gβγ to excised patches. For the whole cell experiments K⁺ currents were measured using two electrode voltage clamp, concurrently with measurements of protein level of the channels in the PM. GIRK1* currents were always measured in 24 mm K⁺, whereas most of GIRK2 recordings were done in 48 mm K⁺, to allow for a better resolution of I_basal of the GIRK2.

Titrated expression of the two channels revealed striking differences which became especially prominent at high expression levels (10–15 ng RNA per oocyte) (see Table 1). The first difference was the magnitude of I_basal, which was many-fold larger in GIRK1* than GIRK2 (3 ± 0.2 μA, n= 44 vs. 0.38 ± 0.06, n= 40; P < 0.001, Fig. 4D–F). This occurred despite the fact that GIRK2 expression was higher compared to GIRK1* (Supplemental Fig. 2), and despite the fact that we used higher K⁺ concentration in GIRK2 measurements. Moreover, I_basal of GIRK1* was strongly RNA dose dependent, whereas I_basal of GIRK2 grew much less with channel density (Supplemental Fig. 1A and B).

Table 1.

Comparison between GIRK1/2, GIRK1* homomers and GIRK2 homomers

Parameter	GIRK1/2	GIRK1*	GIRK2
Ibasal	High, Gβγ dependent	High, Gβγ dependent	Low, largely Gβγ independent
Activation by Gβγ	Weak or medium at high channel densities (Rβγ 1.4–1.8)	None or negative at high channel densities (Rβγ≤ 1)	Strong (Rβγ > 10)
Gαi3GDP on Ibasal	Decreases Ibasal	Decreases Ibasal	Almost no effect
Gαi3GDP on Rβγ	Strong regulation (increases Rβγ with no decrease in Iβγ)	Strong regulation (increases Rβγ with no decrease in Iβγ)	Weak regulation (no increase in Rβγ and Iβγ)
Gαi3-wt on IACh	Increases IACh and Ra with no change in Itotal	Increases IACh and Ra with no change in Itotal	Increases IAch, Ra and Itotal

I_basal is the agonist-independent basal activity. I_ACh is the agonist-evoked response, revealed upon application of ACh via activation of coexpressed m2R. I_total=I_basal+I_ACh. R_βγ, the relative activation by Gβγ, calculated as the ratio of I_βγ measured in a given oocyte which coexpressed Gβγ, to the average I_basal of the control group lacking coexpressed Gβγ (R_βγ=I_βγ/I_basal). Note that I_βγ is the total GIRK current, comprising the basal and the Gβγ-evoked current. See (Peleg et al. 2002; Rubinstein et al. 2007) for additional data regarding GIRK1/2.

The second difference was the effect of coexpression of Gβγ. GIRK1* or GIRK2 currents recorded in control and Gβγ-expressing oocytes are shown in Fig. 4E and F, left panels; the right panels show summaries of the experiments of this series. Surprisingly, coexpression of Gβγ failed to increase whole-cell GIRK1* currents at high channel expression levels (Figs 4E and 6D), whereas GIRK2 was strongly activated by the coexpressed Gβγ (Figs 4F and 8C). Titrated Gβγ expression demonstrated a clear dose-dependent activation of GIRK2 by Gβγ, which contrasted with lack of activation of GIRK1* by Gβγ at all RNA doses tested (Fig. 4H). The differences in the effects of Gβγ were not caused by differential changes in channel expression level (Supplemental Fig. 2). Despite lack of activation of GIRK1*, Gβγ expression almost completely abolished the agonist-evoked response of GIRK1* (Fig. 4E). Interestingly, when GIRK1* was expressed at low levels (1–5 ng RNA per oocyte; I_basal < 0.6μA), the coexpressed Gβγ activated the channel ∼5-fold (Supplemental Fig. 1D), supporting the data reported by (Vivaudou et al. 1997). The mechanism of the complex, even paradoxical, behaviour of GIRK1* is still unclear but seems to be connected with its regulation by Gα (see below).

Inversely, as mentioned above, GIRK2 was strongly activated by the coexpressed Gβγ, as expected from a Gβγ effector (Fig. 4F and H). R_βγ was 46.6 ± 5.17 (Fig. 4G), dramatically higher compared to GIRK1* (0.75 ± 0.05) or even the neuronal GIRK1/2 channel (1.4 ± 0.1, Fig. 2C; Table 1). Notably, I_βγ of GIRK2 was ∼6-fold higher compared to the total current (I_basal+I_ACh) observed without Gβγ expression (Fig. 4F; 16.3 ± 3.1 μA vs. 2.6 ± 0.2 μA) (see Discussion).

Previously, using the Gβγ scavengers m-phosducin and m-cβARK, we showed that up to 90% of the basal current of the neuronal GIRK1/2 channel is Gβγ dependent (Rishal et al. 2005). These two Gβγ scavengers were used to characterize the basal currents of the homomeric channels (Fig. 5 and Supplemental Fig. 3). The PM expression of GIRK1* and especially GIRK2 was substantially reduced by coexpression of m-phosducin (Supplemental Fig. 3), and therefore the currents in Fig. 5 are presented after correction to the relative PM expression. Coexpression of m-phosducin or m-cβARK reduced I_basal of GIRK1* by 89–95%, suggesting that the basal activity of GIRK1* is mostly Gβγ dependent, like in GIRK1/2 (see Fig. 2). In contrast, GIRK2 basal activity appeared largely Gβγ independent as its I_basal was not inhibited by coexpressed m-cβARK or m-phosducin.

GIRK1* is regulated by Gα_i3

The distinct patterns of basal and Gβγ-evoked activities in GIRK1* and GIRK2 and the difference in binding of GIRK subunits to Gα_i in the presence of Gβγ led us to hypothesize that GIRK1* and GIRK2 may be differentially regulated by Gα_i. To test this, homomeric GIRK1* channels were coexpressed with Gα_i3-wt, Gα_i3GA or Gα_i3QL, with or without the coexpression of Gβγ, and the emerging changes in I_basal, I_ACh, I_βγ, R_a, R_βγ and the channel's surface expression were monitored. Analysis of the effects of Gα_i3 were performed at high GIRK1* expression levels (I_basal > 1.4 μA) to minimize the potential interference of endogenous Gα_i (Peleg et al. 2002). GIRK1* PM expression was measured in giant excised PM patches (Singer-Lahat et al. 2000; Peleg et al. 2002; Kanevsky & Dascal, 2006) (Fig. 6A) and found to be largely unaffected by either Gα_i3 variant or by Gβγ (summarized in Fig. 6B). Therefore, the recorded whole-cell currents under different treatments were usually compared without correcting them for changes in PM expression.

Coexpression of Gα_i3-wt or Gα_i3GA reduced GIRK1*I_basal by 73% or 95%, respectively (see Fig. 6A for representative current records, and Fig. 6Da for summary), consistent with the observation that GIRK1*I_basal is Gβγ dependent, as in GIRK1/2. Also, like in GIRK1/2 (Ivanina et al. 2004), coexpression of Gα_i3-wt dramatically increased the relative activation by agonist, R_a (7.6-fold, Fig. 6C). In addition, the extent of activation by Gβγ (R_βγ) was greatly increased by Gα_i3: ∼8-fold by Gα_i3-wt (or ∼5-fold if corrected for PM expression changes) and ∼22-fold by Gα_i3GA (Fig. 6Db). Neither Gα_i3-wt nor Gα_i3GA reduced I_βγ, despite their ability to sequester free Gβγ. Coexpression of Gα_i3QL reduced I_basal by ∼40% (Fig. 6Da), possibly due to this mutant's ability to bind Gβγ (Majumdar et al. 2006); however, R_βγ was not significantly improved by Gα_i3QL (Fig. 6Db). Thus, Gα_i3^GDP regulates GIRK1* and GIRK1/2 similarly: it decreases I_basal and improves R_βγ, while preserving I_βγ (Rubinstein et al. 2007).

Regulation of GIRK1* by Gβγ and Gα_i3GA was further examined in excised patches (Fig. 7A). When oocytes were injected with 10–17 ng of GIRK1* RNA, we observed high cell-to-cell and patch-to-patch variability of channel density (range of I_basal in the absence of Gα_i3GA: 0.03–22 pA; Fig. 7B), therefore not all patches could be defined as having high channel density (Peleg et al. 2002). Nevertheless, GIRK1* showed the main hallmarks of regulation by Gβγ and Gα_i3GA, as in whole cells. (1) Like GIRK1/2, homomeric GIRK1* channels exhibited strong negative correlation between I_basal and R_βγ (Fig. 7B, left panel, P < 0.001), and this correlation was absent in the presence of Gα_i3GA (Fig. 7B, right panel). The negative correlation between I_basal and R_βγ is an important hallmark of a concerted regulation of I_basal by Gα and Gβγ (Peleg et al. 2002; Rubinstein et al. 2007). (2) Gα_i3GA expression reduced I_basal by 70% and improved R_βγ (Fig. 7C). (3) The total Gβγ-evoked current was not reduced by Gα_i3GA.

There were, however, quantitative differences with whole-cell data: the average activation of GIRK1* by added Gβγ in excised patches was stronger than by coexpressed Gβγ in whole cells (R_βγ of 12 ± 3.7), and the improvement in R_βγ by Gα_i3GA expression was milder (3-fold). These differences could, at least in part, result from the inclusion in the analysis of low-density patches, which show high R_βγ in the absence of exogenous Gα. Indeed, in seven patches with high I_basal, above 2 pA, R_βγ was 1.01 ± 0.27 (range: 0.27–2.04; Fig. 7B), resembling the low relative activation of Gβγ in whole cells.

Unfortunately, the same definition of channel density could not be used in cells expressing Gα_i3GA because it strongly reduced I_basal. Thus, we attempted to sort out the cells with high expression levels according to high total current, I_βγ (>2 pA). This criterion is applicable to all patches since I_βγ does not depend on the presence of Gα_i3GA (Fig. 7C). Under this definition, R_βγ was 5 ± 3 (n= 10) in control and significantly higher (44 ± 12, n= 7, P < 0.01) in Gα_i3GA-expressing cells (R_βγ was 32 ± 8, n= 12, in all Gα_i3GA-containing patches). We conclude that Gα_i3GA genuinely improves the relative activation of GIRK1* by Gβγ, as in whole cells. However, in general, activation of GIRK1* by Gβγ in excised patches was better than in whole cells, and therefore an involvement of unidentified cytosolic factors in regulating the Gβγ-dependent gating of GIRK1* cannot be ruled out.

GIRK2 is not regulated by Gα_i3^GDP

Wild-type GIRK2 channels or GIRK2_HA channels containing an extracellular HA tag (Clancy et al. 2005) were expressed at 10–15 ng RNA per oocyte with Gα_i3-wt, Gα_i3GA and Gα_i3QL, with or without Gβγ. GIRK2 expression was very sensitive to Gα_i3 and Gβγ: coexpression of Gα_i3-wt and Gα_i3GA reduced PM levels of GIRK2 by up to 80%, while Gα_i3QL had a mild effect (Fig. 8A and Supplemental Fig. 4Aa). Interestingly, coexpression of Gβγ partly restored GIRK2 expression. Thus, PM expression of GIRK2 was monitored in every experiment. GIRK2 and GIRK2_HA PM levels were measured by imaging in giant PM patches using anti-GIRK2 antibody, and in intact oocytes with an anti-HA antibody. We compared the two methods for assessment of the relative PM expression of GIRK2 and GIRK2_HA under various experimental conditions in the same experiment, and found them to give almost identical results (Fig. 8A). These results further validate the notion that the measurements in giant excised PM patches reliably report the surface expression of membrane or membrane-associated proteins (Kanevsky & Dascal, 2006). The results with GIRK2 and GIRK2_HA were pooled as there were no substantial differences in PM expression or the effects of Gα_i3 and Gβγ on channel currents.

After correcting GIRK2 currents for changes in PM expression, we found that I_basal was largely unaffected by Gα_i3-wt or Gα_i3GA (Supplemental Fig. 4Ab). I_βγ, the total Gβγ-evoked current, and R_βγ were also unaffected by coexpression of Gα_i3 (Fig. 8B and Supplemental Fig. 4Ab). However, the above correction assumes a linear relation between channel PM levels and whole-cell currents, which may not always hold. Therefore, we also titrated the channel's expression by increasing the amount of injected RNA in order to get equal PM expression with or without Gα_i3GA and Gβγ (Fig. 8C; see additional details in Supplemental Fig. 4B). When groups of oocytes with equal PM expression levels were compared, expression of Gα_i3GA only slightly decreased I_basal (P > 0.05) with no effect on I_βγ or R_βγ (Fig. 8C). Thus, Gα_i3^GDP does not alter either basal or Gβγ-evoked activity of GIRK2.

In contrast, expression of Gα_i3-wt dramatically increased the agonist response (Fig. 8D): I_total was increased by ∼7-fold and R_a was improved by ∼12-fold (note that R_a is calculated for each oocyte and is therefore independent of the channel expression variations). In this experimental protocol Gα_i3 serves as the donor of Gβγ following the activation by GPCR. It is plausible that overexpression of Gα_i3 increases the fraction of GIRK2 channels associated with Gα_iβγ trimers prior to activation, and in effect increased the number of functionally responsive channels (see Discussion). The increase in I_total by Gα_i3-wt was not observed with GIRK1/2 or GIRK1* (Fig. 6C and Ivanina et al. 2004).

GIRK2 channels were also tested for activation by purified Gβγ in excised patches. Examples of the currents with or without Gα_i3GA are shown in Fig. 9A. Unlike GIRK1* (Fig. 7) or GIRK1/2 (Peleg et al. 2002), the negative correlation between I_basal and R_βγ, if any, was weak and did not reach statistical significance either in wt GIRK2 (data not shown) or GIRK2_HA, with or without Gα_i3GA (Fig. 9B). Further, in agreement with the whole-cell results, coexpression of Gα_i3GA had no effect on GIRK2 I_basal, I_βγ or R_βγ (Fig. 9C).

In conclusion, although Gα_i3 binds both GIRK1 and GIRK2 subunits, regulation by Gα_i3^GDP of I_basal and of Gβγ-induced activation is a unique feature of GIRK1-containing channels.

The C terminus of GIRK1 is important for Gα_i modulation

GIRK1 and GIRK2 differ in sequence and length, especially in the CT, which is ∼320 amino acids long in GIRK1 and only 220 in GIRK2 (see Ivanina et al. 2003). The distal CT, starting approximately at amino acid (a.a.) 370 of GIRK1 (381 in GIRK2), is the region of lowest homology between the two subunits. Logothetis and collaborators pointed out the importance of the unique CT of GIRK1 for large currents in the context of a GIRK1/GIRK4 chimera (Chan et al. 1997). Accordingly, we hypothesized that this segment may play a role in the unique modulation of GIRK1 by Gα_i3. To this end, we constructed a chimeric channel, G2_CTG1, composed of GIRK2_HA in which the distal CT (a.a. 382–414) was replaced with that of GIRK1, a.a. 371–501 (Fig. 10A). Monitoring of the PM expression was done in whole oocytes using an external HA tag. Like GIRK1*, the chimeric G2_CTG1 displayed large basal currents, which were strongly RNA dose dependent, reaching 5.5 μA at 2 ng RNA per oocyte (Fig. 10C and Supplemental Fig. 1C). On the other hand, coexpression of Gα_i3GA reduced the PM expression levels of G2_CTG1 by ∼60%, resembling in this respect the GIRK2 (Fig. 10B). Coexpression of Gβγ induced weak activation with R_βγ of ∼2 and I_βγ of 8.9 μA (or 12.4 μA if corrected to PM expression, Fig. 10C and D). Gα_i3GA coexpression reduced I_basal by 96–98% but enabled full activation by Gβγ, or even enlargement of the I_βγ if corrected for the reduced expression. R_βγ with Gα_i3GA coexpression was increased from 2 to ∼70 (Fig. 10D). Therefore, the distal part of GIRK1 CT is important for Gα and Gβγ regulation. Transferring this segment confers upon GIRK2 most (though not all) of the unique qualities of GIRK1*: high, Gβγ-dependent basal activity and regulation by Gα_i3GA.

Figure 10. The C terminus of GIRK1 is important for Gα_i modulation.

A, schematic presentation of the G2_CTG1 chimera. B, summary of G2_CTG1 expression levels (1–2 ng RNA per oocyte), as measured in whole oocytes using external HA tag. n= 4–7. C, Gα_i3GA (2.5 ng) regulates I_basal and I_βγ of the G2_CTG1 chimera. Grey bars with horizontal hatching indicate the currents after correction to the PM expression levels (from B), diagonal black hatching indicate coexpression of Gβγ. D, R_βγ with (grey) or without (white) correction to the changes in the PM expression. n= 7–24. **P < 0.05; ***P < 0.001.

Discussion

Here we show that Gα_i regulates the Gβγ gating of the neuronal GIRK1/2 channels in Xenopus oocytes and mammalian (HEK 293) cells. Utilizing ‘constitutively inactive’ and ‘constitutively active’ Gα_i3 mutants in functional assays, we find that the ‘inactive’ Gα_i3 in its GDP-bound form reduces I_basal and predisposes GIRK1/2 to Gβγ activation. This action of Gα_i^GDP is Gβγ dependent and involves the formation of Gα_iβγ heterotrimers. We further explored previously unrecognized differences between GIRK1 and GIRK2 in their interaction with, and functional regulation by, Gα_i and Gβγ. GIRK1 channels, in contrast to GIRK2 homomers, are regulated by Gα_i3, and consequently show a strikingly different pattern of regulation by Gβγ. Our results support the hypothesis that the Gα_i/oβγ heterotrimers or free Gα_i^GDP are regulators of gating of GIRK channels, but potentially limit the list of Gα_i/o^GDP-regulated GIRK channels to those containing the GIRK1 subunit.

‘Constitutively inactive’ mutant of Gα, Gα_i3GA, improves the activation of the neuronal GIRK1/2 by Gβγ

We have previously reported that coexpression of Gα_i3-wt reduced the basal activity of GIRK1/2 and enhanced the direct activation of the channels by added Gβγ, bypassing the GPCR and the Gα_iβγ heterotrimer dissociation. We initially called this phenomenon ‘priming by Gα_i’ and proposed that Gα_i regulates GIRK gating, keeping I_basal low and preparing the channel for activation by ‘free’ Gβγ (Peleg et al. 2002; Rishal et al. 2005; Rubinstein et al. 2007). We suggested that GIRK1/2 modulation was mediated by Gα_i^GDP, or Gα_iβγ heterotrimers. Yet, it was also essential to decisively distinguish between effects of Gα_i^GDP and Gα_i^GTP. Xenopus oocytes reportedly contain high basal levels of ‘free’ Gβγ that help to maintain the meiotic arrest. An unidentified constitutively active GPCR has been hypothesized to cause this condition (Evaul et al. 2007), and in such cases a high level of free Gα^GTP is also expected. Here we showed that only the ‘constitutively inactive’ mutant Gα_i3GA improves GIRK1/2 activation by Gβγ. As with Gα_i3-wt (Rubinstein et al. 2007), the improvement is only on the background of a concomitant decrease in basal activity; the total Gβγ-induced activation is not changed, both in whole cells and in excised patches. The constitutively active Gα_i3QL did not significantly affect Gβγ-induced activation of GIRK1/2 (Figs 2 and 3). However, at present we cannot fully rule out a minor direct effect of Gα_i3QL on GIRK1/2.

Differences in interaction between GIRK subunits and Gα_i3βγ

Despite a larger Gβγ and Gα-interacting surface in GIRK1 compared to GIRK2 (Ivanina et al. 2003; Ivanina et al. 2004), no functional asymmetry was previously reported. Full cytosolic domains of GIRK1 and GIRK2 interact with Gβγ and Gα_i3. However, regulation (enhancement) by Gβγ of the Gα_i3^GDP–GIRK interaction is observed only with GIRK1 (Fig. 4). This implies a central role for the GIRK1 subunit in mediating the effects of Gα_i (or Gα_iβγ) on heterotetrameric GIRK1/2 channels, and possibly on other GIRK1-containing heterotetramers. The enhanced binding of Gα_i3^GDP in the presence of Gβγ could be due to stronger binding of Gα_iβγ heterotrimer to GIRK via Gβγ, or an enhancement of the direct binding of Gα_i3 to a site separate from that of Gβγ. In support, strong binding of Gα_i1βγ heterotrimer to the NT of GIRK1 was reported in the past, and this was suggested to reflect the existence of preformed GIRK-G protein signalling complexes (Huang et al. 1995).

Homomeric GIRK2 and GIRK1* channels are distinctly regulated by Gβγ and Gα_i3^GDP

The homomeric GIRK2 channel seems to behave as a ‘classical’ Gβγ effector, displaying very low basal activity in the absence of agonist and strong activation by added Gβγ (see Table 1). Its gating by Gβγ is not regulated by Gα_i3^GDP, remaining unaltered in the presence of coexpressed Gα_i3-wt or Gα_i3GA in whole cells or excised patches. In agreement with the biochemical data, several functional results suggest a weak interaction with Gα_iβγ heterotrimers. (1) GIRK2's basal currents are small and insensitive to Gβγ scavengers or to the coexpression of Gα_i3. (2) Coexpression of either Gα_i3-wt or Gβγ enhanced the overall GIRK2 currents (I_total and I_βγ, respectively) by ∼7-fold, compared to control cells expressing the channel alone (Figs 4F and 8D). Since Gα_i3^GDP does not alter the extent of the channel's activation caused by direct addition of Gβγ, it is most likely that the addition of Gα_i3 simply increases the amount of Gα_i3βγ heterotrimers available for the activation of this channel. These results indicate that the endogenously present Gα_iβγ heterotrimers cannot fully activate all expressed GIRK2 channels, and the majority of GIRK2 homomers (probably more than 80%) lack pre-associated G-proteins (in contrast to GIRK1*, see below). In summary, GIRK2 channels display very strong activation by added Gβγ and no regulation by Gα_i3^GDP. Lack of regulation of GIRK2 by Gα_i3^GDP further emphasizes the authenticity and uniqueness of this regulation in GIRK1-containing channels.

GIRK1* is similar to GIRK1/2 and differs from GIRK2 in two major aspects (Table 1): (1) GIRK1* exhibits a considerable Gβγ-dependent basal activity; and (2) it is strongly regulated by Gα_i3. I_basal of GIRK1* is highly Gβγ dependent; expression of Gβγ-binding proteins m-phosducin, m-cβARK or Gα_i3 reduces I_basal by more than 70%. It is highly unlikely that the high I_basal of GIRK1* (or GIRK1/2) depends on ‘free’ Gβγ elevated by some ‘constitutively active’ GPCR present in the oocyte, because GIRK2 (which is highly sensitive to both Gβγ and ACh) shows almost no Gβγ-dependent I_basal. Recent studies conducted in vivo using fluorescent energy transfer techniques suggest that GIRK1-containing channels are associated with Gβγ both in endoplasmic reticulum (Robitaille et al. 2009) and in the PM (Riven et al. 2006). Taken together, these considerations suggest that the excessive basal activity of GIRK1* at high expression levels is due to an excess of bound Gβγ, and to an insufficient amount of Gα, as in the case of GIRK1/2 (Peleg et al. 2002; Rishal et al. 2005).

The data presented here establish that Gα_i3^GDP, and not Gα_i3^GTP, is responsible for the regulation of I_basal and R_βγ in GIRK1/2 and GIRK1*. Several facts support regulation by Gα_i^GDP, probably via the formation of Gα_i3βγ heterotrimers. (1) Gα_i3-wt and Gα_i3GA reduced I_basal with no reduction in I_βγ, and greatly improved the extent of activation by Gβγ (R_βγ). (2) In excised patches, R_βγ of GIRK1* exhibited strong negative correlation with I_basal, with almost no activation when I_basal is high. Expression of Gα_i3GA abolished the negative correlation and improved R_βγ by ∼9-fold. (3) Expression of Gα_i3-wt improved the relative activation by agonist, R_a, by ∼8-fold, but I_total did not change. Thus, the improvement of I_ACh and R_a correlated with the decrease in I_basal, indicating that the expressed Gα_i3 enhanced the amount of GIRK1-associated Gα_i3 and corrected the balance between the bound Gα_i3 and Gβγ. Together with the biochemical data, these considerations suggest that most of GIRK1* homomers and GIRK1/2 heterotetramers arrive at the PM in complex with Gα_iβγ or with Gβγ alone.

A plausible model to describe the regulation of GIRK1/2 and GIRK1* by Gα_i3^GDP would be one which includes two sites for G protein binding, somewhat similar to the model proposed by Logothetis and colleagues (He et al. 1999). In our two-site model the GIRK channel would have a docking, or anchoring, site for Gα_iβγ, and a separate activation site that binds Gβγ. In a normal physiological situation, Gα_i/oβγ is anchored to GIRK either via Gα_i (Clancy et al. 2005) or via Gβγ; the Gβγ activation site is free. Activated GPCR triggers partial or full separation of Gα_i/o from Gβγ, causing a shift of Gβγ to the activation site and opening of the channel. For reasons as yet unclear, the overexpressed channel is trafficked to the PM with excess Gβγ over Gα_i/o (Rishal et al. 2005), and Gβγ is free to interact with the activation site, and hence the high I_basal. We further propose that coexpressed Gα_i3 binds Gβγ, and the attachment of the resulting Gα_i3βγ heterotrimer to the anchoring site reduces the excessive I_basal and renders the channel activatable by added Gβγ, which can bind to the activation site. In contrast to GIRK1, it appears that GIRK2 does not possess an anchoring site. In a heterotetrameric GIRK1/2, GIRK1 may serve mainly as the Gα_iβγ-anchoring subunit and GIRK2 as the Gβγ acceptor responsible for channel activation.

Intricate regulation of gating within the GIRK1–G protein complex

While in general GIRK1* behaved similarly to GIRK1/2 (Table 1), it displayed several features unexpected from a Gβγ effector, which further indicate a complex mechanism of regulation by G protein subunits. The most unique and unexpected feature of GIRK1* homomers is the reduction in the total current upon coexpression of Gβγ (Fig. 4E). When the channel was expressed alone, ACh evoked substantial currents. However, Gβγ coexpression failed to activate GIRK1* channels, while completely suppressing the agonist response. The disappearance of I_ACh in cells overexpressing Gβγ is observed also with heterotetrameric GIRK1/4 and GIRK1/2 (Reuveny et al. 1994; Lim et al. 1995; Rubinstein et al. 2007) and has been interpreted as indicating full stimulation of GIRK channels by Gβγ. This is not the case with GIRK1* (as there is no stimulation). These properties are entirely incompatible with the classical idea of activation of GIRK1* by ‘free’ Gβγ.

Although the molecular mechanisms underlying the peculiar behaviour of GIRK1* are currently unclear, several scenarios based upon the existence of a preformed GIRK1–Gα_iβγ complex seem plausible. GIRK1's CT has been proposed to inhibit channel activation by Gβγ without directly competing with Gβγ for binding to GIRK (Dascal et al. 1995; Luchian et al. 1997). Such a cytosolic ‘lock’ may hinder the access of external Gβγ while allowing activation by Gβγ derived from the pre-docked Gα_iβγ heterotrimer. Alternatively (or in addition), Gα_iβγ might allosterically regulate GIRK1* activation, if the proper gating of GIRK1* by Gβγ (via the activation site) requires the docking of Gα_iβγ at the anchoring site. Consequently, overexpression of ‘free’ Gβγ may deplete the channels of docked Gα_iβγ by sequestering Gα_i away from the channel (the ‘lock’ may allow the exit of pre-docked Gα_i), weakening the activation by Gβγ when it binds to the activation site. Coexpression of Gα_i would correct the gating both by restoring the reserve of docked Gα_iβγ, and by regulating I_basal. Allosteric interactions between the ‘lock’, Gα_i and Gβγ, and the existence of two Gβγ-binding sites may produce a complex regulation pattern. Furthermore, additional factors may be involved. We deem it unlikely that the GPCR itself is important in Gα_i^GDP-mediated effects, as regulation of GIRK1/2 by Gα_i3 and Gβγ was identical with or without coexpressed GPCRs (Peleg et al. 2002; Rubinstein et al. 2007). Modulators of Gα activity such as Regulators and Activators of G Protein Signaling (RGSs and AGSs, respectively) were proposed to form complexes with and modulate GIRK channels (Jaen & Doupnik, 2006; Wiser et al. 2006). Yet, the level of endogenous RGSs in the oocytes is low (Doupnik et al. 1997), and titrated coexpression of RGS4 and RGS7 has only marginal effects on I_basal of GIRK1/2 (Keren-Raifman et al. 2001). A full understanding of the molecular details of regulation of GIRK1-containing channels by Gα, Gβγ and other factors and the testing of the model proposed above and its alternatives present a challenge for future work.

The unique distal CT of GIRK1 is essential for Gα_i3-dependent regulations

The distal third of the CT of GIRK1 (a.a. 371–501) does not bind Gα_i and does not strongly interact with Gα_i3 or Gβγ, though it probably possesses a low-affinity Gβγ binding site (Ivanina et al. 2003, 2004). Nevertheless, we find that regulation of GIRK1 by Gα_i3 involves this segment of the CT. Its transfer from GIRK1 to GIRK2 conveyed upon GIRK2 most of GIRK1 properties, notably high I_basal and strong modulation by Gα_i3GA (a high I_basal in a similar GIRK4–GIRK1 chimera has been previously noted by Chan et al. (1997)). These results imply that the end of GIRK1 CT (a.a 371–501) may be important for the anchoring of Gα_iβγ (possibly by interacting with the other parts of the channel rather than with Gα_iβγ itself). Yet, the G2_CTG1 chimera could still be activated by coexpressed Gβγ (weak activation, R_βγ of ∼2), and thus other parts of GIRK1 may be additionally involved in its unique properties. It is also possible that in G2_CTG1 as well as in GIRK1/2 activation is enabled due to the preservation of Gβγ binding sites of GIRK2, as the latter appears to be an excellent sensor of free ‘added’ Gβγ.

Possible physiological consequences of GIRK1/2 asymmetry

The brain expresses all GIRK subunits, predominantly heterotetrameric GIRK1/2 (hippocampus, cerebellum, cortex), GIRK1/3 and GIRK2/3, and homomeric GIRK2 (substantia nigra) (Koyrakh et al. 2005). Variable regional distribution of GIRK2 homomers vs. GIRK1/2 implies distinct properties and roles of these channels in different neurons, rendering the discovered differences between GIRK1 and GIRK2 physiologically relevant. We propose that GIRK1-containing channels contribute to the regulation of both basal and neurotransmitter-induced excitability in neurons (indeed, in hippocampal neurons GIRKs were found to contribute to the resting potential; Chen & Johnston, 2005; Wiser et al. 2006), whereas GIRK2 homomers serve as a low-noise, high-gain neurotransmitter-induced inhibitory relay.

Glossary

Abbreviations

CT

C-terminus

GPCR

G protein-coupled receptor

GST

glutathione-S-transferase

HA

hemagglutinin

NT

N-terminus

PM

plasma membrane

YFP

yellow fluorescent protein

wt

wild-type

Author contributions

N.D., M.R., T.I., S.P., S.B. and C.W.D. conceived the main ideas; N.D., M.R., S.P., T.I., D.B, T.K.-R. and S.B. designed and performed the experiments; M.R. and N.D. wrote the paper; all authors participated in the analysis and interpretation of data, and critically revised and approved the paper.