The G Protein α Subunit Has a Key Role in Determining the Specificity of Coupling to, but Not the Activation of, G Protein-gated Inwardly Rectifying K+ Channels*

In neuronal and atrial tissue, G protein-gated inwardly rectifying K+ channels (Kir3.x family) are responsible for mediating inhibitory postsynaptic potentials and slowing the heart rate. They are activated by Gβγ dimers released in response to the stimulation of receptors coupled to inhibitory G proteins of the Gi/o family but not receptors coupled to the stimulatory G protein Gs. We have used biochemical, electrophysiological, and molecular biology techniques to examine this specificity of channel activation. In this study we have succeeded in reconstituting such specificity in an heterologous expression system stably expressing a cloned counterpart of the neuronal channel (Kir3.1 and Kir3.2A heteromultimers). The use of pertussis toxin-resistant G protein α subunits and chimeras between Gi1 and Gs indicate a central role for the G protein α subunits in determining receptor specificity of coupling to, but not activation of, G protein-gated inwardly rectifying K+ channels.

G protein-gated, inwardly rectifying K ϩ channels were first identified in atrial myocytes where they were shown to be activated by acetylcholine at muscarinic m 2 receptors (1), and stimulation of this current is responsible for slowing of the heart rate in response to vagal nerve stimulation. It was subsequently shown that this activation was sensitive to pertussis toxin (PTx), 1 implicating the family of G i/o proteins (2)(3)(4). It is now apparent that G protein-gated inwardly rectifying K ϩ currents are present in many neuronal cell types, and they are involved in the postsynaptic inhibitory effects of stimulating G i/o -coupled receptors such as GABA B and adenosine A 1 (5).
Cloning efforts from a number of laboratories have revealed the molecular counterparts of these currents (6 -10). The channel is a heteromultimer of members of the inwardly rectifying Kir3.x family of potassium channels. Co-expression of Kir3.1 with Kir3.2, Kir3.3, or Kir3.4 results in currents that show many of the basic characteristics of the native channels in neurons and atria (11)(12)(13). Homomultimers of splice variants of Kir3.2 may also form in certain neuronal regions (14).
Channel activation of these currents in native tissues and of the cloned counterparts in heterologous expression systems was shown to be membrane-delimited (4,15), involving a direct interaction with the G protein ␤␥ dimer (16). This finding was initially controversial, but there are now numerous lines of evidence that point to the involvement of the ␤␥ subunit as the direct channel activator (17). Recent studies have focused on domains and residues involved in G protein ␤␥ dimer binding (18 -24) and the cell biology of channel biogenesis (25)(26)(27).
The question remains as to how receptor specificity is achieved and what are the key determinants involved in this. In native tissues only G i/o -coupled receptors, and not G s -coupled receptors, activate these channels, and yet both should liberate free G protein ␤␥ upon receptor stimulation. Receptor selectivity does not lie at the level of the G protein ␤␥ dimer since a variety of different ␤␥ subunit combinations have been shown to be similarly effective at potentiating these currents (28). In this study we have used a mammalian expression system (HEK293 cells) to investigate the neuronal-type Kir3.1ϩ3.2A channels and their modulation by a variety of G protein-coupled receptors. We have demonstrated the activation of Kir3.1ϩ3.2A channels by G␤␥ and by agonist stimulation of G i/o -coupled receptors but not by G s -coupled receptors. We present evidence to suggest a key role for the G␣ subunit in determining the specificity of coupling between receptor and channel.

Cell Culture and Transfection
HEK293 cells (human embryonic kidney cell line) were maintained in minimum essential medium supplemented with 10% fetal calf serum and 1% penicillin/streptomycin (stocks: 10,000 units/ml penicillin, 1 mg/ml streptomycin), at 37°C in humidified 95% O 2 , 5% CO 2 . A lipidbased transfection procedure was used (LipofectAMINE; Life Technologies, Inc.) and monoclonal cell lines established by picking single colonies of cells following transfection and growth under selective pressure. For the channel-expressing HKIR3.1/3.2 line, 727 g of G418 (Life Technologies, Inc.) was used and for channelϩreceptor-expressing lines, we used a dual selection strategy with 727 g of G418 and 364 g/ml Zeocin (Invitrogen). For transient expression of receptors, G proteins, and other components, cells were co-transfected with cDNA (100 -150 ng) encoding the humanized, red-shifted variant of jellyfish green fluorescent protein (pEGFP-N1; CLONTECH) and cells visualized using a Nikon Diaphot epifluorescent microscope.

Molecular Biology
Standard molecular cloning and mutagenesis techniques were employed throughout. All G protein-coupled receptors and G protein subunits, mutants, and chimeras were cloned into pcDNA1, pcDNA3, or pcDNA3.1/Zeo(ϩ) (Invitrogen). All cDNAs were sequenced to confirm * This work was supported by the Human Frontiers Science Program and the Wellcome Trust. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈTo whom all other correspondence should be addressed. their identity. For the expression of Kir3.1ϩ3.2A, a bicistronic vector was engineered that enabled both subunits to be expressed in the same vector. This vector contained two restriction enzyme polylinkers around a picornavirus internal ribosome entry site (IRES). An IRES-containing plasmid (pIRES1hyg) was obtained from CLONTECH, and the IRES element was amplified using PCR with oligonucleotides containing additional restriction enzyme sites inserted at the 5Ј and 3Ј ends using PCR. This fragment was subcloned into pcDNA3. The channel subunits were then subcloned into the newly engineered pcDNA3 and a stable HEK293 cell line established (HKIR3.1/3.2). For some experiments, we made stable cell lines that expressed both channel and receptor. In these instances, receptors (in pcDNA3.1/Zeo(ϩ)) were transfected into the stable Kir3.1ϩ3.2A line and selected for using a dual selection strategy. Point mutants of G␣ i1 (C351G and C351I) were made as in Ref. 29. COOH-terminal chimeras between G␣ i1 and G␣ s were made using a PCR-based approach with a high fidelity DNA polymerase (Vent; New England Biolabs) and 18 -25 cycles. To replace the COOHterminal 6 amino acids, a single round of PCR was used and appropriate 5Ј and 3Ј restriction enzyme sites added. For the chimeras where the COOH-terminal 13, 16, and 20 amino acids of G␣ i1 were replaced with G␣ s , two rounds of PCR were employed. The first round replaced the COOH-terminal 10, 13, and 17 amino acids, respectively, of G␣ i1 with those of G␣ s , and appropriate 5Ј restriction enzyme sites while the second round added the remaining 3 amino acids, stop codon, and 3Ј restriction enzyme sites.

Biochemistry
Western Blotting-SDS-polyacrylamide gel electrophoresis and transfer of proteins to nitrocellulose was performed using a Bio-Rad minigel system according to the manufacturer's instructions. Cells were harvested into phosphate-buffered saline and total cell homogenate lysed with an equal volume of 2ϫ gel loading buffer and probed with a primary polyclonal rabbit antibody to G␣ s (Santa Cruz) and the appropriate secondary antibody before visualization of bands using the ECL Western blot analysis kit (Amersham Pharmacia Biotech).
Adenylate Cyclase Assay-For the measurement of A 2A and ␤ 1 receptor-mediated accumulation of cAMP, the cell lines HKIR3.1/3.2/A2A and HKIR3.1/3.2/B1 were grown to 20% confluence in six-well dishes. Cells were pre-labeled with 5 Ci of [ 3 H]adenine/well (in minimum essential medium) overnight at 37°C and then incubated with adenosine deaminase (1 unit; for A 2A receptors only) and the phosphodiesterase inhibitor Ro20 -1724 (100 M) in serum-free medium, for 30 min at 37°C. Agonist (1 M NECA for A 2A ; 10 M isoprenaline for ␤ 1 ) was then added and incubated for 15 min at 37°C. Medium was then aspirated and cells washed with serum-free medium. Reactions were terminated by the addition of 2.5% perchloric acid and 0.1 mM cAMP at 4°C. [ 3 H]cAMP was isolated by sequential chromatography using Dowex 50-alumina columns. Each fraction was collected in a scintillation vial containing 5 ml of Ultima Gold MV scintillant (Packard). A Beckman LS6000TA liquid scintillation counter was used to count radioactivity. Reactions were done in triplicate, and data are expressed as percentage conversion of [ 3

H]ATP to [ 3 H]cAMP.
Radioligand Binding-Cells were washed and harvested into binding buffer (50 mM Tris-HCl, 1 mM MgCl 2 , pH 7.4) in the presence of protease inhibitors. They were then hypotonically shocked (10 mM Tris-HCl) and put on ice for 15 min, after which time an equal volume of 500 mM sucrose plus 10 mM Tris-HCl was added to restore tonicity. Cells were homogenized using a glass-on-glass Dounce homogenizer. The homogenate was spun at 4°C and the cell pellet resuspended in binding buffer. Specific binding was assessed using the radioligands [ 3 H]DPCPX (A 1 ) and [ 3 H]CGS21680 (A 2A ) and nonspecific binding determined in the presence of non-labeled R-PIA. Binding reactions were incubated for at least 2 h at room temperature before the reaction was terminated by the addition of 1 ml of ice-cold binding buffer. Bound ligand was separated from non-bound ligand by vacuum filtration onto Whatman GF-B filters, which were then washed four times with 2 ml of ice-cold binding buffer. Bound radioactivity was determined by liquid scintillation. Points were determined in triplicate, and data were pooled. Binding data were fitted using non-linear regression to a saturation binding isotherm.

Electrophysiology
Membrane currents were recorded using the whole-cell configuration of the patch clamp technique (30) with an Axopatch 200B amplifier (Axon Instruments). Patch pipettes were pulled from filamented borosilicate glass (Clark Electromedical) and had a resistance of 1.5-2.5 megohms when filled with pipette solution (see below). Prior to filling, the tips of patch pipettes were coated with a Parafilm/mineral oil suspension. Records were were filtered at 1 kHz, digitized at 5 kHz, and data acquired and analyzed using a Digidata 1200B interface (Axon Instruments) and the pClamp suite of software (version 6.0; Axon Instruments). Cell capacitance was approximately 15 pF, and series resistance (Ͻ10 megohms) was at least 75% compensated. Recordings of membrane current were commenced after an equilibration period of approximately 5 min. Currents were measured at the end of each voltage step. All data are presented as mean Ϯ S.E., and current densities are measured at Ϫ100 mV (unless otherwise stated). Student's t tests were performed to examine statistical significance, and * indicates p Յ 0.05.

Materials and Drugs
Solutions were as follows (concentrations in mM): pipette solution, 107 KCl, 1.2 MgCl 2 , 1 CaCl 2 , 10 EGTA, 5 HEPES, 2 MgATP, 0.3 Na 2 GTP (pH 7.2); bath solution, 140 KCl, 2.6 CaCl 2 , 1.2 MgCl 2 , 5 HEPES (pH 7.4). All materials for cell culture were obtained from Life Technologies, Inc. and Invitrogen. Radioligands were purchased from NEN Life Science Products and antibodies from Santa Cruz. Molecular biology reagents were obtained from New England Biolabs or Roche Molecular Biochemicals. All chemicals were from Sigma, Calbiochem, or RBI. Oligonucleotides were from Genosys Biotechnologies. Drugs were made up as concentrated stock solutions and kept at Ϫ20°C or Ϫ80°C.

Characterization of Kir3.1ϩ3.2A Cell Line-Channel-form-
ing subunits, receptors, and G proteins were expressed in a mammalian cell line (HEK293). These cells are electrically silent (in non-transfected cells, current density measured in symmetrical K ϩ solutions at Ϫ100 mV was 9.6 Ϯ 1.4 pA/pF, n ϭ 24) and do not appear to possess significant endogenous inwardly rectifying K ϩ currents. The native neuronal channel is a heterotetrameric structure thought to comprise both Kir3.1 and Kir3.2 subunits. In initial experiments, the cDNAs encoding Kir3.1 and Kir3.2A were transiently transfected into cells in separate vectors (pcDNA3). No inwardly rectifying K ϩ currents were observed when individual subunits were transfected alone ( Fig. 1A), but transfection of both subunits together resulted in the expression of strong inwardly rectifying K ϩ currents. We subcloned both Kir3.1 and Kir3.2A subunits into a single bicistronic vector around an IRES (Fig. 1B), which allowed the expression of both subunits on a single plasmid. A monoclonal stable line expressing Kir3.1ϩ3.2A (hereafter designated as HKIR3.1/3.2) was then established using this vector, and currents were recorded using the whole cell configuration of the patch clamp technique. The inwardly rectifying K ϩ currents were blocked by external Ba 2ϩ , and an example of these currents and the corresponding current-voltage relationship is shown in Fig. 1 (C and D).
G Protein Regulation of Kir3.1ϩ3.2A Channels-The family of Kir3.x channels are regulated by heterotrimeric G proteins. To mimic the activation of G proteins, we used the non-hydrolyzable analogue of GTP, GTP␥S. Basal currents were potentiated by the inclusion of 500 M GTP␥S in the pipette solution (Fig. 1E). It has been established that free G␤␥ subunits but not G␣ subunits mediate channel activation through a direct protein-protein interaction between determinants on the G␤␥ dimer and cytoplasmic domains on the Kir3.x subunit (18 -24, 31). The currents expressed in our monoclonal cell line demonstrate behavior that is consistent with this body of work. It was possible to transiently transfect the HKIR3.1/3.2 line and identify transfected cells after co-transfection with green fluorescent protein using epifluorescence. Transient transfection of exogenous ␤ 1 ␥ 2 subunits significantly increased basal current density, whereas overexpression of G␣ subunits from G i1 and G i2 significantly reduced basal current densities (Fig. 1E). These data confirm that G␤␥ activates the channel, while the reduction in current density by overexpression of ␣ subunits is consistent with sequestration of free endogenous ␤␥ dimers. Interestingly, G␣ s also acted to suppress basal current (in the absence of receptors) to an extent similar to that observed with G␣ i subunits (data not shown). However, when receptors (either G i/o -or G s -coupled) were co-expressed, no suppression of basal current was seen.
Potentiation of Kir3.  Fig. 2B, upper panel) confirming that channel stimulation was mediated by the G i/o family of G proteins. In contrast, agonist stimulation of the G s -coupled receptors, A 2A , ␤ 1 , and D 1 , was unable to significantly activate currents except for a small increase in current density observed with stimulation of the ␤ 1 adrenergic receptor (Fig. 2B, lower panel). Interestingly this effect was PTx-sensitive, suggesting that ␤ 1 can also couple to G i/o proteins to activate the channel. PTx treatment to inhibit any G i/o proteins that might be competing for G s combined with the overexpression of G␣ s still did not allow the G s -coupled receptors to activate the Kir3.1ϩ3.2A currents (Fig. 2C).
To ascertain that this observation of specificity of receptor coupling was not an anomaly of the expression system, for example, due to stronger or more efficient transient expression of G i/o -coupled receptors than G s -coupled receptors, stable monoclonal cell lines were established that expressed both channel subunits and a receptor (A 1 denoted as HKIR3.1/3. Mean basal current density measured at Ϫ100 mV was 112 Ϯ 17 pA/pF (n ϭ 41). Currents were reversibly inhibited by external Ba 2ϩ (middle traces); 1 mM inhibited currents by 81 Ϯ 2% (n ϭ 17). The lower traces show recovery after removal of Ba 2ϩ . D, corresponding current-voltage relationship for the data shown in C. E, the inclusion of GTP␥S in the pipette solution enhanced basal current density from 103 Ϯ 27 pA/pF to 229 Ϯ 52 pA/pF (n ϭ 9, p Ͻ 0.01). Transfection of exogenous G␤␥ also increased current density from 112 Ϯ 17 pA/pF (n ϭ 41) to 701 Ϯ 226 pA/pF (n ϭ 14, p Ͻ 0.01), in contrast to the reduction in current density observed with the transfection of ␣ subunits from G i1 (41 Ϯ 6 pA/pF, n ϭ 10, p ϭ 0.05), G i2 (20 Ϯ 3 pA/pF, n ϭ 10, p ϭ 0.01) or the PTx-insensitive mutant G␣ i1 C351G (34 Ϯ 8 pA/pF, n ϭ 8, p ϭ 0.05; see "Results").

Confirmation of Expression of Functional Receptors Coupled to an Intact Second Messenger
System-The lack of potentiation of currents by stimulation of G s -coupled receptors was not due to low expression levels of G␣ s since overexpression, together with relevant receptor in PTx-treated cells, did not significantly enhance currents (Fig. 2C). In order to confirm that these cells express endogenous G␣ s , we performed Western blots on total cell homogenate from the HKIR3.1/3.2/A1 and HKIR3.1/3.2/A2A lines. Clearly, the HKIR3.1/3.2/A1 line expressed G␣ s . In the HKIR3.1/3.2/A2A line, expression of G␣ s was lower but was revealed upon longer exposure of the blot. Transfection of the G␣ s -containing plasmid led to increased protein expression (Fig. 3A).
It is clear that the G i/o -coupled receptors were functionally expressed since K ϩ currents were potentiated by stimulation of these receptors. However, since no potentiation was observed with the G s -coupled receptors, it was important to determine that these receptors were actually being expressed at the cell membrane and were functionally coupled to downstream second messenger pathways.
We investigated the biochemical characteristics of the monoclonal lines HKIR3.  (Fig. 3B). It was noted that the A 2A receptor was stably expressed at lower levels than the A 1 receptor; this is an observation that has also been made in CHO cells (32).
To confirm signaling to a downstream effector, namely adenylate cyclase, agonist-induced cAMP accumulation was measured in the HKIR/3.1/3.2/A2A and HKIR3.1/3.2/B1 lines. A large increase in cAMP levels was observed in both lines in response to receptor stimulation (Fig. 3C).
Coupling of Channels to PTx-insensitive G Protein Mutants-Thus far, we report that, as in native tissue, Kir3.1ϩ3.2A currents were strongly activated by G␤␥ but not G␣ subunits and by stimulation of G i/o -but not G s -coupled receptors. To further investigate this specificity of channel activation, we took advantage of the PTx sensitivity of the G i/o family of G proteins. These G proteins have a conserved cysteine residue in the ␣ subunit 4 amino acids from the COOH terminus, which is ADP-ribosylated by PTx (33). Mutation of this residue prevents modification by PTx, thus resulting in a PTx-insensitive G protein (34,35). G␣ i1 was mutated at this position to glycine or isoleucine (29). After transiently transfecting receptors and mutant G␣ i1 subunits into the HKIR3.1/3.2 line, cells were treated with PTx to preclude coupling between receptor and endogenous G i/o , but still allow the study of coupling between receptor and mutant G␣ i1 . Expression of these mutants alone did not enhance membrane currents and in fact significantly reduced basal current density similarly to wild type G␣ i1 (Fig.  1E) and in PTx-treated cells, stimulation of any of the G i/ocoupled receptors tested were unable to enhance Kir3.1ϩ3.2A currents (see Fig. 2B). When the mutant G␣ i1 subunits, G␣ i1 C351G and G␣ i1 C351I, were co-expressed, agonist stimulation of receptor led to a large enhancement of currents (Fig.  4A). This was observed with both A 1 (Fig. 4B) and ␣ 2A (Fig. 4C) receptors. In analogous experiments, expression of G␣ s in PTxtreated cells did not support coupling between Kir3.1ϩ3.2A and any receptors (Figs. 2C and 4B). Thus, the PTx-insensitive G␣ i1 subunits were able to rescue signaling between G i/o -coupled receptors and Kir3.1ϩ3.2A in PTx-treated cells.
A COOH-terminal Chimera between G␣ i1 and G␣ s Completely Swaps Coupling between Receptor and Kir3.1ϩ3.2A Channels-Our data suggest that G␣ is important in the determination of receptor selectivity of channel activation. This hypothesis was further investigated by the use of chimeric G proteins. Since the COOH-terminal 20 amino acids of G protein ␣ subunits are implicated in the selectivity of interactions with G protein-coupled receptors (36), we made a series of COOHterminal chimeras between the G proteins G␣ i1 and G␣ s , where between 6 and 20 amino acids of G␣ i1 were replaced with the corresponding residues of G␣ s . These experiments were done in PTx-treated cells to prevent interaction of receptor with endogenous G i/o , and the data are shown in Fig. 5. Strikingly, a 13-amino acid chimera (GiGs13) allowed a complete swap of receptor coupling; A 2A receptor stimulation by NECA strongly enhanced currents to a similar level observed with A 1 receptor stimulation under control (non-PTx-treated) conditions (Fig.  5A). Further replacements, GiGs16 and GiGs20, had similar, although not as profound, effects (Fig. 5B). The effects of the GiGs13 chimera were not unique to the A 1 /A 2A receptors; this chimera was also able to swap receptor/channel coupling from the ␣ 2A to the ␤ 1 adrenergic receptor (Fig. 5C). Thus, we have managed to potentiate the Kir3.1ϩ3.2A currents via G s -coupled receptors by using chimeric G proteins between G␣ i1 and G␣ s .
Disruption of Coupling between the A 1 Receptor and Kir3.1ϩ3.2A by a COOH-terminal Chimera between G␣ s and G␣ i1 -Further evidence that supports a key role for the G protein ␣ subunit is a dominant negative effect that we observed with a COOH-terminal chimera between G␣ s and G␣ i1 ; this chimera is mainly G␣ s but has the COOH-terminal 13 amino acids of G␣ i1 (GsGi13). This acted to significantly disrupt coupling between the A 1 receptor and Kir3.1ϩ3.2A (a reduction of approximately 70%; Fig. 6). NECA-induced currents in control cells were 217.0 Ϯ 38.9 pA/pF (n ϭ 15), whereas in cells where GsGi13 was expressed, induced currents were significantly reduced: 68.6 Ϯ 21.3 pA/pF (n ϭ 20, p ϭ 0.045). This dominant negative effect is likely to be specific, as overexpression of G␣ s with the A 1 receptor did not suppress basal current (Fig. 4B).

Reconstitution of Specificity of Activation of Kir3.1ϩ3.2A
Channels-One of the aims of this study was to examine whether the specificity of Kir3.1ϩ3.2A channel activation by G protein-coupled receptors, widely observed in native tissue such as neuronal and atrial cells, could be reconstituted in an heterologous expression system. By making a stable channelexpressing mammalian cell line (HKIR3.1/3.2), we were able to reconstitute the specificity of channel activation. We have shown that only G i/o -coupled receptors, but not G s -coupled receptors, activate Kir3.1ϩ3.2A channels in our HEK293 cell line. Even overexpression of G␣ s did not lead to current enhancement through G s -coupled receptors, either in PTx-treated or non-treated cells, suggesting that G␣ s does not participate in the activation of these channels. Thus, in our system, G␤␥ dimers derived from inhibitory heterotrimeric G proteins are able to activate Kir3.1ϩ3.2A heteromultimeric currents while the stimulatory G s is not. A number of studies have demonstrated that, under given conditions, it is possible to activate Kir3.x or native currents through G s -coupled receptors (37)(38)(39). However, some issues arise in relation to these and our work. Often a single receptor pathway has been used. Ruiz-Velasco and Ikeda (38) found they could activate Kir3.1ϩ3.2 and Kir3.1ϩ3.4 channels heterologously expressed in superior cervical ganglion neurons through the vasoactive intestinal peptide (VIP) receptor. VIP does not exclusively stimulate G s -coupled receptors: it has been shown to couple to G␣ i1/2 (40) and also to elevate intracellular Ca 2ϩ levels (41). The VIP-mediated effects were enhanced by pre-treatment with PTx, leading to the authors' suggestion that the uncoupling of PTx-sensitive G proteins allowed more PTx-insensitive G proteins to activate the channels. We found that PTx had no effects on the inability of G s -coupled receptors to activate these channels and did not observe channel activation by G s -coupled receptors either in the presence or absence of PTx. Lim et al. (37) found potentiation of currents though the ␤ 2 adrenergic receptor but it is now appreciated that this receptor is able to couple to G i (42,43). Promiscuous coupling of receptors to more than one G protein is apparent with many receptors (44 -47). One explanation advanced for the specificity of coupling is that there is more G i/o than G s in most cells; indeed, in the above studies, it was often necessary to overexpress G␣ s and in some instances the methods used to elevate G s were not direct. Sorota et al. (39) used infection with an ade-novirus, and this may have nonspecific and toxic effects on the cells. In our hands, even the overexpression of G s did not lead to coupling.
Specificity of coupling may not be an absolute phenomenon and may depend on the relative levels of expression of channel, G protein, and receptor. Even though the current densities in our heterologous system are higher than seen in native neurons, we are still able to demonstrate an essentially absolute selectivity mechanism more analogous to that seen in native tissue. In the studies detailed above, no direct comparison is made with a G i/o -coupled pathway under equivalent conditions. Indeed, in the work of Sorota et al. (39) on the native current in atrial myocytes, the levels of current activation though the ␤ adrenergic receptor are lower than that seen though the muscarinic receptor, even after the overexpression of G␣ s .
A Key Role for the G ␣ Subunit in Determining Specificity of Channel Activation-We have three lines of evidence that the G␣ subunit is the key determinant of selectivity of channel activation by G i/o -linked heptahelical receptors. First, the transfection of PTx-resistant mutants of G␣ i1 was able to rescue coupling between G i/o -coupled receptors and Kir3.1ϩ3.2A, but overexpression of G s with G s -coupled receptors under similar conditions was not able to do so. Second and most tellingly, it was possible to swap coupling between these families of receptor by exchanging only a few amino acids on the COOH terminus of G i with those of G s . Finally, we were able to disrupt coupling between a G i/o -coupled receptor (A 1 ) and Kir3.1ϩ3.2A by using a G␣ s chimera that contained the COOH-terminal 13 amino acids of G␣ i1 . A similar strategy was used by Gilchrist et al. (48), who constructed "minigenes" of the COOH-terminal 11 amino acids of G␣ i1 and showed selective disruption of M 2 muscarinic receptor coupling to Kir3.1ϩ3.4.
Schreibmayer et al. (49) showed inhibition of G␤ 1 ␥ 2 -induced currents by activated G␣ i1 but not G␣ i2 or G␣ i3 . In contrast, we have found that G␣ subunits inhibited basal (non-agonist-stimulated) currents (presumably due to sequestration of free ␤␥ subunits) but that PTx-insensitive mutants of G␣ i1 actually supported coupling between receptor and channel. However, there are a number of methodological differences in expression system and experimental design; for example, our studies are performed in the whole cell configuration, while those studies were largely performed in excised patches in the inside-out configuration.
Expression in PTx-treated cells of a COOH-terminal chimera, in which 13 amino acids of G␣ i1 were replaced with the corresponding residues of G␣ s , swaps the receptor coupling profile, i.e. G s -coupled receptors can now stimulate the channel whereas G i/o -coupled receptors cannot. With 16 and 20 amino acid replacements, a swap of coupling was still observed, although it was not quite so profound. This apparent "drop-off" could simply be due to less efficient protein folding of these chimeras, and a similar finding was observed by Conklin et al. (50). Several investigators have used a chimeric G protein approach to alter the fidelity of receptor activation, i.e. receptors not normally coupled to a particular class of G protein then become able to stimulate that G protein (50 -53). It has been shown that substitution of as few as 4 amino acids can alter receptor signaling to G proteins (50). Replacing the COOHterminal 4 amino acids of G q with those of G␣ i2 allowed the stimulation of phospholipase C by D 2 dopaminergic and A 1 adenosine receptors, which normally couple exclusively to G i . However, whether a swap of coupling occurred was not established, i.e. that G q -coupled receptors are no longer able to stimulate phospholipase C activity with the G␣ q /G␣ i chimera. Our data suggest that it is possible with some receptor groups to achieve a clean switch by COOH-terminal replacement. In addition, the carboxyl terminus of G␣ subunits may not be the sole determinant of mediating specificity as other regions have been implicated in receptor interactions, notably at the NH 2 terminus (36,54). It is likely that the extent of the involvement of the COOH terminus and other regions of the G␣ subunits varies with receptor; indeed, this was found by Conklin et al. (50) with the A 1 and D 2 receptors and their coupling to G q /G i chimeras.
Our data demonstrate that we can reconstitute specificity of coupling between seven helix receptors and Kir3.1ϩ3.2A channels. Basal currents are enhanced by the overexpression of the G␤␥ dimer but not the G␣ subunit. The pivotal role of the G␤␥ dimer in channel activation is not being contested. We have no evidence of channel activation via G s -coupled receptors, even when PTx-sensitive G proteins are uncoupled and/or G␣ s is overexpressed. By using single point mutations of G␣ i we can still activate the channel via G i/o -coupled receptors in PTxtreated cells and by making chimeras between G␣ i1 and G␣ s we can swap coupling from G i/o -to G s -coupled receptors and Kir3.1ϩ3.2A. Our finding that COOH-terminal G␣ i1 /G␣ s chimeras could swap receptor coupling to channel strongly implicates the ␣ subunit as playing an important role.
Do similar considerations apply to other effectors where G␤␥ subunits are the direct mediators such as the inhibition of voltage-gated calcium channels of the N and P/Q type (55)(56)(57)(58)? It is apparent that calcium channel inhibition can be mediated through a number of receptors coupling to a number of different families of G proteins so the mechanisms may be different (59,60).
The molecular mechanisms underlying receptor specificity of Kir3.x activation are largely unknown. It has been proposed without strong supportive evidence that there is compartmentalization of signaling components: either between receptor and channel, G protein and channel, or all three. Our data suggest a key role for the ␣ subunit and seem to make it unlikely that there is compartmentalization between receptor and channel as receptor selectivity can be switched with G␣ i -G␣ s chimeras. So what might the possible relationship be between the G␣ subunit and the channel complex? This could be via a direct protein-protein interaction or indirectly involving other accessory proteins. The heterotrimeric G protein has been found to bind to the NH 2 terminus of Kir3.1 (18,61), but whether G␣ i -containing heterotrimers were preferentially bound in comparison to G␣ s -containing heteotrimers was not addressed. Our data are an important insight in understanding how receptor selectivity is achieved in a G␤␥-activated system and provides a basis for future mechanistic studies.