Identification of a Potassium Channel Site That Interacts with G Protein βγ Subunits to Mediate Agonist-induced Signaling*

Activation of heterotrimeric GTP-binding (G) proteins by their coupled receptors, causes dissociation of the G protein α and βγ subunits. Gβγ subunits interact directly with G protein-gated inwardly rectifying K+ (GIRK) channels to stimulate their activity. In addition, free Gβγ subunits, resulting from agonist-independent dissociation of G protein subunits, can account for a major component of the basal channel activity. Using a series of chimeric constructs between GIRK4 and a Gβγ-insensitive K+ channel, IRK1, we have identified a critical site of interaction of GIRK with Gβγ. Mutation of Leu339 to Glu within this site impaired agonist-induced sensitivity and decreased binding to Gβγ, without removing the Gβγcontribution to basal currents. Mutation of the corresponding residue in GIRK1 (Leu333) resulted in a similar phenotype. Both the GIRK1 and GIRK4 subunits contributed equally to the agonist-induced sensitivity of the heteromultimeric channel. Thus, we have identified a channel site that interacts specifically with Gβγsubunits released through receptor stimulation.

Activation of heterotrimeric GTP-binding (G) proteins by their coupled receptors, causes dissociation of the G protein ␣ and ␤␥ subunits. G ␤␥ subunits interact directly with G protein-gated inwardly rectifying K ؉ (GIRK) channels to stimulate their activity. In addition, free G ␤␥ subunits, resulting from agonist-independent dissociation of G protein subunits, can account for a major component of the basal channel activity.
Using a series of chimeric constructs between GIRK4 and a G ␤␥ -insensitive K ؉ channel, IRK1, we have identified a critical site of interaction of GIRK with G ␤␥ . Mutation of Leu 339 to Glu within this site impaired agonistinduced sensitivity and decreased binding to G ␤␥ , without removing the G ␤␥ contribution to basal currents. Mutation of the corresponding residue in GIRK1 (Leu 333 ) resulted in a similar phenotype. Both the GIRK1 and GIRK4 subunits contributed equally to the agonistinduced sensitivity of the heteromultimeric channel. Thus, we have identified a channel site that interacts specifically with G ␤␥ subunits released through receptor stimulation.
Signaling through GTP-binding (G) proteins depends on dissociation of the heterotrimer G ␣␤␥ into the G ␣ -GTP and G ␤␥ subunits. Direct interactions of G ␣ or G ␤␥ (or both) with effector proteins transduces the external signal into an intracellular response. Atrial potassium (K ϩ ) channels, the first example of a G ␤␥ effector (1), are responsible for the acetylcholine(ACh) 1induced reduction in heart rate during vagal activity (2).
Five members of the G protein-gated inwardly rectifying K ϩ (GIRK1-5) channel subfamily have been reported thus far (3)(4)(5)(6)(7)(8). The presumed topology of these channels includes a cytoplasmic N terminus (ϳ90 amino acids), followed by two transmembrane domains with the "ion selectivity" P-region in between (ϳ100 amino acids) and ending with a long cytoplasmic C terminus (over 200 amino acids) (3,9). GIRK channels can function as highly active heteromultimers (pairing of GIRK1 with any other subtype) or low to moderately active homomultimers (GIRK2-5) (for review, see Ref. 10). Mutations at a specific position within the P-region of these channels ("P-region mutants", e.g. GIRK4-S143T) greatly enhance the activity of homomultimers (11,12). Use of these highly active point mutants simplifies the experimental design of structurefunction studies and allows assessment of the relative contributions of each of the two subunits in the heteromultimeric complex (12).
Several studies have demonstrated direct binding of G ␤␥ subunits to entire GIRK proteins (13) or to segments of channel subunits (14 -19). Although G ␤␥ subunits can interact directly with both N and C termini, interactions with the C terminus of the channel were shown to be the strongest (14,15). In addition, the N terminus also binds to G ␣ subunits alone (14) or to the G ␣␤␥ heterotrimer (14,18).
The ␤␥ subunits of G proteins activate not only native GIRK heteromultimers (1,6), but also recombinant hetero-or homomultimeric GIRK channels (7,20). There is no qualitative difference in the G ␤␥ sensitivity of P-region homomultimeric mutants versus heteromultimeric channels (12). In contrast, the inwardly rectifying K ϩ channel IRK1 (21) is G ␤␥ insensitive (22), despite its high degree of similarity to the five members of the GIRK subfamily.
We sought to identify those residues of GIRK critical for transducing effects of the G ␤␥ subunits. Our strategy was to generate chimeras between the GIRK4(S143T) (referred to as GIRK4*) and IRK1 channels, and screen for differences in G ␤␥ -dependent function and binding to G ␤␥ . Mutagenesis at a single site, namely GIRK4(L339E), reduced binding to G ␤␥ and impaired agonist-induced activity, but left intact the G ␤␥ dependence of the basal activity. Thus, we have identified a site on an effector protein that interacts specifically with G ␤␥ released through receptor stimulation.

EXPERIMENTAL PROCEDURES
Human homologs of GIRK1 and GIRK4 (GenBank TM accession numbers U39195 and U39196) (7) or their point mutated active counterparts (GIRK1-F137S or GIRK1* and GIRK4-S143T or GIRK4*), subcloned in the pGEMHE plasmid vector (23), were used as described previously (11,12). The chimeric cDNA constructs were produced by splicing by overlapping extension polymerase chain reaction (24). Polymerase chain reactions, using Vent DNA polymerase, were performed for only 15 cycles to avoid errors. Point mutations were generated using the Quickchange site-directed mutagenesis kit (Stratagene). The sequence of all constructs was confirmed by automated DNA sequencing (Sequencing facility, Cornell University, Ithaca, NY). The ␤ARK-PH construct (amino acids 452-689) was altered to incorporate the 15 N-terminal residues of Src for membrane targeting. This construct, generously provided by Dr. Eitan Reuveny, was altered and subcloned into pGEMHE.
All constructs were linearized with Nhel and cRNAs were transcribed in vitro using the "message machine" kit (Ambion). RNAs were electrophoresed on formaldehyde gels and concentrations were estimated from two dilutions using an RNA marker (Life Technologies, Inc.) as a standard.
Xenopus oocytes were surgically extracted, dissociated, defolliculated by collagenase treatment, and microinjected with 50 nl of a water solution containing the desired cRNA. Unless otherwise indicated, we used the following approximate quantities: GIRK channel subunits, 1.0 ng/species; IRK1 channel, 0.25 ng; m2 receptor, 1.0 ng; ␤2-adrenergic receptor, 2.0 ng; G ␣ and G ␤ subunits, 1.0 ng; G ␥ subunit, 1.0 ng; ␤ARK-PH, 1.0 ng. G ␤2 and G ␥2 were used in all G ␤␥ coexpression experiments, unless otherwise indicated. Oocytes were incubated for 3 days at 19°C. Whole oocyte currents were then measured by conventional two-microelectrode voltage clamp with a GeneClamp 500 amplifier (Axon Instruments). Agarose cushion microelectrodes were used with resistances between 0.1 and 1.0 megaohms (25). Oocytes were constantly superfused with a high potassium solution having (in mM): 91 KCl, 1 NaCl, 1MgCl 2 , 5 KOH/HEPES (pH 7.4). To block or activate currents, the oocyte chamber was perfused with solutions of the same composition with 3 mM BaCl 2 or 5 M ACh. Typically oocytes were held at 0 mV (E K ), and currents were constantly monitored by 500 ms pulses to a command potential of Ϫ80 mV for 200 ms followed by a step to ϩ80 mV for another 200 ms, and the cycle was repeated every 2 s. Periodically, a protocol was applied with a command potential from Ϫ100 to ϩ100 mV with 10 mV increments. Current amplitudes were measured at the end of the 200 ms pulse at each potential. In such manner, control currents were evaluated 2-5 min after impaling the oocytes just before application of ACh, ACh-activated currents were evaluated at the peak of the response to ACh, and Ba 2ϩ -insensitive currents were evaluated once steady-state inhibition was achieved, 1-3 min after application of 3 mM Ba 2ϩ . Basal current is the difference between control and Ba 2ϩinsensitive currents, and ACh-induced (or ACh-sensitive) current the difference between ACh-activated and control currents. Error bars in the figures represent mean Ϯ S.E. Each experiment shown or described was performed on 3-5 oocytes of the same batch. A minimum of 2-3 batches of oocytes were tested for each experiment shown.
Single-channel activity was recorded on devitellinized oocytes under the cell-attached mode of standard patch-clamp methods (26), as described previously (27). The pipette solution contained 96 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , and 5 mM HEPES, pH 7.4. The bath solution was composed of 96 mM KCl, 1 mM MgCl 2 , 5 mM EGTA, and 10 mM HEPES, pH 7.4. 100 M gadolinium was also included in the pipette solution. G ␤␥ purified from bovine brain was used in the inside-out experiments (gift from Dr. John D. Hildebrandt). Recordings were performed at a holding membrane potential of Ϫ80 mV. Recordings were performed using the EPC-9 patch-clamp amplifier and the PULSE/ PULSEFIT (v. 7.6) data acquisition software (Heka Electronik, Lambrecht, Germany). Data were stored on the hard disk of a PC compatible computer, and single channel analysis made use of TAC (v 2.6.1) software (Skalar Instruments, Inc., Seattle, WA). The sampling rate was 4 kHz for most recordings. Activity expressed as NPo (N, number of channels in the patch; Po, probability of opening) was calculated by integrating the current traces over 30 -60 s intervals and dividing by the unitary current.
Recombinant bovine G ␤1␥2 subunits were purified from Sf9 cells infected with baculoviruses encoding for ␤ 1 , ␥ 2 , and His6-␣i1 as described by Kozasa and Gilman (32). cDNAs encoding the C termini of GIRK4, GIRK4(L339E), or IRK1, and the PH domain of ␤ARK were generated by polymerase chain reaction and cloned in frame with the GST coding sequence in pGEX-4T-3 (Amersham Pharmacia Biotech). The resulting polymerase chain reaction fragments coded for: amino acids 184 -419 for GIRK4, GIRK4(L339E), amino acids 177-428 for IRK1; and amino acids 546 -670 for ␤ARK. Expression of fusion proteins was induced by 0.1 mM isopropyl-1-thio-b-D-galactopyranoside at 37°C for 2 h, and the fusion proteins were purified using glutathione 4B-Sepharose beads. The binding assay of G ␤␥ to the fusion proteins was performed as described by Huang et al. (14). Briefly, 1.0 M GST fusion protein and 0.1 M G ␤␥ were incubated in phosphate-buffered saline, 0.1% lubrol and glutathione-Sepharose beads at 4°C for 30 min. After washing three times with 200 volume of phosphate-buffered saline with 0.01% lubrol, the bound proteins were eluted from beads by heating in protein sample buffer at 70°C for 10 min and were then electrophoresed in a 12% SDS-polyacrylamide gel electrophoresis. GST fusion proteins were visualized by Coomassie staining, and G ␤1 was detected by immunoblotting using a G ␤ antibody (Santa Cruz Biotechnology) and visualized with ECL (Amersham). Densitometry was used to quantify the relative amounts of bound G ␤␥ .

RESULTS
GIRK4* Unlike IRK1 Is G ␤␥ -sensitive-We compared G ␤␥ sensitivities of basal and agonist-induced currents between the highly active GIRK4* and IRK1 channels. All experiments were carried out by expression in Xenopus oocytes. In wholecell experiments, G ␤␥ sensitivity, in the presence of coexpressed G protein-coupled receptor, was assessed by (a) K ϩ current responses to agonist stimulation, (b) coexpression of channels with proteins such as Gi␣ subunits or the PH domain of ␤ARK that can act as "sinks" for endogenous G ␤␥ subunits, and (c) coexpression of channels with exogenous G ␤␥ subunits.
Expression of GIRK4* in oocytes led to large basal and AChinduced currents (Fig. 1, A and C). Coexpression of either ␤ARK-PH or Gi␣1 led to a significant reduction in basal currents. However, oocytes coexpressing Gi␣1, rather than ␤ARK-PH, displayed ACh-induced currents. This result is consistent with the interpretation that Gi␣1, and not ␤ARK-PH, bound to endogenous G ␤␥ may be available for receptor-mediated activation. Coexpression of GIRK4 with exogenous G ␤␥ -enhanced agonist-independent K ϩ currents while preserving the AChinduced response (as in Ref. 12). These results indicate that both the GIRK4* basal and agonist-induced currents are largely mediated by the G ␤␥ subunits. In contrast, oocytes injected with IRK1 exhibited no ACh-induced currents and did not respond to coexpression with Gi␣1, ␤ARK-PH, or G ␤␥ (Fig.  1, B and C).
A Minimal Chimera between GIRK4* and IRK1 with a Defect in Agonist-induced Responses-We constructed chimeras between GIRK4* and IRK1 (Fig. 2, left). We screened for minimal segments of GIRK4* that when replaced by the corresponding IRK1 regions impaired sensitivity to G ␤␥ .
Chimeras were named for the IRK1 segment replacing the corresponding GIRK4 region. We first replaced the full C terminus of GIRK4* with that of IRK1 (GIRK4*(IRK V179-I428 )). This chimera showed intact basal but impaired agonist-induced currents, consistent with a previous report (22). Huang et al. (15) found the GIRK1(Glu 318 -Pro 462 ) segment to be a minimal G ␤␥ binding region. From an alignment of the GIRK1 and GIRK4 primary amino acid sequences, residue GIRK1(Glu 318 ) corresponds to GIRK4(Asp 324 ). Thus, we tested the response of the chimera GIRK4*(IRK L316-I428 ) that replaced the GIRK4(Met 323 -Val 419 ) region with the corresponding IRK1 segment. Again this chimera exhibited intact basal but impaired agonist-induced currents. To narrow the region responsible for the aberration of the GIRK4* agonist-induced currents, we constructed and tested three additional chimeras GIRK4*(IRK L316-Y341 ), GIRK4*(IRK S342-K365 ), and GIRK4*-(IRK Y366-I428 ). All three chimeras displayed intact basal currents. However, the response of GIRK4*(IRK L316-Y341 ) was impaired to agonist. These results suggest that differences between the two channels in this region, GIRK4(Met 323 -Tyr 348 ) and IRK1(Leu 316 -Tyr 341 ), may be important in their differential sensitivity to G ␤␥ . This is unlike the downstream regions where differences were without effects on G ␤␥ sensitivity.
An Agonist-insensitive Chimera with Intact G ␤␥ -mediated Basal Currents-The current resulting from expression of the chimera GIRK4*(IRK L316-Y341 ) had intact basal currents but impaired agonist-induced responses. To test its sensitivity to G ␤␥ , the GIRK4*(IRK L316-Y341 ) chimera was coexpressed with Gi␣1 or ␤ARK-PH. A significant reduction of basal currents was obtained, similar to the GIRK4* control (Fig. 3A). Yet, this chimera differed from GIRK4* (see Fig. 1) in that its expression alone or with Gi␣ or G ␤␥ resulted in impaired ACh-induced currents. This result further supports the conclusion that this chimeric channel is defective in producing agonist-induced currents. Coexpression with G ␤␥ did not stimulate basal levels of activity. Because the G ␤␥ dependence of the basal currents was intact in the GIRK4*(IRK L316-Y341 ) chimera, it is likely that other regions may be involved in G ␤␥ mediation of basal currents.
A Point Mutation Sufficient to Specifically Impair Agonistinduced Currents without Affecting the G ␤␥ Contribution to Basal Activity-We proceeded to test which of the distinct residues within the identified region of the GIRK4* and IRK1 channels were responsible for their differences in sensitivity to G ␤␥ . Eleven point mutations were made in which residues in the Met 323 -Tyr 348 region of GIRK4* were mutated to the corresponding residues found in the Leu 316 -Tyr 341 region of IRK1 (Fig. 3B). Mutant names refer to the position and amino acid of GIRK4 that was mutated to the corresponding IRK1 residue. Only GIRK4*(L339E) showed impaired agonist-induced responsiveness, mimicking the responses obtained with the GIRK4*(IRK L316-Y341 ) chimera.
We next tested the G ␤␥ sensitivity of the basal currents of GIRK4*(L339E), and compared them with that of the GIRK4* control in the same batch of oocytes.
Oocytes coexpressing GIRK4*(L339E), G ␤␥ , Gi␣1, or ␤ARK-PH behaved similar to the chimera GIRK4*(IRK L316-Y341 ), demonstrating an intact G ␤␥ -mediated basal current component. (Fig. 4A). Inside-out patch recordings from oocytes expressing the mutant and control channels were performed to test their responses to G ␤␥ subunits. Fig. 4B (left) compares activity from one batch of oocytes expressing GIRK4* and GIRK4*(L339E) channels. Perfusion of inside-out patches with purified G ␤␥ was ineffective in stimulating GIRK4*(L339E) activity compared with control GIRK4*. Stimulation of currents by endogenous G proteins through GTP␥S application gave similar results as the application of purified G ␤␥ (data not shown, n Ͼ 3). Regardless of their sensitivity to G ␤␥ , control or mutant channels responded to a similar degree to intracellular Na ϩ ions (27), thus providing a positive control for gating by Na ϩ ions. These inside-out patch responses were consistent with the whole-cell data for GIRK4*(L339E). Perhaps the lack of stimulation of whole-cell currents by G ␤␥ coexpressed with GIRK4*(L339E) reflects maximal basal currents for this mutant. To examine this possibility, we coexpressed ␤ARK-PH in oocytes (same batch as the experiments in Fig. 4B, left) with GIRK4* or GIRK4*(L339E) channels. We perfused inside-out patches from such oocytes with G ␤␥ purified from bovine brain (Fig. 4B, right). Inside-out patches of oocytes coexpressing GIRK4*(L339E) and ␤ARK-PH convincingly responded to perfusion with exogenous G ␤␥ , presumably recovering the ␤ARK-PH inhibition of the basal currents seen in the whole-cell experiments. However, these responses were significantly smaller than those of the control GIRK4*. In all cases, GTP␥S application failed to stimulate channel activity by activating endogenous G proteins, serving as a positive control for ␤ARK-PH effectiveness (data not shown, n Ͼ 3). Through these ex- periments we conclude that the GIRK4(L339E) mutation selectively impairs agonist-induced G ␤␥ -mediated responses.
The C Terminus of GIRK4(L339E) Shows Decreased Binding to the G ␤ Subunit-To determine the effects of the GIRK4-(L339E) mutation of the C terminus on G ␤␥ binding, we constructed and purified GST fusion proteins containing the C termini of GIRK4 (GIRK4C), GIRK4(L339E), (GIRK4-(L339E)C) and IRK1 (IRK1C), or ␤ARK-PH. GST fusion proteins were expressed in bacteria and purified (Fig. 5, top). In vitro binding assays were performed with the recombinant bovine G ␤1␥2 subunits purified from Sf9 cells.
As shown in Fig. 5 (middle and bottom panels), the C termini of GIRK4 and GIRK4(L339E) were able to bind G ␤␥ as com-pared with negative controls (GST and IRK1C) and a positive control (␤ARK-PH). GIRK4(L339E)C binding to G ␤␥ was significantly reduced. These results suggest that the critical Leu of GIRK channels, and perhaps neighboring residues, directly interacted with G ␤␥ subunits. Additionally, because GIRK4-(L339E)C has reduced but measurable binding to G ␤␥ , it is likely that additional C-terminal G ␤␥ binding sites exist, which contribute to the G ␤␥ dependence of basal currents.
Wild-type GIRK1/GIRK4 Subunits Contribute Equally to the Agonist-induced Activity of Heteromultimeric Channels-To determine whether the effect seen with the L339E mutant was specific to the GIRK4* subunit, we mutated the corresponding amino acid residue in GIRK1, L333E. We tested for G ␤␥ sensi-

FIG. 2. Chimeras between GIRK4* and IRK1 channels reveal a region important in agonist-induced stimulation of K ؉ currents.
Left, schematic of chimeric constructs between GIRK4* and IRK1. Specific segments of IRK1 (black regions) were used to replace corresponding segments of GIRK4* (white regions). Each chimera was named according to the IRK1 segment (identified by the first and last IRK1 amino acid of the segment), which replaced the corresponding segment of GIRK4*. Note that the corresponding numbers, defining the limits of the grafted segments, are shifted (6 to 7 numbers) as determined by the alignment of the two sequences. Right, basal and agonist-induced currents at Ϫ80 mV of GIRK4*, IRK1, and the five chimeras. Asterisks note significant differences in ACh-induced currents of the particular chimeras from the GIRK4* control (GIRK4*, Ϫ10.5 Ϯ 2.98 A, n ϭ 4; GIRK4*(IRK L316-Y341 ), Ϫ0.62 Ϯ 0.52 A, n ϭ 4; p Ͻ 0.05).
Similar results were obtained with the GIRK1*(L333E) mutant as with the GIRK4*(L339E). Again, although ACh-induced currents were impaired by the mutation, the basal currents were reduced by Gi␣1 and ␤ARK-PH (Fig. 6A, right). Fig. 6B shows that GTP␥S application to inside-out patches expressing GIRK1* or GIRK1*(L333E) resulted in ϳ42-fold increase in GIRK1* activity but caused no increase in the current of GIRK1*(L333E) (n ϭ 5).
We next sought to determine the relative contribution of wild-type GIRK subunits to agonist-induced activation in heteromultimeric channels (Fig. 6C). We introduced the Leu to Glu mutations at the 333 and 339 positions of the wild-type GIRK1 and GIRK4 subunits, respectively. We compared basal and agonist-induced currents of GIRK1/GIRK4 heteromultimeric channels, composed of both wild-type, both Leu to Glu mutants, and each wild-type to mutant combination. Our results suggest that each of the wild-type subunits contribute equally to agonist-induced activity, because heteromultimeric channels containing either Leu to Glu mutants displayed reduced agonist-induced sensitivity. Moreover, heteromultimeric channels, where both the subunits contained the Leu to Glu substitution showed significantly impaired agonist-induced currents. Thus, these results confirmed the importance of the residue for receptor-stimulated currents in heteromultimeric channels.
Mutation of the Critical Leu Residue Does Not Distinguish among Channel Interactions with Specific G ␤ Subunits or Signaling through Specific Receptors-Yeast two-hybrid experi-ments have shown that G ␤1 and G ␤2 interact with the N terminus of GIRK1 more strongly than do G ␤3-5 (19). To determine whether C-terminal mutation of the critical Leu residue could have altered the ability of the channel to interact with specific G ␤ subunits, we coexpressed G ␤1 , G ␤2 , or G ␤3 with G ␥2 and GIRK1/GIRK4 or GIRK1(L333E)/GIRK4(L339E) heteromultimers. All G ␤␥ combinations stimulated wild-type basal currents (2-4-fold, n ϭ 3). When tested with the mutated channel subunits, G ␤1-3␥2 subunits failed to stimulate basal currents (n ϭ 3). These results suggest that mutation of the critical Leu residue does not exert its effects by altering the specificity of channel/G ␤1-3 interactions. However, possible changes in the specificity of Leu mutant channels with the G ␤4 or G ␤5 subunits that were not tested cannot be ruled out. G ␤␥ subunits released from Gs ␣ subunits by ␤2-adrenergic receptor stimulation activate GIRK channels expressed in Xenopus oocytes (28). To test whether the critical Leu residue is involved in G ␤␥ signaling by receptors other than m2, we coexpressed ␤2-adrenergic receptor and Gs␣ subunits with GIRK1/ GIRK4 or GIRK1(L333E)/GIRK4(L339E) heteromultimers. Isoproterenol-induced currents were obtained after expression of wild-type heteromultimers (Ϫ7.65 Ϯ 2.17 A at Ϫ80 mV, n ϭ 3) but not with mutants (Ϫ0.17 Ϯ 0.09 A at Ϫ80 mV, n ϭ 3). These results suggest that the G ␤␥ released after activation of these two different receptors interact in a similar fashion with the critical GIRK Leu residue. DISCUSSION Since Soejima and Noma (29) first reported the membranedelimited nature of the atrial muscarinic K ϩ channel, the mechanism of G protein gating of ion channels has received FIG. 4. Comparison of G ␤␥ sensitivity of GIRK4* and GIRK4*(L339E) in whole-cell and inside-out patch experiments. A, two-electrode voltage clamp experiments plotting currents (at Ϫ80 mV) of GIRK4* and GIRK4*(L339E) channels coexpressed with G ␤␥ , ␤ARK-PH, or Gi␣1. ACh responses were impaired in the GIRK4*(L339E) groups, p Ͻ 0.005, n ϭ 3-6. The basal currents of GIRK4*(L339E) coexpressed with ␤ARK-PH or Gi␣1 were also significantly reduced, p Ͻ 0.005, n ϭ 3-6. B, (left) insideout patches from oocytes expressing the control or point mutant GIRK4* channels. Responses to patch perfusion with G ␤␥ or Na ϩ are shown for a representative patch and a number of patches tested within this batch of oocytes. G ␤␥ increased channel activity significantly compared with control for GIRK4* (p Ͻ 0.005, n ϭ 3) but not for GIRK4*(L339E) (n ϭ 4); (right) inside-out patches from oocytes coexpressing the control or point mutant GIRK4* channels and ␤ARK-PH. Responses to patch perfusion with G ␤␥ or Na ϩ are shown for a representative patch and a number of patches tested within this batch of oocytes. G ␤␥ increased channel activity significantly compared with control for GIRK4* and GIRK4*(L339E) (p Ͻ 0.005, n ϭ 4). The increase in channel activity in response to G ␤␥ was significantly less in GIRK4*(L339E) compared with GIRK4* (p Ͻ 0.005, n ϭ 3-4). great attention. Over a decade ago, G protein-gated inwardly rectifying K channels provided the first example of a G ␤␥ controlled signaling pathway. Yet, despite intense efforts, many questions remain unanswered regarding specific sites of interaction between the channel and G ␤␥ .
Biochemical studies from several groups have pointed to interaction of G ␤␥ with the C and N termini of these channels. Specifically, Huang et al. (14), using deletion mutagenesis, found that deletion of the GIRK1(Val 273 -Pro 354 ) segment reduced G ␤␥ binding of the remaining C-terminal fragment. In subsequent studies, Huang et al. (15) determined the GIRK1(Glu 318 -Pro 462 ) segment as a minimal G ␤␥ binding region. Kunkel and Peralta (16), using a combination of chimeras and deletion mutations, reported the GIRK1(Thr 290 -Tyr 356 ) region to be important in interactions with G ␤␥ .
In this study, we screened the C terminus of GIRK4* for residues that control channel activity. We made chimeras that replaced specific sections of GIRK4* with those from the G protein-insensitive channel, IRK1. Expression of a minimal chimera, the GIRK4*(IRK L316-Y341 ), resulted in normal basal currents that did not respond to ACh when coexpressed with hm2 receptors. Expression of exogenous G ␤␥ did not enhance basal currents in this chimera. Yet, basal currents were inhibited by coexpression of Gi␣1 and ␤ARK-PH.
Of all the amino acid differences between IRK1 and GIRK4* contained in this chimera, only the mutant GIRK4*(L339E) retained all properties of the chimera. This mutant displayed G ␤␥ -sensitive basal currents that were not ACh-sensitive and did not respond to exogenous G ␤␥ . Inside-out patch currents from oocytes expressing GIRK4*(L339E) and hm2 receptors were significantly smaller in response to G ␤␥ or GTP␥S. Binding of the G ␤␥ subunits to the GIRK4 C terminus bearing the L339E mutation was significantly reduced. Within the broader boundaries suggested by others (14 -16), the region surrounding Leu 339 is a G ␤␥ binding site with critical functional consequences.
Basal currents from the GIRK4*(L339E) channel were inhibited by coexpression of G ␤␥ sinks, such as the ␤ARK-PH. Because binding of the L339E mutant of the GIRK4 C terminus was not abolished and because basal currents of GIRK4*-(L339E) could be inhibited by ␤ARK-PH or Gi␣1, it is likely that additional G ␤␥ binding sites contribute to basal channel activity. Moreover, because G ␤␥ perfusion of inside-out patches activated GIRK4*(L339E) channels only when they were coexpressed with ␤ARK-PH, it is likely that this activation reflected reversal of the ␤ARK-PH inhibited basal currents. We hypothesize that the basal binding sites may be high affinity and saturated in both the whole-cell and inside-out patch experiments. Recent evidence has suggested another region of GIRK4 (Ser 209 -Arg 225 ) capable of high affinity binding to G ␤␥ (30). It is possible that such a site accounts for part or all of the basal channel activity.
How does GIRK4*(L339E) impair specifically agonist-induced stimulation? In the simplest model, free G ␤␥ would be bound to high affinity basal sites. The GIRK4*(IRK L316-Y341 ) chimera and the GIRK4*(L339E) mutant may impair a low affinity binding of this region to G ␤␥ subunits. Normally, agonist-induced liberation of G ␤␥ subunits would increase the local free G ␤␥ concentration, allowing interaction with a low affinity site, encompassing GIRK4*(Leu 339 ), and leading to stimulation of channel activity. Further work will be required to test this hypothesis. GIRK1*(L333E) channels displayed similar properties to GIRK4*(L339E). Again, basal currents from this mutant channel were sensitive to G ␤␥ , but no ACh-induced currents could be detected. Double mutations in heteromultimeric GIRK1(L333E)/ GIRK4(L339E) channels expressed in oocytes showed similar properties to the highly active homomultimeric mutants discussed above. Furthermore, mutation of both channels in a heteromultimer was required for the ACh-insensitive phenotype, whereas reduced agonist-induced currents were obtained with one or the other of the two subunits mutated. These results suggest that there is an equivalent contribution of GIRK1 and GIRK4 to G ␤␥ -mediated ACh-induced activity. Additionally, coexpression of different G ␤␥ combinations or different receptors such as the ␤2-adrenergic receptor did not alter the unique properties of these mutant channels. This suggests that signaling through different receptors and by different G ␤␥ combinations activates the channel through conserved interactions.
Biochemical evidence has suggested multiple binding sites in the C-and N-terminal segments of GIRK channels (15,19,30). The multiplicity of G ␤␥ binding sites with effector proteins is in agreement with the finding that the G ␤␥ -phosducin co-crystals show multiple sites of interaction between the two proteins (31). Our data combine biochemical with functional evidence FIG. 5. The Leu 339 mutation of the GIRK4 C terminus causes a significant reduction in G ␤␥ binding. GST fusion proteins were purified using glutathione 4B-Sepharose beads and were detected by Coomassie staining (top). Purified GST fusion protein were incubated with G ␤␥ and glutathione-Sepharose beads. Following wash, the bound proteins were released from the beads by heating in protein sample buffer and were separated by SDS-polyacrylamide gel electrophoresis. G ␤1 was detected by immunoblotting with anti-G ␤ antibody. The position of the ␤ subunit of G ␤␥ is indicated by the arrow. GST was used as negative control, and GST-␤ARK-PH as positive control (middle). Density scanning was used to quantify the relative amounts of bound G ␤␥ (bottom). The results shown represents mean Ϯ S.E. for four separate experiments. *, p Ͻ 0.01 compared with the GST-GIRK4C group.
for more than one G ␤␥ binding site on the channel. Surprisingly, distinct functional roles could be assigned to multiple binding sites; one designed to interact with G ␤␥ released from receptor stimulation, whereas additional site(s) may interact with free G ␤␥ to yield basal activity. Thus, these results suggest that for the K ϩ channel the multiplicity of interactions may subserve distinct functional roles.