Specificity of Gβγ Signaling to Kir3 Channels Depends on the Helical Domain of Pertussis Toxin-sensitive Gα Subunits*

Acetylcholine signaling through muscarinic type 2 receptors activates atrial G protein-gated inwardly rectifying K+ (Kir3) channels via the βγ subunits of G proteins (Gβγ). Different combinations of recombinant Gβγ subunits have been shown to activate Kir3 channels in a similar manner. In native systems, however, only Gβγ subunits associated with the pertussis toxin-sensitive Gαi/o subunits signal to K+ channels. Additionally, in vitro binding experiments supported the notion that the C terminus of Kir3 channels interacts preferentially with Gαi over Gαq. In this study we confirmed in two heterologous expression systems a preference of Gαi over Gαq in the activation of K+ currents. To identify determinants of Gβγ signaling specificity, we first exchanged domains of Gαi and Gαq subunits responsible for receptor coupling selectivity and swapped their receptor coupling partners. Our results established that the G proteins, regardless of the receptor type to which they coupled, conferred specificity to Kir3 activation. We next tested signaling through chimeras between the Gαi and Gαq subunits in which the N terminus, the helical, or the GTPase domains of the Gα subunits were exchanged. Our results revealed that the helical domain of Gαi (residues 63–175) in the background of Gαq could support Kir3 activation, whereas the reverse chimera could not. Moreover, the helical domain of the Gαi subunit conferred “Gαi-like” binding of the Kir3 C terminus to the Gαq subunits that contained it. These results implicate the helical domain of Gαi proteins as a critical determinant of Gβγ signaling specificity.

Heterotrimeric G proteins are versatile messengers that couple directly to a diverse group of effectors that includes potassium (K ϩ ) channels. Early studies in cardiac pacemaker cells found that acetylcholine (ACh) 4 stimulation of muscarinic receptors slowed the heart rate by activating specific K ϩ currents (I KACh ) (1)(2)(3). Subsequent studies demonstrated the involvement of G proteins and also showed that only PTXsensitive G␣ i/o subunits could couple muscarinic receptors to I KACh (4,5). Surprisingly, it was not the G␣ i/o but, rather, the G␤␥ subunits of the G i/o heterotrimers (i.e. G␣ i/o ␤␥) that were shown for the first time to serve as G protein effectors by directly activating the K ACh channel (6).
Studies of K ACh channel activation by G proteins were greatly facilitated by the cloning of five isoforms of the G protein-gated inwardly rectifying potassium (Kir3) channels 1-5 (7)(8)(9)(10)(11)(12). Cardiac Kir3 channels were shown to be heteromers of the Kir3.1 and Kir3.4 subunits (10). Expression of cardiac Kir3 channels in Xenopus oocytes gave large agonist-independent (or basal) K ϩ currents, and co-expression with muscarinic type 2 (M2) receptors resulted in ACh-induced K ϩ currents (e.g. Ref. 12). In Xenopus oocytes G␣ i -GDP applied either to the cytosolic side of inside-out patches or co-expressed with Kir3 channels inhibited K ϩ currents, which could still be activated after agonist stimulation (13,14). These results suggested that G␣ i -GDP sequestered G␤␥, thus preventing it from activating Kir3 channels. In contrast, application of the activated form of G␣ i (G␣ i -GTP␥S) did not activate these currents (13). It is presently widely accepted that upon receptor stimulation, G␤␥ that is liberated specifically from PTX-sensitive G␣ i/o activates Kir3 channels (13,15). Since the initial finding of Kir3 activation by G␤␥, additional G␤␥ effectors have been identified (16). Among them is the neuronal voltage-dependent calcium channel, which is also modulated by the G␤␥ subunits in a PTXsensitive manner (17)(18)(19).
Most G␤ isoforms G␤1-4, with the exception of G␤5, in various combinations with G␥ subunits activate Kir3 channels in a similar manner (21,22). It is therefore unlikely that the specificity of signaling seen with PTX-sensitive G proteins depends exclusively on the identity of specific G␤ and G␥ isoforms that associate selectively with G␣ i/o . Thus, the factors controlling the ability of the G␤␥ associated with PTX-sensitive G␣ subunits in distinguishing between effectors remain unknown.
In addressing this question, we established that the specific receptors involved in the signaling did not predetermine the ability of G␤␥ to signal to Kir3, in agreement with a previous study by Leaney and colleagues (23). Examination of a role for G␣ in conferring specificity to G␤␥ signaling revealed the helical domain of G␣ i as a key determinant.

EXPERIMENTAL PROCEDURES
Molecular Biology-All G␣ subunit chimeras were made using the splice by overlap extension method (24) from wildtype G␣ i1 and G␣ q . DNA primers for generating G␣ subunit chimeras and point mutants were obtained from Invitrogen. The chimeras were constructed as follows: G␣ q -(1-354) i contained residues 1-354 of G␣ q and 350 -354 of G␣ i ; G␣ i -(1-334) q contained residues 1-334 of G␣ i and 340 -359 of G␣ q ; G␣ i -(1-62) q contained residues 1-62 of G␣ i and 69 -359 of G␣ q ; G␣ i -(1-175) q contained residues 1-175 of G␣ i and 181-359 of G␣ q ; G␣ qiq contained residues 63-175 of G␣ i and 1-68 and 181-359 of G␣ q ; G␣ iqi contained residues 1-62 and 176 -354 of G␣ i and 69 -180 of G␣ q . PCR reactions were carried out using Pfu polymerase (Stratagene, La Jolla, CA) for 25 cycles. Amplified PCR products and restriction digestion fragments were gel-purified using the QIAQuick Gel Extraction kit (Qiagen, Valencia, CA). Point mutations were generated using the QuikChange mutagenesis kit (Stratagene). EE-tagged G␣ q subunit was purchased from UMR cDNA Resource Center. EEtagged G␣ qiq was made by ligating the EE tag into the G␣ qiq construct. Human muscarinic type 1 (M1) and type 2 (M2) receptors, the Kir3.4* channel subunit (20), and G␣ subunits were subcloned into the pGEMHE vector, optimized for Xenopus oocyte expression (25), or into the pcDNA3.1 (ϩ/Ϫ) vector (Invitrogen) for transfection into the mammalian HEK 293 (HEK) cells. The pertussis toxin subunit S1 subcloned into pGEMHE was provided by Eitan Reuveny (Weizman Institute of Science, Rehovot, Israel). All mutants were confirmed by sequencing. We discovered at late stages of our work an unexpected point mutation, D108N, in the G␣ i region of the G␣ qiq chimera. We tested the corresponding mutation D102N in the G␣ i (C351A) construct and found it to be innocuous (data not shown). DNA constructs subcloned into pGEMHE were linearized by enzymatic digestion with NheI or SphI (New England Biolabs, Beverly, MA) and in vitro transcribed into cRNA using the mMessage mMachine (Ambion. Austin, TX). cRNA concentration was estimated by subjecting the cRNA sample to formaldehyde gel electrophoresis and comparing the band intensity to an RNA molecular weight standard (Invitrogen).
Electrophysiology in Xenopus Oocytes-Oocytes obtained as previously described (26,27) were injected with 2 ng cRNA each of Kir3.4*, M1 or M2, and G␣ subunit cRNA unless otherwise noted. Electrophysiological recordings were performed 48 h after injection. The two-electrode voltage clamp technique was used to record whole-cell currents from oocytes, as previously described (14,28,29). Briefly, a GeneClamp 500 amplifier (Axon Instruments, Foster City, CA) was used, and data acquisition was carried out with Clampex pClamp6 software. The bath solution (high potassium (HK)) contained 91 mM KCl, 1 mM NaCl, 1 mM MgCl 2 , 5 mM KOH/HEPES (pH 7.4). Oocytes were perfused with HK containing 5 M ACh (Sigma) to activate or HK containing 3 mM BaCl 2 to block inwardly rectifying potassium currents. Currents were monitored in a series of 700-ms sweeps, during which the voltage was stepped from a 0-mV holding potential for 250 ms each to ϩ 80 mV and Ϫ80 mV, repeated every 1 s. Before application of each solution, currents were allowed to reach steady state. Currents were sampled at the end of each 250-ms pulse and were plotted in Clampfit (pClamp6.0, Axon Instruments) as the time course of current amplitudes. The HK current (I HK ), sampled at the end of the Ϫ80-mV pulse, was evaluated after reaching steady state, whereas the ACh-induced current (I ACh ) occurring in the presence of M2 receptor was evaluated at the peak amplitude. I ACh inwardly rectifying currents in the presence of M1 receptor were evaluated after the peak of I CaCl transient (e.g. Fig. 1, B and C, single-headed arrow) to minimize the contribution of its inward component to the rest of the inward current. The time course of I CaCl was assessed by its outwardly rectifying component recorded at ϩ80 mV. Barium-insensitive current (I Ba ) was sampled once inhibition reached steady state. Basal current was calculated as I HK -I Ba , and ACh-induced current was calculated as I ACh -I HK . Data were analyzed in Microcal Origin 6.0 software (Microcal Software, Inc., Northampton, MA). For each injection group, 2-3 batches were tested with the exception of Kir3.4*, M2, and G␣ i with or without PTX (Fig. 2E) and Kir3.4*, M2, and G␣ q -(1-354) i for which only one batch of oocytes was tested (3-6 oocytes per batch). Basal and agonist-induced currents from each oocyte within each batch were normalized to the average of the basal current in oocytes expressing Kir3.4* and M2 of the same batch. Means of normalized basal and agonist-induced currents from each group are presented in bar graphs Ϯ S.E.
Statistical Analysis-The statistical significance of mean normalized currents in groups of oocytes injected with different combination of cRNA was determined using one way analysis of variance (ANOVA) and Holm-Sidak all pairwise multiple comparison as a post hoc test with an overall significance level of 0.05 (SigmaStat, Systat Software, Inc. Richmond). Twotailed independent t test was used to compare basal and total (total ϭ basal ϩ agonist-induced) currents within a group (Microcal Origin 6.0) to determine significance of agonist-induced currents.
Electrophysiology in Adenosine 1 (A1), Kir3.1/Kir3.2 Stable HEK Lines-The HEK 293 cell line stably expressing A1 receptor and Kir3.1 and Kir3.2 channel subunits was a generous gift from Andrew Tinker (University College, London, UK). Maintenance of the cell line and whole-cell patch clamp recordings were performed as previously described (23,30). Briefly, the A1 Kir3.1/2 cell line was maintained in Cellgro Dulbecco's modified Eagle's medium (with 4.5 g/liter glucose with L-glutamate without sodium pyruvate) (Mediatech, Herndon, VA) supplemented with 727 g/ml Geneticin (Invitrogen) and 364 g/ml Zeocin (Invitrogen). Transfections were carried out using the Effectine Transfection kit (Qiagen) with G␣ q , G␣ qiq , or G␣ iqi C351A and/or M1 receptor subcloned into pCDNA3.1(ϩ/Ϫ) vector (Invitrogen). Recordings were performed 48 -72 h after transfection. Co-transfection of pEGFP-N1 (Clontech, Palo Alto, CA) was used to select for efficiently transfected cells. Cells were excited at 488 nm and visualized on a Nikon Diaphot 300 inverted epifluorescence microscope. Patch clamp recordings were performed only on fluorescent cells. An Axopatch 200A amplifier (Axon Instruments), an ITC-16 Computer Interface (Instrutech, Great Neck, NY), and HEKA software (HEKA Electronik, Germany) were used for data acquisition. Currents were filtered at 1 kHz and sampled at 20 Hz. Voltage was held at Ϫ80 mV. Pipette resistances ranged between 6 and 10 megaohms. Cell capacitance was determined using the whole-cell capacitance compensation of the amplifier. The patch pipette solution contained 107 mM KCl, 1. Control currents during HK perfusion (I HK ) were allowed to reach steady state before agonist application. Agonist-induced currents (I ACh or I NECA ) were evaluated at the peak-activated current amplitude. Data were analyzed with Microcal Origin 6.0 software. Basal currents were calculated as I HK -I LK and agonist-induced currents as I ACh/NECA -I HK , as described above. Current densities were obtained by dividing current by cell capacitance. The statistical analysis of current density means was the same as described above for oocytes.
Relative G␣ subunit expression was determined as described in Medina et al. (31). Anti-EE antibodies were generously provided by Dr. Catherine Berlot (Weis Center for Research, Geisinger Clinic, Danville, PA).
Protein Purification: G Protein Subunits-G␣ i , G␣ q , G␣ qiq , and G␣ iqi subunits were purified as described by Kozasa et al. (32) with minor modifications as follows. G␣ subunits were subcloned into the baculovirus transfer vector pVL1392, and a recombinant virus was generated using the BaculoGold baculovirus expression system (BD Biosciences). Sf9 cells were maintained in a serum-free medium (SF900 II SFM) (Invitrogen). G␣ subunits were co-transfected with G␤1 and Histagged G␥2 subunits into Sf9 cells. 10 M GDP was used in all buffers for all G␣ subunits with the exception of the G␣ qiq chimera (50 M GDP). Cells were lysed by sonication of 50-ml aliquots using 4 cycles of pulses (six 10-s pulses with 10-s intervals in between at 60% power).
Mono Q fractions containing G␣ were concentrated using the Amicon Centricon YM-30 (Millipore, Bedford, MA) concentrators and exchanged into 100 mM NaCl buffer S (20 mM HEPES-NaOH (pH 8), 500 M EDTA, 2 mM MgCl 2 , 1 mM DTT, 0.7% CHAPS) (32). Protein concentration was determined by the Bradford assay using Coomassie Plus-the better Bradford TM assay kit (Pierce). Purified G␣ subunits were separated by SDS-PAGE electrophoresis followed by transfer onto a nitrocellulose membrane at 100 mV for 1 h. The membrane was probed either with G␣ i1 (internal epitope), G␣ q (N-terminal epitope), G␣ q / 11 (C-terminal epitope) (Santa Cruz Biotechnol-ogies, Santa Cruz, CA), or G␣ i1/2 (C-terminal epitope) rabbit IgG primary antibodies (Upstate Signaling Solutions, Temecula, CA), depending on the protein being purified, followed by incubation with horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (Santa Cruz Biotechnologies). Western blots were subjected to the enhanced chemiluminescence detection (ECL Plus TM Western blotting detection system; GE Healthcare, Piscataway, NJ). Hyperfilm ECL (GE Healthcare) was exposed to the Western blot chemiluminescence signal. G␣ identity and purity was assessed from the Western blot and SDS-PAGE gel electrophoresis. The protein was aliquoted, frozen in liquid nitrogen, and stored at Ϫ80°C.
Trypsin Protection Assay-The trypsin protection assay was carried out as previously described (33) with minor modifications. Briefly, 200 nM G␣ was incubated for 1 h either with 100 M GDP at 4°C or 100 M GTP␥S and 10 mM MgSO 4 at 30°C. Next, 40 ng of L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (T-1426, Sigma-Aldrich) was added to the reaction mixture at the final concentration of ϳ1 ng/l, and the reaction mixture (total volume 38 l) was incubated for another 30 min at 30°C. The reaction was terminated by the addition of Laemmli gel loading buffer and by heating to 100°C for 10 min. The samples were separated on SDS-PAGE and stained with Coomassie Blue or subjected to Western blotting as described above.
GST Pulldown Assay-GST pulldown assays were carried out as described by Huang et al. (36) with minor modifications. Briefly, G␣ subunits (1 g) were incubated on ice with 40 M GDP for 30 min. GST, GST-Kir3.4-(184 -419), or GST-Kir3.4-(1-92) (100 nM) and phosphate-buffered saline containing 2 mM EDTA and 5 mM ␤-mercaptoethanol were added to 50 l final volume, and the mixture was incubated for 30 min at room temperature. Next 20 l of 50% slurry glutathione-Sepharose-4B (GE Healthcare) in phosphate-buffered saline with 2 mM EDTA and 5 mM ␤-mercaptoethanol were added, and the reaction was incubated on a rotator for 30 min at room temperature. The protein complex bound to the Sepharose was collected by centrifugation at 3000 rpm for 1 min and washed 3 times in 400 l of phosphate-buffered saline containing 2 mM EDTA, 5 mM ␤-mercaptoethanol, and 0.1% CHAPS. Proteins were eluted from Sepharose with 20 l of Laemmli gel loading buffer and heated to 65°C for 10 min. Samples were subjected to SDS-PAGE electrophoresis and Western blotting as described above. Film exposure times were adjusted to achieve the strongest signal with minimal saturation. A control amount of each G␣ (25 ng) was loaded on the gel in parallel with the GST pulldown samples. After probing for G␣ with the appropriate anti-G␣ antibody, nitrocellulose membranes were stripped in 62.5 mM Tris, 2% SDS, 100 mM ␤-mercaptoethanol at 55°C for 30 min and probed with anti-GST rabbit IgG (Santa Cruz Biotechnologies).
Western Blot Densitometry and Analysis-The images were acquired in Adobe Photoshop. The background-corrected signal intensity values were obtained using Kodak Digital Science 1 software (Eastman Kodak Co.). To compare the relative amounts of G␣ pulled down by GST, GST-Kir3.4C, and GST-Kir3.4N, we utilized both G␣ and GST signals from the same Western blot. Differences in the relative amounts of the G␣ pulled down by GST or GST-Kir3.4C and GST-Kir3.4N reflect both specific binding and several nonspecific components. We controlled for these contributions as follows. (a) To account for differences in the binding of GST and GST-tagged fragments to glutathione-Sepharose, we divided the pulled-down G␣ signal intensity by the GST signal intensity in the same lane (G␣/GST- We accounted for differences in G␣ antibody affinities for different G␣ subunits by loading equal amounts of each purified G␣ subunit in parallel with the GST-pulled down G␣ samples and normalizing their signal intensities to that of G␣ q (anti-G␣ q /anti-G␣ other ϭ B). (c) Since we observed nonspecific binding of G␣ subunits to GST alone, we calculated the product of A for GST ϫ B ϭ C. (d) To obtain specific binding of G␣ to either GST-Kir3.4C or GST-Kir3.4N, we calculated the G␣ signal intensity corrected for the amount of GST-tagged protein in the sample A and then normalized it to the antibody affinity A ϫ B and, finally, subtracted the contribution due to nonspecific binding to GST C (D ϭ A ϫ B Ϫ C). The means obtained from five experiments were plotted Ϯ S.E.
Statistical Analysis-One-way ANOVA and Holm-Sidak post hoc multiple comparison tests were carried out on Western blot densitometry values to compare binding of G␣ subunits to either GST-Kir3.4C or GST-Kir3.4N. Student's t test was used to compare binding to GST-Kir3.4C and GST-Kir3.4N of each G␣ subunit.

RESULTS
Kir3 channel stimulation through the PTX-sensitive family of G proteins, G i/o , is well established in native (4) as well as in heterologous expression (15) systems. Fig. 1A shows a representative trace, and Fig. 1E shows summary bar graphs of basal and agonist-induced currents obtained from Xenopus oocytes expressing Kir3.4* and M2 receptor. The recordings were obtained using a two-electrode voltage clamp in HK solutions at membrane potentials of ϩ80 mV (above the zero dashed line) and Ϫ80mV (below the zero dashed line). Double-headed arrows at the beginning and end of the trace indicate the amplitudes of basal and agonist-induced currents. Kir3.4* channel expression yielded large basal inwardly rectifying potassium (K ϩ ) currents that were inhibited by application of barium (Fig.  1, A and E). Basal currents are agonist-independent but are G␤␥-mediated, as they are inhibited by scavengers of G␤␥ (14). The M2 receptor, which couples to G i/o , can activate homomeric Kir3.4* currents upon stimulation with ACh (14,15). Upon application of ACh, Kir3 currents were further potentiated (Fig. 1, A and E) by G␤␥ subunits associated with the endogenous G␣ i/o proteins (14,15). Xenopus oocytes constitute a convenient expression system where G q/11 signaling can be independently assessed by monitoring endogenous calcium-activated chloride currents (I CaCl ) in parallel with effects on exogenously expressed Kir3 currents. Phospholipase C, a downstream effector of the G q/11 family of G proteins, hydrolyzes phosphatidylinositol 4,5-bisphosphate, generating diacylglycerol and inositol 1,4,5-trisphosphate. Inositol 1,4,5-trisphosphate in turn liberates Ca 2ϩ from internal stores, leading to activation of the transient, outwardly rectifying I CaCl (37). Oocytes expressing M1 and Kir3.4* (Fig. 1, C and E) displayed large basal current, but in contrast to oocytes expressing M2, application of ACh did not potentiate Kir3 currents. Instead, agonist application often led to Kir3 current inhibition and potentiation of I CaCl , as previously described (38).
Receptor Swap-Because the G proteins are known to selectively couple to specific receptors (39), the ability of the G␤␥ subunits to specifically activate Kir3 currents during agonistinduced signaling through the PTX-sensitive pathway could be receptor-dependent. Leaney et al. (23) used a chimeric G␣ approach in the HEK cell system and showed that they could couple G i to a G s -coupled receptor and vice versa without affecting G i or G s effector specificity, suggesting that G i -dependent Kir3 activation was receptor-independent.
To investigate receptor dependence of G␤␥ signaling specificity in the Xenopus oocyte system, we carried out a similar receptor swap between G i and G q . The distal C terminus of G␣ subunits plays a crucial role in selective receptor coupling (40,41). Thus, to couple G␣ q to M2 receptors, we constructed a chimera, G␣ q -(1-354) i , where the last five C-terminal amino acids, 355-359, of G␣ q were replaced with the corresponding amino acids of G␣ i (for sequence alignment, see Fig. 2B). To couple G␣ i to M1, we constructed G␣ i -(1-334) q by replacing 20 residues of the G␣ i C terminus (residues 334 -354) with the corresponding G␣ q residues. We co-expressed Kir3.4* and the M2 or M1 receptor with either the G␣ q -(1-354) i or G␣ i -(1-334) q , respectively, and monitored the effects of the chimeras on basal and agonist-induced Kir3 currents. In HK solution, oocytes expressing G␣ q -(1-354) i reduced basal currents (Fig. 1, B and E), suggesting that this G␣ chimera could bind and sequester G␤␥ away from the channel, as previously shown for overexpressed G␣ subunits (14,15). Application of ACh activated I CaCl but failed to activate Kir3 currents (Fig. 1, B and E), indicative of G q signaling but in this case through M2 receptor stimulation. In the presence of the G␣ i -(1-334) q chimera, basal Kir3.4* currents were also reduced compared with control ( Fig.  1, D and E). This result suggested that this chimera could also bind and sequester G␤␥ away from the channel. Application of ACh resulted in activation of small but significant inwardly rectifying Kir3 currents (Fig. 1, D and E). The reason for the reduced ACh-induced currents (e.g. compare M2-mediated response in Fig. 1A) is not clear. It is possible that the 20 amino acids of G␣ q at the C terminus of the G␣ i -(1-334) q chimera were not optimal in coupling M1 to G i as efficiently as M2 couples to G i . Potentially, a more efficient swap might require additional regions of G␣ q . Yet, the response was significant enough to illustrate that G i can signal to Kir3 even when coupled to M1, whereas G q coupled M2 signaled to Kir3 significantly less well than G i . Thus, these experiments suggest that in Xenopus oocytes, G␤␥-specific signaling to Kir3 channels is not exclusively dependent on the identity of the receptor through which signaling is initiated. Because various combinations of G␤ and G␥ subunits could activate Kir3 currents (21,22), it is also unlikely that G␤␥ signaling specificity to Kir3 is dictated by the specific G␤ and G␥ isoforms present. Instead, these results suggested that signaling specificity could be dependent on the specific G␣ subunit with which G␤␥ is associated.
The Helical Domain of G␣ i Confers Specificity to G␤␥ Signaling-When we co-expressed G␣ i with Kir3.4* and M2, we observed a reduction in basal currents (Fig. 2C), suggesting that exogenous G␣ i sequestered G␤␥ away from the channel. ACh perfusion resulted in robust agonist-induced Kir3 currents. Co-expression of G␣ q with Kir3.4* and M1 reduced Kir3 basal currents (Fig. 2C) to a greater extent than G␣ i . Agonist application induced significantly smaller currents than those induced through overexpressed G␣ i . The relative differences in ACh-induced currents between overexpressed G␣ i versus G␣ q subunits seemed greatest at the lowest levels of overexpressed G␣ subunits (data not shown). Because the G␣ subunit was implicated in determining G␤␥ signaling specificity, we asked whether a particular region of G␣ i was responsible for this effect. Guided by the three-dimensional structure of G␣ i -GDP complexed with G␤ 1 ␥ 2 (42) (Fig. 2A), we constructed G␣ chimeras between G␣ i and G␣ q (Fig. 2, B and C). The G␣ i crystal structure reveals three major domains, the N terminus, the helical and GTPase domains (43). Thus, we proceeded to systematically replace the major domains of G␣ q with the corresponding regions of G␣ i and monitor the effect of the chimeras on the agonist-induced current of Kir3. Fig. 2C shows a schematic representation of the chimeras with the corresponding summaries of normalized basal and agonist-induced currents in Xenopus oocytes. G␣ subunit chimeras containing the C terminus of G␣ q were coexpressed with the M1 receptor. When the N terminus and the helical domain of G␣ q were replaced with those of G␣ i , in the G␣ i -(1-175) q chimera, basal currents were reduced by ϳ60% compared with the Kir3.4*/M2 (control) (Fig.  2C), indicating that this chimera could bind and sequester G␤␥ from Kir3. Application of ACh resulted in significantly increased agonist-induced currents compared with G␣ q -expressing oocytes (Fig. 2C), suggesting that the first half of G␣ i contained sufficient determinants to support signaling to the channel. In oocytes co-expressing G␣ i -(1-62) q (G␣ q with the N terminus of G␣ i ) basal currents were also reduced compared with control oocytes. In contrast to G␣ i -(1-175) q , oocytes coexpressing G␣ i -(1-62) q failed to yield significant agonist-induced currents (Fig. 2C). The difference between G␣ i -(1-62) q and G␣ i -(1-175) q is in the region between residues 63 and 175 of G␣ i , which constitutes its helical domain ( Fig. 2A). Thus, we generated the chimera G␣ qiq , containing the helical domain of G␣ i in the background of G␣ q . Co-expression of G␣ qiq with Kir3.4* and M1 reduced basal currents significantly, suggesting that it was able to sequester endogenous G␤␥ from Kir3. Application of agonist in the presence of G␣ qiq resulted in significant Kir3 current potentiation (Fig. 2, C and D, left panel). These results show that the G␣ i region between residues 63 and 175 supports agonist-induced Kir3 activation, suggesting that the G␣ i helical domain is involved in conferring specificity of G␤␥ signaling to Kir3 channels. Although signaling through G␣ qiq was significantly larger than G␣ q , it was also significantly smaller than G␣ i . We suspect that although the helical domain of G␣ i is critical, its context also contributes to the efficiency with which it controls G␤␥ signaling specificity. If indeed specific G␤␥ signaling can be achieved by incorporating the helical domain of G␣ i into G␣ q , could it be removed by replacing the same domain in G␣ i by the corresponding G␣ q domain? To answer this question we constructed the reverse chimera, G␣ iqi , replacing the region of G␣ i between residues 63-175 with the homologous region from G␣ q . Because the G␣ iqi chimera should be activated by M2, we needed to distinguish currents mediated by this chimera from those mediated by the endogenous PTX-sensitive G␣ subunits activated by M2. Expression of the PTX-insensitive point mutant G␣ i (C351A) simultaneously with PTX allows one to abolish M2-coupled signaling through endogenous G proteins while allowing signaling through the over-expressed point mutant (23, 44 -46). PTX abolished agonist-induced Kir3 currents mediated by the wild-type but not by the C351A mutant of G␣ i (Fig. 2E). In the presence of the G␣ iqi C351A Kir3 basal currents were reduced, suggesting that this chimera was able to bind and sequester G␤␥ from the channel. Application of agonist to oocytes expressing G␣ iqi C351A and PTX resulted in reduced agonist-induced currents that were not significantly different from those in the presence of FIGURE 2. G␣ chimeras are used to determine the region of G␣ i that confers specificity onto G␤␥ signaling to GIRK. A, a representation of the x-ray crystal structure of G␣ i ␤␥ heterotrimer (42). Each subunit and three regions of the G␣ subunit (N terminus residues 1-62 are in white, GTPase domain residues 176 -354 are in light gray, and helical domain residues 63-175 are in dark gray) are labeled. B, alignment of G␣ i1 and G␣ q protein sequences was carried out using Clustal (68). Dashes indicate identical residues in G␣ q , dots indicate lack of a corresponding residue. The bar above the alignment corresponds to the code of G␣ subunit domains in panel A (N terminus white, GTPase domain light gray, helical domain dark gray). C, effect of expression of wild-type and chimeric G␣ subunits on K ϩ currents in oocytes co-expressing Kir3.4* and the appropriate receptor. All G␣ subunits with the exception of G␣ i were co-expressed with M1 receptor. Basal and ACh-induced currents from oocytes expressing Kir3.4* and either receptor are included as control. The bar graph shows mean normalized basal and agonist-induced currents Ϯ S.E. of oocytes expressing: Kir3.4*, M2 (n ϭ 44); Kir3.4*, M1 (n ϭ 45); ϩG␣ q (n ϭ 10); ϩG␣ i -(1-175) q (n ϭ 11); ϩG␣ i -(1-62) q (n ϭ 10); ϩG␣ qiq (n ϭ 15). One way ANOVA test followed by a Holm-Sidak pairwise multiple comparison were carried out for groups co-expressing M1 receptor and G␣ subunits. *, p Ͻ 0.05 ACh-induced currents were significantly greater; ACh-induced currents in the ϩG␣ q and ϩG␣ i -(1-62) q groups were not significantly different. D, traces from two-electrode voltage-clamp recordings in oocytes expressing Kir3.4*, M1, and G␣ qiq (left) and Kir3.4*, M2, PTX, and G␣ iqi C351A (right). E, bar graph summarizing mean normalized basal and ACh-induced currents in oocytes injected with: Kir3.4*, M2, ϩG␣ i (n ϭ 6); ϩG␣ i , PTX (n ϭ 5); ϩG␣ i C351A (n ϭ 17); G␣ i C351A, PTX (n ϭ 16), ϩG␣ iqi C351A (n ϭ 10); ϩG␣ iqi C351A, PTX (n ϭ 9). One-way ANOVA and Holm-Sidak multiple comparison were used to analyze Kir3 currents; *, p Ͻ 0.05, ACh-induced currents were significantly greater than those elicited in the presence of Kir3.4*, M2, G␣ i , and PTX. Additionally, ACh-induced currents from oocytes co-expressing Kir3.4*, M2, G␣ iqi C351A, and PTX were not significantly different from those in the presence of Kir3.4*, M2 G␣ i , and PTX. All currents were normalized to basal currents in oocytes of the same batch expressing Kir3.4*, M2 (not shown).
wild-type G␣ i plus PTX. In fact, even in the absence of PTX, when the endogenous G i/o pathway should be stimulated, oocytes overexpressing this chimera did not show significant agonist-induced currents. The reduction of agonist-induced currents through the endogenous G i/o proteins might be due to a dominant negative effect of G␣ iqi C351A, where by associating with G␤␥ it could have sequestered G␤␥ from the channel as well as from the endogenous G i/o , preventing signaling to Kir3. These results suggest that the G␣ q helical domain in the background of G␣ i does not support G␤␥ signaling to Kir3 channels.

Dependence of Specificity of G␤␥ Signaling on the Helical Domain of G␣ i in a Mammalian
Cell System-We examined whether our findings in the Xenopus oocytes could be reproduced in the HEK cell mammalian expression system. HEK cells stably expressing the Kir3.1/Kir3.2 channel subunits and the G i/o -coupled A1 receptor (23) were transiently transfected with various G␣ subunits and the M1 receptor. Fig. 3A shows representative whole-cell traces (top) along with the corresponding mean basal and agonist-induced current densities in the bar graph (bottom) from HEK cells. Replacement of the LK solution with HK increased basal inward Kir3 currents (Fig.  3Ai). Subsequent application of the A1 receptor agonist, NECA, potentiated Kir3 currents through endogenous G i/o proteins (Fig. 3Ai). Accordingly, NECA-activated Kir3 currents were completely abolished when the cells were treated with PTX (Fig. 3Aii). When we transfected stable A1 Kir3.1/2 cells with M1 receptor and treated them with PTX, application of ACh yielded variable Kir3 activation that was not significant over basal currents (Fig. 3Aiii). These results confirm our observations in Xenopus oocytes that in heterologous expression systems G␤␥ subunits signal to Kir3 more efficiently through PTXsensitive than through PTX-insensitive pathways.
We proceeded to test the effect of the G␣ helical domain chimeras in HEK cells. For M1-coupled G␣ subunits (G␣ q and G␣ qiq ), we utilized PTX to inhibit possible promiscuous signaling through G␣ i . For A1-coupled G␣ iqi subunits we utilized the PTX-insensitive chimera (G␣ iqi C351A) and PTX to inhibit endogenous PTX-sensitive signaling induced by NECA. Fig. 3B shows representative traces and summary data of cells transiently transfected with G␣ iqi C351A coupled to the stably expressing A1 receptor (Fig. 3Bi) and those transfected with M1 receptor and G␣ q (Fig. 3Bii) or G␣ qiq (Fig. 3Biii). Application of NECA did not give rise to significant Kir3 currents in cells expressing the PTX-insensitive G␣ iqi C351A chimera (Fig. 3Bi) compared with the PTX-treated control cells (Fig. 3Aii). In contrast, ACh application induced PTX-insensitive Kir3 currents in cells transiently expressing G␣ qiq (Fig. 3Biii), which were significantly larger than agonist-induced currents in G␣ q -expressing cells (Fig.  3Bii).
To verify that the expression levels of G␣ q and G␣ qiq were comparable, we performed Western blot analysis on the cell lysates. We compared expression in A1 Kir3.1/2 stable HEK cell lines transfected with G␣ q and G␣ qiq . We also utilized EE-tagged (31,47) versions of each construct, which ensured that epitope variability in the non-tagged constructs was not a factor influencing the results. The results indicated that it was not enhanced expression levels of G␣ qiq that were responsible for the differences seen in agonist-induced Kir3 currents between G␣ q -and G␣ qiq -expressing cells (data not shown). Our results in HEK cells corroborate those in Xenopus oocytes, suggesting that the helical domain of G␣ i is a critical region that allows specific G␤␥ signaling to Kir3.
Assessment of Chimeric G␣ Subunit Binding to G␤␥ and Nucleotide-induced Active Conformation-The inability of G␣ iqi C351A to support agonist-induced Kir3 currents in PTXtreated oocytes or mammalian cells could have resulted from a loss of function of this chimera. The ability of the chimera to reduce basal Kir3.4* currents suggests this not to be the case. Using biochemical assays, we further examined whether the wild-type and chimeric G␣ subunits could interact appropriately with G␤␥ and whether they could assume active confor- . In PTX-treated cells transiently transfected with M1 receptor basal and total currents were not significantly different. Corresponding sample traces of whole-cell currents recorded at Ϫ80 mV membrane potential. B, bar graph summarizing mean basal and agonist-induced current densities Ϯ S.E. from whole-cell recordings. The agonist-induced current densities in A1 Kir3.1/2 cells treated with PTX are listed along with the cDNAs they were transfected: (i) G␣ iqi (4.8 Ϯ 3.5 pA/pF, n ϭ 11) (ii), M1 and G␣ q (27 Ϯ 11 pA/pF, n ϭ 18) (iii), and G␣ qiq (69 Ϯ 15 pA/pF, n ϭ 10). One way ANOVA followed by Holm-Sidak all-pairwise multiple comparison was used for determination of statistical significance; *, p Ͻ 0.05 ACh-induced currents were significantly different. In PTX-treated cells transiently transfected with G␣ iqi C351A chimera, basal and total currents were not significantly different. Solutions being perfused are indicated above each trace with gaps representing LK application.
mations upon exposure to non-hydrolyzable GTP analogs. First, the purification process from Sf9 cells is based on the protocol developed by Kozasa (32) and is contingent upon G␣ interaction with G␤␥ in the presence of GDP and release of G␤␥ in the presence of the GTP analog GDP-AlF 3 Ϫ . The G␣-GDP bound to His-tagged G␤␥ remains on the nickel-nitrilotriacetic acid matrix column until GDP-AlF 3 Ϫ activates G␣ and elutes it from G␤␥. Using this protocol, we purified G␣ i , G␣ q , G␣ qiq , and G␣ iqi , suggesting that all of these subunits are able to bind G␤␥ and release it upon activation. We next used the trypsin protection assay to test whether in the presence of GDP (inactive) or GTP non-hydrolyzable analogs (active) purified wild-type and chimeric G␣ subunits showed differences in their tryptic digestion patterns (48).
GTP␥S (a non-hydrolysable GTP analog)-bound G␣ takes on an active conformation that protects surface-exposed trypsin cleavage sites from proteolysis and results in loss of only a ϳ2-kDa fragment. In contrast, in the inactive conformation, the GDP-bound G␣ exposes the trypsin-sensitive sites, and proteolysis yields fragments of ϳ10 -20 kDa. Thus, use of the trypsin protection assay can assess whether the G␣ subunits and their chimeras can adopt active conformations as a result of binding to GTP. The G␣ i , G␣ q , G␣ qiq , and G␣ iqi subunits were exposed to trypsin in the presence of GDP or GTP␥S and subjected to SDS-PAGE and Western blotting. In the presence of GDP, the ϳ41-kDa G␣ i subunits were cleaved to yield several low molecular weight fragments in the range of ϳ19 kDa (Fig.  4B). When G␣ i was incubated with GTP␥S, the low molecular weight fragments were greatly reduced and trypsin proteolysis instead resulted in a protected ϳ39-kDa high molecular weight band. This result is consistent with the interpretation that the purified G␣ i subunit upon binding to GTP␥S assumed an active conformation, which partially protected it from trypsin digestion. Similar results were obtained with G␣ q . Incubation with trypsin resulted in extensive proteolysis to fragments not larger than ϳ10 kDa when G␣ q was in the GDP-bound form (Fig. 4C).
In the presence of GTP␥S there was significant protection, but a larger fraction of G␣ q was fully trypsinized compared with G␣ i . This result might be the consequence of the slow nucleo-tide exchange rate of G␣ q (49,50). Nevertheless, in the presence of GTP␥S, G␣ q was partially protected from trypsin, as can be seen by the large increase in the intensity of the 39-kDa molecular weight band and the corresponding reduction in the low molecular weight bands compared with G␣ q -GDP. The trypsin protection assay carried out with G␣ iqi showed that this G␣ subunit is sensitive to trypsin proteolysis, yielding 10-kDa fragments in the presence of GDP (Fig. 4A). When incubated with GTP␥S, G␣ iqi displayed a prominent trypsin-resistant 39-kDA band. G␣ qiq -GDP was mostly digested by trypsin, whereas in the presence of GTP␥S, a small fraction of the protein did display the 39-kDa band characteristic of the trypsin-resistant form in the activated conformation (Fig. 4D). The low intensity of the G␣ qiq signal could be attributed to a lower affinity of the antibody we used to detect both G␣ q and G␣ qiq . Although both control lanes of G␣ q and G␣ qiq (in the absence of trypsin) contained 25 ng of G␣, the G␣ qiq signal was fainter than that of G␣ q (Fig. 4, C and D, first lane).
To further examine the conclusion that the faint band of the trypsin-resistant G␣ qiq was due to a reduced antibody affinity and not to the inability of G␣ qiq to be protected from proteolysis by GTP␥S, we carried out another trypsin protection assay, but this time we used Coomassie Blue stain to visualize the bands on the SDS-PAGE. Fig. 4E shows a convincing protected G␣ qiq band in the presence of GTP␥S. Collectively, these results show that the purified G␣ i , G␣ q subunits, and their chimeras assume a trypsin-resistant, active conformation upon GTP␥S binding. This result is particularly important for the G␣ iqi chimera, suggesting that its inability to support agonist-induced Kir3 currents was not due to a defect in binding to GTP or in assuming an active conformation.
Helical Domain of G␣ i Is Sufficient to Mediate Binding to the Kir3.4 C-terminal Cytosolic Domain-Unlike the N terminus and the GTPase domain, the helical domain of G␣ i does not directly interact with G␤␥ ( Fig. 2A) (42). Thus, if the dependence of the G␤␥-specific activation of Kir3 channels on the helical domain of G␣ i/o involved changes in the G␣ i -G␤␥ interactions, then these effects could only be indirect. Alternatively, G␤␥-specific signaling may be regulated through direct interactions between G␣ and Kir3. Direct interactions between G␣ subunits and the Kir3 channel have been shown both in vivo and in vitro. G␣ subunits were shown to co-immunoprecipitate with the full-length Kir3 channel (51,52). 5 Moreover, G␣ subunits interact directly with Kir3 cytosolic domain fragments in vitro (36,(52)(53)(54)(55). Although both G␣ i and G␣ q can bind the N-terminal domain of Kir3, G␣ i , but not G␣ q or G␣ s , can bind the C terminus of Kir3 (52,55). We tested the hypothesis that the helical domain mediates direct interactions of the G␣ subunit with Kir3. To qualitatively compare binding affinities between G␣ subunits and Kir3 cytosolic domains, we purified GST-tagged N and C termini of Kir3.4 and used them to pull down G␣ i , G␣ q , G␣ qiq , and G␣ iqi . We subjected to Western blotting the samples from GST pulldown assays carried out with either GST-Kir3.4C or GST-Kir3.4N and the four G␣ subunits. GST alone was used as a control for nonspecific binding. Although nonspecific binding was observed to various extents for all G␣ subunits, we corrected for its contribution to pulldown with GST-tagged Kir3.4 N or C termini (see "Western Blot Densitometry and Analysis" under "Experimental Procedures"). Analysis of GST pulldown with the Kir3.4 N terminus revealed that its binding to G␣ i and G␣ q did not show significant differences (data not shown). Similarly, the G␣ chimeras (G␣ qiq and G␣ iqi ) did not exhibit significant differences in binding with the channel N terminus, although these experiments showed large variability (n ϭ 5, data not shown). In contrast, binding of the G␣ subunits to the C terminus of Kir3.4 showed interesting and significant differences (Fig. 5, A and B). The means of the G␣ band intensities shown in Fig. 5B were obtained as described under "Experimental Procedures." The G␣ i and G␣ q binding results with the C terminus as well as studies showing that G␣ q can interact with the N but much less with the C terminus of Kir3 (52,55) are all consistent with the interpretation that it is the Kir3 C terminus that is sensitive to the G␣ subunit type. To determine whether the difference in Kir3 C terminus binding to G␣ i and G␣ q can be attributed to the helical domain, we carried out the GST pulldown assay with G␣ qiq and G␣ iqi . G␣ qiq was pulled down with GST-Kir3.4C to a similar extent as G␣ i and significantly more than G␣ q . This result supported the hypothesis that the helical domain of G␣ i controls the strength of interaction between G␣ and the Kir3 C terminus. Surprisingly, G␣ iqi was pulled down with GST-Kir3.4C to a greater extent than G␣ q , G␣ i , and G␣ qiq . Thus, in contrast to helical domain of G␣ i , the helical domain of G␣ q alone did not mimic the binding behavior of the wild-type G␣ q subunit.

DISCUSSION
The determinants of G␤␥ signaling to Kir3 channels upon activation of PTX-sensitive rather than PTX-insensitive pathways observed in native and heterologous systems have remained unknown. In the present study we compared signaling through the G i and G q pathways in two different heterologous expression systems, Xenopus oocytes and the HEK cell line, and confirmed that in both these systems G␤␥ prefers to signal to Kir3 through G i rather than G q . We showed that swapping receptor coupling of G␣ i and G␣ q does not compromise G protein effector specificity, in agreement with previous studies comparing G␣ i and G␣ s (23). Therefore, because G␤␥ specificity seemed to be dependent on the specific G␣ subunit with which it associated, we set out to explicitly test this possibility. We found a G␣ i region between residues 63 to 175, spanning the helical domain of G␣ i that plays a critical role in confering signaling specificity to G␤␥. A G␣ chimera containing the G␣ i helical domain in the background of G␣ q supported agonistinduced G␤␥ potentiation of Kir3, whereas the reverse construct, in which the helical domain of G␣ i was replaced with that of G␣ q , compromised G␤␥ signaling to Kir3.
G␤␥ binding to the Kir3 channel is critical for K ϩ current activation (35). Several in vitro studies have shown that G␣-GDP prevents G␤␥ from binding the Kir3 C terminus. However, in the absence of G␣ or under conditions that promote dissociation of G␣ from G␤␥ (e.g. treatment of the heterotrimer with GTP␥S) the G␤␥ interaction is restored (36,53). In contrast, G␤␥ could bind the Kir3 N terminus in the presence of G␣-GDP or G␣-GTP␥S (36,53). The importance of the Kir3 C terminus in agonist-induced activation of Kir3 by G␤␥ was further supported by the discovery that mutation of L339E in the C terminus of Kir3.4 (L333E of Kir3.1) abolished agonistinduced Kir3 currents, whereas it did not affect basal currents (14). Taken together these studies suggest that functionally important interactions of G␤␥ with the Kir3 C terminus are regulated by the activation state of the G protein.
Binding of G␣ i subunits both in the presence and absence of G␤␥ to the N-and C-terminal tails of Kir3.1 and Kir3.2 subunits has also been shown by several groups (36,(52)(53)(54). To examine the mechanism of the G␤␥ dependence on the G␣ i helical domain for its signaling to Kir3 channels, we tested the hypothesis that this domain regulates direct interactions of the G␣ subunit with Kir3. We observed a trend of higher binding to the Kir3.4 N terminus than with the C terminus with all G␣ subunits and chimeras tested (data not shown). Moreover, all coexpressed G␣ subunits reduced Kir3 agonist-independent cur-rent in Xenopus oocytes. Thus, it is possible that any G␣ subunit can bind to the Kir3 N terminus and sequester G␤␥ from interacting with Kir3, thereby regulating agonist-independent interactions with Kir3. This is consistent with studies showing that both G␣ i and G␣ q can bind to the N terminus of Kir3, whereas only G␣ i exhibited significant binding to the C terminus (51,52,55). Taken together, work from various laboratories described above points to the C terminus of Kir3 as the region that mediates agonist-induced current activation. Thus, it is not surprising that the C terminus would also be sensitive to the G␣ subtype.
The G␣ i helical domain in the background of G␣ q not only could bind the Kir3 C terminus in a manner similar to that of wild-type G␣ i , but it could also support agonist-dependent G␤␥ activation of Kir3 currents. On the other hand, the chimera lacking the helical domain of G␣ i did not support G␤␥ activation of Kir3 and exhibited significantly greater binding with the Kir3 C terminus than the other G␣ subunits tested. Therefore, the ability of the G␣ subunit to interact with the Kir3 C terminus, although critical, it is not sufficient to bring about activation of Kir3 by G␤␥ alone.
We addressed the possibility that the G␣ iqi chimera failed to signal due to some defect unrelated to its interaction with the Kir3 C terminus. Our results argue against this possibility; first, our purification of G␣ iqi depended upon its ability to reversibly interact with G␤␥, arguing that this chimera can bind G␤␥ in the GDP-bound state and release it in the presence of the transition state GTP analog GDP-AlF 3 Ϫ ; second, purified G␣ iqi chimera exhibited decreased sensitivity to tryptic proteolysis in the presence of a GTP analog similar to the wild-type G␣ i , the G␣ q , and the chimeric G␣ qiq ; third, coexpression of this chimera inhibited basal Kir3 currents, in a manner similar to the other G␣ subunits that showed intact signaling. Thus, our results suggest that the helical domain of G␣ i is a critical determinant in conferring specificity onto G␤␥ signaling to Kir3. Moreover, our results as well as those of others mentioned in this discussion suggest that it is differences in G␣ binding to the Kir3 channel that underlie G␤␥ signaling specificity. Yet differences in other G␣ properties possibly controlled by the G␣ helical domains (e.g. GTPase activity, nucleotide exchange rates, etc.) cannot be ruled out on the basis of the present work as potential contributing mechanisms.
In a recent study Riven et al. (56) explored changes in interactions between the G protein subunits and the Kir3 channel using a combination of the total internal reflection fluorescence and the fluorescence resonance energy transfer techniques. The results from this study point to the existence of a preassembled macromolecular signaling complex between the heterotrimeric G protein and Kir3. At rest G␤␥ is tethered to the channel via its interaction with the G␣ subunit, and upon activation G␤␥ changes its relative position to engage in G␣-independent interactions with the channel. Moreover, G␣ was found to remain stationary with respect to the channel regardless of its activation state.
We propose the following model that is consistent with our results and those of others in the literature. Given our binding results, we propose that the G␣ i subunit interacts with both the N and the C termini at rest. Previous studies have shown that Kir3.4 residues His-64 (N terminus) and Leu-268 (proximal C terminus) are critical G␤␥-interaction sites at rest, accounting for agonist-independent currents (35). Consistent with this view, the G␤1 residues Ser-67 and Thr-128 that do not interact with the G␣ subunit in the heterotrimeric state are critical for agonist-independent channel activity (22). Clancy et al. (52) identified a C-terminal region of the Kir3.2 channel that interacts with G␣ o (conserved residues of interaction correspond to Kir3.4 residues Gly-313, Cys-316, and Ala-318) proximally to residue Leu-339 (distal C terminus), mutations of which abolish agonist-induced K ϩ currents (14). Once an agonist activates G␣ i , it is likely to uncover G␤ residues previously masked by G␣ with which the channel C terminus (e.g. Leu-339) can now interact. The G␣ helical domain is somehow key in enabling the C terminus of Kir3 to interact with G␤␥ leading to channel activation. The proximity of the channel G␣-interacting sites (313-318) to the G␤␥-interacting site (339) makes the role of the G␣ binding rather compelling in a structural sense in terms of its ability to enable G␤␥ interactions with the nearby key channel residue.
In our experiments G␣ qiq behaved in a manner similar to G␣ i both in binding the Kir3 C terminus as well as enabling G␤␥ to activate the channel. Presumably the helical domain of G␣ i facilitates interaction of G␤␥ with the Kir3 C terminus to stimulate channel activity. On the other hand, G␣ q did not mediate G␤␥ activation of Kir3, presumably by failing to facilitate G␤␥ interaction with the Kir3 C terminus. In contrast, interaction between the Kir3 C terminus and G␣ iqi hindered G␤␥ activation of Kir3 by binding the Kir3 C terminus too strongly, so that its helical domain failed to mediate the G␤␥ interaction with the Kir3 C terminus for activation. It is interesting to note that the G␣ q domains represented separately in the chimeric constructs could not reproduce the binding behavior exhibited by the wild-type G␣ q . This result underscores the importance of the context of the entire molecule in enabling the helical domain to both interact with and support G␤␥ activation of Kir3. The helical domain may control allosterically G␣ subunit interaction with the channel. Depending on the G␣ context, the helical domain can yield weaker or stronger interactions with the channel than needed to enable G␤␥-mediated activation.
Further studies will be required to determine the validity of this model. Accumulating evidence suggests that signaling between different proteins involves preassembled macromolecular complexes (56 -59). Moreover, several studies show that some receptors show distinct coupling preference for certain members of the G i/o family (60). For example A1 receptor was suggested to couple preferentially to G␣ o over G␣ i2 , whereas M2 coupled to G␣ i2 over G␣ o in cardiac stem cells (61). Therefore, it is important to keep in mind that in an environment where multiple G protein signaling cascades can be simultaneously activated, segregation of signaling components maybe critical to the fidelity of the response. Our results suggest that within a specific signaling cascade, a mechanism has evolved that relays the signal in the most efficient and specific way such that G␣ q at native levels of expression is not able to activate Kir3 channels.
The role of the helical domain in the function of G␣ subunits has yet to be fully appreciated. The helical domain of G␣t was shown to play a role in both effector cGMP phosphodiesterase and RGS9 regulation (62,63). There is evidence that certain regions of the helical domain are involved in GDP/GTP exchange (64). Moreover, it has been shown that interactions between the GTPase and the helical domain affect nucleotide exchange rates (65). Finally, the GoLoco motif of RGS12 and RGS14 acts as a guanine nucleotide dissociation inhibitor specifically toward G␣ i -GDP, and not G␣ o -GDP (66). Co-crystallization of the RGS14 GoLoco region together with G␣ i -GDP yielded a structure of the GoLoco region interacting with the helix A of the G␣ i helical domain (66,67).
The region corresponding to the helical domain of G␣ is highly divergent even within the G i , G q , G s , and G 12 protein families, consistent with their diverse functions. Here we propose yet another signaling role for the helical domain of G␣ i , as a critical determining factor for G␤␥ signaling specificity to Kir3 channels. G␤␥ couples directly to a number of effectors including voltage-dependent calcium channels and phospholipase C␤ (16). Just as with Kir3, activation of other G␤␥ effectors also occurs specifically. It is possible that the role of the helical domain of G␣ as a determinant of G␤␥ specificity can be generalized to other effectors as well.