Coordination of membrane excitability through a GIRK1 signaling complex in the atria.

Control of heart rate is a complex process that integrates the function of multiple G protein-coupled receptors and ion channels. Among them, the G protein-regulated inwardly rectifying K+ (GIRK or KACh) channels of sinoatrial node and atria play a major role in beat-to-beat regulation of the heart rate. The atrial KACh channels are heterotetrameric proteins that consist of two pore-forming subunits, GIRK1 and GIRK4. Following m2-muscarinic acetylcholine receptor (M2R) stimulation, KACh channel activation is conferred by the direct binding of G protein betagamma subunits (Gbetagamma) to the channel. Here we show that atrial KACh channels are assembled in a signaling complex with Gbetagamma, G protein-coupled receptor kinase, cyclic adenosine monophosphate-dependent protein kinase, two protein phosphatases, PP1 and PP2A, receptor for activated C kinase 1, and actin. This complex would enable the KACh channels to rapidly integrate beta-adrenergic and M2R signaling in the membrane, and it provides insight into general principles governing spatial integration of different transduction pathways. Furthermore, the same complex might recruit protein kinase C (PKC) to the KACh channel following alpha-adrenergic receptor stimulation. Our electro-physiological recordings from single atrial KACh channels revealed a potent inhibition of Gbetagamma-induced channel activity by PKC, thus validating the physiological significance of the observed complex as interconnecting site where signaling molecules congregate to execute a coordinated control of membrane excitability.

Control of heart rate is a complex process that integrates the function of multiple G protein-coupled receptors and ion channels. Among them, the G protein-regulated inwardly rectifying K ؉ (GIRK or K ACh ) channels of sinoatrial node and atria play a major role in beat-tobeat regulation of the heart rate. The atrial K ACh channels are heterotetrameric proteins that consist of two pore-forming subunits, GIRK1 and GIRK4. Following m 2 -muscarinic acetylcholine receptor (M2R) stimulation, K ACh channel activation is conferred by the direct binding of G protein ␤␥ subunits (G␤␥) to the channel. Here we show that atrial K ACh channels are assembled in a signaling complex with G␤␥, G protein-coupled receptor kinase, cyclic adenosine monophosphatedependent protein kinase, two protein phosphatases, PP1 and PP2A, receptor for activated C kinase 1, and actin. This complex would enable the K ACh channels to rapidly integrate ␤-adrenergic and M2R signaling in the membrane, and it provides insight into general principles governing spatial integration of different transduction pathways. Furthermore, the same complex might recruit protein kinase C (PKC) to the K ACh channel following ␣-adrenergic receptor stimulation. Our electrophysiological recordings from single atrial K ACh channels revealed a potent inhibition of G␤␥-induced channel activity by PKC, thus validating the physiological significance of the observed complex as interconnecting site where signaling molecules congregate to execute a coordinated control of membrane excitability.
The response of the heart to parasympathetic stimulation relies on activation of cardiac K ACh channels by acetylcholine, via stimulation of m 2 -muscarinic receptors and consequent activation of the heterotrimeric G proteins coupled to these receptors. The atrial K ACh channels are tetrameric proteins composed of two homologous subunits, GIRK1 1 (1, 2) and GIRK4 (3). Each GIRK subunit consists of two transmembrane helixes (TM1 and TM2), a P region, and cytoplasmic NH 2 -and COOHterminal domains. The transmembrane helixes and the P regions of the GIRK subunits are thought to form the canonical transmembrane pore of the channel (4), whereas the NH 2 and COOH termini are structurally organized in a tetrameric cytoplasmic pore (5). Following M2R stimulation, K ACh channel activation is conferred by direct binding of G protein ␤␥ subunits probably to multiple cytoplasmic regions of the GIRK subunits (for review see Ref. 6).
While the K ACh channels are activated by G␤␥, their function is modulated by a large number of signaling molecules, including cytoplasmic ATP (7), Na ϩ (8,9), and unsaturated free fatty acids (10). Several membrane factors such as phosphatidylinositol 4,5-bisphosphate (11) and derivatives of arachidonic acid (12,13) are also involved in the modulation of K ACh channel activity. Furthermore, G␣ and GTPase-activating proteins such as regulators of G protein signaling can quench K ACh channel activation by G␤␥ (14 -16). Most importantly, K ACh channels are regulated by protein phosphorylation via PKA (17)(18)(19) and dephosphorylation via PP2A (19,20). The large number of signaling proteins involved in the modulation of K ACh channel function raises the possibility that some or all of these molecules might be organized in a macromolecular complex. The molecular size of native channels determined by size exclusion chromatography further supports this hypothesis and is consistent with two K ACh channel configurations, single GIRK1 2 /GIRK4 2 heterotetramers, and large protein complexes containing K ACh channels (21). The composition of such complexes would have important ramifications for the control of membrane excitability and heart rate. We therefore set out to investigate the macromolecular K ACh channel complexes and their components.
We immunopurified K ACh channels from rat atrial membranes and found that these channels are associated with G␤␥, G protein-coupled receptor kinase (GRK), PKA, two protein phosphatases, PP1 and PP2A, receptor for activated C kinase 1 (RACK1), and actin. To identify the functional role of RACK 1 association with K ACh channels we investigated the effect of PKC on G␤␥-activated K ACh channel currents. PKC induced a potent inhibition of these currents, thus confirming the physiological significance of the observed signaling complex.
Immunoprecipitation and Immunoblotting-Atrial tissue, obtained in accordance with institutional guidelines from adult Sprague-Dawley rats (Harlan Sprague-Dawley, Inc.), was homogenized in five volumes of ice-cold HME buffer (20 mM HEPES-KOH, pH 8.0, 2 mM MgCl 2 , and 1 mM EDTA), containing 0.32 M sucrose and protease inhibitor (PI)mix (20 g/ml leupeptin, 3 g/ml aprotinin, 2 g/ml pepstatin, 0.1 mM phenylmethylsulfonyl fluoride) using a Polytron homogenizer at half speed for 10 s, followed by 10 strokes in a glass Potter-Elvehjem type homogenizer. The homogenate was filtered and centrifuged at 500 ϫ g for 15 min at 4°C to pellet nuclei and unbroken cells. The pellet was washed once with HME-PI buffer, and the combined supernatants were centrifuged further at 100,000 ϫ g for 45 min. The resulting crude membrane pellet was washed twice with HME-PI buffer, resuspended in buffer HME-PI containing 0.32 M sucrose in a Potter-Elvehjem homogenizer (tight pestle), quickly frozen in liquid nitrogen, and stored in aliquots at Ϫ70°C until use. For the PKC translocation experiments, excised atria were equilibrated in oxygenated solution containing (in mM) 140 NaCl, 5.4 KCl, 1 CaCl 2 , 1.6 MgCl 2 , 5 HEPES, 5 glucose, pH 7.4, stimulated with 10 M phenylephrine for 10 min at room temperature and frozen in liquid nitrogen. Unstimulated tissue processed in parallel was used as a control.
Membrane-associated proteins were solubilized according to a protocol previously used by us for purification of heterotrimeric G proteins (22). The membranes were incubated for 3 h in MEB buffer (20 mM HEPES, pH 8.0, 2 mM MgCl 2 , 1 mM EDTA, 100 mM NaCl, 10 mM ␤-mercaptoethanol, 1% sodium cholate, and 0.45% IGEPAL CA-630) with PI. The cholate-insoluble material was removed by centrifugation at 100,000 ϫ g for 1 h, and the supernatant was used for immunoprecipitations. Protein concentration was determined by Bio-Rad Dc protein assay against a standard curve of BSA. For the experiments, labeled "GDP" in Fig. 7, all buffers in the procedures described above contained 100 M GDP.
Immunoprecipitations were performed with Dynal magnetic beads according to the manufacturer's protocol using a magnetic particle concentrator (Dynal, Lake Success, NY). The use of magnetic beads eliminates the centrifugation steps in the immunoprecipitation process. Consequently, a possible contamination of the immunoprecipitation product by large protein complexes that might be trapped within antibody-protein A/G beads latticework during centrifugation is avoided. Briefly, beads were washed thoroughly with PBS (pH 7.4) containing 0.1% (w/v) BSA (PBS/BSA) and incubated for 1 h with rotation at 4°C with the anti-phosphoserine (4A9 clone) or anti-G␤ antibodies at a concentration of 3 g per 1.2 ϫ 10 7 M-450 rat anti-mouse IgM beads, or anti-M2R (Oncogene) antibody, or normal mouse IgG at a concentration of 3 g per 1.8 ϫ 10 7 M-280 sheep anti-mouse IgG beads. All antibodies used for immunoprecipitation recognized only the protein of interest in the atrial membranes. Coated beads were washed twice with PBS/BSA and once with PBS. 400 g of solubilized membrane proteins (at 5 mg/ml) was diluted 6.7-fold with appropriate buffer to a final concentration of 0.15% sodium cholate, 0.3%, IGEPAL, 25 HEPES, pH 8.0, 3 mM MgCl 2 , 1 mM EDTA, 100 mM NaCl, 10 mM ␤-mercaptoethanol and incubated with the coated beads for 2 h at 4°C with rotation. The complexed beads were collected by magnetic precipitation, washed four times with PBS/0.1% IGEPAL, resuspended in Laemmli sample buffer (62.5 mM Tris-HCl, pH 6.8, 2% SDS, 10% v/v glycerol, 100 mM dithiothreitol, 0.02% bromphenol blue), and heated for 5 min at 95°C (70°C for M2R detection). The eluted proteins were separated by Tris-glycine SDS-PAGE using an 11% gel and electrophoretically transferred to Immobilon-P sq membranes. Following blocking in LTB (50 mM Tris-HCl, pH 8.0, 80 mM NaCl, 2 mM CaCl 2 , 0.1% Tween 20, containing 5% dry milk) for 1 h at room temperature, sections of the membrane containing the molecular mass range of proteins of interest were incubated with the appropriate antibodies at 4°C for 14 h. After three washes in LTB, the membrane was incubated for 1 h at room temperature with horseradish peroxidase-conjugated secondary antibodies, and antibody complexes were visualized by using an enhanced chemiluminescence detection system.
The modal behavior of G␤ 1 ␥ 7 -activated K ACh channels was analyzed in the context of a model that assumes four functional modes of a single K ACh channel arising from different occupancy of four equivalent and independent G␤␥ sensors in the channel, as previously described (22). The continuous records were divided into 400-ms consecutive segments, and the frequency of apparent openings, f, was calculated for each data segment. The results from all active data segments recorded under identical experimental conditions were then combined to generate a frequency distribution. The f histograms were subsequently fitted by a sum of gamma components, to determine the characteristic f k values and the relative occupancy of different gating modes, F k (F k ϭ N k /⌺N k ). For each f histogram, the probability of G␤␥ binding, P, was computed from the relative occupancy of mode 1, F 1 , according to the Equation 2.
The relative occupancy of the remaining modes, F k , was then examined as a function of the parameter P to verify that experimental F k estimates were consistent with the theoretically derived probability of observing mode k, where P is the probability of G␤␥ binding to one of four G protein sensors in K ACh channel structure, and N ϭ 4. For each experimental condition, the fraction of blank data segments due to K ACh channel desensitization, F D , was calculated as the difference, (F 0E Ϫ F 0T ), where F 0E is the experimentally determined fraction of blank data segments, and F 0T is the predicted fraction of time the channel spends with no G␤␥ bound to it. The F 0T fraction, F 0T ϭ (1 Ϫ P) 4 , was calculated from the probability of G␤␥ binding, P, according to Equation 3.
G protein G␤ 1 ␥ 7 dimers were expressed and purified from Sf9 cells in the Laboratory of Dr. Janet Robishaw as described previously (22).

Immunopurification of a Macromolecular Complex Contain-
ing GIRK 1-Previous immunopurifications using antibodies against N-and C-terminal domains of GIRK1 yielded only the single GIRK1 2 /GIRK4 2 heterotetramers but not large K ACh channel protein complexes (3,24). In such multimeric complexes, N-and C-terminal domains of GIRK1 might be bound to other proteins and, thus, rendered unavailable for antibody binding. Therefore, initially we sought to identify an antibody suitable for immunoprecipitation of the macromolecular K ACh channel complexes. In rat atrial membranes, GIRK1 protein exists in two isoforms, migrating at 54 and 64 -65 kDa (Fig.  1A), which correspond to the core and the glycosylated form of GIRK1 (3). The native GIRK1 is also constitutively phosphorylated on serine/threonine residue(s) (20). We therefore screened a panel of monoclonal antibodies, specific for different serine and threonine phosphorylated sites, for their ability to recognize GIRK1. Six of the phospho-specific antibodies either did not recognize any proteins in the atrial membrane extracts or identified multiple protein bands. In contrast, the phosphoserine-specific 4A9 clone consistently recognized a single protein band of ϳ54 kDa in the atrial membranes (Fig. 1B). Ac-cording to the supplier (Calbiochem), the 4A9 clone was generated against a phosphoserine conjugated to keyhole limpet hemocyanin, using a proprietary technology, and preferentially recognizes phosphoserine residues directly neighbored to positively charged amino acids. To determine whether the 54-kDa polypeptide is the core form of GIRK1, we immunoprecipitated K ACh channels from atrial membrane lysate with antibodies against the N terminus of GIRK1 (25). Subsequent immunoblotting confirmed that the 4A9 antibody detected the core form of the immunoprecipitated GIRK1 (Fig. 1C); interestingly, this antibody did not recognize the glycosylated form of GIRK1. In a reciprocal experiment we investigated the ability of 4A9 antibodies to immunoprecipitate the cardiac GIRK1. Atrial membrane lysates were incubated with 4A9 antibodies, and the resulting immunocomplexes were analyzed by silver staining (Fig. 1D). The 4A9 immunocomplexes contained four prominent bands with molecular masses of 80 kDa (labeled P80), 65 kDa (labeled Gly-GIRK1), 54 kDa (labeled GIRK1), 44 kDa (labeled P44), and four additional bands migrating between 37 and 32 kDa (Fig. 1D, lane 2). The 54-and 65-kDa bands corresponded in size to the two forms of GIRK1, and their identity was confirmed by immunoblotting of a gel run in parallel with antibodies to GIRK1 (Fig. 1E). The 44-kDa protein migrated at the predicted size of GIRK4, which is known to co-immunoprecipitate with GIRK1 (3). None of the protein bands observed in the 4A9 immunoprecipitate were detected in the product from a parallel immunoprecipitation using normal mouse IgG (Fig. 1D, lane 1). Because the phosphoserine-specific 4A9 antibody recognized only the GIRK1 protein in atrial membranes as shown in Fig. 1B, the additional proteins in the 4A9-immunoprecipitate should be components of a native multimeric complex containing GIRK1. We further sought to establish the identity of the proteins associated with the K ACh channels.
Composition of GIRK1 Signaling Complex-The spatial organization of GPCR, G protein, and GIRK channel within a pre-coupled complex in the membrane was suggested as a potential mechanism for achieving specificity of the signal transduction from a particular receptor to the channel (26). To determine whether the K ACh channels pre-assemble with the M2R and G␤␥, solubilized atrial membranes were immunoprecipitated with antibodies against M2R, phosphoserine (clone 4A9), or G␤, followed by immunoblotting for M2R, GIRK1, and G␤. We found that GIRK1 co-precipitated with G␤␥ and, conversely, G␤ immunoprecipitates contained GIRK1 ( Fig. 2A). Furthermore, parallel analysis of G␤-and 4A9-immunocomplexes by silver staining revealed that G␤ antibodies immunoprecipitated the same set of proteins found in the 4A9 product (data not shown). The M2R, however, was not detected in both 4A9 and G␤ immunoprecipitates ( Fig. 2A). In a reciprocal immunoprecipitation of the endogenous M2Rs, we did not detect an association between M2R and GIRK1 or G␤ ( Fig. 2A). Thus, in atrial membranes we demonstrated the presence of an endogenous GIRK1-G␤␥ complex that, under the described experimental conditions, did not contain M2R.
Recruiting of G protein-coupled receptor kinases to specific membrane domains of active GPCR signaling is a ubiquitous function of G␤␥ that leads to termination of G protein activa- . The immunoprecipitation products were collected, and the components were separated by 11% SDS-PAGE. Lanes containing a sample of 5 and 20% of the lysate used for immunoprecipitation were also run on this gel (labeled "5% load" and "20% load," respectively). Proteins were transferred to a PVDF membrane, which was cut into sections, and each probed with the antibodies indicated to the left (Blot). tion (27). We therefore investigated whether GRK2, which regulates the desensitization of both ␤-adrenergic and m 2 -muscarinic receptors in the heart (28), is also associated with the GIRK1-G␤␥ complex. GRK2/3 was detected as a single band of 80 kDa in the atrial membrane lysate (Fig. 2B). Immunoblotting of GIRK1 and G␤ immunocomplexes for GRK2/3 revealed the association of these proteins with the oligomeric complex (Fig. 2B) and identified the P80 as GRK2/3.
Binding of GRK2 and G␣ to G␤␥ is mutually exclusive due to their overlapping binding sites on the surface of G␤␥ (29,30). To determine whether a fraction of the G␤ in the GIRK1 complex is in the form of a G protein heterotrimer, we immunoblotted the 4A9 and G␤ immunoprecipitates for G␣ (Fig. 2B). G␣ was not detected in these experiments, indicating the absence of heterotrimeric G proteins in the GIRK1-G␤␥ complexes described here. However, the possibility that K ACh channel-heterotrimeric G protein complexes exist in atrial cells but are disrupted during detergent solubilization of membrane proteins cannot be excluded.
Phosphorylation by PKA (17)(18)(19) and dephosphorylation by PP2A (19,20) of the K ACh channels are key regulatory mechanisms that control channel sensitivity to G protein stimulation. We therefore investigated whether PKA and PP2A are also associated with GIRK1-G␤␥ complexes. PKA, PP1, and PP2A were present in relatively small amounts in both 4A9 and G␤ immunoprecipitates (Fig. 2B), indicating that these proteins are not only involved in the regulation of K ACh channel function but are also components of the channel complex. PP2A was also found in the M2R immunoprecipitation product (Fig. 2B).
The purified GIRK1 protein is also a substrate for PKC phosphorylation (20). Following activation, PKC is recruited to the plasma membrane and co-localized with its substrates by scaffold proteins, RACKs (31). Stimulation of atrial ␣-adrenergic receptors by phenylephrine is associated with translocation of PKC␤ to the plasma membrane (Fig. 3). Because RACK1 has been identified as the specific anchoring protein for PKC␤ (32), we investigated whether RACK1 is associated with the oligomeric K ACh channel complex by probing the M2R, 4A9, and G␤ immunoprecipitates with antibodies against RACK1. Endogenous atrial RACK1, migrating at 33 kDa, was found present in both 4A9 and G␤ immunoprecipitates (Fig. 2B).
Suppression of K ACh Channel Activity by PKC-Several studies have implicated PKC in the regulation of heterologously expressed GIRK channels (reviewed in Ref. 33). Because RACK1 might provide the structural link between PKC and the native atrial K ACh channels, we evaluated the effect of PKC phosphorylation on channel activation by G␤␥. Single K ACh channels residing in membrane patches excised from rat atrial myocytes were activated by purified recombinant G␤␥, and then purified PKC was applied to the cytoplasmic side of the patch in the presence of Mg-ATP. Application of a mixture of purified PKC isotypes (40 pg/l) caused a potent suppression of G␤␥-activated K ACh channel currents (Fig. 4). The suppression of the K ACh channel activity developed within 2 min following PKC application and was sustained during the continued presence of PKC. In control experiments, heat-inactivated PKC had no effect on the G␤␥-induced K ACh channel activity (Fig. 5A).
Recently, mutagenesis analysis of heterologously expressed GIRK1 and GIRK4 subunits identified Ser-185 in GIRK1 and Ser-191 in GIRK4 as the putative PKC phosphorylation sites in the K ACh channel (34). PKC might employ at least two different mechanisms to inhibit the K ACh channel currents. It could either reduce the affinity of the K ACh channels for G␤␥ binding, thus impairing the activation process, or accelerate K ACh channel desensitization, affecting channel availability for G␤␥ activation. To distinguish between these two possibilities, we studied the effect of Mg-ATP and PKC on the gating behavior of G␤␥-activated K ACh channels. Upon G␤␥ activation atrial K ACh channels fluctuate between four functional modes rendered active by the apparent binding of increasing number of G␤␥ subunits to four identical and independent G protein sensors on each channel (22). To define the characteristic frequency of gating for individual modes and the equilibrium among them, we generated f histograms (see "Experimental Procedures") from data recorded from the same membrane patches in the presence of G␤ 1 ␥ 7 alone, G␤ 1 ␥ 7 /Mg-ATP, and G␤ 1 ␥ 7 /Mg-ATP/ PKC (Fig. 5B). Neither the characteristic frequency of gating within individual functional modes, f k , nor the mean open time was affected in the presence of PKC (Fig. 5B). However, comparison of the f histograms revealed that PKC suppression of the K ACh channel activity was associated with a shift in the equilibrium between the different gating modes. Furthermore, this shift was accompanied by a ϳ50% reduction in the probability of G␤␥ binding to the channel, P (Fig. 5B). At the same time, the comparison of the fractions of blank data segments attributed to channel desensitization revealed a ϳ20% increase in the F D value when the K ACh channels were switched from G␤ 1 ␥ 7 /Mg-ATP to G␤ 1 ␥ 7 /Mg-ATP/PKC. No significant difference in the F D value was detected when the channels were switched from G␤ 1 ␥ 7 to G␤ 1 ␥ 7 /Mg-ATP. Thus, analysis of K ACh channel gating indicates an efficient dual mechanism for PKC suppression of K ACh channel activity affecting both G␤␥-binding affinity and desensitization of the ion channel.
Association of Actin with GIRK1-G␤␥ Complex-The actin FIG. 4. Suppression of K ACh channel activity by PKC. Inside-out patch clamp recording of a single K ACh channel activity from a neonatal rat atrial myocyte. The membrane potential was clamped at Ϫ90 mV. Current traces show channel activity (downward deflections) in the presence of 5 nM G␤ 1 ␥ 7 (A); 10 min after addition of 40 M Mg-ATP (B); and 10 min after addition of 40 pg/l PKC to the bath (C). Calibration: 2 pA and 500 ms. A mixture of ␣, ␤, and ␥ isoforms of PKC purified from rat brain (Upstate Biotechnology) was used in these experiments. cytoskeleton provides the structural framework necessary for the assembly of highly structured protein complexes in the cells. In addition, the cytoskeleton is implicated in regulation of a large number of ion channels and pumps (reviewed in Ref. 35). Mechanosensitivity of the K ACh channels has been also reported (36,37). We therefore tested whether actin is also associated with the macromolecular channel complex. Anti-actin-specific antibodies recognized a protein band migrating at the predicted molecular size (43 kDa) of actin in both 4A9 and G␤ immunoprecipitates, confirming an associ-ation of actin with the GIRK1-G␤␥ complexes (Fig. 6). We also investigated whether the small GTPases, Rac, Rho, and Cdc42, which control the organization and dynamics of actin cytoskeleton (reviewed in Ref. 38), are components of the GIRK1 signaling network. Neither Rac nor Rho was found in the M2R, 4A9, and G␤ immunoprecipitates (Fig. 6); Cdc42 was not detected in the atrial membrane lysates (data not shown). Ras and two caveolae markers, caveolin 1 (data not shown) and caveolin 3, were also not present in the immunoprecipitates (Fig. 6).
Evidence for Transient Association of GIRK1 and G␤␥ in the Atrial Membrane-Finally, we sought to determine whether some components of the GIRK1-G␤␥ macromolecular complex are contributed through their association with G␤␥. The 4A9 antibodies were incubated with atrial membrane lysates prepared in the presence of 100 M GDP. These conditions favor the formation and stabilization of inactive G␣␤␥ heterotrimers. The resulting 4A9 immunocomplexes contained GIRK1, GRK2/3, PKA, PP2A, and actin, whereas G␤␥, RACK1, and PP1 were not present under these experimental conditions (Fig. 7). This observation is consistent with a recent report demonstrating a direct binding of RACK1 to G␤␥ (39) and suggests that G␤␥ might facilitate the association of RACK1 and PP1 with the macromolecular GIRK1-G␤␥ complex. A direct binding of GRK to heterologously expressed GIRK1/GIRK2 heterotetramers has also been reported (40). Altogether, these experimental data indicate that (i) GIRK1 and G␤␥ bring different partners to the macromolecular signaling complex and (ii) the composition of this complex is a function of G protein activation in the atrial membrane. Lysates from rat atrial membranes were immunoprecipitated with antibodies against M2R, phosphoserine (clone 4A9), G␤, and normal mouse IgG (IP). The immunoprecipitates and lysates (lanes labeled "5% load" and "20% load") were blotted with the antibodies indicated to the left (Blot).

DISCUSSION
In the present study we demonstrate that the pore-forming GIRK subunits of native atrial K ACh channels are associated with G␤␥, GRK2/3, PKA, two counterpart phosphatases PP1 and PP2A, RACK1, and actin. The finding, that independent immunoprecipitations with an antibody specific for GIRK1 and with an antibody specific for G␤ yield the same set of associated proteins, presents a strong argument for existence of a large signaling GIRK1-G␤␥ complex in vivo. The presence of such a multimeric entity might explain the significant variability in the sensitivity of the native atrial K ACh channels to M2R stimulation. This variability is manifested by a wide range of open probability estimates (0.005-0.250) obtained from individual K ACh channels activated in cell-attached patches by 1 M ACh (22). We have also observed similar functional differences among the K ACh channels activated by purified recombinant G␤␥. This observation indicates that at least some molecular determinants of channel sensitivity to G protein activation are localized in the membrane and continue to exercise their control even after the channels have been removed from their cytoplasmic environment. A recent study demonstrated the ability of heterologously expressed ␤ 2 -adrenergic, dopamine D 2 , and D 4 receptors to form stable complexes with different GIRK channels in COS-7 and HEK 293 cells (41). In this case, however, the assembly of the receptor/channel complexes is likely to be a biosynthetic event, driven by the co-expression of these proteins and does not occur when the receptors and channels are separately expressed, solubilized, and mixed together. The likely existence of similar complexes in other heterologous expression systems might account for the different sensitivity of the expressed GIRK1/GIRK4 channels to G␤␥ stimulation observed by different groups (9,42).
The assembly of macromolecular signaling complexes containing defined GPCRs, G proteins, and their downstream effectors can provide a mechanism that ensures the specificity of G protein-mediated signal transduction. The existence of such signaling complexes was recently demonstrated for the neuronal ␤ 2 -adrenergic receptors, adenylyl cyclase, and L-type Ca 2ϩ channels (Ca v 1.2) (43). This receptor/channel complex also contained heterotrimeric G proteins, PKA and PP2A. Similar ␤ 2adrenergic receptor/L-type Ca 2ϩ channel complexes might also exist in the heart (44). In atrial cells, the M2R stimulation leads to inhibition of the adenylyl cyclase involved in the regulation of L-type Ca 2ϩ channels. Data from our laboratory, however, indicate that the atrial GIRK1-G␤␥ complexes are not associated with detectable amounts of L-type Ca 2ϩ channel Ca v 1.2 protein, demonstrating the specificity of interactions between the constituents of the K ACh channel signaling complexes. Interestingly, our experiments showed no evidence for a stable association of M2Rs with the native atrial K ACh channels. However, because detergent solubilization of membrane proteins is known to interfere with some protein-protein interactions, the existence of receptor-effector complexes cannot be completely ruled out.
Most importantly, our findings reveal that protein oligomerization might be essential not only for the specificity of signal transduction but also for the coordination of different G protein-coupled pathways. Formation of macromolecular K ACh channel complexes would allow several signaling pathways to coordinate their function through an intricate network of synergistic and antagonistic events converging on the channel. Although the dynamics and the precise role of individual signaling components in this network remain to be established, it is tempting to speculate that one critical function of the K ACh channel complex would be to terminate M2R activation of the channel. The GRK from this complex would phosphorylate activated M2Rs thereby leading to their rapid desensitization and termination of G protein signaling in the vicinity of the K ACh channel. Another important function of the identified complex might be to effectively control the sensitivity of the K ACh channels to G␤␥ stimulation in response to ␣and ␤-adrenergic signaling. The ␤-adrenergic signaling via PKA phosphorylation increases K ACh channel activity (17)(18)(19), whereas ␣-adrenergic signaling via PKC activation and translocation to the channel would attenuate its activation. Because of these distinct modulatory effects of PKA and PKC on K ACh channel-G␤␥ interactions, the ensuing membrane hyperpolarization would depend not only on M2R activation but also on the stoichiometry and the dynamics of GIRK1 signaling complex. Transient association of different signaling molecules with this complex would enable the K ACh channels to generate a vast repertoire of hyperpolarization responses, coordinating sympathetic and parasympathetic control of membrane excitability. Future studies will rigorously test these hypotheses.