NMR Analyses of the Gβγ Binding and Conformational Rearrangements of the Cytoplasmic Pore of G Protein-activated Inwardly Rectifying Potassium Channel 1 (GIRK1)*

G protein-activated inwardly rectifying potassium channel (GIRK) plays crucial roles in regulating heart rate and neuronal excitability in eukaryotic cells. GIRK is activated by the direct binding of heterotrimeric G protein βγ subunits (Gβγ) upon stimulation of G protein-coupled receptors, such as M2 acetylcholine receptor. The binding of Gβγ to the cytoplasmic pore (CP) region of GIRK causes structural rearrangements, which are assumed to open the transmembrane ion gate. However, the crucial residues involved in the Gβγ binding and the structural mechanism of GIRK gating have not been fully elucidated. Here, we have characterized the interaction between the CP region of GIRK and Gβγ, by ITC and NMR. The ITC analyses indicated that four Gβγ molecules bind to a tetramer of the CP region of GIRK with a dissociation constant of 250 μm. The NMR analyses revealed that the Gβγ binding site spans two neighboring subunits of the GIRK tetramer, which causes conformational rearrangements between subunits. A possible binding mode and mechanism of GIRK gating are proposed.

G protein-activated inwardly rectifying potassium channel (GIRK) 3 is a member of the inwardly rectifying potassium channel (Kir) family, which regulates heart rate and neuronal excitability (1,2). The Kir proteins function as tetramers, consisting of a transmembrane (TM) region and a cytoplasmic pore (CP) region. The helix bundle at the cytoplasmic side of the TM region is assumed to be a K ϩ -ion gate. The opening and closing in the gate (gating) of Kirs are regulated by a variety of cytoplasmic factors. The gating of GIRK is triggered by the binding of its CP region with the heterotrimeric G-protein ␤␥ subunits (G␤␥), which are released from the pertussis toxin-sensitive G protein ␣ subunit (G␣ i/o ), subsequent to the stimulation of a G protein-coupled receptor, such as M2 acetylcholine receptor.
Extensive mutational analyses to identify the GIRK residues that are critical for the G␤␥-induced activation revealed several critical residues such as His 57 , Leu 262 , Leu 333 , and Gly 336 of GIRK1 (3)(4)(5). However, when these residues were mapped on the recently reported crystal structures of Kirs, they did not form a cluster on the protein surface (6 -9). Therefore, no clear consensus has been obtained regarding the region of GIRK that is essential for G␤␥ binding and/or GIRK activation. One of the reasons might be a structural alteration introduced by the mutagenesis (10), which could change the gating property of the channel.
Although the various crystal structures have provided little information about the conformational change involved in the gating of the channel, FRET analyses have clearly demonstrated the conformational rearrangements in the CP region of GIRK upon G␤␥ binding (11). However, the resolution of the structural information obtained by the FRET analyses is low, primarily due to the large size of the fluorescent probe proteins.
To reveal the structural mechanism by which G␤␥ binding activates GIRK, we have investigated the direct interaction between the CP region of GIRK and G␤␥ by isothermal titration calorimetry (ITC) and NMR spectroscopy. Our ITC results indicated that four G␤␥ molecules bind to a tetramer of the CP region of GIRK, with a dissociation constant (K d ) value of 250 M. NMR analyses revealed that the binding of G␤␥ occurs at two adjacent contiguous surfaces, clustered between the two neighboring subunits of the CP region of GIRK, and causes the conformational rearrangements of the tetramer. Based on our analysis, we propose a possible binding mode and structural mechanism of GIRK gating.

EXPERIMENTAL PROCEDURES
Expression and Purification of the Cytoplasmic Regions of Mouse GIRK1-The N-and C-terminal cytoplasmic regions of mouse GIRK1 (residues 41-63 and 190 -371) were fused into a single polypeptide, which is hereafter referred to as * This work was supported in part by grants from the Japan New Energy and Industrial Technology Development Organization (NEDO) and the Ministry of Economy, Trade, and Industry (METI) (to I. S.), a Grant-in-aid for Scientific Research on Priority Areas from the Japanese Ministry of Education, Culture, Sports, and Technology (to M. O. and I. S.), and a grant from Takeda Science Foundation (to M. O.). □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental "Experimental Procedures" and Figs. S1-S5. 1 Research fellow of the Japan Society for the Promotion of Science (JSPS). 2 To whom correspondence should be addressed: 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Construction of Recombinant Baculoviruses-The cDNA encoding the G␥ 2 subunit was amplified from a human heart cDNA library (BD Biosciences) by PCR and inserted into the baculovirus transfer vector pVL1393 (Invitrogen). The sitedirected mutagenesis of the cysteine residue at position 68 of G␥ 2 to serine and the insertion of a hexahistidine tag at the N terminus (G␥ 2 His-C68S) were performed by the QuikChange method (Stratagene). The recombinant baculovirus, encoding the bovine G␤ 1 subunit, was kindly provided by Drs. Y. Fukada and M. Katadae (The University of Tokyo, Japan). The amino acid sequences of G␤ 1 are fully conserved among the bovine, human, and mouse forms.
Expression and Purification of G␤␥-The G␤ 1 and G␥ 2 subunits were co-expressed in expresSFϩ insect cells (Protein Sciences) by infection with the amplified recombinant baculoviruses, encoding the G␤ 1 and G␥ 2 His-C68S subunits at a multiplicity of infection of 2.5 for each baculovirus. After an incubation at 27°C for 60 -72 h, the cells were harvested and re-suspended in 100 ml of buffer A (20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM dithiothreitol (DTT)) containing 1ϫ protease inhibitor mixture (EDTA free) (Nacalai Tesque) per 1 liter of culture cells. All of the following purification procedures were carried out at 4°C. The cell suspension was homogenized with Teflon homogenizer (15 strokes) or by nitrogen cavitation (Parr bomb) at 800 p.s.i. for 30 -60 min. Cell lysates were centrifuged at 750 ϫ g for 10 min to remove intact cells and nuclei. The supernatants were further centrifuged at 100,000 ϫ g for 30 min, and the supernatants were used for purification. The supernatant was purified using HIS-Select Nickel Affinity Gel (Sigma), followed by further purification using His-Trap HP (GE Healthcare). The elution was applied to a HiLoad 16/60 Superdex 200 pg column (GE Healthcare). The purified proteins were concentrated by ul-trafiltration and buffers were adjusted for the following experiments using an Amicon Ultra filter (Millipore).
ITC Analyses-The binding of G␤␥ to GIRK CP was measured by ITC, using a MicroCal iTC200 calorimeter (GE Healthcare), with stirring at 1000 rpm at 25°C. The protein samples were dialyzed against a buffer containing 10 mM HEPES-NaOH (pH 7.5), 50 mM KCl, and 1 mM tris-(2-carboxyethyl)phosphine. The titration of the GIRK CP tetramer with G␤␥ involved 25 injections of 1.5 l of the GIRK CP solution (1.8 mM as the tetramer) at intervals of 180 s into a sample cell containing 200 l of G␤␥(380 M). The heat of dilution of the titrant (the GIRK CP tetramer) was subtracted from the titration data for the GIRK CP titration into G␤␥. The data were analyzed with the MicroCal Origin TM 5.0 software to determine the enthalpy (⌬H), dissociation constant (K d ), and stoichiometry of binding (N). Thermal titration data were fit to a single site binding model and thermodynamic parameters ⌬H, K d , and entropy change (⌬S) were obtained by fitting to the model. The error of each parameter shows the fitting error.
NMR Analyses-All NMR experiments were performed at 303 K on a Bruker Avance 600 spectrometer equipped with a cryogenic probe. All spectra were processed by the Bruker TopSpin 2.1 software and the data were analyzed by Sparky (T. D. Goddard and D. G. Kneller, Sparky 3, University of California, San Francisco). The error bars were presented based on the signal to noise ratio calculated by the Sparky software. The assignments of the backbone NMR signals from GIRK CP were reported (BioMagResBank accession number 11067 (12)).
The TCS experiments were carried out as described with minor modifications (13). A solution containing the uniformly 2 H, 15 N-labeled GIRK CP (300 M as a tetramer) and the nonlabeled G␤␥ (200 M) was prepared in buffer (10 mM HEPES-NaOH (pH 7.5), 50 mM KCl, 2 mM DTT, 1 mM sodium 2,2dimethyl-2-silapentane-5-sulfonate, 20% 1 H 2 O, 80% 2 H 2 O). The selective saturation for the aliphatic protons of G␤␥ was performed with a 15-ms IBURP2 pulse centered at 1.0 ppm. The saturation duration and the relaxation delay were set at 2.5 and 3.5 s, respectively. To evaluate the effect of the residual aliphatic protons within the GIRK CP protein, a TCS experiment was also carried out for the control sample without G␤␥, containing the uniformly 2 H, 15 N-labeled GIRK CP (250 M as a tetramer) alone. It should be noted that the residues with overlapping resonances were excluded from the analyses.

RESULTS
The Binding Affinity of the Cytoplasmic Pore Region of GIRK for G␤␥-In this study, we used the cytoplasmic pore regions of mouse GIRK1, composed of residues 41-63 and 190 -371 as a single polypeptide (hereafter referred to as GIRK CP ), and the prenylation-deficient mutant of G␤␥, composed of the wild-type G␤ 1 and the C68S mutant of G␥ 2 , which has a hexahistidine tag at the N terminus (hereafter referred to as G␤␥).
We previously reported that the purified GIRK CP protein forms a tetramer determined by size exclusion chromatography combined with multiple angle light scattering (10). The 1 H-15 N TROSY spectrum of GIRK CP was well dispersed and provided a number of signals somewhat consistent with the number of the residues within a single subunit of GIRK CP , indicating that the four subunits within the tetramer are in an equivalent environment in solution (12). The validity of this construct was confirmed by its crystal structures (7,8), which were essentially identical to the structures of the CP region in the full-length Kir channels, such as KirBac1.1 (14) and a chimeric channel of KirBac1.3 and GIRK1 (6). The purified G␤␥ contains approximately equimolar amounts of G␤ and G␥, and its binding activity to G␣ was confirmed by size exclusion chromatography (supplemental Fig. S1).
ITC experiments were carried out to determine the binding affinity and stoichiometry between GIRK CP and G␤␥. Fig. 1 shows the isotherm upon the titration of GIRK CP into the G␤␥ solution (upper panel), and the amounts of the heat exchange obtained by the integration of each peak, after the subtraction of the heat of GIRK CP dilution (lower panel). The best fit of the integrated isotherm was obtained, by using a single-site model, which generated an apparent dissociation constant (K d ) value of 250 M (Fig. 1, lower panel). Although the concentration of G␤␥ in the ITC cell (380 M) was not sufficient for the determination of the precise binding stoichiometry when the K d value is 250 M, the fitting ended with a binding stoichiometry of 0.25 eq of GIRK CP tetramer per one G␤␥ molecule. Because the 0.25 eq of GIRK CP tetramer corresponds to the concentration of the GIRK CP monomer, the ITC result suggests that four G␤␥ molecules bind to one GIRK CP tetramer, where the K d value for the binding of GIRK CP monomer to G␤␥ is 250 M.
The G␤␥ Binding Site on GIRK CP , Revealed by Transferred Cross-saturation Experiments-To identify the G␤␥ binding site on GIRK CP , TCS experiments were performed. The radio frequency irradiation to the sample containing G␤␥ resulted in selective intensity reductions of the 1 H-15 N TROSY signals of the backbone amide groups in the G␤␥ binding site ( Fig.  2A) (15,16). Our relaxation matrix calculations (17), in which on and off rates of the GIRK CP -G␤␥ interaction are also taken into account, suggested that amide protons of GIRK CP within 6 Å from G␤␥ exhibited significant intensity reductions.
The intensity reductions of the TROSY peaks upon irradiation are mainly caused by the cross-saturation from G␤␥. However, even with the highly selective saturation by IBURP2 at 0.5 to 1.5 ppm, the direct saturation of the small amount of residual protons of the highly deuterated GIRK CP and/or the exchangeable protons in the hydroxyl and thiol groups, in principle, would reduce the signal intensity of the proximal amide groups, because of the enhanced dipole-dipole interactions due to the high molecular weight of GIRK CP (nearly 100,000). To exclude the effects of the direct saturation of the GIRK CP resonances, a control experiment in the absence of G␤␥ was also performed (Fig. 2B). We evaluated the crosssaturation from G␤␥, by subtracting the intensity reduction ratios of this control experiment, which were scaled up considering the slower tumbling time of GIRK CP caused by the binding to G␤␥ (supplemental Fig. S2, gray; see the legend of supplemental Fig. S2 for details) from the ratios in the presence of G␤␥ (supplemental Fig. S2, beige) (16). The differences in the reduction ratios, ⌬RR, are shown in Fig. 2C with error bars calculated based on the signal-to-noise ratios. The minimum values of ⌬RR within the error ranges (⌬RR min ) were utilized for the evaluation.
The residues with large intensity reductions (⌬RR min Ͼ 0.20) are Arg 236 , Gly 241 , Phe 243 , and Leu 333 , and those with small but significant intensity reductions (0.083 Ͻ ⌬RR min Յ 0.20) are Gln 237 , Thr 238 , Glu 240 , Glu 242 , Leu 244 , Met 308 , Glu 320 , Glu 334 , Phe 337 , Lys 339 , Asp 341 , and Glu 350 . These residues mostly belong to one of two regions: the region from the ␤D to ␤E strands (Arg 236 -Thr 238 and Glu 240 -Leu 244 ) and the region from the ␤L to ␤M strands (Leu 333 , Glu 334 , Phe 337 , Lys 339 , and Asp 341 ) (Fig. 2C). The other three residues, Met 308 , Glu 320 , and Glu 350 , are located in the loop between the ␤H and ␤I strands, the loop between the ␤I and ␤J strands, and in the ␤N strand, respectively.
These residues were mapped on a single subunit of GIRK CP (PDB code 1N9P) (Fig. 2D). Although the mapped residues are dispersed on a single subunit of GIRK CP , the residues except for Met 308 , Glu 320 , and Glu 350 formed two adjacent contiguous surfaces in the two neighboring subunits of the GIRK CP tetramer (Fig. 2E). The two surfaces are separated by two proline residues (Pro 245 and Pro 329 ), which lack amide protons and thus were excluded from the analysis, suggesting that the two surfaces could form a single G␤␥ binding site (supplemental Fig. 3A).
The intensity reduction of Met 308 would be induced by the direct saturation of the hydrogen atoms in the hydroxyl groups of the three proximal threonine residues (Thr 305 , Thr 306 , and Thr 309 ). The distances between the nitrogen atom of Met 308 and the ␥1 oxygen atoms of Thr 305 , Thr 306 , and Thr 309 are 7.4, 3.1, and 7.0 Å, respectively, in the crystal structure. Thus, the direct saturation of the threonine hydroxyl groups would result in the large ⌬RR value of Met 308 , by decreasing the distances due to the conformational change (see below) and/or enhancing the dipole-dipole interactions upon G␤␥ binding, even though significant intensity reductions were not observed in the control experiment.
Glu 320 and Glu 350 might be included in the G␤␥ binding site. Although these residues are separated from the contiguous surfaces, the intervening residues are difficult to detect by cross-saturation. Glu 320 is separated from the clustered surface by Arg 43 and Gln 44 of the N-terminal region (supplemental Fig. S3A). Because the exchange rates for the amide protons of several residues in the N-terminal and loop regions under the TCS experimental conditions seem to be faster than those of the other regions (supplemental Fig. S3B), the intensity reductions of the signals for these regions might be underestimated. Glu 350 is separated by two phenylalanine residues (Phe 328 and Phe 349 ), and their side chain aromatic rings are oriented toward the protein surface (supplemental Fig. S3A). The amide groups of these two phenylalanine residues are buried in the protein and thus are more than 6 Å away from the contiguous surface for G␤␥ binding. There-fore, Phe 328 and Phe 349 are unlikely to be detected as binding residues by the TCS experiment, even when they are involved in the G␤␥ binding interface. Altogether, we concluded that the G␤␥ binding site on GIRK CP is composed of the following residues clustered between the two neighboring surfaces of the adjacent subunits: Arg 236 , Gln 237 , Thr 238 , Glu 240 , Gly 241 , Glu 242 , Phe 243 , Leu 244 , Glu 320 , Leu 333 , Glu 334 , Phe 337 , Lys 339 , Asp 341 , and Glu 350 .
Chemical Shift Perturbation of GIRK CP upon Binding to G␤␥-The G␤␥ binding effects on GIRK CP were also investigated by CSP experiments, which reflect the effects of the direct binding of G␤␥ as well as the conformational change of GIRK CP induced by the G␤␥ binding. Upon the addition of 50, 100, 200, and 400 M G␤␥ to 50 M uniformly 2 H, 15 Nlabeled GIRK CP tetramer, no signals exhibited chemical shift changes larger than 0.05 ppm, but significant intensity reductions were observed for a number of signals (Fig. 3A).
The observed intensity reductions of the NMR signals are caused by slower tumbling, due to the increased average molecular weight of the complex of GIRK CP and G␤␥. Furthermore, the intensity reductions are induced by the chemical shift differences of GIRK CP between the free and bound states (18). The differences reflect the changes of the microenvironments on the G␤␥ binding site on GIRK CP , including the disruption of the 4-fold symmetry of the GIRK CP tetramer, as well as the concomitant conformational changes of the remote region from the binding site.
The intensity reduction ratios of the TROSY signals are plotted in Fig. 3B for the sample containing 50 M of the 2 H, 15 N-labeled GIRK CP tetramer mixed with 100 M G␤␥, in which the GIRK CP tetramers in the free form and 1:1 complex with G␤␥ are predominantly present, assuming the K d value of 250 M obtained by the ITC analysis (supplemental Fig.  S4). In the analysis, the signal intensity of free GIRK CP was divided by a factor of 1.36, considering the increase in the averaged molecular weight of GIRK CP in complex with G␤␥. Several residues such as Gly 61 -Glu 63 , Ser 368 , Ser 369 , and Leu 371 exhibited the negative RR values lower than Ϫ0.20, suggesting that these residues tumble faster than other regions of GIRK CP , due to their local flexibility. The color representation of the label and frame corresponds to the intensity reduction ratios for each signal: higher than 0.2 (red), within the 0.1-0.2 range (orange), and lower than 0.1 (black). B, a selected portion and one-dimensional cross-sections of the 1 H-15 N TROSY spectra of the uniformly 2 H, 15 N-labeled GIRK CP , in the absence of G␤␥. The observed intensity reductions of the signals were moderate for Leu 249 and small for Phe 243 and Trp 323 in the absence of G␤␥. C, plot of the difference in the reduction ratios (⌬RR) originating from the backbone amide groups with and without irradiation between in the presence and absence of G␤␥ (see also supplemental Fig. S2). The residues indicated by asterisks were those with no data mostly due to the overlapping resonances. The error bars were calculated based on the signal-to-noise ratios. Bars corresponding to the residues with significant intensity reductions, which have the minimum values of ⌬RR within the error ranges (⌬RR min ) larger than 0.20 and those within the 0.083 to 0.20 range, are colored red and orange, respectively. The primary sequence of GIRK CP (residues 41-63 and 190 -371) is displayed in the single letter amino acid code followed by the residue number. The secondary structure elements of GIRK CP are depicted in green below the sequence, based on the crystal structure (PDB code 1N9P). D, mapping of the affected residues in the TCS experiment on a single subunit of the GIRK CP structure (PDB code 1N9P). Side views of the GIRK CP structures parallel to the membrane plane are depicted by a ribbon diagram with stick representations (left) and surface representations (center and right), viewed from the outside (left and center) or the inside of the pore (right), where the membrane side is above the molecules. All of the affected residues are colored according to the ⌬RR values, as in C. In the surface representations (center and right), the proline residues and the residues with no data are colored black and gray, respectively, because they lacked the cross-saturation data. E, mapping of the affected residues in the TCS experiment on either of the two adjacent subunits of a GIRK CP tetramer (white and cyan), viewed from the outside (left) and the inside (right) of the pore, with the same color code as in C. Schematic drawings of the GIRK CP tetramer (subunits A, B, C, and D) viewed from the membrane side are included to indicate the view of the surface models. The dashed blue line indicates the boundary between the two neighboring subunits.
Overall signal intensity reductions should be proportional to the ratio of the averaged molecular weight for the free GIRK CP and the GIRK CP -G␤␥ complex against the molecular weight of the free GIRK CP , which was calculated to be 1.36, in the same way as described under supplemental Fig. S2, by using concentrations of 200 and 100 M for GIRK CP as a monomer and G␤␥, respectively. Thus, the intensities of the free GIRK CP were divided by a scaling factor of 1.36, and then, used for the calculation of the intensity reduction ratios. The error bars were calculated based on the signal-to-noise ratios. Bars corresponding to the signals with intensity reduction ratios higher than 0.32 are colored red and those within the 0.18 -0.32 range are orange. Asterisks indicate the residues with no data, mainly due to the overlapping. C, mapping of the residues with significant intensity reductions in the CSP experiments on the surface of a single subunit of the GIRK CP structure (PDB code 1N9P), colored according to the intensity reductions, as in B. The GIRK CP structures are viewed from the side with the membrane side above, where the left and right figures are rotated by 180°about the vertical axis relative to each other. Proline residues and the residues with no data are colored black and gray, respectively. D, mapping of the residues with significant intensity reductions clustered in the interface on the outside surface of the two neighboring subunits of GIRK CP (left) and mapping of the other residues, which are not exposed to the outside surface, on both of the two neighboring subunits of GIRK CP viewed from the inside of the pore (right), with the same color scheme as in B. Schematic drawings of the GIRK CP tetramer (subunits A, B, C, and D) viewed from the membrane side are included to indicate the view of the surface models. The buried residues are labeled with dashed lines. The dashed blue line indicates the boundary between the two neighboring subunits.
The GIRK CP tetramer seems to bind to four G␤␥ molecules, and no significant affinity difference was observed among the multiple G␤␥ binding sites. The binding stoichiometry is consistent with the results of the cross-linking experiments indicating that 1-4 G␤␥ molecules bind to a tetramer of GIRK (19). The binding of 1-4 G␤␥ molecules to a GIRK tetramer might be relevant to the G␤␥-dependent multiple modes of GIRK activation, where increased G␤␥ binding seems to result in increased current frequencies (open probability) of GIRK, observed by the single-channel analyses (20). The dissociation constant between GIRK CP and G␤␥ (K d value of 250 M) determined in this study is considerably larger than the reported constant for the G␤␥-induced activation of GIRK (K act value of 11 nM) determined by electrophysiological experiments (21). Theoretically, the affinity between membrane proteins could be enhanced by 10 6 -fold as compared with that in solution, because of the effects of the excluded volume, the substantially higher local concentra-tions, and the preorientation of proteins in membranes (22). Indeed, the affinities of protein-protein interactions within membranes were reported to be enhanced by 10 1 -10 3 -fold (23,24). Because G␤␥ is anchored to the membrane by the lipid moiety covalently attached to the C terminus of G␥, the orientational constraint of G␤␥ should be smaller than those of integral membrane proteins. Therefore, the enhancement of the binding affinity in the membrane is expected to be significantly smaller than 10 6 -fold, suggesting that the K d value of the GIRK-G␤␥ interaction falls in the nanomolar to micromolar range. Considering that GIRK activation is terminated by the removal of G␤␥ from GIRK through the competitive binding of the GDP-bound G␣ to G␤␥ (IC 50 ϭ 10 nM (21)), the affinity between GIRK and G␤␥ should be lower than that between G␣ and G␤␥ (K d ϭ 6 -9 nM (25)) to accomplish the rapid closure of GIRK by the GDP-bound G␣. Therefore, the K d value of 250 M, determined in this study, would be reasonable for the interaction between GIRK and G␤␥ in solution.
The Possible Binding Mode between GIRK and G␤␥-The TCS method identified the site of the direct G␤␥ binding, which is mainly located on the region from the ␤D to ␤E strands of one subunit and the region from the ␤L to ␤M strands of the adjacent subunit of GIRK CP (Fig. 2). The latter region contains Leu 333 and Gly 336 , which were previously identified by the mutational analyses as important residues for the G␤␥-induced activation of GIRK (3,5,26). Although the other residues postulated by the mutational studies, His 57 and Leu 262 , were excluded from the present analyses, unfortunately, due to insufficient signal intensity and lack of the backbone assignment of the NMR signal, respectively.
On the other hand, the mutational analyses have suggested several residues of G␤␥ that are important for GIRK activation (residues Leu 55 , Lys 78 , Ile 80 , Lys 89 , Trp 99 , Asp 228 , Asp 246 , and Trp 332 of G␤) (27,28). It should be noted that most of these residues are located in the G␣ binding site, in the crystal structure of the heterotrimeric G␣␤␥ complex (Fig. 4A). The former four G␤ residues (Leu 55 , Lys 78 , Ile 80 , and Lys 89 ) bind to the N-terminal helix of G␣, which undergoes a large conformational change upon G␤␥ binding (29), suggesting that this site on G␣ is formed specifically in the complex with G␤␥. Meanwhile, the latter four residues (Trp 99 , Asp 228 , Asp 246 , and Trp 332 ) of G␤ reside on the "top" face of the G␤␥ torus, also known as the "hot spot" (Fig. 4A, left), which binds to the GTPase domain of G␣ (Fig. 4A, right). The center of the binding site of the G␤␥ torus on G␣ (Fig. 4A) contains the hydrophobic residues, Ile 180 , Phe 195 , and Phe 211 , which are located in the proximity of the Trp 99 side chain of G␤. These three hydrophobic residues of G␣ are flanked by two acidic residues, Glu 182 and Glu 212 , where Glu 182 hydrogen bonds with the Trp 99 side chain of G␤ and Glu 212 makes an electrostatic interaction with Lys 57 of G␤. Indeed, the importance of Trp 99 was indicated by the mutational analyses, and affected the activation of G␤␥ effectors besides GIRK, such as phospholipase C-␤2 and adenylyl cyclase 2 (30). Although G␤␥ binding affects the conformation of the GTPase domain of G␣, the structural motif comprising "the hydrophobic region" . Important residues for the interaction between GIRK CP and G␤␥. A, surface representations of G␤␥ (left) and G␣ (right) viewed from the contacting surfaces in the crystal structure of the G␣␤␥ complex (PDB code 1GOT). The residues involved in the intermolecular interaction between G␣ and G␤␥, within a distance of 4 Å in the G␣␤␥ complex structure, are colored according to their side chain properties: hydrophobic, hydrophilic, acidic, and basic residues are green, yellow, red, and blue, respectively. The interacting residues between the N-terminal helix of G␣ and G␤ are labeled in gray. The important residues of G␤␥ for the GIRK binding identified by mutational analyses are labeled with asterisks. The region of the G␤␥ binding site on G␣ (the G␤␥ binding motif), composed of the hydrophobic region flanked by two acidic residues, and the interacting surface on G␤␥ are enclosed in circles. B, mapping of the G␤␥ binding surface of GIRK, determined by TCS experiments, on the GIRK CP structure (PDB code 1N9P), are colored according to the side chain properties as in A. The similar G␤␥ binding motif to that found on the G␣ surface in A is circled and enlarged, where the residues without TCS information are also labeled in black and blue, for the residues from one subunit colored in white and the other subunit, colored cyan, respectively. The dashed magenta lines represent the edge of each subunit between two neighboring subunits. and "the two flanking acidic residues" is preserved in GDPbound G␣, in both the presence and absence of G␤␥.
A closer inspection of the G␤␥ binding sites on GIRK CP revealed a similar structural motif, the hydrophobic region consisting of the residues on the ␤E strand (Phe 243 , Leu 244 , and Pro 245 ) of one subunit and the residues at the beginning of the ␤L strand (Pro 329 and Ile 331 ) of the neighboring subunit, and the two flanking acidic residues (Glu 242 and Glu 334 ) (Fig. 4B). Thus, we conclude that the G␤␥ binding sites on GIRK possess surface properties similar to those on G␣. It is tempting to speculate that, upon binding, the Trp 99 side chain of G␤ is accommodated in the hydrophobic region supported by the interactions with the two flanking acidic residues of GIRK. The binding orientation would be possible for intact GIRK and G␤␥ that anchors to the membrane with a lipid moiety attached to the C terminus of G␥. Because the hydrophobic region spans two neighboring subunits, G␤␥ binding to this site could cause the rearrangements of the CP regions of GIRK, as evidenced by the CSPs at distant sites far away from the G␤␥ binding site (see below).
Propagation of the G␤␥-induced Rearrangements of GIRK CP to the K ϩ -ion Gate-The G␤␥ binding induced significant intensity reductions originating from CSPs not only for the residues at the direct binding site, but also for those in other regions (Fig. 3, C and D). In particular, the significant CSPs observed for the residues located at the subunit-subunit interface away from the G␤␥ binding site (Fig. 3D, right) strongly suggest that the rearrangements of the four subunits were induced by G␤␥ binding. Indeed, two distinct conformations of the CP region by the rigid body subunit rearrangements were observed in the crystal structure of a chimeric channel of KirBac1.3 and GIRK1 (PDB code 2QKS) (6). The quite recently reported structures of KirBac3.1 (31) also revealed conformational changes in the orientation of the CP region relative to the TM region (PDB codes 2WLI and 2WLJ, as the representative structures).
In the crystal structure of the chimeric channel, several residues around the K ϩ -ion gate (Phe 72 and Thr 73 in the slide helix, Ile 182 and Ser 185 in the inner TM helix, and Gln 186 and Pro 187 in the C-linker) directly interact with the residues in the loop between the ␤H and ␤I strands (H-I loop) of the CP region of GIRK1 (Val 303 , Thr 305 , Thr 306 , Met 308 , and Cys 310 ), as summarized in Fig. 5. In addition, Thr 73 (slide helix) and Lys 188 (C-linker) interact with His 222 and Arg 219 , respectively, which are in the loop between the ␤C and ␤D strands (C-D loop) of the CP region (supplemental Fig. S5A). The interactions of the residues in the H-I loop with the TM region residues including the slide helix, inner helix, and C-linker are also observed in the crystal structure of KirBac3.1, whereas more interactions were observed between the C-D loop and the slide helix in one form of the KirBac3.1 structures (supplemental Fig. S5, panels B and C).
As shown in Fig. 3, significant CSPs upon G␤␥ binding were observed for Gly 307 and Met 308 on the H-I loop, suggesting that the relative positions between the loops of the four subunits change upon G␤␥ binding, which could result in the opening of the K ϩ -ion gate. It should be noted that the residue at the top of the H-I loop, Thr 305 , provided a very weak NMR signal, and was excluded from the analysis, and no significant CSP was observed for the adjacent residue, Thr 306 , due to the lack of the interacting TM residues in GIRK CP , used in this study. The residues in the C-D loop were also excluded from the analysis, because their backbone NMR signals were not assigned. Further studies are awaited to reveal the mode of the G␤␥-induced structural rearrangements of the CP region of GIRK, and their effects on the conformation of the K ϩ -ion gate.