Gβ Residues That Do Not Interact with Gα Underlie Agonist-independent Activity of K+ Channels

Gβγ subunits interact directly and activateG protein-gated Inwardly RectifyingK + (GIRK) channels. Little is known about the identity of functionally important interactions between Gβγ and GIRK channels. We tested the effects of all mammalian Gβ subunits on channel activity and showed that whereas Gβ1−4 subunits activate heteromeric GIRK channels independently of receptor activation, Gβ5 does not. Gβ1 and Gβ5 both bind the N and C termini of the GIRK1 and GIRK4 channel subunits. Chimeric analysis between the Gβ1 and Gβ5 proteins revealed a 90-amino acid stretch that spans blades two and three of the seven-propeller structure and is required for channel activation. Within this region, eight non-conserved amino acids were critical for the activity of Gβ1, as mutation of each residue to its counterpart in Gβ5 significantly reduced the ability of Gβ1 to stimulate channel activity. In particular, mutation of residues Ser-67 and Thr-128 to the corresponding Gβ5 residues completely abolished Gβ1 stimulation of GIRK channel activity. Mapping these functionally important residues on the three-dimensional structure of Gβ1 shows that Ser-67, Ser-98, and Thr-128 are the only surface accessible residues. Gαi1 interacts with Ser-98 but not with Ser-67 and Thr-128 in the heterotrimeric Gαβγ structure. Further characterization of the three mutant proteins showed that they fold properly and interact with Gγ2. Of the three identified functionally important residues, the Ser-67 and Thr-128 Gβ mutants significantly inhibited basal currents of a channel point mutant that displays Gβγ-mediated basal but not agonist-induced currents. Our findings indicate that the presence of Gβ residues that do not interact with Gα are involved in Gβγ interactions in the absence of agonist stimulation.

Heterotrimeric GTP-binding (G) proteins are made of ␣, ␤, and ␥ subunits. Ligand activation of G protein-coupled receptors catalyzes the exchange of GDP for GTP on G␣ subunits, leading to their dissociation from G␤␥. Both the G␣-GTP and G␤␥ subunits are thus activated and capable of interacting with and stimulating downstream effector proteins. Hydrolysis of GTP yields G␣-GDP that re-associates with G␤␥ forming the inactive heterotrimer. The three-dimensional structures of G protein subunits in the active and inactive states or in complex with each other have been determined (1). Yet, the details of how G protein subunit interactions with effectors lead to signaling are poorly understood.
Acetylcholine-activated K ϩ channels were the first effectors identified for the G␤␥ subunits (2), providing a highly sensitive assay for G␤␥ signaling. Despite intensive study, little is known about the nature of the interactions between these two proteins. Acetylcholine-activated K ϩ channels have been cloned and classified as G protein-gated Inwardly Rectifying K ϩ channels (GIRK) 1 (3). To date four distinct mammalian GIRK subunits have been identified (GIRK1-4) (3-7). Heterotetrameric channels containing the GIRK1 and GIRK4 subunits form channels identical to those found in heart (4). Coexpression of GIRK heterotetramers reveals agonist-independent (basal) currents that can be stimulated further in the presence of agonist. Both basal and agonist-induced currents depend on G␤␥ subunits, as proteins that bind G␤␥ abolish all K ϩ currents (8). Interestingly, mutation of a Leu residue (GIRK4-L339, GIRK1-L333) abolished agonist-induced currents while leaving basal activity intact (8). This result has suggested that agonist-dependent and -independent currents are mediated by distinct channel-G␤␥ interactions.
A number of studies have identified channel regions that interact with G␤␥. These regions are contained within the intracellular N and C termini of the channel (8 -14). Interestingly, it has been shown that the N terminus of the channel can bind the trimeric GDP-bound G␣␤␥ complex as well (9,15). In atrial and neuronal cells, GIRK channels are activated specifically by G␤␥ subunits complexed to pertussis toxin-sensitive G␣ subunits. However, their activation by receptors coupled to G␣ s has been reported in some heterologous expression systems (16,17), but not in others (18). It has been suggested that G␤␥ activation of GIRK channels can be mediated by several different G␤␥ subunit combinations suggesting a promiscuous interaction between G␤ and GIRK channels (19). This assertion was based on the activity of two G␤ (G␤1 and G␤2) subunits in combination with four different G␥ subunits (G␥1, G␥2, G␥5, and G␥7). Complete promiscuity of G␤␥ subunits leading to activation of GIRK channels would exclude the large number of distinct G␤␥ complexes as possible contributors to G␤␥-specific signaling. Lei and colleagues (20) have recently reported that although G␤5 subunits bound GIRK channels, as other mammalian G␤ subunits did, they failed to stimulate channel activity. Possible differences between G␤1 and G␤5 accounting for their distinct effects remain to be elucidated.
Since G␣ subunits compete with effectors for interactions with G␤␥ subunits, Ford et al. (21) identified specific G␣ contact points on G␤ critical for activation of different effectors, including GIRK channels. Furthermore, Albsoul-Younes and colleagues (22) showed that G␤ chimeras that incorporated some residues from Saccharomyces cerevisiae G␤ displayed reduced activity in stimulating GIRK channels. Some of these regions were G␣-independent suggesting a role for multiple regions of G␤␥ for channel activation. However, the significance of G␣-independent interactions with the channel was not further explored.
In the present study, we have tested the interaction of all mammalian G␤ subunits with GIRK channels. These channels provide an excellent model for effector-G␤␥ interactions and specificity of G␤␥ signaling. By comparing G␤1 and G␤5 subunits that exhibit differences in their abilities to regulate GIRK currents, we have identified several G␤ residues important for activation of GIRK channels. These amino acid residues map onto different surfaces of the G␤ three-dimensional structure, suggesting multiple interactions between the channel and G␤. Interestingly, these residues include not only G␣ contact points, but in addition, amino acids that do not interact with G␣ subunits. Residues that do not interact with G␣ subunits were found to be important for the agonist-independent or basal activity of these channels. These results identify for the first time functionally significant G␤ residues that are independent of interactions with G␣. EXPERIMENTAL PROCEDURES cDNAs, Mutant, and Chimeric Constructs-All the cDNAs used were subcloned into pGEMHE (23) to accommodate sufficient expression in oocytes. Chimeric constructs were made by the method of Splice by Overlap Extension as described previously (8). The PCR reactions were performed using Vent polymerase (New England Biolabs, Beverly, MA) for only 12 cycles to reduce undesirable mutations. All the chimeric constructs were confirmed by sequencing. The mutants were made using the QuikChange method (Stratagene, La Jolla, CA) using high fidelity Pfu polymerase for 12-16 cycles only. All the mutants were confirmed by DNA sequencing.
Expression in Oocytes-cDNA constructs were linearized and subjected to in vitro transcription using the mMessage mMachine kit (Ambion, Austin, TX). The resulting cRNAs were quantified by comparison of two dilutions to a standard on a formaldehyde gel. Oocytes were isolated from Xenopus laevis frogs, enzymatically digested with collagenase and incubated in ND-96 solution containing calcium and nutrients (6). Oocytes were injected with cRNAs, 2 ng each of GIRK1/GIRK4, 1 ng of hM 2 and 2 ng of each G protein subunit or chimera. Following injection, oocytes were kept for 48 -96 h in an 18°C incubator before recordings.
Electrophysiology-Two-electrode voltage clamp recordings were carried out as previously described (8). Briefly, the oocytes were placed in a chamber and perfused with a solution containing high potassium (96 mM). Currents were recorded using a voltage step protocol from Ϫ100 to ϩ50 mV (control current). Barium (3 mM) was used to measure the left over current that was not inwardly rectifying (barium-insensitive current). We determined the basal current by subtracting barium-insensitive current from the control current.
Immunoblot Analysis-Oocytes were homogenized in a Dounce homogenizer in lysis buffer with protease inhibitors on ice and centrifuged for 5 min at 3000 ϫ g. The supernatant was centrifuged using a Beckman-ultracentrifuge at 100,000 ϫ g for 30 min. The supernatant was discarded, and the pellet was resuspended in lysis buffer.
Equal amounts of protein were loaded onto the gels based on estimations from Coomassie Blue staining for each sample. For immunoblotting, samples were boiled in equal volume of loading buffer and run on a 12% polyacrylamide gel. The proteins were transferred onto nitrocellulose paper using a semi-dry transfer apparatus from Bio-Rad. Nonspecific sites were blocked overnight using a blocking buffer containing 5% non-fat dry milk. The N terminus of G␤5 was tagged with a FLAG epitope and a mouse monoclonal antibody (Sigma) was used for detection in the immunoblot. For the mutants, a polyclonal antibody to the common region of G␤ was used (Santa Cruz Biotechnology, Santa Cruz, CA). The nitrocellulose membranes were incubated with the proper antibody in the blocking solution for 1 h. The filters were washed and incubated with a peroxidase-conjugated secondary in the blocking buffer for 1 h. The blots were washed extensively and then subjected to detection using the enhanced chemiluminescence method. Mutants from three different batches of oocytes were tested for expression level.
G␤␥ Purification-The G proteins used for trypsin digestion experiments were generated using the BaculoGold Expression System (PharMingen) and were purified by coexpression of a hexahistidinetagged ␥2 according to Kozasa and Gilman (24) except that Lubrol was replaced with CHAPS in the final washes. Final protein concentrations were estimated against known amount of bovine serum albumin standard on SDS-PAGE following Coomassie staining.
GST-GIRK Pulldown Assays-GST-GIRK1 C-terminal domain (GIRK1, amino acids 182-501), GST-GIRK1 N-terminal domain (amino acids 1-86), GST-GIRK4 C-terminal domain (GIRK4, amino acids 186 -419), or GST-GIRK4 N-terminal domain (amino acids 1-91) fusion proteins were used in an in vitro protein interaction assay as described previously (17) with the following modifications. Interaction assays were performed using solubilized membranes from Sf9 cells infected with recombinant baculoviruses. Confluent cells were infected for 48 h with viruses encoding either G␤1␥2 or G␤5␥2. Membranes were prepared and solubilized with 0.3% ␤-dodecyl maltoside (25). For the interaction assay, solubilized membranes were brought up to 0.01% Lubrol-PX, and the assays were performed in phosphate-buffered saline with 0.01% Lubrol-PX. Purified G␤1␥2 was used at different concentrations as a standard to determine the approximate concentration of G␤␥ in Sf9 membrane preparations and as a positive control in the GST pulldown assay. GST fusion proteins (200 nM) were incubated with solubilized Sf9 membranes at an estimated G␤␥ concentration of 30 nM for 30 min at 4°C with gentle agitation. Glutathione-4B-Sepharose beads (40 l) were subsequently added to the reaction and incubated for an additional 1-2 h at 4°C. The beads were washed with three volumes of phosphate-buffered saline with 0.06% Lubrol-PX and eluted with SDS loading buffer. SDS-PAGE and Western blotting were performed as described (17,25) using the common anti-G␤ (T-20 antibody raised against the C terminus, Santa-Cruz Biotech) and anti-GST antibodies (Amersham Biosciences, Inc.) and developed by ECL (Amersham Biosciences, Inc.).

G␤5␥2 Inhibition of Basal GIRK Currents-
We used the oocyte expression system to test the ability of different G␤ subunits to activate GIRK channels. The GIRK1/GIRK4 heteromeric channel was expressed in oocytes. Fig. 1A shows representative recordings using two-electrode voltage clamp in oocytes expressing GIRK1/GIRK4 channels. Oocytes were held at 0 mV, and currents were recorded in response to voltage steps from Ϫ100 mV to ϩ50 mV in bathing solutions containing potassium concentrations similar to those found intracellularly. Similar experiments in oocytes from the same frog that expressed G␤1␥2 or G␤5␥2 subunits in addition to GIRK1/ GIRK4 are shown in the middle and lower traces of Fig. 1A, respectively.
Expression of G␤1␥2 subunits significantly enhanced GIRK1/GIRK4 channel activity (middle trace), whereas G␤5␥2 did not (lower trace). Fig. 1B summarizes these results in current-voltage plots from the same batch of oocytes expressing GIRK1/GIRK4 alone, with G␤1␥2, or with G␤5␥2. Fig. 1C shows the effectiveness of different G␤ subunits co-expressed with G␥2 to enhance GIRK1/GIRK4 currents in oocytes. For comparison, currents at Ϫ80 mV were normalized to the levels of GIRK1/GIRK4 expressed alone. In the presence of G␥2, G␤1Ϫ4 enhanced GIRK1/GIRK4 basal currents significantly. G␤5␥2, on the other hand, showed a reduction of the basal currents when co-expressed with GIRK1/GIRK4. To confirm that G␤5 was expressed in oocytes, we tagged the G␤5 subunits with the FLAG-epitope. The FLAG-tagged G␤5 reduced basal currents similarly to the non-tagged G␤5 (data not shown). Western blot analysis (Fig. 1D) showed that the G␤5 subunit is expressed in oocytes, and therefore lack of expression cannot explain its inability to activate GIRK1/GIRK4 channels.
To address the question of whether the G␤5 effect on GIRK channel activity could be the result of direct interactions of this G␤ subunit with the channel, we used a biochemical-binding assay. GST fusion proteins of the N-and C-terminal domains of GIRK1 or GIRK4 were purified and pulldown assays were performed with solubilized membranes from Sf9 cells expressing G␤1␥2 or G␤5␥2 combinations. Fig. 1E shows that 30 nM G␤1␥2 or G␤5␥2 both bound the N-and C-terminal domains of the GIRK subunits. These results are in close agreement with those obtained recently by Lei et al. (20), showing that purified G␤5␥2 interacted directly with the GIRK heteromer. It is likely that the effect of G␤5␥2 on basal currents is the result of direct binding and competition with endogenous G␤ for binding sites on the channel. We next proceeded to identify critical functional regions in G␤1 that are important for activation of GIRK channels.
Identification of the G␤ Domain Required for Channel Activation-Chimeric constructs between G␤1 and G␤5 were de-signed to identify a minimal region of G␤1 that when grafted into the background of G␤5 would activate the channel ( Fig.  2A). These G␤ chimeras were designed using an alignment based on the known structural motifs of G␤ (26,27). cRNAs of chimeric constructs G␤1/␤5 along with GIRK channels and G␥2 were co-injected into oocytes. Wild-type G␤1␥2 and G␤5␥2 were tested each time on the same batch of oocytes. Fig. 2B shows basal oocyte currents expressed as a percentage of GIRK current achieved by co-expression of wild-type G␤1␥2. Results from oocytes injected with channel only, wildtype G␤1␥2, and G␤5␥2 are shown for comparison. The chimera ␤1(1-142)␤5 (amino acids 1-142 of G␤1 and 152-353 of G␤5) activated the channel to a similar extent as the wild-type G␤1, indicating that the N-terminal 142 amino acids of G␤1 harbored the critical differences between G␤1 and G␤5 in activating GIRK currents. Chimera ␤5␤1(52-340) (amino acids 52-340 of G␤1 and 1-59 of G␤5) also activated GIRK1/GIRK4 similarly to the wild-type G␤1, suggesting that the 51-N-terminal-residue differences were not important for GIRK activa-FIG. 1. Activation of GIRK channels by different G␤ subunits. A, currents from oocytes expressing GIRK1/GIRK4. The membrane potential was held at 0 mV, and currents were recorded in symmetrical potassium solutions. Currents were measured using a voltage step protocol from Ϫ100 to ϩ50 mV in 10 mV steps. Upper trace, inwardly rectifying basal K ϩ currents (barium-sensitive currents) from oocytes expressing GIRK1/GIRK4 channels. Middle trace, same conditions as upper traces except oocytes also express G␤1␥2 in addition to the channel subunits. The current is stimulated under these conditions compared with control oocytes. Lower trace, same as upper traces except oocytes also expressed G␤5␥2 in addition to the channel subunits. The trace shows reduced currents in the presence of G␤5␥2. B, current-voltage plots from a batch of oocytes expressing GIRK1/GIRK4 (squares, n ϭ 4), GIRK1/GIRK4 and G␤1␥2 (circles, n ϭ 4), and GIRK1/GIRK4, G␤5␥2 (triangles, n ϭ 4). C, currents at Ϫ80 mV in oocytes expressing GIRK1/GIRK4 in the presence of G␥2 and different G␤ subunits. Data are expressed as percentage of the control GIRK1/GIRK4 current. Expression of G␤1-4 all enhance currents significantly in comparison to GIRK1/GIRK4 alone (G␤1-3: *, p Ͻ 0.01, unpaired t test, n ϭ 8 -20; G␤4: #, p Ͻ 0.05, unpaired t test, n ϭ 4). Expression of G␤5 reduced channel activity significantly (*, p Ͻ 0.01, unpaired t test, n ϭ 19). D, immunoblot showing expression of the FLAG-tagged G␤5 in oocytes. E, both G␤1 and G␤5 bind the N and C termini of GIRK channels. GST fusion proteins were purified using glutathione 4B-Sepharose beads. Purified GST fusion proteins were incubated with membranes from Sf9 cells expressing mammalian G protein ␤1␥2 and ␤5␥2 subunits (30 nM) and incubated for 30 min. Glutathione-Sepharose beads were added and incubated for 2 h. GST was used as negative control. The beads were washed extensively, and bound proteins were eluted with SDS loading buffer and separated by SDS-PAGE. G␤ was detected using the common anti-G␤ antibody. Data shown are representative of three independent experiments. tion. The chimera ␤5␤1(52-142)␤5 (amino acids 1-59 of G␤5, 52-142 of G␤1, and 152-353 of G␤5) activated GIRK1/GIRK4 channels when expressed in oocytes. This chimera contained a minimal region of G␤1 that when inserted into G␤5 caused GIRK channel activation. Fig. 2C shows the position of this region on the crystal structure of G␤1␥2 as adapted from Wall et al. (26). The white colored areas identify the region of G␤1 between residues 52-142 that includes predominantly blades two and three in the seven-propeller structure. Fig. 2D shows an alignment of this region between G␤1 and G␤5. Of the 90 amino acids contained within this region, 51% are identical between the two proteins. These data suggest that specific residues of G␤1 within this region are critical for the ability to activate GIRK channels.
Role of G␣ Contact Points of G␤1 in Channel Activation-Recently, Ford et al. (21) identified which G␣ contact points on G␤ were important in activating distinct effectors. Each of the G␣ contact points on G␤1 (as identified from the crystal structure) was mutated to an alanine, and the mutant tested for its ability to activate several effectors, including GIRK channels. In these studies GIRK1/GIRK4 were co-expressed in oocytes with G␥2 and wild-type or mutant G␤1 subunits and were tested for effects on basal current levels. Substitution by alanine of any of six G␤1 residues contacting G␣ reduced the ability of G␤1 to activate the channel. Five of the six amino acids are contained within the G␤1(52-142) region. We compared the ability to activate GIRK1/GIRK4 of each mutant versus that of wild-type G␤1␥2 in the same batch of oocytes. The G␤1 mutants K78A, I80A, K89A, and W99A, all of which are conserved in G␤5, failed to activate the channel in agreement with the reported results by Ford et al. (21) (37 Ϯ 7%, 40% Ϯ 5%, 42% Ϯ 5%, and 41 Ϯ 2% of wild-type G␤1␥2). However, in our hands, L55A, the only one of the six G␣-contacting residues identified by Ford et al. that was not conserved between G␤1 and G␤5, showed enhancement of GIRK channel activity that was not significantly different from that with wild-type G␤1␥2 (76 Ϯ 16% of wild-type G␤1␥2, p Ͼ 0.2, n ϭ 10). Therefore, the residues involved in G␣ interactions cannot account for the difference between G␤1 and G␤5 in their ability to stimulate channel activity. Thus, we proceeded to identify specific residues in G␤1 that are important for channel activation based on the differences between G␤1 and G␤5.
G␤1 Residues Critical for Activation of GIRK Channels-To identify specific residues within the G␤1 (52-142) region, The minimal region of G␤1 that was grafted into G␤5 and caused stimulation of GIRK1/GIRK4 currents is identified between residues 52 and 142 of G␤1. C, depiction of the seven-propeller structure of G␤1 based on the published coordinates (26). Blades two and three are shown in white, the rest of the protein is in black. The critical region is almost entirely contained within blades two and three. D, alignment of G␤1 and G␤5 in the identified region. From the 90 amino acids, 46 are identical between the two proteins. The alignment is based on the published structural motifs of G␤1 (26,27). which are responsible for the difference in stimulation of GIRK currents by G␤1 and G␤5, we mutated each G␤1 residue to its corresponding G␤5 counterpart and tested its ability to stimulate GIRK currents in oocytes. Each mutant was co-expressed with GIRK1/GIRK4 and G␥2. The ability of each mutant to stimulate channel activity was expressed as a percentage of that achieved by the wild-type G␤1␥2 that was tested in parallel. We screened for mutations that significantly (p Ͻ 0.01) impaired the ability of G␤1 to stimulate GIRK activity. Fig. 3A shows the results from 34 of the 44 mutants tested. These mutants enhanced GIRK1/GIRK4 currents to levels that were not significantly different from the wild-type G␤1␥2 (p Ͼ 0.01 unpaired t test, n ϭ 6 -14). Fig. 3B shows a different set of mutants that showed significant reduction in their ability to enhance GIRK1/GIRK4 cur-rents compared with the wild-type G␤1␥2 (p Ͻ 0.01, unpaired t test, n ϭ 12-24). The lower panel of Fig. 3B shows a representative Western blot from oocytes expressing each of these mutants (n ϭ 3). An antibody raised against a common G␤ region (M-14, Santa Cruz Biotechnology) was used for immunoblotting. This antibody recognized the endogenous oocyte G␤ subunits (two left lanes) in oocytes that were either non-injected or were injected with channel subunits alone. A larger band corresponding to the exogenous G␤ can only be seen in oocytes that were injected with either the wild-type or the mutant G␤1 subunit. Thus, expression levels of the mutant G␤1 subunits with significantly impaired ability to stimulate GIRK1/GIRK4 currents were similar to the wild-type G␤1. Fig.  3C shows results from two additional mutants that failed to activate GIRK channels. S67K and T128F showed no enhance-FIG. 3. Critical G␤1 residues for activation of GIRK channels. Non-conserved G␤1 residues contained between residues 52 and 142 were mutated to the corresponding residues in G␤5 and tested for their ability to activate the channel. A, 34 of the mutations showed current enhancement not significantly different (p Ͼ 0.01, n ϭ 6 -14) from the wild-type G␤1. Mutations L95M, R96P, and N132M showed reduced enhancement as compared with the wild-type G␤1; however, this reduction was not present in all batches of oocytes. Statistically, these mutants are different from the wild-type G␤1 at the p Ͻ 0.05 level but not p Ͻ 0.01. However, due to the variability seen with these mutants, they were not grouped with the negative mutants. B, mutants A73S, L79V, I93V, S98T, T102A, and I123V showed significantly reduced enhancement of GIRK1/GIRK4 basal current as compared with the wild-type G␤1␥2 (*, p Ͻ 0.01, unpaired t test, n ϭ 12-24). Lower panel shows a representative immunoblot indicating that these mutants were expressed in levels comparable with the wild-type G␤1; therefore differences in expression levels cannot account for their reduced effectiveness. The left two lanes in the blot are from uninjected oocytes or oocytes that were injected with only the channel subunits. A band at ϳ36 kDa corresponding to the endogenous oocyte G␤ was detected in these oocytes. In lanes that are loaded with proteins from oocytes expressing exogenous G␤1 an additional band was detected at about 39 kDa. C, mutants S67K and T128F failed to enhance basal currents of GIRK1/GIRK4 channels compared with the wild-type G␤1 (*, p Ͻ 0.01, unpaired t test, n ϭ 12). In addition, the S67K mutant significantly reduced basal GIRK1/GIRK4 currents (#, p Ͻ 0.01, unpaired t test, n ϭ 12). Immunoblot analysis (lower panel) showed that both mutants were expressed in levels comparable with the wild-type G␤1. ment of the channel activity as compared with the wild-type G␤1␥2 (p Ͻ 0.01 unpaired t test, n ϭ 12). Furthermore, oocytes expressing the S67K mutant showed significantly reduced currents compared with oocytes that only expressed the channel subunits (p Ͻ 0.01, unpaired t test). Again, Western blot analysis indicated expression levels for these two mutants comparable with the wild-type G␤1 (n ϭ 3). Two additional mutants, L139V and G141M, also did not enhance channel activity. However, we could not detect expression of these mutants in oocytes using Western blot analysis. Therefore, these two mutants may result in conformational changes in the G␤1 structure that lead to degradation or misfolding preventing antibody detection.
Overall, these data identify several residues in blades two and three of G␤1 that are involved in mediating activation of GIRK channels. These residues differ between G␤1 and G␤5 and appear responsible for the failure of G␤5 to stimulate GIRK currents. However, mutation of these residues in G␤5 to the corresponding G␤1 residues did not affect the functional effects of G␤5 on the channel (data not shown). This suggests that these individual residues are necessary in G␤1 for functional interactions but they are not sufficient in mediating channel activation by G␤5. Since our chimeric design focused on important functional differences between G␤1 and G␤5, conserved residues involved in interactions with the channel were not probed.
Mapping of Critical G␤-Effector Interacting Residues on the Three-dimensional Structure of G␤1␥2-To gain a better understanding of the nature of the interactions between the channel and G␤1, we mapped the residues of G␤1 that are critical for functional interactions with the GIRK channel onto the known crystal structure of the G␣ i 1␤1␥2 complex (26). Fig. 4A shows a ribbon diagram of the G␤1␥2 complex adapted from the published coordinates (PDB IGP2). Blades two and three are shown in pink and green with the rest of G␤1 in white, G␥2 is shown in yellow. Residues Ser-67 and Thr-128, which did not enhance channel activity upon mutation to their G␤5 counterparts, are colored red. The six residues that, when mutated, showed partial but significantly impaired channel activation are shown in blue. The six residues in blue all cluster in the same region of the protein, whereas the two residues in red are distant from the others. Furthermore, residues Ser-67, Ser-98, and Thr-128 are localized in loops, while the rest of the critical residues FIG. 4. Localization of the critical G␤ residues on the protein structure. Three-dimensional structure of the G protein complex was constructed from the published coordinates (26). A, ribbon model of the G␤1␥2 complex rendered from the coordinates using MolScript (28). G␤1 is depicted in white with G␥2 in yellow. Blades two and three that contain the minimal activation domain of G␤1 are in pink and green. The residues identified to be important in channel activation by G␤1 are colored in red or blue. Mutation of Ser-67 or Thr-128 (red) caused G␤1 to fail to enhance GIRK1/GIRK4 currents. Mutation of Ala-73, Leu-79, Ile-93, Ser-98, Thr-102, or Ile-123 (blue) caused a significant reduction in the ability of G␤1 to stimulate GIRK1/GIRK4 currents. The left panel depicts the complex as it would be positioned in the cell with the C terminus of G␥ that is lipid-modified (prenylated) on top. The right panel is in the same orientation but rotated clockwise by 90°around the y axis. The residues depicted in blue are all clustered in the same region while the residues in red are distant from the others. Additionally, residues Ser-67, Ser-98, and Thr-128 are localized in loops while the remaining residues are found in ␤ sheets. B, similar depiction as in A with the surface of the G␤1␥2 built onto the figure from the published coordinates (26) using GRASP (29). G␤1 is in white, and G␥2 is in yellow. All the residues are color coded as in A. Here Ser-67 and Thr-128 are apparent in the left panel, Thr-128 and Ser-98 are apparent in the right panel. Only these three residues are surface-accessible in the structure. Additionally, these residues are positioned almost orthogonal to one another placing them on three distinct faces of the protein. C, same as B except for addition of the G␣ i 1 in green. Addition of G␣ i 1 to the complex does not cover Ser-67 or Thr-128, but partially covers Ser-98. Ser-67 and Thr-128 could be readily available for intermolecular interactions in the presence of G␣ i 1.
are in ␤ sheets. The right panel of Fig. 4A shows a view of the same structure rotated clockwise around the y axis by 90°. Fig.  4B is a surface model of Fig. 4A, using the same coloring scheme. In this view, residues Ser-67 and Thr-128 are apparent on the surface (left). Following the 90°clockwise rotation around the y axis (4B, right), residue Ser-98 can be seen along with Thr-128, while Ser-67 has been rotated to the back face of the structure. Mapping the eight functionally important residues onto the G␤1␥2 structure clearly indicates that only three residues are surface-accessible. Furthermore, these three residues are positioned on three different faces of the G protein suggesting that they interact with distinct parts of the channel. In Fig. 4C G␣ i 1 (in green) has been added to the complex. In the trimeric G␣␤␥ depicted here, residues Ser-67 and Thr-128 reside at a distance from G␣ i 1, while Ser-98 is partially covered by G␣ i 1.
Structural Fidelity of Mutant Proteins-We determined the structural fidelity of the mutant proteins by first testing for their interaction with G␥2. Association of G␥ with G␤ partially protected it from tryptic digestion (30,31). We used this property to test for interactions between mutant G␤1 proteins and G␥2. The three mutants (S67K, S98T, and T128F) as well as wild-type G␤1 were expressed in Sf9 cells and purified using a co-expressed hexahistidine-tagged G␥2. Purification of all G␤ proteins by their associated hexahistidine-tagged G␥2 indicated that the two subunits interacted. Furthermore, purified proteins were subjected to digestion by trypsin, and both treated and untreated proteins were detected by immunoblotting using an antibody against a C-terminal epitope in G␤ (T-20 antibody, Santa Cruz). Fig. 5 shows the results from these experiments.
Whereas in the untreated samples the 36-kDa band corresponding to the predicted molecular mass for G␤1 is detected, in samples treated with trypsin a protected 27-kDa band is detected. This G␥-protected band indicated proper interactions between G␥2 and the three tested mutants (30,31).
The Ser-67 and Thr-128 Residues of G␤1 That Do Not Interact with G␣ Subunits Are Involved in the Generation of Agonistindependent Channel Activity-We have previously shown that a Leu residue in GIRK1 and GIRK4 controls their agonistinduced activity (8). In that study, co-expression of mutants GIRK1(L333E) and GIRK4(L339E) in oocytes displayed significant agonist-independent K ϩ currents. The G␤␥ subunits were responsible for generation of these basal currents, since coexpression of G␤␥ scavenger, such as ␤ARK-PH or G␣ subunits, inhibited basal currents (8). However, the basal currents for GIRK1(L333E)/GIRK4(L339E) could not be enhanced by exogenous G␤␥ (8).
Since the mutant channels lack agonist-induced currents, reduction of their agonist-independent currents by S67K,T128F suggest that these G␤1␥2 mutants selectively affect basal currents. These data suggest that the Ser-67 and Thr-128 residues, which are not in contact with G␣ subunits, are involved in the G␤␥ mediation of agonist-independent K ϩ currents.

DISCUSSION
Four distinct G␤ subunits (G␤1-4) in complex with G␥2 activate receptor-independent GIRK channel activity, while G␤5 does not. G␤5 binds both the N and C termini of the GIRK subunits. Lack of channel activation by G␤5 suggests that some specificity of signaling could arise from the appropriate G␤ subunits employed by different G protein-coupled receptors.
Based on differences between G␤1 and G␤5, we have identified a minimal region on G␤1 that is required for stimulation of GIRK channel activity. Within this region we identified two different classes of functionally important residues. Mutation of six residues that are clustered in the same area of the G␤ structure significantly reduced the effectiveness of G␤1 in stimulating GIRK channel activity. Another two mutants, S67K and T128F, abolished G␤ stimulation of GIRK currents completely. Furthermore, S67K reduced basal GIRK currents in oocytes. All eight of the functionally relevant mutants were expressed in levels comparable with the wild-type G␤␥, thus excluding the possibility that alterations in expression were the cause of the lack of G␤ effects.
A previous study has shown that several G␣ contact points on G␤1 are crucial in interactions with a number of effectors, including GIRK (21). Five of the six G␣ contact points reported reside in the region that we have identified. All these residues are localized on the surface of G␤. Residue Trp-99 interacts FIG. 5. All G␤1 mutants interact with G␥2. Wild-type and mutant G␤1 were purified from Sf9 cells using a coexpressed hexahistidinetagged ␥2 according to Kozasa and Gilman (24). Purified proteins were digested by trypsin for 30 min, and both treated and untreated samples were separated by SDS-PAGE. G␤1 was detected using the common anti-G␤ antibody. After trypsin digestion, a 27-kDa band is visible that corresponds to the protected C terminus of G␤1. This blot is representative of two independent experiments. The reduced signal in S67K is due to loading of smaller amount of protein. with the switch II region of G␣ i 1, and the other four interact with the ␣ helical N terminus. We have shown that four of these five residues play an important role in stimulating channel activity. However, these four residues are conserved between G␤1 and G␤5 and therefore cannot account for the differences we observe between the two G protein subunits.
Wall et al. (32) have identified four G␣ interaction sites in the 52-142 region of G␤1 that had not been previously tested for effector interactions and are not conserved in G␤5 (Arg-52, Tyr-59, Val-90, and Asn-132 in G␤1). The data presented here show that mutation of each of these G␤1 residues to its G␤5 counterpart did not significantly affect the ability of G␤1 to enhance GIRK channel activity. Furthermore, tyrosine 85 of G␤1 is the only residue in this region that is in contact with G␥2 and is not conserved between G␤1 and G␤5. Again, our data indicate that mutation of tyrosine 85 to phenylalanine (the corresponding G␤5 residue) did not affect its ability to enhance GIRK channel activity. A recent study showed that chimeras that replaced parts of the G␤1 with S. Cerevisiae-G␤ (STE4) did not activate GIRK channels (22). Based on these chimeras, two regions were identified on G␤1 that are important for GIRK channel activity. Both of these regions are contained in the outer strands of blades 2 and 3 in the sevenpropeller structure. However, only two of the eight identified residues in our study reside in these regions, and one is conserved between STE4 and G␤1.
G protein signaling proceeds following dissociation of G␣ and G␤␥ upon receptor activation. The sites required for agonistinduced activation of effectors by G␣ or G␤␥ are likely not to be accessible in the GDP-bound G␣␤␥ trimer, thus necessitating at least partial dissociation of the complex for activation of downstream effectors. From the eight functionally important residues that we have identified, only three are surface-accessible, and two of these three do not interact with G␣. We have shown that the two residues Ser-67 and Thr-128 that do not interact with G␣ subunits are critically involved in the generation of agonist-independent K ϩ currents. Of these two residues, mutation of ␤1 (Ser-67) to the corresponding Lys residue in G␤1 mimics most closely the G␤5␥2 inhibition of basal currents. However, reverse mutation of this residue in G␤5 was not sufficient to confer channel activation to G␤5.
Biochemical evidence has suggested that the G protein heterotrimer binds the N terminus of the channel via the G␤␥ subunits (9,15). Overall, the location of the critical residues we identified on the three-dimensional structure of G␤1 is consistent with biochemical evidence of multiple interaction points between the channel and G␤1 (9,12,14,33). We have localized the three surface-accessible and functionally important residues on three orthogonal faces of the G␤1 subunit. Since both the N and C termini of the channel bind G␤1, we postulate that they bind distinct surfaces of the G␤1 protein. In this scheme, the N terminus would bind the surface of G␤1 that is not covered by G␣ to allow for G␣␤␥ binding. When the G protein complex is activated, G␣ dissociates from G␤␥, exposing sites previously occupied by G␣ that can now efficiently interact with the channel C terminus and stimulate channel activity. This model is consistent with the requirement of both N and C termini for channel activation by the G␤␥ subunits (10). Multiple interaction sites of G␤␥ have been found with other effectors, such as phosducin and PLC␤ (34 -36). Our two identified G␣-independent residues are candidates for mediating the interactions with the heterotrimer. Such interactions may act as anchors between G proteins and their effectors serving to increase the efficiency and/or specificity of signaling. These residues may form contact points with the channel or they may alter channel interaction sites in their proximity. Our data further suggest that categorization of such contacts as those shared by G␣ subunits from those that are not may be of functional significance (also see Ref. 8). Mutation of the G␣independent residues abolished the ability of exogenous G␤␥ to stimulate channel activity, while mutation of G␣-shared residues significantly attenuated the responses to exogenous G␤␥. This points to an effective design in G protein signaling in which G␤␥ subunits in their inactive state (bound to G␣) can find effectors but can only signal efficiently once the G␣ is dissociated. The functional significance of such a design for G␤␥ signaling remains to be further elucidated.