Specific regions of heteromeric subunits involved in enhancement of G protein-gated K+ channel activity.

Heterologous coexpression of recombinant, G protein-gated, inwardly rectifying K+ (GIRK) channel subunits has yielded large currents, severalfold greater than those obtained from expression of the individual subunits. Such current enhancement has been obtained from coexpression of the inactive GIRK1 subunit with the low activity GIRK2-5 subunits in Xenopus oocytes. Using deletion and chimeric constructs, we now report the identification of a C-terminal region unique to GIRK1 and a larger central region of GIRK4 highly homologous to GIRK1, both of which are critical for production of large currents. Chimeras containing these two regions produced homomeric channels, exhibiting currents severalfold greater than those from either wild-type subunit alone. G protein regulation of such chimeric channel currents resembled that of wild-type currents. Green fluorescent protein-tagged channels showed that the amount of chimeric channel expressed on the oocyte cell surface was similar to its wild-type counterpart, suggesting that the enhanced activity was not due to differences in relative levels of expression but rather to the coexistence of the chimeric regions. Single-channel recordings of the active chimeras exhibited patterns of activities with open-time kinetics and conductance characteristics representative of those of GIRK4, indicating that the presence of the GIRK1 C-terminal region caused an increase in the frequency of channel openings without affecting their duration.

G protein-gated inwardly rectifying K ϩ (GIRK) 1 channels use a membrane delimited mechanism to provide cardiac, neuronal, and endocrine cells with rapid inhibition of membrane excitability in response to extracellular signals. The best studied case, the atrial ACh-induced K ϩ channel, has been shown to be directly activated by the ␤␥ subunits of G proteins in response to the binding of acetylcholine to the muscarinic m2 receptor (1). Similar GIRK channels have also been found in neuronal tissues (2). Five homologous members, GIRK1 through GIRK5, have been identified in this channel subfamily (3)(4)(5)(6)(7). Heterologous expression of individual GIRK members has produced relatively small currents with short-lived singlechannel events (6). Recent studies from coimmunoprecipitation (6,8) immunocytochemistry (9), in situ hybridization (10 -12), and functional coexpression (6 -8, 13-17) have demonstrated the heteromeric nature of GIRK channels. The GIRK1 subunit has been found to greatly enhance the activity and change the single-channel kinetics of other GIRK members, even though when expressed alone it fails to form functional homomeric channels (7). Recently, we identified a single P-region residue of GIRK1, Phe 137 , which when associated with other GIRK subunits contributed to the current enhancement and prolonged the duration of single-channel openings (17). However, additional determinants, other than Phe 137 , were implicated for the high activity of heteromeric channel currents (17).
We set out to identify such additional structural determinants of activity using the human GIRK1 and KGP (or GIRK4) channel subunits and monitoring their ability to cause current enhancement as a result of deletions and chimeric manipulations. In this study, we report that both the unique C terminus of GIRK1 and a homologous core region of KGP are involved in generation of enhanced currents. We have engineered chimeras containing these critical regions, which are capable of producing homomeric channels exhibiting large currents. These results revealed that specific regions of the two channel subunits are involved in producing large currents. Identification of these critical regions is an important step in elucidating how the two subunits interact to give rise to the highly active heteromeric channel.

Construction of Chimeric and Deletion Channels-GIRK1 and KGP
were each subcloned in pGEMHE and were used as template for PCR (8,18). The K 1 K 2 G 3 chimera (see Fig. 1 for nomenclature) was constructed using a three-fragment ligation reaction. The K 1 K 2 fragment was amplified by PCR with BamHI and NaeI sites incorporated in the 5Ј and 3Ј ends, respectively. Similarly, the G 3 fragment was amplified with DraI and EcoRI sites incorporated at the ends. Both PCR fragments were subcloned into the region between BamHI and EcoRI sites of pGEMHE to create the full-length chimera. Restriction digestion with SacI yielded a fragment spanning the junction between K 1 K 2 and G 3 regions and expanding through the stop codon (one SacI site was located within the codons for amino acids 319 -321 of KGP, whereas the other was in the pGEMHE vector). Sequencing (19) of this SacI cassette confirmed its integrity and revealed a single silent mutation corresponding to Pro 370 of GIRK1 (CCC was changed to CCT). The SacI cassette was subcloned back into pGEMHE-KGP. A similar three-fragment ligation strategy was used to construct the G 1 G 2 K 3 and G 1 K 2 K 3 chimeras. G 1 K 2 G 3 was constructed by splicing G 1 K 2 K 3 and K 1 K 2 G 3 using the common KpnI site found in the K 2 region. The integrity of all chimeras was confirmed by sequencing. K 1 G 2 G 3 and K 1 K 2 I 3 were constructed using a PCR splicing method (20). The full-length chimeras were subcloned into pGEMHE. The sequence of the full-length K 1 G 2 G 3 was confirmed. As with K 1 K 2 G 3 , the sequence of a SacI cassette in the K 1 K 2 I 3 chimera was verified after subcloning in pGEMHE-KGP.

FIG. 1. Sequence comparison between GIRK1 and KGP and construction of chimeras and deletion mutants.
A, an alignment of GIRK1 and KGP reveals three regions of differing homology between the two channels. Regions 2 (G 2 and K 2 ) showed a 66% identity (or 92% similarity). B, schematic diagram of wild-type, chimeric, and C-terminal truncated channels. Based on the alignment in A, three regions are identified and denoted with corresponding subscript numbers preceded by a G (for GIRK1) or a K (for KGP). Cartoon representations of the various channels and constructs used are shown and used in subsequent figures. leted) was sequenced and subcloned back to pGEMHE-GIRK1. Similarly, a SacI-cut cassette in the C terminus of pGEMHE-K 1 K 2 was sequenced and subcloned back to pGEMHE-KGP.
The stop codon was removed, and an EcoRV site was introduced to the 3Ј end of both KGP and K 1 K 2 G 3 coding region using PCR, all of which were confirmed by manual sequencing. The coding region of GFP was amplified by PCR and subcloned in frame into EcoRV and NotI sites to produce pGEMHE-KGP-GFP and pGEMHE-K 1 K 2 G 3 -GFP.
Expression in Xenopus Oocytes and Electrophysiological Studies-All constructs were linearized with NheI, and cRNAs were transcribed as described previously (17). Oocytes were isolated and microinjected as described previously (21,22). About 4 ng of each RNA species was injected per oocyte.
Two-electrode voltage clamp on Xenopus oocytes was performed as described previously (8,23). The high potassium bath solution contained (in mM): 91 KCl, 1 NaCl, 1 MgCl 2 , 5 KOH/HEPES, 1.8 CaCl 2 , pH 7.4; high potassium with Ba 2ϩ contained high potassium solution with 3 mM BaCl 2 . Each figure shows a representative experiment that includes oocytes from the same batch. A minimum of two to three batches of oocytes gave similar results.
Single-channel activity was recorded on devitellinized oocytes under the cell-attached mode of standard patch-clamp methods (24,25) using an Axopatch 200A amplifier (Axon Instruments, CA) as described previously (8). Single-channel recordings were performed at a holding membrane potential of Ϫ80 mV in the absence of ACh from the high potassium bath solution.
Immunoprecipitations of Channel Subunits from Xenopus Oocyte Membranes-An antibody raised against the rat N-terminal 21 amino acids of CIR (Upstate Biotechnology, Lake Placid, NY) and an affinity purified C-terminal peptide antibody raised against the rat GIRK1 C-terminal end (Gln 375 -Thr 501 ) (11) were used in immunoprecipitation experiments as described previously (8).

Evaluation of Surface Expression of Channel Proteins by Confocal
Microscopy-Xenopus oocytes were fixed in 4% paraformaldehyde at room temperature overnight. Fixed oocytes were embedded in 3% agarose, and 50-m sections were cut, mounted, and viewed under a confocal microscope (Leica TCS 4D). In order to compare relative surface expression level of the KGP versus K 1 K 2 G 3 channels fused to GFP (Clontech, CA), images of sections from oocytes expressing the different channels were recorded using the same values for parameters such as laser intensity, pinhole, and offset.

RESULTS
The C-terminal Region of GIRK1 Is Involved in Enhancement of Currents with the KGP Subunit-Heterologous coexpression of the GIRK1 and KGP channel subunits in Xenopus oocytes produces greatly enhanced currents (8). Comparison between the GIRK1-and KGP-deduced protein sequences revealed 66% identity or 92% similarity in a central region (region 2), whereas little similarity was found between either the N-(region 1) or C-terminal (region 3) regions of the two proteins (Fig.  1A). We will refer to each of the three regions of the two proteins using the appropriate subscript numbers (1, 2, and 3) preceded by a "G" for GIRK1 or a "K" for KGP. We hypothesized that the coexistence of the nonhomologous regions between the two proteins could be involved in the current enhancement seen upon their assembly. Fig. 1B describes the nomenclature of the specific chimeric and deletion constructs between GIRK1 and KGP used to test this hypothesis.
Because the C terminus of GIRK1 has been implicated in gating interactions of recombinant inwardly rectifying K ϩ channels (26), we tested the involvement of the G 3 region A, Ba 2ϩ -sensitive current traces with 96 mM K ϩ in the bath at Ϫ80 mV and ϩ80 mV (center) and current-voltage relationships (right). Ba 2ϩ -sensitive currents were obtained by subtracting currents obtained in the presence from those in the absence of 3 mM BaCl 2 . B, bar graph of Ba 2ϩ -sensitive currents at Ϫ80 mV and ϩ80 mV from oocytes injected with GIRK1 or its C-terminal deletion constructs alone or together with KGP.
(region 3 of GIRK1) in affecting the basal currents of heteromeric K ϩ channels. The effects of sequential deletions of the G 3 region of GIRK1 (GIRK1⌬1, GIRK1⌬2, and G 1 G 2 ) on current size resulting from coexpression with KGP were assessed. Xenopus oocytes were injected with GIRK1, KGP, GIRK1/KGP, or the G 3 deletion constructs of GIRK1 expressed alone or in combination with KGP cRNAs, and Ba 2ϩ -sensitive membrane currents were measured using two-electrode voltage clamp. Fig. 2A shows representative Ba 2ϩ -sensitive current traces and corresponding mean current-voltage relationships from such an experiment. Fig. 2B summarizes in bar graph form data comparing current magnitudes from oocytes injected with GIRK1 or its C-terminal deletion constructs alone or together with KGP. Incremental deletions of the G 3 region of GIRK1 in the GIRK1/KGP heteromer resulted in corresponding decreases in the magnitude of the enhanced inwardly rectifying current. These results suggested the involvement of the G 3 region in the enhancement of the heteromeric GIRK1/KGP channel currents.
A Chimera between GIRK1 and KGP Yields Large Inwardly Rectifying Currents-We proceeded to construct chimeras that exchanged the nonhomologous C-terminal ends of the two channel subunits (i.e. K 1 K 2 G 3 and G 1 G 2 K 3 ) in order to test the ability of these regions (G 3 or K 3 ) to produce large currents as parts of a homomeric complex of a single protein. Oocytes were injected with GIRK1, KGP, GIRK1/KGP, K 1 K 2 G 3 , or G 1 G 2 K 3 cRNAs, and Ba 2ϩ -sensitive currents were measured. The K 1 K 2 G 3 chimera yielded large inwardly rectifying currents (Fig. 3A). In contrast, expression of the G 1 G 2 K 3 chimera alone gave small currents. These results indicate that regions represented by the K 1 K 2 G 3 chimera can produce large currents, suggesting that the presence of G 3 or possibly the absence of K 3 plays an important role in the enhanced currents of heteromers. Because of the small currents displayed by the G 1 G 2 K 3 chimera, we further tested its functional expression by coinjections with either the wild-type GIRK1 or KGP subunits. As shown in Fig. 3A, the G 1 G 2 K 3 /KGP coinjection produced large currents, ensuring that the G 1 G 2 K 3 chimera produced a functional protein. Interestingly, because the enhanced currents obtained with the G 1 G 2 K 3 /KGP coinjection do not involve the G 3 region of GIRK1, regions of GIRK1 other than G 3 are also involved in producing the large currents of the heteromeric channel. Fig. 3B shows immunoprecipitations from oocyte membranes of metabolically labeled recombinant channel subunits using antibodies directed against either KGP or GIRK1 (8). The 45-kDa band corresponded to KGP (lane 1), whereas the 56 -58-kDa doublet corresponded to GIRK1 (lane 2). Both channel FIG. 3. Chimera K 1 K 2 G 3 expresses inwardly rectifying Ba 2؉ -sensitive currents, severalfold greater than those of KGP or GIRK1 alone. Recordings were performed 4 days after injection and values for A and C are averages of six oocytes (Ϯ S.E.). A, bar graph of Ba 2ϩsensitive currents shows that the K 1 K 2 G 3 cRNA produced large currents similar to the coinjection of GIRK1/KGP cRNA and unlike the G 1 G 2 K 3 chimera. Coinjection of the G 1 G 2 K 3 with KGP yielded enhanced currents, ensuring that the chimera produced a functional protein. B, immunoprecipitations of wild-type and chimeric channel subunits from Xenopus oocyte membranes metabolically labeled with 35 S-labeled methionine and cysteine. KGP-targeted antibody was used in lanes 1, 3, and 4, and GIRK1-targeted antibody was used in lanes 2 and 5. The 45-kDa band corresponded to KGP (lane 1), whereas the 56 -58-kDa doublet corresponded to to GIRK1 (lane 2). When coexpressed in oocytes, both channel subunits were coprecipitated by the KGP-targeted antibody (lane 3). Immunoprecipitation of the chimera K 1 K 2 G 3 with either antibody gave a single ϳ60-kDa band, as expected from expression of a homomeric protein (lanes 4 and 5). C, Ba 2ϩ -sensitive current traces with 96 mM K ϩ in the bath at Ϫ80 mV and ϩ80 mV, and current-voltage relationships for the K 1 K 2 G 3 chimera and the heteromeric GIRK1/KGP channel. D, bar graph of Ba 2ϩ -sensitive K 1 K 2 G 3 currents coexpressed with human muscarinic receptor and normalized to the basal K 1 K 2 G 3 currents. The K 1 K 2 G 3 chimera showed ACh-induced currents that were pertussis toxin-sensitive. Coinjection of G ␤␥ subunits yielded enhanced K 1 K 2 G 3 chimeric currents. G protein-coupled experiments were carried out as described previously (8). subunits could be coprecipitated by either antibody (as for example by the antibody directed against KGP (lane 3)) when coexpressed in oocytes, as shown previously (8). Immunoprecipitation of the chimera K 1 K 2 G 3 with either antibody gave a single ϳ60-kDa band, as expected from expression of a homomeric protein (lanes 4 and 5).
Ba 2ϩ -sensitive current traces and corresponding currentvoltage relationships of the K 1 K 2 G 3 chimera and the GIRK1/ KGP heteromer are shown in Fig. 3C. Currents of the K 1 K 2 G 3 chimera showed faster activation kinetics than the heteromeric wild-type currents. Thus these experiments demonstrated that homomers of the K 1 K 2 G 3 chimera were capable of producing enhanced basal currents with activation kinetics similar to those of KGP.
Coexpression of the K 1 K 2 G 3 chimera with m2 receptor enabled ACh-induced activity that was pertussis toxin-sensitive (Fig. 3D). Similarly, coexpression with G ␤␥ subunits resulted in stimulation of K 1 K 2 G 3 chimeric currents. These results indicated that the regulation of the K 1 K 2 G 3 chimera by G proteins resembled that of the wild-type subunits (8).
The Large Currents of the K 1 K 2 G 3 Chimera Are Not Due to the Absence of the K 3 Region of the KGP Subunit-To test whether the large currents observed with the K 1 K 2 G 3 chimera were due to a release of inhibition exerted by the K 3 region, we replaced the K 3 region with I 3 , the corresponding region from the related but not G protein-gated channel, IRK1 (Fig. 4A). Control experiments showed that coinjection of IRK1 with either GIRK1 or KGP did not further increase the already high IRK1 current levels (data not shown). Oocytes injected with the K 1 K 2 I 3 chimera gave rise to small currents comparable in size with those obtained with the wild-type channel subunits, GIRK1 and KGP (Fig. 4B). Further functional tests for expression of the K 1 K 2 I 3 chimera involved its coinjection with the wild-type GIRK1 subunit. Coinjection of the K 1 K 2 I 3 chimera with GIRK1 produced greatly enhanced currents, ensuring that the K 1 K 2 I 3 chimera produced a functional protein (Fig.  4B). Thus, the results from these experiments suggest that it is not the absence of the K 3 region that is responsible for the large currents seen with the K 1 K 2 G 3 chimera.
The G 3 Region of GIRK1 Is Involved in Producing the Large K 1 K 2 G 3 Chimeric Currents-Sequential deletions of the G 3 region in the K 1 K 2 G 3 chimera were performed (i.e. K 1 K 2 G 3 ⌬1, K 1 K 2 G 3 ⌬2, and K 1 K 2 ) to assess the effects of G 3 on the magnitude of the currents. Note that the ⌬1 and ⌬2 deletions of G 3 in the K 1 K 2 G 3 chimera were identical to those presented earlier for GIRK1 (Fig. 2B). Fig. 4C shows current magnitudes from oocytes injected with the K 1 K 2 G 3 chimera or its C-terminal deletion constructs. Incremental deletions of the G 3 region in the context of the K 1 K 2 G 3 chimera resulted in corresponding decreases in the magnitude of the current. Further functional tests for expression of the K 1 K 2 deletion construct, which showed small currents, ensured that K 1 K 2 produced a functional protein, because coexpression with GIRK1 resulted in enhanced currents. These results, taken together with those presented in Fig. 2, demonstrate the crucial role of the G 3 region of GIRK1 in generation of enhanced currents both in the K 1 K 2 G 3 chimera, as well as in the GIRK1/KGP heteromer. Moreover, the similar pattern of the effect of the sequential deletions on both enhanced currents in the chimera or the heteromer suggests that elements throughout the length of the G 3 region are important determinants of the large current size.
The K 2 Region of KGP Is Sufficient to Produce Enhanced Currents When Coexpressed with GIRK1 Regions-Because the absence of the K 3 region of KGP did not prevent constructs containing the K 1 and K 2 regions from forming active heteromeric channel currents with GIRK1 (Fig. 4, B and C), we proceeded to test the involvement of the K 1 and K 2 regions in current enhancement. We constructed two additional chimeras, K 1 G 2 G 3 and G 1 K 2 G 3 , to separate each of the two KGP regions from one another and to test them in the context of GIRK1 regions. Oocytes injected with the K 1 G 2 G 3 chimera did not give rise to large currents (Fig. 5A). Coinjection of the K 1 G 2 G 3 chimera with the KGP (but not with the GIRK1) subunit produced greatly enhanced currents, ensuring that this chimera gave rise to a functional protein (Fig. 5A). These experiments suggest that the K 1 region of KGP is not sufficient to produce enhanced currents both in chimeras and in the heteromer with GIRK1. In contrast, oocytes injected with the G 1 K 2 G 3 chimera did give rise to sizable currents (Fig. 5B). In addition coinjection of the G 1 K 2 G 3 chimera with GIRK1 resulted in FIG. 4. The large currents of the K 1 K 2 G 3 chimera are not due to the absence of the K 3 region but are directly related to the G3 sequence. Recordings were done 3 days after injection, and the values are averages of five oocytes (Ϯ S.E.).. A, schematic representation of the K 1 K 2 I 3 chimera. The I 3 segment represents the C-terminal end region of IRK1 between Arg 359 and Ile 428 amino acid residues (28). B, expression of the K 1 K 2 I 3 chimera gave rise to small currents comparable in size with those obtained with the wild-type channel subunits, GIRK1 and KGP. When coinjected with GIRK1 subunit, the K 1 K 2 I 3 chimera produced greatly enhanced currents, ensuring that the chimeric construct produced a functional protein. C, effects on the current magnitude of oocytes injected with the K 1 K 2 G 3 chimera or its C-terminal deletion constructs, K 1 K 2 G 3 ⌬1, K 1 K 2 G 3 ⌬2, and K 1 K 2 were assessed. Incremental deletions of the G 3 region in the context of the K 1 K 2 G 3 chimera resulted in corresponding decreases in the magnitude of the large chimeric current. Coexpression of the K 1 K 2 deletion construct, which showed small currents, with GIRK1 resulted in large currents, ensuring that the K 1 K 2 construct produced a functional protein.
greatly enhanced currents, compared with the currents of the G 1 K 2 G 3 chimera alone (Fig. 5B). Because K 2 was the only KGP region present in this coexpression experiment, we concluded that K 2 is sufficient to enhance currents in the heteromeric channels. Interestingly, the enhanced currents of the G 1 K 2 G 3 chimera were much smaller than those of the K 1 K 2 G 3 chimera, consistent with the notion that the presence of the N terminus of GIRK1 may be somehow decreasing basal currents (27). Single-channel recording showed that G 1 K 2 G 3 had conductance and mean open time similar to KGP and an NP o in between that of KGP and K 1 K 2 G 3 (data not shown).
Single-channel Characteristics of K 1 K 2 G 3 Reveal a Higher NP o than That of KGP-Single-channel recordings of the K 1 K 2 G 3 were obtained and compared with the corresponding wild-type KGP and GIRK1/KGP channel activities. Cell-attached recordings from oocytes expressing wild-type KGP or chimeric K 1 K 2 G 3 channels displayed similar open time kinetics, distinct from those obtained by the GIRK1/KGP coinjections (Fig. 6). Although single-channel activity was highly variable, the activity (NP o ) trend obtained from cell-attached patches expressing KGP, K 1 K 2 G 3 , or GIRK1/KGP was similar to the corresponding whole-cell current amplitudes. The singlechannel conductance of K 1 K 2 G 3 was similar to that previously shown for KGP (8). These data suggest that the structural determinants represented in the K 1 K 2 G 3 chimera show enhanced unitary activity with KGP-like characteristics.

K 1 K 2 G 3 and KGP Proteins Show Similar Surface Expression in Oocyte
Membranes-To investigate whether there was an increase in the amount of K 1 K 2 G 3 chimeric proteins expressed on the oocyte cell membrane compared with that of KGP, we fused the GFP to the C-terminal ends of both channels and examined the surface protein level by fluorescence confocal microscopy. Uninjected oocytes gave very weak fluorescence background (Fig. 7A). Oocytes injected with 2 ng of KGP-GFP RNA showed strong fluorescence localized in the plasma membrane (Fig. 7B). Injection of a 5-fold higher concentration of KGP-GFP RNA resulted in an apparent increase in the amount of fluorescence detected on the cell membrane (Fig. 7C). Injection of corresponding amounts of K 1 K 2 G 3 -GFP RNA resulted in qualitatively similar fluorescence signals on the cell surface as with KGP-GFP (Fig. 7, D and E). Both fusion proteins produced similar macroscopic currents (Fig. 7F) and single-channel char-FIG. 5. The K 2 region of KGP is involved in the production of the large currents exhibited by the K 1 K 2 G 3 chimera. Recordings were done 4 days after injection. The values are averages of six oocytes (Ϯ S.E.). A, expression of the K 1 G 2 G 3 chimera did not give rise to large currents unlike the K 1 K 2 G 3 chimera. In contrast, coexpression of the K 1 G 2 G 3 chimera with KGP but not GIRK1 produced greatly enhanced currents, ensuring that the K 1 G 2 G 3 chimera produced a functional protein. B, expression of the G 1 K 2 G 3 chimera did give rise to large currents when expressed alone or with GIRK1. acteristics (data not shown) compared with the parental counterparts, indicating that the 238-amino acid-long GFP does not interfere with normal channel function. Moreover, the increase in fluorescence on the oocyte surface correlated well with the larger magnitude of the basal currents recorded from oocytes injected with increasing amount of K 1 K 2 G 3 -GFP RNA (Fig. 7F). Hence, there was no qualitative difference on the surface expression of the two channel proteins when injected in comparable amounts in Xenopus oocytes. Yet 2 ng of K 1 K 2 G 3 -GFP produced much greater currents than 10 ng of KGP-GFP despite the inverse relationship on surface expression, suggesting that the K 1 K 2 G 3 channels were more active compared with the wild-type KGP channels. These results, together with those obtained from the single-channel measurements, suggest that it is the frequency of opening of K 1 K 2 G 3 that must be higher than that of KGP, because the two channels have similar mean open times. DISCUSSION We investigated the role of the homologous and nonhomologous regions between GIRK1 and KGP in the current enhancement seen upon coexpression of the two subunits, using deletion mutants and chimeric constructs between the two proteins. We present evidence that the G 3 region of GIRK1 is involved in the current enhancement of KGP currents. Although the G 3 region is involved in enhancing KGP currents, its presence was not absolutely necessary, because enhanced currents could result even in its absence (as were seen when G 1 G 2 K 3 and KGP were coexpressed). This result is consistent with a previous report concluding that the GIRK1 segment from amino acids 357 to 501 (same as the G 3 region) was not needed for CIR-dependent agonist-induced current enhancement (29). This result is also consistent with our previous finding that a single P-region residue, Phe 137 of GIRK1 (in the G 2 region), is also involved in enhancing KGP current and is responsible for the prolonged single-channel openings of the heteromer (17). K 2 , on the other hand, was the only single region of KGP that together with GIRK1 regions produced large currents, as could be demonstrated with the G 1 K 2 G 3 chimera alone or its coexpression with GIRK1. The necessity of the presence of K 2 was suggested by all cases where current enhancement was observed.
The mechanism by which the G 3 and K 2 regions produce enhanced currents is not clear. The gating mechanism of the heteromeric channel is undoubtedly complex and presently not fully understood. On one hand, the channel can be gated by a mechanism independent of G protein activation that involves a functional modification dependent on MgATP followed by intracellular Na ϩ gating (30). It is possible that direct interactions of the G 3 and K 2 regions in the context of a tetrameric channel result in large basal currents, independent of G pro- FIG. 7. Surface expression of KGP and K 1 K 2 G 3 channels on Xenopus oocyte membranes as shown by tagging the channels with the GFP. A, section of an uninjected oocyte to illustrate the low fluorescent background. B and C, representative sections of oocyte membranes injected with 2 and 10 ng of KGP-GFP RNA, respectively. D and E, representative sections of oocyte membranes injected with 2 and 10 ng of K 1 K 2 G 3 -GFP RNA, respectively. F, bar graph of Ba 2ϩsensitive currents recorded from oocytes 3 days after injecting the corresponding RNA species. Each bar is an average of three oocytes. tein gating. On the other hand, the G ␤␥ subunits of G proteins have been shown to directly bind to both GIRK1 and CIR channels (31), to the C termini of both GIRK1 (32) and CIR (or GIRK4) (33), or to specific regions within the C terminus of GIRK1 (29,34). Moreover, it has also been demonstrated (6,35) that G ␤␥ subunits stimulate recombinant heteromeric GIRK subunit activity in a manner similar to that of the native ACh-induced K ϩ channel (1). In addition, pertussis toxin treatment of GIRK1/KGP heteromeric or KGP homomeric channels reduced basal and abolished agonist-induced currents, suggesting that channel gating proceeds to a large extent, but not exclusively, via a pertussis toxin-sensitive G protein (8). Thus, although the K 1 K 2 G 3 chimera remains sensitive to G protein regulation, it is possible that it brings together regions of the two subunits that alter the number or quality of G ␤␥ -binding sites, thus fine tuning its response to endogenous or to overexpression of exogenous G ␤␥ subunits. Such a complicated scenario will require testing by further detailed studies. It has been previously shown that the region of the C terminus encoded between Lys 405 and Thr 501 does not bind G ␤␥ subunits (34). Our GIRK1⌬2 construct, which lacked the segment from Arg 409 to Thr 501 , exhibited greatly reduced currents when coexpressed with KGP. This result would argue against the possibility that the current enhancement seen in the heteromeric or chimeric channels is controlled by the mere presence of G ␤␥ -binding sites. A recent study has implicated the proximal C-terminal region of IRK1 region between amino acids 220 and 300) as partly responsible for the assembly of inwardly rectifying potassium channels (36). GIRK1, GIRK4, and IRK1 show high homology in a region spanning amino acids 50 -350. If GIRK channels use this same region as other inwardly rectifying K ϩ channels for assembly, then the effects of the G 3 region on KGP activity are unlikely to be due to an enhanced assembly of the individual subunits.
The full-length C-terminal cytoplasmic tail (amino acids 183-501) (26) has been shown to block GIRK1 and ROMK1 macroscopic currents in Xenopus oocytes. Our data on deletion mutants of both GIRK1 and K 1 K 2 G 3 chimera in the G 3 region (amino acids 359 -501) did not result in an increase in macroscopic current as would be expected from a release of physical block on the channel. These results would suggest that the sequence constituting the physical block proposed by Dascal et al. (26) does not lie in the last 150 amino acids of GIRK1.
The activation kinetics of the K 1 K 2 G 3 or G 1 K 2 G 3 chimeras were faster than those of the currents of the GIRK1/KGP heteromers, resembling those of KGP alone (8). Similarly, singlechannel characteristics of the active chimeras K 1 K 2 G 3 or G 1 K 2 G 3 resembled those of KGP in open-time kinetics and conductance. These results suggest that the K 2 region contains the structural determinants responsible for these KGP channel characteristics. The increase in the magnitude of the whole-cell current of the K 1 K 2 G 3 or G 1 K 2 G 3 chimeras correlated well with the increase in single-channel activity (NP o ). Yet, this increase in NP o did not include a contribution from a change in the mean open time for the chimeras as it did for the heteromers. These observations suggest that region 2 of the subunits could determine open times (K 2 promoting short openings, whereas G 2 promoting longer openings), whereas region 3 (or its regulators) could modulate the frequency of channel opening. This scheme is consistent with the notion that Phe 137 of GIRK1 controls the gating kinetics of other channel family members with which it associates (17,37).