Receptor-mediated inhibition of G protein-coupled inwardly rectifying potassium channels involves G(alpha)q family subunits, phospholipase C, and a readily diffusible messenger.

G protein-coupled inwardly rectifying K+ (GIRK) channels can be activated or inhibited by distinct classes of receptor (G(alpha)i/o- and G(alpha)q-coupled), providing dynamic regulation of cellular excitability. Receptor-mediated activation involves direct effects of G(beta)gamma subunits on GIRK channels, but mechanisms involved in GIRK channel inhibition have not been fully elucidated. An HEK293 cell line that stably expresses GIRK1/4 channels was used to test G protein mechanisms that mediate GIRK channel inhibition. In cells transiently or stably cotransfected with 5-HT1A (G(alpha)i/o-coupled) and TRH-R1 (G(alpha)q-coupled) receptors, 5-HT (5-hydroxytryptamine; serotonin) enhanced GIRK channel currents, whereas thyrotropin-releasing hormone (TRH) inhibited both basal and 5-HT-activated GIRK channel currents. Inhibition of GIRK channel currents by TRH primarily involved signaling by G(alpha)q family subunits, rather than G(beta)gamma dimers: GIRK channel current inhibition was diminished by Pasteurella multocida toxin, mimicked by constitutively active members of the G(alpha)q family, and reduced by minigene constructs that disrupt G(alpha)q signaling, but was completely preserved in cells expressing constructs that interfere with signaling by G(beta)gamma subunits. Inhibition of GIRK channel currents by TRH and constitutively active G(alpha)q was reduced by, an inhibitor of phospholipase C (PLC). Moreover, TRH- R1-mediated GIRK channel inhibition was diminished by minigene constructs that reduce membrane levels of the PLC substrate phosphatidylinositol bisphosphate, further implicating PLC. However, we found no evidence for involvement of protein kinase C, inositol trisphosphate, or intracellular calcium. Although these downstream signaling intermediaries did not contribute to receptor-mediated GIRK channel inhibition, bath application of TRH decreased GIRK channel activity in cell-attached patches. Together, these data indicate that receptor-mediated inhibition of GIRK channels involves PLC activation by G(alpha) subunits of the G(alpha)q family and suggest that inhibition may be communicated at a distance to GIRK channels via unbinding and diffusion of phosphatidylinositol bisphosphate away from the channel.

G protein-coupled inwardly rectifying K ϩ (GIRK 1 ; Kir3.x) channels are targets for up-and down-regulation by receptors that couple to different classes of heterotrimeric G proteins, providing a mechanism for dynamic regulation of cellular excitability (1,2). Activation of GIRK channels involves receptors that couple to pertussis toxin (PTx)-sensitive G␣ subunits of the G␣ i or G␣ o family, whereas inhibition of GIRK channels is mediated by receptors that couple via PTx-insensitive G␣ subunits. This dual regulation of GIRK channels has been noted in a number of cellular contexts, including atrial cells of the myocardium (3,4), aminergic neurons of the brainstem (5,6), and enteric neurons of the peripheral nervous system (7). Because demonstration of this dual regulation requires simultaneous activation of different receptor classes, the phenomenon may be even more widespread than is currently realized.
Of the dual components of receptor-mediated GIRK channel regulation (activation and inhibition), most emphasis has been on understanding cellular mechanisms underlying GIRK channel activation. It is now well established that, following agonist stimulation, G␤␥ subunits released from receptor-bound heterotrimers bind directly to GIRK channels to increase channel activity (reviewed in Refs. 1, 2, and 8). It was recently found that G␤␥ activation of GIRK channels requires permissive levels of membrane phosphatidylinositol bisphosphate (PIP 2 ) and that interactive effects of these mediators on channel activity result from G␤␥-mediated stabilization of PIP 2 binding to GIRK channels (9 -11).
In contrast to GIRK channel activation, mechanisms that contribute to receptor-mediated inhibition of GIRK channels are not well understood. It is clear that GIRK channel inhibition involves receptors that typically couple via PTx-insensitive G proteins of the G␣ q family, and it therefore seems reasonable to suspect involvement of that class of G␣ subunit. However, GIRK channel inhibition by these and/or other classes of PTxinsensitive G␣ subunits has not been directly examined. Moreover, it remains to be determined if the signal for GIRK channel inhibition derives from the ␣ subunit or ␤␥ dimer of the heterotrimeric G protein. In this respect, we recently found that G␤ 5 -containing dimers can inhibit G␤␥-activated GIRK channels (12), a finding that is especially intriguing in the current context of receptor-mediated GIRK channel inhibition since G␤ 5 ␥ dimers associate preferentially with G␣ q -coupled receptors (13,14). Signaling mechanisms that contribute to GIRK channel inhibition downstream of G protein subunits are also not known with any certainty, although recently it was suggested that phospholipase C (PLC) activation could contribute to GIRK channel inhibition by G␣ q -coupled receptors in atrial myocytes by decreasing membrane levels of PIP 2 rather than by production of downstream mediators (15)(16)(17). It remains to be determined if a similar mechanism contributes to GIRK channel inhibition by similar receptors in other settings.
Here, we prepared mammalian cell lines expressing GIRK channel subunits together with two different classes of G protein-coupled receptors in order to recapitulate dual regulation of GIRK channels in a heterologous expression system. Using this system, we found that G␣ q -coupled thyrotropin-releasing hormone (TRH) type 1 (TRH-R1) receptors can inhibit GIRK channels pre-activated by G␣ i/o -coupled 5-hydroxytryptamine (5-HT) type 1A (5-HT 1A ) receptors. GIRK channel inhibition involved G␣ q family subunits, PIP 2 , and activation of PLC, but it was independent of downstream signaling mediators typically associated with PLC activation such as protein kinase C (PKC), inositol trisphosphate (IP 3 ), and intracellular calcium. These data are consistent with the possibility that PLC-mediated decreases in membrane PIP 2 levels provide the signal for GIRK channel inhibition by G␣ q -coupled receptors (15)(16)(17); if this is the case, PIP 2 may be readily diffusible since inhibition was robust even in cell-attached patches that were not exposed to agonist. These results further indicate that G␣ and G␤␥ subunits, each derived from a different class of receptor, can converge with opposite effects on the same channel to dynamically regulate cellular excitability. 1A Receptors-The derivation and maintenance of stable human embryonic kidney HEK293 cells expressing GIRK1 (Kir3.1) and GIRK4 (Kir3.4) channels (referred to as G1/4 cells) were described previously (12). This cell line was further stably transfected with the TRH-R1 and 5-HT 1A receptors (referred to as G1/4R cells). Briefly, TRH-R1 (pBS; M. C. Gershengorn, Cornell University) and 5-HT 1A receptor (pGEM3z; D. K. Grandy, Vollum Institute) cDNAs were subcloned into pcDNA3 and transfected together with pBabe-Puro (K. R. Lynch, University of Virginia) into G1/4 cells and maintained under G418 (400 g/ml; Life Technologies, Inc.) and puromycin (200 ng/ml; Sigma) selection. A G418-and puromycin-resistant cell line was selected based on strong expression of inwardly rectifying K ϩ currents and robust regulation of those currents by TRH and 5-HT (G1/4R cells). Both the G1/4 and G1/4R cell lines were cultured in Dulbecco's modified Eagle's medium/ nutrient mixture F-12 with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin in the continued presence of G418 (G1/4 cells) or G418 and puromycin (G1/4R cells).
Western Blots-Crude cell lysates were prepared from transfected cells in the presence of a mixture of protease inhibitors (2 g/ml each leupeptin, pepstatin, and aprotinin and 100 M phenylmethylsulfonyl fluoride). Proteins were resolved by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose, and immunoblotted. Expression of FLAG epitope-tagged constitutively active G␣ q subunits (FLAG-G␣ q *) was detected using M2 anti-FLAG monoclonal antisera (10 g/ ml; Sigma), and GIRK channel subunit expression was determined with a rabbit anti-GIRK1 antibody (1:500 dilution; Alomone Labs). Primary antibodies were detected with horseradish peroxidase-conjugated mouse or rabbit IgG using enhanced chemiluminescence.
Electrophysiological Recordings from Transfected Cells-Cells were plated onto glass coverslips at a confluency appropriate to obtain single cells for electrical recording 48 -72 h after transfection (except where noted). Coverslips were submerged in a recording chamber at room temperature on microscopes equipped with epifluorescent optics (Zeiss Axioskop FS or Nikon TE300). Individual cells that expressed GFP were identified using standard fluorescein isothiocyanate filter sets and targeted for recording.
Electrophysiological data were acquired and analyzed using the pCLAMP suite of programs (Axon Instruments, Inc.). Series resistance was typically Ͻ10 megaohms and compensated by ϳ70 -80%. Membrane voltages were corrected for a 10-mV liquid junction potential. To record inwardly rectifying whole cell currents, cells were held at Ϫ50 mV, and a slow voltage ramp command (⌬Ϫ90 mV, 0.1 V/s) was applied at 0.1 Hz. Membrane current was filtered at 0.5-1 kHz and sampled at 1-2 kHz. Measured variables included holding current at Ϫ50 mV and slope conductance, obtained from a linear fit to current-voltage data over the range of Ϫ100 to Ϫ120 mV. The effect of TRH on GIRK channel conductance was determined following subtraction of endogenous currents, defined by their resistance to 0.2 mM Ba 2ϩ . Single channel data were filtered at 2 kHz (four-pole Bessel) and collected on line using gap-free recording with a 10-kHz sampling frequency. The slope of current amplitudes obtained from gaussian fits to all-points histograms at three different patch potentials was used to determine single channel conductance. Analysis of the effects of agonist stimulation on GIRK channel activity in cell-attached patches was performed using Fetchan (Axon Instruments, Inc.) and the nPo freeware program (provided by J. L. Sui). All data are presented as means Ϯ S.E., and each data point includes results obtained from at least two transfections.
Drugs and Reagents-Compounds were prepared as concentrated stock solutions, stored at Ϫ20°C, and diluted to the indicated concentrations in recording or pipette solutions, as needed. 5-HT (serotonin, Sigma) was made up as a 10 mM stock solution with 100 mM ascorbate and applied to cells at 50 M. TRH (100 M stock solution; Peninsula Laboratories, Inc.) was applied at 100 nM. Toxins from Bordetella pertussis (132 g/ml) and Pasteurella multocida (50 g/ml) were generously provided, respectively, by E. L. Hewlett (University of Virginia) and L. J. Eaton (List Biological Laboratories, Inc.); cells were incubated in toxins overnight at 2 and 1 g/ml, respectively. The PLC inhibitor U73122 and its inactive analog, U73343, were obtained from Research Biochemicals Inc. and prepared as 2 mM stock solutions in dimethyl sulfoxide; preincubation with U73122 or U73443 (both at 1 M) in the culture medium was begun at least 30 min prior to recording to test effects on receptor-mediated GIRK channel inhibition and ϳ36 -48 h before recording to test effects in cells transfected with G␣ q *. The bisindolylmaleimide PKC inhibitor (GF 109203X, Calbiochem) was stored as a 5 mM stock solution, and continuous exposure to 5 M bisindolylmaleimide was started Ն30 min before recording. The phorbol ester phorbol 12,13-dibutyrate (PDBu; Research Biochemicals Inc.) and its inactive analog, 4␣-PDBu (Alexis), were prepared as 1 mM stock solutions in dimethyl sulfoxide and applied in the bath at 1 M. IP 3 (10 mM; Alexis) was diluted to 100 M in pipette solution.

Dual Modulation of GIRK1/4 Channel Currents in a Stable
HEK293 Cell Line-To study mechanisms contributing to receptor-mediated inhibition of recombinant GIRK channels in the context of a mammalian cell system, we prepared a stable HEK293 cell line that expresses Kir3.1/3.4 (GIRK1/4) channels. As we described previously for G1/4 cells, hyperpolarizing ramp voltage commands evoked substantial inwardly rectifying currents under non-stimulated conditions that were due, in large part, to channel activation by free G␤␥ subunits endogenous to those cells (12). In cells transfected with the G␣ i/ocoupled 5-HT 1A receptor and G␣ q -coupled TRH-R1, whether expressed transiently in G1/4 cells or stably in G1/4R cells, the dual regulation of GIRK channels that is seen in various native cell systems (5-7) was fully recapitulated, i.e. GIRK channel currents pre-activated by G␣ i/o -coupled receptors were inhibited by G␣ q -coupled receptors. Thus, as illustrated in the representative cell of Fig. 1A, the basal inwardly rectifying conductance was increased by 5-HT (50 M), and this enhanced conductance was inhibited by TRH (100 nM). Although it was often more convenient to use the G1/4R cell line stably expressing the 5-HT 1A and TRH-R1 receptors, we obtained essentially identical results with transient receptor expression; therefore, where available, data from experiments using transient and stable receptor expression were combined. Averaged data from cells treated with both 5-HT and TRH (n ϭ 9) revealed that 5-HT caused an increase in GIRK channel conductance (42.2 Ϯ 8.7% of the initial conductance), whereas TRH inhibited GIRK channel conductance (80.4 Ϯ 3.8% of the conductance in 5-HT); when referenced to the initial GIRK channel conductance before 5-HT, the inhibition by TRH was often Ͼ100% (average of 118.3 Ϯ 11.5%, n ϭ 9), indicating that signaling from 5-HT 1A and TRH-R1 receptors converges on the same GIRK channels, but with opposite effects.
It is well known that activation of GIRK channels by 5-HT 1A and other receptors in native systems is mediated by PTxsensitive G proteins of the G␣ i/o class. Likewise, we found that GIRK channel activation by 5-HT 1A receptors was completely blocked by PTx (2 g/ml, overnight; 56.2 Ϯ 7.3% activation in control cells (n ϭ 15) versus 11.9 Ϯ 3.8% inhibition after PTx (n ϭ 16)). On the other hand, receptor-mediated GIRK channel inhibition in native systems is PTx-insensitive; and accordingly, we found no effect of PTx on TRH-induced GIRK channel inhibition (Fig. 1B, right panel). Because TRH-R1 is typically associated with G proteins of the G␣ q family, we tested the effects of P. multocida toxin (PMT) (Fig. 1B, left panel), which appears to directly activate and permanently uncouple G␣ q subunits from receptors (25, 26); compared with the typical TRH-induced inhibition of GIRK channel conductance in a control cell (closed circles), the inhibition by TRH was completely blocked in a representative cell treated with PMT (1 g/ml, overnight; open circles). The degree of inhibition of GIRK channel conductance by TRH was Ͼ55% in nearly all control cells (38 of 41 cells tested; 93%), whereas it reached that level in only three PMT-treated cells (of 15 cells tested; 20%); as shown in Fig. 1B (right panel), the averaged percent inhibition of GIRK channel conductance by TRH was significantly decreased by PMT (34.5 Ϯ 5.7%, n ϭ 15) compared with control cells (72.6 Ϯ 2.0%, n ϭ 41; p Ͻ 0.05). Thus, these data suggest that TRH-R1 couples to G␣ q family subunits to mediate inhibition of GIRK channels.
G␣ q Family Subunits and G␤ 5 -containing G␤␥ Dimers Are Capable of GIRK Channel Inhibition-Receptor-mediated modulation of native GIRK channel currents is invariably associated with G␣ q -coupled receptors (reviewed in Ref. 1). To determine if members of the G␣ q family can cause inhibition of GIRK channels independent of receptor activation, we transfected constitutively active (i.e. GTPase-deficient, Gln 3 Leu mutant) G␣ q * subunits into G1/4 cells. We chose to use constitutively active G␣ subunits because effects of wild-type G␣ subunits could result from promiscuous coupling to receptors; from activity due to intrinsic GDP/GTP cycling; or indirectly, from effects due to sequestering free G␤␥ subunits.
As evident in the sample records from representative cells shown in Fig. 2A, basal GIRK channel currents were substantially smaller in a cell transfected with FLAG epitope-tagged G␣ q * compared with a control G1/4 cell. Currents were essentially identical in cells transfected with either FLAG-tagged or untagged G␣ q *, and data from experiments with those constructs were combined in subsequent analyses. As shown in the averaged data of Fig. 2C, GIRK channel currents were significantly decreased in G1/4 cells transfected with G␣ q *; indeed, the conductance in cells transfected with G␣ q * approximated that seen in parental HEK293 cells not expressing GIRK channels (data not shown), suggesting that channel activity was essentially completely eliminated by G␣ q *. Likewise, GIRK channel currents were drastically reduced in cells transfected with constitutively active mutants of all other G␣ q family members tested, including G␣ 11 *, G␣ 14 *, and G␣ 15 * (Fig. 2C). Expression of GIRK channel subunits was apparently unaffected by G␣ q * transfection inasmuch as we found no difference in GIRK1 levels by Western blot analysis ( Fig. 2A, inset). In addition, this inhibitory effect was specific to G␣ q family subunits since GIRK channel currents were unaffected by representative constitutively active members of all other major classes of G␣ proteins. These included all members of the G␣ i/o family (G␣ i1 *, G␣ i2 *, G␣ i3 *, and G␣ o *) as well as the PTxinsensitive G␣ s *, G␣ 12 *, G␣ 13 *, and G␣ z * subunits (Fig. 2C).
G␤␥ subunits also convey signaling information to effectors following receptor activation (reviewed in Refs. 1, 2, and 8). Many G␤ and G␥ subunit genes have been identified by molecular cloning, and it is now well established that multiple different combinations of G␤␥ are capable of activating GIRK channels, including dimers composed of G␤ 1 , G␤ 2 , G␤ 3 , and G␤ 4 , together with various G␥ subunits (12, 27, 28). By contrast, we recently found that G␤ 5 forms G␤␥ dimers that cause inhibition, rather than the customary G␤␥-mediated activation of GIRK channel currents (12). As shown in the representative records of Fig. 2B and in averaged data of Fig. 2C, GIRK channel currents were consistently smaller in cells transfected with G␤ 5 ␥ 2 than in control cells not expressing exogenous G protein subunits. The observation that G␤ 5 -containing dimers are unique among G␤␥ dimers in causing GIRK channel inhibition is particularly cogent in the current context of GIRK channel inhibition by G␣ q -coupled receptors since G␤ 5 -containing dimers associate preferentially with G␣ subunits of the G␣ q family (13,14). Together, these data indicate that G␣ q family subunits and G␤␥ dimers that contain G␤ 5 are capable of inhibiting GIRK channel currents.
Receptor-mediated GIRK Channel Inhibition Involves G␣ q , but Does Not Require G␤␥ Dimers-Our experiments using chronic exogenous expression of G protein subunits appear to constrain the possible mediators of GIRK channel inhibition either to G␣ subunits of the G␣ q family or to G␤␥ dimers containing G␤ 5 (12). The next series of experiments was designed to determine the role of these subunits in acute receptormediated inhibition of GIRK channels.
First, we examined TRH-induced inhibition of GIRK channel currents in cells transfected with either G␣ q * or G␤ 5 ␥ 2 , as shown in Fig. 3. Consistent with the results presented above, expression of both G␤ 5 ␥ 2 and G␣ q * inhibited basal GIRK channel currents. When challenged with TRH, however, GIRK channel currents were further diminished in G␤ 5 ␥ 2 -transfected cells, but not in G␣ q *-transfected cells. When inhibition was expressed as percent of control (Fig. 3, inset), it was clear that TRH inhibited GIRK channel currents equally effectively in control and G␤ 5 ␥ 2 -expressing cells. These data are consistent with the possibility that TRH receptor-mediated GIRK channel inhibition was occluded by G␣ q *, but not by G␤ 5 ␥ 2 .
It is important to note that, although TRH was without effect on GIRK channel currents in G␣ q *-expressing cells, the basal GIRK channel currents were essentially eliminated even before TRH application. Thus, although the results are consistent with the interpretation that G␣ q * occluded receptor inhibition, it is also possible that chronic overexpression of G␣ q * inhibited GIRK channel currents through a separate mechanism and that TRH was without effect simply because there was no residual basal GIRK channel current to inhibit. Therefore, we performed additional experiments in which cells were trans-FIG. 2. GIRK1/4 channel currents are inhibited by G␣ q family subunits and G␤ 5 -containing G␤␥ dimers. A, sample current traces from a representative control G1/4 cell and from a cell expressing FLAG-G␣ q *. The basal GIRK channel conductance was essentially completely eliminated in the cell expressing FLAG-G␣ q *. Inset, immunoblots of G1/4 cell lysates from control cells and from cells transfected with FLAG-G␣ q *. There was no effect on GIRK1 channel expression (upper panel) in cells expressing FLAG-G␣ q * (lower panel). B, sample current traces from a control G1/4 cell and from a cell expressing G␤ 5 ␥ 2 dimers. Conductance was reduced in the cell expressing G␤ 5 ␥ 2 . C, averaged data (means Ϯ S.E.) from cells expressing the indicated constitutively active G␣ subunits and G␤ 5 ␥ 2 . Asterisks denote statistically significant differences from control cells (p Ͻ 0.05 by ANOVA with post hoc Bonferroni's test; the number of cells in each group is shown in parentheses). Note that among constitutively active G␣ subunits, only members of the G␣ q family inhibited GIRK channel conductance; ␤ 5 ␥ 2 also inhibited GIRK channel conductance. NS, not significant.
FIG. 3. Effect of G␣ q * and G␤ 5 ␥ 2 on TRH-induced inhibition of GIRK channel currents. The effect of TRH was tested in control cells (n ϭ 41) and in cells transfected with either G␤ 5 ␥ 2 (n ϭ 8) or G␣ q * (n ϭ 9). Averaged data (means Ϯ S.E.) show basal conductance (black bars) and conductance in the presence of TRH (gray bars). Inset, TRH-induced inhibition of conductance (percent of control conductance; means Ϯ S.E.). Asterisks denote statistically significant differences from control cells (p Ͻ 0.05 by ANOVA with post hoc Bonferroni's test). Inhibition of GIRK channel currents was preserved in cells expressing G␤ 5 ␥ 2 , but not in cells expressing G␣ q *.
fected with minigene constructs designed to interfere with signaling via either G␣ q or G␤␥ subunits (Fig. 4).
We chose to use two different inhibitors of G␣ q signaling: RGS2 and a GFP-tagged C-terminal construct of PLC␤1 (18). Both of these constructs selectively bind activated G␣ q -like proteins, and their overexpression interferes specifically with receptor signaling mediated by G␣ q (18,29,30). Although each of these constructs exhibits GTPase activity, the inhibition of signaling appears to be independent of that activity and due rather to their ability to compete with effectors for G␣ q binding (i.e. as "sinks" or "effector antagonists") (30). We tested the effect of TRH on GIRK channel currents in cells transfected with these two different inhibitors of G␣ q signaling. As is evident in the sample records provided in Fig. 4A, the inhibition of GIRK channel currents by TRH was diminished by PLC␤1-ct or RGS2. Indeed, unlike the control condition, in which GIRK channel current inhibition was Ͼ55% in the overwhelming majority of cells (38 of 41 cells tested; 93%), inhibition by TRH was Ͼ55% in only 3 of 25 of cells transfected with PLC␤1-ct (12%) and in just 3 of 16 cells transfected with RGS2 (19%). This was also borne out in summary data from these cells, where the averaged percent inhibition by TRH was significantly reduced by both constructs (Fig. 4C). These data indicate that inhibition of GIRK channels mediated by the TRH receptor involves signaling by G␣ q .
In contrast to the abrogating effects of the minigene inhibitors of G␣ q signaling, we found that receptor-mediated GIRK channel inhibition was largely preserved in cells transfected with a number of different G␤␥ buffers. First, we overexpressed two RGS proteins, RGS6 and RGS11, which, by virtue of their G␥-like domains that bind specifically to G␤ 5 (19,20), can interfere with signaling mediated by G␤ 5 -containing G␤␥ pairs (21). As shown in Fig. 4B, TRH produced a strong inhibition of GIRK channel currents in cells transfected with either of the G␥-like domain-containing RGS proteins; the averaged percent inhibition in cells transfected with RGS6 and RGS11 was not different from that in control cells (Fig. 4C). To rule out possible involvement of other G␤␥ dimers in receptor-mediated GIRK channel inhibition (e.g. by their ability to activate PLC) (8), we tested two additional, relatively nonselective G␤␥ sinks, wildtype G␣ t and ␤ARK-ct. Neither had any effect on TRH-induced inhibition of GIRK channel currents (Fig. 4C). Attesting to the efficacy of G␤␥ sequestration by G␣ t and ␤ARK-ct, the basal GIRK channel conductance was significantly reduced in G␣ tand ␤ARK-ct-expressing cells (7.1 Ϯ 0.7 nanosiemens in control cells versus 3.7 Ϯ 0.6 and 2.2 Ϯ 0.4 nanosiemens in G␣ t -and ␤ARK-ct-expressing cells, respectively; both p Ͻ 0.05) (see also Ref. 12); and moreover, expression of each completely blocked GIRK channel activation by 5-HT 1A receptors, a known G␤␥mediated effect (data not shown). Thus, these data indicate that GIRK channel inhibition by TRH-R1 is due, at least in large part, to G␣ q rather than G␤␥ signaling.
Receptor-mediated GIRK Channel Inhibition Involves PLC-In light of the above results that suggest a primary role for G␣ q signaling in receptor-mediated GIRK channel inhibition, we tested if activation of PLC, a well known G␣ q effector, might also play a role. As illustrated in Fig. 5, TRH-induced inhibition of GIRK channels was markedly diminished in cells pretreated with the PLC inhibitor U73122, but not with the control compound U73343 (both at 1 M for at least 30 min.). The inhibition by TRH was Ͼ55% in only three cells and averaged just 35.7 Ϯ 4.4% (n ϭ 19) during U73122 exposure, significantly less than the inhibition recorded in the presence of U73343 (75.2 Ϯ 3.9%, n ϭ 18) or in control cells that were not pretreated with either compound (72.6 Ϯ 2.0%, n ϭ 41).
We also tested if PLC activation contributes to the sustained inhibition of GIRK channels obtained by expression of constitutively active G␣ q * (Fig. 5C). Throughout the period following transfection with G␣ q * (or empty vector), cells were incubated continuously with U73122 at 1 M (ϳ36 -48 h). Note that we observed an apparent nonspecific inhibition of GIRK channels in control cells chronically treated with U73122 (see also Ref. 17). Thus, the averaged conductance in those U73122-treated cells was smaller than that seen in untreated control cells (3.1 Ϯ 0.5 versus 9.2 Ϯ 0.6 nanosiemens, respectively; p Ͻ 0.05) (compare Figs. 5C and 2C, first bars). Nevertheless, as shown in Fig. 5C, the averaged conductance was further reduced in all cells transfected with G␣ q *; and importantly, the reduction FIG. 4. Receptor-mediated GIRK channel inhibition involves G␣ q . A, the effect of TRH on GIRK channel conductance in representative cells transfected with the G␣ q sinks PLC␤1-ct (left panel) and RGS2 (right panel). Note that TRH had little effect on conductance in cells expressing either of these constructs, which interfere with signaling via G␣ q family subunits. B, the effect of TRH on GIRK channel conductance in cells expressing the G␤ 5 sinks RGS6 and RGS11, which inhibit signaling by G␤ 5 -containing G␤␥ dimers (21). TRH caused a strong inhibition of GIRK channel conductance in these cells. C, summary data showing inhibition of GIRK channel conductance by TRH (percent of control conductance; means Ϯ S.E.) in cells transfected with constructs that disrupt signaling by G␣ q family subunits (i.e. G␣ q sinks) or G␤␥ dimers (i.e. G␤␥ sinks). Asterisks denote statistically significant differences from control cells (p Ͻ 0.05 by ANOVA with post hoc Bonferroni's test; the number of cells in each group is indicated in parentheses). Note that inhibition of GIRK channel currents by TRH was significantly diminished by the G␣ q sinks, but not by the G␤␥ sinks, consistent with a primary role for G␣ q in mediating the effects of TRH. associated with expression of G␣ q * was significantly attenuated in cells treated with the PLC inhibitor U73122. As a control, inhibition of GIRK channel currents by G␣ q * was unaffected by identical treatment with the inactive analog, U73343. Thus, activation of PLC by TRH and G␣ q * contributes, at least in part, to GIRK channel inhibition.
Lowering Plasma Membrane PIP 2 Levels Diminishes Receptor-mediated GIRK Channel Inhibition-It has recently been established that the membrane phospholipid PI(4,5)P 2 is an important coactivator of GIRK channels (9, 10, 31), and it has been suggested that PLC-mediated decreases in PI(4,5)P 2 levels might account for GIRK channel inhibition following receptor activation in atrial cells (15)(16)(17). Since we found that TRHinduced inhibition of GIRK channel currents also appears to involve PLC, we tested the possibility that GIRK channel inhibition might be sensitive to membrane PI(4,5)P 2 levels.
To this end, we used two minigenes that deplete membrane PI(4,5)P 2 levels via different mechanisms (23). The first was a GFP-tagged fusion protein that includes the pleckstrin homology domain of PLC␦, a construct that sequesters PI(4,5)P 2 ; as a control for PLC␦-PH, we used a different GFP-tagged pleckstrin homology domain derived from AKT (AKT-PH), which interacts with a separate isoform of the membrane phospholipid PI(3,4)P 2 . The second construct was a GFP fusion protein that included a constitutively active yeast 5Ј-PI-PTase that was targeted to the plasma membrane by incorporation of a myristoylation-palmitoylation sequence. When tested within 24 h of transfection, both the PLC␦-PH and 5Ј-PI-PTase constructs are localized to the plasma membrane, and both lower membrane PI(4,5)P 2 levels (23). As shown in the example records of Fig. 6, TRH caused very little inhibition of GIRK channel currents in cells expressing either the membrane-targeted 5Ј-PI-PTase (Fig. 6A, upper panel) or PLC␦-PH (middle panel) construct, but was highly effective in cells expressing the control AKT-PH construct (lower panel). Indeed, whereas GIRK channel inhibi-tion by TRH was Ͼ55% in nearly all control cells (38 of 41 cells tested; 93%) and AKT-PH-transfected cells (9 of 11 cells tested; 82%), TRH inhibition was Ͼ55% in only one 5Ј-PI-PTase-transfected cell (of 12 cells tested; 8%) and in just four PLC␦-PHtransfected cells (of 15 cells tested; 27%). These differences were also reflected in summary data presented in Fig. 6B, in which the averaged TRH-induced inhibition of GIRK channel currents was significantly diminished in cells expressing the two inhibitors of PIP 2 compared with either control cells or cells expressing AKT-PH.
It should be pointed out that the basal GIRK channel conductance tended to be somewhat lower in cells transfected with 5Ј-PI-PTase or PLC␦-PH compared with control cells (ϳ4 versus ϳ7 nanosiemens), as expected given the demonstrated facilitating effects of PI(4,5)P 2 on GIRK channels (9 -11). In many cells, however, a substantial basal GIRK channel conductance was retained (e.g. see records in Fig. 6A, upper and middle panels), suggesting that differences in initial current amplitudes could not account for the diminished effect of TRH. It is also noteworthy that PLC␦-PH can bind IP 3 (32), a product of PLC-mediated hydrolysis. Although it is therefore conceivable that buffering IP 3 could be responsible for inhibitory effects of the PLC␦-PH construct (but see below), this could not account for the effects of 5Ј-PI-PTase. Thus, depletion of membrane PI(4,5)P 2 levels appears to disrupt receptor-mediated inhibition of GIRK channel currents.
Receptor-mediated Inhibition of GIRK Channel Currents Is Independent of PKC or IP 3 -The data presented up to this point indicate a role for PLC and its substrate, PI(4,5)P 2 , in GIRK channel inhibition by TRH receptors. Therefore, we tested if signaling pathways typically activated as a result of PLC-mediated PI(4,5)P 2 hydrolysis are involved in GIRK channel current inhibition. We were particularly interested in investigating a role for PKC since it has earlier been implicated in inhibition of recombinant GIRK channels (Ref. 33, but see  , ϳ36 -48 h) compared with that in G␣ q *-transfected cells that were untreated or treated with U73122 or U73343. GIRK channel conductance was smaller in all cells transfected with G␣ q *, but this effect was significantly diminished in cells treated with the PLC inhibitor U73122 (i.e. the current reduction was not as great). *, statistically significant difference from control cells; ‡, statistically significant difference different from G␣ q *-transfected cells (p Ͻ 0.05 by ANOVA with post hoc Bonferroni's test; the number of cells in each group is indicated in parentheses). Fig. 7A, neither the PKC-activating phorbol ester PDBu nor its inactive analog, 4␣-PDBu, had any effect on GIRK channel currents (both applied via the perfusate at 1 M). Because some PKC isozymes are insensitive to phorbol esters, we also tested if a competitive inhibitor of ATP binding to the catalytic site of PKC, the bisindolylmaleimide compound GF 109203X, could interfere with TRH-induced inhibition of GIRK channel currents. As depicted in the records from the representative cell shown in Fig. 7B (left panel), TRH was highly effective at inhibiting GIRK channel currents in cells pretreated with bisindolylmaleimide (5 M for Ն30 min). Indeed, TRH inhibited 81.1 Ϯ 2.6% of the initial GIRK channel conductance in bisindolylmaleimide-treated cells (n ϭ 8), clearly not less than the TRH-induced inhibition in control cells (ϳ73%). In control experiments using HEK293 cells expressing mouse TREK-1 channels, we found that PDBu strongly inhibited TREK-1 currents and that bisindolylmaleimide blocked those effects of PDBu, attesting to the efficacy of both PDBu and bisindolylmaleimide compounds (data not shown). These data argue against a role for PKC in TRH receptor-mediated inhibition of GIRK channel currents.

Ref. 34). As illustrated in
Activation of PLC also leads to production of IP 3 and increases in intracellular calcium, either of which could conceivably lead to GIRK channel inhibition. It is unlikely that changes in intracellular calcium contribute to the effects of TRH since our whole cell experiments were performed with intracellular calcium buffered to ϳ10 Ϫ8 M by including 10 mM EGTA in the internal solution. To determine if IP 3 could have effects independent of altering intracellular calcium, we introduced IP 3 into cells by including it in the recording pipettes (100 M). As illustrated in the example cells of Fig. 7C, we found that the time-and TRH-dependent decreases in GIRK channel current in IP 3 -dialyzed cells were not different from those in control cells recorded with pipettes that did not contain IP 3 . As is evident in the averaged data of Fig. 7C, the TRHinduced inhibition of GIRK channel currents in IP 3 -containing cells (75.5 Ϯ 4.1%, n ϭ 7) was essentially identical to that in control cells (i.e. ϳ73%). Thus, intracellular perfusion with IP 3 neither mimicked nor occluded the TRH receptor-mediated inhibition of GIRK channel currents.
Receptor-mediated GIRK Channel Inhibition Involves a Readily Diffusible Messenger-Our experiments provide no evidence to implicate the major signaling pathways downstream of PLC (either PKC or IP 3 /Ca 2ϩ ) in receptor-mediated GIRK channel inhibition. To test whether other diffusible messengers might be involved, we recorded the effects of bath-applied TRH on GIRK channel activity in cell-attached patches. In this recording configuration, the agonist does not have direct access to channels recorded within the patch. Therefore, any actions of the agonist must be transmitted by some intermediary that can move between the channels inside the patch and the receptors outside the patch, either via the cytosol or within the plane of the membrane.
Cell-attached patch recordings of K ϩ channels were obtained in G1/4 cells either from the cell line stably expressing the TRH-R1 and 5-HT 1A receptors (Fig. 8A) or from the parental G1/4 cell line, which does not express these receptors (Fig. 8B). Single channel currents were clearly inwardly rectifying (data not shown), with a unitary conductance of ϳ40 picosiemens  , two constructs that deplete membrane PI(4,5)P 2 , as well as a control AKT-PH construct that does not bind to PI(4,5)P 2 (lower panel). Note that TRH had little effect on conductance in cells expressing either of the constructs that lower membrane PI(4,5)P 2 levels, but was highly effective in the cell expressing the control construct that does not bind PI(4,5)P 2 . B, summary data quantifying the effects of TRH on GIRK channel currents in cells transfected with the indicated constructs. Note that relative to the control, inhibition of GIRK channel currents by TRH was significantly diminished by the two constructs that lower membrane PI(4,5)P 2 levels (0.9 Ϯ 0.1 ms, n ϭ 9) (Fig. 8C), as expected for GIRK1/4 channels (35). Moreover, when exposed to extracellular 5-HT in an inside-out patch configuration, the channels were vigorously activated by adding GTP to the inside face of the patch (data not shown). These are defining properties of GIRK channels (1,2,8).
As illustrated in Fig. 8A (left panel), GIRK channel activity (i.e. NP o ) in a cell-attached patch from a TRH-R1-expressing G1/4 cell was strongly diminished by bath-applied TRH. Based on our whole cell recordings, we expect that the TRH-induced decrease in basal GIRK channel currents would cause a membrane depolarization of ϳ30 mV. To correct for effects of this membrane depolarization on the patch potential, we adjusted the patch holding potential by Ϫ30 mV in the presence of TRH. This adjustment was apparently appropriate since the unitary current was approximately the same amplitude in the presence of TRH, after correcting for the membrane depolarization (see dashed line in sample records of Fig. 8A, right panel). Furthermore, since channel activity remained strongly inhibited even after this correction, the inhibition by TRH could not be attributed solely to a TRH-induced change in membrane potential.
TRH inhibited GIRK channel activity by Ͼ55% in 8 of 10 cell-attached patches, with an average inhibition in all patches of ϳ68% (NP o decreased from 0.09 Ϯ 0.02 to 0.03 Ϯ 0.02, n ϭ 10; p Ͻ 0.05) (Fig. 8C, middle panel), which is very similar to the magnitude of TRH-induced inhibition of whole cell GIRK channel currents. In those patches with a robust inhibition of channel activity by TRH (n ϭ 8), the inhibition involved, at least in part, a decrease in mean channel open time (Fig. 8C,  right panel). As expected, there was no effect of TRH on GIRK channel activity in cell-attached patches from control G1/4 cells that did not express TRH-R1, as shown in Fig. 8 (B and C,  middle panel). These data therefore indicate that GIRK channel inhibition involves a readily diffusible messenger. DISCUSSION In this study, we examined dual regulation of GIRK channels using mammalian cell lines that express GIRK1/4 channels together with two G protein-coupled receptors that couple preferentially to distinct classes of G␣ subunit. Unlike GIRK channel activation by G␣ i/o -coupled receptors, which is mediated by G␤␥ dimers (reviewed in Refs. 1, 2, and 8), we found that GIRK

FIG. 7. Receptor-mediated GIRK channel inhibition is independent of PKC activation and intracellular IP 3 .
A, the effect on GIRK channel conductance of a PKC-activating phorbol ester (PDBu) and its inactive analog (4␣-PDBu) in a representative cell (left panel). The compounds were applied in the bath (1 M) as indicated; neither had any effect on GIRK channel currents, as is also apparent in averaged data from cells treated with the two phorbol ester compounds (percent of control conductance; means Ϯ S.E.) (right panel). B, the effect of a bisindolylmaleimide inhibitor of PKC (GF 109203X) on TRH-induced inhibition of GIRK channel currents in a representative cell pretreated with 5 M bisindolylmaleimide for Ն 30 min (left panel). TRH caused a strong inhibition of GIRK channel currents in this bisindolylmaleimidetreated cell; averaged values for the magnitude of TRH-induced inhibition (percent of control conductance; means Ϯ S.E.) indicated that it was equal in magnitude in control and bisindolylmaleimide-treated cells (right panel). C, the effect of IP 3 (100 M in the whole cell recording pipette) on GIRK channel conductance and on TRH-mediated inhibition of GIRK channel conductance in a representative cell (left panel). The conductance at all time points (Gt) was normalized (norm.) to the initial conductance (Gi) to illustrate the close correspondence of the current rundown and the inhibition by TRH in control and IP 3 -treated cells. Averaged data show the magnitude of TRH-induced inhibition (percent of control conductance; means Ϯ S.E.) in these experiments and indicate that there was no difference between control cells and those recorded with IP 3 in the pipette (right panel). channel inhibition by G␣ q -coupled receptors involves principally the G␣ subunit. Exogenous expression of constitutively active G␣ subunits revealed little selectivity within the G␣ q family for GIRK channel inhibition, but clear specificity for G␣ q family subunits over all other classes of G␣ subunits. Moreover, receptor-mediated inhibition of GIRK channel currents was diminished by minigene constructs that interfere with G␣ q signaling, but not by those that target G␤␥ subunits. Signaling downstream of G␣ q appeared to involve PLC since receptorand G␣ q *-mediated GIRK channel inhibition was reduced by the PLC inhibitor U73122 and because GIRK channel inhibition by G␣ q -coupled receptors was diminished by two different constructs that reduce membrane PIP 2 levels. We found no evidence for involvement of PKC, IP 3 , or intracellular Ca 2ϩ in GIRK channel inhibition, although cell-attached patch recordings indicated involvement of a readily diffusible messenger. These data support burgeoning evidence that PIP 2 , itself a potent activator of GIRK channels (9 -11), represents a diffusible signal responsible for GIRK channel inhibition by G␣ qcoupled receptors (15)(16)(17). It remains possible, however, that receptor activation of G␣ q subunits and PLC initiates a separate signaling pathway, yet undefined, that leads to GIRK channel inhibition.
G␣ q Family Subunits Mediate GIRK Channel Inhibition-In and plotted as a function of time (lower trace); TRH was added to the perfusate for the indicated period. The patch holding potential is provided (upper trace). Note that bath-applied TRH caused a strong inhibition of channel activity within the patch. This was true even after the patch potential was adjusted (to E m Ϫ60 mV) to correct for the anticipated TRH-induced depolarization of cell membrane potential (30 mV). For clarity, periods of transient instability in the patch recordings were blanked (indicated by stars). Right panel, shown are the continuous records (ϳ1 s) of GIRK channel activity under control conditions (at a patch potential of E m Ϫ30 mV) and in the presence of TRH (at a patch potential of E m Ϫ60 mV). Note that the membrane potential correction resulted in current amplitudes in TRH that approximated those in the control (see the dashed line). B: the effect of TRH on GIRK channels recorded in a cell-attached patch of G1/4 cells not expressing TRH-R1. Left panel, GIRK channel activity was quantified as NP o (in 1-s bins) and plotted as a function of time (lower trace); TRH was added to the perfusate for the indicated period. The patch holding potential is provided (upper trace), and periods of brief patch instabilities were blanked (stars). As expected, TRH had no effect in control cells that do not have TRH receptors. Right panel, shown are continuous records (ϳ1 s) of GIRK channel activity under control conditions and in the presence of TRH (at a patch potential of E m Ϫ30 mV). C: left panel, shown is a schematic of the experimental configuration. To modulate channels contained within the patch, activation of receptors outside the patch must either produce an inhibitory mediator that can diffuse to the channels or induce the withdrawal of an activating substance from the patch. Middle panel, shown are the averaged data (percent inhibition of NP o by TRH; means Ϯ S.E.) in patches from G1/4R cells (i.e. with TRH-R1; n ϭ 10) and from control G1/4 cells (i.e. without TRH-R1; n ϭ 6). TRH inhibited channel activity by ϳ68%, similar to the TRH-induced inhibition of whole cell GIRK channel currents. Right panel, mean channel open time (ϮS.E.) was determined before and during TRH application in those patches of G1/4R cells that showed a robust response to TRH (n ϭ 8). Note that TRH caused a significant decrease in mean open time (determined by paired t test). native systems, agonist-evoked GIRK channel current inhibition is typically associated with receptors that couple via PTxinsensitive G␣ q subunits (e.g. ␣ 1 -adrenoreceptors in atrial myocytes (3) and NK1 receptors in locus ceruleus neurons (6)). Accordingly, we found that receptor-mediated GIRK channel inhibition was not affected by PTx, but was diminished by PMT, a toxin that selectively uncouples G␣ q -type subunits from receptors (25,26). Moreover, we found that GIRK channel inhibition was a property unique among G␣ subunits to members of the G␣ q family; constitutively active mutants of G␣ q family subunits caused essentially complete inhibition of basal GIRK channel currents in G1/4 cell lines stably expressing GIRK1/4 channels, whereas the corresponding constitutively active members of other G␣ subunit families were without effect. These data are consistent with most previous studies in which other (non-G␣ q family) G␣ subunits pre-activated with GTP␥S had little effect on native and/or recombinant GIRK channels in excised patches (1,8,36). In one earlier study, however, purified recombinant G␣ s -GTP␥S and G␣ i1 -GTP␥S, but not G␣ i2 -GTP␥S or G␣ i3 -GTP␥S, inhibited G␤␥-activated GIRK channels in inside-out patches from Xenopus oocytes (37). We found no effect of any of these subunits in our whole cell assay using G␣ subunits activated by a Gln 3 Leu point mutation. Although the reason for these discrepant findings is not entirely clear, it is possible that the inhibition of GIRK channels observed in oocyte patches with G␣ i1 and G␣ s was due to contamination with non-activated G␣-GDP subunits, which will sequester G␤␥ and indirectly inhibit G␤␥-activated GIRK channel activity. Indeed, we have found that multiple different wild-type G␣ subunits (i.e. those not activated by the Gln 3 Leu mutation), including G␣ i and G␣ s subunits, strongly inhibit basal GIRK channel currents when expressed in G1/4 cells (12). Moreover, it is important to point out that those earlier results are difficult to reconcile with current understanding of receptor signaling to GIRK channels since receptors that couple to G␣ s typically have no effect on GIRK channel currents, and those that couple to G␣ i cause GIRK channel activation, not inhibition (1,36,38). By contrast, our results with G␣ q * and closely related subunits are entirely concordant with observations that inhibition of GIRK channels in native settings involves receptors that couple via PTx-insensitive G␣ proteins of the G␣ q family (1).
Our results provide additional evidence indicating that it is indeed the G␣ subunit, and not the associated G␤␥ dimer, that provides the signal for receptor-mediated GIRK channel inhibition. Thus, we showed that so-called "effector antagonists" of G␣ q subunits (i.e. RGS2 and PLC␤1-ct) substantially decreased GIRK channel current inhibition evoked by TRH-R1 stimulation. Likewise, in atrial myocytes, a peptide that interferes with G␣ q -mediated PLC activation disrupted GIRK channel current desensitization by muscarinic receptors (15). By contrast, expression of constructs that sequester G␤␥ subunits relatively non-selectively (i.e. G␣ t and ␤ARK-ct), as well as those that bind G␤ 5 specifically (i.e. RGS6 and RGS11), did not disrupt receptor-mediated GIRK channel current inhibition. This result is perhaps not surprising since most combinations of G␤ and G␥ subunits are known to activate, rather than inhibit, GIRK channels (12,27,28). However, we showed here that activation of PLC, a known G␤␥ effector (8), is necessary for GIRK channel inhibition, and we recently found that G␤␥ dimers including the G␤ 5 subunit are unique in causing inhibition of GIRK channels, perhaps by competing with activating G␤␥ pairs for binding to the channel (12,14). Nevertheless, we found no evidence to indicate that receptor-mediated inhibition of GIRK channel currents requires downstream effects of G␤␥ dimers, including G␤ 5 ␥ pairs. If G␤ 5 -containing dimers do as-sociate preferentially with G␣ q in vivo (13) and exclude GIRK channel-activating G␤␥ pairs from receptor-bound heterotrimers, this function is apparently not required for receptor-mediated GIRK channel inhibition. Moreover, these data indicate that G␤␥ subunits do not contribute appreciably to the PLC activation that appears to be necessary for receptor-mediated GIRK channel inhibition.
GIRK Channel Inhibition May Occur by PLC-mediated Decreases in PIP 2 -Consistent with the demonstrated involvement of G␣ q family subunits, our data also indicate a key role for PLC, the major downstream effector of G␣ q , in receptormediated GIRK channel current inhibition. Thus, agonist-induced GIRK channel inhibition was diminished pharmacologically by U73122, a PLC inhibitor, and also disrupted by minigene constructs that lower membrane levels of a PLC substrate, PIP 2 . Despite this, however, we found no evidence to implicate further effectors in the classical signaling pathway downstream of PLC since GIRK channel inhibition was apparently independent of effects on PKC, IP 3 , and intracellular calcium. Similarly, PLC involvement has been demonstrated in other studies of receptor-mediated GIRK channel inhibition (15,16), but there has been no evidence to indicate that either IP 3 or Ca 2ϩ plays a role (3,4,17,33). Although it was suggested that an atypical PKC could mediate inhibition of GIRK channels expressed in Xenopus oocytes (33), in other native and heterologous expression systems, receptor-mediated inhibition of GIRK channel currents was independent of PKC activation (3,4,16,17,34), as we found here.
Therefore, inasmuch as our data implicate PLC, but exclude its usual downstream mediators, and given compelling evidence indicating that PIP 2 is a potent activator of GIRK channels (9 -11), the data support the following mechanism for receptor-mediated GIRK channel inhibition: decreased membrane concentrations of PIP 2 that ensue following hydrolysis by receptor-activated PLC result in GIRK channel inhibition by promoting release of PIP 2 from its channel binding site. This same mechanism was proposed to account for muscarinic inhibition of K ATP channels in COS-7 cells (39) and, more recently, for inhibition of native GIRK channels in atrial myocytes by a variety of G␣ q -coupled receptors (i.e. m3 muscarinic, ␣ 1 -adrenergic, and endothelin-1 receptors) (15)(16)(17). As in the present work with recombinant GIRK1/4 channels, the molecular correlate of those atrial GIRK channels (35), involvement of PIP 2 was usually inferred by exclusion when PLC was implicated in receptor-mediated GIRK channel current inhibition, but downstream mediators were not. In each study, a different experimental approach was used to test PIP 2 involvement; for example, receptor-mediated inhibition of atrial GIRK channel currents was disrupted by using the phosphatidylinositol kinase inhibitor wortmannin to lower membrane PIP 2 levels (16) or by using intracellular perfusion of excess PIP 2 to overwhelm the GIRK channel PIP 2 -binding site (17). Using a particularly elegant and non-pharmacological approach, it was found that M1 muscarinic receptor-mediated inhibition of homomeric mutant GIRK4 channels expressed in Xenopus oocytes was diminished by amino acid substitutions that increased the affinity of the channel for PIP 2 (15). In accord with these earlier results, we found here that GIRK channel inhibition was disrupted by membrane-targeted 5Ј-PI-PTase and PLC␦-PH, two constructs that lower PIP 2 levels (23). It is also interesting to note that PIP 2 -induced increases in GIRK channel activity are associated with increased mean channel open times (9), and further consistent with a role for diminished levels of PIP 2 , we found that receptor-mediated GIRK channel inhibition involved decreased mean channel open time (see Fig. 8C). In sum, it therefore appears that data from both native and heterologous systems converge on the conclusion that decreases in PIP 2 could account for receptor-mediated GIRK channel inhibition. However, it is important to point out that this interpretation is based on indirect tests of PIP 2 involvement and by exclusion of other pathways downstream of PLC; since all other potential signaling pathways have not yet been excluded, the possibility remains that some other mechanism for inhibition of GIRK channels by G␣ q -coupled receptors may yet be discovered. PIP 2 as a Diffusible Mediator of GIRK Channel Inhibition-We found that bath-applied TRH was able to inhibit GIRK1/4 channels in cell-attached patches. Similar results were noted for native GIRK channel inhibition in atrial myocytes by ␣ 1 -adrenoreceptors and endothelin-1 receptors (3,16,17) as well as for metabotropic glutamate receptor inhibition of recombinant GIRK channels in Xenopus oocytes (33). These type of results are usually interpreted to imply that cytosolic second messengers are involved in the modulatory mechanism. However, if PIP 2 maintains its membrane association, the GIRK channel inhibition would be membrane-delimited even though it was initiated by G␣ q -coupled receptors outside the patch. By contrast, membrane-delimited receptor activation of GIRK channels, mediated by direct binding of G␤␥ subunits to the channels, requires agonist stimulation of G␣ i/o -coupled receptors within the patch (2,40). Thus, if decreased membrane PIP 2 levels are indeed involved in GIRK channel inhibition, as we and others now suggest (15)(16)(17), the implication is that PIP 2 may be more readily diffusible within the membrane than are G␤␥ dimers. In this respect, it has been suggested that the non-activated G protein heterotrimer may serve to tether G␣ i/o -coupled receptors in close proximity to GIRK channels (41), and this could keep membrane signaling by G␤␥ dimers relatively more localized than that by PIP 2 .
Dual Regulation of GIRK Channels-In cells of the heart and brain, dual regulation of native GIRK channels by receptors that couple via distinct classes of the G␣ subunit represents a mechanism by which convergent effects on a single target can provide dynamic control of cell excitability. Here, we used a heterologous expression system that fully recapitulates dual regulation in order to demonstrate that receptor-mediated inhibition of GIRK channels is mediated by G␣ q subunits and to support the proposition that it may be due to PLC-catalyzed decreases in membrane PIP 2 levels. This mechanism is unlike that causing GIRK channel activation, which is mediated by G␣ i/o -coupled receptors and involves direct interactions of G␤␥ dimers with GIRK channels. Interestingly, a similar dual regulation of AKT by G proteins was recently discovered; as with GIRK channel modulation, AKT was activated by G␤␥ dimers and inhibited by G␣ q (42). Thus, opposing actions of G␣ and G␤␥ to provide up-and down-regulation of a single effector may be a more prevalent phenomenon than is currently appreciated.