Novel Inhibition of G bg -activated Potassium Currents Induced by M 2 Muscarinic Receptors via a Pertussis Toxin-insensitive Pathway*

G i protein-coupled receptors such as the M 2 musca- rinic acetylcholine receptor (mAChR) and A 1 adenosine receptor have been shown to activate G protein-acti-vated inwardly rectifying K 1 channels (GIRKs) via pertussis toxin-sensitive G proteins in atrial myocytes and in many neuronal cells. Here we show that muscarinic M 2 receptors not only activate but also reversibly in- hibit these K 1 currents when stimulated with agonist for up to 2 min. The M 2 mAChR-mediated inhibition of the channel was also observed when the channels were first activated by inclusion of guanosine 5 * - O -(thiotri-phosphate) in the pipette. Under these conditions the M 2 mAChR-induced inhibition was quasi-irreversible, suggesting a role for G proteins in the inhibitory process. In contrast, when GIRK currents were maximally activated by co-expressing exogenous G bg , the extent of acetylcholine (ACh)-induced inhibition was significantly reduced, suggesting competition between the re-ceptor-mediated inhibition and the large pool of available G bg subunits. The signaling pathway that led to the ACh-induced inhibition of GIRK channels was unaf-fected by pertussis toxin pretreatment. Furthermore, the internalization and agonist-induced phosphorylation of M 2 mAChR was not required because a phospho- rylation- and internalization-deficient mutant of the M 2 mAChR

G i protein-coupled receptors such as the M 2 muscarinic acetylcholine receptor (mAChR) and A 1 adenosine receptor have been shown to activate G protein-activated inwardly rectifying K ؉ channels (GIRKs) via pertussis toxin-sensitive G proteins in atrial myocytes and in many neuronal cells. Here we show that muscarinic M 2 receptors not only activate but also reversibly inhibit these K ؉ currents when stimulated with agonist for up to 2 min. The M 2 mAChR-mediated inhibition of the channel was also observed when the channels were first activated by inclusion of guanosine 5-O-(thiotriphosphate) in the pipette. Under these conditions the M 2 mAChR-induced inhibition was quasi-irreversible, suggesting a role for G proteins in the inhibitory process. In contrast, when GIRK currents were maximally activated by co-expressing exogenous G␤␥, the extent of acetylcholine (ACh)-induced inhibition was significantly reduced, suggesting competition between the receptor-mediated inhibition and the large pool of available G␤␥ subunits. The signaling pathway that led to the ACh-induced inhibition of GIRK channels was unaffected by pertussis toxin pretreatment. Furthermore, the internalization and agonist-induced phosphorylation of M 2 mAChR was not required because a phosphorylation-and internalization-deficient mutant of the M 2 mAChR was as potent as the wild-type counterpart. Pharmacological agents modulating various protein kinases or phosphatidylinositol 3-kinase did not affect the inhibition of GIRK currents. Furthermore, the signaling pathway that mediates GIRK current inhibition was found to be membrane-delimited because bath application of ACh did not inhibit GIRK channel activity in cell-attached patches. Other G protein-coupled receptors including M 4 mAChR and ␣ 1A adrenergic receptors also caused the inhibition, whereas other G proteincoupled receptors including A 1

and A 3 adenosine receptors and ␣ 2A and ␣ 2C adrenergic receptors could not induce the inhibition. The presented results suggest the existence of a novel signaling pathway that can be activated selectively by M 2 and M 4 mAChR but not by adenosine receptors and that involves non-pertussis toxinsensitive G proteins leading to an inhibition of G␤␥activated GIRK currents in a membrane-delimited fashion.
An enormous variety of G protein-coupled receptors (GPCRs) 1 allow for a large number of extracellular signals to converge on a relatively small number of heterotrimeric G proteins (1), which in turn cause activation of downstream effectors. When a cell receives multiple signals that are transduced through different GPCRs, the integration of multiple signaling events is complex not only because of the fact that the G proteins have multiple effector systems but also because of "cross-talk" between signaling pathways. Recently, evidence has emerged that cross-talk between GPCRs not only can occur between pathways that utilize different classes of G proteins but also within one G protein family (2). The current study presents a new case of cross-talk between signals that are produced by activation of adenosine and muscarinic receptors that are known to couple to the same class of G proteins.
The initial goal of this study was to investigate the regulation and the desensitization of M 2 mAChR-and A 1 adenosine receptor-activated K ϩ channels that give rise to the current known as I KACh . I KACh channels were first identified in the supraventricular tissue of the heart, and their regulation by M 2 mAChRs and other GPCRs, such as the A 1 adenosine or lysosphingolipid receptors, has been the topic of many different studies over the past two decades (3)(4)(5). I KACh channels were characterized as G protein-activated inwardly rectifying K ϩ channels (GIRKs) that are activated by pertussis toxin (PTX)sensitive G proteins in a membrane-delimited fashion (3, 6 -9). It has become clear that GIRK channels are activated upon binding of the G␤␥ subunits to the channel (10 -12). GIRK channels are heteromultimers of two homologous subunits. GIRK1 and GIRK4 form the cardiac channel (13,14), whereas GIRK1 and GIRK2 or GIRK3 form certain neuronal channels (15,16). The GIRK channels have been functionally expressed in a variety of cells including Xenopus oocytes and mammalian HEK and CHO cells (13,(17)(18)(19).
Desensitization of GPCR-activated GIRK currents has been observed in native tissues and in heterologous expression systems. We have utilized the GIRK channels as a readout system to study desensitization of G protein-mediated signals for two reasons. First, electrophysiological recording of GIRK currents with the patch clamp technique allows for real time measurement of G protein-mediated signals in intact cells. Second, the GIRK channels conduct potassium ions, which diffuse very rapidly, and the concentration of K ϩ can be controlled; therefore, even after activation of GIRK channels for a long period of time, depletion of intra-or extracellular K ϩ does not occur, and hence desensitization at the level of the channels does not play a major role in desensitization events. Desensitization of GPCR-generated signals can be homologous, where only the stimulated GPCR desensitizes, or heterologous, where the signaling by heterologous GPCRs is inhibited. To distinguish between homologous and heterologous desensitization processes, two different G i -coupled receptors (A 1 adenosine and M 2 mAChR) were co-transfected into HEK293 cells with GIRK1 and GIRK4. Before and after an initial treatment of one of these receptors with agonist, the response to stimulation of the other receptor was determined. If the response of the treated receptor only was reduced after the initial treatment with agonist, this was considered homologous desensitization. However, if there was a reduction of the response of the untreated receptor after an initial treatment of the other receptor, this was considered to reflect heterologous desensitization. We report novel observations concerning heterologous desensitization of GPCR-activated GIRK channels.
The transfection method used in this study was based on a method (22) in which replication-deficient adenoviruses lacking the E1 gene were coated with DEAE-dextran and mammalian expression vectors to import the cDNAs of interest. Ad5dl312 viruses (a gift from Dr. K. Rundell) were allowed to replicate in HEK293 cells and were harvested when 70 -90% of the HEK293 cells were rounded up. The cells were centrifuged gently, and the pellet was resuspended in 1 ml of Ham's F-12 medium/10-cm dish of cells. For cell lysis, cells were frozen and thawed three times and centrifuged. The supernatant was used as the viral stock solution. The viral stock (10 l) was diluted with 130 l of Ham's F-12 medium, and 80 g of DEAE-dextran was added. The mixture was incubated at room temperature for 5-10 min. The total amounts of plasmid DNA (2.2 g) were added (empty pcDNA 3.1 was used to balance cDNA amounts), and the mixture was incubated for another 2-5 min. Ham's F-12 medium (2 ml) was added, and this transfection solution was added onto one plate of 60 -70% confluent CHO K1 cells. Prior to the incubation the cells were washed three times with 3 ml of serum and antibiotic-free medium. All assays were performed 48 -72 h posttransfection.
Measurement of Membrane Currents-Membrane currents were recorded under voltage clamp using conventional whole cell-patch clamp techniques (23). Patch pipettes were fabricated from borosilicate glass capillaries (GF-150 -10, Warner Instruments) using a horizontal puller (P-95, Fleming & Poulsen). The DC resistance of the filled pipettes ranged from 2 to 5 megohms. Membrane currents were recorded using a patch clamp amplifier (Axopatch 200, Axon Instruments). Signals were analog-filtered using a low-pass Bessel filter (1-3 kHz corner frequency). Data were digitally stored using an IBM-compatible PC equipped with a hardware/software package (ISO2 by MFK, Frankfurt/ Main, Germany) for voltage control, data acquisition, and data evaluation. I KACh was measured as an inward current using a holding potential of Ϫ90 mV as described (24). Voltage ramps (from Ϫ120 mV to ϩ60 mV in 500 ms, every 10 s) were used to determine current-voltage relationships.

RESULTS
ACh Induced an Inhibition of Ado-evoked GIRK Currents-M 2 mAChRs, as well as A 1 adenosine receptors, can activate inwardly rectifying potassium currents carried by GIRK channels when the receptors and channel subunits are transiently co-expressed in CHO K1 cells. Agonist-activated currents were measured, using the whole cell-patch clamp configuration, as inward currents by setting the membrane potential negative (Ϫ90 mV) to the potassium equilibrium potential of about Ϫ50 mV (20 mM external potassium). In cells transfected with the GIRK subunits and the M 2 mAChRs, application of ACh (1 M) caused an activation of GIRK currents that rapidly desensitized during a 2-min application of ACh (Fig.  1A, left panel). In contrast, in cells transfected with the GIRK subunits and the A 1 adenosine receptors, rapid superfusion of the cells with adenosine (1 M) induced an inwardly rectifying K ϩ current that exhibited only a marginal desensitization after a 2-min application of adenosine, and after washout of the agonist, GIRK currents deactivated to basal levels ( Fig. 1A, right panel).
To determine whether the desensitization of the ACh response took place at the level of the M 2 mAChR or downstream of it, we analyzed the effects of a series of successive exposures to Ado and ACh in cells co-expressing both the A 1 adenosine receptors and the M 2 mAChRs (Fig. 1B). An initial exposure to adenosine induced GIRK currents that were stable after an initial rapid, small desensitization (this type of desensitization is not the topic of this study but was studied in recent papers (25)(26)(27)). After washout of Ado and a return of currents to basal levels, the cells were exposed to ACh. The ACh-induced currents were of similar amplitude compared with those induced by Ado; however, again the ACh-induced GIRK currents markedly desensitized during the 2-min exposure to ACh. To test if the ACh treatment would affect subsequent effects of Ado, we exposed the cells to a second application of Ado within 1 min after the washout of ACh (Fig. 1B). These Ado-elicited GIRK currents were substantially reduced compared with the initial Ado response (Fig. 1B), demonstrating that the prior treatment with ACh caused an inhibition of the subsequent Ado response. That the ACh-induced reduction of the Ado response was attributable to the preceding stimulation of M 2 mAChR receptors was demonstrated by the observation that in cells lacking the M 2 mAChRs, no inhibition of the Ado response by pretreatment with ACh was observed (Fig. 1C). In addition, prior treatment with Ado did not substantially reduce the effects of a second stimulus with Ado ( Fig. 1C), indicating that the activation of GIRK currents itself was not sufficient to cause desensitization. Furthermore, that prior application of Ado was not responsible for the M 2 mAChR-induced desensitization and inhibition of the GIRK currents was demonstrated by results showing similar responses elicited by the M 2 mAChR in cells not previously stimulated with Ado ( Fig. 1A and data not shown). The summarized results from a series of similar experiments demonstrated that prior exposure to ACh caused a reduction of the subsequent Ado response to 56 Ϯ 4% (n ϭ 13) of the initial Ado response, whereas prior exposure to Ado only caused a reduction of 5 Ϯ 2.2% (n ϭ 7) (Fig. 1D). The results suggested that activation of the mAChRs caused an initial activation followed by an inhibition of GIRK currents and that the inhibition persisted during a subsequent exposure to Ado and thus diminished the response to Ado.
The Inhibition of Ado-evoked GIRK Currents via Stimulation of M 2 mAChR Is Reversible-To test if the ACh-induced inhibition of the Ado currents was reversible, we performed similar experiments as described above, in which we tested the effects of successive treatments with Ado at different times following a desensitizing treatment with ACh. The cells were co-transfected with the GIRKs and both receptors and initially exposed to a brief pulse of Ado, which activated the currents (Fig. 2). After washout of Ado, the cells were exposed to ACh for 2 min during which there was an initial activation of the GIRK currents to ϳ75% of the extent observed with Ado, followed by a substantial inhibition (Fig. 2). Immediately following washout of the ACh the cells were exposed briefly to Ado, and the response was substantially reduced compared with the initial Ado response (Fig. 2). However, the size of the Ado-induced currents increased over time following the washout of ACh, as the subsequent treatments with Ado resulted in larger currents. After 6 min following the washout of ACh, the response to Ado was almost comparable with that of the initial Ado response (Fig. 2, last trace). These results demonstrated that the ACh-induced inhibition of the currents was reversible, albeit with a relatively slow time scale (t1 ⁄2 ϭ 2-5 min).

The M 2 mAChR-induced Inhibition of Ado-evoked GIRK Currents Occurs Downstream of Adenosine Receptors-To test
whether the ACh-induced inhibition of the currents was due to cross-desensitization of A 1 receptors and whether GTP-binding proteins were participating in the inhibitory pathway, we performed experiments in which we activated the currents by preactivating the G i proteins with GTP␥S added to the pipette solution. This strategy allowed us to eliminate the need to first activate the currents with Ado, thus bypassing the Ado receptor, and allowed us to ask whether ACh could inhibit currents activated directly by G proteins. The amplitude of the GTP␥Sactivated GIRK currents was determined by a short application of 0.5 mM Ba 2ϩ , which is sufficient to completely inhibit GIRK currents but does not significantly affect endogenous currents and therefore is used to define the base line (28) (Fig. 3). The difference in currents before and after Ba 2ϩ reflected the amplitude of the GTP␥S-activated currents, and it can be seen in Fig. 3 that GTP␥S caused a substantial activation of the currents. After a complete washout of Ba 2ϩ and the return of the currents to the activated state, the cell was superfused with ACh for 1 min. Surprisingly, following an initial small, further activation of the currents, the activation of the M 2 mAChRs by ACh caused a marked inhibition of the GTP␥S-induced K ϩ currents (Fig. 3). The extent of inhibition of the GTP␥S-activated currents by ACh was 57.5 Ϯ 7.2% (n ϭ 4, summarized in Fig. 4C). Moreover, in contrast to experiments performed without GTP␥S (for example Fig. 2), no recovery from the ACh- . Treatment with ACh caused an initial activation of GIRK currents followed by a pronounced desensitization during a 2-min period of agonist application (left panel). In contrast, a similar treatment with Ado resulted in an activation of the GIRK currents that was followed by a minimal desensitization (right panel). B, in cells expressing both M 2 mAChR and A 1 Ado receptors, an initial activation by Ado caused a stable activation of GIRK current with minimal desensitization during a 2-min application except for a fast initial, small desensitization. After washout of Ado, currents returned to base line. Subsequent application of ACh to the same cell caused a transient activation of GIRK currents to a similar magnitude as Ado but was followed by a substantial reduction during the 2-min exposure. After a 1-min washout of ACh, the reapplication of Ado induced markedly reduced currents compared with the initial Ado response. C, in cells not transfected with M 2 mAChR, ACh had no effect alone and no effect on Ado-induced GIRK currents. In addition, prior treatment with Ado did not induce a decrease in the second Ado response. D, summarized data for the reduction of Ado-induced GIRK currents measured 20 -60 s after a 2-min application of either ACh or Ado in cells expressing both receptors.

FIG. 2. M 2 mAChR-mediated inhibition of GIRK currents was
slowly reversible. The figure shows a representative experiment in which the experimental protocol was similar to that described in Fig. 1B except that successive pulses of Ado were given following the exposure to and washout of ACh to test the reversibility of the ACh-mediated inhibition of the Ado-evoked currents. The half-time of recovery of the Ado response following the ACh-mediated inhibition was between 2 and 5 min.
induced inhibition of the GIRK currents was observed, even after 20 min after washout of ACh (Fig. 3). The inhibition of GTP␥S-activated GIRK currents by ACh suggested that the ACh-induced inhibition was due to an inhibitory effect at the level of the G proteins or GIRKs. In addition, the results suggested that the inhibition of the Ado-induced GIRK currents by the M 2 mAChRs (Figs. 1B and 2) was probably not due to a cross-desensitization of the A 1 adenosine receptors. To avoid confusion we refer to this phenomenon as "ACh-induced inhibition of the GIRK currents" rather than heterologous desensitization. The observation that the M 2 mAChR-induced inhibition of the GIRK currents became quasi-irreversible in the presence of intracellular GTP␥S strongly suggested the involvement of GTP-binding proteins in this inhibitory pathway, as the poorly hydrolyzed GTP␥S should lead to long lasting activation of GTP-dependent processes.
The Inhibition of GIRK Currents Was Observed in the Presence of Heterologously Expressed G␤␥-To further define where the inhibition of GIRK currents by M 2 mAChR occurred, we studied the M 2 mAChR-induced inhibition of GIRK channels that were prestimulated with heterologously expressed G␤␥ subunits. CHO cells that were transiently transfected with M 2 mAChR, GIRK1, GIRK4, and G␤ 1 ␥ 2 exhibited constitutively activated GIRK currents that were blocked by 0.5 mM Ba 2ϩ (Fig. 4A). The currents obtained with G␤␥ expression were activated to a much larger degree than with receptor activation or GTP␥S (compare Figs. 1, 2, and 3) and presumably reflected full activation of the channels. Consistent with this notion, stimulation of M 2 mAChR with ACh did not cause any further increase in Ba 2ϩ -sensitive GIRK currents (Fig. 4A). Notably, in most cases ACh still caused a reduction of basal currents (Fig.  4A). These ACh-inhibited currents were found to have identical current-voltage relationships as GIRK currents (Fig. 4B), indicating that the G␤␥-induced channel activity was inhibited by stimulation of M 2 mAChR. However, the ACh-induced inhibition of GIRK currents in the presence of expressed G␤␥ was significantly less (20 Ϯ 3.8%, n ϭ 8 after a 2-min application of 1 M ACh) compared with GIRK currents activated by GTP␥S and endogenous G proteins (Fig. 4C). Because these results were obtained under conditions in which full activation of the channels was induced by overexpression of G␤␥, the results suggested that the inhibitory effect of the M 2 mAChRs on GIRK currents was due to a reduced interaction between the G proteins and the channels. The smaller inhibitory effect observed under these conditions presumably reflected competition between the receptor-induced inhibitory signal and the large pool of active G␤␥ subunits.
The Inhibition of GIRK Currents by M 2 mAChR Was Mediated via Non-PTX-sensitive G Proteins-Because GTP-binding proteins seem to be involved in mediating the inhibitory effect of M 2 mAChR on GIRK currents (Fig. 3), we tested whether PTX-sensitive G proteins were involved. To do this, transfected CHO cells expressing G␤␥ subunits, GIRK1, GIRK4, and M 2 mAChR were pretreated with 200 ng/ml PTX for 4 -6 h and subsequently tested for an ACh-induced inhibition of the G␤␥activated GIRK currents (Fig. 5). The PTX pretreatment did not prevent or reduce the inhibitory effect of M 2 mAChR on GIRK currents (inhibition of Ba 2ϩ -sensitive GIRK currents after a 2-min incubation with 1 M ACh was 17.5 Ϯ 6.6%, n ϭ 4). In cells in which G␤␥ subunits were not co-expressed, the PTX pretreatment completely blocked activation of GIRK currents by A 1 receptors or M 2 mAChR (n ϭ 4, data not shown), CHO cells were transfected with cDNAs encoding for G␤ 1 and G␥ 1 subunits as well as GIRK1/4 and M 2 mAChR. A, basal GIRK current levels under these conditions were determined by brief applications of Ba 2ϩ (the arrow at the right of the current trace indicates current level in the presence of Ba 2ϩ ). Addition of ACh did not result in any further current increase, indicating that GIRK channels were fully activated by co-expressed G␤␥. However, during the 2-min application of ACh a moderate inhibition of the GIRK currents was observed. B, characteristics of the ACh-inhibited current as determined by subtracting the current-voltage curves obtained prior to (A, a) and during (A, b) the application of ACh. C, summarized data of the inhibition (expressed as % inhibition) of GIRK currents either activated by GTP␥S (as described in Fig. 3) or by heterologously expressed G␤␥ in response to ACh for 2 min.
indicating that the PTX treatment was 100% effective in blocking receptor-mediated activation of G i and G o proteins. Thus, the inhibitory effects of the M 2 mAChR on the GIRK channels, in contrast to their stimulatory effects, did not appear to involve a PTX-sensitive G protein. In addition, pretreatment with PTX did not alter the time course of recovery from inhibition after removal of ACh (data not shown).
Can Other GPCRs Inhibit GIRK Currents?-To test whether other GPCRs could also inhibit GIRK currents, M 3 and M 4 mAChRs were tested as well as A 3 purinergic receptors and ␣ 2A and ␣ 2C adrenergic receptors. The M 4 mAChRs, which, like the M 2 mAChRs are known to couple to PTX-sensitive G proteins, were as effective as M 2 mAChRs in activating GIRK currents and in inducing the inhibition of GIRK currents upon treatment of transfected cells for 2 min with ACh (Fig. 6). Furthermore, activation of the M 4 mAChRs caused an inhibition of a subsequent activation of the A 1 receptors in a manner that was very similar to that caused by the M 2 mAChRs (compare Figs. 1B and 6). In contrast, the other G i /G o -linked GPCRs tested, namely the ␣ 2A and ␣ 2C adrenergic receptors and the A 3 adenosine receptors, exhibited no obvious inhibitory effect on GIRK currents and basically behaved the same as A 1 adenosine receptors (n ϭ 5, data not shown). On the other hand, activation of M 3 mAChRs, which activate G q proteins rather than PTXsensitive G proteins, caused a strong inhibition of Ado-activated GIRK currents. The inhibition of GIRK currents by the M 3 mAChRs was not blocked by PTX in experiments similar to those described in Fig. 5. To test whether the inhibitory action on GIRK currents was specific for muscarinic receptors, the ␣ 1A adrenergic receptor (AR), another G q -coupled receptor, was tested. Whereas activation of ␣ 1A adrenergic receptors with phenylephrine caused a very modest activation of the GIRK currents compared with Ado (Fig. 7A), these receptors caused a robust inhibition of GIRK currents as observed by the striking reduction in the ability of subsequent pulses of Ado to activate the currents (Fig. 7A). Thus the inhibition of the GIRKs was observed with several subtypes of the mAChRs as well as with the ␣ 1A adrenergic receptor. M 2 mAChR-mediated Inhibition of GIRK Currents Was Not Mediated via G q Proteins-Because the inhibition of GIRK currents by the G q -coupled ␣ 1A -AR or M 3 mAChRs as well as by the normally G i /G o -coupled M 2 and M 4 mAChRs did not appear to involve PTX-sensitive GTP-binding proteins, we questioned whether the inhibition caused by these receptors utilized a common pathway. Therefore cells transfected with both the M 2 FIG. 5. Pretreatment with PTX did not affect inhibition of G␤␥-activated GIRK currents. Cells were transfected as described in Fig. 4, and inhibition of G␤␥-activated GIRK currents by ACh was determined after pretreatment of the cells with PTX (200 ng/ml) for 2-4 h. Note the lack of effect of the PTX treatment on the ACh-mediated inhibition of the G␤␥-activated currents and the sustained inhibition of currents by ACh even after washout of agonist.

FIG. 6. Activation of M 4 mAChRs induced inhibition of A 1 Ado receptor-activated GIRK currents.
Experiments similar to the one described in Fig. 1B were performed using cells that expressed M 4 mAChRs instead of M 2 mAChRs. This representative experiment illustrates the reduction of Ado-activated GIRK currents after superfusion of the cell with 1 M ACh for 2 min.

FIG. 7. The inhibition of GIRK currents could be mediated by G q -dependent and -independent pathways.
CHO cells were transiently transfected with cDNAs for M 2 mAChRs, A 1 Ado receptors, GIRK1/4, and the G q -coupled ␣ 1A ARs. A, experiments similar to those described in Figs. 1B and 2 were performed except that the AR agonist PhE (0.1 mM) was used instead of ACh. Stimulation of the ␣ 1A AR resulted only in a modest activation of GIRK current. However, the PhE treatment resulted in a strong and reversible inhibition of GIRK current activations elicited by subsequent brief applications of Ado. B, cells were pretreated with the G q -specific P. multocida toxin (1 g/ml for 20 h). Application of Ado caused a robust activation of the currents. Simultaneous application of PhE had no effect, demonstrating that P. multocida toxin treatment resulted in a complete loss of the PhEinduced inhibition of Ado-activated GIRK currents. A subsequent exposure of the cell to ACh caused a strong inhibition of Ado-evoked currents, demonstrating the lack of a role of G q in the ACh-mediated inhibition of the currents. mAChR and ␣ 1A -AR as well as A 1 adenosine receptors were pretreated with Pasteurella multocida toxin, a protein toxin that was previously shown to irreversibly uncouple G q proteins from their receptors, presumably by first activating G proteins and modifying them in a manner that prevented reactivation after deactivation (29,30). 2 After treatment of cells with 1 g/ml of P. multocida toxin for 20 h, activation of the A 1 adenosine receptors was still able to cause activation of the GIRK currents (Fig. 7B). However, co-application of phenylephrine did not produce any attenuation of the current, demonstrating that P. multocida toxin treatment resulted in a complete loss of the ␣ 1A -AR-induced inhibition of GIRK currents (Fig. 7B). In marked contrast, P. multocida toxin treatment did not affect the inhibition of the GIRK current by the M 2 mAChRs, since application of ACh caused a marked inhibition of the Ado-activated GIRK currents (Fig. 7B). These results suggested the existence of two different pathways of GIRK current inhibition, one that is both PTX-and P. multocida toxin-insensitive and induced by the M 2 and presumably the M 4 mAChRs, and another induced by the G q -linked ␣ 1A -ARs and the M 3 mAChRs that can be selectively blocked by P. multocida toxin.
M 2 mAChR Inhibited GIRK Currents via a Membrane-delimited Pathway-To determine whether cytosolic soluble messengers were involved in the M 2 mAChR-induced inhibition of GIRK currents, we utilized the cell-attached configuration of the patch clamp technique. Channels in the membrane area under the tip of the patch pipette were preactivated with Ado that was included in the patch pipette, and currents were measured in the cell-attached configuration. During the measurement ACh was applied to the bath solution (not to the patch), and effects on the Ado-activated GIRK channel activity were studied. No change in GIRK channel activity was observed in response to the bath application of ACh, suggesting that bath application of ACh was unable to mediate inhibition of the channels being measured under the patch pipette (Fig.  8A). This suggested that ACh was not producing inhibition by generating a soluble second messenger that could theoretically diffuse through the cell to inhibit the channels under the pipette. As a control to demonstrate that the M 2 mAChRs were indeed functionally expressed in this cell, 40 min after washout of the bath-applied ACh, the identical cell was measured a second time, now in the whole cell configuration. Under these conditions, a strong inhibition of the Ado-evoked GIRK currents was observed (Fig. 8B), demonstrating that both receptors were functional in this cell and that inhibition by ACh could be observed in the whole cell configuration. These results demonstrated that no cytosolic soluble messenger molecules were needed for the M 2 mAChR-mediated inhibition of GIRK currents and that the inhibition occurs via a membrane-delimited pathway.

The M 2 mAChR-induced Inhibition of GIRK Currents Was Not Dependent on Internal Calcium Mobilization and Did Not
Require Protein Kinases-Recent studies demonstrated that certain G q -coupled receptors could inhibit GIRK channels expressed in Xenopus oocytes (31,32). In two studies, contradictory results were reported with regard to the calcium dependence and the role of protein kinases in the G q -coupled receptormediated inhibition of GIRK currents. One study (32) reported that injection of the calcium chelator BAPTA into the oocytes resulted in a decrease of endothelin-1A receptor-induced inhibition of opioid receptor-activated GIRK currents, whereas no effect was found of pretreatment with the non-selective protein kinase inhibitor staurosporine. In contrast, another study (31) reported that there was no calcium dependence but involvement of protein kinase C in the bombesin receptor-mediated inhibition of basal GIRK currents.
We tested whether intracellular Ca 2ϩ or various protein kinases participated in the M 2 mAChR-induced inhibition of GIRK currents. Perfusion of cells with 5 mM EGTA and 5 mM BAPTA, which should be sufficient to prevent any substantial rise in the concentration of internal Ca 2ϩ , did not prevent the inhibition of Ado-evoked GIRK currents by activation of the M 2 mAChR (n ϭ 4, data not shown). This suggested that inositol 1,4,5-trisphosphate-mediated calcium release was not involved in this signaling pathway. Interestingly, the inhibition of GIRK currents by stimulation of ␣ 1A adrenergic receptors was not prevented by buffering internal Ca 2ϩ under our conditions. Pretreatment of transfected cells with the nonspecific protein kinase inhibitor staurosporine (100 nM for 30 -90 min) did change the morphology of the cells but did not have an effect on the M 2 mAChR-induced inhibition of Ado-evoked GIRK currents (n ϭ 4, data not shown). Similarly, no effects on the M 2 mAChR-mediated inhibition were observed after pretreatments with wortmannin (100 nM, 30 -60 min, n ϭ 5) or genistein (0.025 mM, 30 -90 min, n ϭ 3). Thus, the M 2 mAChRmediated inhibition of the GIRK currents in the mammalian HEK cells does not appear to require a rise in intracellular calcium, and this inhibition is not blocked by non-selective serine/threonine or tyrosine protein kinase inhibitors or by an inhibitor of the phosphatidylinositol 3-kinase pathway.

The Inhibition of GIRK Currents via M 2 mAChRs Was Not Correlated with the Phosphorylation or Internalization of These
Receptors-The activation of M 2 mAChRs leads to a rapid agonist-dependent phosphorylation of the receptors that is correlated with homologous desensitization and a rapid and extensive internalization of the receptors from the cell surface within a few minutes (21,33). A 1 adenosine receptors have been demonstrated to undergo agonist-induced phosphorylation, but in contrast to M 2 mAChR they internalize and desensitize very slowly (time course takes several hours) (34, 35). Therefore we questioned whether the observed inhibition of 2 B. Wilson, personal communication. Channel activity was measured at a pipette potential of ϩ80 mV (seen as inward deflections). Prior to and after adding ACh (indicated by the bar above the traces) to the bath solution, rectangular voltage pulses to test potentials ranging from Ϫ100 mV to ϩ80 mV were applied to identify GIRK currents by their inwardly rectifying properties. Note that application of ACh did not lead to any changes in channel activity when channels were measured in the cell-attached configuration. B, ACh was washed out for 40 min, and the same cell as in A was patched again; however, this time currents were measured in the whole cell configuration, as described in Fig. 1B. Note that under these conditions the application of ACh for 2 min resulted in an inhibition of Ado-activated GIRK currents.
GIRK currents, which is induced by M 2 mAChR but not by A 1 and A 3 adenosine receptors, is mechanistically connected to the phosphorylation and internalization of muscarinic receptors. Mutants of the M 2 mAChR that either lack the middle part of the third intracellular loop (M 2 ⌬1 mAChR) (21) or in which two clusters of serine or threonine residues in that loop are mutated to alanines (36) are unable to undergo agonist-induced phosphorylation and are severely impaired in their ability to internalize. We utilized these mutants to test whether or not the inhibition of GIRK currents by the M 2 mAChR depends on the agonist-induced phosphorylation and internalization of these receptors. When cells were transiently transfected with M 2 ⌬1 mAChR, A 1 adenosine receptors, and GIRK1 and GIRK4, ACh rapidly induced inwardly rectifying K ϩ currents and caused inhibition of subsequent adenosine-evoked GIRK currents (Fig. 9). No difference was found between the ability of the M 2 ⌬1 and wild-type M 2 mAChRs to mediate the inhibition of the GIRK currents. Therefore, the inhibitory effects on the GIRK currents do not depend on either agonist-mediated phosphorylation of the receptors or on the internalization of M 2 mAChRs.

Inhibition of GIRK Currents Was Mediated through Muscarinic but Not A 1 or A 3 Purinergic Receptors-
The M 2 mAChRs are known to activate several signaling systems, including GIRK channels, via PTX-sensitive G proteins. In the study presented here we demonstrate that M 2 mAChRs and M 4 mAChR not only activated GIRK currents but also inhibited GIRK currents activated via themselves, A 1 adenosine receptors, or directly by G proteins. The inhibitory effect of ACh on GIRK currents followed a much slower time course (t1 ⁄2 ϳ20 -50 s) than the activation of GIRK currents, which peaks after a few seconds (t1 ⁄2(act) ϭ 0.2-1 s) (19), and therefore might represent a mechanism of heterologous desensitization. ACh neither elicited GIRK currents nor inhibited Ado-activated GIRK currents when no muscarinic receptor was co-expressed, indicating that both the activation as well as the subsequent inhibition of GIRK currents specifically required activation of the heterologously expressed mAChRs. Prolonged activation of GIRK currents via A 1 adenosine receptors or A 3 adenosine receptors as well as via ␣ 2A and ␣ 2C adrenergic receptors, which, like the M 2 mAChR, activate GIRK channels via PTXsensitive G proteins, did not cause an inhibition of GIRK currents. However, the M 4 mAChRs, which also couple to PTXsensitive G proteins, caused effects similar to those of the M 2 mAChR. These results indicated that the subtypes of the mAChR that couple to G i /G o proteins induced an additional inhibitory signaling pathway, which was not activated by the Ado or ␣ 2 adrenergic receptors.
The Inhibition of GIRK Currents by M 2 mAChRs Was Independent of Receptor Phosphorylation and Internalization-The M 2 mAChRs are known to internalize in a phosphorylation-dependent manner within a few minutes of agonist treatment (21,37). In contrast, A 1 adenosine receptors require much longer agonist treatment to internalize into intracellular compartments (34,35). Whereas phosphorylation of the M 2 mAChR is required for homologous desensitization (21,36) and mutants of the M 2 mAChR that are impaired in agonist-induced phosphorylation have been demonstrated to be severely impaired in their ability to internalize in an agonist-dependent manner (21,37), the results obtained in this study demonstrate that a phosphorylation-deficient mutant of the M 2 mAChR is fully able to cause inhibition of the GIRK currents. Therefore the inhibitory action of the M 2 mAChRs on Ado-evoked GIRK currents was independent of internalization and phosphorylation of the receptors.

Inhibition of Ado-activated GIRK Currents by Muscarinic Receptors Was Mediated by PTX-insensitive G Proteins and
Occurred Downstream of GPCRs-Under conditions where GIRK channels were constitutively activated either by including GTP␥S in the pipette solution or by heterologous expression of G␤␥, stimulation of M 2 mAChR resulted in an inhibition of these currents, indicating that this inhibition takes place downstream of GPCRs. Furthermore, the observation that the inhibition of GIRK currents by M 2 mAChRs was significantly reduced in the presence of overexpressed G␤␥ subunits compared with the inhibition observed following activation of the currents by Ado or GTP␥S may point to the G proteins as the target of the M 2 mAChR-induced inhibition. Thus, if some factor or event reduces the ability of G␤␥ and the GIRKs to interact, the inhibition caused by the hypothetical factor would be predicted to be greater if G␤␥ is limiting (which appears to be the case in CHO cells) than under conditions where G␤␥ is abundant, as is likely to be the case when G␤␥ is overexpressed. However, further studies are required to demonstrate the exact mechanism of the inhibition demonstrated in this study.
The possibility of activating GIRK currents through heterologously expressed G␤␥ and of inhibiting the activated currents by stimulation of either M 2 or M 3 mAChRs allowed us to test whether uncoupling of those receptors from G i or G o proteins by PTX had any impact on the inhibition of the GIRK by ACh. Because a PTX pretreatment that resulted in a complete loss of receptor-mediated activation of GIRK currents in cells expressing only endogenous G proteins did not affect the receptor-mediated inhibition of the GIRK currents in cells overexpressing G␤␥, the conclusion can be drawn that the inhibitory effect of the M 2 mAChR was not mediated through PTXsensitive G proteins. However, other yet to be identified GTPbinding proteins may be involved in the pathway that links M 2 mAChR to the inhibition of GIRK currents. The M 2 mAChRinduced inhibition of Ado-evoked GIRK currents was at least partially reversible (Fig. 2); however, the inhibition of GTP␥Spreactivated currents was not reversible (Fig. 3). Because PTXsensitive G proteins do not mediate the inhibitory effect of M 2 mAChR on GIRK currents, the conversion of the inhibitory FIG. 9. The inhibition of GIRK currents did not require agonist-induced phosphorylation or internalization of muscarinic receptors. Cells were transfected with cDNAs for GIRK1/4, A 1 Ado receptor, and a mutant M 2 mAChR (M 2 ⌬1 mAChR, lacking a significant portion of its third intracellular loop) that has been shown to be deficient in agonist-induced phosphorylation and internalization (21). GIRK currents were measured as described in Fig. 1B. ACh caused activation and then inhibition of the currents during the 2-min exposure as well as the subsequent inhibition of Ado-induced currents. effect of the M 2 mAChR from a reversible to an irreversible effect in the presence of the slowly hydrolyzable GTP analog suggests that other GTP-binding proteins must be involved. A potential candidate might be G z , which is a member of the G i /G o family but PTX-insensitive.
G q Protein-coupled Receptors Inhibited GIRK Currents through a Different Pathway-ACh-induced muscarinic K ϩ currents can be inhibited via ␣ 1 adrenergic receptors in atrial myocytes (38), most likely via activation of G q proteins and production of soluble second messengers. We present in this study two major pieces of evidence that M 2 mAChRs do not cause inhibition of GIRK channels via activation of G q proteins. First, preincubation of the cells with P. multocida toxin, which is known to irreversibly modify and uncouple G q proteins from their cognate receptors (29, 30), 2 did not block the M 2 mAChRinduced inhibition of GIRKs, although P. multocida toxin was able to block the ␣ 1A -AR/G q -mediated inhibition of the GIRKs. Second, the pathway that leads to inhibition of GIRK channels via activation of M 2 mAChRs is a membrane-delimited pathway that did not appear to involve soluble second messengers. In contrast, the pathway that leads to inhibition of GIRK currents via the ␣ 1 receptors in cardiac myocytes was found in similar experiments not to be membrane-restricted (38).
The Signaling Pathway That Leads to Inhibition of GIRKs Seems Not to Involve Protein Kinases or Phosphatidylinositol 3-Kinase-We made attempts to pharmacologically dissect the signaling pathway of the M 2 mAChR-induced inhibition of GIRK currents. Several recent studies in Xenopus oocytes tried to dissect the pathway that causes inhibition of GIRK currents through G q -coupled receptors such as the metabotropic glutamate receptors, bombesin, or endothelin 1A receptors (31,32,39). In Xenopus oocytes the endothelin-induced inhibition of -opioid receptor-activated GIRK currents through endothelin 1A receptors was partially sensitive to Ca 2ϩ chelating agents (BAPTA) but not affected by protein kinase inhibitors such as staurosporine (32). Inhibitors of calcium-dependent phospholipase A 2 caused a 40% reduction in the inhibitory effect of endothelin 1A receptors at the highest phospholipase A 2 inhibitor concentration used. Because arachidonic acid can be generated through phospholipase A 2 activity and because extracellularly applied arachidonic acid inhibited GIRK currents, Rogalski et al. (32) postulated that endothelin 1A receptors inhibited GIRK currents through mobilization of intracellular Ca 2ϩ and subsequent activation of phospholipase A 2 and generation of arachidonic acid. In contrast to this finding, in another study it was concluded that metabotropic glutamate receptors inhibited basal and ACh-activated GIRK currents through activation of a staurosporine-and bisindolylmaleimide-sensitive kinase that may have been protein kinase C because phorbol esters were able to mimic this effect (39). Similar results were reported for the inhibition of basal GIRK currents through activation of bombesin receptors (31). Whereas each of these studies found that G q -coupled receptors can cause inhibition of GIRK currents in Xenopus oocytes, the contradictory results concerning the inhibitory mechanism(s) will require future studies to determine the exact pathway(s) that mediates inhibition of GIRK currents through G q -coupled receptors. The M 2 mAChRs did not mediate their inhibitory effect via G q since the effect was not blocked by P. multocida toxin, whereas this toxin did block the action of the G q -coupled receptors tested in this study. Furthermore, we found that the M 2 mAChR-induced inhibition of GIRK currents was insensitive to staurosporine, H-7, wortmannin, genistein, and chelation of intracellular Ca 2ϩ with BAPTA and EGTA, indicating that the pathway utilized by the M 2 mAChR receptor is different from the pathway that mediates the inhibition of GIRK currents by G q -coupled receptors. Recently, another study described that dopamine D 3 dopaminergic receptors and somatostatin SST1 receptors can activate homomeric GIRK2 channels via PTXsensitive G proteins in CHO cells. Interestingly, activation of the D 3 dopamine receptors induced an attenuation of the somatostatin response but not vice versa (40), suggesting that the D 3 dopamine receptors may act by a pathway that is similar to that used by the M 2 and M 4 mAChRs to cause inhibition of GIRK currents.
Further evidence that some receptors that couple to PTXsensitive G proteins can induce inhibition of other G i /G o -coupled pathway proteins came from a study that compared the inhibition of N-type Ca 2ϩ channels byand -opioid receptors and GABA B receptors. Results from this study indicated that agonists ofand -opioid receptors inhibited N-type Ca 2ϩ currents to a lesser extent than did GABA B receptors, but if opiate agonists were added together with GABA B receptors, the opiate agonists attenuated the GABA B response (2). It is unclear if this cross-talk is related to the inhibition of Ado-evoked GIRK currents by muscarinic receptors, but these studies and our results presented here clearly demonstrate that signaling of G protein-coupled receptors that are known to couple to the same family of G protein-coupled receptors is more divergent and complex than previously anticipated. Further studies will be required to identify the components that mediate the inhibition of the GIRK channels by the receptors described here and to ascertain whether the inhibitory signals are specific for the GIRK channels or whether they participate in regulating other effectors downstream of these GPCRs.