The M3 Receptor-mediated K+ Current (IKM3), a Gq Protein-coupled K+ Channel*

Stimulation of muscarinic acetylcholine receptors (mAChRs) can activate an inward rectifier K+ current (IKACh), which is mediated by the M2 subtype of mAChR in cardiac myocytes. Recently, a novel delayed rectifier-like K+ current mediated by activation of the cardiac M3 receptors (designated IKM3) was identified, which is distinct from IKACh and other known K+ currents. While IKACh is known to be a Gi protein-gated K+ channel, the signal transduction mechanisms for IKM3 activation remained unexplored. We studied IKM3 with whole-cell patch clamp and macropatch clamp techniques. Whole cell IKM3 activated by choline persisted with minimal rundown over 2 h in presence of internal GTP. When GTP was replaced by guanyl-5′-yl thiophosphate, IKM3 demonstrated rapid and extensive rundown. While IKACh (induced by ACh) was markedly reduced in cells pretreated with pertussis toxin, IKM3 was unaltered. Intracellular application of antibodies targeting α-subunit of Gi/o protein suppressed IKACh without affecting IKM3. Antibodies targeting the N and the C terminus, respectively, of Gq protein α-subunit substantially depressed IKM3 but failed to alter IKACh. The antibody against β-subunits of G proteins inhibited both IKACh and IKM3. IKM3 activated by choline in the cell-attached mode of macropatches persisted in the cell-free configuration. Application of purified Gq protein α-subunit or βγ-subunit of G proteins or guanosine 5′-O-(thiotriphosphate) to the internal solution activated IKM3-like currents in inside-out patches. Our findings revealed a novel aspect of receptor-channel signal transduction mechanisms, and IKM3 represents the first Gq protein-coupled K+ channel. We propose that the G protein-coupled K+ channel family could be divided into two subfamilies: Gi protein-coupled K+ channel subfamily and Gq protein-coupled K+ channel subfamily.

While M 2 receptors are commonly believed to be the only functional mAChRs 1 in cardiac tissues, this concept has been challenged by recent findings revealing the presence of M 3 receptors in the hearts of various species including guinea pig (1,2), rat (3), dog (4 -6), and human (7)(8)(9)(10)(11). We discovered that the cardiac M 3 receptors mediate the activation of a novel delayed rectifier K ϩ current (we have named it I KM3 ) distinct from I KACh and other known K ϩ currents. I KACh is characterized by strong inward rectification, whereas I KM3 conducts a delayed rectifier-like K ϩ current. We also found that M 3 receptors and I KM3 play a significant role in regulating heart rates, cardiac resting membrane potential, and membrane repolarization (1,2). The findings suggest that we are no longer able to consider parasympathetic control of the heart as due to a simple ACh-M 2 interaction; we have to understand cholinergic effects in terms of the consequence of activating multiple subtypes of mAChRs, with potentially varying signal transduction and effector systems (such as different K ϩ channels) (6).
The M 1 and M 3 receptors are characterized biochemically by stimulation of a large inositol phosphate response while having a small stimulatory effect on adenylate cyclase activity. The M 2 and M 4 isoforms are typically linked to an inhibition of adenylyl cyclase activity and only a modest stimulation of inositol phosphate release (12,13). I KACh is a G protein-gated K ϩ channel and its activation is a result of interactions between M 2 receptors and K ϩ channels coupled directly via PTX-sensitive G i proteins (14 -16), i.e. activation of I KACh is critically determined by G i protein activity and is independent of the downstream components of the G i protein signaling pathway. It remained unknown how the M 3 receptors mediate activation of I KM3 . Since M 3 receptors have been found to mainly activate the G q -phospholipase C (PLC)-protein kinase C (PKC) pathway, we hypothesized that G q protein plays a critical role in I KM3 activation upon M 3 receptor stimulation in cardiac cells. The present study was designed to examine this hypothesis so as to elucidate the signal transduction mechanisms of M 3mediated activation of I KM3 , as compared with M 2 -I KACh coupling.

EXPERIMENTAL PROCEDURES
Cell Isolation and Culture-Single canine atrial myocytes were isolated as previously described (1, 2, 4 -6). The dispersed cells were stored in KB medium (20 mM KCl, 10 mM KH 2 PO 4 , 25 mM glucose, 70 mM potassium glutamate, 10 mM ␤-hydroxybutyric acid; 20 mM taurine, 10 * This work was supported in part by the Heart and Stroke Foundation of Quebec and the Fonds de la Recherche de l'Institut de Cardiologie de Montreal (awarded to Z. W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ mM EGTA, 0.1% albumin, and 40 mM mannitol, pH adjusted to 7.4 with KOH) at 4°C for later electrophysiological experiments. Use of animals was approved by the Animal Care Committee of Montreal Heart Institute. Human ether-a-go-go-related gene (HERG)-expressing HEK293 cells were culture as described previously (17).
Whole-cell Patch Clamp Recording-Patch clamp techniques have been described in detail elsewhere (1, 2, 4 -6). Experiments were conducted at 36 Ϯ 1°C. The following blockers were included in the perfusate (Tyrode solution) to minimize contaminating currents: CdCl 2 (200 M, L-type Ca 2ϩ current), 4-aminopyridine (1 mM, transient outward K ϩ currents), dofetilide (1 M, rapid delayed rectifier K ϩ current), and chromanol 293B (10 M, slow delayed rectifier K ϩ current). I KACh was activated in the presence of 1 M ACh in the superfusate, and the amplitude was measured as the current amplitude at the end of the test pulses. I KM3 was activated in the presence of 10 mM choline in the superfusate, and the amplitude was measured as the activating currents after subtraction of the time-independent leak current. Dog atrial ultra-rapid delayed rectifier K ϩ current (I Kur.d ) was recorded in the presence of atropine (100 nM) to prevent I KM3 and I KACh (18). For better comparison, all mean data were expressed as current density calculated by normalizing to the cell capacitance.
Macropatch Recording-Macropatch currents were measured in cellattached and inside-out configuration by using pipettes made from borosilicate glass with tip openings of ϳ20 m as previously described (19). The antibodies used in this study were all purchased from Santa Cruz Biotechnology (the specificity of the antibodies was tested by the company). Antibodies were added to the pipette solution and applied intracellularly by dialysis. Currents were recorded right after membrane rupture before dialysis took place as base-line control data, and the same recordings were repeated 20 min after membrane rupture with complete dialysis, and the data reflected the effects of antibodies.
Cloning of Cardiac M 3 Receptor cDNA Fragment-Gene-specific primer pairs (tggaacaacaatgatgctgc (forward) and ccttttccgcttagtgatctg (reverse)) were designed based on published sequence of human m 3 (20). Total RNA was isolated from dog atrium as previously described in detail (19). First-strand cDNA resulting from reverse transcription was used as a template for PCR amplification, and the PCR products of desired size were subjected to sequencing analysis.
Data Analysis-Group data are mean Ϯ S.E. Paired t tests were used for single comparisons. Kinetics were analyzed with CLAMPFIT (pCLAMP 8.0) or Graphpad Prism.

RESULTS AND DISCUSSION
Depolarizing voltage steps activated I KM3 in the presence of continued stimulation of M 3 receptors by choline (10 mM) in the bathing solution, which was otherwise absent before choline application. The current was highly sensitive to M 3 -selective antagonists p-F-HHSiD (20 nM), darifenacin (20 nM), or 4-DAMP (10 nM) (Fig. 1B). To test whether G protein is required for I KM3 activation by M 3 receptor stimulation, GTP of the pipette solution was replaced by GDP␤S (1 mM, a nonphosphorylatable analogue of GDP) to deplete the cytosolic GTP. Recordings were made right after membrane rupture before dialysis took place as base-line control data and from 5 to 30 min after membrane rupture to determine changes caused by omission of intracellular GTP. As shown in Fig. 1A, I KM3 had little rundown over a 2-h recording period with a GTP pipette. However, with GDP␤S in the pipette, the current demonstrated a quick decay starting within 5 min after formation of whole-cell configuration and virtually disappeared 10 min after membrane rupture, indicating a requirement of G proteins for I KM3 activation.
It is known that stimulation of M 2 receptors activates the PTX-sensitive G i protein, which in turn interacts with I KACh to open up the channels. That is, M 2 receptors are coupled to I KACh via the membrane-delimited G i/o proteins, or in other words, I KACh is a G i protein-gated K ϩ channel (14 -16). Our experiments confirmed this point. The possible role of PTX-sensitive G proteins in M 3 -I KM3 coupling was also investigated. As shown in Fig. 1B, while I KACh was clearly suppressed in cells pretreated with PTX (2 g/ml) for 90 min in the recording solution, I KM3 was unaltered. Similar results were obtained when PTX (2 g/ml) was applied intracellularly by dialysis (data not shown). In another set of experiments, anti-G␣ i/o antibody targeting the C terminus of G i/o protein ␣-subunit was applied to the cytosol of the cells by dialysis through the pipette solution. Consistent with the PTX data, anti-G␣ i/o antibody (15 g/ml) remarkably diminished I KACh but left I KM3 unchanged (Fig. 1C). The results from the above experiments, while confirming the role of G i/o protein in the coupling between M 2 receptors and I KACh , ruled out contribution of G i/o protein to I KM3 activation by choline. Studies on mAChR function by delivering antibodies into the cytoplasm of cells by pipette solution dialysis have been frequently employed by other researchers (21,22).
We then turned to test the possible role of G q protein. We used two different antibodies: anti-G␣ q (N) and anti-G␣ q/11 (C), antibodies targeting the N and the C-terminal regions of a G q protein ␣-subunit, respectively. Clearly, both anti-G␣ q (N) and anti-G␣ q/11 (C) at a concentration of 10 g/ml produced substantial suppression of I KM3 , whereas I KACh was unaffected. Elevating anti-G␣ q (N) antibody concentration to 50 g/ml nearly abolished I KM3 , and the inhibitory effects were not reproduced when the antibody was preinactivated by boiling ( Fig. 2A). There is evidence indicating that the C-terminal region of G q protein ␣-subunits is responsible for interaction with receptor proteins (14,(23)(24). Thus, the results from anti-G␣ q/11 (C) antibody would attest the requirement of whole G q protein for I KM3 activation but do not allow us to distinguish which G q protein heterotrimer subunit (G␣ or G␤␥) transduces the signal from M 3 receptors to I KM3 . It is also known that the N terminus of G q protein bears a region critical for effector function (15,

25-26) and
␤␥-subunit binding (23). The anti-G␣ q (N) antibody is designed to target the "active" ␣-subunit-GTP complex. The ability of this antibody to suppress I KM3 would therefore suggest the importance of the ␣-subunit of G q protein in interacting with I KM3 .
The data from anti-G␣ q (N) antibody experiments indicated a role of ␣-subunit of G q protein for I KM3 activation, because the antibody targets the N terminus of ␣-subunit, which is thought to be responsible for effector interactions and for ␤␥-subunit binding (25,26). The ␤␥-subunit of G i/o protein is known to play a major role over ␣-subunit for I KACh activation (14 -16). We then further studied the effects of anti-␤ com (against various ␤-subunits including ␤ 1 , ␤ 2 , ␤ 3 , and ␤ 4 ) on I KM3 . Our data demonstrated the ability of anti-␤ com antibody (10 g/ml) to inhibit both I KACh and I KM3 (Fig. 2B). This finding indicates that G␤␥ may also be a functional subunit in M 3 -I KM3 signaling system.
Activation of G q protein can result in activation of PLC that hydrolyzes the phospholipid phosphatidylinositol 4,5-bisphosphate (PIP 2 ) to form 1,4,5-inositol trisphosphate (IP 3 ) and diacylglycerol (DAG). IP 3 stimulates Ca 2ϩ release from intracellular stores. Increase in [Ca 2ϩ ] i and DAG further activates the downstream PKC activity. Our data showed that PLC inhibitor U73122 (1 M), neutralizing monoclonal PIP 2 -specific antibody (60 nM), a DAG analogue OAG (10 M 1-oleoyl-2-acetyl-snglycerol), IP 3 (100 M), PMA (a phorbol ester activator of PKC, 100 nM), and Bis (bisindolylmaleimide, a selective PKC inhibitor, 40 nM) all failed to affect I KM3 (Fig. 2C). Positive control experiments were performed to assure the effectiveness of for these agents. The effects of U73122, OAG, PMA, and Bis were verified with I Kur.d , which we have previously shown to be modulated by PKC (18). While bath application of the ␣ 1 - . Macropatch I KM3 -like current was measured in the presence of 2 mM Ba 2ϩ to inhibit the I KACh -like current. C, activation of I KACh -and I KM3 -like currents by purified G␣ I -(100 nM, n ϭ 5) and G␣ q -(100 nM, n ϭ 6) subunits, respectively. D, activation of I KACh -like (n ϭ 4) and I KM3 -like (n ϭ 5) currents by purified G␤␥-subunit (10 M/ml). Macropatch I KM3 -like current was measured in the presence of 2 mM Ba 2ϩ to inhibit the I KACh -like current.

FIG. 2. Role of G q protein in I KM3 activation.
A, effects of anti-G␣ q (N) antibody (10 g/ml) on I KACh and I KM3 . t test indicates significant reduction of I KM3 by anti-G␣ q antibodies (p Ͻ 0.05 for potentials from Ϫ20 to ϩ50 mV, n ϭ 10) but no effect on I KACh (n ϭ 8). Boiling inactivated anti-G ␣q (N) antibody lost the effect on I KM3 (p Ͼ 0.05, n ϭ 5). The inset shows the suppressing effects of anti-G␣ q (C) antibody (10 g/ml) on I KM3 (n ϭ 3). B, effects of anti-G␤ com antibody (10 g/ml) targeting various ␤-subunits of G protein on I KACh and I KM3 . t test indicates significant reduction of both I KACh (p Ͻ 0.05 for all potentials tested, n ϭ 3) and I KM3 (p Ͻ 0.05 for potentials from Ϫ20 to ϩ50 mV, n ϭ 4). C, effects of various components belonging to the G q -PKC signaling pathway on I KM3 activation. Data shown are from at least three cells for each group. *, p Ͻ 0.05 versus Ctl and ϩp Ͻ 0.05 versus phenylephrine (Phen), OAG, or PMA. adrenoreceptor agonist phenylephrine (10 M) caused enhancement of I Kur.d , co-application with U73122 or Bis reversed the effects (Fig. 2C). Superfusion with either PMA or OAG directly increased I Kur.d , and the increases were reversed by Bis. The effectiveness of PIP 2 was verified by experiments showing that PIP 2 (10 M) included in the pipette increased the K ϩ current carried by human ether-a-go-go channels (I HERG ) and shifted the activation of I HERG toward hyperpolarizing potentials, effects consistent with the previous findings (17,27). Since our pipette solution contained 10 mM EGTA, and L-type Ca 2ϩ current was inhibited by Cd 2ϩ present in the superfusate, it is unlikely that I KM3 activation was related to [Ca 2ϩ ] i . This possibility was further excluded by additional experiments showing the lack of effects of intracellular application of heparin (1 mM) on I KM3 (Fig. 2C).
To investigate whether the interaction between G q protein and I KM3 is membrane-delimited, we further employed macropatch to study the activation of I KM3 . With choline-or ACh-free pipette (external) solution, I KM3 or I KACh was observed in neither cell-attached nor inside-out patches. Addition of choline (10 mM) or ACh (1 M) to the superfusate (internal solution) still failed to activate any currents in the inside-out mode. However, when choline was included in the pipette solution, a current characteristic of I KM3 was activated in the cell-attached patch and sustained after a membrane excision into an insideout patch during the 30-min recording period with little rundown (Fig. 3A). Similarly, when ACh was present in the pipette solution I KACh was induced in both cell-attached and inside-out configurations. When GTP␥S (10 M, a nonhydrolyzable analogue of GTP) was present in the superfusate (internal solution), but with choline and ACh absent in the pipette in the inside-out patches, only I KACh -like currents were activated in seven out of eight patches tested, and there was no apparent concomitant activation of I KM3 -like currents (Fig. 3B). Only when the internal GTP␥S was elevated to 50 M did I KM3 -like currents become manifested in five out of eight patches, consistent with the previous finding that higher concentrations of GTP␥S are required for G␣ q activity relative to G␣ i/ o (28). To separate the I KM3 -like currents from concomitantly activated I KACh -like currents, 2 mM Ba 2ϩ was added to the superfusate to inhibit the latter. The I-V relationships of I KACh -like and I KM3like currents resemble those of whole-cell I KACh and I KM3 , with inwardly and outwardly rectifying properties, respectively (Fig.  3B). The results from our macropatch recordings indicate that I KM3 is activated by G protein in a membrane-delimited fashion.
To further investigate the role of ␣and ␤␥-subunits of G proteins in I KM3 activation, we performed experiments using the G␣ i and G␣ q proteins purified from Sf9 cells infected with baculoviruses encoding the recombinant G␣ i -and G␣ q -subunits, respectively, and the purified ␤␥-subunit of G proteins (Chemicon International, Inc.). Application of G␣ i to intracellular side of the inside-out macropatch induced small I KACh , but failed to induce I KM3 in all eight patches tested. By comparison, application of G␣ q activated an I KM3 -like current in four out five cells, but in none of six cells tested was I KACh induced by G␣ q (Fig. 3, C and D). On the other hand, perfusion with the ␤␥-subunit of G protein activated large I KACh current and I KM3 too (Fig. 3D). It is therefore possible that the signal generated upon M 3 activation requires G␣ q for specificity, but the signal is transduced by both G␣-subunit and G␤␥ dimer. Interestingly, one elegant study focusing on the M 3 receptor signal transduction for Ca 2ϩ release in Xenopus oocytes (29) revealed a similar mechanism to what was found in this study.
If it is true that I KM3 activation by the M 3 receptors is mediated by G q protein, then the conserved sequence of mACh for G protein coupling must be present in the heart. Indeed, we cloned a cDNA fragment representing dog cardiac M 3 receptor (GenBank TM accession number AF056305) spans a part of the third intracellular loop thought to contain a region critical for G protein binding. Particularly, our fragment covers the first 16 -21 amino acids of the third intracellular domain, which determine the G protein coupling specificity (30,31). This cardiac m 3 cDNA fragment shares ϳ81% homology to the same region of human brain m 3 gene (20) in the amino acid level, but only 4% homology to human m 2 receptor and 3.8% homology to the dog cardiac m 2 receptor (GenBank TM accession number AF084483). The data provide an evidence for the presence of G q protein coupling domain of the M 3 receptor in dog heart.
The K ϩ channel superfamily is composed of several K ϩ channel families (voltage-gated, inward rectifier, ligand-gated, second-messenger-gated, and G protein-gated K ϩ channel families). I KACh is a prototype member of the G protein-gated K ϩ channel family identified to date and probably the only current, among the cardiac currents identified to date, belonging to a member of the G protein-coupled K ϩ channel subfamily, although some evidence suggests that ATP-sensitive K ϩ current I KATP is also gated by G protein (15). A major distinction of G protein-coupled K ϩ channels from others is that activation of G protein is absolutely required for channels to open. Although other K ϩ channel families like voltage-gated K ϩ channels are also modulated by G protein activity, their activation is not dependent on G protein. Our present finding that I KM3 is a G q protein-coupled K ϩ channel can be viewed as an addition to the G protein-coupled K ϩ channel family. We therefore propose that the G protein-coupled K ϩ channel family could be divided into two subfamilies: G i protein-coupled K ϩ channel subfamily and G q protein-coupled K ϩ channel subfamily.
I KACh has been shown to be down-regulated in both its function and expression (6,32,33), whereas I KM3 was found to be increased (6), in atria with atrial fibrillation in both animal models and human hearts. Correspondingly, the atrial M 2 receptor density decreased, whereas M 3 receptor density increased (6), and G␣ i protein level decreased by 12% (34), in atrial fibrillation. It is tempting to speculate that an increase in I KM3 may contribute to initiation and perpetuation of atrial fibrillation. Yet future studies are needed to test this notion.