Coupling of M2 Muscarinic Receptors to Membrane Ion Channels via Phosphoinositide 3-Kinase γ and Atypical Protein Kinase C*

We report a novel signaling pathway linking M2 muscarinic receptors to metabotropic ion channels. Stimulation of heterologously expressed M2 receptors, but not other Gi/Go-associated receptors (M4 or α2c), activates a calcium- and voltage-independent chloride current in Xenopus oocytes. We show that the stimulatory pathway linking M2 receptors to these chloride channels consists of Gβγ stimulation of phosphoinositide 3-kinase γ (PI-3Kγ), formation of phosphatidylinositol 3,4,5-trisphosphate (PIP3), and activation of atypical protein kinase C (PKC). The chloride current is activated in the absence of M2 receptor stimulation by the injection of PIP3, and PIP3 current activation is blocked by a pseudosubstrate inhibitory peptide of atypical PKC but not other PKCs. Moreover, the current is activated by injection of recombinant PKCζ at concentrations as low as 1 nm. M2 receptor-current coupling was disrupted by inhibiton of PI-3K and by injection of βγ binding peptides, but it was not affected by expression of dominant negative p85 cRNA. We also show that this pathway mediates M2 receptor coupling to metabotropic nonselective cation channels in mammalian smooth muscle cells, thus demonstrating the broad relevance of this signaling cascade in neurotransmitter signaling.

M 2 muscarinic receptors mediate numerous cellular functions including presynaptic regulation of transmitter release by neurons in the brain and autonomic nervous system and postsynaptic control of the heart, smooth muscle, and secretory cells. Stimulation of M 2 receptors in heart cells opens inward rectifier K ϩ channels through a direct interaction between released G protein ␤␥ subunits and channel proteins (1)(2)(3)(4)(5)(6), but the signaling pathways linking M 2 receptors to ion channels in nerve, smooth muscle, and secretory cells are poorly understood. Hormone-stimulated phosphoinositide 3-kinase (PI-3K) 1 plays an important role in cell growth, adhesion, and survival and in actin assembly (7). The identification of the G␤␥-stimulated PI-3K␥ (8 -10) extends the potential range of processes mediated by PI-3K to G protein-coupled receptors, although specific physiological processes mediated by PI-3K␥ have not been identified. One potential target of lipid second messengers generated by PI-3K are atypical protein kinase C enzymes (aPKCs), which lack a diacylglycerol binding site and are activated in vitro by phosphatidylinositol phosphates (11)(12)(13)(14). Here we show that stimulation of heterologously expressed M 2 receptors, but not other G i /G o -linked receptors, opens endogenous metabotropic chloride channels in Xenopus oocytes by activation of PI-3K␥, generation of PIP 3 , and stimulation of aPKC. We also show that this signaling pathway mediates physiological coupling between M 2 receptors and nonselective cation channels in mammalian smooth muscle cells.

EXPERIMENTAL PROCEDURES
Xenopus Oocyte Procedures-Surgical removal of oocytes was performed in the laboratory of Dr. Peter Drain in accordance with a protocol approved by the University of Pennsylvania Animal Care and Use Committee. Oocyte defolliculation, injection, and dual-electrode voltage clamp were as described previously (15). Currents were amplified (OC-725C, Warner Instruments, Hamden, CT), filtered at 200 Hz (Ϫ3 dB; 8-pole Bessel filter, model 902, Frequency Devices, Haverhill, MA), digitized at 1 kHz (TL125, Axon Instruments, Foster City, CA), and monitored and simultaneously stored on disk (Axotape, Axon Instruments). All currents shown were leak-subtracted using identical voltage paradigms before exposure to mACH. Pipettes with resistances between 0.5 and 1 megaohm were filled with 3 M KCl. Extracellular bath solution was (concentrations in mM): 115 NaCl, 2.8 KCl, 1 MgCl 2 , and 10 Hepes. Intracellular injection of all substances consisted of 50-nanoliter volumes, and the indicated concentrations assume a 20fold dilution in the oocyte cytosol (1 l volume). Injections were made 10 min before oocyte stimulation. Antibodies directed against specific G␣ subunits were injected at the obtained concentration (1:20 final titer). Solution changes were made by washing the bath with at least 25ϫ bath volume (1 ml).
Preparation of cRNAs-cRNA was prepared using the mMessage mMachine kit (Ambion). Plasmid DNAs were linearized with appropriate restriction enzymes, and cRNAs were synthesized using the appropriate RNA polymerase. The integrity of the cRNAs was tested on ethidium bromide-stained agarose gels, and concentrations were estimated by spectrophotometry. The ⌬P85 construct (16) in PGEX was obtained from Dr. M. Kasuga and subcloned in pBlueScript KSϩ (Stratagene). M 2 , M 3 , and M 4 clones were kindly provided by Dr. E. Peralta and Dr. T. Morelli.
Patch-clamp and Myocyte Dispersion-Equine trachealis tissue was obtained in accordance with protocols approved by the University of Pennsylvania Animal Care and Use Committee. Cell isolation, whole cell recording, and agonist application were as described previously (17). Seals were formed with 3-5-megaohm pipettes, and cells were dialyzed with the following (concentrations in mM): 130 CsCl, 1.2 MgCl 2 , 1 MgATP, 0.1 EGTA, and 10 Hepes, pH 7.3. The bath solution was (concentrations in mM): 125 NaCl, 5 KCl, 1 MgSO 4 , 1.8 CaCl 2 , 10 glucose, and 10 Hepes, pH 7.4. Cells were allowed to adhere to a glass coverslip, and recordings in relaxed cells were made at room temperature. Following break-in, cells were dialyzed for 5 min before activation of currents by application of the muscarinic agonist using a puffer pipette.
Chemicals-Chelerythrine, GF109203X, Gö 6976, and cPKC pseudosubstrate peptide were obtained from Calbiochem and PIP 3 and * This work was supported by National Institutes of Health Grants HL45239 and HL41084 (to M. I. K.) and a grant from the American Heart Association (to Y.-X. 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.

RESULTS AND DISCUSSION
Heterologous expression of M 2 receptors in Xenopus oocytes indicated that receptor stimulation activates a novel metabotropic chloride current. The muscarinic agonist methacholine (mACH) activated only a sustained inward current in oocytes injected with M 2 receptor cRNA and recorded in calcium-free conditions, whereas a transient inward current was observed in oocytes expressing G q/11 -coupled M 3 receptors (Fig. 1A). The transient current was shown to be the ubiquitous endogenous calcium-activated chloride current, as it was blocked by chelating intracellular calcium or by inhibiting calcium release with heparin, whereas activation of the M 2 current could be obtained repeatedly in calcium-free solutions and was not affected by intracellular calcium chelation or by blockade of calcium release. These currents were identified in early original experiments characterizing muscarinic acetylcholine receptor subtypes in Xenopus oocytes (18), although the sustained current was not isolated and was reported to be cation selective.
Ion substitution experiments clearly identified the M 2 current as anion selective, with a selectivity sequence of I Ϫ Ͼ Cl Ϫ Ͼ isethionate, whereas substitution of more than 90% of the sodium for Tris had no effect on current reversal potential ( Fig.  1, B and C). The chloride current activated following M 2 receptor binding was voltage-independent, and no measurable current was available in the absence of M 2 stimulation (no shift in background current observed with anion substitution), indicating that receptor binding is required for channel opening.
To examine the linkage between M 2 receptors and the novel chloride current, we used antibodies, specific peptides, dominant negative constructs, and enzyme inhibitors to disrupt receptor-effector coupling. M 2 receptor-chloride current coupling was blocked by preinjection of antibodies directed against G␣ i or G␣ o , but not G␣ q , proteins (Fig. 2, A and B). Anti-G␣ i1 / G␣ i2 and anti-G␣ i3 /G␣ o antibodies blocked 83 Ϯ 3% (n ϭ 6) and 52 Ϯ 4% (n ϭ 6) of the current, respectively. Whereas M 2 signaling was clearly coupled by G i /G o proteins, the M 2 current was not activated by heterologously expressed adrenergic ␣ 2C receptors or muscarinic M 4 receptors, which coupled weakly to intracellular calcium release and the associated calcium-activated chloride current (Fig. 2C). These receptors also preferentially associate with G i /G o proteins (19,20), indicating that the signaling pathway leading to activation of the M 2 chloride current discriminates between receptors signaling through pertussis toxin-sensitive G proteins. Moreover, whereas both M 2 and M 4 receptors are capable of activating inward rectifying K ϩ channels through ␤␥ proteins (21), the stimulatory pathway linking M 2 receptors and chloride channels effectively distinguishes between these closely related receptors.
Protein kinase C (PKC) molecules that are activated by diacylglycerol and calcium following stimulation of phospholipase C by G protein-coupled receptors have been implicated in M 2 receptor-ion channel coupling (22)(23)(24)(25). M 2 coupling to the novel chloride current was disrupted by exposure of oocytes to the nonselective protein kinase C inhibitor chelerythrine. However, GF109203X and Gö 6976 (not shown), more selective inhibitors of several conventional and novel PKC isoforms, had no effect on current activation (Fig. 3, A and B). None of the protein kinase C inhibitors affected M 3 coupling to phospholipase C, and conversely, M 2 receptor coupling was not affected by phospholipase C inhibition (Fig. 3A). aPKCs that are activated by phosphatidylinositol 4,5-diphosphate, PIP 3 , and cisfatty acids have been implicated as effectors in mitogen, apoptotic, and contractile signaling (11-13, 26, 27), although involvement of aPKCs in ion channel signaling has not been reported. We examined the role of protein kinase C subtypes in M 2 receptor coupling using peptides that selectively bind and inhibit conventional PKCs or aPKCs. Preinjection of aPKC pseudosubstrate inhibitory peptides (28), but not cPKC inhibitory peptides (29), inhibited the M 2 chloride current coupling, suggesting that aPKC activation is necessary for M 2 -chloride Cation substitution (equimolar Tris-Cl (TrisCl) substituted for NaCl in bath solution) did not alter the current-voltage relationship of the M 2 current, but anion substitution (sodium isethionate and NaI for NaCl in the bath solution) markedly altered the magnitude and reversal potential of the current, as predicted for an anion-selective current. Currents shown are from voltage clamp steps to between Ϫ90 and 60 mV in 10-mV increments (V h ϭ Ϫ60 mV), imposed during activation of the M 2 current. Current families were obtained before and after changing bath solution from the control to the test solution. We confirmed the role of aPKC in receptor-effector coupling by direct injection of aPKC into the cytosol (Fig. 4). Injection of recombinant PKC activated the chloride current in a concentration dependent fashion, with currents observed at final concentrations of PKC as low as 1 nM. Moreover, injection of phosphatidylinositol 3,4,5-trisphosphate (PIP 3 ), which stimulates PKC in vitro (12), activated the current. Current activation was specific to PIP 3 injection. Phosphatidylinositol 4,5diphosphate, phosphatidic acid, phosphatidylcholine, linolenic acid, phosphatidylserine, and arachidonic acid failed to induce any current (data not shown). PIP 3 and PKC currents were indistinguishable from that observed following M 2 receptor stimulation in several respects. First, PIP 3 and PKC activated an anion selective current with a permeability sequence of I Ϫ Ͼ Cl Ϫ Ͼ isethionate and a current-voltage relationship identical to the M 2 current. Second, the current was sustained in the absence of extracellular calcium following activation by either agent and was insensitive to calcium chelation. Third, currents activated by M 2 receptor stimulation and by PIP 3 or PKC injection were not additive. Following activation of the current by 50 M mACH, little or no further current was elicited by the injection of PIP 3 or PKC (n ϭ 6), and PIP 3 or PKC, current activation abrogated subsequent mACH current (n ϭ 5) (Fig.  4A). Finally, the aPKC pseudosubstrate inhibitor blocked both the M 2 and PIP 3 -induced chloride currents (Fig. 4, A and B). PIP 3 has been shown to stimulate novel PKC-⑀ and PKC- (30) as well as atypical PKC activity, and the block of PIP 3 currents by the aPKC pseudosubstrate peptide indicates the specific nature of PKCs mediating current activation. Moreover, neither diacylglycerol analogues nor phorbol esters activated the current (although 1-oleoyl-2-acetyl-sn-glycerol weakly activated the calcium-activated chloride current), and M 2 receptorchloride current coupling was not affected by prior exposure to these agents (Fig. 4A). Thus pharmacologic and peptide inhibitors that are selective for aPKCs blocked activation of the M 2 chloride current, and lipid activators of aPKCs (but not known activators of conventional and novel PKCs) and PKC evoked the current in the absence of M 2 receptor stimulation, indicating that stimulation of aPKC is both necessary and sufficient to open M 2 -coupled chloride channels.

FIG. 2. Activation of the metabotropic chloride current is not common to all G i /G o -linked receptors.
A and B, injection of subtypespecific antibodies directed against G i /G o proteins inhibited muscarinic receptor-chloride current coupling with no effect on the M 3 current, whereas anti-G q antibodies had no effect on the M 2 current, but blocked M 3 calcium release and the attendant current. C, stimulation of heterologously expressed inhibitory G protein-coupled M 4  Hormone-stimulated PI-3K activity results in the formation of PIP 3 , which has been implicated as a second messenger in a wide variety of cellular processes such as glucose transport, actin rearrangement, chemotaxis, and apoptosis (see Ref. 7). Activation of chloride channels by PIP 3 suggested the involvement of PI-3K in the stimulatory pathway linking M 2 receptor stimulation to chloride channel opening. Consistent with such a signaling cascade, the covalent PI-3K inhibitor wortmannin blocked the M 2 -stimulated current (Fig. 5), with no effect on M 3 coupling (not shown). Two groups of hormone-stimulated (class I) PI-3Ks have been identified based on activation by tyrosine kinase or by G protein-coupled signaling pathways. The former kinases are heterodimers composed of a catalytic p110 subunit that is stimulated following interaction with a p85 or p55 regulatory subunit containing SH2 and SH3 domains, whereas G protein-stimulated PI-3 kinase activity occurs through the direct activation of p110␥ by the binding of G␤␥ subunits (9,10,(31)(32)(33). To determine the identity of the PI-3K involved in M 2 coupling to aPKC, we used a peptide fragment of adenylyl cyclase 2 (QEHA peptide) containing a putative G␤␥ binding motif, which blocks G␤␥ signaling to several protein targets (34) and has been shown to inhibit ␤␥ stimulation of PI-3K␥ in vitro (33). The QEHA peptide inhibited coupling between M 2 receptors and chloride channels by as much as 80%. Concen-tration-dependent inhibition of M 2 receptor-effector coupling by the peptide was quite similar to that observed for PI-3K␥ (33) (Fig. 5, A and B). The peptide had no effect on PIP 3 activation of the current, indicating that the block is upstream of PIP 3 formation (data not shown). Similarly, purified bovine transducin ␣ proteins, which bind free G␤␥ subunits with high affinity, strongly inhibited M 2 -chloride channel coupling (Fig.  5C). Injection of proteins to a final concentration of approximately 5 M inhibited the current by 77.5 Ϯ 5.3% (n ϭ 12), whereas injection of boiled transducin ␣ proteins was without effect. We also expressed a dominant negative p85 cRNA (⌬p85), which has been shown to prevent the activation of p85-regulated PI-3K (16). Expression of ⌬p85 or wild-type p85 had no effect on M 2 signaling when cRNAs were injected at concentrations up to 4-fold higher than M 2 receptor cRNA.
Finally, we sought to determine whether the PI-3K⅐aPKC coupling pathway was involved in physiological M 2 neurotransmission. Release of acetylcholine from vagal nerves stimulates M 3 and M 2 receptors on smooth muscle cells at neuromuscular junctions in the lung, bladder, viscera, and some vessels. Stimulation of M 2 receptors on isolated myo- In the experiment shown, the ⌬P85 construct was injected at a 3.6-fold higher concentration than the M 2 cRNA (1.8 g/l ⌬P85 and 0.5 g/l M 2 ). B, mean data from experiments described in A. Wort, wortmannin. C, concentration-dependent inhibition of the M 2 current by the QEHA peptide relative to the current evoked by mACH alone in paired experiments.
cytes activates a sustained cation current (I cat ) that mediates slow excitatory postsynaptic potentials (35)(36)(37)(38), although the coupling process is poorly understood. In voltage-clamped single tracheal smooth muscle cells, dialysis of the aPKC pseudosubstrate peptide selectively abolished I cat without affecting the large calcium-activated chloride current (I Cl(Ca) ) that is associated with M 3 -mediated calcium release (see Ref. 37) (Fig. 6), indicating that aPKC activation is necessary for M 2 coupling to I cat in smooth muscle. As shown, dialysis of the cPKC peptide had no effect on coupling. M 2 coupling to I cat , but not M 3 coupling to I Cl(Ca) , was also disrupted by dialysis of the QEHA peptide and by the PI-3 kinase inhibitor wortmannin, indicating that key elements of the signaling pathway linking M 2 receptors to chloride channels in Xenopus oocytes are required for coupling of these receptors in mammalian cells. It should be noted, however, that coupling of muscarinic receptors to nonselective cation is an example of signaling convergence in which discrete signals generated by the simultaneous stimulation of the M 2 and M 3 receptors are required for current activation (37,39,40). That is, cur-rent activation requires the release of intracellular Ca 2ϩ , although such release is not itself sufficient for channel opening without simultaneous M 2 receptor engagement (37). Not surprisingly, dialysis of tracheal myocytes with PIP 3 did not result in the activation of I cat (data not shown). Taken together, however, these data indicate that the M 2 ⅐PI-3K␥⅐aPKC coupling pathway likely underlies physiological postsynaptic muscarinic signaling in smooth muscle.
In summary, we demonstrate a novel signaling pathway linking M 2 receptors to metabotropic ion channels. Receptor binding results in the stimulation of the G␤␥-regulated PI-3K␥, formation of PIP 3 , and activation of aPKC. This signaling pathway leads to the opening of a novel, second messenger-activated chloride current in Xenopus oocytes and mediates activation of nonselective cation channels in smooth muscle cells. These findings define a novel signaling cascade linking G protein-coupled receptors to membrane ion channels and provide further insight into the intricate role of lipid second messengers in receptor signaling.