I. Introduction

Top Next References

Upon stimulation of vagal nerves, acetylcholine (ACh)^c is released from axonal terminals and decelerates the heart beat. This historic discovery by Otto Loewi in the 1920s established the concept of chemical synaptic transmission (Loewi, 1921; Loewi and Navaratil, 1926). Since then, many physiologists have been trying to elucidate the mechanism(s) underlying neurotransmitter (Vagusstoff)-induced bradycardia. Del Castillo and Katz (1955) first described hyperpolarization of the membrane induced by ACh in frog heart. Hutter and Trautwein (1955) measured an increase of K⁺ efflux across the cardiac cell membrane with vagal stimulation. Trautwein and Dudel (1958) showed an increase of K⁺ conductance under voltage-clamp conditions. Trautwein and colleagues analyzed the kinetics of the ACh-induced K⁺ current in the rabbit sinoatrial node and proposed that ACh induces activation of a specific population of K⁺ channels, named muscarinic K⁺ (K_ACh) channels, to decelerate pacemaker activity (Noma and Trautwein, 1978; Osterrieder et al., 1981). The single channel currents of the K_ACh channels were recorded for the first time by Sakmann et al. (1983), who showed that the channel exhibited kinetic properties that clearly differed from those of the background inwardly rectifying K⁺ (I_K1) channel in cardiac myocytes.

The next big step was the discovery that pertussis toxin (PTX)-sensitive heterotrimeric G proteins are involved in the activation of the K_ACh channel by M₂-muscarinic and A₁ adenosine receptors (Pfaffinger et al., 1985; Breitwieser and Szabo, 1985; Kurachi et al., 1986a and b). Because the K_ACh channel could be activated by intracellular guanosine 5'-triphosphate (GTP) (in the presence of agonists) and GTPS (even in the absence of agonists) in cell-free inside-out patches, the system seemed to be delimited to the cell membrane, which led to the proposal that the channel is directly activated by G proteins (Kurachi et al., 1986a,b,c). The G protein responsible for activation of the K_ACh channel was designated G_K according to its function (Breitwieser and Szabo, 1985).

It was quite a surprise that the subunit (G) but not the  subunit (G) of the G_K protein, was proposed to mediate the G_K-induced activation of K_ACh channels (Logothetis et al., 1987, 1988; Kurachi et al, 1989a), because it was strongly believed at that time that regulation of different effectors by G proteins was mediated only by G, although G merely served to bind to the GDP-form of G (G_-GDP) to anchor the trimeric G protein to the cell membrane (Gilman, 1987). Actually, Brown, Birnbaumer, and their colleagues proposed G_K and not G_K as the physiological activator of K_ACh channels (Yatani et al., 1987, 1988; Codina et al., 1987; for review see Brown and Birnbaumer, 1990). The dispute concerning the G protein subunit responsible for the physiological activation of K_ACh channels continued for nearly a decade (Ito et al., 1992; Yamada et al., 1993, 1994a,b; Nanavati et al., 1990; Kurachi, 1989, 1990, 1993, 1994, 1995; Kurachi et al., 1992; Clapham and Neer, 1993; Wickman and Clapham, 1995) until the functional interaction between the channel and G was shown at the molecular level with cloned G protein-gated K⁺ (K_G) channel and/or G protein subunits (Kubo et al., 1993b; Dascal et al., 1993; Wickman et al., 1994; Reuveny et al., 1994; Krapivinsky et al., 1995a; Inanobe et al., 1995b). Now it is established that G_K is the physiological activator of K_G channels not only in cardiac myocytes, but also in neurons and endocrine cells. Recently, it was indicated that G protein-inhibition of neuronal Ca²⁺ channels is also mediated by G and not by G (Herlitze et al., 1996; Ikeda, 1996). Efforts are now being made to elucidate the molecular mechanisms underlying G-control of K_G and N-type Ca²⁺ channels.

The importance of the G protein-activation of K_G channel system in receptor-mediated regulation of cell responses is now more widely appreciated than before because a wide variety of membrane receptors, such as M₂-muscarinic, A₁ adenosine, ₂-adrenergic, D₂ dopamine, µ-, -, and -opioid, 5-HT_1A serotonin, somatostatin, galanin, m-Glu, GABA_B, and sphingosine-1phosphate receptors, have been shown to use this system in inhibiting cell excitation in various organs (North et al., 1987; Lacey et al., 1988; Hille, 1992a; Grudt and Williams, 1993; Oh et al., 1995; Saugstad et al., 1996; Sharon et al., 1997; Bünemann et al., 1995; Koppen et al., 1996). In this review, we will first summarize ACh-activation of cardiac K_ACh channels, the prototype of this system, and then recent progress in molecular dissection of the K_G channel system.

	II. Functional Analysis of G Protein-Mediated Activation of Muscarinic K+ Channels in Cardiac Atrial Myocytes

Top Previous Next References

ACh added to the extracellular solution elicits a K_ACh channel current in cardiac atrial myocytes (fig. 1). The activation time-course is sigmoidal and takes several hundred milliseconds to reach a peak (Breitwieser and Szabo, 1988). Thereafter, the evoked current gradually decreases to a quasi-steady-state level within 1 min in the presence of high concentrations of ACh (> 0.3 µM). This reduction of cell K⁺ current in the continuous presence of ACh is called "short-term" desensitization (Kurachi et al., 1987b). After wash-out of the agonist, the current disappears within several seconds (deactivation). It is worth noting that in the inside-out patch configuration of the patch-clamp method, one measures K_ACh channel activity only in the steady-state phase. Thus, in these experiments, limited information is available regarding the desensitization of the channel.

View larger version (9K): [in this window] [in a new window]   Fig. 1.   Time-dependent response of the whole-cell muscarinic K+ channel current to acetylcholine. By using the whole-cell voltage clamp method of the patch-clamp technique, the response of the whole-cell current of a guinea-pig atrial myocyte to 11 µM acetylcholine (ACh) was measured. In the presence of normal Tyrode solution that contained 5.4 mM external K+, the cell membrane potential was clamped at 53 mV. The patch pipette contained (in mM): 150 KCl, 2 MgCl2, 5 EGTA, 5 HEPES, and 0.1 GTP (pH = 7.3). ACh was applied to the bath for the period indicated by the horizontal bar above the cell membrane current trace. An arrowhead indicates the zero current level. An upward deflection of the cell current record indicated an outwardly directed cell membrane current that would be carried by the movement of K+ ions under these circumstances.

These three phases of the response involve interactions between an agonist (i.e., ACh), an M₂-muscarinic receptor, a PTX-sensitive G protein, and the K_ACh channel. Therefore, to understand the reaction of the K_ACh channel to ACh, it is necessary to know how the receptor-generated signal is transferred to the channel through the G protein and how this signal transmission might be modulated by other factors interacting with these different reactions.

1. The G protein cyclic reaction mediating the receptor-to-channel signal transmission. Activation of K_ACh channel induced by M₂-muscarinic receptor stimulation is mediated by a heterotrimeric G protein (G_K) (fig. 2). The heterotrimeric G proteins are membrane-bound proteins which transduce signals from receptors to effectors such as adenylyl cyclase, phospholipase C, the K_ACh channel, and other ion channels (Gilman, 1987). These proteins are composed of , , and  subunits (G, G, and G, respectively). Up to now, at least 16 G, 5 G, and 11 G genes have been identified (Bourne, 1997). Heterotrimeric G proteins interact with receptors through G. It is well known that the interaction between M₂-muscarinic receptors and G_K is blocked by the toxin from Bordetella pertussis (PTX) (Ui, 1984; Kurose et al., 1986). PTX modifies covalently a cysteine residue at the carboxyl-terminal end of G subunits belonging to G_i, G_o, and G_t families by transferring an ADP-ribose group from the nicotinamide adenine dinucleotide moiety to the cysteine residue (Gilman, 1987). Because the receptor-mediated activation of K_G channels in cardiac atrial myocytes and neurons are inhibited by PTX (Pfaffinger et al., 1985; Kurachi et al., 1986a), G_K seems to belong to one of these G protein families. However, its molecular identity has not been fully elucidated, although G_K is proposed to be a member of the G_i class of G proteins in some systems (Kozasa et al., 1996; Takano et al., 1997)

View larger version (41K): [in this window] [in a new window]   Fig. 2.   Schematic representation of the G protein cycle involved in the activation of the muscarinic K+ channel in response to acetylcholine.

- GDP

-GDP

-GTP

-GTP

View this table: [in this window] [in a new window]   TABLE 1 G protein effectors regulated by G protein subunits

2. Activation phase. The time to peak of the ACh-induced response of the K_ACh channel is dependent on ACh concentration: the higher the concentration of ACh, the faster the activation. In the presence of a maximum effective concentration of ACh, the time to peak is several hundred milliseconds. If the M₂-muscarinic receptor, G_K, and the K_ACh channel encountered by simple diffusion in the membrane, the response time requires that all these signaling molecules be within less than 1.5 µm of each other (Hille, 1992a). The molecular mechanism satisfying such a topological requirement has not been clearly identified. However, it was recently suggested that K_ACh channel subunits may directly interact with not only G_K but also G_K, trimeric G_K, and the receptor and thereby might form a complex with these proteins (Huang et al., 1995; Slesinger et al., 1995) (detailed in the Sections III.F.3. and 4.).

3. The phase of short-term desensitization. Short-term desensitization becomes more prominent as the concentration of ACh is increased above 0.3 µM (Kurachi et al., 1987b). This may at least partly arise from the transition of M₂-muscarinic receptors from the high to low affinity-binding state due to dissociation of G_K from receptors after agonist application (Gilman, 1987). Recent studies demonstrated that heterologous coexpression of RGS proteins with M₂-receptors and recombinant K_G channels reestablishes the short-term desensitization, which normally cannot be seen in the absence of RGS proteins in the reconstituted system (Doupnik et al., 1997). Therefore, this protein may also be one of the molecules responsible for the short-term desensitization of the K_G channel system. Other possible candidates for the short-term desensitization include phosphorylation of M₂-muscarinic receptors by -adrenergic receptor kinase (ARK), dephosphorylation of K_ACh channels and functional modulation of G proteins.

4. Deactivation of the response of the muscarinic K⁺ channel. The ACh-induced K⁺ current disappears quickly when the agonist is washed out from the extracellular solution (fig. 1). The rate of deactivation of the whole-cell K_ACh channel current was estimated as ~30 to 200/min, which is much more rapid than either the GTP hydrolysis rate of G proteins (~1 to 5/min) or the rate of dissociation of G from K_G channel subunits (~0.01/min) estimated in vitro (Breitwieser and Szabo, 1988; Nakajima et al., 1992; Gilman, 1987; Doupnik et al., 1997; Krapivinsky et al., 1995c). This discrepancy might in part be attributed to positive cooperativity in the interaction between the channel and G_K that will be described in the Section II.B.2., where even a slight decrease in free G_K concentration in the membrane should cause a larger reduction of the channel activity.

4

1. Single-channel characteristics of the muscarinic K⁺ channel. Fig. 3A shows single-channel recording of the K_ACh channel obtained from a cell-attached membrane of a guinea-pig atrial myocyte (Kurachi et al., 1986a). In general, K⁺ ions flow through K⁺ channels depending on the electrochemical gradient for K⁺ ions across the plasma membrane. This gradient is the difference between the membrane potential (V_m) and the K⁺ equilibrium potential (E_K): V_m  E_K. The single-channel current flowing through a K⁺-selective channel can be described as follows:

(1)

where i is the single-channel current amplitude, and  is the single-channel conductance of the channel. The current is positive (outwardly flowing across the membrane) at V_m positive to E_K, although it is negative (inwardly flowing) at V_m negative to E_K.

View larger version (26K): [in this window] [in a new window]   Fig. 3.   Single-channel properties of the muscarinic K+ channel. A: Cell-attached recordings of muscarinic K+ channel currents from a single guinea pig atrial myocyte at different membrane potentials. The patch pipette contained 150 mM K+ and 5.5 µM acetylcholine, although K+ concentration in the bath was 5.4 mM. The resting membrane potential (Er) of the cell was 52 mV. Membrane potentials are indicated to the left of each trace as the difference from Er. Arrowheads indicate the zero current level. Downward reflection of the current record represent ion current passing inwards into the cell from the pipette. Upward deflections represent passing from the cell outwards into the pipette. B: The single-channel current-voltage relationship of the muscarinic K+ channel shown in A. The line was fitted to the data by eye, and the single channel conductance was 46 pS at potentials between Er 60 and Er +40 mV. C: Open time histogram of the muscarinic K+ channel at Er. The line is the fit of the data with a single exponential curve with a time constant of 1.35 msec. [Modified from Kurachi et al. (1986a)].

13

1

2. Positive cooperative effect of GTP on muscarinic K⁺ channel activity. Activation of K_ACh channels by intracellular GTP (GTP_i) can be reproduced in inside-out patch membranes of atrial myocytes in the presence of ACh in the pipette (Kurachi et al., 1986a, 1990; Ito et al., 1991). Fig. 4 shows the concentration-dependent effect of GTP_i in the presence of different concentrations of ACh.

View larger version (29K): [in this window] [in a new window]   Fig. 4.   Concentration-dependent effect of intracellular GTP on the muscarinic K+ channel in the absence and presence of acetylcholine. A: Examples of inside-out patch experiments obtained from guinea-pig atrial myocytes. The channel currents were recorded at 80 mV with the symmetrical 145 mM K+ solutions. The concentration of acetylcholine (ACh) in the pipette was 0 or 1 µM as indicated. The bars above each trace indicates the protocol of application of different concentrations of GTP or 10 µM GTPS to the internal side of the patch membrane. The 3- to 10-fold increase in GTP concentration resulted in a dramatic increase of N.Po of muscarinic K+ channels, indicating the existence of a highly cooperative process. B: The relation between the concentration of GTP and the N.Po of muscarinic K+ channels normalized to the maximum N.Po induced by 10 µM GTPS in each patch. Symbols and bars are mean ± SD. The continuous lines indicate the fit of the relationship between GTP and channel activity in the presence of each concentration of ACh with the following Hill equation: where f is the relative N.Po; Vmax, the maximum N.Po available in the presence of 10 µM GTPS; Kd, the apparent dissociation constant of GTP; and n, the Hill coefficient. [Reproduced with permission from Ito et al. (1991)].

(2)

K

K

K

3. Spectral analysis of the muscarinic K⁺ channel currents in the presence of different concentrations of intracellular GTP. Precise and reliable analysis of the single-channel kinetics of the K_ACh channel is difficult because multiple K_ACh channels are usually included in a single membrane patch of atrial myocytes (fig. 4A). In these cases, the spectral analysis of the channel currents (an analysis based on a frequency domain) is one of the most reliable and powerful ways to assess the channel kinetics (fig. 5). The power spectrum constructed from inside-out patch recordings of the K_ACh channel currents is always well fitted with the sum of two Lorenzian curves irrespective of GTP_i concentration (Hosoya et al., 1996). These observations indicate that the K_ACh channel possesses three distinct open/closed states. Because the channel possesses a single open state (Sakmann et al., 1983), the equilibrium of the states can be described as C2C1O, where O represents the open state although C1 and C2 are closed states. It is likely that the transition among these three states is responsible for the open and closure of K_ACh channel currents observed at the single-channel level (figs. 3A, 4A and 5A). The corner frequencies of the two Lorenzian functions (the frequencies at which the power of the each component is the half-maximum) were constant irrespective of GTP_i concentration (fig, 5B). The ratio of the powers of the two Lorenzians at 0 Hz was also unaffected by GTP_i concentration. These results indicate that the kinetics of the fast open-close transition of the channel is not a function of G_K activity. In other words, G_K activates the K_ACh channel without altering the channel's fast open-close kinetics.

View larger version (32K): [in this window] [in a new window]   Fig. 5.   Spectral analysis of the muscarinic K+ channel currents in an inside-out patch. A: The muscarinic K+ channel currents in the inside-out patch membrane of a guinea pig atrial myocyte. The channel currents were recorded at 60 mV with the symmetrical 150 mM K+ solutions. The patch pipette also contained 0.5 µM acetylcholine. Different concentrations of GTP were applied to the internal side of the patch membrane as indicated above each current trace. B: Power density spectra calculated from the data shown in A. Each spectrum could be fitted with the sum of two Lorentzian functions. F1 and F2 indicated by arrows indicate the corner frequencies of the slow and the fast Lorentzian components, respectively. [Reproduced with permission from Hosoya et al. (1996)].

U A

C2C1O

View larger version (34K): [in this window] [in a new window]   Fig. 6.   The relationship between GTP concentration and the fraction of the "available" state of the muscarinic K+ channel, and the concerted allosteric model of Monod, Wyman, and Changeux. A: The fraction of the "available" state (A/(A + U)) was calculated from inside-out membrane patch experiments such as shown in fig. 5. Symbols indicate the relationship between GTP concentration and the calculated fraction of the "available" state. Lines indicate the fit of the data with the Monod, Wyman, and Changeux allosteric model with different assumed numbers of n (see Section II.B.4.). B: Schematic representation of the Monod, Wyman, and Changeux allosteric model. In this scheme, each muscarinic K+ channel is assumed to be an oligomer composed of four identical subunits (i.e., n = 4). Each subunit is in either the tense (T) or the relaxed (R) state, which is represented by squares and circles, respectively. Each subunit in the T or R state binds with one dissociated G protein subunit (solid circles) independently of each other with the microscopic dissociation constant of KT or KR, respectively. In this model, all subunits in the same oligomer must change their conformations simultaneously. Therefore, the channel can be either T4 or R4. T4 and R4 are in the equilibrium through an allosteric constant L. [Reproduced with permission from Hosoya et al. (1996)].

4. A possible mechanism for the G protein-mediated increase in the functional numbers of muscarinic K⁺ channels. Recent studies have revealed that Kir channels, including K_ACh channel, have an oligomeric structure (Yang et al., 1995b; Krapivinsky et al., 1995a) that may underlie the positive cooperativity of the G_K protein/K_ACh channel interaction (Monod et al., 1965).

K

K

K

K

K

K

K

K

Although the G_K/K_ACh channel interaction is the essential step of G protein-mediated activation of the K_ACh channel, this reaction is modulated by many factors such as intracellular ATP, Na⁺ ions, and arachidonic acid metabolites. Intracellular ATP has been shown to activate native and recombinant K_G channels in an Mg²⁺_i-dependent manner (Otero et al., 1988; Heidbüchel et al., 1990; Kaibara et al., 1991; Kim, 1991; Lesage et al., 1995; Sui et al., 1996). Although the molecular mechanism underlying this phenomenon has not been unequivocally identified, PIP₂ may be involved in this phenomenon (Huang et al., 1998).

The activity of K_G channels pretreated with intracellular MgATP could be further enhanced by intracellular Na⁺ (Lesage et al., 1995; Sui et al., 1996). The site of action of Na⁺ is unknown. Sui et al. (1996) showed that intracellular Na⁺ increased the activity of the K_ACh channel (and also the corresponding recombinant K_G channel) with an EC₅₀ of ~40 mM mainly by increasing the frequency of the channel's opening. They found that priming of channels with MgATP was a prerequisite for the action of Na⁺. Lesage et al. (1995), however, found that 20 mM intracellular Na⁺ activated recombinant K_G channel whether or not they had been pretreated with MgATP. This discrepancy might have occurred due to the different subunit composition of the K_G channels used in these two studies. Interestingly, Sui et al. (1996) showed that a cardiac glycoside ouabain, an inhibitor of the Na⁺/K⁺ pump, induced the opening of the K_ACh channel. They found that the N*P_o value of the channel increased although the mean open time was unchanged, indicating that the activating effect of ouabain was probably mediated by accumulation of intracellular Na⁺ but not a possible local increase in ATP concentration. However, they did not directly measure intracellular Na⁺ concentration nor reported the apparent change in the reversal potential of the K_ACh channel that might be expected when intracellular K⁺ concentration decreased due to blockade of Na⁺/K⁺ pump. Therefore, further studies may be necessary to conclude that cardiac glycosides activate the K_ACh channel through accumulation of intracellular Na⁺. This phenomenon might, at least in part, underlie the "direct" negative chrono- and dromo-tropic effects of the agent on the heart.

Arachidonic acid (AA) metabolites are known to modulate K_ACh channels (Kurachi et al., 1989c; Kim et al., 1989; Yamada et al., 1994b). The effect of AA is mimicked by leukotriene C₄ (LTC₄) and specifically blocked by AA861, a 5-lipoxygenase inhibitor (Kurachi et al., 1989c). Therefore, the effect of AA may be mediated by LTC₄ or its metabolites. Although the site of action of LTC₄ has not been clearly identified, the complete dependency of the LTC₄ effect on the presence of GTP_i indicates that LTC₄ does not directly act on the K_ACh channel (Kurachi et al., 1989c). In the absence of receptor agonists, GTP_i usually induces only 20% of the maximum K_ACh channel activity in the inside-out patch membranes even when Cl is used as an intracellular anion. However, GTP_i fully activated the channel in an agonist-independent manner when the patches were pretreated with AA before patch excision (Kurachi et al., 1989c). Thus, AA metabolites may stimulate the basal turn-on reaction of G_K. Stimulation of K_ACh channels by platelet-activating factor or ₁-adrenergic receptors may be mediated by this second-messenger pathway (Nakajima et al., 1991, Kurachi et al., 1989b).

	III. Molecular Analysis of G Protein-Gated K+ Channels

Top Previous Next References

In 1993, the molecular structure of inwardly rectifying K⁺ channels (Kir) was disclosed. The cDNAs encoding an ATP-dependent Kir channel, ROMK1 (Ho et al., 1993), and a classical Kir channel, IRK1 (Kubo et al., 1993a), were isolated by expression cloning from the outer medulla of rat kidney and a mouse macrophage cell line, respectively (fig. 7). The primary structure of these channels were similar with two putative membrane-spanning regions (M1 and M2) and one potential pore-forming region (H5). This structure resembles that of the S5, H5, and S6 segments of the voltage-gated K⁺ (K_v) channels. Because the voltage-sensor of the K_v channel subunit exists in the S4 segment that possesses repeated positively-charged amino acid residues, Kir channel subunits lack an obvious voltage-sensor region. This is consistent with electrophysiological studies that show the kinetics of Kir channels apparently depends on the difference of Vm from E_K and not on Vm itself.

View larger version (22K): [in this window] [in a new window]   Fig. 7.   Evolutionary tree of Kir subunits. The tree was made using the UPGMA (Unweighted Pair Group Method with Arithmetic Mean) Tree Window in Geneworks (IntelliGenetics, Inc., Mountain View, CA).

After the cloning of ROMK1 and IRK1, the cDNAs encoding the main subunits of K_G and K_ATP channels (GIRK1 and BIR) were also cloned (Kubo et al., 1993b; Dascal et al., 1993; Inagaki et al., 1995a). All of these Kir channel subunits exhibit basically the same primary structure. So far, at least 11 cDNAs encoding Kir channel subunits have been isolated. The evolutionary tree of this family is depicted in fig. 7.

These cloned Kir subunit cDNAs encode proteins composed of 327 to 501 amino acids. The identity of the predicted amino acid sequences is ~30 to 40% among the members of the different Kir subfamilies and more than 60% among those in the same subfamilies. The highest level of sequence identity (50 to 60%) is found in the H5 region and the proximal part of the C-terminal cytosolic domain. The cloned Kir channel subunits have been classified at least into four groups (Doupnik et al., 1995a): (a) IRK (Kir2.x) subfamily made of the classical constitutively active "inward rectifier" Kir channels: IRK1 (Kubo et al., 1993a; Morishige et al., 1993), IRK2 (Koyama et al., 1994; Takahashi et al., 1994) and IRK3 (Morishige et al., 1994; Makhina et al., 1994; Pärier et al., 1994); (b) GIRK (Kir3.x) subfamily, corresponding to G protein-regulated K⁺ channels: GIRK1 (Kubo et al., 1993b; Dascal et al., 1993), GIRK2 (Lesage et al., 1994, 1995; Isomoto et al., 1996; Tsaur et al., 1995; Stoffel et al., 1995; Bond et al, 1995; Ferrer et al., 1995), GIRK3 (Lesage et al., 1994), GIRK4 (Ashford et al., 1994; Krapivinsky et al., 1995a; Chan et al., 1996), and GIRK5 (Hedin et al., 1996); (c) K_AB subfamily of ATP-dependent K⁺ channels (Kir1.1 and Kir4.1): ROMKs (Ho et al., 1993; Zhou et al., 1994; Yano et al., 1994; Shuck et al., 1994; Boim et al., 1995; Kondo et al., 1996) and K_AB-2 (Bond et al., 1994; Takumi et al., 1995); and (d) K_ATP subfamily (Kir6.x), the ATP-sensitive K⁺ channels: uK_ATP-1 and BIR (Inagaki et al., 1995a,b; Sakura et al., 1995).

Recent progress in the molecular biology of Kir channels has enabled us to study the structure-function relationship of biophysics, physiological regulation, and pharmacology of these channels at the molecular level.

GIRK1 was first isolated from the rat atrium (Kubo et al., 1993b; Dascal et al., 1993). From a mouse brain cDNA library, two additional homologues of GIRK1 were isolated and designated GIRK2 and GIRK3 (table 2) (Lesage et al., 1994). Furthermore, it has been shown that at least three different isoforms of mouse GIRK2 are generated by alternative splicing of transcripts from a single gene, and we designated them GIRK2A, GIRK2B, and GIRK2C in the order of identification (Isomoto et al., 1997). These alternatively spliced transcripts share an N-terminal end and a central core, and differ at their C-terminal ends. GIRK2B was isolated from mouse brain cDNA library and shown to be ubiquitously expressed in various tissues (Isomoto et al., 1996). Its amino acid sequence is shorter than that of GIRK2A by 87 amino acids. The eight amino acid residues in the C-terminal end of GIRK2B are different from those of GIRK2A. GIRK2C has a C-terminus which is longer than that of GIRK2A by 11 amino acids. GIRK2C was isolated from cDNA libraries of insulinoma cells and brain (Lesage et al., 1994, 1995; Tsaur et al., 1995; Stoffel et al., 1995; Bond et al., 1995; Ferrer et al., 1995).

GIRK2C was originally termed K_ATP-2 because it was thought to be a subunit of the K_ATP channel (Stoffel et al., 1995; Tsaur et al., 1995) due to its sequence similarity to cK_ATP-1, which was isolated by Ashford et al. (1994). However, GIRK4, which is virtually identical with rat cK_ATP-1, reconstitutes cardiac K_ACh channel with GIRK1 and does not contribute to the K_ATP channel as described in the Section III.D. (Krapivinsky et al., 1995a,b). Thus, it is now clear that both cK_ATP-1 and K_ATP-2 belong to the GIRK subfamily. GIRK5 was cloned from Xenopus oocytes (Hedin et al., 1996). Although its mammalian homologue has not been reported, the amino acid sequence of GIRK5 is most homologous to that of GIRK4 among mammalian GIRKs.

The GIRK clones contain various known functional motifs in their amino acid sequences that may be important for the physiological functions of the subunits in K_G channels (fig. 8). GIRK1 possesses an amino acid sequence homologous to the G-binding domain of ARK1 in its C-terminus, which is therefore the candidate for the site of G-binding to the K_G channel (Reuveny et al., 1994). As with all the other Kir channel subunits, GIRKs possess conserved cationic residues adjacent to the C-terminal end of the M2 domain. One of these positively charged residues, arginine (R) at position 188 of ROMK1, was shown to be critically involved in PIP₂-induced activation of rundown ROMK1 channels (Huang et al., 1998). Thus, it is conceivable that the corresponding residues in GIRK subunits (R190 for GIRK1; R201 for GIRK2s; R167 for GIRK3; and R196 for GIRK4) also participate in the PIP₂-induced activation of K_G channels. All of the GIRK clones have an arginine-glycine-aspartate (RGD) motif in their linker region between M1 and H5. This motif could be an integrin receptor-site (Hynes et al., 1992), whose role in K_G channels has not been examined yet. The characteristic feature of GIRK2C is the serine/threonine-X-valine/isoleucine (S/T-X-V/I) motif at its C-terminus end (Gomparts, 1996). This motif has been