Regulation of the Inward Rectifying Properties of G-protein-activated Inwardly Rectifying K+ (GIRK) Channels by Gβγ Subunits

Gβγ subunits are known to bind to and activate G-protein-activated inwardly rectifying K+channels (GIRK) by regulating their open probability and bursting behavior. Studying G-protein regulation of either native GIRK (IKACh) channels in feline atrial myocytes or heterologously expressed GIRK1/4 channels in Chinese hamster ovary cells and HEK 293 cells uncovered a novel Gβγ subunit mediated regulation of the inwardly rectifying properties of these channels. IKACh activated by submaximal concentrations of acetylcholine exhibited a ∼2.5-fold stronger inward rectification than IKACh activated by saturating concentrations of acetylcholine. Similarly, the inward rectification of currents through GIRK1/4 channels expressed in HEK cells was substantially weakened upon maximal stimulation with co-expressed Gβγ subunits. Analysis of the outward current block underlying inward rectification demonstrated that the fraction of instantaneously blocked channels was reduced when Gβγ was over-expressed. The Gβγ induced weakening of inward rectification was associated with reduced potencies for Ba2+ and Cs+ to block channels from the extracellular side. Based on these results we propose that saturation of the channel with Gβγ leads to a conformational change within the pore of the channel that reduced the potency of extracellular cations to block the pore and increased the fraction of channels inert to a pore block in outward direction.

G␤␥ subunits are known to bind to and activate Gprotein-activated inwardly rectifying K ؉ channels (GIRK) by regulating their open probability and bursting behavior. Studying G-protein regulation of either native GIRK (I KACh ) channels in feline atrial myocytes or heterologously expressed GIRK1/4 channels in Chinese hamster ovary cells and HEK 293 cells uncovered a novel G␤␥ subunit mediated regulation of the inwardly rectifying properties of these channels. I KACh activated by submaximal concentrations of acetylcholine exhibited a ϳ2.5-fold stronger inward rectification than I KACh activated by saturating concentrations of acetylcholine. Similarly, the inward rectification of currents through GIRK1/4 channels expressed in HEK cells was substantially weakened upon maximal stimulation with co-expressed G␤␥ subunits. Analysis of the outward current block underlying inward rectification demonstrated that the fraction of instantaneously blocked channels was reduced when G␤␥ was over-expressed. The G␤␥ induced weakening of inward rectification was associated with reduced potencies for Ba 2؉ and Cs ؉ to block channels from the extracellular side. Based on these results we propose that saturation of the channel with G␤␥ leads to a conformational change within the pore of the channel that reduced the potency of extracellular cations to block the pore and increased the fraction of channels inert to a pore block in outward direction.
G-protein-activated inwardly rectifying K ϩ channels (GIRKs) 1 are expressed in many areas of the brain and in supraventricular myocytes of the heart (1,2). Activation of G-protein-coupled receptors that couple to G i -proteins such as the M 2 muscarinic acetylcholine receptors (M 2 -mAChRs) lead to a dissociation of heterotrimeric G-proteins into activated ␣ subunits and ␤␥ dimers. G␤␥ subunits are known to bind to GIRK channels and increase the open probability of these channels (3,4). Cardiac I KACh channels are formed by heteromultimers of GIRK1 and GIRK4 subunits (4). The binding site of G␤␥ subunits to GIRK channels was mapped primarily to the C terminus of GIRK1 and GIRK4 (4 -8). Cross-linking studies have demonstrated that the heterotetrameric channel can bind up to 4 G␤␥ subunits (9). However, despite much experimental effort the mechanism by which G␤␥ activates these channels is not well understood.
GIRK channels belong to the family of strong inwardly rectifying K ϩ channels, which are characterized by their strong inwardly rectifying current-voltage relationships. The inward rectification has been linked to the presence of intracellular Mg 2ϩ and polyamines (10 -12). These positively charged cytoplasmic ions are thought to block outward K ϩ currents by blocking the pore of channels from the inside (10 -13); however, for a related inwardly rectifying channel Kir2.1 this hypothesis has recently been questioned (14). Inward rectification of K ϩ channels is not only voltage-dependent but also dependent on the extracellular K ϩ concentration (11). The inward rectification of these K ϩ channels is closely related to their function in the heart as well as in many neuronal tissues. In cardiac myocytes activation of inwardly rectifying K ϩ channels such as I KACh causes the cell membrane to hyperpolarize between action potentials because the conductivity for K ϩ generated by these channels is high at membrane potentials close to E K . This hyperpolarization induced by I KACh appears to be at least partially responsible for the negative chronotropic effect induced by vagal activity (1,2,15). During action potentials, however, the conductivity of I KACh for K ϩ declines several-fold with the rise of voltage enabling the myocyte to generate prolonged action potentials, which are critically important for cardiac function (16).
The initial observation that led to the study presented here was the discovery that the agonist-induced I KACh in cardiac myocytes were quite variable in their degree of inward rectification, 2 indicating that the modulation of the open probability of these channels (15) by ACh may not be the only property of these channels that is regulated by ACh. It seemed possible that, in addition, inward rectification of these channels may be modulated as well by ACh. The present experiments have tested this possibility.

EXPERIMENTAL PROCEDURES
Preparation of Feline Atrial Myocytes-Isolation of feline atrial myocytes was performed as described (17). Animal procedures used were in accordance with guidelines of the Animal Care and Use Committee of Northwestern University. Briefly, adult cats were first anesthetized with pentobarbital sodium (70 mg/kg body weight, intraperitoneally). The heart was quickly removed and retrograde perfused with Krebs-Henseleit buffer. It was digested by perfusion with collagenase-containing solution. After 10 -15 min of digestion the atria were collected and cut into small pieces, followed by a 5-min incubation with fresh enzyme solution. Isolated atrial myocytes were collected, placed in M199 (In-vitrogen), and plated in cell culture dishes. The cells were kept at 37°C under 7% CO 2 until further use.
Measurement of Membrane Currents-Membrane currents were recorded under voltage-clamp conditions, using conventional whole cell patch clamp techniques (20). Patch-pipettes were fabricated from borosilicate glass capillaries, (GF-150 -10, Warner Instrument Corp.) using a horizontal puller (P-95, Fleming & Poulsen). The DC resistance of the filled pipettes ranged from 3-6 M⍀. Membrane currents were recorded using either a patch-clamp amplifier (Axopatch 200, Axon Instruments) or an EPC 9 (HEKA Instruments) as described previously (21,22). Signals were analog-filtered using a lowpass Bessel filter (1-3 kHz corner frequency). Data were digitally stored using either a Mac (Centrion 640 with pulse software) or 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 (23). Voltage ramps (from Ϫ120 mV to ϩ60 mV in 500 ms, every 10 s) were used to determine current-voltage (I-V) relationships.
All measurements were performed at room temperature. Summarized results are presented as mean values Ϯ S.E. Student's t tests (two population) were performed to test for significance of differences between groups of data.

RESULTS
The atrial muscarinic K ϩ current (I KACh ) is regulated by muscarinic receptors, and the underlying pathway has been studied in detail by many groups (15). The inwardly rectifying properties of this channel have been the topic of many detailed studies that provided interesting insights into the mechanisms of inward rectification (10,11). However, so far no physiological modulation of the inward rectification of this current has been reported. The following study was based on the surprising observation that the inward rectification of ACh-evoked K ϩ currents in feline atrial myocytes varied as a function of the agonist concentration.

The Inward Rectification of Feline Atrial I KACh Was Modulated by Stimulus
Strength-I KACh in isolated feline atrial myocytes was measured in response to two different concentrations of ACh either in the inward or outward direction using the whole cell patch technique. The membrane potential was clamped to either Ϫ90 mV or ϩ60 mV in the presence of 20 mM extracellular K ϩ . When the holding potential was negative (Ϫ90 mV) to the potassium equilibrium potential (E K ) (Fig. 1A), superfusion of the cell with 0.1 M ACh gave rise to inward currents that were about 70% in amplitude compared with currents activated by 10 M ACh. In contrast, at ϩ60 mV outward currents induced by 0.1 M ACh were barely detectable and were only about 10% in amplitude compared with currents activated by 10 M ACh. I-V curves of ACh-induced currents ( Fig. 1B) were determined by subtracting currents measured in the absence of agonist from currents measured in the presence of agonist in response to linear voltage ramps from Ϫ120 mV to ϩ60 mV. I-V curves of the currents elicited by 0.1 M ACh or by 10 M ACh exhibited inward rectification and identical reversal potentials close to the E K (Fig. 1B), suggesting that the currents generated by either concentration of ACh were attributable to activation of I KACh . However, inward rectification of the current activated by 0.1 M ACh was found to be considerably stronger than that of the current activated by 10 M (Fig. 1B). Plotting GIRK conductance (normalized to the amplitude measured at Ϫ90 mV) against voltage shows a 15-20 mV shift to more positive potentials when currents were activated by 10 M compared with 0.1 M ACh (Fig. 1C).

Heterologously Expressed GIRK Currents Exhibited Strong Inward Rectification When Activated via Endogenous G-proteins-
Whole cell currents were measured in CHO-K1 or HEK 293 cells transfected with GIRK 1 and 4 as described before to prove that GIRK channels are responsible for the observed currents in feline atrial myocytes (21,24). G i -coupled receptors such as the M 2 -mAChR or A 1 adenosine receptors activated GIRK currents in response to agonist (Fig. 2, A and B). Although the size of the Heterologously Expressed GIRK Currents Exhibited Weakened Inward Rectification When Activated via Co-expressed G␤␥-Agonist-induced GIRK currents obtained from cells heterologously transfected with GIRK1/4 were activated via endogenous G-proteins (Fig. 2, A and B). It seemed likely that the pool of endogenous G-proteins might have been limiting for the extent of maximal GIRK current activation. Therefore, G␤ 1 ␥ 2 subunits were co-expressed with GIRK channels. GIRK currents were constitutively active due to G␤␥ subunits. The amplitude of GIRK currents was determined via inhibition by Ba 2ϩ (Fig. 2C). In most cases, activation of co-expressed A 1 adenosine receptors by 10 M Ado induced no further stimulation of Ba 2ϩ -sensitive GIRK currents, indicating a maximal stimulation of GIRK channels by G␤␥. Under these conditions, total Ba 2ϩ -sensitive GIRK currents compared with control conditions (activation via receptor and endogenous G-proteins) were about 2-fold larger in amplitude (147 Ϯ 12.7 pA/pF with co-expressed G␤ 1 ␥ 2 versus 71.2 Ϯ 20.7 pA/pF activated via A 1 adenosine receptors and endogenous G-proteins) and exhibited a weaker inward rectification (Fig. 2D versus Fig. 2B). Comparing the conductance-voltage relationship revealed a shift to more positive voltages for GIRK currents activated by heterologously expressed G␤␥ subunits compared with GIRK currents activated by agonist only (Fig. 2E). To quantify the relative inward rectification of GIRK currents the ratio of GIRK current conductance in outward versus inward direction (Ba 2ϩsensitive GIRK currents at reversal potential (E rev ) Ϯ 50 mV) was calculated. The ratio of outward/inward currents of Ba 2ϩsensitive GIRK currents activated by heterologously expressed G␤␥ was significantly increased compared with Ba 2ϩ -sensitive GIRK currents activated via A 1 adenosine receptors and endogenous G-proteins (0.39 Ϯ 0.11, n ϭ 9 versus 0.14 Ϯ 0.05, n ϭ 8) (Fig. 2F). The voltage-dependence of GIRK currents maximally activated by heterologous expression of G␤␥ was comparable with atrial I KACh activated by saturating concentrations of ACh (10 M), whereas agonist-induced GIRK currents activated via endogenous G-proteins exhibited similar strong inward rectification as submaximally activated atrial I KACh . This result indicated that the inwardly rectifying properties of GIRK channels were modulated depending on the internal G␤␥ concentration.
In a minority of cells transfected with G␤␥ subunits addition of adenosine to stimulate A 1 adenosine receptors resulted in a further increase in GIRK currents (Fig. 3A), indicating submaximal stimulation of GIRK channels by heterologously expressed G␤␥. Under these circumstances, basal G␤␥-induced GIRK currents exhibited strong inward rectification, whereas addition of adenosine resulted in a pronounced weakening of inward rectification (Fig. 3B), demonstrating that inward rectification of heterologously expressed GIRK currents can be modulated via stimulation of G-protein-coupled receptors similar to atrial myocytes. Taken together, these results suggested that G␤␥ at submaximal concentrations induces strong inwardly rectifying GIRK currents, whereas at maximal concen-FIG. 2. G␤␥ modulates inward rectification of heterologously expressed GIRK currents. HEK 293 cells were transiently transfected with cDNAs for GIRK1, GIRK4, and A 1 adenosine receptors and with G␤ 1 ␥ 2 (C, D, E, F, indicated as ϩG␤␥) or without additional exogenous G-protein subunits (A, B, E, F, indicated as agonist). GIRK currents activated via A 1 adenosine receptors or by co-expression with G␤␥ subunits were measured using whole cell voltage clamp recording similar as described in the legend to Fig. 1. GIRK current-voltage curves were calculated by subtracting either background currents in the absence of agonist (B) or currents insensitive to 1 mM Ba 2ϩ (D). Normalized GIRK current conductance in cells co-transfected with or without G␤␥ subunits were plotted against voltage (E). To quantify the degree of inward rectification, an inward rectification factor was defined (F ir ϭ I(E rev Ϫ 50 mV)/I(E rev ϩ 50 mV)) (F). Summarized data were compared for adenosine-evoked currents in the absence of exogenous G␤␥ (agonist) and Ba 2ϩ -sensitive currents evoked by heterologous expression of G␤␥ (F) (n ϭ 9 each, the two groups were significantly different at p Ͻ 0.05).

FIG. 3. Agonist-mediated modulation of inward rectification in cells submaximally stimulated with transfected G␤␥.
In a minority of cells transfected with G␤␥, stimulation of co-expressed A 1 adenosine receptors caused, in addition to the constitutively active GIRK currents, a further increase in GIRK currents (A). I-V curves for the Ba 2ϩ -sensitive currents in the absence (a-c) and presence of 1 M adenosine (Ado, b-c) as well as the currents that were stimulated by Ado in addition to the constitutively active currents (b-a) were determined (B). Note a substantial weakening of GIRK current inward rectification in response to adenosine.
The Ratio of G␤␥ to GIRK Channel Expression Is Critical for Regulation of the Inward Rectification of GIRK Currents-If the ratio of GIRK channels versus available G␤␥ in the cells is important for the degree of inward rectification as suggested by these results, it should be possible to achieve a high ratio of endogenous G␤␥ to GIRK channels by lowering the GIRK channel expression. Contrarily, a strengthening of the inward rectification should occur when GIRK channel expression is increased relative to the G␤␥ expression. We attempted to counteract the G␤␥ mediated weakening of inward rectification by transfecting HEK 293 cells with steady amounts of G␤␥ subunits but increasing amounts of GIRK1 and GIRK4 subunits as illustrated in Fig. 4. G␤␥ expression in the presence of co-expressed GIRK channels decreased cell survival. Therefore, we choose to co-express an ␣ 2A -adrenergic receptor fused to a G i ␣ 1 -protein (26) to reduce constituitive G␤␥ signals. GIRK channels were stimulated using saturating concentrations of norepinephrine (10 M), and subsequently GIRK currents were blocked by superfusion of the cells with 1 mM BaCl 2 to determine background currents. The ratio of outward to inward GIRK currents significantly declined with increasing amounts of GIRK channels transfected (Fig. 4, upper panel). The GIRK current density measured at Ϫ90 mV increased with increasing amounts of GIRK1/4 channel expression (Fig. 4, lower panel), suggesting that G␤␥ expression was not limiting for maximal inward GIRK currents in cells transfected with 0.1 g cDNA/5 cm dish of GIRK1/4. These results supported the hypothesis that inward rectification of GIRK channels is modulated dependent on the ratio of G␤␥ subunits relative to GIRK channels. In addition, we tried to lower GIRK expression relative to endoge-nous G-proteins by prolonging the time after transfection and found a significant increase in the ratio of outward to inward currents from day 3 to 4 post-transfection in transiently transfected CHO cells (I out /I in : 0.22 Ϯ 0.06 d.4 versus 0.095 Ϯ 0.025 d.3) accompanied by a small reduction in GIRK current density determined at Ϫ90 mV (43 Ϯ 13 pA/pF, d.4 n ϭ 12 compared with 64 Ϯ 10 pA/pF, d.3 n ϭ 6). This weakening of inward rectification of GIRK currents reflected most likely a decrease in GIRK channel expression in the individual cells, resulting in an increase of the ratio of G-proteins versus GIRK channels.

The Weakened Inward Rectification Was Not Accompanied by Changes in Slow Blocking Kinetics of Outward GIRK Currents Attributed to Polyamine-induced Inward Rectification-
For further analysis experimental conditions were chosen to consistently induce either strong inward rectifying currents (control) or weak inward rectifying currents (G␤␥-induced) in HEK cells stably expressing GIRK1 and 4. Strong inward rectifying currents were induced via agonist stimulation of ␣ 2A adrenergic receptors in the absence of exogenous G␤␥, whereas weak inward rectifying currents were evoked by additional co-transfection of G␤ 1 ␥ 2. As described above the current model of the inward rectifying mechanism is a voltage-dependent open channel block by internal Mg 2ϩ and polyamines such as spermine and spermidine. To test whether an alteration of the polyamine-and Mg 2ϩ -induced open channel block was the cause for the observed weakening of inward rectification, blocking and unblocking kinetics were determined using whole cell recording. According to Refs. 13, 27, and 28, the polyamine block is responsible for the time-dependent (slow) activation and inactivation of K ϩ currents through GIRK channels (or other inward rectifier channels) in response to voltage steps, whereas current block induced by internal Mg 2ϩ occurs almost instantaneously. Therefore, whole cell currents resulting from voltage steps (Ϫ120 mV to 60 mV; 60 mV to Ϫ120 mV) were measured to determine the time constants of polyamine block onset and offset. In case inward rectification was weakened due to lowered polyamine block affinity, a faster polyamine unbinding from the channel and/or a slower-polyamine binding to the channel should be observed. In contrast, if Mg 2ϩ block was altered, the fraction of channels blocked instantaneously in outward direction should be decreased, whereas changes in blocking and unblocking kinetics should not be observed. Background currents were determined by inhibiting GIRK channels via Ba 2ϩ and subtracted from each measured whole cell current. A second-order exponential function was used to fit the current curves and determine time constants. Comparison of currents measured under control (strong inward rectification) and G␤␥ over-expressed (weak inward rectification) conditions showed no striking alteration of the slow blocking kinetics (Fig.  5A). As expected, the unblocking appeared to be faster (Fig.  5B), however, this effect did not reach statistical significance (1.28 ms Ϯ 0.1 versus 1.05 ms Ϯ 0.46; 10 ms Ϯ0.95 versus 9.1 ms Ϯ 2.8). In contrast to the proposition, blocking of the channel in the outward direction (reflecting binding of polyamines) was faster, too (4.9 Ϯ 0.97 ms versus 2.75 Ϯ 0.62 ms; 58.8 Ϯ 16 ms versus 36 Ϯ 9.3 ms). Normalizing to the maximum inward current revealed that the probability of channel opening at voltages positive to E K was increased under weak inward rectifying conditions compared with control conditions. Normalizing to the outward maximum current demonstrated that the same percentage of channels underwent a slow blockade under control as well as under weak inward rectifying conditions. Because the fraction of channels instantaneously blocked in the outward direction was lower when G␤␥ was over-expressed the potency of internal Mg 2ϩ to block the channels might have been reduced. Therefore, we increased internal Mg 2ϩ up to 20 mM to compensate for a reduced potency of Mg 2ϩ to block GIRK channels, however, no change in inward rectification was observed (data not shown).
Affinity for Ba 2ϩ Block Was Reduced under Weak Inward Rectifying Conditions-A hallmark for strong inward rectifier potassium channels is a high affinity block by external Ba 2ϩ . Studies using crystal structures of the bacterial KcsA channel complexed with Ba 2ϩ have located a single Ba 2ϩ -binding site on the cytosolic side of the selectivity filter (29,30). In close proximity to this site are some of the residues that have been implicated to be critical for strong inward rectification (11,28). To test if G␤␥ mediates a conformational change of the GIRK channel that causes weakening of inward rectification by altering structures close to the selectivity filter, we questioned whether or not GIRK channel block by Ba 2ϩ was affected by G␤␥. Whole cell currents at a holding potential of Ϫ90 mV were measured in the presence of 1 M, 10 M, 40 M, 140 M, 1 mM, and 2 mM extracellular Ba 2ϩ under strong and weak inward rectifying conditions (Fig. 6). Ba 2ϩ effectively inhibited GIRK currents under both conditions, however, the potency of Ba 2ϩ to block GIRK currents was substantially decreased when channels were maximally activated by G␤␥ (IC 50 : 73 M versus 20 M; Hill coefficient: n ϭ 1.14 versus n ϭ 2). These results strongly suggested that interaction with G␤␥ subunits induced conformational changes of GIRK channel structures close to the Ba 2ϩ -binding site.
Cs ϩ -induced Block of GIRK Channels Was Attenuated under Weak Inward Rectifying Conditions-Inwardly rectifying K ϩ channels can be blocked efficiently by external Cs ϩ . This block is highly voltage-dependent and most prominent at negative potentials (11,31). Binding sites for Cs ϩ in the channel have been mapped to pore-lining residues of transmembrane domain 2 (M2) (32) and to a site close to selectivity filter (32). Therefore, possible G␤␥-dependent modulation of GIRK current block by external Cs ϩ (3 mM) was studied (Fig. 7). At a membrane potential of Ϫ90 mV, whole cell GIRK currents were inhibited under control (strong inward rectifying) conditions by 85 Ϯ 2.6%, whereas whole cell currents in the presence of heterologously expressed G␤␥ (weak inward rectifying conditions) were inhibited only by 28 Ϯ 4% (Fig. 7, A-C). To verify whether or not the attenuation of the Cs ϩ block by co-expression of G␤␥ was correlated to the G␤␥-mediated weakening of inward rectification, the degree of inward rectification (defined as F ir ϭ I(E rev Ϫ 50 mV)/I(E rev ϩ 50 mV)) was plotted against the potency of Cs ϩ to block GIRK channels. We obtained a close inverse correlation of the degree of inward rectification and the ability of Cs ϩ to block GIRK currents (Fig. 7D). This result suggested that a G␤␥-mediated conformational change of GIRK channels caused the reduced inward rectification and was mechanistically coupled to a reduction of the Cs ϩ block. We further analyzed the voltage-dependencies of the Cs ϩ block by comparing GIRK currents activated via endogenous G-proteins and selected GIRK currents activated via co-expressed G␤␥ subunits, but exhibiting a different degree of inward rectification (most likely due to different expression levels of G␤␥ subunits). Background-subtracted, current-voltage relationships of strong inward rectifying (F ir ϭ 0.10; no G␤␥ co-transfected) and medium and weakly inward rectifying currents (F ir ϭ 0.14, F ir ϭ 0.20; both with co-expression of G␤␥) were determined in the presence or absence of 3 mM external Cs ϩ and fitted according to the Woodhull model (33, 34) (Fig. 7E).
The half-blocking voltage E Block 1 ⁄2 was shifted in the negative direction by up to Ϫ30 mV by G␤␥ (E Block 1 ⁄2 ϭ Ϫ64 mV for F ir ϭ 0.10; E Block 1 ⁄2 ϭ Ϫ79 mV for F ir ϭ 0.14; E Block 1 ⁄2 ϭ Ϫ93 mV for F ir ϭ 0.20). Interestingly, the apparent voltage dependence of the Cs ϩ -induced current block as indicated by the electrical distance ␦ was up to 3-fold steeper under weak inwardly rectifying conditions (␦ ϭ 2.2 for F ir ϭ 0.10; ␦ ϭ 3.0 for F ir ϭ 0.14; ␦ ϭ 6.9 for F ir ϭ 0.20), suggesting a deeper penetration of Cs ϩ into the pore or a change in voltage-dependent binding parameters for Cs ϩ within the pore. This result strongly suggests that G␤␥ induced a significant conformational change within the GIRK channel pore.

FIG. 5. The slow component of outward current block is not altered by co-expression of G␤␥ subunits.
Illustrated are representative current recordings measured in response to voltage steps (Ϫ120 mV, 60 mV, and Ϫ120 mV, as indicated) from cells, which did (red) or did not express exogenous G␤␥ (black). Currents were normalized either to the maximal inward (upper and lower right panel) or outward (lower left panel) currents, and the time course of the onset of outward current block (lower left panel) as well as the recovery from outward current block (lower right panel) was fitted best by a biexponential decay. Summarized data for the resulting time constants are illustrated in the figure (n ϭ 5-7).
FIG. 6. Concentration-response curve for Ba 2؉ -induced GIRK current inhibition. The inhibition of steady GIRK currents (holding potential: Ϫ90 mV) in response to extracellular Ba 2ϩ was determined in CHO-K1 cells transfected with the same set of cDNAs as described in the legend to Fig. 2. Curve fitting using conventional dose-response equations (Origin 6.1 software) determined the concentration for halfmaximal inhibition of GIRK currents to be 30 M Ba 2ϩ for agonistactivated (strong inward rectifying) currents and 70 M for G␤ 1 ␥ 2activated currents. Hill slopes were n ϭ 1.1 (agonist) and 2.0 (ϩG␤ 1 ␥ 2 ), respectively (n ϭ 3-5).

Weakening of Inward Rectification of I KAch in Feline Atrial
Myocytes Is Due to Binding of G␤␥ to GIRK Channels-This study discovered that G-proteins do not only activate GIRK channels, but in addition also regulate the degree of inward rectification of these channels. In isolated atrial myocytes from adult cats, submaximally activated I KACh exhibited strong inward rectification, whereas maximal stimulation resulted in a 2-3-fold weakening of the inward rectification of I KACh . The fact that GIRK channels heterologously expressed in cell lines devoid of any other measurable inward rectifying currents were modulated in their inward rectifying properties by coexpression of G␤␥ subunits, strongly suggested that the inward rectification of GIRK channels themselves can be modulated by G␤␥ and that the observed stimulus-dependent weakening of I KACh inward rectification was the result of a G␤␥-mediated modulation of GIRK channels.
The maximal activation of GIRK channels expressed in HEK 293 or CHO cells was clearly limited by the availability of endogenous G␤␥ subunits as the co-expression of G␤␥ subunits boosted GIRK currents 2.5-4-fold (Figs. 2 and 4 and Refs. 21 and 35). This may explain why in the absence of G␤␥ coexpression no agonist-mediated modulation of GIRK inward rectification was observed (unless GIRK channel expression was very low). In cells exhibiting strong inwardly rectifying GIRK currents despite expression of exogenous G␤␥, additional stimulation via G␣ i -coupled receptors led to a dramatic weakening of the inward rectification of these currents (Fig. 3). This observation suggested that modulation of inward rectification was not an artifact of G␤␥ over-expression. At a constant expression level of G␤␥ subunits an increase of GIRK expression strengthened inward rectification and increased inward current density (Fig. 4), suggesting that the ratio of available G␤␥ subunits to expressed GIRK channels is critical for regulating inward rectification. Taken together these results point to a bimodal regulation of GIRK channels by G␤␥ subunits: at submaximal concentrations G␤␥ increased the open probability of GIRK channels (as demonstrated before (15,36)), whereas at saturating concentrations G␤␥ weakened inward rectification of GIRK channels giving rise to a substantial increase in outward K ϩ current conductance.
The Physiological Role of Weakened Inward Rectification-The inward rectification of I KACh channels is important for their physiological function to stabilize the membrane potential at negative voltages but not for blocking the generation of the plateau phase of action potentials (16). Because under physiological conditions net-potassium flux through this channel will always be in outward direction, one would predict that 2-3-fold increases in potassium outward currents, due to weakening of the inward rectification as observed in this study, will have a great impact on the shape and duration of supraventricular action potentials. It seems likely that the local in vivo concentration of ACh in the synaptic cleft can reach levels high enough, at least for very short periods, to cause weakening of inward rectification of atrial I KACh , because high frequency stimulation of the vagal nerves can induce a hyperpolarization in atrial tissue similar in amplitude as if directly evoked by ACh in the low M range (37).
The Weakening of Inward Rectification Is Not Due to a Reduced Polyamine Affinity-It has been shown that open channel block by polyamines and Mg 2ϩ ions contributes to inward rectification in GIRK channels. Therefore, changing the inward rectification in the observed way may be related to polyamine and/or Mg 2ϩbinding properties to the channel. Mg 2ϩ is known to block instantaneously, whereas polyamine block exhibits slow voltage-dependent blocking and unblocking kinetics (13,27,28). In whole cell patch clamp experiments the polyamine block is found to be responsible for the slow inactivation/activation of GIRK currents measured resulting from voltage steps (28). Our investigation of polyamine block revealed no striking alteration of the blocking/ unblocking time constants in the presence of G␤␥ over-expression. If a decrease of the polyamine affinity had been the cause for weakened inward rectification, a major increase in the blocking time constants and/or a major decrease in the unblocking time constant should have been observed. However, we found the contrary. Under weak inwardly rectifying conditions blocking time constants were slightly decreased and no major differences in unblocking time constants was observed. The observed weakening of inward rectification could be attributed to a decrease of the fraction of channels that were blocked instantaneously in outward direction (Fig. 3), pointing to attenuation of either the Mg 2ϩ -induced channel blockade or some yet unknown intrinsic outward current block (14). However, no change in inward rectification was observed when increasing internal Mg 2ϩ up to 20 FIG. 7. G␤␥ over-expression attenuates Cs ؉ -induced inward current block of GIRK channels. Current-voltage relationships of GIRK currents were recorded in the presence or absence of 3 mM Cs ϩ in the bath solution. GIRK currents were evoked either via ␣ 2A adrenergic receptors and endogenous G-proteins (A and C, blue-colored bar) or by co-expression of G␤␥ (B and C, red-colored bar). The potency of 3 mM Cs ϩ to block GIRK currents at Ϫ90 mV were plotted against the degree of inward rectification in cells expressing or not expressing exogenous G␤␥ (D). The voltage dependence of the current block induced by 3 mM extracellular Cs ϩ was determined in dependence of the degree of inward rectification (E). Representative experiments have been selected for strong (F ir ϭ 0.1, control; no G␤␥ transfected), medium, and weak (F ir ϭ 0.14, F ir ϭ 0.20; both with co-transfection of G␤␥ subunits) inwardly rectifying currents and were fitted according to the Woodhull model (Equation 1 under "Results"). mM to compensate for a possibly reduced potency. So far, there is no direct experimental evidence to attribute the weakening of inward rectification to altered binding properties of polyamines or Mg 2ϩ to the channels. However, we cannot exclude that Mg 2ϩinduced outward current block was completely impaired in weak inwardly rectifying GIRK channels.
The Weakening of Inward Rectification Is Associated with a Reduction of Ba 2ϩ and Cs ϩ Affinity-In Kir2.1 channels there exists an overlap between sites important for inward rectification and blocking by external cations such as Cs ϩ and Ba 2ϩ (11,38). Therefore, G␤␥-induced reduction of the affinity of Ba 2ϩ to block GIRK currents supports the assumption that weakening of inward rectification is induced by conformational changes in the pore region of GIRK channels. Extracellular Cs ϩ is known to block strongly inwardly rectifying K ϩ channels in a highly voltagedependent manner. The Cs ϩ -binding site is also located within the channel pore, probably deeper in the channel than the blocking site for Ba 2ϩ . Similarly to Ba 2ϩ -induced GIRK channel block, Cs ϩ -induced block was attenuated under weak inward rectifying conditions and the weakening of inward rectification correlated to weakening of Cs ϩ -induced current block. G␤␥induced weakening of inward rectification was correlated as well with a stronger voltage dependence of Cs ϩ block and a shift to more negative potentials. These G␤␥-mediated changes in the pore blocking properties of GIRK channels compare well to the differences of the pore blocking properties between members of the week and strong inwardly rectifying K ϩ channel family. Weak inwardly rectifying K ϩ channels exhibit usually a weaker affinity for Cs ϩ and Ba 2ϩ compared with strong inwardly rectifying K ϩ channels (11,28,39). Based on these results we propose that maximal activation of GIRK channels by G␤␥ subunits induces a conformational change within the channel pore that tunes GIRK channels from a strong to a weak inwardly rectifying channel.
G␤␥ Induces a Conformational Change in the Pore of GIRK Channels-How does binding of G␤␥ to the channel induce a conformational change of the channel that leads to reduced affinity for Cs ϩ and Ba 2ϩ ? The fact that G␤␥ binds near the intracellular C terminus makes it unlikely that the cations and G␤␥ subunits share common binding sites within the pore. Considering that a tetrameric channel can bind up to four G␤␥ subunits (9) and the open probability is gradually regulated by at least three G␤␥-binding sites (40), we propose that binding of the third or more likely the fourth G␤␥ subunit to the channel may force the channel into a weak inward rectifying conformation. How could this work? Recently L.Y. Jan and co-workers (41) presented convincing data, which suggested that opening of GIRK channels requires a rotation of the M2 transmembrane helix. Because the residues important for cation pore block are located either on the M2 helix or are in close proximity to the M2 helix, a rotation of these helices may likely alter the position of these residues. Assuming that the model of Jan and co-workers is correct and binding of a G␤␥ subunit to a GIRK channel subunit causes a rotation of the M2 helix of this particular GIRK channel subunit, it is obvious that the structures close to the cation-binding site(s) within the channel pore of a tetrameric GIRK channel will be different depending on how many G␤␥ subunits are bound. According to Refs. 36 and 42, single channel characteristics in respect to open and closed times GIRK channels are different depending on the concentration of available G␤␥ subunits. If strong inward rectification and high affinity Ba 2ϩ and Cs ϩ block require one or two of the four M2 helices not to be rotated, rotation of the last two helices (induced by binding of the 3 rd or 4 th G␤␥ subunit to the tetrameric channel) could potentially weaken inward rectification. This working hypothesis needs to be verified in future studies.
The G-protein-mediated regulation of inward rectification of atrial and heterologously expressed GIRK channels described in this study represents to our knowledge the first description of a regulatory mechanism that alters the inward rectifying properties of an ion channel. Furthermore, we demonstrate that binding of G␤␥ subunits to the channel alter the conformation at known cation-binding sites within the channel pore, supporting the hypothesis that G␤␥ might gate the channel at the selectivity filter rather than at a cytoplasmic gate.