Redox-dependent Gating of G Protein-coupled Inwardly Rectifying K+ Channels*

G protein-coupled inwardly rectifying K+ channels (GIRK) play a major role in inhibitory signaling in excitable and endocrine tissues. The gating mechanism of these channels is mediated by a direct interaction of the Gβγ subunits of G protein, which are released upon inhibitory neurotransmitter receptor activation. This gating mechanism is further manifested by intracellular factors such as anionic phospholipids and Na+ and Mg2+ ions. In addition to the essential role of these components for channel function, phosphorylation events can also modulate channel activity. In this study we explored the involvement of redox modulation on GIRK channel function. Extracellular application of the reducing agent dithiothreitol (DTT), but not reduced glutathione, activated GIRK channels without affecting their permeation or rectification properties. The DTT-dependent activation was found to mimic receptor activation and to act directly on the channel in a membrane delimited fashion. A critical cysteine residue located in the N-terminal cytoplasmic domain was found to be essential for DTT-dependent activation in hetero- and homotetrameric contexts. Interestingly, when mutating this cysteine residue, DTT-dependent activation was abolished, but receptor-mediated channel activation was not affected. These results suggest that intracellular redox potential can play a major role in tuning GIRK channel activity in a receptor-independent manner. This sort of redox modulation can be part of an important cellular protective mechanism against ischemic or hypoxic insults.

pears to drive a rearrangement of the pore-forming second transmembrane ␣-helix (TM2) to open the channel (13,14). The G␤␥-mediated gating of the channel can be aided by other intracellular components. Sodium and Mg 2ϩ ions have been found to interact with the C-terminal cytoplasmic domain of the channel to fine-tune channel gating mediated by the anionic phospholipid, phosphatidylinositol-4,5-bisphosphate (15)(16)(17)(18)(19), which is an obligatory component for the stability of the open state of the channel (14). The Na ϩ -phosphatidylinositol-4,5-bisphosphate interaction can also serve as a mechanism for modulating by external ligands (20). In addition to the modulatory action of the various ionic species mentioned above, channel function can also be tuned by phosphorylation/dephosphorylation events. Activation of TrkB receptors by neurotrophin brain-derived neurotrophic factor affects channel function via tyrosine phosphorylation (21), and activation of cAMP formation via ␤2-adrenergic receptors facilitates channel function (22). These phosphorylation events may act by affecting the interaction of the G␤␥ subunits with the channel by occluding the G␤␥ binding site (23). Additionally, inhibitory signaling can occur via protein kinase C modulation by an as yet unknown mechanism (24 -26).
Several studies have shown that redox signaling can serve as an additional mechanism to modulate ion channel activity to link the metabolic state of the cell to its electrical properties (for review see Refs. 27 and 28). Redox signaling has been found to affect the activity of several ion channels, such as the N-methyl-D-aspartate receptor, NR1 (29), the Ca 2ϩ -activated K ϩ channel, hslo (30), ryanodine receptor (31), a voltage gated K ϩ channel, Kv1.4 (32), and the inwardly rectifying K ϩ channel, IRK1 (33). In this study we examined the role of redox potential in mediating GIRK channel activity expressed in Xenopus oocytes. Exposure of GIRK channels to the membranepermeable redox agent, dithiothreitol (DTT), but not impermeable glutathione (GSH), increased channel activity in a reversible manner when monitored under intact whole cell conditions. This effect of DTT was specifically attributed to an N-terminal cytosolic cysteine residue, as mutating this residue abolished DTT action without affecting receptor-mediated channel activation. In this study, we suggest that an increase of GIRK channel activity by redox signaling can serve as a protective cellular mechanism under hypoxic or ischemic insults.
Site-directed Mutagenesis-Site-directed mutagenesis was done based on polymerase chain reaction of full-length plasmid using the * This work was supported by the Israeli Science Foundation, the Minerva Foundation (Germany), the Human Frontier Science Program, and the Buddy Taub Foundation. 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.
high fidelity Pfu polymerase (method by Stratagene). Positive clones were verified by sequencing.
Electrophysiology-Currents through the expressed channels were recorded using the two-electrode voltage clamp technique as previously described (14). Oocytes were held at 0 mV unless otherwise indicated, and voltage ramps from Ϫ100 mV to ϩ50 mV for 0.6 s were applied. The oocytes were washed for at least 2 min with 90K solution containing 90 mM KCl, 10 mM HEPES, 2 mM MgCl 2 , pH 7.4 (KOH) before data collection. In all experiments presented, oocytes expressing less than 10 A of current at Ϫ100 mV were used.
Statistical and Data Analysis-Data are presented as fold induction means Ϯ S.E., and n denotes the number of oocytes assayed. A t test was used to calculate the statistical significance of differences between different populations.

RESULTS
We were interested in testing the effect of the reducing agent, DTT, on GIRK1/4 channel function. Two electrode voltage clamp recordings of Xenopus oocytes expressing GIRK1 and GIRK4 display inwardly rectifying currents upon application of voltage ramp protocols from Ϫ100 to ϩ50 mV. Upon the application of 10 mM DTT to the extracellular solution in the absence of expressed G protein coupled receptors and agonists of native receptors, a gradual increase in basal inwardly rectifying current is apparent that reaches a maximum steady state induction within 10 min of application and reverses upon DTT washout to base-line levels (Fig. 1, A and B). The average induction of GIRK1/4 channel currents was 2.62 Ϯ 0.20-fold (n ϭ 12) (Fig.  1C). The DTT-induced currents did not change the rectification characteristics monitored at ϩ50 mV (Fig. 1, A and B) and did not change ion selectivity (K ϩ versus Na ϩ , data not shown). Thus, these results suggest that the selective increase in membrane current by DTT is mainly due to an increase in activity of GIRK1/4 channels. In control experiments, DTT had no effect on uninjected oocyte basal current levels. To assess the dose dependence of DTT action on GIRK1/4 channel currents, we measured current levels in response to a series of DTT solutions of increasing concentrations. We found that the DTT concentration that induced 50% of the maximal increase in basal GIRK1/4 currents was 1.93 Ϯ 0.89 mM, with a Hill coefficient of 0.84 Ϯ 0.32 (Fig. 1D). To test whether DTT action is intracellular, we also tested the effect of the non-permeant reducing agent GSH on channel currents. GSH, at 10 mM, did not induce an increase in current (0.68 Ϯ 0.05-fold (n ϭ 6) induction), suggesting that the action site of DTT is intracellular (Fig. 1C).
We also wanted to test whether the effect seen with DTT on GIRK1/4 channel currents was specific for the GIRK family. We therefore tested the effect of DTT using the same protocols as above on the G protein-independent inwardly rectifying K ϩ channel, IRK1. DTT had no stimulatory effect on the IRK1 currents but, rather, induced a slight inhibition 0.79 Ϯ 0.17fold (n ϭ 6) (Fig. 1D). The decrease in IRK1 currents by DTT has been seen before by Ruppersberg et al. (33). We then tested whether the DTT effect depends on soluble components endogenously present in the oocyte cytosol.
To further verify that the lack of GSH to induce channel activity was due to its inability to cross the plasma membrane, we performed single channel recordings on excised inside-out patches and tested the effect of 1 mM DTT or 10 mM GSH applied to the cytosolic face of the patch. Extensive washes of the inner face of the patches did not abolish the DTT or the GSH effect. Both DTT and GSH increased the probability of channel openings without affecting the single channel ampli-FIG. 1. DTT activates GIRK channels expressed in Xenopus oocytes in a membrane-delimited manner. A, time course of GIRK current induction by 10 mM DTT applied to the extracellular solution for 10 min and the return to basal current levels after washout. Current levels at ϩ50 and Ϫ50 mV were taken from ramps (Ϫ100 to ϩ50 mV for 1 s) applied every 2 s from a holding potential of 0 mV. B, representative ramps before (1), during (2), and after (3)  tude measured at Ϫ100 mV (Fig. 2, A and B). Interestingly, extensive washout (for at least 5 min) was not able to reverse either the DTT or the GSH effects, and the channels remained as active as in the presence of the corresponding reducing agent. Thus, these results suggest that the action of DTT to induce channel activation is membrane-delimited, and the reversibility of this effect (as seen under whole cell measurements, Fig. 1A) may depend on the cytoplasmic factor(s) that is absent in the excised patch.
In native tissue, GIRK1/4 channels are activated by neurotransmitter receptor stimulation, and so we were interested in examining the nature of DTT channel induction in relation to activation by the type 2 muscarinic receptor (m2R). Oocytes were co-injected with GIRK1/4 channel and m2R, and the effect of DTT was tested in the presence and absence of receptor stimulation using 3 M carbachol. Receptor-mediated channel activity was measured after 30 s of application of carbachol in the presence and in the absence of 10 mM DTT (Fig. 3A). Carbachol application alone increased the GIRK1/4 currents by 6.6 Ϯ 0.7-fold (n ϭ 8), and carbachol application in addition to DTT increased the current by 6.7 Ϯ 0.5-fold (Fig. 3B). This indicates that DTT induction of the GIRK1/4 channel is not additive to receptor stimulation and, thus, may act on the same activation pathway or on a converged one. To further explore this possibility, we tested whether DTT is able to induce channel currents once channels are fully activated, independent of receptor stimulation. One way to fully activate the channels is by the co-expression of the channel with the G␤ 1 ␥ 2 subunits of the G protein (4). Application of 10 mM DTT to oocytes co-expressing GIRK1/4 channels and G␤ 1 ␥ 2 induced an increase in currents with 1.34 Ϯ 0.11-fold (n ϭ 5) induction (Fig. 4A), but as with receptor stimulation, this effect was occluded in that it was considerably smaller that the 2.6-fold induction seen in the absence of G␤␥. An alternate way to test the effect of DTT on fully activated GIRK1/4 channels is to use constitutively active, G␤␥-independent channel mutants (14). These mutations, localized to the TM2 domain of the channel, lock the channel in its activated conformation. We tested two of these mutants, GIRK1(C179A)/GIRK4(C185A) and GIRK1(S170P)/ GIRK4(S176P). Once again, DTT had a considerably attenu-ated induction of channel currents, with 1.24 Ϯ 0.06 (n ϭ 5) and 1.19 Ϯ 0.07-fold (n ϭ 8) induction for the two mutants, respectively (Fig. 4A). The reduced ability of DTT to induce channel activity in activated channels suggests that the action of DTT may be to mimic G␤␥-dependent activation.
Two possibilities are likely to explain the above results. The action of DTT could be the consequence of direct reduction of the channel, and the second possibility involves DTT activation of either the G protein trimers to release G␤␥ or of another second messenger system. To address the latter possibility, we tested the ability of DTT to induce channel currents under three different conditions that reduce the levels of free G␤␥ in the oocyte. The first way was by co-expressing the catalytic subunits of pertussis toxin (PTX-S1). PTX-S1 induces ADPmediated ribosylation of the G␣ i/o subunit of the G protein and prevents the release of G␤␥. The second way to reduce free G␤␥ levels in the oocyte was by expressing the C-terminal end of the ␤-adrenergic receptor kinase (C-␤ARK), which has been shown to bind the G␤␥ subunits of G proteins (38) and, thus, to reduce channel activity (4). The third way to reduce G␤␥ levels was by the co-expression of G␣ s , with the idea that an excess of G␣ s will act as a large sink for free G␤␥ subunits (39). As expected, under the first two conditions, m2R activation was impaired (Fig. 4C). However, the DTT induction was not significantly affected (Fig. 4B). DTT induction for co-expression with PTX-S1, C-␤ARK (C-terminal end of the ␤-adrenergic receptor kinase), or G␣ s was 2.2 Ϯ 0.2, 1.9 Ϯ 0.3, and 2.9 Ϯ 0.8-fold, respectively. These results suggest that DTT activates GIRK1/4 channels without a direct activation of G proteins.
As discussed above, protein kinases may regulate GIRK channel activity. To rule out any involvement of protein kinase A or protein kinase C in DTT action, we used the potent nonspecific protein kinase inhibitor 1-(5-isoquinolinesulfonyl)-2-methylpiperazine dihydrochloride (H7). Oocytes expressing GIRK1/4 channels were incubated in 100 M H7 for at least 20 min before recordings, and the action of DTT was then tested. Incubation with H7 failed to affect DTT stimulatory action, with DTT induction of 2 Ϯ 0.3-fold. These results suggest that the action of DTT on the channel is not mediated through the activation of protein kinase A or protein kinase C in the oocytes.
Recently, it has become evident that many proteins undergo specific post-translational modification on cysteine residues, mainly by the use of nitric oxide donors, to form nitrosylated cysteines (for review, see ref. 40). Thus, another possible target for DTT action is the breakdown of a nitrosylated cysteine that participates in channel regulation. We explored this possibility using two independent approaches to modify nitric oxide levels; those approaches were either by using a nitrosothiol derivative that releases nitric oxide, S-nitroso-N-acetylpenicillamine, or by direct inhibition of nitric oxide synthase using N G -nitro-Larginine methyl ester. These compounds had no affect on channel function, receptor activation, or basal channel activity (not shown). In addition, it is also worth noting that the critical negative residue essential for the acid-base catalysis of sulfahydryl nitrosylation, which should flank potential nitrosylated cysteines, is not present around cysteines in either GIRK1 or GIRK4 (41). These results thus suggest that GIRK channels are not regulated by nitrosylation, and the effect of DTT is probably not mediated through this pathway.
The above results suggest that DTT may mediate channel induction via direct reduction of cytoplasmic cysteines of the channel molecule. GIRK1 and GIRK4 contain seven and eight cysteine residues, respectively, including two extracellular conserved residues at positions 123 and 156 in GIRK1 and at positions 129 and 162 in GIRK4. These cysteines are fully conserved in all inwardly rectifying K ϩ channels and have been shown to be important for the assembly of IRK-like channels (42)(43)(44). There is an additional cysteine located in the TM2 domain that is conserved in all members of the GIRK and K-ATP channel families. This cysteine has been shown to be involved in G␤␥-mediated gating conformation (14). Alignment of the Kir family members reveals another conserved residue at the intracellular N terminus of the channel, at positions 53 and 60 in GIRK1 and GIRK4, respectively. The remaining intracellular cysteines are located in the C-terminal cytosolic domain of the channel (Fig. 5A). All six GIRK4 intracellular cysteines were replaced, and the channel mutants were co-expressed with wild type GIRK1 channel. We found that none of the single cysteine substitutions affected the ability of DTT to induce channel currents (Fig. 5B). The following GIRK4 mutants co-expressed with GIRK1 were induced by DTT by: C60A, 1.98 Ϯ 0.20 (n ϭ 5)-fold; C216A, 3.32 Ϯ 0.66 (n ϭ 5)-fold; C316T, 1.85 Ϯ 0.18 (n ϭ 5)-fold; C362A ϩ C363S, 1.84 Ϯ 0.11 (n ϭ 5)-fold; C389A, 1.86 Ϯ .07 (n ϭ 5)-fold (Fig. 5C).
Since removing individual cysteines from GIRK4 in the heteromultimeric context was not sufficient to affect DDT action, we wanted to test whether removing all intracellular cysteines in a homomultimeric context would have an effect. To simplify the possible mutant combinations we used a GIRK4 mutant, GIRK4(S143T), that has been shown to form functional homotetramers (36) and tested the ability of DTT to induce its currents. Fig. 6A represents a time course of DTT induction of the

FIG. 4. DTT is ineffective on fully activated channel and does not directly activate G proteins.
A, bar graph summarizing the effect of 10 mM DTT application on fully activated GIRK1/4 channels by either co-expressing G␤ 1 ␥ 2 subunits or on constitutively active GIRK1(C179A)/GIRK4(C185A) and GIRK1(S170P)/GIRK4(S176P) mutant channels. B, bar graph summarizing the full effectiveness of 10 mM DTT in activating GIRK1/4 channels co-expressed with the G␤␥ binding domain of ␤-adrenergic receptor kinase 1 (c␤ARK), the catalytic subunit of pertussis toxin (PTX-S1), or G␣ s subunit of the G protein. C, bar graph summarizing the inability of 3 M carbachol to activate GIRK1/4 channels co-expressed with either c␤ARK (C-terminus of ␤-adrenergic receptor kinase) or PTX-S1, demonstrating that either trapping G␤␥ or preventing it from dissociating from the G protein trimer is sufficient to block m2R stimulation without affecting DTT induction (B). GIRK4(S143T) channel. DTT was able to induce GIRK4(S143T) channel currents in a similar manner to the GIRK1/4 heteromultimer with 2.3 Ϯ 0.3-fold induction. We then removed all intracellular cysteines in the background of the S143T mutation and tested the effect of DTT (Fig. 6B). DTT failed to affect the homomeric GIRK4(S143T), lacking all six intracellular cysteines (C60A, C216A, C316T, C362A, C363S, and C389A), with 0.8 Ϯ 0.23-fold induction (Fig. 6D), without affecting m2R activated currents (Fig. 6C). This first suggests that the action of DTT on induction of channel currents is probably mediated through the channel intracellular cysteines and, second, that the mechanism by which DTT induces activity and receptor-mediated activation are via two separate pathways.
We set out to identify the cysteine residue responsible for DTT induction in the homomeric GIRK4(S143T) context. We therefore mutated all intracellular cysteines in groups according to their location in the linear sequence of the channel. All channel mutants that contained the C60A mutation failed to respond to DTT (Fig. 6D). Thus, these results suggest that the cysteine residue at position C60 is involved in channel activation by DTT. Since in native tissues most of the GIRK channel combinations form heteromultimers, we also wanted to test whether mutating the analogous cysteines at position Cys-53 and Cys-60 in GIRK1 and GIRK4, respectively, would affect DTT induction. GIRK1(C53A)/GIRK4(C60A) mutant channels failed to respond to DTT, as seen in the homomultimeric context (Fig. 7A), without affecting the ability of carbachol to induce currents (Fig. 7B). These results again suggest that the N-terminal cysteines at positions Cys-53 and Cys-60 in GIRK1 and GIRK4, respectively, play a critical role in DTT action without affecting receptor-mediated activation. DISCUSSION In this paper, we report a novel mechanism to modulate GIRK channel gating; that is, reduction of a conserved cysteine residue in the N-terminal cytosolic domain. This form of gating may operate in parallel to the action of the classical activators of the channel, the G␤␥ subunits of the G protein, phospholipids and Na ϩ ions. From the experiments presented above we suggest that reduction by DTT mimics gating of the channel by G␤␥ subunits. Several lines of evidence contribute to this conclusion. First, GIRK currents are activated by DTT without a change in rectification and selectivity. Second, the effect of DTT is reduced with receptor-mediated GIRK channel activation and with the constitutive activation caused by overexpressing G␤ 1 ␥ 2 . Third, channels that are constitutively active and G␤␥-FIG. 5. DTT does not require all cytosolic cysteines in a heterotetrameric context for its action. A, a schematic depicting the extracellular and transmembrane (light gray) and presumed cytosolic (black) cysteine residues of GIRK4 channels. B, time course of DTT action on oocytes co-expressing various GIRK4 cysteine mutants with GIRK1 wild-type. The dashed line represents a zero current level; above and below are the measured current levels at ϩ50 mV and Ϫ50 mV, respectively. Scale bars are 1 A and 60 s. C, bar graph summarizing the DTT induction of the various GIRK4 cysteine mutants co-expressed with GIRK1 wild-type channel. There is no statistical difference between the induction of the GIRK1/GIRK4 and the GIRK1/GIRK4 cysteine mutants.
FIG. 6. DTT induction of the GIRK4(S143T) homomeric mutant and the lack of induction once the cytoplasmic cysteines are removed. A, time course of 3 M carbachol (Carb.) and of 10 mM DTT induction of the GIRK4(S143T) homomultimer. Current measurements were taken as described in Fig. 1. B, time course of 3 M carbachol and 10 mM DTT induction of GIRK4(S143T) channel lacking all intracellular cysteines (C60A, C216A, C316T, C362A, C363S, and C389A). Note the intact ability of carbachol to gate the mutated channel and the impaired DTT induction. C, bar graph summarizing the extent of channel induction by carbachol of GIRK4(S143T) and of the mutant lacking all intracellular cysteines. D, bar graph summarizing fold induction by 10 mM DTT of GIRK4(S143T) and its various cysteine mutants (as indicated). Note the lack of effect of DTT when the N-terminal C60A mutation is present. independent (14) abolish the DTT-mediated induction. These results may suggest that the cysteine reduction mechanism involves a similar conformation of the gating apparatus as induced by receptor stimulation or the G␤␥ subunits but, interestingly, without affecting channel rectification.
The failure of DTT to affect channel rectification is the most apparent difference in channel biophysical properties between activation by receptor stimulation and DTT. In the former case, the extent of channel rectification during activation is reduced, and hence, more current is apparent at potentials above the equilibrium for K ϩ ions (for example, see Fig. 3A traces at ϩ50 mV). In the latter situation, DTT activation gating does not affect channel rectification. One plausible explanation for this difference is that DTT opens GIRK1/4 channels to a state where the gate is open but does not undergo an additional transition normally induced by a full receptor-dependent activation. This full transition gating state induced by receptor stimulation of the channel may suggest that G␤␥ interaction, unlike DTT induction, affects channel conformation by reducing the binding of either Mg 2ϩ ions or polyamines. This may be due to the ability of G␤␥ to strongly interact with the C-terminal cytoplasmic domain of the channel and induce a conformational change in the second transmembrane domain of the channel (13,14). These two regions are known to interact with the components responsible for inward rectification, Mg 2ϩ ions and polyamines (Ref. 45 and 46; for review, see also Ref. 47). The robust capacity of G␤␥ to gate the channel compared with DTT is also evident by the ϳ2-fold less efficient channel activation by DTT.
Does DTT directly act on the channel molecule, or is its action mediated via a second mediator? Such mediated action could involve DTT activation of the G protein-coupled receptor, the G protein, or of another second-messenger system. First, we can exclude the possibility that DTT acts directly on the G protein-coupled receptor (48) because 1) the effect of DTT was seen with oocytes expressing only the GIRK channel, without the m2R receptor, and 2) co-expression of the channel and the receptor with the catalytic subunit of pertussis toxin abolished receptor-mediated activation (36) but did not affect the ability of DTT to induce channel currents. Second, we can also exclude the possibility that DTT directly activates the G protein trimer to release G␤␥ subunits. This is because co-expression of the GIRK channel with either G␣ s or the C terminus of ␤ARK, both of which reduce the availability of free G␤␥ (35,39), did not affect DDT induction, even though it dramatically reduced receptor-mediated activation. This indicates that DTT-mediated channel induction does not involve a direct activation of G proteins in the oocyte. Third, we can also exclude the involvement of protein kinase activation by DTT, since the nonspecific kinase inhibitor, H7, was unable to abolish DTT induction, and fourth, we can exclude the involvement of nitrosylation in DTT action. We therefore can suggest that DTT action is a direct effect on the channel molecule to induce an activated conformation.
The proposal that the DTT effect on GIRK gating is direct is further supported by the finding that the DTT effect is eliminated by removal of all of the intracellular cysteines from the GIRK4(S143T) homomultimeric channel. By systematically eliminating individual cysteines from this homomeric channel, it becomes apparent that only the N-terminal cytosolic cysteine at position 60 is essential for DTT-mediated activation. Moreover, from experiments using co-expression of GIRK1 wild type and GIRK4(C60A) mutant channels, it is apparent that it is not obligatory to have N-terminal cysteines in all four channel subunits for DTT to gate the channel (Fig. 5C). Since experiments involving tandemly linked different GIRK channel subunits point toward a specific preferred architecture of functional channels, alternating GIRK1/GIRK4 monomers (49,50), it becomes plausible to conclude that N-terminal cysteines from neighboring subunits are not involved in DTT action. Once the N-terminal cysteines are removed from both subunits, GIRK1(C53A) and GIRK4(C60A), DTT is no longer able to activate the channel. Thus, if indeed the action of DTT is to mediate the reduction of disulfide bridges, they should exist between diagonal subunits, which will require the opposing N-terminal cytosolic domains to be within a few angstroms of each other, as the length between two thio groups forming disulfide bond is about 2 Å. Alternatively, the N-terminal cysteines can be oxidized at their sulfhydryl group (at physiological pH) to become reactive cysteines (thiolate, S Ϫ ). This reactivity can be stabilized by positively charged amino acids neighboring these cysteines, as seen in the bacterial transcription factor OxyR (51). Interestingly, in the GIRK channel family, there is a lysine side chain immediately preceding the N-terminal cysteines, which is not found in the IRK channel family. These reactive cysteines can form sulfenic ion (S-OH) under very mild oxidation that can cause them to interact with the high levels of intracellular GSH to form mixed disulfides (28). Due to the lower reducing potential of DTT compared with GSH, DTT may act to reduce this mixed disulfide to the sulfhydryl form. The equilibrium between the mixed disulfide and sulfhydryl or reactive cysteine species may determine the receptor-independent activity of the channel. The fact that DTT induction is reversible only under intact whole cell and not under excised patch configuration suggests that cytosolic factors, e.g. thioredoxin or glutaredoxin, may be involved in the reversal of DTT induction. We thus hypothesize that the cysteines in the N-terminal cytoplasmic domain of the channel can undergo an oxidation-reduction cycle to modulate channel activity.
What is the physiological relevance of redox-mediated GIRK channel activation? In the past few years it has become more appreciated that redox signaling at low concentrations of reac- FIG. 7. Impaired DTT induction of the N-terminal cysteine mutation in a heteromultimeric context. A, bar graph summarizing the inability of 10 mM DTT to activate the GIRK1(C53A)/ GIRK4(C60A) mutant. For purposes of comparison, the bar graph of the wild-type is the same as in Fig. 1. B, bar graph summarizing the ability of 3 M carbachol (Carb.) to gate the above N-terminal cysteine channel mutant. Note that carbachol induction is not affected, similar to the situation with the homomeric channel described in Fig. 6. tive oxygens may play a major role in regulating many cellular events such as cell-cell adhesion, immunological responses, and cellular excitability (for review, see Refs. 52 and 53). Under hypoxic or ischemic conditions, when the oxygen tension is reduced, activation of GIRK channels may be of great importance, specifically in reducing cellular excitability by hyperpolarization. For example, in CA1 pyramidal neurons hypoxia causes hyperpolarization, which is independent of opening of K-ATP, Ca 2ϩ -activated K ϩ or Cl Ϫ channels (54). This form of GIRK channel control by redox modulation can serve as a control to reduce excitability independent of synaptic activity. In the case of the heart it has been proposed that ischemic preconditioning, a protective mechanism in which a brief period of myocardial ischemia renders the myocardium resistant to a more sever ischemic insult (55), is mediated by a pertussis toxin-sensitive mechanism that involves acetylcholine and adenosine (56). Although most of the published observations point toward the involvement of mitochondrial K-ATP channels in this protection during preconditioning ischemic insults (57), it is interesting to note that both neurotransmitters also activate GIRK channels in atrial myocytes. This relation raises the possible involvement of GIRK channels in this protective action, specifically the anti-arrhythmic aspect of preconditioning.
In summary, the present work identifies a novel aspect of GIRK channel gating in which the redox state can modulate channel gating. An increase in redox potential opens the GIRK channel without affecting the permeation properties of the channel. We identify a single conserved residue located in the N terminus of the channel as the main mediator of this action. This redox-dependent gating is an additional gating mechanism to neurotransmitter-receptor activation, which underlies its importance as a potential protective mechanism in cases of ischemic or hypoxic shock, mainly by membrane hyperpolarization to reduce cellular excitability.