A Novel Mechanism for Human K2P2.1 Channel Gating

The mammalian K2P2.1 potassium channel (TREK-1, KCNK2) is highly expressed in excitable tissues, where it plays a key role in the cellular mechanisms of neuroprotection, anesthesia, pain perception, and depression. Here, we report that external acidification, within the physiological range, strongly inhibits the human K2P2.1 channel by inducing “C-type” closure. We have identified two histidine residues (i.e. His-87 and His-141), located in the first external loop of the channel, that govern the response of the channel to external pH. We demonstrate that these residues are within physical proximity to glutamate 84, homologous to Shaker Glu-418, KcsA Glu-51, and KCNK0 Glu-28 residues, all previously argued to stabilize the outer pore gate in the open conformation by forming hydrogen bonds with pore-adjacent residues. We thus propose a novel mechanism for pH sensing in which protonation of His-141 and His-87 generates a local positive charge that serves to draw Glu-84 away from its natural interactions, facilitating the collapse of the selectivity filter region. In accordance with this proposed mechanism, low pH modified K2P2.1 selectivity toward potassium. Moreover, the proton-mediated effect was inhibited by external potassium ions and was enhanced by a mutation (S164Y) known to accelerate C-type gating. Furthermore, proton-induced current inhibition was more pronounced at negative potentials. Thus, voltage-dependent C-type gating acceleration by protons represents a novel mechanism for K2P2.1 outward rectification.

The best studied member of the K 2P family is the mammalian mechanosensitive K 2P 2.1 (KCNK2, TREK-1) channel, expressed at high levels in excitable tissues such as the nervous system (2), heart (3), and smooth muscle (4). K 2P 2.1 channels have attracted increasing interest in recent years as their activity and biophysical properties are strongly regulated by various physical and chemical signals (5). In addition to activation by mechanical stretch, K 2P 2.1 opens in response to high temperatures (6), lysophospholipids (7), internal acidosis (8), arachidonic acid (9), volatile general anesthetics (10), and other agents. K 2P 2.1 activity is down-regulated upon phosphorylation by protein kinases A and C (2) and inhibited by various compounds such as fluoxetine (Prozac) (12), caffeine, and theophylline (13). It has been suggested that K 2P 2.1 is an important target for volatile anesthetics and participates in protection against epilepsy and neuroprotection during brain and spinal chord ischemia (14). Recently, it has been shown that K 2P 2.1 channels expressed in small dorsal root ganglion neurons (6) can act as one of the molecular sensors involved in pain perception (15).
Although the regulation and contribution of the carboxylterminal domain to the modulation of K 2P 2.1 activity have been studied in detail, the conformational changes and gating events that take place at the ion-conducting pore during channel regulation remained unknown. It had been previously shown that gating of another member of the K 2P family, i.e. the Drosophila KCNK0, entails protein rearrangement of the externally oriented portion of the pore and shares many characteristics with the slow inactivation (C-type inactivation) of voltage-gated potassium channels (16). Now, we provide evidence that K 2P 2.1 channels rely on this mechanism for gating as well.
We report that external acidification, within the physiological range, strongly inhibits the human K 2P 2.1 channel. Mutations in two outer pore surface-exposed histidine residues (i.e. His-87 and His-141), located in the first turret loop of the channel, dramatically decreased channel responsiveness to external acidification. Our results also indicate that external protons inhibit the channel by inducing closure of the outer pore gate, similar to the C-type inactivation of voltage-gated potassium channels. Moreover, we propose a mechanistic model for proton-induced pore destabilization of potassium leak channels. Finally, as the magnitude of the pH effect was voltagedependent, we present a novel mechanism for K 2P 2.1 outward rectification.

EXPERIMENTAL PROCEDURES
Molecular Biology-The human K 2P 2.1 channel (NM_ 001017425) was cloned into the pRAT plasmid as described previously (17). The murine K 2P 2.1 channel, cloned into the pcDNA3.1 plasmid, was generously provided by Prof. K. Sanders from the University of Nevada. Mutants were generated using the QuikChange site-directed mutagenesis technique (Stratagene, La Jolla, CA). The mutations were confirmed by DNA sequencing. cRNA was transcribed in vitro using T7 polymerase and the AmpliCap High Yield Message Maker (EPI-CENTRE Biotechnologies) kit.
Electrophysiology-Xenopus laevis oocytes were isolated and injected with 23 nl containing 0.2-8 ng of cRNA. Whole-cell currents were measured 1-3 days after injection by the twoelectrode voltage clamp technique, using a GeneClamp 500B amplifier (Axon Instruments, Union City, CA). Data were sampled at 2 kHz and filtered at 0.5 kHz with Clampex 9.0 software (Axon Instruments).
For two-electrode voltage clamp experiments, the pipette contained 3 M KCl and the bath solution contained, unless otherwise noted (in mM): 4 KCl, 96 NaCl, 1 MgCl 2 , 0.3 CaCl 2 , 5 HEPES, pH 7.4, with NaOH (standard solution). HEPES buffer was replaced by Tris in alkali solutions (pH 8.5, 9.0) and MES 2 in acidic solutions (pH 5.5-6.5). When required, bath solution sodium ions were isotonically replaced by potassium ions. To avoid possible effects of pH on channel regulation (18), the following protocol was applied: The currents were first stabilized at pH 9.0 at Ϫ80 mV and then pulsed from Ϫ140 to ϩ40 mV in 20-mV voltage steps for 60 ms with 1-s interpulse intervals. Next, the external solution pH was altered to the desired value for a period of 15 s, the same measurement protocol was applied, and the pH was returned to pH 9.0. This procedure was repeated for each pH value. The pH dependence was calculated from steady state currents measured at ϩ40 mV at a specific pH value and compared with the measured current observed at pH 9.0 from the same episode.
Data from patch clamp studies were sampled with an Axon 200B amplifier (Axon Instruments). Bath and pipette solutions contained 100 mM KCl, 1 mM EGTA, and 5 mM Tris-HCl, pH 9.0, or 5 mM HEPES, pH 7.5, or 5 mM MES, pH 6.5. Experiments were performed at room temperature. Data were sampled at 5 kHz and filtered at 1 kHz with Clampex 9.0 software. Chemicals were purchased from Sigma.
Calculations-Dose-response curves were fitted to the Hill equation, according to the following parameters: . The fitting was separately performed for each experiment using the Origin 6.1 software (Northampton, MA). Values are presented as mean Ϯ S.E. The voltage dependence of the proton-mediated blockage was modeled by simplification of the approach of Woodhull (19), as described previously (20).

RESULTS
Human K 2P 2.1 Channels Sense External pH-Recent work has suggested the existence of unspecified, pH-sensitive, leak-like K ϩ -selective conductance in different neuronal tissues, such as dorsal root ganglion neurons and the retrotrapezoid nucleus (21,22). We thus examined the sensitivity of K 2P 2.1 leak channels to external pH (pH O ), given the wide distribution of this channel in the central nervous system. We found human K 2P 2.1 channels to be highly sensitive to pH O (at values within the physiological range) when expressed and studied in X. laevis oocytes. Under physiological potassium concentrations (4 mM), lowering the external pH from 9.0 to 5.5 decreased current levels by Ͼ70% (Fig. 1). For steady state currents recorded at ϩ40 mV, inhibition by acidic pH values was characterized by an apparent pK of 7.47 Ϯ 0.2 (Fig. 1C). A calculated Hill coefficient of 0.51 Ϯ 0.07 implies the existence of multiple proton-binding sites in the K 2P 2.1 channels, acting with negative cooperative interactions. Single channel analysis shows that at low pH O values, channel open probability, but not single channel conductance, is decreased (Fig. 1D), suggesting an effect on channel gating.
Identifying the pH Sensor of K 2P 2.1 Channel-To date, three mechanisms have been proposed as being responsible for the pH O sensitivity of K 2P channels. In members of the K 2P 3.1 (TASK1) group, current decreases are due to protonation of a histidine residue, located next to the first pore domain selectivity filter signature (i.e. GYGH) (23,24). In K 2P 5.1 (TASK2) channels, several extracellular charged residues have been implicated as carrying the channel response to alkalization (25). This was later disputed and an alternative mechanism in which a single arginine residue, located near the second pore domain (Arg-224), was proposed as being responsible for the pH-dependent activity of K 2P 5.1 as well as TALK (K 2P 16.1 and K 2P 17.1) channels (26). None of these explanations seemingly applies to K 2P 2.1. K 2P 2.1 channels neither possess a histidine residue at the selectivity filter, like K 2P 3.1, nor do they sense pH O by Lys-286 (homologous to Arg-224 in K 2P 5.1), as mutating Lys-286 to alanine failed to abolish K 2P 2.1 inhibition by protons (data not shown).
In seeking the pH sensor of human K 2P 2.1, we studied the mouse variant of the channel, not considered to be highly pH Osensitive (27). In our hands, the murine K 2P 2.1 channel indeed showed low sensitivity to external pH, with only 35% inhibition noted at pH 5.5 (Fig. 2B). A comparison of the two sequences revealed the presence of a histidine residue (His-87) in the human sequence not found in its murine counterpart ( Fig. 2A). Indeed, mutating the human His-87 to glutamine, as found at position 87 of the murine K 2P 2.1 sequence, resulted in a pH sensitivity profile almost identical to that of the murine variant (Fig. 2B). In light of this finding, we evaluated the roles of the three other human K 2P 2.1 histidine residues predicted to reside in extracellular loops of the channel by individually mutating them to alanine (i.e. Fig. 2A, H106A, H141A, and H247A). The H106A and H247A mutants did not significantly affect the pH O sensitivity of human K 2P 2.1 (Fig. 2B). By contrast, His-141 was found to participate in the channel response to pH O , as mutating this residue dramatically decreased channel responsiveness to external acidification, shifting the sensitivity curve away from the physiological range (Fig. 2B). The H87Q,H141A double mutant, however, showed a pH sensitivity similar to the single H141A mutant, suggesting that both residues act through a shared mechanism.
Voltage-dependent Proton-mediated Inhibition Increases Outward Rectification-We found that the proton-mediated inhibition is voltage-dependent, being more pronounced at negative potentials, with V1 ⁄ 2 of Ϫ79 mV Ϯ 1.7 (Fig. 3, A and B). This differential inhibition causes the K 2P 2.1 currents to be significantly more outwardly rectifying at acidic than at basic pH values (Fig. 3A). Accordingly, in the absence of proton-mediated inhibition (H141A mutant) or Mg 2ϩ blockade, the channel behaves as an open rectifier channel (Fig. 3, C and D). These results suggest a new mechanism contributing to the exceptional capability of K 2P 2.1 to function in a voltage-gated mode. We believe that the voltage dependence of the channel reflects, in part, release from voltage-dependent proton-mediated inhibition.
External Acidification Facilitates C-type Closure of K 2P 2.1-What is the mechanism by which external protons inhibit K 2P 2.1 currents? Neither His-87 nor His-141 is predicted, according to the known K ϩ channel structures (28), to lie along the ion conduction pathway. This, therefore, rules out the possibility that protons act through pore blockade. Hence, to elu-cidate the manner by which these residues sense pH O levels, we studied the effect of pH inhibition on K 2P 2.1 gating. In a homologous channel, i.e. the Drosophila KCNK0 channel, closing of the channel involves conformational changes in the external pore (16) through a mechanism resembling the C-type inactivation of voltage-gated potassium channels (29 -31). Might K 2P 2.1 pH sensors, i.e. His-87 and His-141, located at the first turret loop, induce closure of the external gate by a mechanism similar to C-type inactivation? C-type inactivation is known to be slowed by permeable external ions (29). As such, high external potassium concentrations might reduce the extent of K 2P 2.1 inhibition observed at acidic pH values. Indeed, elevating external potassium concentrations from 4 to either 20 or 100 mM dramatically interfered with pH inhibition of channel activity (Fig. 4A). Although lowering the pH from 9.0 to 7.0 produced a 42% inhibition in K 2P 2.1 currents attained in the presence of 4 mM external potassium, only 18% inhibition was measured in 20 mM potassium solution and a mere 11% inhibition was observed in the presence of 100 mM external potassium (Fig. 4A, right panel). In light of these results, we measured the ionic selectivity of K 2P 2.1 at different pH O values as conformational modification of the pore area alters the selectivity filter arrangement, thereby increasing sodium permeability (32)(33)(34). Lowering the pH of a 100-mM sodium solution from 9.0 to 7.0 caused a significant right shift of 20 mV (Ϫ134.4 Ϯ 0.6 to Ϫ114.3 Ϯ 0.9, mean Ϯ S.E., n ϭ 17) in the measured reversal potential of oocytes expressing wild type (WT) K 2P 2.1 channels (Fig. 4B, left panel). To exclude any nonspecific influences of low pH in these studies, the same experiment was repeated using the pH-insensitive mutant K 2P 2.1-H141A. In cells expressing the mutant, pH changes failed to alter the reversal potential (Fig. 4B, right panel). Accordingly, the K ϩ /Na ϩ permeability ratio in WT channels was considerably decreased (close to a 3-fold change) as the external pH was reduced (Fig. 4C). By contrast, the H141A mutant presented only a mild decrease (Ͻ20%) in the K ϩ /Na ϩ permeability ratio, with the change only being noted at pH 6.5, when a slight inhibition by protons was noted as well (Fig. 4C). The changes in the K ϩ /Na ϩ permeability ratio most likely result from an increase in Na ϩ permeability rather than a decrease in K ϩ permeability, as the single channel current amplitude, measured in symmetrical K ϩ concentrations (140 mM K ϩ and no Na ϩ in the pipette solution), was not affected by external pH changes (Fig. 1D). Rb ϩ and Cs ϩ selectivities were not changed in either the WT or H141A channels under acidic conditions (data not shown).
We next considered the effects on K 2P 2.1 pH sensitivity of two mutations, homologous to those shown previously to accelerate C-type gating in other potassium channels. In Shaker channels, mutation of an external threonine residue (i.e. Thr-449) located next to the selectivity filter alters C-type gating kinetics (31). Mutating the homologous position in the KCNK0 first pore domain (i.e. S112Y) enhanced C-type gating (16). A similar enhancement resulted from mutation of an extracellular glutamate, present in most potassium channels in the amino-terminal of the pore turret (i.e. Glu-28 in KCNK channels and Glu-418 in Shaker channels) (35,36). Ser-164, the K 2P 2.1 homolog of Shaker Thr-449 and KCNK0 Ser-112, was mutated to tyrosine (S164Y) and Glu-84, homologous to Shaker Glu-418 and KCNK0 Glu-28, was mutated to alanine (E84A). The K 2P 2.1-S164Y mutant demonstrated increased pH sensitivity, with a pK of 8.1 and Ͼ90% inhibition at low pH values (Fig. 4D). Surprisingly, the K 2P 2.1 E84A mutation, expected to affect C-type gating in the same manner, presented currents that were significantly less sensitive to external pH, with a maximal inhibition of only 40% (Fig. 4D).
The Molecular Mechanism of pH Sensing-The insensitivity of the E84A mutant to changes in pH O implies that Glu-84 takes part in the pH-sensing mechanism itself. It was previously suggested that Shaker Glu-418, which lies adjacent to the pore, stabilizes the open conformation of the slow inactivation gate by forming hydrogen bonds with residues in the P-S6 loop (36). Breaking this bond during conformational changes enables the rotation of the P-S6 loop and the subsequent collapse of the slow inactivation gate. Accordingly, we hypothesized that protonation of the external histidines, His-87 and His-141, would add positive charges that could interact with the negative charge of Glu-84 to draw this residue away from its natural interactions, thus causing a collapse of the selectivity filter region.
In Fig. 5A, we present one subunit of the bacterial KcsA potassium channel, based on its published structure (28). The predicted interactions of K 2P 2.1 at the homologous sites in KcsA are indicated as follows: at the primary sequence level, KcsA residues Glu-51, Val-84, and Thr-85 align with K 2P 2.1 residues Glu-84, Arg-166, and Thr-167, respectively. The side chain of KcsA Glu-51 is predicted to form hydrogen bonds (orange lines) with the backbone amides of Arg-166 and Thr-167 and the side chain hydroxyl group of Thr-167. KcsA Leu-59 and Ala-54 are replaced in this presentation by histidine resi-  JULY 11, 2008 • VOLUME 283 • NUMBER 28

JOURNAL OF BIOLOGICAL CHEMISTRY 19451
dues, as present at the homologous positions in K 2P 2.1 (i.e. His-141 and His-87, respectively). Potential hydrogen bonds between the Glu-84 hydroxyl groups and the histidine amide groups are drawn as orange lines. For the proposed mechanism of K 2P 2.1 pH O sensitivity to be valid, the three residues, namely Glu-84, His-87, and His-141, must be close enough to one another to interact. To test whether this is indeed the case, we first produced paired cysteine mutants, i.e. E84C and H87C, and E84C and H141C. Formation of either of the disulfide bonds (i.e. Cys-84-Cys-87 or Cys-84-Cys-141) would be expected to modify K 2P 2.1 currents whereas reduction would be expected to reverse this effect. Although channels containing a single cysteine mutation (i.e. E84C, H87C, or H141C) were active, channels including double mutations failed to produce any measurable currents (data not shown). Adding the reducing agent dithiothreitol (10 mM) to the bath solution did not affect this situation (data not shown), possibly because the introduced disulfide bonds locking the channel into the closed state were not accessible to the reducing agent. However, we cannot exclude other possibilities at this point, such as improper assembly or massive conformational changes of the channel caused by introducing two mutations in this vital domain of the channel.
Seeking an alternative manner to demonstrate the physical proximities of Glu-84 and His-141/His-87, we exploited the ability of three or more histidine and cysteine residues to coordinate the binding of Cd 2ϩ ions. If Glu-84 is indeed sufficiently  . B, pH affects K 2P 2.1-mediated reversal potential in oocytes. Whole-cell currents were assessed in oocytes by two-electrode voltage clamp at pH 9.0 or 7.0, as indicated, in 100 mM sodium bath solution. Oocyte membrane potential was held at Ϫ80 mV and pulsed from Ϫ150 to Ϫ50 mV in 10-mV voltage steps for 60 ms with 1-s interpulse intervals. Currents of a representative oocyte expressing K 2P 2.1 WT (left) or the H141A mutant (right) channels are shown. A 20-mV shift in reversal potential was measured in WT channels (Ϫ134.4 Ϯ 0.6 to Ϫ114.3 Ϯ 0.9 mV; mean Ϯ S.E., n ϭ 17) but not in the H141A mutant channels (Ϫ128.9 Ϯ 1.2 to Ϫ126.8 Ϯ 1.4 mV; mean Ϯ S.E., n ϭ 17). C, pH affects the ion selectivity of K 2P 2.1 channels. Whole-cell macroscopic reversal potentials (E rev ) were measured by two-electrode voltage clamp in 100 mM sodium and 100 mM potassium bath solutions at different pH levels. Permeability ratios were approximated (for each pH value separately) according to: P K /P Na ϭ exp(ϪFE rev /RT), where P K and P Na represent the permeabilities of potassium and sodium, respectively. Results are shown as means Ϯ S.E. (n ϭ 10 -29). D, extracellular pH dependence curves of mutant E84A (triangles) and S164Y (squares) channels as compared with human WT channels (circles). Currents were studied and analyzed as in Fig. 1. Results are shown as means Ϯ S. E. (n ϭ 7). The solid lines represent a fit of the data to the Hill equation.
adjacent to both His-141 and His-87, replacing this glutamate with a cysteine residue (E84C) should create a new Cd 2ϩ -binding site. As proposed, K 2P 2.1-E84C was Ͼ10-fold more sensitive to Cd 2ϩ ions than was the WT channel (Fig. 5B). Furthermore, damaging the newly created Cd 2ϩ -binding site by mutating either of the histidine residues in the K 2P 2.1-E84C mutant background (i.e. E84C,H141A; E84C,H87Q; or E84C,H87Q,H141A) reduced Cd 2ϩ binding back to WT levels (Fig. 5B).

DISCUSSION
In the present report, we have shown that extracellular protons within the physiological range strongly inhibit the human K 2P 2.1 channel by affecting the channel open probability (Fig.  1). Transient pH variations occur in all three compartments of nervous tissue, i.e. in neurons, glial cells, and extracellular spaces, in response to neuronal stimulation to neurotransmitters and hormones as well as in a secondary manner to metabolic activity and ion transport (37). Shifts in blood flow, synchronous activation of nerve cells, seizure, or spreading depression have been associated with interstitial alkaline shift or acidosis that could persist for minutes in different segments of the central nervous system (reviewed in Ref. 38). Such changes are known to dramatically modulate ion channel activity and hence affect neuronal tissue excitability. As decreased tissue pH is a common feature of the ischemic brain that can lead to neuronal injury, Xiong et al. (39) had recently suggested acid-sensing ion channels as novel therapeutic targets for ischemic brain injury. Being modulated by intra-and extracellular acidification, the K 2P 2.1 channel is an important sensor for pH changes in the central nervous system.
We have identified two histidine residues located in the first turret loop (i.e. His-87 and His-141), predicted to lie in the outer vestibule of the conduction pathway, as being responsible for the K 2P 2.1 response to external pH changes (Fig. 2). The location of the two histidine residues implies a new pH-sensing mechanism, different from those previously suggested for K 2P channels (23)(24)(25)(26). In this context, the His-87 residue is especially interesting, as it is present in the human K 2P 2.1 variant (as well as in all other known non-rodent, yet unstudied, mammalian homologs) but is replaced by glutamine in the mouse K 2P 2.1 channel, which is only mildly affected by pH O (Fig. 2). It is possible that the two variants thus serve distinct physiological roles as their general expression patterns, as well as their central nervous system distribution, diverge (2,40). His-87 is also absent from the human K 2P 10.1 (TREK-2) channel, closely related to K 2P 2.1 (78% homology) and presenting significant similarity to K 2P 2.1 in terms of biophysical properties, as well as with respect to channel regulation traits (41,42). By placing a phenylalanine residue instead of histidine at position 102 (homologues to K 2P 2.1 His-87), the K 2P 10.1 channel was found to modestly react to extracellular pH changes (42). Hence, the addition of a single histidine residue in the turret loop may offer an efficient means to expand the physiological diversity of K 2P channels. This may represent a more general phenomenon to govern functional variance of potassium channels. For example, the pH sensitivity of the rK V 1.5 channel is mediated by a histi-  JULY 11, 2008 • VOLUME 283 • NUMBER 28 dine residue in the turret loop (His-452), absent in the closely related, pH-insensitive, K V 1.2 channel (43).

Inhibition of Human K 2P 2.1 Channel by External Protons
Our results indicate that the outward rectification of K 2P 2.1 is more pronounced at low external pH values and almost completely lost in the pH-insensitive mutant H141A at physiological pH values (Fig. 3). Because neither of the two studied histidine residues is predicted to lie within the membrane electrical field (Fig. 5A), we presuppose a non-direct effect of the membrane potential. Thus, the identity of the actual membraneembedded voltage sensor has yet to be revealed. Two mechanisms were previously suggested to explain the limitation of K 2P 2.1 currents at negative potentials, namely voltage-dependent external Mg 2ϩ blockade or an intrinsic voltage-sensing mechanism, argued to depend on an essential carboxyl-terminal protein kinase A phosphorylation site in human K 2P 2.1, Ser-348 (V1 ⁄ 2 ϭ 10 Ϯ 7 mV) (17), or on the carboxyl-terminal domain of the mouse K 2P 2.1 variant (44). In addition, it was shown that agonist-induced phosphatidylinositol-4,5-bisphosphate hydrolysis shifts the voltage sensitivity of K 2P 2.1 channels toward depolarized potentials (45). Here, we present a novel mechanism for outward rectification of K 2P 2.1, depending on extracellular pH sensors, presenting a V1 ⁄ 2 of Ϫ79 mV, likely to play a significant physiological role in humans. Voltage dependence of K 2P 2.1 activity could have profound physiological significance by modulating membrane resting potential, action potential duration, firing rate, and other biophysical properties of human excitable systems.
We further propose that protonation of the pH sensors promotes the collapse of the external pore gate, similar to the C-type inactivation of voltage-gated potassium channels (Fig.  4). Several lines of evidence support this hypothesis. High levels of external potassium, known to slow C-type inactivation, decrease the effect of protons (Fig. 4A). The observation that pH O alters K 2P 2.1 potassium selectivity suggests that protoninduced conformational modifications occur at the selectivity filter (Fig. 4, B and C). Finally, mutating serine 164 to tyrosine, a mutation that accelerated outer pore gating in Shaker (31) and KCNK0 (16) channels, dramatically increased the pH O effect observed here (Fig. 4D). Indeed, our observation represents the first reported indications for outer pore gating in K 2P 2.1 channels. These results are consistent with previous findings indicating that gating of the KCNK0 leak channel involved a mechanism resembling C-type inactivation (16) and that external acidification accelerates C-type gating in K 2P 3.1, Shaker, Kv1.4, and K V 1.5 potassium channels (11,32,43,46).
How can protonation of His-87 and His-141 induce destabilization of the K 2P 2.1 external gate? We attained our first insight into this putative mechanism by analyzing the effect of an E84A mutation on pH O sensitivity. Neutralization of this residue was expected to accelerate C-type gating (as with mutation of KCNK0 Glu-28 and Shaker Glu-418) and hence elicit increased pH O sensitivity, similarly to the K 2P 2.1-S164Y mutant. Surprisingly, E84A currents were significantly less sensitive to external pH, implying a more direct role for Glu-84 in pH sensing. Is Glu-84 sufficiently close to the pH-sensing histidines to sense their charge variations? The introduction of a cysteine residue at this site created a Cd 2ϩ -binding site in the K 2P 2.1 channel. We speculate that this Cd 2ϩ -binding site is formed by the original histidines together with the introduced cysteine, as it was abolished when either histidine residue was eliminated (Fig. 5B). Accordingly, the double cysteine-substituted mutants (i.e. Cys-84-Cys-87 and Cys-84-Cys-141) were inactive, probably due to the formation of dithiothreitol-inaccessible disulfide bonds. These results thus indicate that position Glu-84 is sufficiently proximal to the pH-sensing histidines (positions 87 and 141), in accordance with the predicted positions of homologous residues in the KcsA structure (28). Therefore, we hypothesize that adding positive charges to the external histidines draws the negatively charged Glu-84 residue away from its stabilizing interactions (with Arg-166 and Thr-167), thus causing the outer pore gate to collapse (Fig. 5A).
Taken together, our results highlight the physiological importance of human K 2P 2.1 channels as important sensors of pH changes in the central nervous system, with external pH changes modifying both their currents and membrane potential dependence. Moreover, we offer new insight into the molecular mechanism and structural rearrangements that occur during the K 2P 2.1 response to pH O , involving an external pH sensor, identified here for the first time.