A pH-sensitive yeast outward rectifier K+ channel with two pore domains and novel gating properties.

YORK is a newly cloned K+ channel from yeast. Unlike all other cloned K+ channels, it has two pore domains instead of one. It displays eight transmembrane segments arranged like a covalent assembly of a Shaker-type voltage-dependent K+ channel (without S4 transmembrane segments) with an inward rectifier K+ channel. When expressed in Xenopus oocytes, YORK does not pass inward currents; it conducts only K+-selective outward currents. However, the mechanism responsible for this strict outward rectification is unusual. Like inward rectifiers, its activation potential threshold closely follows the K+ equilibrium potential. Unlike inward rectifiers, the rectification is not due to a voltage-dependent Mg2+ block. The blocking element is probably intrinsic to the YORK protein itself. YORK activity is decreased at acidic internal pH, with a pKa of 6.5. Pharmacological and regulation properties were analyzed. Ba2+ ions and quinine block YORK currents through high and low affinity sites, while tetraethylammonium displays only one affinity for blocking. Activation of protein kinase C indirectly produces an increase of the current, while protein kinase A activation has no effect.

YORK is a newly cloned K ؉ channel from yeast. Unlike all other cloned K ؉ channels, it has two pore domains instead of one. It displays eight transmembrane segments arranged like a covalent assembly of a Shakertype voltage-dependent K ؉ channel (without S4 transmembrane segments) with an inward rectifier K ؉ channel. When expressed in Xenopus oocytes, YORK does not pass inward currents; it conducts only K ؉ -selective outward currents. However, the mechanism responsible for this strict outward rectification is unusual. Like inward rectifiers, its activation potential threshold closely follows the K ؉ equilibrium potential. Unlike inward rectifiers, the rectification is not due to a voltage-dependent Mg 2؉ block. The blocking element is probably intrinsic to the YORK protein itself. YORK activity is decreased at acidic internal pH, with a pK a of 6.5.
Pharmacological and regulation properties were analyzed. Ba 2؉ ions and quinine block YORK currents through high and low affinity sites, while tetraethylammonium displays only one affinity for blocking. Activation of protein kinase C indirectly produces an increase of the current, while protein kinase A activation has no effect.
A large number of K ϩ channel subunits have been identified in the last few years (1)(2)(3)(4)(5)(6). All of the pore-forming K ϩ channel subunits cloned in vertebrates as well as in plants, paramecia, and bacteria (7-10) share a highly conserved sequence called the pore (or H5) domain (P-domain), 1 which has been shown to be part of the K ϩ -selective pore (11)(12)(13)(14). The presence of this sequence in a protein is considered the signature of a K ϩ channel structure. In all K ϩ channels so far cloned, there is never more than one P-domain per subunit and a variable number of transmembrane segments (TMS). Voltage-gated K ϩ channels have six TMS, inward rectifier K ϩ channels have two TMS (1)(2)(3)(4), and Ca 2ϩ -dependent K ϩ channels have 10 TMS (15,16). All six-TMS channels have a positively charged TMS (S4) that is responsible for voltage sensitivity. The two-TMS K ϩ channels do not possess such a domain. Their apparent voltage sensitivity for gating results from a voltage-dependent block by intracellular Mg 2ϩ or polyamines (17)(18)(19)(20)(21)(22). YORK, the yeast outward rectifying K ϩ channel described in this work, presents both novel structural features and novel functional properties.

EXPERIMENTAL PROCEDURES
Molecular Biology-YORK cDNA was amplified from Saccharomyces cerevisiae DNA by polymerase chain reaction with a low error rate polymerase (Pwo DNA pol, Boehringer Mannheim), cloned into the pEXO vector, and sequenced on both strands by automatic sequencing using an Applied Biosystems Model 373A sequencer. Site-directed mutagenesis was carried out using a Transformer system (CLONTECH) according to the manufacturer's protocol. Wild-type and mutant plasmids were linearized by BamHI digestion, and then capped cRNAs were synthesized using T7 RNA polymerase (Stratagene).
For inside-out patch recordings, pipettes were filled with a high K ϩ solution (40 mM KCl, 100 mM potassium methanesulfonate, 1.8 mM CaCl 2 , 2 mM MgCl 2 , 5 mM HEPES adjusted to pH 7.4 with KOH, and 10 M GdCl 3 ) to prevent endogenous stretch-activated channel activities. The bath solution contained 140 mM KCl, 3 mM MgCl 2 , 5 mM EGTA, and 5 mM HEPES adjusted to pH 7.2 with KOH. For outside-out recordings, pipettes were filled with a solution identical to that used for the bath solution in the inside-out patch recordings. The external medium contained 140 mM KCl, 0.3 mM CaCl 2 , 1 mM MgCl 2 , and 10 mM HEPES adjusted to pH 7.1 with KOH. The external divalent-free solution contained 140 mM KCl, 2 mM EGTA, 2 mM EDTA, and 10 mM HEPES adjusted to pH 7.1 with KOH. Single-channel recording data were filtered at 5 kHz and digitized at 50 kHz. CO 2 -enriched solution was prepared by bubbling 100% CO 2 into the standard ND96 solution for 5 min. For inside-out patch experiments, internal solutions at different pH values were prepared by adding Na 2 HPO 4 and NaH 2 PO 4 at different ratios. To obtain pH values of 6, 6.4, 6.8, 7.2, 7.6, and 8, the Na 2 HPO 4 /NaH 2 PO 4 ratios were 1.2/8.76, 2.6/7.34, 4.8/5.1, 7.2/2.8, 8.6/1.3, and 9.4/0.5 mM, respectively. The curve was the least-square fit of the following equation: I ϭ I max /(1 ϩ (exp(pK Ϫ pH))/a), with pK ϭ 6.5, I max ϭ 1.4 pA, and a ϭ 0.38.

RESULTS AND DISCUSSION
Structural Properties of YORK-A search in the GenBank/ EMBL Data Bank for sequences homologous to the mSlo Pdomain (15) led to the identification of a DNA sequence that presented the particularity of displaying two P-like domains in the same open reading frame. This sequence (accession number X77087) was originally obtained from chromosome X of S. cerevisiae in the framework of the European Yeast Genome Sequencing Project (24). The full coding sequence was amplified by polymerase chain reaction from yeast DNA and cloned in pEXO, a vector designed for preparation of efficiently transcribed cRNA in Xenopus oocytes (25).
The YORK gene encodes a 691-amino acid-long protein that does not share significant sequence conservation with previously cloned K ϩ channels, except in the two P-domain se-* This work was supported by CNRS and the Association Française contre les Myopathies and by Commission of the European Communities Contract CHRX-CT93-0167. 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) U37254.
‡ To whom correspondence should be addressed. Tel.: 33-93-95-77-00; Fax: 33-93-95-77-04; E-mail: douy@unice.fr. 1 The abbreviations used are: P-domain, pore domain; TMS, transmembrane segment(s); PMA, phorbol 12-myristate 13-acetate. quences ( Fig. 1, a and c). The hydropathy profile shown in Fig.  1b predicts eight membrane-spanning hydrophobic segments (S1 to S8). The P-domains P1 and P2 are flanked by TMS S5 and S6 and TMS S7 and S8, respectively. Placing the NH 2 terminus on the cytoplasmic side, in agreement with the absence of signal peptide, leads to the topology model proposed in Fig. 1b. In this model, the two P-domains are inserted into the membrane from the outside, corresponding to the known orientation of these loops in all K ϩ channels. Moreover, the overall structural motif of YORK is similar to the motif that would be obtained by adding a six-TMS subunit of a classic voltage-dependent K ϩ channel to a two-TMS inward rectifier subunit. However, none of the YORK TMS shows the highly conserved charged and regularly spaced residues that are responsible for the voltage-sensing ability of outward rectifier K ϩ channels. The alignment of P-domain sequences shown in Fig. 1c indicates that the K ϩ -selective pore regions of YORK are well conserved. In all K ϩ channels, 2 Gly residues are perfectly conserved (Gly-16 and Gly-18) (Fig. 1c), while nearly exclusive conservative changes are found for the residues at positions 7 (Phe or Trp, except for Shal, which has Tyr), 14 (Thr or Ser), and 17 (Tyr or Phe). YORKP2 is perfectly canonical with regard to all of these residues. In the case of YORKP1, a Leu residue is found in place of Tyr-17, which is present in all but one of the other K ϩ channels. In addition to these five positions, the global conservation of YORKP1 and YORKP2 with other Pdomains is high and suggests the participation of both segments in the K ϩ channel permeation pathway. Site-directed mutagenesis was used to try to gain experimental evidences that this is indeed the case. In the mutant YORKP1-L17Y, the unusual Leu residue was replaced by Tyr, while in YORKP2-Y17L, the classical Tyr residue was replaced by Leu. Both mutants were found to be nonfunctional. The absolute necessity for function of the Tyr residue in the second P-domain was not surprising since several mutations of the corresponding Tyr in Shaker channels resulted in production of nonfunctional channels (12). Conversely, the loss of function observed for the YORKP1-L17Y mutant was unexpected. Although improper protein folding or sorting cannot be excluded, this negative result suggests that the GLG motif is likely to belong to a functional P-domain. The double mutant YORKP1-L17Y/ P2Y17L, in which the Leu and Tyr residues were mutually exchanged, was also nonfunctional. Clearly, more mutations will be necessary to definitively establish the participation of both P-domains in the formation of the K ϩ permeation of YORK. Other relevant features of the YORK protein are the presence of three N-glycosylation sites located in the external S1-S2 and S7-P2 loops and several phosphorylation consensus sites, including seven sites for protein kinase C and one site for protein kinase A and tyrosine kinase (Fig. 1a).
Functional Expression of YORK Channels-The biophysical properties of the YORK channels were characterized after injection of their cRNA into Xenopus oocytes. Large currents of several microamperes were recorded that were strictly outward rectifying. Depolarization pulses induced a two-step current composed of instantaneous and delayed slow components (Fig.  2a). The delayed current could be fitted by an exponential function, with a time constant varying from 0.7 s at Ϫ50 mV to 0.2 s at ϩ50 mV (Fig. 2b, inset). The instantaneous and delayed I-V relationships are similar. No K ϩ inward currents were recorded (Fig. 2b), not even as a detectable tail current regardless of the repolarization level (see Fig. 7a). The threshold potential for activation followed the change in the K ϩ equilibrium potential (E K ) as the external K ϩ concentration decreased (55.4 Ϯ 0.7-mV shift (n ϭ 5) for a 10-fold change in the external K ϩ concentration) (Fig. 2c). Hence, the YORK activation potential depends on the driving force for K ϩ (V Ϫ E K ) as described for inward rectifier K ϩ channels (26) rather than on the membrane potential (V) as described for voltage-dependent outward rectifier K ϩ channels. Moreover, as for inward rectifier K ϩ channels (26), the maximum chord conductance of YORK was approximately proportional to the square root of the external K ϩ concentration (Fig. 2c). Chord conductance was independent of the presence of Na ϩ in the bath, demonstrating a high K ϩ versus Na ϩ selectivity (data not shown).
In inside-out patches containing several YORK channels, responses to depolarizing voltage steps, from a holding potential of Ϫ20 mV, were consistent with those obtained in whole cell measurements, i.e. an instantaneous activity followed by a progressive increase in the number of open channels (Fig. 2d). Single-channel recordings showed that the current was strictly outward, with a unitary conductance of 26 Ϯ 2 picosiemens (n ϭ 5), and had a high tendency to flicker, with a mean open time of 0.19 Ϯ 0.01 ms (six inside-out patches) (Fig. 3, a and b).
The absence of an S4 voltage sensor structure, associated with striking similarities to the biophysical properties of inward rectifier K ϩ channels, suggested that the gating mechanism for the outwardly rectifying YORK channel might be analogous to that of inward rectifiers, i.e. a voltage-dependent block by penetration of an impermeant ion in the pore. Mg 2ϩ or polyamine cations play this role for inward rectifier channels (17)(18)(19)(20)(21)(22). Surprisingly, Mg 2ϩ and Ca 2ϩ removal from the external solution has no effect on the outward rectifying properties of YORK whole cell currents (data not shown). This lack of effect of divalent cations on YORK rectification was fully confirmed on excised patches. In outside-out configuration, the complete chelation of divalent cations by EGTA and EDTA in the external bathing solution did not allow inward currents to be recorded (Fig. 3c). This finding has been validated on seven independent patches and differs from the results of Ketchum et al. (27) that appeared while this paper was submitted for publication. Therefore, it cannot be concluded, as these authors did, that the mechanism leading to YORK outward rectification is strictly the same as in inward rectifiers, but in the opposite direction. The blocking element responsible for outward rectification probably belongs to the channel protein itself and obstructs the permeation pathway when the membrane potential is below E K .
Pharmacological Properties-The pharmacology of YORK channels was studied under two-microelectrode conditions. Currents are blocked by Ba 2ϩ (Fig. 4a). The instantaneous component of the current is 150 times more sensitive than the delayed component. The complete dose-response curve is bimodal, with IC 50 values of 20 M and 3 mM for the instantaneous and delayed components, respectively (Fig. 4b). Quinine is also a good blocker of YORK current. Like Ba 2ϩ , this compound is more potent on the instantaneous component, with an IC 50 of 90 M (Fig. 4d). The sensitivity of the delayed component to quinine is by far lower and was not accurately determined. Both current components are equally sensitive to tetraethylammonium (IC 50 ϭ 5.5 mM) (Fig. 4c). Known blockers of other K ϩ channels such as Cs ϩ (10 mM), 4-aminopyridine (1 mM), dendrotoxin (100 nM), charybdotoxin (3 nM), apamin (300 nM), tedisamil (30 M), and glibenclamid (10 M) and K ϩ channel openers such as pinacidil and P1060 (both at 100 M) had no effect on YORK currents.
Regulation of YORK Channels by Internal pH and Phosphorylations-YORK channel activity was inhibited at acidic pH (Fig. 5, a and b). Inhibition by protons was conserved in insideout excised patches and shown to result from a decrease in the number of active channels rather than from changes in their unitary conductance or kinetic parameters (Fig. 5c). The pH dependence of the open probability (NPo) had a pK a of 6.5, which suggests participation of one of the YORK histidine residues in the blockade (Fig. 5d). Since external pH (from 4.3 to 8) had no effect on YORK activity, it is likely that the pH-sensitive groups are located intracellularly.
Activation of protein kinase C by PMA (30 nM) increased the YORK current (Fig. 6, a and b). The time course of activation (Fig. 6c) and the range of PMA concentrations needed to observe the effect (between 1 and 60 nM) (Fig. 6e) are those expected for the protein kinase C activation in Xenopus oocytes (28,29). Moreover, the inactive phorbol ester 4␣-phorbol 12,13didecanoate was without effect (Fig. 6c), as was PMA after preincubation with the protein kinase C inhibitor staurosporine (30 M) (Fig. 6e). Activation of protein kinase C can also be achieved in a more physiological way by activating a receptor coupled to phospholipase C through G proteins. When the cloned 5-HT 2 serotonin receptor (30) was coexpressed with YORK, serotonin, similarly to PMA, induced an increase of the K ϩ current (Fig. 6d). All these results taken together clearly indicate that protein kinase C is involved in YORK regulation. To our knowledge, this is the first description of activation of a cloned K ϩ channel by protein kinase C. However, when protein kinase C was directly applied on excised patches, it did not increase channel activity (data not shown). It is therefore likely that the protein kinase C regulation is indirect and requires the involvement of soluble cytoplasmic components. Elevation of [Ca 2ϩ ] i by application of A23187 (1 M) or microinjection of inositol triphosphate (1 M) did not affect YORK activity. Therefore, YORK seems insensitive to variations of internal Ca 2ϩ . Protein kinase A activation by application of 8-chloro-cAMP (300 M) or forskolin (10 M) had no effect on channel function.
Gating Kinetics and Proposed Model for YORK Rectification-Tail currents were absent in the voltage range of Ϫ120 to ϩ30 mV following voltage pulses to ϩ30 mV, at which the K ϩ channels are fully activated (Fig. 7a). In a simplified kinetic model of the YORK channel comprising only one open state and one closed state, the lack of any detectable time course of the tail currents suggests a very fast transition from the open to the closed state. The ratio between the slow and the instantaneous components of the YORK currents during step depolar-izations is dependent on the holding potential. Membrane hyperpolarization clearly increases this ratio, as shown in Fig. 7b, where the same outside-out patch was stimulated repeatedly, first in a polarized condition and then in a depolarized condition. Voltage-dependent changes in the distribution of the two components of the YORK channel currents are also shown in Fig. 7 (c and d). A brief (40 ms) membrane potential change during the onset of the YORK current following a step depolarization is enough to modify the ratio between the slow and the instantaneous components of the K ϩ current. The majority of the initial slow component reappeared after a brief hyperpolarization to Ϫ200 mV. The time dependence for recovery of the slow component from a voltage pulse to ϩ30 mV is analyzed in Fig. 7 (e and f) by changing the time interval between pulses. The time for 50% recovery of the slow component was ϳ0.5 s. The absence of tail current kinetics and the voltage and time dependences for the recovery of the slow component suggest a conversion of the slow component into an instantaneous component.
The model presented in Fig. 7g takes into account all the properties of YORK channels described in this study. (i) The currents are outwardly rectifying. (ii) As for inward rectifiers, the activation potential is coupled to E K . (iii) The rectification persists in the absence of external divalent cations. (iv) Two kinetically distinct current components are observed.
In this model, an endogenous blocking particle such as an external peptide domain of the channel enters the permeation pathway upon hyperpolarization (Fig. 7g). The presence of two binding sites is hypothesized to account for the instantaneous and the delayed slow currents. A "deep" binding site for the blocking particle is favored in hyperpolarization against a "shallow" binding site. Under polarized conditions, YORK channels are either in the deep blocked state (C1) or in the shallow blocked state (C2). Upon depolarization, the channels undergo either a "delayed" conformational change from C2 to C1 or an "instantaneous" transition from C1 to the open state (O). The transition C1 3 O will give the instantaneous component of the current, while the transitions C1 3 C2 3 O will impose the delayed component. Upon repolarization, the only "observable" transition is O 3 C1, which is instantaneous. The model involves two binding sites for the putative blocking element. This view is supported by the fact that both Ba 2ϩ and quinine display two affinities for YORK blockade (Fig. 4). Conclusions-Outward rectifier K ϩ channels have been described in yeast (31,32). Unlike YORK, they are Ca 2ϩ -sensitive, but like YORK, they are tetraethylammonium-sensitive, and their relevant gating parameter appears to be V Ϫ E K instead of V (32). This peculiar property suggests that these channels function in balancing charge movements during nutriment transport in Saccharomyces (33,34). The gating properties of YORK, its slow activation kinetics, and its inhibition by intracellular acidosis are similar to the properties of K ϩ channels described in stomatal guard cells (35,36). It is thus likely that the YORK channel has at least a functional counterpart in plant cells and perhaps in other higher organisms as well. A search in the GenBank/EMBL Data Bank with the YORK sequence led to the identification of a Caenorhabditis elegans gene characterized in the framework of the Nematode Sequencing Project (accession number Z35719). This gene encodes a putative K ϩ channel containing two P-domains with GIG and GYG sequences. It seems likely that the presence of two P-domains with the motifs GYG and GL/IG is a hallmark shared by members of a novel class of K ϩ channels. Whether these structural relationships are associated with functional similarities remains to be determined after cloning and expression of the C. elegans gene. FIG. 7. Voltage dependence of gating kinetics of YORK currents. a, absence of tail current following a voltage pulse to ϩ30 mV. Inset, voltage clamp protocol. b, same outside-out patch in symmetrical K ϩ solution (140 mM). Average currents were obtained from 30 successive voltage pulses to ϩ60 mV from holding potentials (HP) of Ϫ60 mV (left trace) and 0 mV (right trace) (n ϭ 10). c and d, voltage dependence of the recovery of the slow component of the YORK current obtained after a short interruption (40 ms) of the test pulse to potentials from Ϫ200 to ϩ30 mV. c, superimposed current traces obtained with the pulse protocol illustrated in the inset of d (1-min interval between each pulse). d, plot of the percentage of recovery for the slow component as a function of membrane potential; same experiment as in c. e and f, effect of the time interval between two successive test pulses on the recovery of the slow component. e, superimposed current traces at increasing pulse intervals. f, plot of the percentage of recovery for the slow component as a function of the time between two successive pulses; same experiment as in e. g, proposed model for YORK channel gatings with an intrinsic blocking particle. Em, membrane potential.