An ACTH- and ATP-regulated Background K+ Channel in Adrenocortical Cells Is TREK-1*

Bovine adrenal zona fasciculata (AZF) cells express a background K+channel (IAC) that sets the resting potential and acts pivotally in ACTH-stimulated cortisol secretion. We have cloned a bTREK-1 (KCNK2) tandem-pore K+ channel cDNA from AZF cells with properties that identify it as the native IAC. The bTREK-1 cDNA is expressed robustly in AZF cells and includes transcripts of 4.9, 3.6, and 2.8 kb. In patch clamp recordings made from transiently transfected cells, bTREK-1 displayed distinctive properties of IAC in AZF cells. Specifically, bTREK-1 currents were outwardly rectifying with a large instantaneous and smaller time-dependent component. Similar to IAC, bTREK-1 increased spontaneously in amplitude over many minutes of whole cell recording and was blocked potently by Ca2+ antagonists including penfluridol and mibefradil and by 8-(4-chlorophenylthio)-cAMP. Unitary TREK-1 and IAC currents were nearly identical in amplitude. The native IAC current, in turn, displayed properties that together are specific to TREK-1 K+ channels. These include activation by intracellular acidification, enhancement by the neuroprotective agent riluzole, and outward rectification. bTREK-1 current differed from native K+ current only in its lack of ATP dependence. In contrast to IAC, the current density of bTREK-1 in human embryonic kidney-293 cells was not increased by raising pipette ATP from 0.1 to 5 mm. Further, the enhancement of IAC current in AZF cells by low pH and riluzole was facilitated by, and dependent on, ATP at millimolar concentrations in the pipette solution. Overall, these results establish the identity of IAC K+ channels, demonstrate the expression of bTREK-1 in a specific endocrine cell, identify potent new TREK-1 antagonists, and assign a pivotal role for these tandem-pore channels in the physiology of cortisol secretion. The activation of IAC by ATP indicates that native bTREK-1 channels may function as sensors that couple the metabolic state of the cell to membrane potential, perhaps through an associated ATP-binding protein.

Bovine adrenal zona fasciculata (AZF) cells express a background K ؉ channel (I AC ) that sets the resting potential and acts pivotally in ACTH-stimulated cortisol secretion. We have cloned a bTREK-1 (KCNK2) tandempore K ؉ channel cDNA from AZF cells with properties that identify it as the native I AC . The bTREK-1 cDNA is expressed robustly in AZF cells and includes transcripts of 4.9, 3.6, and 2.8 kb. In patch clamp recordings made from transiently transfected cells, bTREK-1 displayed distinctive properties of I AC in AZF cells. Specifically, bTREK-1 currents were outwardly rectifying with a large instantaneous and smaller time-dependent component. Similar to I AC , bTREK-1 increased spontaneously in amplitude over many minutes of whole cell recording and was blocked potently by Ca 2؉ antagonists including penfluridol and mibefradil and by 8-(4-chlorophenylthio)-cAMP. Unitary TREK-1 and I AC currents were nearly identical in amplitude. The native I AC current, in turn, displayed properties that together are specific to TREK-1 K ؉ channels. These include activation by intracellular acidification, enhancement by the neuroprotective agent riluzole, and outward rectification. bTREK-1 current differed from native K ؉ current only in its lack of ATP dependence. In contrast to I AC , the current density of bTREK-1 in human embryonic kidney-293 cells was not increased by raising pipette ATP from 0.1 to 5 mM. Further, the enhancement of I AC current in AZF cells by low pH and riluzole was facilitated by, and dependent on, ATP at millimolar concentrations in the pipette solution. Overall, these results establish the identity of I AC K ؉ channels, demonstrate the expression of bTREK-1 in a specific endocrine cell, identify potent new TREK-1 antagonists, and assign a pivotal role for these tandem-pore channels in the physiology of cortisol secretion. The activation of I AC by ATP indicates that native bTREK-1 channels may function as sensors that couple the metabolic state of the cell to membrane potential, perhaps through an associated ATP-binding protein.
Bovine AZF 1 cells express a novel K ϩ -selective channel (I AC ) that functions pivotally in the regulation of cortisol secretion (1)(2)(3). I AC displays a number of properties typical of background or leak-type channels that function in setting the resting membrane potential of various cells. However, native I AC channels also exhibit a combination of properties not observed previously in a single type of K ϩ channel.
In whole cell patch clamp recordings, I AC appears as a noninactivating, outwardly rectifying K ϩ current that displays weak voltage dependence and grows spontaneously over many minutes (1,3,4). At the single channel level, the conductance and rectifying properties of these channels depend strongly on the presence of divalent cations (5). In symmetrical K ϩ , with physiological concentrations of Ca 2ϩ and Mg 2ϩ , the channels are outwardly rectifying, whereas they become nearly ohmic in the absence of divalent cations (4,5).
In whole cell patch clamp recordings, I AC is inhibited potently by ACTH and angiotensin II, the two peptide hormones that regulate cortisol secretion physiologically (1,6). I AC channels are also inhibited by paracrine factors, including ATP and adenosine, which act through specific G protein-coupled nucleotide receptors (7,8).
Second messengers synthesized or released upon activation of these receptors, including Ca 2ϩ and cAMP, inhibit I AC channels. However, inhibition appears to involve atypical signaling pathways, because responses to peptides and second messengers are insensitive to inhibition of cAMP and Ca 2ϩ -activated protein kinases (3,5,6,9).
Among background K ϩ channels, I AC channels are unique in their activation by hydrolyzable and non-hydrolyzable forms of ATP, applied intracellularly at physiological concentrations through a recording pipette (4). These background channels are also activated by UTP, GTP, and by inorganic polytriphosphate (PPP i ) (4,5). I AC channels also display a unique pharmacological profile. Specifically, these channels are relatively insensitive to many standard K ϩ channel antagonists, including 4-AP and TEA (respective IC 50 values of 2.8 and 24.3 mM), whereas they are blocked by others such as quinidine (IC 50 ϭ 24.1 M) (10). Surprisingly, I AC K ϩ channels are blocked potently by the Ca 2ϩ antagonists penfluridol and mibefradil at far lower concentrations (respective IC 50 values of 0.20 and 0.50 M) (10,11).
Although I AC K ϩ channels function as background or leak K ϩ channels that set the resting potential of AZF cells, their aggregate properties do not match the profile of any K ϩ chan-nel yet described. Regardless, these channels appear to function critically in the physiology of cortisol secretion. In this capacity, they act as sensors that integrate hormonal and metabolic signals and couple these to depolarization-dependent Ca 2ϩ entry.
Because of their distinctive properties as a metabolic sensor and role in cortisol secretion, it will be important to determine the molecular identity of I AC channels. In this regard, K ϩselective ion channels in mammals can be divided into several families, based on structure and membrane topology. Voltagegated K ϩ channels include a major subunit that consists of one pore domain and six transmembrane segments (1P/6TMS) (12). A second family that consists of channel subunits containing a single pore domain and two membrane-spanning segments (1P/ 2TMS) includes inward rectifiers many of which contribute to the resting potential and some of which are inhibited directly by intracellular ATP (12)(13)(14).
K ϩ -selective leak or background channels comprise the third and most recently described family of K ϩ channels. In mammals, the largest group of leak K ϩ channels possesses distinctive subunits that consist of two pore domains and four transmembrane segments. These 2P/4TMS channels are typically non-inactivating K ϩ channels that set the resting potential in a range of cells (15)(16)(17). Since 1995, more that a dozen members of the 2P/4TMS family have been identified and characterized by molecular cloning and patch clamp techniques (15)(16)(17).
Although similar in structure, 2P/4TMS K ϩ channels can be distinguished based on a variety of functional properties including unitary conductance, rectification, sensitivity to mechanical stimulation, pH and temperature, modulation by cAMP, fatty acids, membrane phospholipids, general anesthetics, and the neuroprotective agent riluzole (15)(16)(17)(18)(19). These leak-type channels also show variable sensitivity to standard K ϩ channel blockers (15)(16)(17). At least one member of this family is reported to acquire voltage-dependent gating upon phosphorylation (20).
Although I AC functions in AZF cells as a classic leak-type K ϩ channel, its distinctive properties are not duplicated in any of the 2P/4TMS K ϩ channels yet described. In particular, direct modulation of K ϩ channels by ATP has only been demonstrated in K ATP channels of the 1P/2TMS type, which are inhibited rather than activated by this nucleotide (13,14).
We have cloned a 2P/4TMS channel TREK-1 (KCNK2) K ϩ channel cDNA from AZF cells that appears to code for I AC K ϩ channels. When expressed in cell lines, bTREK-1 currents share several distinctive features with native I AC currents. Evidence is presented that indicates that, in AZF cells, bTREK-1 channels acquire additional properties including ATP sensitivity.

Methods
Isolation and Culture of AZF Cells-Bovine adrenal glands were obtained from steers (age range 1 to 3 years) within 1 h of slaughter at a local slaughterhouse. Fatty tissue was removed immediately, and the glands were transported to the laboratory in ice-cold phosphate-buffered saline containing 0.2% dextrose. Isolated AZF cells were prepared as described previously (4). After isolation, cells were either resuspended in Dulbecco's modified Eagle's medium/F12 (1:1) with 10% fetal bovine serum, 100 units/ml penicillin, 0.1 mg/ml streptomycin and the antioxidants 1 M tocopherol, 20 nM selenite and 100 M ascorbic acid (Dulbecco's modified Eagle's medium/F12ϩ) and plated for immediate use or resuspended in fetal bovine serum/5% Me 2 SO, divided into 1-ml aliquots each containing about 2 ϫ 10 6 cells, and stored in liquid nitrogen for future use. Cells were plated in either 60-or 35-mm dishes containing 9-mm 2 glass coverslips. Dishes or coverslips were treated with fibronectin (10 g/ml) at 37°C for 30 min then rinsed with warm, sterile phosphate-buffered saline immediately before adding cells. Cells were plated in Dulbecco's modified Eagle's medium/F12ϩ and were maintained at 37°C in a humidified atmosphere of 95% air and 5% CO 2 .
Cloning of the b-TREK-1 K ϩ Channel cDNA-RNeasy columns (Qiagen, Valencia, CA) that had been treated with RNase-free DNase (Qiagen) to remove genomic contamination were used to extract total RNA from AZF cells that had been cultured overnight. poly(A) ϩ mRNA was extracted from total RNA using a Poly(A) Pure kit (Ambion, Austin, TX). Reverse transcription and cDNA amplification were performed using a Marathon cDNA amplification kit (Clontech, Palo Alto, CA). The resulting AZF double-stranded cDNA was used as the template for all PCR and RACE PCR reactions. RACE and PCR reactions were run as per Clontech instructions on a Thermolyne Temptronic thermal cycler.
Noting the sequence similarities among double-pore channels, particularly at the transmembrane and pore regions, degenerate primers were designed using human TREK-1 (AF171068), and human TRAAK (AF247042) pore (P1) and transmembrane (M2) sequences as templates. For an initial 3ЈRACE reaction, a degenerate primer designed from P1 5Ј-CTG GGA CCG TCA TCA CVA CCA TMG G-3Ј) was used as the forward primer, and the Clontech adaptor primer 1 was used as the reverse primer. Using the product of this initial reaction, nested 3ЈRACE was performed using a degenerate primer from the M2 region (5Ј-TGG GGA TTC CGC TGT TTG GKW TYY T-3Ј) as the forward primer, and the Clontech adaptor primer 2 was used as the reverse primer. The 3Ј nested RACE product was separated on a 1.2% agarose gel and visualized with ethidium bromide/UV transillumination. A 750-bp 3Ј RACE product was excised from the gel and cloned into pCR-TOPO vector (Invitrogen). Clones were sequenced using an Applied Systems sequencer, model 3700, version 3.6.1 software. Using this initial 750-bp product to design specific nested primers for 5Ј RACE PCR, the 5Ј end of bTREK-1 was obtained. For transfection, the fulllength bTREK-1 clone was produced by PCR using a low error rate DNA polymerase (Pfu DNA polymerase; Invitrogen), bTREK-1-specific primers 5Ј-/5phos/GCG ATT CCG TGT CTT CTC-3Ј as the forward primer and 5Ј-AAG GTC TAA TTG CTA TGC CTG AG-3Ј as the reverse primer, and AZF double-stranded cDNA as the template. The PCR product obtained was subcloned into the pCR3.1-Uni expression vector (Invitrogen) for transfection into HEK-293 cells or CHO-K1.
Measurement of bTREK-1 mRNA-7 g of poly(A) ϩ mRNA, isolated as described above, was separated on a denaturing 8% formaldehyde, 1.0% agarose gel, and transferred to a nylon membrane (Gene Screen Plus; PerkinElmer Life Sciences). The RNA was fixed to the membrane by UV-cross-linking using a Stratalinker (Stratagene, La Jolla, CA). Northern blots were prehybridized in heat-sealable plastic bags for 2 h at 42°C in ULTRAhyb (Ambion, Austin, TX) for probing using either a 1415-bp fragment that contained the entire coding region of bTREK-1 or a 206-bp BamHI fragment primarily from the bTREK-1 5Ј-untranslated end. The blots were hybridized with a random-primed 32 P-␣-dCTP radiolabeled probe (Prime-it-II; Stratagene, La Jolla, CA) overnight in minimal volume of hybridization solution. After 18 -24 h the blots were washed twice at room temperature in 2ϫ saline/sodium phosphate/ EDTA for 15 min, twice at 40°C in 2ϫ saline/sodium phosphate/EDTA, 1% SDS for 30 min with a final wash at 65°C with 0.1 ϫ saline/sodium phosphate/EDTA, 1% SDS for 20 min. Autoradiograms were obtained by exposing the blots for 2 to 5 h to Eastman Kodak Co. X-Omat AR film at Ϫ70°C.
AZF cells were used for patch clamp experiments 2-12 h after plating. Typically, cells with diameters of Ͻ15 m and capacitances of 8 -15 picofarads were selected. Coverslips were transferred from 35-mm culture dishes to the recording chamber (volume, 1.5 ml), which was perfused continuously by gravity at a rate of 3-5 ml/min. Drugs were applied externally by bath perfusion controlled manually by a six-way rotary valve. Patch electrodes with resistances of 2.0 -5.0 megohms were fabricated from Corning 0010 glass (WPI, Sarasota, FL). K ϩ currents were recorded at room temperature (20 -23°C) following the procedure of Hamill et al. (22), using an Axopatch 1D patch clamp amplifier (Axon Instruments, Inc., Union City, CA) or List EPC 7 patch clamp amplifier.
HEK-293 or CHO-K1 cells were used for patch clamping 24 -72 h after transfection with bTREK-1. Transfected cells were plated on 9-mm glass coverslips as described above. Fifteen min before initiating a patch clamp experiment anti-CD8 antibody-coated beads were added to the culture dish. Upon transferring coverslips to the recording chamber, transfected cells were identified based on decoration with the beads. Whole cell bTREK-1 K ϩ currents in transfected cells were recorded as described for AZF cells. Smaller cells with capacitances of 10 -15 pF were selected for recording.
Pulse generation and data acquisition were done using a personal computer and PCLAMP software with a TL-1 interface (Axon Instruments, Inc., Union City, CA). Currents were digitized using an 8-pole Bessel filter (Frequency Devices, Haverhill, MA) Linear leak and capacity currents were subtracted from current records using hyperpolarizing steps of 1 ⁄3 to 1 ⁄4 amplitude. Data were analyzed and plotted using PCLAMP 5.5 and 6.04 (Clampan and Clampfit) and SigmaPlot (version 5.0).

RESULTS
bTREK-1 in AZF Cells-Previous studies using whole cell and single channel patch clamp recording showed that I AC channels resembled 2P/4TMS channels of the TREK-1 and TRAAK with regard to single channel conductance and rectification (5,16). Degenerate primers were designed using the P1 and M2 regions of hTREK-1 and hTRAAK as templates, and 3ЈRACE PCR was performed using AZF cell double-stranded cDNA generated from poly(A) ϩ mRNA. A 750-bp product was obtained that showed 92% nucleotide sequence identity to human TREK-1. No PCR products were obtained that were homologous to hTRAAK. Specific primers were designed from this TREK-1-like product for extension in the 5Ј direction using nested 5Ј RACE PCR. Combining the 3Ј and 5Ј RACE products yielded a 1534-nt cDNA (GenBank TM accession number AY148474). This coding sequence consists of an open reading frame of 411 amino acids. The predicted amino acid sequence for bTREK-1 is given in Fig. 1, with the four transmembrane and two pore sequences noted. The bovine sequence AAN37591-1 shows 99% protein sequence identity to human TREK-1 (protein ID, NP_055032.1), 96% identity to rat TREK-1 (protein ID, AAL95708), and 96.4% to mouse TREK-1 (Protein ID, XP_123605).
Northern blot analysis of poly(A) ϩ mRNA showed that bTREK-1 mRNA is expressed highly in the bovine adrenal cortex. A 1415-bp DNA probe that included the full-length coding region of bTREK-1 hybridized to separate ϳ4.9-, 3.6-, and 2.8-kb mRNA transcripts isolated from AZF cells (Fig. 1B, left lane). To eliminate the possibility that these transcripts could be the result of the full-length TREK-1 probe hybridizing to mRNA coding for homologous channels present in the adrenal cortex, a TREK-1-specific probe consisting of a 206-bp BamHI fragment derived primarily from the 5Јuntranslated end of bTREK-1 was used to probe a separate poly(A) ϩ blot. The 206-bp fragment also hybridized to the same three transcripts (Fig. 1B, right lane), with the strongest hybridization signal originating from the 3.6-and 2.8-kb transcripts.
Properties of bTREK-1 Currents-Properties of bTREK-1 currents were studied in transiently transfected HEK-293 cells. Successfully transfected cells were identified based on decoration with anti-CD-8 antibody-coated beads, as described under "Methods." In whole cell patch clamp recording, bTREK-1 was remarkably similar in several respects to the native I AC current in bovine AZF cells. Specifically, in physiological saline bTREK-1 appeared as a non-inactivating current that increased spontaneously in amplitude to a maximum value over many minutes of whole cell recording ( Fig. 2A).
In the experiment illustrated in Fig. 2A, during 17 min of recording, bTREK-1 amplitude doubled from its initial value of ϳ1250 pA. This spontaneous time-dependent increase in the amplitude of bTREK-1 current was observed in nearly all of the more than 75 cells tested. It occurred at pipette solution pHs ranging from 6.4 to 7.3 and was independent of ATP concentrations between 0.1 and 5.0 mM. Further, this phenomenon was not specific to HEK-293 cells. The growth of bTREK-1 current with time was also observed in transfected CHO-K1 cells (data not shown). Although the time-dependent growth of I AC in whole cell recordings has been established as a hallmark of this current in AZF cells (1,3,4), this phenomenon has not been reported previously for other native or cloned TREK-1 channels.
Similar to I AC , when bTREK-1 is activated from a holding potential of Ϫ80 mV to various test potentials, this current consists of a large instantaneous component and a smaller Either full-length bTREK-1 cDNA (left) or a 206-bp BamHI fragment from the 5Јuntranslated end of bTREK-1 (right) labeled with [ 32 P]␣-dCTP was hybridized to size-fractionated poly(A) ϩ mRNA (7 g) extracted from AZF cells. Northern blot analysis was carried out as described under "Methods." Bands correspond to transcripts of ϳ4.9, 3.6, and 2.8 kb.
time-dependent component that could be fit with a single exponential ( Fig. 2, A (left) and B). In physiological saline, bTREK-1 is outwardly rectifying (Fig. 2B).
Comparative Pharmacology of I AC and Cloned bTREK-1-If native I AC and cloned bTREK-1 K ϩ channels are identical then they should display a similar pharmacological profile. The pharmacology of 2P/4TMS K ϩ channels, including TREK-1, is currently being investigated. Murine and human TREK-1 channels are relatively insensitive to standard organic K ϩ channel antagonists, including TEA and 4-AP (23,24). High concentrations of these two drugs are also required to inhibit native I AC channels (10). However, I AC channels are inhibited potently by several organic Ca 2ϩ antagonists, including diphenylbutylpiperidines (DPBPs) and mibefradil (10, 11).

FIG. 3. Penfluridol blocks bTREK-1 in transfected cells.
K ϩ currents were recorded from bTREK-1-transfected HEK-293 cells in response to voltage steps to ϩ20 mV applied at 30-s intervals from a holding potential of Ϫ80 mV. When bTREK-1 reached a maximum value, cells were superfused with penfluridol at concentrations between 0.1 and 2.5 M. A, current traces and associated plot of bTREK-1 current amplitudes for a cell superfused with 2.5 M penfluridol. Numbers on traces correspond to those on graph at right. B, penfluridol on current-voltage relationship. bTREK-1 K ϩ currents were recorded from transiently transfected HEK-293 cells in response to voltage steps applied from Ϫ80 mV to test potentials between Ϫ50 and ϩ40 mV, before and 6 min after superfusing the cell with 2.5 M penfluridol. Current amplitudes are plotted against test potential. C, inhibition curve for penfluridol constructed from experiments as in A. Fraction of unblocked bTREK-1 current is plotted against penfluridol concentration. Data were fit with an equation of the form, I/I max ϭ 1/[1 ϩ B/IC 50 )] X , where B is the penfluridol concentration, IC 50 is the concentration that produced half-maximal inhibition, and X is the Hill slope. The IC 50 estimated from the fit is 3.5 ϫ 10 Ϫ7 M. Data are normalized mean values Ϯ S.E. of the indicated number of measurements.
FIG. 2. Properties of bTREK-1 K ؉ current in HEK-293 cells. Whole cell K ϩ currents were recorded from HEK-293 cells that had been transiently co-transfected with pCR ® 3.1-Uni-bTREK-1 and P3-CD8 cDNAs. Cells decorated with anti-CD8 antibody-coated beads were selected for patch clamping (see "Methods"). Mock-transfected cells were transfected with the expression plasmid pCR ® 3.1-Uni (no insert) and P3-CD8 cell surface antigen. A, time-dependent increase in bTREK-1 currents. K ϩ currents were activated by voltage steps to ϩ20 mV applied at 30-s intervals from a holding potential of Ϫ80 mV. Traces and associated graph illustrate currents recorded from a single cell at indicated times after initiating whole cell recording. B, bTREK-1 current-voltage relationship. Whole cell K ϩ currents were recorded from bTREK-1transfected HEK-293 and control cells in response to voltage steps applied from Ϫ80 mV in 10-mV increments to test potentials between Ϫ70 and ϩ60 mV. Current traces from bTREK-1-transfected and control cells and associated I-V plots are shown.
In the present study, we found that the DPBP penfluridol and mibefradil also potently inhibit cloned bTREK-1 channels expressed in mammalian cell lines. DPBPs such as penfluridol inhibit I AC K ϩ channels with IC 50 values of ϳ0.2-0.4 M (10). Penfluridol inhibited bTREK-1 expressed in HEK-293 cells with an IC 50 of 0.35 M (Fig. 3A and C). Inhibition of bTREK-1 by penfluridol was independent of voltage at test potentials of Ϫ40 to ϩ40 mV (Fig. 3B). bTREK-1 was also inhibited by a second DPBP, pimozide (data not shown).
Cloned TREK-1 channels and bovine I AC channels are inhibited by cAMP (3,23). The membrane-permeable cAMP analog 8-pcpt-cAMP inhibited native I AC and cloned bTREK-1 channels with equal effectiveness (Fig. 4, D and E).
Unitary bTREK-1 and I AC Currents-Unitary bTREK-1 currents were recorded from transfected CHO-K1 cell membrane patches in the inside-out configuration and compared with unitary I AC currents recorded from AZF cells under the same conditions. Amplitude analysis of single channel bTREK-1 and native I AC currents showed that these were nearly identical in size (Fig. 5, A and B). In the experiment illustrated unitary bTREK-1 and I AC currents measured at ϩ30 mV had amplitudes of 3.57 and 3.60 pA, respectively. Solutions in contact with the cytoplasmic membrane face in these recordings contained 0.1 mM ATP. In excised inside-out patches activity of I AC and bTREK-1 channels were both independent of ATP.
Activation of I AC Current in AZF Cells by Intracellular Acidification-When expressed in HEK-293 cells, bTREK-1 current displayed several distinctive properties of the native I AC of AZF cells. Experiments were done to determine whether I AC , in turn, possessed well established properties of TREK-1 channels cloned previously. These include activation by acidification and the neuroprotective agent riluzole (18,25).
Extensive patch clamp and molecular cloning studies have identified only two types of K ϩ channels in bovine AZF cells, a rapidly inactivating, voltage-gated Kv1.4-type current and the non-inactivating I AC current (1,4,26,27). Although present at low density upon initiating whole cell recordings, I AC often grows dramatically over a period of minutes provided that ATP is present at millimolar concentrations in the pipette solution (1,4,5).
The absence of time-and voltage-dependent inactivation of the I AC current allows it to be isolated and measured in whole cell recordings, using either of two voltage-clamp protocols. When voltage steps of 300 ms duration are applied from a holding potential of Ϫ80 mV, I AC can be measured near the end of a voltage step when the transient current has inactivated (Fig. 6, left traces). Alternatively, I AC can be activated selectively after a 10-s prepulse to Ϫ20 mV has fully inactivated the Kv1.4-type current (Fig. 6A, right traces).
Of the 2P/4TMS K ϩ channels described thus far, TREK-1 and TREK-2 channels (KCNK2 and KCNK10) are unique in the pronounced increase in activity observed in response to lowering intracellular pH (25,28). To determine whether the activity of the ATP-dependent I AC current was enhanced by cytoplasmic acidification, K ϩ currents were recorded with pipette solution that was supplemented with 5 mM MgATP at pH levels between 6.2 and 7.5.
As shown in Fig. 6, the time-dependent development of the non-inactivating K ϩ current was enhanced markedly and selectively at successively lower pH levels. Overall, I AC current density varied from a minimum of 14.6 Ϯ 1.7 pA/pF (n ϭ 15) at pH 7.4 to 118.3 Ϯ 9.2 pA/pF (n ϭ 8) at pH 6.2 (Fig. 6, A and B).
Cooperative Activation of I AC by pH and ATP-The magnitude of I AC K ϩ current recorded in the presence of 5 mM ATP was enhanced markedly at low pH levels. Conversely, the enhancement of I AC by acidification was found to depend on the concentration of ATP in the pipette solutions.
The cooperative actions of ATP and pH on I AC expression are FIG. 6. pH-dependent activation of I AC K ؉ current. Whole cell K ϩ currents were recorded at 30-s intervals in response to voltage steps to ϩ20 mV from a holding potential of Ϫ80 mV using patch pipettes containing standard solutions at pH levels between 6.2 and 7.5. A, time-dependent expression of I AC current at intracellular pH levels of 6.2 and 7.4. Current traces recorded with (right) and without (left) depolarizing prepulses immediately after initiating whole cell recording (1) and after I AC had reached a stable maximum amplitude (2). I AC amplitude with (E) and without (q) depolarizing prepulses are plotted against time at right. B, maximum average I AC current densities (expressed as pA/pF) were determined from experiments as in A at seven different pH levels and plotted against pH. Values are mean Ϯ S.E. of indicated number of determinations.

FIG. 5. Comparison of unitary native I AC and cloned bTREK-1 currents.
Unitary I AC and bTREK-1 currents were recorded in excised inside-out patches from AZF cells and bTREK-1transfected CHO-K1 cells, respectively. Patches were excised into standard internal solutions used in whole cell recording, supplemented with 0.1 mM MgATP (see "Methods"). Voltage steps of 300-ms duration to ϩ30 mV were applied at 10-s intervals from a holding potential of Ϫ40 mV. Amplitude histograms were constructed from idealized unitary currents that were distributed into bins 0.180 pA in width. A and B, unitary currents (top) and corresponding amplitude histograms (bottom) from AZF cells (A) and bTREK-1-transfected CHO-K1 cells (B). Currents were filtered at a cutoff frequency of 2 kHz and sampled at a frequency of 5 kHz. Continuous lines in the histograms are fits of Gaussian distributions to the data. illustrated in Fig. 7. When I AC K ϩ currents were recorded at pH 6.5 (Fig. 7A) or pH 7.0 (Fig. 7B), the time-dependent increase in I AC was ϳ4-fold greater in the presence of 5.0 mM ATP compared with 0.5 mM ATP. Overall, with 0.5 ATP in the pipette lowering pH from 7.5 to 7.0 and 6.5 increased maximum I AC current density from a value of 11.7 Ϯ 3.4 pA/pF (n ϭ 5) to 13.1 Ϯ 3.0 pA/pF (n ϭ 4) and 40.4 pA/pF (n ϭ 5), respectively. By comparison, with pipettes containing 5.0 mM ATP, reducing pH from 7.5 to 7.0 and 6.5 increased maximum I AC current density from 15.0 Ϯ 3.1 (n ϭ 7) to 63.5 Ϯ 8.4 pA/pF (n ϭ 5) and 114 Ϯ 9.8 pA/pF (n ϭ 6), respectively (Fig. 7C).
Effects of Penfluridol, ACTH, and Riluzole on pH-activated K ϩ Current-The cooperative enhancement of the background K ϩ current of AZF cells by acidity and ATP suggest that both agents act on the same channels. They suggest further that I AC is a TREK-1 type K ϩ channel that acquires ATP sensitivity when expressed in its native cell type.
If the pH-activated K ϩ current and I AC are one and the same, selective I AC antagonists such as penfluridol should inhibit the pH-activated current with similar potency. Accordingly, penfluridol (2.5 M) inhibited the non-inactivating K ϩ current recorded at pH 6.4 by 85 Ϯ 5% (n ϭ 5) (Fig. 8A).
ACTH inhibits I AC current in AZF cells at subnanomolar concentrations (1). This peptide hormone also inhibited the acid-stimulated K ϩ current potently in these cells. In four experiments, ACTH (200 pM) completely inhibited the noninactivating K ϩ current activated at pH 6.4 with pipette solutions containing ATP at either 0.5 or 5.0 mM (Fig. 8B). In addition to low pH, cloned human and rat TREK-1 K ϩ channels are activated by the neuroprotective agent riluzole (2-amino-6-(trifluoromethoxy) benzothrazole) (18). However, the riluzole-induced increase in K ϩ current through cloned TREK-1 channels was reported to be transient and followed by inhibition, in response to a secondary rise in intracellular cAMP (18).
If the pH-activated K ϩ current of AZF cells is because of TREK-1 channels, then riluzole should also modulate this current. In the experiment illustrated in Fig. 8C, whole cell currents were recorded at pH 6.4 with 0.5 mM ATP in the pipette. After the pH-dependent current had reached maximum amplitude, AZF cells were superfused with saline containing 100 M riluzole, whereas continuing to monitor currents at 30-s intervals.
Riluzole produced a pronounced increase in the non-inactivating K ϩ current, which reached a maximum within several minutes and then declined toward the control value. In a total FIG. 7. ATP facilitates pH-dependent expression of I AC . Whole cell K ϩ currents were recorded from AZF cells using the voltage protocols described in the legend of Fig. 4 with pipettes containing standard solution supplemented with either 0.5 or 5 mM MgATP and with pH levels adjusted to 6.5, 7.0, or 7.5. A and B, K ϩ currents recorded at pH 6.5 (A) or pH 7.5 (B) with pipettes containing 0.5 or 5 mM MgATP. Numbered traces recorded with (left) or without (right) depolarizing prepulses were recorded at times indicated on corresponding graph. I AC amplitudes are plotted against time at right. Open and closed symbols indicated currents recorded with and without depolarizing prepulses, respectively. C, maximum I AC current densities expressed as pA/pF were determined from experiments such as those shown in A and B and plotted against pH at two ATP concentrations. Values are mean Ϯ S.E. for the indicated number of determinations. of eight cells, riluzole increased the non-inactivating current to 331 Ϯ 59% of its control value. In six of these cells, the noninactivating K ϩ current declined to 71.9 Ϯ 6.6% of its maximum value after 10 min in the continued presence of riluzole. In the remaining two cells, the riluzole-induced increase was sustained, with no transient component. The riluzole-induced increase in non-inactivating K ϩ current is consistent with the notion that the drug acts on native TREK-1-type K ϩ channels. The variation in the kinetics of the response likely reflects differences in the extent of dilution of intracellular components (see below). The inhibition of the acid-stimulated current by penfluridol and ACTH, and its enhancement by riluzole are consistent with the hypothesis that I AC is a pH and ATP-dependent bTREK-1 current expressed in AZF cells.
Rectifying Properties of Native and Transfected bTREK-1 Channels-If the ATP-and pH-dependent background K ϩ currents of AZF cells are because of K ϩ flux through bTREK-1 channels, then each of these currents should display similar voltage-dependent rectification. The rectifying properties of non-inactivating K ϩ currents in AZF cells, activated by acidification of pipette solution to pH 6.4 or by 5 mM ATP were compared with those of bTREK-1 currents in transfected HEK-293 cells using ramp voltage protocols. Fig. 9A shows noninactivating K ϩ currents recorded in response to 2-s linear ramp voltages, applied between ϩ60 and Ϫ140 mV. Each of the three currents displayed strong outward rectification, crossing the voltage axis near the Nerst equilibrium potential for K ϩ . Scaling of the three currents using factors derived from the maximum current amplitudes measured at ϩ60 mV resulted in three nearly superimposable traces (Fig. 9B). This result suggests that these currents result from K ϩ flow through identical channels.
Enhancement of ATP-dependent I AC by Riluzole-If the ATPdependent I AC K ϩ channels are of the TREK-1 type, then riluzole should increase the amplitude of I AC in AZF cells. In the experiments illustrated in Fig. 10A, whole cell recordings were made with pipettes containing 5 mM ATP at pH 7.1. When I AC reached a stable maximum amplitude, cells were superfused with riluzole at concentrations between 1 and 100 M.
Although riluzole (50 -100 M) induced rapid 2-to 8-fold increases in the amplitude of the non-inactivating K ϩ current in each of 33 cells tested, the temporal pattern of the responses varied widely (Fig. 10A). In the majority of cells (21/33) the riluzole-induced increase was characterized by distinct transient and sustained components (Fig. 10A, middle panel). In four cells, the transient increase in current was followed by near complete inhibition of the non-inactivating current (Fig.  10A, left panel). In these cells, a second application of riluzole, after washing with control saline, failed to produce any increase.
When riluzole was applied to the remaining 12 cells, it induced a sustained increase in current with no detectable decrement during at least 10 min of exposure (Fig. 10A, right  panel). In these cells, a second application of riluzole was again effective at enhancing the non-inactivating K ϩ current.
The varying temporal pattern of the riluzole-induced responses was correlated with the measured series resistance of the recording pipette. In the cells where riluzole-induced increase in current was sustained, series resistance averaged 3.6 Ϯ 0.3 megohms (n ϭ 12). By comparison, in experiments where the riluzole response displayed a distinct transient component, averaged series resistance was more than twice as great (8.6 Ϯ 0.7 megohms) (n ϭ 21). These results suggest that the inhibitory component of the riluzole response is dependent on the continued presence of cytoplasmic contents that are lost more rapidly when currents are recorded with low resistance pipettes.
The varying transient and sustained components of the riluzole-induced increases in non-inactivating current could suggest that this drug enhances the activity of two different K ϩ channels in AZF cells. Several lines of evidence indicate that this is not the case and that riluzole enhances current flow only through pre-activated I AC channels.
Ramp voltage protocols applied before and after exposing AZF cells to riluzole showed that the transient and sustained components of non-inactivating K ϩ current induced by this drug were indistinguishable from I AC . In Fig. 10C (left), ramp voltage protocols were applied after the K ϩ current reached a maximum value in control saline (control), immediately after the current reached a maximum value in riluzole (Riluzole max) and after a steady state value was reached (Riluzole SS).
Scaling of the three currents using factors derived from the maximum current amplitudes measured at ϩ60 mV and plotting these on a separate set of axis yielded three current traces that were nearly superimposable (Fig. 10C, right traces). Thus, the transient and sustained components of the riluzole-induced current had rectifying properties identical to I AC .
Cooperative Activation of I AC by Riluzole and ATP-The riluzole-stimulated increases in non-inactivating K ϩ current were observed to be proportional to the amplitude of the preexisting I AC current present in individual cells. This suggested that riluzole-enhanced current flow through I AC channels that had been pre-activated by ATP. To test this hypothesis, we compared the effects of riluzole in experiments with pipettes containing ATP at several concentrations.
In whole cell recordings made with pipette solutions containing ATP at concentrations of 1 mM or less, I AC was small and failed to grow significantly with time. Further, under these conditions, riluzole (100 M) was much less effective at increasing the amplitude of the non-inactivating K ϩ current (Fig. 11,  A (left traces) and B).
By comparison, when recordings were made with pipettes containing 5 mM ATP, I AC currents reached much larger maximum values, whereas riluzole produced far greater absolute increases in a non-inactivating component of K ϩ current that was not distinguishable from I AC (Fig. 11, A (middle traces) and  B).
The non-hydrolyzable ATP analog AMP-PNP is more potent than ATP as an activator of I AC K ϩ channels (5) (4). Accordingly, with 1 mM AMP-PNP in the recording pipette, I AC reached current densities severalfold greater than those observed with 1 mM ATP. Further, in these experiments, riluzole again produced much larger increases in non-inactivating K ϩ current in the presence of 1 mM AMP-PNP, compared with 1 mM ATP (Fig. 11, A (right panel) and B).
In addition to nucleotide triphosphates, PPP i activates I AC effectively in whole cell recordings (5). With pipettes containing 1 mM ATP and 5 mM PPPi, I AC current density was severalfold larger than that observed with ATP alone. Accordingly, riluzole produced 6-fold larger absolute increases in non-inactivating K ϩ current in the presence of PPPi and ATP, compared with ATP alone (Fig. 11B). Overall, these results are consistent with the hypothesis that riluzole enhances only the activity of I AC K ϩ channels that have been pre-activated by ATP and PPP i .
Inhibition of Riluzole-activated K ϩ Current by ACTH and cAMP-Results demonstrating cooperative effects of ATP, PPP i , and riluzole on expression of background K ϩ current in AZF cells suggest that these agents all enhance only the activity of I AC channels and that these are bTREK-1 channels. However, the possibility that multiple K ϩ channels are involved has not been excluded. For example, TRAAK-type K ϩ channels exhibit some properties of I AC , including activation by riluzole (18,29). In this regard, unlike native I AC and cloned TREK-1 channels, TRAAK channels are not inhibited by cAMP (29). If riluzole activates TRAAK in AZF cells, this current would be resistant to inhibition by ACTH and cAMP.
ACTH was found to inhibit all non-inactivating K ϩ currents completely in AZF cells, including that induced by riluzole. In the experiment illustrated in Fig. 12A, I AC reached a stable maximum value before the cell was superfused with 100 M riluzole, which produced a 3-fold enhancement of the current. Superfusion of the cell with ACTH (200 pM) in the continued presence of riluzole produced complete inhibition of the noninactivating current. Nearly identical results were obtained in each of four experiments.
In other experiments, I AC was allowed to grow to a stable maximum amplitude before sequentially perfusing 200 pM ACTH and 100 M riluzole, respectively. After complete inhibition of I AC by ACTH, riluzole was totally ineffective in activating any current (Fig. 12B). Similar findings were obtained in three other experiments.
cAMP was equally effective as ACTH in suppressing riluzoleinduced K ϩ currents. In the experiment illustrated in Fig. 12C, I AC was permitted to reach a maximum amplitude before sequentially superfusing a membrane-permeable cAMP analog (8-pcpt-cAMP; 500 M), followed by riluzole (100 M). After complete inhibition of I AC current by 8-pcpt-cAMP, riluzole failed to produce any increase in K ϩ current.
When 8-pcpt-cAMP was applied intracellularly through the recording pipette, I AC was suppressed, and riluzole was again ineffective (Fig. 12D). In a total of six similar experiments, riluzole increased I AC from a control value of 53.6 Ϯ 7.6 pA (n ϭ 6) to a final value of only 68.8 Ϯ 6.1 pA. These experiments demonstrate that the riluzole activated K ϩ channel is identical to I AC with respect to sensitivity to inhibition by cAMP and ACTH.
Effect of Penfluridol on Riluzole-activated Current-Penfluridol is a potent antagonist of both native I AC and cloned bTREK-1 currents. If the riluzole-activated current in AZF cells is composed only of I AC , then it should be inhibited by this drug with similar potency. In the experiment shown in Fig.  13A, I AC was allowed to approach a maximum value (trace 2) before superfusing with saline containing riluzole (100 M), which produced transient (trace 3) and sustained (trace 4) increases in the non-inactivating current. Subsequent superfusion of saline containing riluzole and penfluridol (2.5 M) resulted in near complete inhibition of the remaining non-inactivating K ϩ current (trace 5). Overall, in four similar experiments, penfluridol inhibited the sustained K ϩ current by 92.6 Ϯ 5.8%.
Experiments such as those illustrated in Fig. 13A did not eliminate the possibility that the transiently activated compo- Left traces, linear voltage ramps of 100 mV/s were applied from a holding potential of 0 mV to potentials between ϩ60 and Ϫ140 mV after I AC reached a maximum in control saline (control), after a maximum increase was reached in response to riluzole (Riluzole (max)), and after a steady increase in riluzole was achieved (Riluzole (SS)). Right traces, traces at left were scaled against the riluzole (max) trace, using the maximum value of each trace measured at ϩ60 mV to determine the appropriate scaling factors.
FIG. 11. ATP-and PPP i -dependent activation of I AC by riluzole. Whole cell K ϩ currents were recorded from AZF cells at 30-s intervals in response to voltage steps to ϩ20 mV with or without depolarizing prepulses, using patch pipettes containing MgATP (0.1, 0.5, or 5.0 mM), AMP-PNP (1 mM), or 5 mM PPP i plus 1 mM MgATP. When I AC reached a maximum amplitude, the cell was superfused with 100 M riluzole. A, K ϩ current records made with pipettes containing MgATP or AMP-PNP at the indicated concentrations. Numbers on traces recorded with (bottom) and without (top) depolarizing prepulses correspond to currents recorded immediately after initiating whole cell recording (1), after I AC reached a maximum value in control saline (2), and after the maximum current in the presence of riluzole (3). B, the effect of riluzole in combination with ATP, AMP-PNP, and PPP i on I AC current density was determined from experiments as in A above. I AC current density, expressed as pA/pF, was determined by dividing the maximum I AC current amplitude for each condition by the cell capacitance. Values are mean Ϯ S.E. of the indicated number of determinations. nent of non-inactivating current induced by riluzole was a separate penfluridol-insensitive K ϩ current. This possibility was excluded by experiments like that illustrated in Fig. 13B in which I AC was permitted to reach a maximum amplitude (trace 2) before subsequently superfusing the cell with saline containing penfluridol (2.5 M), followed by one containing both penfluridol (2.5 M) and riluzole (100 M). Penfluridol inhibited the I AC current completely (trace 3) and nearly eliminated any riluzole-induced increase (trace 4). In this experiment, the previously reported spontaneous time-dependent decrease in Kv1.4 current that accompanies the growth of I AC is evident after treatment with penfluridol (trace 3) (30). At this concentration, penfluridol blocks I AC selectively. Similar results were obtained in each of three experiments. DISCUSSION In this study, a b-TREK-1 K ϩ channel cDNA was cloned and shown to be expressed robustly in bovine AZF cells. When bTREK-1 cDNA was transiently expressed in HEK-293 or CHO-K1 cells, the corresponding membrane current displayed prominent characteristics of the native I AC background K ϩ current of AZF cells. Conversely, I AC displayed distinctive properties of cloned bTREK-1 channels (Table I). Overall, these results provide convincing evidence that I AC is a bTREK-1 channel expressed in AZF cells. Apparently, when expressed in AZF cells, these native channels retain properties, such as ATP sensitivity, that are lost when the same channels are expressed in cell lines.
Comparison of bTREK-1 to Other TREK-1 Channels- FIG. 12. Inhibition of Riluzole-activated K ؉ Current by ACTH and cAMP. K ϩ currents were recorded from AZF cells at 30-s intervals in response to voltage steps of increasing size, applied from a holding potential of Ϫ80 mV with (E) or without (q) depolarizing prepulses to Ϫ20 mV. I AC amplitudes are plotted against time. A and B, inhibition of riluzole-activated current by ACTH. A, after I AC reached a maximum in control saline, cell was superfused sequentially with saline containing 100 M riluzole followed by 100 M riluzole plus 200 pM ACTH. B, after I AC reached a maximum amplitude in control saline, cell was superfused sequentially with saline containing 200 pM ACTH and 200 pM ACTH plus 100 M riluzole. C and D, inhibition of riluzole-activated K ϩ current by cAMP. C, after I AC reached a maximum amplitude, cell was superfused sequentially with saline containing 500 M 8-pcpt-cAMP followed by 500 M 8-pcpt-cAMP and 100 M riluzole. D, K ϩ currents were recorded with a patch pipette containing 100 M cAMP. After 7 min, cell was superfused with saline containing 100 M riluzole as indicated.
FIG. 13. Inhibition of riluzole-activated K ؉ current by penfluridol. K ϩ currents were recorded from AZF cells at 30-s intervals in response to voltage steps to ϩ20 mV from a holding potential of Ϫ80 mV with (E) and without (q) depolarizing prepulses to Ϫ20 mV. Cells were superfused with 100 M riluzole and/or 2.5 M penfluridol as indicated. A, after I AC reached a maximum amplitude in control saline, cell was superfused sequentially with saline containing 100 M riluzole followed by one containing 100 M riluzole and 2.5 M penfluridol. Numbers on traces at left correspond to those on plot of I AC amplitude against time at right. B, after I AC reached a maximum amplitude in control saline, cell was superfused with saline containing 2.5 M penfluridol followed by one containing 2.5 M penfluridol and 100 M riluzole. Numbers on traces at left correspond to those on plot of I AC amplitude against time at right. bTREK-1 codes for a 411-amino acid protein that is 96 and 99% homologous in amino acid sequence to mouse and human TREK-1 channels of the same length (24,31). However, Northern blot analysis of TREK-1 transcripts from several mouse tissues and human brain identified a single 3.8-kb transcript (24,31). In contrast, Northern blot analysis of AZF cell mRNA identified three bTREK-1 transcripts of 4.9, 3.6, and 2.8 kb. This result indicates that several splice variants of bTREK-1 channels may be expressed in bovine AZF cells. Multiple variants of a number of K ϩ channel subtypes are expressed in various tissues (32). Kv1.4 transcripts of 3.4 and 4.4 kb are expressed by bovine AZF cells (27).
Comparison of bTREK-1 Currents to I AC -The time-dependent increase in bTREK-1 current observed in transiently transfected HEK-293 and COS-1 cells had not been observed previously in whole cell recordings of cloned TREK-1 channels but is typical of I AC in AZF cells (1,(3)(4)(5). The molecular mechanism underlying this phenomenon is unknown, but dilution of an endogenous cytoplasmic inhibiting factor common to all three cell types appears likely.
Whole cell bTREK-1 current waveforms resemble I AC currents with respect to kinetics of activation. Both consist of an instantaneous component and a smaller time-dependent fraction that activates with first order kinetics (3). This two-component current has also been described for cloned mouse TREK-1 expressed in a mammalian cell line (31).
The time-dependent component of bTREK-1 and I AC currents observed in response to membrane depolarization are indicative of voltage-dependent gating. In a previous study, we demonstrated a weak voltage-dependent activation of I AC in which open probability increased by 0.3 between test voltages of Ϫ40 and ϩ40 mV (4). Recently, others have shown a voltage dependence activation of native and cloned mouse and rat TREK-1 channels that may or may not depend on the phosphorylation state of the channel (20,33). Regardless, bTREK-1, I AC , and TREK-1 currents from several species display similar kinetics and voltage-dependent gating.
In physiological solutions, cloned bTREK-1 current displayed identical rectification with both the ATP-and pH-activated background K ϩ current recorded from AZF cells. This rectification is typical of that observed for mouse and human TREK-1 channels (24,31).
bTREK-1 channels displayed a pharmacological profile similar to that described previously for I AC in AZF cells (10,11). In particular, the potent inhibition of bTREK-1 and I AC currents by penfluridol and mibefradil establishes an interesting pharmacological similarity between the two channels. It also identifies these two Ca 2ϩ channel blockers as the most potent antagonists of TREK-1 channels yet described. In this respect, penfluridol and mibefradil are at least 1000-to 10,000-fold more potent than K ϩ channel antagonists such as TEA and 4-AP (16,23,31). It will be interesting to determine whether these Ca 2ϩ channel antagonists will also inhibit other tandempore K ϩ channels.
In whole cell recordings from transfected HEK-293 cells, bTREK-1 currents were expressed independently of the concentration of ATP in the pipette solution. This is the single significant property distinguishing these cloned channels from native I AC channels. Direct activation by nucleotides or phosphates has not been reported for any cloned TREK-1 channel. It is possible that cloned TREK-1 channels lose their ATP dependence and sensitivity, because an ATP-binding protein similar to the K ATP channel associated sulfonylurea receptor is missing in the host cell (34). If ATP binding to a TREK-associated inhibitory subunit frees this channel for activation, then absence of this protein in the host cell could explain the ATPindependent activity of bTREK-1 channels expressed in HEK-293. In this regard, we have been unable to demonstrate ATPdependent activation of I AC channels in excised inside-out patches from AZF cells. In fact, on patch excision, channel activity typically increases spontaneously with time, regardless of the ATP concentration at the cytoplasmic membrane surface (5). This result indicates that ATP-mediated modulation of I AC activity requires AZF cell components that are lost after patch excision.
Relatedly, cloned TREK-1 channels are temperature-sensitive and activated dramatically by heat. The thermal activation of TREK-1 is lost on patch excision (35). It appears that the thermal sensitivity of TREK-1 channels also depends on factors present in an intact cell.
Cooperative Activation of I AC by H ϩ and ATP-ATP and pH acted cooperatively in promoting the activity of I AC channels. Expression of the native I AC current in the presence of 5 mM ATP was enhanced strongly at acidic pHs and nearly eliminated by raising the intracellular pH to 7.5 or above. These results suggest that I AC is a 2P/4TMS channel of the TREK variety. Of the tandem pore K ϩ channels cloned thus far, only murine and human TREK-1 and TREK-2 channels are activated by acidification of the cytoplasm (17,25,28).
In this regard, it is unlikely that I AC is TREK-2, because TREK-2 is strongly inwardly rectifying in symmetrical K ϩ solutions and insensitive to 100 M quinidine (28). By comparison, in symmetrical K ϩ solutions, I AC channels are nearly ohmic whereas they are blocked by quinidine with an IC 50 of 26 M (5, 10). Further, Northern blot analysis of AZF cell mRNA using a rat TREK-2 cDNA as a probe indicated that TREK-2 mRNA is not expressed in these cells. 2 The cooperative activation of I AC by pH and ATP was also evident in the ATP-dependent enhancement of I AC current by acidification. With pipettes containing ATP at a concentration of 0.5 mM, successive reductions in pH were far less effective at enhancing I AC than in the presence of 5 mM ATP.
There seems to be little doubt that the non-inactivating K ϩ current activated by ATP and/or acidification is due to a single type of K ϩ channel. In addition to the cooperative actions of ATP and pH described above, currents activated by either agent were inhibited by ACTH and penfluridol equivalently. Further, riluzole enhanced both pH-and ATP-activated K ϩ current in AZF cells. Finally, pH-and ATP-activated currents displayed identical rectification at potentials between ϩ60 and Ϫ140 mV. Enhancement of I AC by Riluzole-Several lines of evidence indicated that the riluzole-activated K ϩ current in AZF cells was I AC and not a second non-inactivating K ϩ current. Specifically, the riluzole-activated current was indistinguishable from I AC with regard to current waveform, voltage-dependent rectification, and inhibition by penfluridol, ACTH, and cAMP.
Further, the activation of non-inactivating K ϩ current by riluzole varied directly with the concentration of ATP in the pipette solution. At low ATP concentrations (Յ1 mM), I AC was poorly expressed, and riluzole induced only very small absolute increases in I AC current. These results are consistent with the hypothesis that riluzole enhances current through I AC channels that have been pre-activated by ATP. The facilitory effects of AMP-PNP and PPP i on riluzole-mediated increases in K ϩ current are also consistent with this model.
The identification of I AC as the riluzole-activated K ϩ current in AZF cells provides further evidence that this current flows through bTREK-1 channels. Of the 2P/4TMS K ϩ channels riluzole has been reported to increase only the activity of TRAAK and TREK-1 (18). Of these two, only TREK-1 type channels are inhibited by cAMP (18,31). Because cAMP and ACTH inhibited riluzole-induced increases completely in the non-inactivating current, it is unlikely that this current was TRAAK and consistent with the hypothesis that it is TREK-1. In addition, TRAAK has been shown to be a neuron-specific channel (15,29).
The riluzole-induced increase in I AC current was similar to that reported for activation of cloned rat TREK-1 channels with respect to concentration dependence but differed with respect to kinetics. The inhibition that followed transient activation of cloned TREK-1 channels has been attributed to a riluzoleinduced increase in cAMP, subsequent to inhibition of phosphodiesterase (18). The temporal pattern of this response, and the inhibition in particular, would depend on the presence of required cytoplasmic components. Within this framework, the absence of the inhibitory component of the riluzole response that we observed with low resistance electrodes would occur through more rapid dilution of cytoplasmic contents by the pipette solution. We believe that the varying temporal pattern of the riluzole response in AZF cells is consistent with the activation of only I AC (bTREK-1)-type channels.
Identity of I AC and bTREK-1 Channels-In summary, we have cloned a bTREK-1 K ϩ channel cDNA that is expressed robustly in bovine AZF cells. When expressed in cell lines, bTREK-1 current displays many of the properties of the endogenous I AC current of AZF cells. These include time-dependent growth in whole cell recordings, instantaneous and time-dependent components in response to depolarization, potent inhibition by penfluridol and mibefradil, similar rectification, and unitary current amplitude.
Native I AC in turn displays essential properties of bTREK-1, as well as those of human, rat, and mouse TREK-1 channels expressed in cell lines (20,24,29). In addition to similarities mentioned previously in single channel conductance, rectification, and inhibition by cAMP, I AC channels have now been shown to be activated by riluzole and intracellular acidification. In other experiments, we have observed that additional agents that activate TREK-1 channels, including unsaturated fatty acids, lysophospholipids, and the volatile anesthetic chloroform also activate I AC current in AZF cells. 3 Overall, convincing evidence indicates that the cloned bTREK-1 channel is the primary subunit of AZF cell I AC channels.
Significance-Earlier studies have shown that TREK-1 channels are expressed primarily in the CNS and in primary sensory neurons (31,35). TREK-1 channels may also be present in heart, kidney, lungs, ovary, and stomach (16,24,31,36). This is the first study demonstrating that TREK-1 channels are expressed by a specific endocrine cell. Further, when combined with previous studies, the present work assigns a pivotal role to bTREK-1 in the physiology of cortisol secretion. Apparently, these channels set the resting potential of bovine AZF cells and couple ACTH and angiotensin II receptor activation to depolarization-dependent Ca 2ϩ entry through T-type channels (1)(2)(3)6). Because I AC channels are also expressed in bovine glomerulosa cells where they are inhibited by ACTH and angiotensin II, it is also likely that bTREK-1 channels have a similar function in regulating aldosterone secretion (6). It will be interesting to determine whether TREK-1 is expressed in other steroid hormonesecreting cells.
Previous studies showed that TREK-1 K ϩ channels set the resting membrane potential and conferred sensitivity of the cell to diverse stimuli, including mechanical force, temperature, pH, fatty acids, phospholipids, and transmitters (25,31,(35)(36)(37). The activation of I AC by ATP and inorganic phosphate expands the role of TREK-1 channels to that of a metabolic sensor that may couple the metabolic state of the cell to membrane potential and electrical activity. In AZF cells, bTREK-1 channels may integrate both hormonal and metabolic signals linking these to depolarization-dependent Ca 2ϩ entry and cortisol secretion.