TWIK-2, an Inactivating 2P Domain K 1 Channel*

We cloned human and rat TWIK-2 and expressed this novel 2P domain K 1 channel in transiently transfected COS cells. TWIK-2 is highly expressed in the gastroin-testinal tract, the vasculature, and the immune system. Rat TWIK-2 currents are about 15 times larger than human TWIK-2 currents, but both exhibit outward rectification in a physiological K 1 gradient and mild inward rectification in symmetrical K 1 conditions. TWIK-2 currents are inactivating at depolarized potentials, and the kinetic of inactivation is highly tempera-ture-sensitive. TWIK-2 shows an extremely low conductance, which prevents the visualization of discrete single channel events. The inactivation and rectification are intrinsic properties of TWIK-2 channels. In a physiological K 1 gradient, TWIK-2 is half inhibited by 0.1 m M Ba 2 1 , quinine, and quinidine. Finally, cysteine 53 in the M1P1 external loop is required for functional expression of TWIK-2 but is not critical for subunit self-assembly. TWIK-2 is the first reported 2P domain K 1 channel that inactivates. The base-line, transient, and delayed activ-ities of TWIK-2 suggest that this novel 2P domain K 1 channel may play an important functional role in cell electrogenesis.

We cloned human and rat TWIK-2 and expressed this novel 2P domain K ؉ channel in transiently transfected COS cells. TWIK-2 is highly expressed in the gastrointestinal tract, the vasculature, and the immune system. Rat TWIK-2 currents are about 15 times larger than human TWIK-2 currents, but both exhibit outward rectification in a physiological K ؉ gradient and mild inward rectification in symmetrical K ؉ conditions. TWIK-2 currents are inactivating at depolarized potentials, and the kinetic of inactivation is highly temperature-sensitive. TWIK-2 shows an extremely low conductance, which prevents the visualization of discrete single channel events. The inactivation and rectification are intrinsic properties of TWIK-2 channels. In a physiological K ؉ gradient, TWIK-2 is half inhibited by 0.1 mM Ba 2؉ , quinine, and quinidine. Finally, cysteine 53 in the M1P1 external loop is required for functional expression of TWIK-2 but is not critical for subunit self-assembly. TWIK-2 is the first reported 2P domain K ؉ channel that inactivates. The base-line, transient, and delayed activities of TWIK-2 suggest that this novel 2P domain K ؉ channel may play an important functional role in cell electrogenesis.
TREK-1 and TRAAK are mechano-gated K ϩ channels that are opened by polyunsaturated fatty acids including arachidonic acid (AA) 1 (17)(18)(19). Additionally, TREK-1 is opened by lysophospholipids, mild intracellular acidosis, and inhalational anesthetics (11,17,19,20). TREK-1 is highly expressed in temperature-sensitive neurons of the preoptic hypothalamus and dorsal root ganglions and moreover is activated by heat (21). Finally, opening of TREK-1 has been proposed to be a key event in the neuroprotective effects of polyunsaturated fatty acids against the deleterious effects of brain ischemia and epilepsy (22). TASK subunits are background K ϩ channels that are inhibited by mild external acidosis (3,7,23). The opening of TASK-1 is stimulated by inhalational anesthetics including halothane and isoflurane (11). TASK-1 has been recently proposed to encode the background K ϩ channel present in motoneurons, cerebellum granular cells, and type I carotid body cells (24 -26). In motoneurons and cerebellar neurons, the background TASK-1-like K ϩ current is reversibly inhibited by the activation of G q -coupled receptors, including the muscarinic receptor (25,26). In the chemoreceptor type I carotid body cells, TASK-1-like K ϩ channels are reversibly inhibited by acidosis and hypoxia (24,27).
TWIK-1, the founding member of the 2P domain mammalian family, is widely expressed in human tissues and is particularly abundant in brain and heart (8). hTWIK-1 currents expressed in Xenopus oocytes are K ϩ -selective, are time-independent, and present a nearly linear I-V relationship that rectifies for depolarizations positive to 0 mV (8). TWIK-1 is blocked by Ba 2ϩ , quinine, and quinidine (8). Recently, hTWIK-2 (also called hTOSS), a TWIK-1-related gene, was cloned by two independent groups (1,9). Although both hTWIK-2 and hTOSS sequences are identical, conflicting results were published concerning functional expression in heterologous systems (1,9). hTWIK-2 expressed in Xenopus oocytes was shown to be a noninactivating, time-independent, weak inward rectifier with biophysical properties identical to TWIK-1 (1). Pharmacologically, hTWIK-2 was reported to be different from TWIK-1 with a lack of sensitivity to quinine, quinidine, and Ba 2ϩ (1). On the contrary, no significant current was observed in hTOSS cRNAinjected Xenopus laevis oocytes or in hTOSS cDNA-transfected HEK293T cells (9). Co-injection of equimolar concentrations of hTWIK-1 and hTOSS cRNA also failed to generate currents in Xenopus oocytes (9). These negative findings have led these authors and others to propose that hTOSS may be targeted to locations other than the plasma membrane or that it may possess a regulatory function, modulating the properties of other principal channel-forming subunits with tissue-specific implications (2,6,9,10).
In the present report, we cloned both human and rat TWIK-2 channels. We demonstrate that TWIK-2 encodes a low conductance inactivating K ϩ channel that is functionally distinct from that previously described for TWIK-1 (8).

EXPERIMENTAL PROCEDURES
Cloning of hTWIK-2, rTWIK-2, and Splice Variants-Two partial expressed sequence tags were identified in the NCBI data base corre-* This work was supported by CNRS and the Association Française contre les Myopathies. 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.
Site-directed Mutagenesis-The polymerase chain reaction (PCR) was used as described previously to generate the following point mutations: cysteine 53 to serine in hTWIK-2 and rTWIK-2, leucine 217 to phenylalanine in hTWIK-2, tyrosine 109 to alanine in hTWIK-2 (19). All PCRs were performed using the Advantage-GC cDNA polymerase mix (CLONTECH) according to the manufacturer's protocol. PCR products were cloned into pCI⅐IRES⅐CD8. The clones obtained in this manner were sequenced in their entirety using an automatic sequencer (Applied Biosystems).
Preparation of Affinity-purified Antibodies-Rabbit polyclonal antibodies were raised against GST fusion proteins containing the M1P1 loop (amino acids 29 -92) and the C-terminal region of TWIK-2 (amino acids 253-313) as described previously (13,21,29). The antibodies were affinity-purified using the GST fusion proteins used for the immunization. Briefly, the crude antisera were preincubated for 4 h at 4°C with 800 g of GST protein previously transferred to Hybond C-extra nitrocellulose membranes (Amersham Pharmacia Biotech), followed by a similar treatment with GST fusion protein strips. After four washes in PBS (0.1% Tween 20), the anti-TWIK-2 antibodies were recovered by a 1-min elution of each strip with 0.1 M glycine, 20 mg/ml bovine serum albumin, pH 2.8. After the elution, the purified antibodies were rapidly brought to pH 7.6 with 1 M Tris (pH 8.5).
Protein Preparation and Immunodetection-COS cells were transfected using Fugene 6 (Roche Molecular Biochemicals) and harvested 48 h later. Cells were washed three times with PBS and scraped at 4°C into PBS supplemented with a mixture of protease inhibitors (Sigma  TWIK-2, an Inactivating 2P Domain K ϩ Channel catalog no. P8340) and 20 mM iodoacetamide. The cell suspension was homogenized with a cell disrupter (Vibra cell 72423, Bioblock Scientific) set at a potency of 5 watts. The insoluble material was removed by centrifugation at 12,000 ϫ g for 15 min at 4°C. 5 g of proteins were separated on 12% SDS-polyacrylamide gel (with or without 2% ␤-mercaptoethanol) and transferred to nitrocellulose membranes (Hybond C-extra; Amersham Pharmacia Biotech). All incubations and washes were carried out in NETG (150 mM NaCl, 5 mM EDTA, 50 mM Tris-Cl, pH 7.4, 0.05% Triton X-100, 0.25% gelatin). The membranes were incubated for 1 h with affinity-purified antibodies (1:700) and washed with NETG. The binding of the primary antibodies was detected by adding goat anti-rabbit horseradish peroxidase-conjugated secondary antibodies (1:25,000) (Jackson) for 1 h at room temperature followed by enhanced chemiluminescence (Super Pierce; Pierce) detection. N-Glycosidase treatment was performed according to the manufacturer's protocol (Roche Molecular Biochemicals).
RNA Expression Analysis-A multiple tissue expression array (MTE CLONTECH) was hybridized with a probe against the first 350 bp of hTWIK-2 according to the manufacturer's protocol. The membrane was then exposed to a phosphor-imaging screen (Fuji) for 1 h. The signals obtained were quantified using TINA version 2.09 g for PC.
Total RNA was isolated from various rat tissues using the SV total RNA kit (Promega). 1 g of total RNA was reverse transcribed using 250 ng of random primers and SuperscriptII (Life Technologies, Inc.) according to the manufacturer's instructions. 2 l of each template was used for the PCR (final volume of 25 l) using Advantage-GC cDNA polymerase mix (CLONTECH) according to the manufacturer's protocol. PCRs for rTWIK-2 (oligonucleotides (5Ј to 3Ј): cggaattccaccatgcggcggggcgcgctc and acgcgtcgacctacctggggatggaggc) were carried out with an initial denaturation of 95°C for 3 min and then 28 cycles of 94°C for 45 s, 55°C for 45 s, and 68°C for 90 s. PCRs for actin (oligonucleotides (5Ј to 3Ј): ttgtaaccaactgggacgatatgg and gatcttgatcttcatggtgctagg) were carried out with an initial denaturation of 95°C for 3 min and then 20 cycles of 94°C for 45 s, 60°C for 45 s, and 68°C for 90 s. 5 l of each reaction was then run out on a gel, transferred to nylon, and probed with an internal oligonucleotides (rTWIK-2 (5Ј-ccacaagcagcgcgccgagcg-3Ј) or rat actin (5Ј-tctacaatgagctgcgtgtg-3Ј)). The membrane was then exposed to a phosphor-imaging screen for 30 min and analyzed as described above. Signals were normalized to actin.
Electrophysiology-COS-7 cell culture, transfection, and electrophysiology have been extensively described elsewhere (11,(17)(18)(19)(20)(21). Briefly, COS cells were routinely transfected using the DEAE-dextran protocol with 1 g of DNA/20,000 cells seeded in a 35-mm plate. Transfected cells were visualized with anti-CD8-coated beads. For whole cell experiments, the internal solution was 150 mM KCl, 3 mM MgCl 2 , 5 mM EGTA, and 10 mM Hepes at pH 7.2 with KOH. In some experiments (regulation of TWIK-2 by second messengers), 5 mM ATP was included in the internal medium. The external medium contained 150 mM NaCl, 5 mM KCl, 3 mM MgCl 2 , 1 mM CaCl 2 , 10 mM Hepes at pH 7.4 with NaOH. A K ϩ -rich (155 mM) solution was made by substituting external NaCl with KCl. The same solutions were used for outside-out patch experiments. In certain experiments, internal Mg 2ϩ was omitted. For cell-attached patch experiments, the bath contained the internal solution, and the pipette contained the external solution. Cells were continuously superfused with a microperfusion system during the time course of the experiments (0.3 ml/min). Temperature of the superfusing solution could be changed and monitored throughout the experiment using a homemade heating device (21).
Data analysis was performed using Clampfit (Pclamp) and Biopatch (Biologic). Student's t test was used for statistical analysis (p Ͻ 0.01).

RESULTS
A partial human expressed sequence tag (AA604914) was identified, and the full coding sequence was subsequently obtained from human brain using 5Ј-and 3Ј-RACE PCR. This gene encodes a 313-amino acid polypeptide 100% identical to the previously reported sequences of hTWIK-2 and hTOSS (1,9). We subsequently identified a partial rat expressed sequence tag (AI454696) coding for rTWIK-2. 5Ј-and 3Ј-RACE were used to clone rTWIK-2 in its entirety from rat heart. rTWIK-2 shares 84% identity with hTWIK-2. The greatest divergence between FIG. 3. Functional expression of TWIK-2 in transfected COS cells. Whole cell patch clamp configuration was used to record TWIK-2 currents in transiently transfected cells. A, hTWIK-2 recorded in a physiological K ϩ gradient. A hTWIK-2-expressing cell was voltage-clamped at a holding potential of Ϫ80 mV, and steps were applied every 10 s with an increment of 20 mV from Ϫ110 mV to 90 mV. The zero current is indicated by a dotted line. B, rTWIK-2 currents recorded in a physiological K ϩ gradient. The holding potential was Ϫ80 mV, and voltage steps were applied every 20 s with an increment of 20 mV from Ϫ110 mV to 90 mV. The zero current is indicated by a dotted line. C, currents elicited with voltage ramps of 800-ms duration from a holding potential of Ϫ80 mV were recorded in physiological (5 mM K ϩ , NaCl) and symmetrical K ϩ gradients (155 mM K ϩ ; KCl). A mock CD8-transfected cell is represented in gray, and a hTWIK-2 transfected cell is represented in black. The inset shows the relationship between the reversal potential of hTWIK-2 and the external K ϩ concentration. D, current density of CD8, hTWIK-2, hTWIK-2 splice variant, rTWIK-2, and rTWIK-2 splice variant in COS transfected cells. Currents were measured at Ϫ130 mV in symmetrical K ϩ (KCl; lower bars) and at 100 mV in physiological K ϩ (NaCl; upper bars). Numbers of experiments are indicated. **, p Ͻ 0.01; ***, p Ͻ 0.001. the homologues occurs in the COOH terminus ( Fig. 1). A TWIK-2 splice variant was cloned from both rat and human tissues. These splice variants encode proteins that initiate in transmembrane segment 2 ( Fig. 1).
Hybridization of a human multiple tissue expression array with a hTWIK-2 probe and PCR analysis of rat cDNA templates were used to examine the pattern of expression of TWIK-2 in various tissues (Fig. 2). The highest expression of hTWIK-2 was found in placenta, esophagus, stomach, salivary gland, spleen, and aorta ( Fig. 2A). A substantial expression was also detected in lymph node, bone marrow, peripheral blood leukocytes, and thymus ( Fig. 2A). Rat TWIK-2 was abundant in aorta, kidney, spleen, stomach, lung, and pulmonary artery (Fig. 2B). Rat blood vessels were denuded of endothelium, and colon and jejunum were free of epithelium, indicating an important expression of rTWIK-2 in smooth muscle. As observed in the human heart, rTWIK-2 is homogeneously expressed at lower levels in the septum, right ventricle, left ventricle, and atria, but almost absent in the brain (Fig. 2, A and B). We analyzed by reverse transcriptase-PCR the differential distribution of the full-length TWIK-2 and the splice variant in various human tissues. In most tissues tested, the full-length TWIK-2 and the splice variant had the same pattern of expression (not shown). However, in colon, fetal heart, atrium, and atrioventricular node, the splice variant was not significantly expressed, while the full-length form was abundant.
Co-expression of rTWIK-2 with hTWIK-1 (n ϭ 14) or KCNK7 (n ϭ 6) did not alter channel properties and current density (not shown). TWIK-1, TWIK-2, and KCNK7 channels are characterized by the presence of a leucine residue in the second pore P2 instead of a phenylalanine found at a conserved position in all the other 2P domain K ϩ channels. Substituting leucine 217 FIG. 4. Inactivation characteristics of rTWIK-2. A, a two-pulse voltage protocol was used to monitor the inactivation of rTWIK-2. A conditioning voltage prepulse of 10 s in duration was applied before a test pulse at 100 mV. The holding potential was Ϫ80 mV, and increments of 20 mV were applied every 20 s from Ϫ110 to 90 mV. B, rTWIK-2 I-V curves measured at peak current (open circles) and at the end of the 10-s duration prepulse (filled circles). The same cell is shown as in A. C, steady-state inactivation curve of rTWIK-2. Normalized currents measured during the test pulse were represented as function of the preconditioning prepulse voltage. Experimental data (n ϭ 9) were fitted with a Boltzmann relationship, and half-inactivation occurred at ϩ65 mV. The inset represents the time constant of inactivation (single exponential decay) as a function of voltage of the cell illustrated in A. D, effects of a K ϩ -rich solution (155 mM KCl) on the inactivation kinetic of rTWIK-2. The holding potential was Ϫ80 mV, and the cell was depolarized every 20 s to 100 mV. The same cell is shown as in A. TWIK-2, an Inactivating 2P Domain K ϩ Channel with a phenylalanine in hTWIK-2 had no significant effect on current density (Ϫ13.2 Ϯ 4.4 pA/pF in symmetrical K ϩ at Ϫ130 mV and 8.8 Ϯ 1.9 pA/pF in physiological K ϩ at 100 mV, n ϭ 14), on the reversal potential (Ϫ84.0 Ϯ 2.9 mV, n ϭ 14) or on current kinetics. On the contrary, substitution of tyrosine 109 with an alanine in the first pore domain killed channel activity (Ϫ1.5 Ϯ 0.7 pA/pF in symmetrical K ϩ at Ϫ130 mV and 2.7 Ϯ 0.4 pA/pF in physiological K ϩ at 100 mV, n ϭ 13).
The inactivation of rTWIK-2 was studied using a standard double voltage pulse protocol (Fig. 4A). rTWIK-2 currents recorded during the conditioning prepotential showed a gradual inactivation upon depolarization. Inactivation became evident at potentials above Ϫ30 mV (Fig. 4B). At room temperature, the time constant of inactivation was 1935 Ϯ 159 ms (at 70 mV, n ϭ 7) and remained constant at all potentials studied (Fig. 4C, inset). The current amplitude elicited during the test pulse at 100 mV was gradually decreased with depolarizing preconditioning voltage pulses (Fig. 4, A and C). A steady-state inactivation curve was constructed and indicates that rTWIK-2 was inactivated over the whole voltage range with half-inactivation at ϩ65 mV (Fig. 4C). Finally, the inactivation rate of rTWIK-2 at depolarized potential was largely impaired when currents were recorded in symmetrical K ϩ conditions (n ϭ 6; Fig. 4D). Similar inactivation properties were observed for hTWIK-2 (Fig. 3A).
The inactivation of rTWIK-2 was highly temperature-sensitive (Fig. 5). At physiological temperature, the outward current inactivated with a time constant of 218 Ϯ 42 ms (at 0 mV, n ϭ 15) (Fig. 5, A, B, and D). Moreover, the steady-state outward current became inwardly rectifying at 37°C (Fig. 5, B and C). The effect of temperature on both current kinetics and rectification occurred within seconds and was fully reversible on returning to room temperature (Fig. 5A). The inward current recorded during hyperpolarization remained time-independent at physiological temperature (Fig. 5B). Inactivation was only partial, and a steady outward current was still present at 37°C (Fig. 5, A and D). Recovery from inactivation at 37°C occurred within 200 ms (n ϭ 3) (Fig. 5D).
Outside-out patches were excised from COS cells expressing rTWIK-2. The pipette solution was Mg 2ϩ -and Ca 2ϩ -free. Under these conditions, we observed currents that displayed the same kinetic and rectification characteristics as the macroscopic currents (Fig. 6, A and C). The example illustrated in Fig. 6A shows a patch with a high level of channel expression. The I-V curve recorded in a physiological K ϩ gradient was outwardly rectifying and became mildly inwardly rectifying in a symmetrical K ϩ gradient. Similar data were obtained with patches expressing low levels of currents as well as in the presence of 3 mM Mg 2ϩ in the intracellular medium (not shown). In a patch held at Ϫ80 mV and expressing a low level of current, application of an external K ϩ -rich solution induced an inward current of about 2 pA (Fig. 6A, inset). Discrete channel gating could not be clearly identified (n ϭ 18), and single channel conductance was lower than 5 pS. I-V curves were constructed using voltage pulses (as previously shown for the macroscopic currents) in both physiological and symmetrical K ϩ gradients (Fig. 6B). In a physiological K ϩ gradient, the current displayed the same rectification and inactivation characteristics as the whole cell current (Fig. 6B, left panel). In a symmetrical K ϩ gradient, the current became inward at negative potentials, and as observed in the macroscopic conditions, the inactivation was markedly slowed down (Fig. 6B, right  panel). Although the current became more noisy at extreme negative and positive potentials in symmetrical K ϩ gradient, no single-channel event could be clearly observed. In the excised outside-out patch configuration, increasing temperature from 25 to 35°C speeded up the inactivation kinetic, and the effect was fully reversible (n ϭ 3) (Fig. 6C).
TWIK-2 was characterized immunologically by Western blot analysis with polyclonal TWIK-2 antibodies (Fig. 8). Under reducing conditions, no signal was obtained with either mocktransfected COS cells (pCI⅐IRES⅐CD8) or TWIK-1 (Fig. 8B). Two complexes were detected with molecular masses of 34 and 37 kDa for both hTWIK-2 and rTWIK-2 (Fig. 8B). The size of 34 kDa is in good agreement with that of 33.7 kDa calculated from the sequence of hTWIK-2. Since TWIK-2 has two potential sites for N-linked glycosylation in the external M1P1 loop (Fig. 8A), the protein lysates were treated with N-glycosidase before immunoblotting to determine whether the upper complex (37 kDa) might correspond to a glycosylated form of TWIK-2. After this treatment, only the complex of 34 kDa was detected (Fig.  8B). The splice variant of rat TWIK-2 expressed a protein that migrated at the predicted molecular mass of 20 kDa (Fig. 8B). An upper band migrating at 38 kDa was detected that could correspond to a dimer form.
The proposed topology for TWIK-2 indicates that a cysteine is present at position 53 in the extracellular M1P1 linker domain (Fig. 8A). hTWIK-1 subunits self-associate via a disulfide bridge formed by this conserved cysteine at position 69, and this assembly is critical for hTWIK-1 channel activity (29). We substituted cysteine 53 with a serine and examined the effects by Western analysis on transfected COS cells using either nonreducing (absence of 2-mercaptoethanol (Ϫ␤Me)) or reducing conditions (ϩ␤Me). Under nonreducing conditions, major bands were detected corresponding to molecular masses of 64 -70 kDa with lysates from the wild-type rTWIK-2 transfected conditions (as observed for hTWIK-2; data not shown), probably corresponding to a TWIK-2 dimer (Fig. 8C). Substitution of cysteine 53 with a serine did not alter this migration profile (Fig. 8C) (also observed for hTWIK-2; not shown). Under reducing conditions, the bands moving at 64 -70 kDa disappeared as observed for wild-type rTWIK-2 (Fig. 8, B and C). 6. rTWIK-2 single channel activity in excised membrane patches. A, membrane currents in outside-out membrane patches excised from rTWIK-2-transfected COS cells. Currents elicited with voltage ramps of 800 ms in duration from a holding potential of Ϫ80 mV were recorded in a physiological K ϩ gradient (NaCl; 5 mM K ϩ ) and in a symmetrical K ϩ gradient (KCl; 155 mM K ϩ ). The internal medium was Mg 2ϩ -free. The inset shows rTWIK-2 current in an outside-out patch measured at Ϫ80 mV in the presence of a physiological (NaCl) and a symmetrical K ϩ gradient (KCl). The current amplitude corresponding to a conductance of 5 pS is indicated by a vertical bar. B, left panel shows rTWIK-2 currents recorded in a physiological K ϩ gradient (NaCl); right panel shows the same patch in a symmetrical K ϩ gradient (KCl). The holding potential was Ϫ80 mV, and voltage steps are applied every 15 s with an increment of 20 mV from Ϫ90 to 90 mV. The zero current is indicated by a dotted line. The intracellular medium was Mg 2ϩ -free. C, effects of increasing temperature from 25 to 35°C (within 15 s) on rTWIK-2 inactivation kinetic. The holding potential was Ϫ80 mV, and the outside-out patch was depolarized to 100 mV every 15 s. The internal medium contained 3 mM Mg 2ϩ .

TWIK-2, an Inactivating 2P Domain K ϩ Channel
Substitution of cysteine 53 with a serine in the external loop M1P1 of hTWIK-2 and rTWIK-2 significantly reduced current amplitudes recorded at both negative and positive potentials in the two different K ϩ conditions (Fig. 8D). DISCUSSION We cloned human and rat TWIK-2 and successfully expressed both channels in transiently transfected COS cells. TWIK-2 is abundantly expressed in visceral and vascular smooth muscle cells. Additionally, the substantial expression of hTWIK-2 in spleen, peripheral blood leukocytes, thymus, lymph node, and bone marrow suggests that TWIK-2 is also a component of the K ϩ channel repertoire of the immune system. TWIK-2 may thus constitute an interesting pharmacological target for the control of immunoreactivity as well as smooth muscle tone.
The currents generated by hTWIK-2 in COS cells were substantially smaller (about 15-fold) than rTWIK-2-mediated currents. The weak currents recorded with hTWIK-2 may thus explain the unsuccessful functional expression of hTOSS in cRNA-injected Xenopus oocyte (9). The most variable region between rTWIK-2 and hTWIK-2 is located in the carboxyl terminus. Interestingly, we have previously demonstrated that the carboxyl-terminal regions of TREK-1 and TASK-1 are critical for channel activity (11, 18 -21). Besides the difference in current density, the general biophysical and pharmacological characteristics of the currents are identical between human and rat TWIK-2 channels and moreover are unique among the previously cloned 2P domain K ϩ channels.
So far, TWIK-2 is the only member of the 2P domain K ϩ channel family to display a time-dependent inactivation at depolarized potentials. Increasing temperature from 22 to 37°C dramatically sped up the rate of inactivation by about 10-fold. Increased inactivation at physiological temperature resulted in a decrease in the steady-state current amplitude and produced a marked inward rectification. TWIK-2 is thus a heat-inactivated K ϩ channel, contrasting with TREK-1, a heatactivated K ϩ channel in hypothalamic and dorsal root ganglion thermosensitive neurons (21). At physiological temperature and even at very depolarized potentials, TWIK-2 was only partially inactivating, and a steady current was still present. Interestingly, the inactivation of TWIK-2 was impaired in K ϩrich conditions. The inactivation of Shaker-type K ϩ channels is similarly highly temperature-dependent and sensitive to cations (30,31). The rate of the conformational change that underlies Shaker C-type inactivation is determined largely by the exit rate of K ϩ from the pore, and occupancy of the selectivity filter by K ϩ slows the rate of inactivation (31). Inactivation of Shaker K ϩ channels involves the amino-terminal domain (fast N type inactivation) as well as the carboxyl-terminal region (slow C type inactivation). The exact molecular mechanism of TWIK-2 inactivation remains to be determined. rTWIK-2 was insensitive to TEA and 4-aminopyridine but was reversibly inhibited by external Ba 2ϩ , quinine, and quinidine with a half-inhibition at about 0.1 mM. Interestingly, the inhibition by these blockers was absent in K ϩ -rich conditions whatever the direction of the current. It has previously been reported that hTWIK-2 was a noninactivating weak inward rectifier Kϩ channel (1). Moreover, hTWIK-2 was found to be FIG. 7. Pharmacology of rTWIK-2. A, whole cell recording of rTWIK-2 recorded in physiological K ϩ conditions (NaCl) and in symmetrical K ϩ conditions (KCl). The addition of 1 mM Ba 2ϩ in physiological K ϩ conditions blocked rTWIK-2 currents at both negative and positive potentials. In symmetrical K ϩ conditions, rTWIK-2 was resistant to 10 mM Ba 2ϩ . The holding potential was Ϫ80 mV, and the cell was stimulated with voltage ramps of 800 ms in duration every 20 s. B, histogram showing the dose-dependent inhibition of rTWIK-2 by Ba 2ϩ in a physiological (NaCl) and a symmetrical K ϩ gradient (KCl) at 100 mV. The numbers of experiments are indicated. C, effect of 0.1 mM quinidine on rTWIK-2. The holding potential was Ϫ80 mV, and the cell was hyperpolarized to Ϫ120 mV and depolarized to ϩ60 mV. D, effect of 10 mM Cs ϩ on rTWIK-2. Conditions were the same as in C.
resistant to Ba 2ϩ , quinine, and quinidine and thus was pharmacologically different from TWIK-1 (1). However, these experiments were performed in high K ϩ conditions at voltage values below ϩ40 mV, thus masking the time-dependent inactivation and altering channel pharmacology (1). This observation demonstrates that analysis of these channels should be performed under physiological K ϩ conditions. Moreover, these results suggest that occupancy of the pore by K ϩ might alter the conformation of the channel as observed for the inactivation and thus influence the pharmacology. Interestingly, the pharmacology of Shaker-type K ϩ channels, in addition to the inactivation (see above), is also cation-dependent (31).
As previously observed for TWIK-1, KCNK7, and TREK-1 (13,21,29), TWIK-2 self-assembles as a dimer under nonreducing conditions. However, substitution of the conserved cysteine with a serine at position 53 in the M1P1 external loop of TWIK-2 does not impair self-assembly. In the absence of cysteine 53, TWIK-2 current density was strongly reduced, thus demonstrating, as previously observed for TWIK-1 (29), the critical role of this residue for functional expression of TWIK channels. This cysteine is lacking in TASK-1 and TASK-3 channels (3, 6, 10).
A TWIK-2 splice variant encoding a truncated channel was identified in both human and rat tissues. The TWIK-2 splice variant does not express a current. Using reverse transcriptase-PCR analysis, we identified the colon, the atrium, and the atrioventricular node as differentially expressing the full-length and the splice variant TWIK-2 channels. The physiological relevance of the TWIK-2 splice variant remains to be determined.
The high expression of rTWIK-2 allowed us to record this channel in the outside-out patch configuration. Channel events were not visualized under our recording conditions, indicating a very small single channel conductance. Current inactivation as well as rectification were observed in excised patches in the absence of internal Mg 2ϩ . These results suggest that both inactivation and rectification are Mg 2ϩ -independent mechanisms and may be intrinsic properties of the TWIK-2 channels.
The characterization of TWIK-2 clearly differentiates this channel from the other closest 2P domain family member, TWIK-1 (8). First, the patterns of expression of the two channels are completely different, as illustrated by the strong expression of TWIK-1 in the human brain, while TWIK-2 is almost absent. Moreover, mouse TWIK-2 is only expressed in the liver (9), while mouse TWIK-1 is strongly expressed in multiple tissues including brain and heart (8). Second, the functional properties of both channels are radically different. TWIK-1 is time-independent, while TWIK-2 is inactivating at depolarized potentials (8). TWIK-1 shows a prominent Mg 2ϩdependent inward rectification, while the mild inward rectification of TWIK-2 is Mg 2ϩ -independent (8). The single channel conductance of TWIK-1 is large (34 pS), whereas TWIK-2 single channel conductance is below 5 pS (8). The cysteine in the M1P1 loop is critical for TWIK-1 dimerization (29), whereas it is not for TWIK-2 self-assembly. The common functional features between TWIK-1 and TWIK-2 are the background timeindependent activity at negative potentials; the sensitivity to Ba 2ϩ , quinine, and quinidine in physiological K ϩ ; the insensitivity to TEA and 4-aminopyridine; and the importance of the conserved cysteine at position 53 for channel activity (8,29).

FIG. 8. Immunological characterization of TWIK-2 in COS transfected cells.
A, schematic diagram illustrating the proposed membrane topology of TWIK-2. B, polyclonal rabbit antibodies were raised against the M1P1 external loop and the carboxyl-terminal region of hTWIK-2. 48 h after transfection, cell lysates were prepared, and approximately equal amounts of proteins were separated by 12% SDS-PAGE and transferred onto a nitrocellulose membrane after treatment with 2-mercaptoethanol (␤Me) (2% final concentration). 5 g of whole cell lysate were loaded per lane, and control conditions were CD8-and TWIK-1-transfected COS cells. The molecular mass markers (in kDa) are indicated on the left, and the migration patterns of the full-length (WT) and the splice variant (splice) of rTWIK-2 are indicated on the right. The glycosylated form of wild-type rTWIK-2 is indicated by an arrow on the right. C, rTWIK-2 and cysteine 53-mutated rTWIK-2 polypeptides were detected by the affinity-purified antibodies. Control lanes show CD8-transfected COS cells. The molecular mass markers (in kDa) are indicated on the left. The glycosylated form of wild-type rTWIK-2 is indicated by an arrow on the right. Treatment with or without 2-mercaptoethanol is indicated by a plus and minus sign, respectively. D, current density of hTWIK-2, Cys 53 hTWIK-2, rTWIK-2, and Cys 53 rTWIK-2 in COS transfected cells. Currents were measured at Ϫ130 mV in symmetrical K ϩ (KCl; lower bars) and at 100 mV in physiological K ϩ (NaCl; upper bars). The numbers of experiments are indicated. TASK-1 is highly sensitive to external acidosis and is fully blocked at pH 6.4 (3), while TWIK-2 is only depressed by 36%. TREK-1 and TRAAK are strongly stimulated by polyunsaturated fatty acids including arachidonic acid (5,17,19,20), while TWIK-2 is only stimulated by 76%. TREK-1, TREK-2, and TRAAK are mechano-gated K ϩ channels (2,(17)(18)(19), while TWIK-2 is constitutively active at atmospheric pressure. TREK-1 is blocked by the cAMP/protein kinase A pathway as well as the protein kinase C pathway (19,20), while TWIK-2 is only slightly potentiated. TREK-1 is opened by heat (21), while TWIK-2 is inactivated. TREK-like channels are characterized by a high conductance (100 pS) (2,17,19), and TASK-like channels are characterized by intermediate conductances (14 -27 pS) (6,7,10,19), while TWIK-2 is characterized by a conductance lower than 5 pS. Taken together, these functional properties clearly differentiate TWIK-2 from the other 2P domain K ϩ channels.
The base-line activity of TWIK-2 suggests that these channels may contribute to the setting of cellular resting membrane potential. The inactivation of TWIK-2, which is fast at physiological temperature, transforms this background K ϩ channel into an A-type K ϩ channel at depolarized potentials. TWIK-2 behaves like Kv1.4, Kv4.2, and Kv4.3 at depolarized potentials and thus might contribute to the early repolarization of the action potential. The lack of complete inactivation of TWIK-2 even at very depolarized potentials also implies that the steady outward TWIK-2 current will probably contribute to the late phase of repolarization of the action potential. TWIK-2 should then be seen as a K ϩ channel contributing to the setting of the resting membrane potential and influencing both early and late phases of action potential repolarization.