ION CHANNELS, RECEPTORS AND TRANSPORTERS

A splice variant of the two-pore domain potassium channelTREK-1 with only one pore domain reduces the surfaceexpression of full-length TREK-1 channels

Susanne Rinné & Vijay Renigunta & Günter Schlichthörl &Marylou Zuzarte & Stefan Bittner & Sven G. Meuth &

Niels Decher & Jürgen Daut & Regina Preisig-Müller

Received: 23 August 2013 /Revised: 9 October 2013 /Accepted: 12 October 2013# Springer-Verlag Berlin Heidelberg 2013

Abstract We have identified a novel splice variant of thehuman and rat two-pore domain potassium (K2P) channelTREK-1. The splice variant TREK-1e results from skippingof exon 5, which causes a frame shift in exon 6. The frameshift produces a novel C-terminal amino acid sequence and apremature termination of translation, which leads to a loss oftransmembrane domains M3 and M4 and of the second poredomain. RT-PCR experiments revealed a preferentialexpression of TREK-1e in kidney, adrenal gland, andamygdala. TREK-1e was nonfunctional when expressed inXenopus oocytes. However, both the surface expression andthe current density of full-length TREK-1 were reduced by co-expression of TREK-1e. Live cell imaging in COS-7 cellstransfected with GFP-tagged TREK-1e showed that this splicevariant was retained in the endoplasmic reticulum (ER).Attachment of the C-terminus of TREK-1e to two differentreporter proteins (Kir2.1 and CD8) led to a strong reduction inthe surface expression of these fusion proteins. Progressivetruncation of the C-terminus of TREK-1e in these reporterconstructs revealed a critical region (amino acids 198 to 205)

responsible for the intracellular retention. Mutagenesisexperiments indicated that amino acids I204 and W205 arekey residues mediating the ER retention of TREK-1e. Ourresults suggest that the TREK-1e splice variant may interferewith the vesicular traffic of full-length TREK-1 channels fromthe ER to the plasma membrane. Thus, TREK-1e mightmodulate the copy number of functional TREK-1 channelsat the cell surface, providing a novel mechanism for finetuning of TREK-1 currents.

Keywords Alternative splicing . K2P channel . Potassiumchannels . Splice variant . TREK-1

Introduction

The subunits of K2P channels have four helical trans-membrane domains (M1–M4) and two pore-forming domains(P1 and P2); two subunits assemble to form a functionaldimeric K+ channel [3, 12]. The mammalian genome encodes15 different mammalian K2P channels, but not all of them canbe functionally expressed. TREK-1 is a mechano-sensitivemember of the K2P channel family that is also sensitive tolipids, polyunsaturated fatty acids (including arachidonicacid), temperature, pH, and a range of clinically relevantcompounds including volatile anesthetics [5, 8]. It isexpressed at high levels in excitable cells such as neurons,cardiac muscle cells, and smooth muscle cells. In neurons,TREK-1 plays a prominent role in controlling restingmembrane potential and electrical excitability. The diversityof functional K2P channel proteins can be increased byalternative translation initiation [17, 19] and by alternativesplicing [6, 7, 22]. Two TREK-1 isoforms resulting fromalternative splicing of exon 1 have been described previously(TREK-1a and TREK-1b) [22]. In addition, a truncated

S. Rinné :V. Renigunta :G. Schlichthörl :M. Zuzarte : J. Daut :R. Preisig-MüllerInstitute for Physiology and Pathophysiology, Cell Physiology,University of Marburg, Marburg, Germany

S. Rinné :N. DecherInstitute for Physiology and Pathophysiology, Vegetative Physiology,University of Marburg, Marburg, Germany

S. Bittner : S. G. MeuthDepartment of Neurology, University ofMünster, Münster, Germany

S. Rinné (*) :R. Preisig-Müller (*)Institute for Physiology and Pathophysiology, Philipps-Universityof Marburg, Deutschhausstraße 1-2, 35037 Marburg, Germanye-mail: rinne@staff.uni-marburg.dee-mail: preisigm@staff.uni-marburg.de

Pflugers Arch - Eur J PhysiolDOI 10.1007/s00424-013-1384-z

neuronal splice variant has been reported, TREK-1ΔEx4,which encodes a heavily truncated protein with only a singletransmembrane domain [20]. This variant displayed nochannel activity itself, but reduced TREK-1 whole-cell currentamplitudes. Finally, myometrial TREK-1 splice variants,lacking either the pore, transmembrane domains, or both areknown [21]. Here, we describe a novel splice variant of theK2P channel TREK-1. In this variant, denoted TREK-1e, theskipping of exon 5 results in a protein lacking the second poredomain (M3 and M4) due to a frame shift in the M2–M3linker which produces a premature stop codon. We havestudied the function of TREK-1e and found that it exerts adominant-negative effect on other TREK-1 isoforms. Thiseffect is mediated by a short amino acid stretch in the C-terminus of TREK-1e and might represent a new mechanismfor the fine tuning of the functional expression of TREK-1 indifferent tissues.

Materials and methods

Cloning, expression analysis, and mutagenesis of TREK-1splice variants

Using the basic local alignment search tool (BLAST) with theamino acid sequence of TREK-1 in the Expressed SequenceTags database (dbEST) dynamically translated in all readingframes, we identified two splice variants of TREK-1 (humanand rat) which differ in their extreme N-terminus (TREK-1cand TREK-1d). The BLAST programs tblastn and blastn wereused to localize the different exons 1 on human chromosome 1or rat chromosome 13. Total RNAs from human and rattissues were purchased from BD Biosciences Clontech. TotalRNA from the cell line NCI-H295R was extracted using theHigh Pure RNA Isolation Kit (Roche) according tomanufacturer's protocol. Reverse transcription (RT) wasperformed to prepare first-strand cDNAs using Superscript IIreverse transcriptase (Life Technologies Inc., Gaithersburg,MD) according to the manufacturer's instructions. Humanbrain microvascular endothelial cells (HBMECs, ScienCellResearch Laboratories, Carlsbad, CA, USA) were culturedin fibronectin-coated (2 μg/cm2, Sigma-Aldrich, Munich,Germany) T-75 flasks with microvascular endothelial cellgrowth medium (Provitro, Berlin, Germany). Cells wereharvested using accutase (PAA, Cölbe, Germany) accordingto the manufacturer's instructions and counted by CASYModel TT (Innovatis AG, Reutlingen, Germany). TotalRNAwas isolated using TRIzol reagent (Invitrogen, Carlsbad,Germany). Reverse transcription was performed to preparefirst-strand cDNAs using TaqMan Reverse TranscriptionReagents (Applied Biosystems, Darmstadt, Germany). Full-length hTREK-1a, hTREK-1b, and hTREK-1c and rTREK-1a, rTREK-1b, rTREK-1c, and rTREK-1d were obtained by

RT-PCR; the primers are shown in Tables 1 and 2. Amplifiedproducts were subcloned into pGEM-T-Easy (Promega) andsequenced. Plasmid DNAwas isolated from four independentcolonies during each cloning procedure. Sequencing ofplasmids from different hTREK-1c and rTREK-1a coloniesrevealed splice variants formed by skipping of exon 5. Thesesplice variants were denoted hTREK-1e and rTREK-1e. Thenew nucleotide sequences reported here have been submittedto GenBank with the following accession numbers: hTREK-1c (AY552980), hTREK-1e (AY552981), rTREK-1c(DQ403851), rTREK-1d (AY555072), and rTREK-1e(AY555073). For TREK-1 and TREK-1e expression analysis,the primers were located on exons flanking exon 5 (hTREK-1and hTREK-1e: forward and reverse primers located on exons3 and 6, respectively; rTREK-1 and rTREK-1e: forward andreverse primers located on exon 4 and exon 7, respectively(for details, see Tables 1 and 2). RT-PCR of GAPDHwas usedas a control. For single amino acid exchanges, insertions ordeletions the QuikChange Site-directed mutagenesis kit(Stratagene) was used. All constructs were verified bysequence analysis.

Western blot analysis

About 24 h after transfection, HEK-293 cells were washedwith phosphate-buffered saline, collected by centrifugation,and extracted in lysis buffer containing 50 mM Tris–HCl, pH7.4, 150 mM NaCl, 1 % TX-100, and 10 μl/ml proteaseinhibitor cocktail (Sigma). Porcine adrenal gland washomogenized in homogenization buffer (10 mM HEPES, pH7.4, 350 mM sucrose, 5 mM EDTA, 10 μl/ml proteaseinhibitor cocktail (Sigma)) and centrifuged at 100,000×g for1 h. The pellet was lysed in homogenization buffer containing

Table 1 Primers used for cloning and qualitative expression analysis

Primer sequence in 5′ to 3′ orientation Primer name Primernumber

CTGCCCGTGCAGCTCGGAGCG hTREK-1a-For 1

GAATGCTGCATGCCTCATGCTT TREK-1b-For 2

GAAACCTTGGAGGAAGAACGGCATT hTREK-1c-For 3

TCAGTCACTGGGATTTGGGAAGTTC hTREK-1-For3 4

GTCTTTTTAGATATCACTCGGACCA hTREK-1-Rev5 5

CTACCAGCAACAGTCATAAAGAGC hTREK-1-Rev4 6

CTGCCCGTGCAGCTCGGAGCG rTREK-1a-For 7

GAAACCTTGGAGGAAGAGCAGCGTT rTREK-1c-For 8

GCATGACAGCTAGAGCAAGAGGTG rTREK-1d-For 9

CAGCTTGGAACCATATTTG rTREK-1-For3 10

TCCCTATGGCTCCACCTCAGACTTC rTREK-1-Rev3 11

CATCACCATCTTCCAGGAGCGA GAPDH-For 12

GTCTTCTGGGTGGCAGTGATGG GAPDH-Rev 13

Pflugers Arch - Eur J Physiol

0.5 % (w /v ) deoxycholate. The protein extracts weresuspended in sodium dodecyl sulfate (SDS) sample buffer,denatured for 10 min at 95 °C, separated on 12 % SDSpolyacrylamide gels, and visualized by immunoblotting usinga TREK-1-specific antibody (Santa Cruz Biotechnology,TREK-1 (N20) sc-11554). To show the specificity of theTREK-1 antibody, the antigenic peptide (Santa CruzBiotechnology, TREK-1 (N20)P sc-11554P blocking peptide)was mixed in a 1:1 (w /w) ratio with the antibody for a pre-absorption assay. The binding of the primary antibody wasdetected with appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies (DAKO Cytomation,polyclonal rabbit anti goat HRP, P0449, 1:2000) and achemiluminescent substrate (Pierce, Super Signal West DuraExtended Duration Substrate, 34075). For the detection of theCD8 constructs, a rabbit anti-CD8 antibody (Santa Cruz,1:1000) and an HRP-conjugated anti-rabbit antibody (Pierce,1:2,000) were used. Detection of hemagglutinin (HA)constructs in Xenopus laevis oocytes lysates was performedas described previously [25].

Imaging

COS-7 cells were seeded on 35-mm dishes. About 24 h later,cells were transfected with the particular DNA construct usingFugene 6 (Roche). Transfected COS-7 cells were seeded onPetri dishes, fixed, and stained with a TREK-1-specific orCD8-specific antibody or live cell imaging was performed.For staining of the nucleus, Hoechst Dye 33258 (MolecularProbes) was used. ER-Tracker Red (BODIPY TRGlibenclamide, Molecular Probes) was used as a marker forthe endoplasmic reticulum (ER). To obtain a suitable plasma

membrane marker, the 20 amino terminal amino acids ofneuromodulin were fused to the amino terminus of DsRedmonomer [25]. Fluorescence images were acquired using aninverted microscope (Olympus IX71) equipped with a ×60objective (numerical aperture, 1.45) and a cooled CCDcamera (SensiCam QE 12 bit camera, PCO-CCD Imaging,Kelheim, Germany). Images were further analyzed by usingImage-Pro® Plus software, version 4.5.

Yeast two-hybrid assay

The N-termini of rTREK-1a (aa 1-36), rTREK-1b (aa 1-51),and rTREK-1d (aa 1-39) were tested for interaction with thecytosolic C-terminus of rTREK-1e (aa 179-226) using theMatchmaker yeast two-hybrid system (Clontech, Palo Alto,CA). The C-terminus of guinea pig Kir2.1 served as a negativecontrol. The DNA fragments were amplified by PCR andcloned into the vectors pGBT-9 (for binding domain fusion;bait vector) and pGAD-424 (for activation domain fusion;prey vector). All constructs were verified by sequenceanalysis. The assay was performed by co-transformation ofthe yeast strain HF7c with 150 ng of both vectors as describedpreviously [15]. Growth of colonies on the selective dropoutmedium lacking leucine, tryptophan, and histidine (-LWH)was scored as positive for interaction.

Current and surface expression measurements in Xenopusoocytes

Full-length cDNAs of the different TREK-1 splice variantswere subcloned from the pGEM-T-easy plasmids into theNotI sites of the oocyte expression vector pSGEM. In ratTREK-1a, an HA tag was introduced at amino acid position266. The construction of Kir2.1-HA-TREK-1e-CT wasperformed by introducing the HA tag at amino acid position117 of the guinea pig Kir2.1 and by fusing the rat TREK-1e C-terminus (amino acids 179–226) to the distal C-terminus ofKir2.1 via EcoRI sites we had introduced before. CappedcRNA transcripts were synthesized in vitro using the T7MessageMachine kit (Ambion, Austin, TX, USA). ThecRNAs were purified, photometrically quantified, andinjected individually or in combination into oocytes atconstant amounts (2.5 ng cRNA for TREK-1 splice variantsand 0.25 ng cRNA for the fusion constructs). For hTREK-2c[6], 5 ng cRNAwas injected. Oocytes were incubated at 19 °Cfor 24–48 h in ND96 solution containing (millimolar) 96NaCl, 2 KCl, 1 MgCl2, 1.8 CaCl2, and 5 HEPES (pH 7.4–7.5), supplemented with 50 mg/l gentamicin, 90 mg/ltheophylline, and 275 mg/l sodium pyruvate. Two-microelectrode voltage clamp measurements were performedwith a Turbo Tec-10 C amplifier (npi, Tamm, Germany), anddata were recorded at a sampling rate of 120 Hz. The oocyteswere placed in a small-volume perfusion chamber and

Table 2 Primer combinations from Table 1 used for cloning andqualitative expression analysis; expected sizes of the amplificationproducts are given in base pairs (bp)

Expression analysis Full-length cloning

Primercombination

Expectedsize (bp)

Primercombination

Expectedsize (bp)

hTREK-1 4.5 538 – –

hTREK-1a – – 1.6 1,354

hTREK-1b – – 2.6 1,370

hTREK-1c – – 3.6 1,410

hTREK-1e 4.5 350 3.6 1,223

rTREK-1 10.11 734 – –

rTREK-1a – – 7.11 1,314

rTREK-1b – – 2.11 1,311

rTREK-1c – – 8.11 1,371

rTREK-1d – – 9.11 1,322

rTREK-1e 10.11 547 7.11 1,127

GAPDH 12.13 343 – –

Pflugers Arch - Eur J Physiol

superfused with ND96 solution. Since current amplitudesvaried from one batch of oocytes to the next, currents werenormalized for each batch. Surface expression analysis inoocytes was performed as previously described [23, 25].Chemiluminescence was quantitated in a luminometer(Berthold EG&G LB9507 or Promega GLOMAX 20/20)and results are given in relative light units.

CD8 reporter protein assay

pcDNA3 hCD8-KKXX and pcDNA3 hCD8-AAXXconstructs [23] were kindly provided by Blanche Schwappach(University of Göttingen). The C-terminus of rTREK-1e wasamplified by PCR and cloned into a pcDNA3-CD8 vectorusing NotI and XhoI restriction sites. Surface expressionanalysis in COS-7 cells was performed as describedpreviously [15].

Data analysis

Data are reported as means ± SEM. Statistical significancewas calculated using unpaired Student's t test. All experimentswere performed at room temperature (22–24 °C); A singleasterisk (*) indicates P <0.05, two asterisks P <0.01, threeasterisks P <0.001, and “n.s.” indicates no significant change.

Results

Cloning of a TREK-1 splice variant with two transmembranedomains

Performing BLASTsearch in ESTand genomic databases, wepreviously identified TREK-1 splice variants, TREK-1c andTREK-1d, which we deposited in the NCBI database with theaccession numbers hTREK-1c (AY552980) for the humanand rTREK-1c (DQ403851) and rTREK-1d (AY555072) forthe rat clones, respectively. These isoforms differ in their N-terminal sequence, due to an alternative splicing of the firstexon (Fig. 1a, b). The genomic organization of the human andrat TREK-1 (KCNK2 ) gene locus is illustrated in Fig. 1a. Notethat the TREK-1d splice variant was only found in rat, whilethe other N-terminal isoforms are also present in humans(Fig. 1a, b). Figure 1b shows a sequence alignmentdemonstrating the differences between the TREK-1 isoformscaused by alternative N-terminal splicing.

We now report a novel splice variant, named TREK-1e,which is characterized by skipping of the entire exon 5. Whilethe “classical” TREK-1 isoforms are spliced from exon 5 toexon 6 (Fig. 1c), TREK-1e in humans and rats is spliced fromexon 4 to exon 6 (Fig. 1c). This exon skipping in TREK-1eleads to a frame shift in exon 6 (Fig. 1c), starting at the distalpart of the M2–M3 linker. The frame shift in TREK-1e leads

to a premature stop codon and to a deletion of transmembranedomains 3 and 4 and the second pore domain (Fig. 1d). Thus,TREK-1e encodes only the first two transmembrane domainsof TREK-1, followed by a novel C-terminal sequence(Fig. 1b). TREK-1e can be found in rat (226 aa) and humantissue (228 aa). The new nucleotide sequences were submittedto GenBank with the following accession numbers: hTREK-1e (AY552981) and rTREK-1e (AY555073).

Expression analysis

Expression and functional properties of TREK-1a and TREK-1b were previously studied in cardiomyocytes and neuronaltissue [2, 4, 22]. We have now analyzed the expression patternof the novel TREK-1e splice variant by RT-PCR fromdifferent human and rat tissues. In human tissue, the forwardPCR primer was located in exon 3 and the reverse primer inexon 6 (Fig. 2a). In rat tissue, using a similar approach, theforward primer was located in exon 4 and the reverse primerwas located in exon 7 (not illustrated). Thus, the PCRs lead toamplification products containing exon 5 (corresponding tothe classical isoforms TREK-1a-d) and amplification productsnot containing exon 5 (the novel splice variant TREK-1e)(Fig. 2a). In rat, TREK-1e is expressed in the brain and kidneyand in humans additionally in the adrenal gland (Fig. 2b,lower bands). Within the human brain, TREK-1eis preferentially expressed in the amygdala (Fig. 2c). NoTREK-1e expression was observed in thalamus, hypo-thalamus, hippocampus, and substantia nigra (not shown).As we found TREK-1e transcripts in the human adrenal gland(Fig. 2b), we also tested for the presence of this isoform in thehuman adrenocortical cell line NCI-H295R. Here, TREK-1eexpression is very prominent (Fig. 2d, lower band). Since arecent study indicated that endothelial TREK-1 regulatesimmune cell trafficking into the central nervous system [1],we also tested for TREK-1e expression inHBMECswhich arethe major element of the human blood–brain barrier (Fig. 2e).We found a very strong expression of TREK-1; however, notranscripts of TREK-1e were detected in these cells (Fig. 2e).

To address the question whether TREK-1e protein isexpressed in vivo, we analyzed protein extracts from porcineadrenal gland cells. As controls, we expressed TREK-1a,TREK-1b, or TREK-1e splice variants in HEK293 cells(Fig. 2f). Western blots from porcine adrenal glands using aTREK-1 antibody detected a ∼47-kD protein for TREK-1aand TREK-1b (in agreement with the expected molecularweight) and a ∼25-kD protein for TREK-1e (Fig. 2f). Thus,the western blots led us to the conclusion that TREK-1e isexpressed in vivo. Pre-absorption of the antibody with itsantigenic peptide prevented any staining (not illustrated). Inline with previous reports, our western blots indicate theexistence of additional, smaller TREK-1 isoforms as well inadrenal gland cells [20, 21].

Pflugers Arch - Eur J Physiol

Intracellular localization of TREK-1 splice variants expressedin COS-7 cells

To analyze the intracellular localization of the newly identifiedsplice variant, live cell imaging or immunostainingexperiments using a TREK-1-specific antibody wereperformed. COS-7 cells were transiently transfected withdifferent TREK-1 splice variants and analyzed 48 h aftertransfection. As previously reported [22], TREK-1b wasprimarily localized at the surface membrane (Fig. 3a–c).Surface membrane expression was confirmed using themembrane marker NM-DsRed [25] (Fig. 3a–c). In contrast,live cell imaging of GFP-tagged TREK-1e channels showedlocalization to the ER, as indicated by co-localization with ER-Tracker Red (BODIPY TR Glibenclamide, Molecular Probes)(Fig. 3d–f). Even 72 h after transfection, the intracellularTREK-1e localization was not changed (data not shown).

TREK-1e reduces current amplitude of classical TREK-1isoforms

Injection of TREK-1a cRNA into Xenopus oocytes produced atypical outwardly rectifying current (Fig. 4a). As expected inview of its intracellular localization, no currents could berecorded after expression of TREK-1e subunits in oocytes

(Fig. 4a) or in HEK-293 cells (not illustrated). As TREK-1e isretained in the ER (Fig. 3d–f), we next tested whether TREK-1emight interact with classical N-terminal TREK-1 splice variantsand reduce their surface expression.We co-expressed TREK-1awith TREK-1e (Fig. 4a) and measured the resulting currentamplitude at 0 mV (Fig. 4b). We found that co-expression ofTREK-1e reduced the currents produced by TREK-1a in adose-dependent manner (Fig. 4b).When the ratio of the injectedTREK-1e to TREK-1a cRNA was increased, the reduction ofTREK-1 current was more pronounced (ratio 1:2; currentnormalized to TREK-1a alone, 0.64±0.06; ratio 1:5, 0.43±0.06; ratio 1:10, 0.37±0.03; Fig. 4b). For the other classicalsplice variants, TREK-1b and TREK-1d, a similar currentreduction by TREK-1e was observed (not illustrated). As acontrol, TREK-1e was co-injected with the potassium channelKir2.1 (ratio 1:10); current amplitudes were found to beunaffected by TREK-1e (Fig. 4b).We also studied the effectof TREK-1e on the mechano-sensitive K2P channel TREK-2.We found that co-expression of TREK-1e did not alter thecurrent amplitude of TREK-2 (Fig. 4c).

Although our results strongly suggest that TREK-1e interactswith classical full-length TREK-1 isoforms (TREK-1a, TREK-1b, or TREK-1d), the quaternary structure of the resultingcomplex is still unknown (see “Discussion”). One possibilityis that one full-length TREK-1 subunit and one TREK-1e

a c

b

dTREK-1a-d TREK-1e

C

N

C

N

Fig. 1 Genomic organization, sequence alignment, splice sitelocalization, and membrane topologies of TREK-1 splice variants. aGenomic organization of TREK-1 (KCNK2) on chromosome 1 in human(upper panel) and chromosome 13 in rat (lower panel), respectively. bPartial alignment of amino acid sequences of rat and human TREK-1splice variants. The amino acid positions are indicated for rat TREK-1b.Red-colored amino acids (TREK-1e) indicate changes as a result of exon5 skipping. Please note that TREK-1e starts with exon 1a (rat) or exon 1c

(human). c Partial gene structures of human (h) and rat (r) TREK-1. Theintronic splice sites are shown in italics , bold, and underlined ; the codingsequences are given in capital letters ; and non-coding and intronicsequences are given in lower-case letters. The selectivity filter of thesecond pore region is underlined . d Putative membrane topologies ofTREK-1a-d and TREK-1e. The new TREK-1e carboxyl terminus (C)encoded by exon 6 is indicated in red

Pflugers Arch - Eur J Physiol

subunit form a heterodimer and that the unique C-terminus ofTREK-1e interacts with a cytosolic domain of full-lengthTREK-1. To test for a direct interaction between the C-terminus of TREK-1e and the N-termini of TREK-1a, TREK-1b, and TREK-1d, we utilized a yeast two-hybrid assay. The N-termini of classical TREK-1 isoforms were cloned into the

activation domain-containing vector pGAD-424, and the C-terminus of TREK-1e was cloned into the yeast DNA-bindingdomain-containing vector pGBT-9. Both constructs were co-transformed into the yeast strain HF7c and transformants wereselected on a leucine and tryptophan double drop-out medium(−LW). Interaction was assayed on triple drop-out medium

TREK-1b PM mergecba

egreme1-KERT ERd e f

5 µm

Fig. 3 Subcellular localization ofthe rat splice variants TREK-1band TREK-1e in transfected COS-7 cells. a–c Immunocytochemistry48 h after transfection of theTREK-1b splice variant usingTREK-1-specific antibodies(1:100). As a plasma membranemarker, NM-DsRed was usedwhich shows a typical staining ofthe entire cell. d–f Live cellimaging 48 h after transfection ofGFP-TREK-1e. The ER wasstained red using ER-Tracker

ahTREK-1

c

bt

dNCI H295R

TREK-1

TREK-1e

1 2 3 4 5 76

1 72 3 4 6

hTREK-1e

HBMECe

rat

TREK-11

GAPDH

NCI-H295R HBMECs

TREK-1eTREK-1

TREK-1eTREK-1

TREK-1e

human

TREK-1TREK-1e

GAPDH

TREK-1

f

TREK-1e

705540352515

kD

human

GAPDH

α

Fig. 2 Expression analysis of the TREK-1e splice variant. a Schematicdrawing of the cDNAs of human TREK-1 and TREK-1e. The primers(arrows) recognizing all TREK-1 splice forms (upper part) or the TREK-1e splice variant (lower part) yield two PCR products with different sizes(indicated by arrows in b–e). b Expression analysis of TREK-1 splicevariants in different human and rat tissues. The black arrows indicate therespective PCR fragments. c Expression analysis of the TREK-1e splicevariant in different regions of human brain. d Expression analysis ofTREK-1 splice variants in total RNA from the adrenocortical cell line

NCI-H295R and e in total RNA of human brain microvascularendothelial cells (HBMECs). f Western blot analysis of protein lysatesof non-transfected HEK293 cells (nt), heterologously expressed TREK-1splice variants TREK-1a, TREK-1b, and TREK-1e and porcine adrenalgland (adr. gland). TREK-1 proteins were immunostained with a TREK-1-specific antibody (molecular weights: TREK-1a and TREK-1b, 47 kD;TREK-1e, 25 kD). The specificity of the antibody was tested by pre-absorption with the antigenic peptide (not shown). Black arrow indicatesthe corresponding TREK-1e protein in porcine adrenal gland lysate

Pflugers Arch - Eur J Physiol

lacking also histidine (−LWH). We found that the C-terminus ofTREK-1e can interact with the N-termini of TREK-1a, TREK-1b, and TREK-1d (Fig. 4d). As a negative control, we co-transformed the vectors containing the C-terminus of TREK-1e and the C-terminus of Kir2.1 into HF7c cells. For the lattercombination, no interaction was detectable (Fig. 4d).

TREK-1e reduces the surface expression of classical TREK-1isoforms

To examine whether the reduction of the TREK-1 currentamplitude was due to reduced surface expression of thechannel, we quantified the surface expression of TREK-1a in

ba1 2TREK 15

0.2

0.4

0.6

0.8

1.0

1.2

No

rmal

ized

cu

rren

t

***

**

******

TREK-1

TREK-1e

0

1

2

3

4

5

Cu

rren

t (

A)

TREK-1+1e(1:5)

dc

TREK-1 1e

TREK-1+

1e (1

:2)

TREK-1+

1e (1

:5)

TREK-1

+1e (

1:10)

Kir2.1

Kir2.1+

1e(1

:10)

0.0

-LW -LWH

-120 -80 -40 0 40

0

Voltage (mV)

e

0.2

0.4

0.6

0.8

1.0

1.2

No

rmal

ized

cu

rren

t

1e-ct / 1a-nt

1e-ct / 1b-nt

1e-ct / 1d-nt

1e-ct / Kir2 1-ct0.2

0.4

0.6

0.8

1.0

1.2

lati

ve s

urf

ace

exp

ress

ion

***

***

C

N

HA

hTREK-2

c

+1e (

1:1)

+1e (1

:3)0.0

1e-ct / Kir2.1-ct

TREK-1

-HA

+1e (

1:2)

+1e (

1:5)

0.0Rel

μ

Fig. 4 Functional characterization of TREK-1e. a Current voltagerelationship of rat TREK-1a (black), rat TREK-1e (light gray), or afterco-expression of TREK-1a and TREK-1e (ratio 1:5, gray ). bConcentration-dependent inhibitory effect of TREK-1e on TREK-1a.Kir2.1 currents were not affected by co-injection of TREK-1e cRNA(n =20). c Co-expression of TREK-1e with TREK-2c with the ratios 1:1or 1:3. Two to three independent sets of experiments were analyzed (n =10–33). d Yeast two-hybrid assay probing the direct interaction betweenthe N-termini of TREK-1a (1a-nt), TREK-1b (1b-nt), and TREK-1d (1d-nt) with the C-terminus of TREK-1e (1e-ct). Transformed HF7c yeastcells were plated on selective double dropout medium lacking leucine andtryptophan (−LW) and growth indicates positive transformation of bothconstructs. Growth of colonies on the selective −LWH medium was

scored as positive for interaction. The C-terminus of Kir2.1 (Kir2.1-ct)served as a negative control. e Surface expression analysis in oocytes.Surface expression (background subtracted) of extracellularly HA-taggedrTREK-1a injected alone (normalized to 1) or co-injected with TREK-1ein different ratios (1:2 and 1:5). Four to five independent sets ofexperiments were analyzed (n =52–76). Schematic drawing of theinsertion of the HA tag after the second pore region in the TREK-1achannel protein (at amino acid 266) is indicated in the inset. Control ofprotein expression in oocyte lysates after injection of TREK-1a-HA,TREK-1a-HA co-injected with TREK-1e (1:5), or non-injected (ni )oocytes. Proteins of the western blots were stained with HA-specificantibodies and detected with a peroxidase-conjugated secondary antibody

Pflugers Arch - Eur J Physiol

Xenopus oocytes using a luminometric assay. For this purpose,an extracellular HA tag was inserted in the extracellular P2-M4loop (amino acid position 266) of TREK-1a, as illustrated in theinset of Fig. 4e. Co-expression of TREK-1e with the HA-tagged TREK-1a channel reduced the surface expression in aconcentration-dependent manner (Fig. 4e). Relative surfaceexpression was reduced to 0.48±0.05 (ratio 1:2) or 0.32±0.03(ratio 1:5) (Fig. 4e). Western blot analysis of cell lysates usingan antibody against the extracellular HA epitope showed thatsimilar amounts of HA-tagged TREK-1a protein wereexpressed in the oocytes (Fig. 4e). We conclude from thesefindings that the reduced TREK-1 currents caused by the co-expression of TREK-1e were caused by a reduction in thesurface expression of the channel.

Reporter assays with Kir2.1 and CD8 fusion proteins showintracellular retention caused by the C-terminus of TREK-1e

We have previously used the inward rectifier channel Kir2.1and CD8 proteins for studying the effects of C-terminal motifsin determining the surface expression of potassium channels[15, 25]. For the Kir2.1 reporter assay, an extracellular HA tagwas introduced at amino acid position 117. The surfaceexpression of the constructs (Fig. 5b) was quantified with aluminometric assay in Xenopus oocytes, as describedpreviously [25, 26]. As a control experiment for the reliabilityof our surface expression assay, we truncated Kir2.1 atposition 374. This truncation (F374*), which removes thedi-acidic ER-export signal (FCYENE) starting at amino acidposition 374 [11], led to an approximately fourfold reductionof the surface expression of Kir2.1 (0.26±0.02) (Fig. 5b).Next, we fused the C-terminus of rat TREK-1e (Fig. 5a) tothe C-terminus of the Kir2.1 reporter protein (Fig. 5b). Fusionof the TREK-1e C-terminus to the Kir2.1 reporter construct(1e-ct) caused an approximately threefold reduction in surfaceexpression (0.32±0.03) (Fig. 5b). These findings support theidea that the C-terminus of TREK-1e contains a peptide motifcausing intracellular retention.

Next, we tested the possible effect of the C-terminus ofTREK-1e on intracellular transport using a different reporterassay. The C-terminus of TREK-1e was attached to the shortintracellular C-terminus of CD8 (Fig. 5c) and the resulting fusionprotein was transfected in COS-7 cells [15]. The surfaceexpression of the construct was quantified with an antibody-based luminometric assay [15, 25, 26]. As a positive control forretention, the amino acids KKTNwere attached to C-terminus ofCD8 (KKXX construct), and as a negative control, AATN wasattached (AAXX construct), as previously described [15, 26](Fig. 5c). The AAXX construct was strongly expressed at theplasma membrane (normalized, 1±0.05), whereas the KKXXconstruct was not (0.023±0.008). Fusion of the TREK-1 C-terminus to the CD8 reporter construct (1e-ct) almost completelyabolished surface expression of the CD8 fusion protein (0.02±

0.004) (Fig. 5c). Protein expression of all constructs was probedby western blot analysis, using a CD8-specific antibody (notillustrated). These results are consistent with the hypothesis thatthe C-terminus of TREK-1e contains a region that causesintracellular retention.

Identification of a critical region in the TREK-1e C-terminusfor the intracellular retention

The C-terminus of TREK-1e does not contain any sequencemotif resembling one of the canonical retention signals. Thus,to further characterize the region responsible for intracellularretention, we successively truncated the Kir2.1-HA-rTREK-1e fusion protein (construct 1e-ct) from the distal end and

a179 198coded by exon 4

1.0

1.2

sio

n

b c1.0

sio

n

179 198coded by exon 4

622302 coded by exon 6 (frame shift)SN I W T CT S S S SG S G S TL G W P RL L F

D Q L G T I F G K G I A K V E D T F I LV G R T

Kir2.1

374*

1e-c

t0.0

0.2

0.4

0.6

0.8

Rel

ativ

e su

rfac

e ex

pre

ss

KKXXAAXX

1e-c

t nt0.0

0.2

0.4

0.6

0.8

Rel

ativ

e su

rfac

e ex

pre

ss

******

*** ***

Kir2 F37 1e KK AA 1e

KK

XX

AA

XX

1e-ct

CD

8

CD

8

CD

8

HA HA HA

Kir2.1 1e-ctF374*

Fig. 5 Kir2.1 and CD8 reporter assays for surface expression. aSchematic drawing of the rat TREK-1e C-terminus. Amino acids encodedby exon 4 are shown in blue , and the new sequence due to exon 5skipping and a resulting frame shift encoded by exon 6 is illustrated inred . b Surface expression analysis after fusion of the cytosolic C-terminus of rat TREK-1e to an extracellularly HA-tagged Kir2.1 channel.The cartoon shows schematic drawings of the fusion constructs. F374*indicates a truncated Kir2.1 channel, in which the di-acidic ER-exportsignal starting at amino acid position 374 (FCYENE) was removed. Fourto 11 independent sets of experiments were analyzed (n =40–114). cFusion of the cytosolic C-terminus of rat TREK-1e to CD8 led to areduced surface expression of the reporter protein in COS-7 cells. TheER retention motif KKXX served as a positive control for retention andAAXX as a negative control. The constructs are shown in a schematicdrawing (lower illustration). Five to six independent sets of experimentswere analyzed (n =10–12)

Pflugers Arch - Eur J Physiol

measured the surface expression in Xenopus oocytes (Fig. 6a).We found that removal of the last 21 amino acids of theTREK-1e C-terminus (Δc21 construct) did not changeintracellular retention (0.39±0.09), whereas removal of oneadditional amino acid (Δc22) completely restored the surfaceexpression (1.04±0.12) (Fig. 6a). To show that comparable

amounts of proteins were expressed, western blots with anHA-specific antibody were performed (Fig. 6a, bottom).

Similar results were obtained when fusion proteins of CD8with the C-terminus of TREK-1e were transfected in COS-7cells (Fig. 6b). Again, the Δc21 construct showed only verylow surface expression (0.13±0.02) and deletion of a single

a

1 4

1.6

n ***

b1.0n

***

***0 0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Rel

ativ

e su

rfac

e ex

pre

ssio

n

******

******

***

0 0

0.2

0.4

0.6

0.8

1.0

Rel

ativ

e su

rfac

e ex

pre

ssio

n

***

***

Kir2.1

1e-c

t c9 c18

c19

c20

c21

c22

c23

c27

c36

VGRT-AAAA

RTLN-A

AAA0.0R

KKXXAAXX

1e-c

tc2

1 c2

2

IW-A

A

RTLN-A

AAA nt0.0R

cKKXX 1e-ctAAXX

c22 ntc21

10 µM

Fig. 6 Motif dissection in the TREK-1e C-terminus using Kir2.1 andCD8 reporter assays. a Kir2.1 reporter assay of truncations and aminoacid exchanges of the Kir2.1-TREK-1e fusion construct. Two to elevenindependent sets of experiments were analyzed (n =9–114). VGRT-AAAA: mutation of V198 to T201 to alanine; RTLN-AAAA : mutation ofR200 to N203 to alanine. The surface expression of the Kir2.1-TREK-1econstruct (1e-ct) was significantly reduced in comparison to Kir2.1 (p <0.001). For all other constructs, statistical significance was probed incomparison to 1e-ct. Western blot analysis of Kir2.1-TREK-1e fusionconstructs stained with HA-specific antibodies are shown at the bottom.b CD8 reporter assay in COS-7 cells confirming the results in a and

identifying residues I204 and W205 as important determinants for theretention. Two to six independent sets of experiments were analyzed (n =4–12). IW-AA : mutation of I204 andW205 to alanine. nt: non-transfectedcells. The surface expression of the CD8-TREK-1e construct (1e-ct) wassignificantly reduced compared to CD8-AAXX (p<0.001). For all otherconstructs, significance was probed in comparison to 1e-ct. Bottom :Western blot analysis of cell lysates from COS-7 transfected with theindicated CD8 fusion constructs and stained with CD8-specific antibody,indicating a similar protein expression for all constructs. c Localization ofthe CD8 TREK-1e fusion constructs in COS-7 cells using CD8-specificantibodies. The nucleus is stained in blue. nt: non-transfected cells

Pflugers Arch - Eur J Physiol

additional amino acid (Δc22) fully restored surface expression(Fig. 6b). For the Δc22 construct, the surface expression wascomparable (1±0.02) to the CD8-AAXX control constructshowing no retention (1±0.05). To show that comparableamounts of proteins were expressed, western blots with aCD8-specific antibody were performed (Fig. 6b, bottom).

The clear change in surface expression between the Δc21and Δc22 constructs was also obvious from the imagesobtained by immunohistochemistry using a CD8-specificantibody (Fig. 6c). The positive control (CD8-KKXX)showed complete intracellular retention, whereas the negativecontrol (CD8-AAXX) displayed strong surface expression(Fig. 6c). The CD8-TREK-1e fusion protein (1e-ct), like theKKXX construct, was retained in a perinuclear region, mostlikely the ER. While the CD8-TREK-1e Δc21 construct wasalso retained in the ER, the Δc22 construct was efficientlytransported to the cell surface.

To further dissect the region responsible for theintracellular retention, we mutated various residues in theregion from amino acid 198 to 205 of rTREK-1e, justproximal to the Δc21 truncation (Fig. 7). Within this region(VGRTLNIW), the amino acid substitutions, VGRT-AAAAand RTLN-AAAA, did not alter the retention properties of theTREK-1e C-terminus (Figs. 6a, b and 7). However, the doublemutation of I204 and W205 (IW-AA), which are 23 and 22amino acids away from the rTREK-1e C-terminal end,changed the retention efficiency of the C-terminus (Figs. 6band 7). These results suggest that the amino acids I204 andW205 might be involved in the retention of TREK-1e.

Figure 7 summarizes the mutations of the TREK-1e C-terminus studied with the two reporter protein assays (Kir2.1and CD8 fusion proteins). Successive truncation in the TREK-1e C-terminus retained the ER localization of the fusionproteins (Δc9, Δc18, Δc19, Δc20, Δc21) in both assays.However, deletion of one further amino acid (Δc22 construct)resulted in a localization at the plasma membrane with bothassays (Fig. 7). Further truncations until deletion of 36 aminoacids (Δc36 construct) also showed localization to the plasmamembrane (Fig. 7). The pronounced effect of the IW-AAmutation, which is just proximal to the transition pointbetween ER localization (Δc21) and surface expression(Δc22), indicates that residues I204 and W205 are part ofthe critical region responsible for ER localization.

Discussion

We have identified a novel splice variant of TREK-1 which wenamed TREK-1e. The classical splice variants TREK-1a-dhave an alternatively spliced exon 1 and thus differ in theiramino terminal sequence. These splice variants all possess thecharacteristic K2P channel topology of four transmembranedomains and two-pore-forming domains. In contrast, TREK-

1e lacks the last two transmembrane domains, due to skippingof exon 5 and a shift in the open reading frame of exon 6,resulting in a premature stop codon. Expression of TREK-1eproduced no current, and the channel protein was found to belocalized to the ER. Co-expression of TREK-1e with full-length TREK-1 isoforms caused a decrease in TREK-1 currentamplitude, due to a reduction in surface expression. Thus, theeffect of TREK-1e was caused by a change in the intracellulartraffic of the classical TREK-1 channels. Therefore, it is likelythat TREK-1e interacts with the full-length splice variants atsome point in the biosynthetic pathway.

The intracellular retention caused by TREK-1e could betransferred to the reporter proteins Kir2.1 and CD8 (assayed inXenopus oocytes and in COS-7 cells, respectively) by fusing theC-terminus of TREK-1e with these proteins. These findingsindicate that the C-terminus of TREK-1e contains a criticalamino acid sequence responsible for the intracellular retention.We dissected the C-terminal sequence of TREK-1e by successivetruncations and by alanine mutations, using surface expressionassays with Kir2.1 and CD8 reporter proteins. As summarized inFig. 7, we found that the amino acids I204 and W205 play amajor role for the ER retention of TREK-1e. The region betweenamino acids 202 and 207 of TREK-1e (LNIWTS) is conservedbetween human and rat (Fig. 1b). However, fusion of the IWTSmotif to the CD8 reporter protein did not result in ER retention(not illustrated). This may be attributable to the fact that theamino acid context and the distance of these residues to theplasmamembranemay also be relevant, as is the case with RxRand lysine-based retention signals [24]. Despite the fact that wehave identified a critical region responsible for the ER retention,the precise mechanism leading to the ER retention is stillunclear. The simplest interpretation of our results is that themotif containing the sequence IW may interact with the COPIcoatomer or with another (unknown) protein. Similar short andunusual motifs for ER retention were also found in other ionchannel splice variants. In the case of BK channels, alternativesplicing of the SV1 variant generates an unconventional ERlocalization signal, CVLF, leading to a retention of the isoformin the ER. Another C-terminal splice variant of BK channels,the DEC variant, is primarily localized in the ER and is, likeTREK-1e, capable of reducing the surface expression of otherchannel isoforms [10].

Now that the basic architecture of K2P channels is known [3,12], we can speculate what the TREK-1/TREK-1e heteromersmight look like. It is unlikely that classical TREK-1 splice variantsand TREK-1e can form a trimer consisting of one TREK-1subunit and two TREK-1e subunits because of the resulting grossasymmetry: The protein would have three M1–P1 loops. Thesimplest, and probably most likely, case is that classical TREK-1subunits and TREK-1e form a heterodimer which is not afunctional channel because it has only three pore domains. Thesemisfolded dimeric proteins might then be retained in theendoplasmic reticulum by means of its exposed retention signal,

Pflugers Arch - Eur J Physiol

similar to the classical quality control mechanism describedpreviously [16, 23]. While it is well known that the alpha-helices of the M1–P1 linker play a role in dimerization of K2P

channels, this is certainly not the only interacting region. In mostpotassium channels, the formation of the cytosolic pore involvesan interaction betweenN- andC-termini, and in view of their longC-terminal domains, this may also apply to K2P channels. To testthe idea that one TREK-1e subunit may form a dimer withclassical TREK-1, we asked the question whether the “new” C-terminus can directly interact with the N-terminus of TREK-1a,TREK-1b, and TREK-1d, and our yeast-two hybrid analysisshowed that this was indeed the case (Fig. 4d). These findingsare consistent with the idea that TREK-1/TREK-1e dimers areassembled in the ER via interacting domains in the extracellularcap and in the cytosolic loops and retained via the IWTSmotif ofTREK-1e. However, the possibility that TREK-1e, via its C-terminus, associates with the cytosolic domains of homo-dimeric TREK-1 channels cannot be excluded.

Alternative splicing was also found in other K2P channels,including TREK-2 and TRAAK of the mechano-sensitive K2P

subgroup [6, 7, 9, 13, 14, 20]. Similar as for TREK-1, thesplicing of alternative exon 1 generates TREK-2 and TRAAKisoforms with varying amino terminal sequences. It is not yetclear why the short N-terminus of variousK2P channels is subjectto such extensive and diverse alternative splicing. Since some ofthese K2P channels are also subject to functionally relevantalternative translation initiation (ATI) [17, 19], it is likely thatthe alternative N-terminal splicing might interfere with orregulate ATI [18].

The TREK-1e variant is found in combination with differentexons 1. In rat, TREK-1e is spliced in combination with exon1a; in humans, it is combined with exon 1c; and inchimpanzees, it can be found in public databases in associationwith exon 1b. This suggests that the splicing-associated loss ofexon 5, which generates TREK-1e, can occur independently ofthe transcript variant of exon 1 (TREK-1a-d).

For the K2P channel TALK-1, alternative splicing generatesa channel protein with three transmembrane segments [7]. Thissplice variant was found to have no channel activity by itselfand to have no influence on the single channel conductance ofTALK-1 in co-expression experiments [7]. Veale et al. recentlydescribed a TREK-1 splice variant, TREK-1ΔEx4, whichencodes a heavily truncated protein retaining a singletransmembrane domain [20]. This variant also had no channelactivity by itself, but reduced TREK-1 whole-cell currentamplitude. Wu et al. found TREK-1 variants in myometriumfrom women in preterm labor. These were lacking either thepore domain or the transmembrane domains or both [21].Whilethe splice variants described by Wu et al. were shown tophysically interact with TREK-1, there is no data available toindicate whether these variants alter the electrophysiologicalproperties of TREK-1. The presence of these truncated splicevariants seems to be a mechanism to alter the functionaldiversity of TREK-1 currents in different tissues. Consideringall these examples mentioned above, it is tempting to speculatethat the modulation of channel trafficking by alternativesplicing may be a common mechanism for regulating surfaceexpression of ion channels.

TREK-1e is preferentially expressed in the brain, kidney,adrenal gland, and amygdala. At present, it is not known onwhich factors the expression level of this splice variantdepends and how it might be regulated. Our results suggestthat in cells expressing a relatively high amount of TREK-1eprotein, the intracellular transport of classical TREK-1channel isoforms between the ER and the plasma membranemay be impeded by interaction with TREK-1e. In this way,the copy number of functional TREK-1 channels at the cellsurface might be modulated in a tissue-specific manner.

Acknowledgments This work was supported by grants of the DeutscheForschungsgemeinschaft SFB 593, TPA4 to J.D. and 1482-3/2 to N.D.,the P.E. Kempkes Foundation (to S.R. and R.P.M.), and the Anneliese

ERlocalization

PMKir2.1 CD8reporter protein

TREK-1e ctc9c18c19c20c21

502891971

VGRTLNIW

c21c22c23c27c36

VGRT-AAAA AAAA

RTLN-AAAAIW-AA

AAAA

AAn.t.

Fig. 7 Summary of the motifdissection. Illustration of theTREK-1e truncations (countedfrom the distal C-terminus) oralanine mutants analyzed usingthe reporter proteins Kir2.1 orCD8. The localization of therespective construct is indicatedby endoplasmic reticulum (ER)and plasma membrane (PM). n.t.:not tested

Pflugers Arch - Eur J Physiol

Pohl Stiftung to S.R.. We thank Caroline Rolfes and Thorsten Steinfeldtfor the isolation of porcine adrenal gland. We are grateful to AndreaSchubert for excellent technical support.

Conflict of interest The authors declare no conflict of interest.

References

1. Bittner S, Ruck T, Schuhmann MK, Herrmann AM, Maati HM,Bobak N, Gobel K, Langhauser F, Stegner D, Ehling P, BorsottoM, Pape HC, Nieswandt B, Kleinschnitz C, Heurteaux C, Galla HJ,Budde T, Wiendl H, Meuth SG (2013) Endothelial TWIK-relatedpotassium channel-1 (TREK1) regulates immune-cell trafficking intothe CNS. Nat Med 19:1161–1165. doi:10.1038/nm.3303

2. Bockenhauer D, Zilberberg N, Goldstein SA (2001) KCNK2: reversibleconversion of a hippocampal potassium leak into a voltage-dependentchannel. Nat Neurosci 4:486–491. doi:10.1038/87434

3. Brohawn SG, del Marmol J, MacKinnon R (2012) Crystal structureof the human K2P TRAAK, a lipid- and mechano-sensitive K+ ionchannel. Science 335:436–441. doi:10.1126/science.1213808

4. Fink M, Duprat F, Lesage F, Reyes R, Romey G, Heurteaux C,Lazdunski M (1996) Cloning, functional expression and brainlocalization of a novel unconventional outward rectifier K+ channel.EMBO J 15:6854–6862, PMID: 9003761

5. Franks NP, Honore E (2004) The TREK K2P channels and their rolein general anaesthesia and neuroprotection. Trends Pharmacol Sci 25:601–608. doi:10.1016/j.tips.2004.09.003

6. GuW, Schlichthörl G, Hirsch JR, Engels H, Karschin C, Karschin A,Derst C, Steinlein OK, Daut J (2002) Expression pattern andfunctional characteristics of two novel splice variants of the two-pore-domain potassium channel TREK-2. J Physiol 539:657–668.doi:10.1113/jphysiol.2001.013432

7. Han J, Kang D, Kim D (2003) Functional properties of four splicevariants of a human pancreatic tandem-pore K+ channel, TALK-1. AmJ Physiol Cell Physiol 285:C529–C538. doi:10.1152/ajpcell.00601.2002

8. Kim D (2005) Physiology and pharmacology of two-pore domainpotassium channels. Curr Pharm Des 11:2717–2736. doi:10.2174/1381612054546824

9. Kim Y, Bang H, Kim D (1999) TBAK-1 and TASK-1, two-pore K+

channel subunits: kinetic properties and expression in rat heart. Am JPhysiol 277:H1669–H1678, PMID: 10564119

10. MaD, Nakata T, Zhang G, Hoshi T, LiM, Shikano S (2007) Differentialtrafficking of carboxyl isoforms of Ca2+-gated (Slo1) potassiumchannels. FEBS Lett 581:1000–1008. doi:10.1016/j.febslet.2007.01.077

11. Ma D, Zerangue N, Lin YF, Collins A, YuM, Jan YN, Jan LY (2001)Role of ER export signals in controlling surface potassium channelnumbers. Science 291:316–319. doi:10.1126/science.291.5502.316

12. Miller AN, Long SB (2012) Crystal structure of the human two-poredomain potassium channel K2P1. Science 335:432–436. doi:10.1126/science.1213274

13. Ozaita A, Vega-Saenz de Miera E (2002) Cloning of two transcripts,HKT4.1a and HKT4.1b, from the human two-pore K+ channel gene

KCNK4 . Chromosomal localization, tissue distribution andfunctional expression. Brain Res Mol Brain Res 102:18–27

14. Patel AJ, Maingret F, Magnone V, Fosset M, Lazdunski M, Honore E(2000) TWIK-2, an inactivating 2P domain K+ channel. J Biol Chem275:28722–28730. doi:10.1074/jbc.M003755200

15. Renigunta V, Yuan H, Zuzarte M, Rinné S, Koch A, WischmeyerE, Schlichthörl G, Gao Y, Karschin A, Jacob R, Schwappach B,Daut J, Preisig-Müller R (2006) The retention factor p11 confersan endoplasmic reticulum-localization signal to the potassiumchannel TASK-1. Traffic 7:168–181. doi:10.1111/j.1600-0854.2005.00375.x

16. Schutze MP, Peterson PA, Jackson MR (1994) An N-terminaldouble-arginine motif maintains type II membrane proteins in theendoplasmic reticulum. EMBO J 13:1696–1705

17. Simkin D, Cavanaugh EJ, Kim D (2008) Control of the singlechannel conductance of K2P10.1 (TREK-2) by the amino-terminus:role of alternative translation initiation. J Physiol 586:5651–5663.doi:10.1113/jphysiol.2008.161927

18. Staudacher K, Baldea I, Kisselbach J, Staudacher I, Rahm AK,Schweizer PA, Becker R, Katus HA, Thomas D (2011) Alternativesplicing determines mRNA translation initiation and function ofhuman K2P10.1 K+ channels. J Physiol 589:3709–3720. doi:10.1113/jphysiol.2011.210666

19. Thomas D, Plant LD, Wilkens CM, McCrossan ZA, Goldstein SA(2008) Alternative translation initiation in rat brain yields K2P2.1potassium channels permeable to sodium. Neuron 58:859–870. doi:10.1016/j.neuron.2008.04.016

20. Veale EL, Rees KA, Mathie A, Trapp S (2010) Dominantnegative effects of a non-conducting TREK1 splice variantexpressed in brain. J Biol Chem 285:29295–29304. doi:10.1074/jbc.M110.108423

21. Wu YY, Singer CA, Buxton IL (2012) Variants of stretch-activatedtwo-pore potassium channel TREK-1 associated with preterm laborin humans. Biol Reprod 87:96. doi:10.1095/biolreprod.112.099499

22. Xian Tao L, Dyachenko V, Zuzarte M, Putzke C, Preisig-Müller R,Isenberg G, Daut J (2006) The stretch-activated potassium channelTREK-1 in rat cardiac ventricular muscle. Cardiovasc Res 69:86–97.doi:10.1016/j.cardiores.2005.08.018

23. Zerangue N, Schwappach B, Jan YN, Jan LY (1999) A new ERtrafficking signal regulates the subunit stoichiometry of plasmamembrane KATP channels. Neuron 22:537–548

24. Zerangue N, Malan MJ, Fried SR, Dazin PF, Jan YN, Jan LY,Schwappach B (2001) Analysis of endoplasmic reticulum traffickingsignals by combinatorial screening in mammalian cells. Proc NatlAcad Sci U S A 98:2431–2436. doi:10.1073/pnas.05163019898/5/2431

25. Zuzarte M, Rinné S, Schlichthörl G, Schubert A, Daut J, Preisig-Müller R (2007) A di-acidic sequence motif enhances the surfaceexpression of the potassium channel TASK-3. Traffic 8:1093–1100.doi:10.1111/j.1600-0854.2007.00593.x

26. Zuzarte M, Heusser K, Renigunta V, Schlichthörl G, Rinné S,Wischmeyer E, Daut J, Schwappach B, Preisig-Müller R (2009)Intracellular traffic of the K+ channels TASK-1 and TASK-3: role ofN- and C-terminal sorting signals and interaction with 14-3-3 proteins. JPhysiol 587:929–952. doi:10.1113/jphysiol.2008.164756

Pflugers Arch - Eur J Physiol