KCNQ1/KCNE1 potassium channels in mammalian vestibular dark cells

KCNQ1/KCNE1 potassium channels in mammalian vestibular dark cells

Marie-There©se Nicolas a;*, Danielle Demeªmes a, Agne©s Martin a, Sabina Kupershmidt b,Jacques Barhanin c

a INSERM U432, Universite Montpellier 2, Place Euge©ne Bataillon, P.O. Box 089, 34095 Montpellier Cedex 05, Franceb Vanderbilt University School of Medicine, Department of Pharmacology, Nashville, TN 37235-6602, USA

c Institut de Pharmacologie Moleculaire et Cellulaire, CNRS, 660 Route des Lucioles, 06560 Valbonne, France

Received 21 August 2000; accepted 27 November 2000

Abstract

The high [K�] in the inner ear endolymph is essential for mechanosensory transduction in hearing and balance. Several ionchannels, including a slowly activating, voltage-dependent, outwardly conducting K� channel composed of the KCNQ1 (KvLQT1)and KCNE1 (IsK/minK) subunits, are expressed at the apical surface of vestibular dark cells. We investigated the underlyingmolecular mechanisms of this conductance using in situ hybridization, RT-PCR, and immunocytochemistry and by tracking theultrastructural changes of vestibular structures in kcne1(3/3) mice. In the wild type mice, the KCNE1 and KCNQ1 proteins areexpressed specifically at the apical membrane of dark cells, as early as gestational day (GD) 17 for KCNE1 while KCNQ1 mRNAscan be detected at GD 18. This is the first demonstration that the two protein components of this potassium channel co-localize in apolarized fashion at the cellular level. Although the vestibular end-organs are normal at birth in kcne1(3/3) mice, they begin toshow modifications during postnatal development: we observed an increase in the height of the dark cells, in their number ofmitochondria, and in basolateral membrane infoldings. Subsequently, the epithelium degenerates and the endolymphatic spacecollapses. Similar changes are known to occur in the cardio-auditory Jervell^Lange-Nielsen syndrome which is caused by mutationsin the same channel. ß 2001 Published by Elsevier Science B.V.

Key words: Dark cell ; KCNQ1/KCNE1 (KvLQT1/IsK); Knock-out mouse; Immunocytochemistry; In situ hybridization;Electron microscopy

1. Introduction

Endolymph is the extracellular £uid bathing mecha-noreceptor sensory cells of the inner ear. It is uniqueamong extracellular £uids, in that the major salt is notNaCl but KCl. The concentration of K� is unusuallyhigh, around 150 mM and close to the intracellular[K�], whereas that of Na� is low, around 10 mM (Fer-rary and Sterkers, 1998). In vivo, the maintenance of ahigh outside potassium concentration is essential in or-

der to ensure good mechanotransduction which is car-ried by K� through the sensory hair cells. Numeroushearing and balance disorders are due to disturbancesin endolymph secretion and composition (Barhanin etal., 1999; Cooper and Jan, 1999; Steel and Bussoli,1999). It is therefore important to understand how en-dolymph is secreted and how its ionic composition isregulated.

All vestibular sensory organs except the saccule pos-sess areas of dark cells with intense enzymatic activities(Sterkers et al., 1988) and characterized at the ultra-structural level by numerous basal membrane infoldings(Kawamata et al., 1986). They are situated at the baseof each crista ampullaris (Fig. 1) and around the utricle.The extent of these areas is variable but is particularlywide between the utricle and the lateral and anteriorcristae (Kimura, 1969). The dark cells in the vestibularsystem secrete endolymphatic K� (Marcus and Marcus,

0378-5955 / 01 / $ ^ see front matter ß 2001 Published by Elsevier Science B.V.PII: S 0 3 7 8 - 5 9 5 5 ( 0 0 ) 0 0 2 6 8 - 9

* Corresponding author. E-mail: [email protected]

Abbreviations: DEPC, diethyl pyrocarbonate; DIG, digoxigenin;GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GD, gestation-al day; IKs, slowly activating potassium channel; LQTS, long QTsyndrome; PD, postnatal day; PB, phosphate bu¡er; PBS, phos-phate-bu¡ered saline solution; RT-PCR, reverse transcription-poly-merase chain reaction; SSC, saline sodium citrate; WT, wild type

HEARES 3634 15-2-01

Hearing Research 153 (2001) 132^145

www.elsevier.com/locate/heares

1987; Ferrary et al., 1989; Marcus and Shen, 1994;Marcus and Shipley, 1994; Ferrary and Sterkers, 1998).

The identi¢cation of the individual K� channel(s)responsible for endolymphatic K� secretion has beenthe subject of intense research. Electrophysiologicalstudies have shown that a non-selective cation current(Marcus et al., 1992), maxi-K� conductances (Takeuchiet al., 1992) and a slow, voltage-dependent, delayedrecti¢er K� current (Marcus and Shen, 1994; Marcusand Shipley, 1994) are all present in the apical mem-brane of dark cells. The latter current shows the bio-physical properties of cardiac slowly activating potassi-um channels (IKs) and is involved in secretion of K�

into the endolymph (Shen and Marcus, 1998). Recentmolecular genetic studies have identi¢ed the KCNQ1(also called KvLQT1) (Wang et al., 1996) andKCNE1 (also called IsK or minK) (Takumi et al.,1988) gene products as the components of the IKs chan-nel (Barhanin et al., 1996; Sanguinetti et al., 1996). Inhumans, loss of function through mutation in either ofthe two genes is known to cause the congenital long QTsyndrome (LQTS) which may lead to syncope and sud-den cardiac death (reviewed in Barhanin et al., 1998). Inaddition, mutation of either KCNQ1 (Neyroud et al.,1997; Splawski et al., 1997) or KCNE1 (Schulze-Bahr etal., 1997; Tyson et al., 1997) can cause the recessiveform of LQTS, Jervell^Lange-Nielsen syndrome, a dis-ease characterized by cardiac arrhythmias and bilateraldeafness (Jervell and Lange-Nielsen, 1957; Friedmannet al., 1966).

We investigated the subcellular distribution and theonset of appearance of the IKs subunits KCNE1 andKCNQ1 during the development of vestibular receptorsin mice and rats. We used reverse transcription-poly-merase chain reaction (RT-PCR), in situ hybridizationand immunocytochemistry. In addition, we performeddetailed ultrastructural analysis of the postnatalchanges in kcne1(3/3) knock-out mice (Vetter et al.,1996) to assess the e¡ects of the de¢cit in the kcne1gene on the vestibular epithelial structures.

2. Materials and methods

2.1. Animals

Swiss and 129 Sv wild type (WT) mice, from gesta-tional day (GD) 16 to postnatal day (PD) 24 andSprague^Dawley rats between PD 3 and PD 6 wereused for molecular biology and immunocytochemistryexperiments. The animals were housed and handled ac-cording to the approved guidelines of the French De-partment of Agriculture, Forestry and Education.kcne1(3/3) knock-out mice (Vetter et al., 1996) wereraised at the Institut de Pharmacologie Moleculaire et

Cellulaire (CNRS, Valbonne) in full compliance withthe French Government animal welfare policy. Animalswere anesthetized with sodium pentobarbital (i.p. 60mg/kg), followed by either decapitation, ¢xation, ortranscardial perfusion (as described below) dependingon age. The vestibular end-organs were then dissectedout in the ¢xative.

2.2. Transmission electron microscopy

For electron microscopy studies, mouse inner earspecimens were collected on PD 0 (day of birth), PD7, PD 10, and PD 69. Dissected vestibules were ¢xedovernight at 4³C in 2% glutaraldehyde in 0.1 M phos-phate bu¡er (PB) pH 7.2. After rinsing in 0.2 M PB, thesamples were post¢xed in 2% osmium tetroxide in 0.1 MPB, for 1 h, at room temperature, then washed in PB,transferred quickly to distilled water, and dehydrated ingraded ethanol solutions, impregnated with epoxy resin(Araldite 502/EMbed 812, EMS-Tebu, France), ori-ented under a dissecting microscope, and embedded at60³C for 24 h, before polymerization. Thin sectionswere cut with a diamond knife on an ultramicrotome(Reichert OMU2, Vienna, Austria) and collected onFormvar-covered copper grids, stained with uranyl ace-tate and lead citrate and observed with a Jeol JEM 1200Ex II transmission electron microscope (Central Labo-ratory of Electron Microscopy, University of Montpel-lier II, France).

2.3. Immunocytochemistry

For immunocytochemistry experiments, samples were¢xed in 4% paraformaldehyde in 0.1 M phosphate-bu¡-ered saline solution (PBS) pH 7.2, for 24 h at 4³C. Thevestibular end-organs were carefully dissected out undera dissecting microscope and rinsed in PBS. They wereembedded in 3.75% agarose in PBS and cut into 60 Wmthick sections with a vibratome (TPI, Series 1000).

Two primary a¤nity-puri¢ed polyclonal antibodieswere used: one speci¢cally directed against KCNE1(Lesage et al., 1993), the other against KCNQ1. Anti-KCNQ1 antibodies were generated using the services ofQuality Controlled Biochemicals, Inc. (QCB, Hopkin-ton, MA, USA). The immunogen was the syntheticpeptide acetyl-CVAKKKFQQARKPYDVRD-amidecoupled to keyhole limpet hemocyanin via a male-imide-cysteine reaction and puri¢ed to s 90% usingHPLC. Rabbits were immunized with the peptide andbled ¢ve times at 2-week intervals. Sera were tested forreactivity using ELISA followed by Western blot assayson crude membrane extracts from mouse heart and liv-er. Membrane extracts and Western blotting were essen-tially done as described in Deal et al. (1994) except thatthe tissues were ground up in bu¡er (0.32 M sucrose,

HEARES 3634 15-2-01

M.-T. Nicolas et al. / Hearing Research 153 (2001) 132^145 133

5 mM Na2HPO4, pH 7.4 with protease inhibitors) usinga Polytron before using the Dounce homogenizer. Us-ing this assay, a serum (1104.2) was identi¢ed whichresulted in a band of the expected size (about 62kDa) against extract from heart, but not liver, whichwas then immunoa¤nity-puri¢ed on a thiol couplinggel. In addition, the speci¢city of the immunoa¤nity-puri¢ed 1104.2 antiserum was tested in Western blots,where a band of the expected size was detected with theimmune serum, but not with the preimmune serum andin protein extracts from yeast cells expressing the intra-cytoplasmic C-terminus of KCNQ1 (which contains theepitope against which the serum was raised) fused inframe to the Gal4 DNA binding domain of yeast.The same band was detected by antiserum directedagainst the Gal4 DNA binding domain further indicat-ing the correct expression and detection of the Gal4^KCNQ1 protein fusion.

The secondary antibodies were Texas red-conjugateda¤nity pure donkey anti-rabbit IgG (Jackson Labora-tories, PA, USA). All the following steps were done at4³C under gentle rotation. Free £oating vibratome sec-tions were preincubated for 1.5 h in a blocking bu¡ercontaining 10% normal donkey serum, to minimizenon-speci¢c immunolabelling, and 0.3% Triton X-100in PBS, to enhance reagent penetration. The sectionswere then incubated overnight with the primary anti-body: either anti-KCNE1 antibody diluted 1/2000 inPBS containing 1% normal serum, or anti-KCNQ1antibody diluted 1/250 in PBS containing 1% normalserum and 0.1% Triton X-100. The samples were rinsedin PBS (4U15 min) and the secondary antibody diluted1/100 in PBS was applied for 1 h, at 4³C, and in the

dark. The sections were rinsed in PBS (4U15 min),mounted in antifading FluorSave Reagent (Calbio-chem, San Diego, CA, USA) and observed with a laserscanning confocal microscope (Bio-Rad MRC 1024)mounted on a Zeiss, Axiovert 100TV microscope,with the oil objective U40.

2.4. RT-PCR

Mouse labyrinths were isolated by microdissection insterile conditions, at 4³C, then frozen at 380³C beforeRNA extraction. For the standard two-step method,dissected labyrinths (utricular maculae and cristae)from 25^30 inner ears were pooled. Total RNA (frompieces of heart and from labyrinths) was extracted atroom temperature, using the RNeasy kit (Qiagen). Theextracted RNA was collected in 40 Wl of sterile diethylpyrocarbonate (DEPC)-treated water, and stored at380³C. Two sets of primers (synthesized by Eurogen-tec, Belgium) were used: one for KCNQ1 transcripts,and one for glyceraldehyde-3-phosphate dehydrogenase(GAPDH) transcripts for quality control. The primersequences for KCNQ1 were 5P-CTG-AGA-AAG-ATG-CGG-TGA-AC-3P for upstream and 5P-TGG-GGG-TCA-GCA-GTG-TCT-CC-3P for downstream.The sequences for GAPDH were 5P-ATG-ACA-ACT-TTG-GCA-TCG-TG-3P for the upstream and 5P-GAA-TGG-GAG-TTG-CTG-TTG-AAG-3P for the down-stream primer. Total RNA from heart, a known sourceof KCNQ1 transcripts (Barhanin et al., 1996; Sangui-netti et al., 1996), was used as positive control. RT wasperformed, at 50³C for 30 min, with Superscript II en-zyme using oligo-dT for a speci¢c antisense primer(Gibco, Life Technology, France). RT controls includedleaving the Superscript II enzyme out of the reactionmixtures. Standard methods were used for PCR (Dief-fenbach and Dveksler, 1995).

To analyze single animals (two vestibular labyrinthspooled), we used one-step RT-PCR. RT and PCR re-actions were done using the Titan kit (Boehringer-Mannheim, Germany) according to the manufacturer'sinstructions. PCR products (15 Wl) plus 5 Wl of orangeethidium bromide were analyzed by 2% agarose (GibcoBRL) gel electrophoresis and ethidium bromide stain-ing. Molecular weight markers (100 bp DNA ladder)were from Promega (Madison, WI, USA).

2.5. In situ hybridization

The KCNQ1 probe was a riboprobe, labeled by in-corporation of dUTP coupled to digoxigenin (DIG).The KCNQ1 cDNA construct plasmid (Barhanin etal., 1996) was linearized with NcoI and puri¢ed withthe High Pure PCR product puri¢cation kit (Boehr-inger Mannheim). Probes were prepared with the

Fig. 1. Mouse crista ampullaris showing the topological localizationof the dark cell area, on each side of the basal part of a crista. Theupper part of the crista is constituted by the sensory epithelium.The transitional epithelium extends between the dark cells and thesensory cells. Semithin section, stained with toluidine blue. Scalebar, 10 Wm.

HEARES 3634 15-2-01

M.-T. Nicolas et al. / Hearing Research 153 (2001) 132^145134

DIG-UTP labeling kit (Boehringer Mannheim) accord-ing to the manufacturer's instructions using T7 (sense)and T3 (antisense) polymerase transcription enzymes.The probe was puri¢ed with the High Pure PCR prod-uct puri¢cation kit and the amount of the probe wasestimated with DIG-quanti¢cation test strips (Boehr-inger Mannheim). The length of the probe was 786 bp.

A technique was developed for in situ hybridizationusing the entire vestibular labyrinth dissected understerile conditions. Prehybridization was done in PBS-1% Tween in 2Usaline sodium citrate (SSC), 1UDen-hardt's solution and 0.2Usarcosyl made up in DEPC-treated water, for 1 h at room temperature. Hybridiza-tion was done with 4000 pg of the probe diluted in asolution of 50% deionized formamide and 50%RPN3310 bu¡er (Amersham), overnight, at 56³C, ina humid atmosphere. After hybridization, sampleswere washed in: 5USSC at 50³C for 2 h, 0.2USSC at

50³C for 1 h, and 0.2USSC at room temperature for 5min. An anti-DIG-alkaline phosphatase polyclonalantibody, diluted 1/500 (Boehringer Mannheim), wasadded. Phosphatase activity was revealed by overnightincubation in the dark, at room temperature, with 4-nitroblue tetrazolium chloride (100 mg/ml)^x-phos-phate/5-bromo-4-chloro-3-indolyl-phosphate (50 mg/ml) and levamisole (0.24 mg/ml) at pH 9.5. The sampleswere rinsed in water, embedded in OCT compound(Sakura, Torrance, CA, USA) and frozen in isopentanecooled at 340³C by liquid nitrogen. The specimenswere then cut with a cryostat (Leica CM 3050) andfrozen 18 Wm thick sections were collected on sterileProbeOn Plus slides. Slides were equilibrated at roomtemperature and were mounted in FluorSave medium(Calbiochem, San Diego, CA, USA). The labeled senseprobe was used as a control for the speci¢city of thehybridization signals.

Fig. 2. Immunocytochemical localization of KCNE1 protein in mouse vestibular cristae, observed by confocal microscopy. (A) At the base ofeach side of a crista, dark cells show intense staining, whereas the other parts of the crista, sensory and transitional epithelium, are devoid ofany labeling. (B) Optical section showing that the labeling is restricted to the apical membrane of the dark cells, facing the endolymph. The cy-toplasm and the basal infoldings are not labeled. (C) Reconstructed 18 Wm section of a dark cell area at the base of a crista, showing their un-dulating apical surface. Scale bars, 50 Wm (A); 10 Wm (B,C).

HEARES 3634 15-2-01


3. Results

3.1. Distribution of the KCNE1 channel subunit

In adult 129 Sv WT mice, as well as in adult rats, weobserved speci¢c localization of KCNE1 immunoreac-tivity to the apical membrane of dark cells, which canbe seen at the base of the cristae ampullares (Fig. 2) andaround the utricle. Staining was intense and highly spe-ci¢c for the dark cell membranes facing the endolym-phatic space. The cytoplasm and the basolateral mem-

brane of the dark cells were devoid of labeling as werethe other cell types of the vestibular end-organs, i.e.sensory hair cells, supporting cells, and transitionalcells. We visualized the precise localization of theKCNE1 protein at the apical membrane using an 18Wm reconstructed section which showed that the entiresurface of each dark cell was homogeneously labeled(Fig. 2C).

As negative controls, we included sections incubatedwithout primary anti-KCNE1 antibody or with normalserum, neither of which showed labeling. Moreover, no

Fig. 3. KCNE1 immunolabeling during mouse embryonic development. Confocal microscopy. (A) In this optical section on GD 16, noKCNE1 labeling is detectable. (B) On GD 17, KCNE1 immunolabeling is visible on the apical side of the dark cells. Scale bar, 10 Wm.

Fig. 4. Detection of KCNQ1 mRNA by RT-PCR in the vestibular end-organs. (A) Two-step RT-PCR: cDNA product from KCNQ1 mRNA,of 6-day-old mice, in vestibule (lane 1) and in heart, the positive control (lane 2). In both lanes, a band at 631 bp is present. cDNA productfrom the GAPDH control reaction in heart (lane 3). (B) One-step RT-PCR in one vestibular end-organ of an 8-day-old mouse. cDNA pro-duced from KCNQ1 (lane 1) and GAPDH (lane 3) transcripts. Blank RT control (lane 2). (C) One-step RT-PCR on vestibular end-organ ofone animal on GD 18. cDNA products from KCNQ1 (lane 1) and GAPDH (lane 2) transcripts. Lane M: molecular weight markers.

HEARES 3634 15-2-01


labeling of the dark cell area was observed when weused the anti-KCNE1 antibodies in kcne1(3/3)knock-out mice.

In order to determine the onset of appearance of theKCNE1 protein during mouse inner ear development,we studied sections at di¡erent developmental timepoints. We detected no KCNE1 immunostaining onGD 16 in the tiny crista ampullaris (Fig. 3A), but asignal was detected on GD 17 in a few cells at thebase of the crista ampullaris (Fig. 3B). Ultrastructuralobservation of the dark cell area on GD 17 showed thatthe cells did not present the numerous basolateral mem-brane infoldings characteristic of mature dark cells. Atbirth, the labeling of the apical membrane of the darkcells was almost as intense as on PD 15.

3.2. Distribution of the KCNQ1 channel subunit

We tested the vestibular system for KCNQ1 mRNA

transcripts using RT-PCR. Both the one- and two-stepRT-PCR methods, using RNA from isolated labyrinthsfrom 6^8-day-old mice, resulted in a single 631 bpband, identical to that detected in the heart (positivecontrol) (Fig. 4A,B). KCNQ1 mRNA was detected onGD 18 (Fig. 4C), the earliest age tested. In the absenceof RT enzyme, no transcription and ampli¢cation wereobserved.

In addition, we used non-radioactive in situ hybridi-zation to localize KCNQ1 message at the cellular level.We observed a strong mRNA signal in the cytoplasm ofthe dark cells which was restricted to their apical side(Fig. 5A,B) Controls, using a sense riboprobe, synthe-sized with T7 DNA polymerase, revealed no positivesignal in the dark cells.

In order to determine KCNQ1 expression at the pro-tein level, i.e. KCNQ1 protein, we used a rabbit poly-clonal anti-KCNQ1 antibody for immunocytochemis-try. Like the KCNE1 protein, KCNQ1 was localized

Fig. 5. (A) Detection of KCNQ1 transcripts by in situ hybridization, with a DIG-labeled riboprobe, in a 12-day-old mouse vestibular end-or-gan showing the intense labeling in the dark cell area, between the arrows, next to the utricle (u). (B) Magni¢cation of the dark cell area fromA: the cytoplasm of the dark cells, located in the apical part, is heavily stained. The basal part corresponding to the basolateral membrane in-foldings appear unstained. Scale bars, 10 Wm (A); 5 Wm (B).

Fig. 6. (A) Immunolocalization of KCNQ1 in the mouse vestibular end-organs of a 15-day-old mouse. Confocal microscopy. At the base ofthe crista ampullaris, the dark cells present intense labeling on their apical side, facing the endolymph. Scale bar, 50 Wm. (B) Reconstructed 18Wm section of a dark cell area at the base of a crista ampullaris labeled with anti-KCNQ1 antibodies. Note that the labeling is restricted at theapical membrane of the dark cells. Scale bar, 10 Wm.

HEARES 3634 15-2-01


at the apical surface of the dark cells in WT mice (Fig.6A,B), as well as in rats on PD 15. Sometimes, raredark cells appeared not to be labeled although the ap-ical side of all the others was intensely positive. The¢rst stage at which we stained for, and detected,KCNQ1 protein was at birth. Leaving out the anti-KCNQ1 antibody (negative control) resulted in abroga-tion of the signal.

3.3. Postnatal morphological changes of the vestibularend-organs in the KCNE1 knock-out mouse

We focused our studies on the postnatal ultrastruc-tural changes of dark cells and sensory cells in thekcne1(3/3) mutant.

The vestibular end-organs in the kcne1(3/3) knock-out mouse looked completely normal at birth, but ma-jor abnormalities appeared soon thereafter. Dark cellsincreased in height by 3^5 times (Fig. 7B) compared toWT animals (Fig. 7A). Ultrastructural observation alsorevealed increases in the number of mitochondria andof basal membrane infoldings, which were accompaniedby dilatation of the extracellular spaces between theseinfoldings (Fig. 8). By PD 69, the endolymphatic spacehad completely collapsed, the cells of the ampulla wallwere apposed to the apical surface of the dark cells(Fig. 9), and the vestibular end-organs had completelydegenerated. However, the dark cells were still presentand di¡erentiated, with a morphology similar to thatobserved 1 or 3 weeks after birth, except for a morepycnotic nucleus.

The modi¢cations in the sensory epithelia of the cris-tae ampullares were much less spectacular during the¢rst few days after birth (Fig. 10). Thus, we noticed nomajor morphological abnormalities at birth or on PD

10. The earliest modi¢cations in the cellular structure ofthe sensory epithelium were observed at 3 weeks afterbirth (PD 22). At that time we noticed the presence ofbundles of ¢laments along the baso-apical axis of nu-merous sensory hair cells (Fig. 10B), and the presenceof a few cystic cavities (Fig. 10C). However, the grossmorphology of the sensory cells, with their hair bundlesand the supporting cells, seemed to be normal. Neithervacuolization of the cytoplasm, nor dense cytoplasmicinclusion, nor pycnotic nuclei were observed in the sen-sory epithelium. A¡erent and e¡erent nerve endingsalso appeared normal (Fig. 10A).

By PD 69 most of the hair cells had degenerated. Theendolymphatic space had completely collapsed and thecells of the wall of the membrane labyrinth of the am-pulla were pressing on only a few remaining sensorycells of the sensory epithelium (Fig. 10D). However,we could still distinguish organelles within the cyto-plasm of the rare remaining sensory cells. In addition,the space between the a¡erent calyx and the cells wasenlarged.

Signi¢cantly, immunostaining of the KCNQ1 proteinin the kcne1(3/3) mutant mouse revealed no speci¢clabeling of the apical membrane of the dark cells; wenoticed a low labeling all over the dilated cytoplasm inthe kcne1(3/3) mouse (Fig. 11).

4. Discussion

Vestibular dark cells are known to express a voltage-dependent, outward K� conductance which exhibits thesame biophysical and pharmacological properties as thecardiac delayed recti¢er IKs (Marcus and Shen, 1994;Sunose et al., 1997; Barhanin et al., 1999). Since the

Fig. 7. Comparison of the dark cell area of WT (A) and kcne1(3/3) mutant mice (B) on PD 10. In A, the dark cells present a homogeneousaspect at the base of a crista and are much smaller than all the other parts of the sensory epithelium. In B, both the height and volume of thedark cells are larger. The cytoplasm and nucleus are restricted to the apical side facing the endolymph, and the basolateral membrane infold-ings occupy more than two thirds of the cell height. Semithin sections stained with toluidine blue. Scale bar, 10 Wm. se = sensory epithelium,tc = transitional cells.

HEARES 3634 15-2-01


Fig. 8. Electron micrographs illustrating the postnatal changes of dark cells in the kcne1(3/3) mutant mice. (A,B) At birth (PD 0), the cells ap-pear completely normal and look like cells in WT mice. (C) On PD 7, the basal membrane infoldings and the size of the cells have started toincrease. (D) On PD 10, note details of the increase of the basolateral membrane infoldings. (E,F) On PD 22, note the substantial modi¢cationsof the dark cells: increase in size, multiplication of mitochondria, and basal membrane infoldings. The cytoplasm of the dark cell is reduced toa tiny area around a pycnotic nucleus and the spaces between the basolateral infoldings have increased. Scale bars, 2 Wm (A,C,E); 1 Wm(B,D,F).

HEARES 3634 15-2-01


expression of both the KCNE1 and KCNQ1 proteinsubunits is required to generate IKs (Barhanin et al.,1996; Sanguinetti et al., 1996), and co-expression ofthe two proteins has so far not been directly demon-strated at the cellular level, we investigated their expres-sion patterns throughout development in this cell type.

The KCNE1/KCNQ1 conductance is crucial for in-ner ear function as shown by studies conducted previ-ously on a kcne1(3/3) mouse which resulted in massiveinner ear damage in these animals (Vetter et al., 1996).

4.1. KCNE1 and KCNQ1 K+ channel subunits:localization and functional implications

In order to investigate the distribution of the KCNE1protein in the vestibular end-organs, we used immuno-cytochemistry. The protein was found at the apicalmembrane ^ facing the endolymph ^ of the vestibulardark cells. This observation is in accordance with pre-vious reports (Mori et al., 1993). Additionally, it couldbe detected at the apical side of the marginal cells of thestria vascularis (Sakagami et al., 1991), which have alsobeen shown to contain IKs (Shen et al., 1994). In thecochlea, these cells are specialized to secrete endolymphand are therefore equivalent to the dark cells of thevestibular system (Wangemann, 1995).

We used RT-PCR of the vestibular end-organs toinvestigate the distribution of KCNQ1 mRNA andfound that its distribution coincided with that of theKCNE1 protein. In situ hybridization and immunocy-tochemistry further corroborated these ¢ndings. Thus,using KCNQ1-speci¢c antibodies we demonstrated the

presence of the protein at the apical membrane of darkcells, i.e. in the same location as the KCNE1 protein.These data further support the view that the highlyspecialized dark cells of the vestibular end-organs areinvolved in secretion of K� into the endolymph.

KCNQ1 and KCNE1 transcripts are present in avariety of tissues including heart and kidney, althoughtheir co-localization has not previously been determinedat the cellular level in those tissues (Sanguinetti et al.,1996; Chouabe et al., 1997; Suessbrich and Busch,1999). But they are absent in brain and liver. In strialmarginal cells of the cochlea, KCNQ1 mRNA has beendetected by in situ hybridization (Neyroud et al., 1997).

4.2. KCNQ1/KCNE1 channel during development

Our next goal was to determine the onset of appear-ance of the channel subunits responsible for secretion ofK� into the endolymph a few days after birth. Usingimmunostaining and RT-PCR, we observed that theKCNE1 and KCNQ1 proteins were already present atbirth. Indeed, the KCNE1 protein was present in miceon GD 17 while KCNQ1 mRNA was visible by GD 18.At this early stage of development (Dechesne et al.,1994), inner ear morphogenesis is completed althoughcellular di¡erentiation and synaptogenesis are not. Nu-merous calcium binding proteins (Fig. 12) appear onGD 17^18, such as calbindin and calretinin, indicatorsof the biochemical maturation of the sensory end-or-gans (Dechesne et al., 1994). Hair cell di¡erentiationand a¡erent nerve synapses are also established at thisstage (Mbiene et al., 1988; Scarfone et al., 1988). The

Fig. 9. On PD 69, the cells of the roof of the ampulla are crushed (arrow) onto the dark cells. The endolymphatic space has completely col-lapsed. Scale bar, 2 Wm.

HEARES 3634 15-2-01


e¡erent synaptic contacts start to di¡erentiate only atbirth (Demeªmes and Broca, 1998) and physiologicalmaturation takes place a few days after birth (Desma-dryl et al., 1986).

Although we show that the KCNE1 and KCNQ1proteins are present at birth, it has been shown before(Anniko et al., 1979) that the K� concentration of theendolymph starts to increase only 5 days after birth to

Fig. 10. Postnatal morphological changes of the sensory epithelium in the kcne1(3/3) mutant mouse crista ampullaris. (A) On PD 22, the gen-eral morphology of the sensory hair cells (type I and type II) is not disturbed, except for the presence of intracytoplasmic rods of ¢laments (ar-row). The nucleus is not pycnotic, the hair bundles are present but in some cases scattered. (B) Detail of rods of ¢laments in a type I hair cellsurrounded by an a¡erent nerve calyx (nc). The rods are orientated along the apico-basal axis of the cell. (C) Cystic cavities (star) surroundedby an a¡erent nerve calyx (arrows). This may indicate the loss of a sensory hair cells. (D) On PD 69, the rare remaining sensory hair cellsseem completely wrapped in the few remaining supporting cells. The cells of the roof of the crista (large arrow) are lying on the few remainingcuticular plates of the sensory epithelium (short arrows). Scale bars, 2 Wm (A,C,D); 1 Wm (B).

HEARES 3634 15-2-01


reach its mature concentration of 150 mM 1 week afterbirth.

The apparent discrepancy between the physiologicalobservations and the detection of the two K� channelsubunits at birth and at earlier developmental stagescan be explained by non-functionality of the channelat birth and before birth. For example, the subunitsare present but may not be properly complexed. Anoth-er possibility is that the two protein subunits are notcorrectly targeted to the membrane, i.e. they might berequired to be placed in a speci¢c microdomain of themembrane by an as yet unidenti¢ed protein. A thirdpossibility is that regulatory proteins or second messen-gers, such as cAMP, involved in K� secretion or regu-lation pathways may not be fully e¡ective at birth. Forexample, it is known that the activity of adenylate cy-clase increases during the ¢rst postnatal week (Zajic et

Fig. 11. Immunolocalization of KCNQ1 in a utricle (U) of akcne1(3/3) mutant mouse, on PD 28. Note the absence of labelingat the apical membrane of the dilated dark cells. Scale bar, 10 Wm.

Fig. 12. Schematic drawing summarizing the developmental time course of KCNE1 and KCNQ1 proteins in the dark cells and the mammaliansensory organ development. The KCNE1 protein ¢rst appears on GD 17. KCNQ1 mRNA was also detected at this stage, and the KCNQ1protein is present at birth. However, the endolymph K� concentration starts to increase only on PD 3 (Anniko et al., 1979) to reach its maturelevel 1 week after birth.

HEARES 3634 15-2-01


al., 1983), and in the inner ear, the IKs current has beenshown to be activated by cAMP (Sunose et al., 1995,1997).

4.3. Postnatal morphological changes of the vestibularend-organs in the kcne1(3/3) mutant mouse

We observed some discrepancies with the ultrastruc-tural changes reported previously in kcne1(3/3) mice(Vetter et al., 1996) which may be accounted for byinter-individual di¡erences of mice. For example, theseauthors describe vacuolization of the sensory epitheliumas soon as PD 10 compared to 3 weeks in our studies.In addition, our detailed ultrastructural analysis ofpostnatal morphological changes using electron micros-copy allowed us to reveal abnormalities of vestibulardark cells in the mutant mouse. Although the structureand morphological characteristics (intracytoplasmic or-ganelles, desmosomes, basal infoldings) of dark cellswere preserved and were still easily recognizable even69 days after birth when the rest of the vestibule wascompletely degenerated, we observed changes consistingof structural modi¢cations rather than degeneration.Thus, the extracellular spaces between the basal mem-brane infoldings were signi¢cantly larger than normal.This enlargement may be explained by an ionic imbal-ance due to the missing transepithelial voltage-depen-dent current in the kcne1(3/3) mice (Vetter et al.,1996). In vitro studies of the mutant mice also showedno K�-stimulated K� secretion, which is characteristicof WT vestibular dark cells (Wangemann et al., 1996).

On the other hand, the changes in the sensory epi-thelium of the cristae can be considered a true degen-eration and may also be a consequence of the ionicimbalance of the endolymph. The ¢rst indications ofdegeneration are the presence of intracytoplasmic ¢la-ment rods in sensory hair cells, and of cystic cavitiesobserved 3 weeks after birth. This type of inclusion investibular hair cells has previously been described indi¡erent degenerative processes of the inner ear (Sobinand Flock, 1981; Issa and Forge, 1997) and duringcontinuous degeneration with aging (Sans and Ray-mond, 1994).

The mature K� concentration of the endolymph (150mM) is reached only 8^10 days after birth (Anniko etal., 1979). Hence, during the ¢rst week after birth, themedium bathing the vestibular sensory cells is probablysimilar in WT and kcne1(3/3) mutant mice. We hy-pothesize that with increasing age and enzymatic activ-ities, the secretion through the IKs channel no longeroccurs in the mutant because of the missing KCNE1subunit, although pumping of K� from the perilymphthrough the Na,K-ATPases and Na�/Cl3/K� cotrans-porter (Wangemann, 1995) is still functional. The maxi-

K� channel and the non-speci¢c cation channel are note¤cient enough to secrete endolymphatic K� to its nor-mal mature concentration. The resulting K� imbalancemay be the starting point of the degeneration of thesensory hair cells. Signi¢cantly, in humans, a similarmorphological degeneration of the inner ear was de-scribed in Jervell^Lange-Nielsen syndrome (Jervell andLange-Nielsen, 1957; Friedmann et al., 1966), a form ofcardiac LQTS, which is known to be related to IKs

dysfunctions (Guicheney et al., 1997; Schulze-Bahr etal., 1997; Tyson et al., 1997; Barhanin et al., 1998,1999; Drici et al., 1998).

5. Conclusion

In this study we have investigated the molecularstructures underlying the KCNQ1/KCNE1 channelwhich is known to be involved in K� secretion intothe endolymph in the inner ear. We have co-localizedthe two protein subunits of the channel to the apicalsurface of vestibular dark cells which represents the ¢rsttime that the two proteins have been identi¢ed in apolarized fashion at the cellular level. In addition, wede¢ned the ultrastructural changes observed in the in-ner ear of kcne1(3/3) mouse model. These changes areconsistent with those observed in the cardio-auditoryJervell^Lange-Nielsen syndrome which a¡ects humanpatients with mutations in KCNE1 or KCNQ1.

Another signi¢cant ¢nding is that the two subunits ofthe IKs channel can be considered speci¢c markers ofthe dark cells of the inner ear. This is of particularinterest because up to now, in cell biological studies,these cells have been characterized mainly by theirdark content and their basolateral infoldings. KCNQ1and KCNE1 subunits are thus speci¢c markers foundonly in the dark cells of the vestibular end-organs. Inthe future, we therefore expect that a dark cell culturepreparation may be a useful model to better understandand study K� endolymph secretion and regulation aswell as its dysfunctions.

Acknowledgements

This work was done at INSERM U432, Montpellier,France. We wish to thank D. Orcel, B. Arnaud, E.Poiraud, C. Travo for their technical assistance, Dr.A. Rabier (UPR 9955, Montpellier, France) for theprimer sequences of GAPDH and Drs. C.J. Caussid-ier-Dechesne and J. Raymond for their helpful com-ments. Some of the results were presented in 1999, atthe 4th International Menie©re Symposium in Paris andat the 2nd FEPS Congress in Prague.

HEARES 3634 15-2-01


References

Anniko, M., Wroblewsri, R., Wersa«ll, J., 1979. Development of endo-lymph during maturation of the mammalian inner ear. Arch. Oto-rhinolaryngol. 225, 161^163.

Barhanin, J., Lesage, F., Guillemare, E., Fink, M., Lazdunski,M., Romey, G., 1996. KvLQT1 and IsK (minK) proteins associateto form the IKS cardiac potassium current. Nature 384, 78^80.

Barhanin, J., Attali, B., Lazdunski, M., 1998. IKs, a slow and intrigu-ing cardiac K� channel and its associated long QT diseases.Trends Cardiovasc. Med. 8, 207^214.

Barhanin, J., Romey, G., Lazdunski, M., 1999. IsK: A novel type ofpotassium channel regulatory subunit. Curr. Top. Membr. Biol.46, 67^84.

Chouabe, C., Neyroud, N., Guicheney, P., Lazdunski, M., Romey,G., Barhanin, J., 1997. Properties of KvLQT1 K� channel muta-tions in Romano-Ward and Jervell and Lange-Nielsen inheritedcardiac arrhythmias. EMBO J. 16, 5472^5479.

Cooper, E.C., Jan, L.Y., 1999. Ion channel genes and human neuro-logical disease: recent progress, prospects, and challenges. Proc.Natl. Acad. Sci. USA 96, 4750^4756.

Deal, K.K., Lovinger, M.M., Tamkun, M.M., 1994. The brain Kv1.1potassium channel: in vitro studies on subunit assembly and post-translational processing. J. Neurosci. 14, 1666^1676.

Dechesne, C.J., Rabejac, D., Desmadryl, G., 1994. Development ofcalretinin immunoreactivity in the mouse inner ear. J. Comp. Neu-rol. 346, 517^529.

Demeªmes, D., Broca, C., 1998. Calcitonin gene-related peptide immu-noreactivity in the rat e¡erent vestibular system during develop-ment. Dev. Brain Res. 108, 59^67.

Desmadryl, G., Raymond, J., Sans, A., 1986. In vitro electrophysio-logical study of spontaneous activity in neonatal mouse vestibularganglion during development. Dev. Brain Res. 25, 133^136.

Die¡enbach, C.W., Dveksler, G.S. (Eds.), 1995. PCR Primer: A Lab-oratory Manual. Cold Spring Harbor Laboratory Press, ColdSpring Harbor, NY, pp. 1^682.

Drici, M.D., Arrighi, I., Chouabe, C., Mann, J.R., Lazdunski, M.,Romey, G., Barhanin, J., 1998. Involvement of KCNE1-associatedK� channel in heart rate control of repolarization in a murineengineered model of Jervell and Lange-Nielsen syndrome. Circ.Res. 83, 95^102.

Ferrary, E., Sterkers, O., 1998. Mechanisms of endolymph secretion.Kidney Int. 36, S98^S103.

Ferrary, E., Bernard, C., Oudar, O., Sterkers, O., Amiel, C., 1989.Sodium transfer from endolymph through a luminal amiloride-sensitive channels. Am. J. Physiol. 257, F182^F189.

Friedmann, I., Fraser, G.R., Froggan, P., 1966. Pathology of the earin the cardioauditory syndrome of Jervell and Lange-Nielsen (re-cessive deafness with electrocardiographic abnormalities). J. Lar-yngol. 80, 451^470.

Guicheney, P., Barhanin, J., LeMarec, H., 1997. Molecular basis ofinherited arrhythmias. Med. Sci. 14, 1025^1035.

Issa, A., Forge, A., 1997. Possible e¡ects of gentamicin on the darkcell region of the vestibular system. Br. J. Audiol. 31, 80^81.

Jervell, A., Lange-Nielsen, F., 1957. Congenital deaf-mutism, func-tional heart disease with prolongation of Q-T interval and suddendeath. Am. Heart J. 54, 59^68.

Kawamata, S., Harada, Y., Tagashira, N., 1986. Electron microscopicstudy of the vestibular dark cells in the crista ampullaris of theguinea pig. Acta Otolaryngol. (Stockh.) 102, 168^174.

Kimura, R.S., 1969. Distribution, structure and function of dark cellsin the vestibular labyrinth. Ann. Otol. Rhinol. 78, 542^561.

Lesage, F., Attali, B., Lakey, J., Honore, E., Romey, G., Faurobert,E., Lazdunski, M., Barhanin, J., 1993. Are Xenopus oocytes

unique in displaying functional IsK channel heterologous expres-sion? Receptors Channels 1, 143^152.

Marcus, D.C., Shen, Z., 1994. Slowly activating voltage-dependentK� conductance in apical pathway for K� secretion in vestibulardark cells. Am. J. Physiol. 267, C857^C864.

Marcus, D.C., Shipley, A.M., 1994. Potassium secretion by vestibulardark cell epithelium demonstrated by vibrating probe. Biophys. J.66, 1939^1942.

Marcus, D.C., Takeuchi, S., Wangemann, P., 1992. Ca2� activatednon selective cation channel in apical membrane of vestibulardark cells. Am. J. Physiol. 262, C1423^C1429.

Marcus, N.Y., Marcus, D.C., 1987. Potassium secretion by non-sen-sory region of gerbil utricle in vitro. Am. J. Physiol. 253, F613^F621.

Mbiene, J.P., Favre, D., Sans, A., 1988. Early innervation and di¡er-entiation of hair cells in the vestibular epithelia of mouse embryos:SEM and TEM study. Anat. Embryol. 177, 331^340.

Mori, N., Sakagami, M., Fukazawa, K., Matsunaga, T., 1993. Animmunohistochemical and electrophysiological study on IsK pro-tein in the stria vascularis of the guinea pig. Eur. Arch. Oto-Rhi-no-Laryngol. 250, 186^189.

Neyroud, N., Tesson, F., Denjoy, I., Leibovici, M., Donger, C., Bar-hanin, J., Faure, S., Gary, F., Coumel, P., Petit, C., Schwartz, K.,Guicheney, P., 1997. A novel mutation in the potassium channelgene KvLQT1 causes the Jervell and Lange-Nielsen cardioaudi-tory syndrome. Nature Genet. 15, 186^189.

Sakagami, M., Fukazawa, K., Matsunaga, T., Fujita, H., Mori, N.,Takumi, T., Ohkub, H., Nakanishi, S., 1991. Cellular localizationof rat IsK protein in the stria vascularis by imunohistochemicalobservation. Hear. Res. 56, 168^172.

Sanguinetti, M.C., Curran, M.E., Zou, A., Shen, J., Spector, P.S.,Atkinson, D.L., Keating, M.T., 1996. Coassembly of KvLQT1and minK (IsK) proteins to form cardiac IKS potassium channel.Nature 38, 80^83.

Sans, A., Raymond, J., 1994. Le vieillissement du vestibule. Vertiges93, 91^94.

Scarfone, E., Demeªmes, D., Jahn, R., De Camilli, P., Sans, A., 1988.Secretory function of the vestibular nerve calyx suggested by pres-ence of vesicles, synapsin I and synaptophysin. J. Neurosci. 8,4640^4645.

Schulze-Bahr, E., Wang, Q., Wedeking, H., Haverkamp, W., Chen,Q., Sun, Y., Rubie, C., Hordt, M., Towbin, J.A., Borggrefe, M.,Assmann, G., Qu, X., Somberg, J.C., Breithardt, G., Oberti, C.,Funke, H., 1997. KCNE1 mutations cause Jervell and Lange-Niel-sen syndrome. Nature Genet. 17, 267^268.

Shen, Z., Marcus, D.C., 1998. Divalent cations inhibit IsK/KvLQT1channels in excised membrane patches of strial marginal cells.Hear. Res. 123, 157^167.

Shen, Z., Marcus, D.C., Sunose, H., Chiba, T., Wangemann, P., 1994.IsK channel in strial marginal cells. Audit. Neurosci. 3, 215^230.

Sobin, A., Flock, A., 1981. Sensory hairs and ¢laments rods in ves-tibular hair cells of the waltzing guinea pig. Acta Otolaryngol. 91,247^254.

Splawski, I., Tristani-Firouzi, M., Lehmann, M.H., Sanguinetti,M.C., Keating, M.T., 1997. Mutations in the hminK gene causelong QT syndrome and suppress IKs function. Nature Genet. 17,338^340.

Steel, N.P., Bussoli, T.J., 1999. Deafness genes, expressions of sur-prise. Trends Genet. 15, 207^211.

Sterkers, O., Ferrary, E., Amiel, C., 1988. Production of inner ear£uids. Physiol. Rev. 68, 1083^1128.

Suessbrich, H., Busch, A.E., 1999. The IKs channel: coassembly ofIsK (minK) and KvLQT1 protein. Rev. Physiol. Biochem. Phar-macol. 137, 191^226.

HEARES 3634 15-2-01


Sunose, H., Liu, J., Marcus, D.C., 1995. Stimulation of KCNE1channel current by cAMP in K� secretory epithelia of inner ear.Proc. Sendai Symp. 5, 73^76.

Sunose, H., Liu, J., Shen, Z., Marcus, D.C., 1997. cAMP increasesapical I-sK channel current and K� secretion in vestibular darkcells. J. Membr. Biol. 156, 25^35.

Takeuchi, S., Marcus, D.C., Wangemann, P., 1992. Maxi K� channelin apical membrane of vestibular dark cells. Am. J. Physiol. 267,C857^C864.

Takumi, T., Ohkubo, H., Nakanishi, S., 1988. Cloning of a membraneprotein that induces a slow voltage-gated potassium current. Sci-ence 242, 1042^1045.

Tyson, J., Tranebj×rg, L., Bellman, S., Wren, C., Taylor, J., Bathen,J., Aslaken, B., SÖrland, S.J., Lund, O., Malcom, S., Pembrey,M., Bhattacharya, S., Bimer-Glindzicz, M., 1997. IsK andKvLQT1: mutation in either of the two subunits of the slowcomponent of the delayed recti¢er potassium channel can causeJervell and Lange-Nielsen syndrome. Hum. Mol. Genet. 6, 2179^2185.

Vetter, D.E., Mann, J.R., Wangemann, P., Liu, J.Z., McLaughlin,K.J., Lesage, F., Marcus, D.C., Lazdunski, M., Heinemann,S.F., Barhanin, J., 1996. Inner ear defects induced by null muta-tion of the IsK gene. Neuron 17, 1251^1264.

Wang, Q., Curran, M.E., Splawski, I., Burn, T.C., Millholland, J.M.,van Raay, T.J., Shen, J., Timothy, K.W., Vincent, G.M., de Jager,T., Swartz, P.J., Towbin, J.A., Moss, A.J., Atkinson, D.L.,Landes, G.M., Connors, T.D., Keating, M.T., 1996. Positionalcloning of a novel potassium channel gene: KvLQT1 mutationscause cardiac arrhythmia. Nature Genet. 12, 17^23.

Wangemann, P., 1995. Comparison of ion transport mechanisms be-tween vestibular dark cells and strial marginal cells. Hear. Res. 90,149^157.

Wangemann, P., Shen, Z., Liu, J., 1996. K�-induced stimulation ofK� secretion involves activation of the KCNE1 channel in vestib-ular dark cells. Hear. Res. 100, 201^210.

Zajic, G., Anniko, M., Schacht, J., 1983. Cellular localization of ade-nylate cyclase in the developping and mature inner ear of themouse. Hear. Res. 10, 249^261.

HEARES 3634 15-2-01


KCNQ1/KCNE1 potassium channels in mammalian vestibular dark cells

Documents

Transcript of KCNQ1/KCNE1 potassium channels in mammalian vestibular dark cells