Biochimica et Biophysica Acta - CORE40, 0.1% SDS, 2 mM EDTA, 6 mM Na 2HPO 4, and 4 mM NaH 2PO 4) containing the protease inhibitor PMSF (Benza and Leuptin), and incubated on ice for

Biochimica et Biophysica Acta 1812 (2011) 536–543

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r.com/ locate /bbad is

Pathogenic effects of a novel mutation (c.664_681del) in KCNQ4 channels associatedwith auditory pathology

Jeong-In Baek a,1, Hong-Joon Park b,1, Kyungjoon Park c, Su-Jin Choi a, Kyu-Yup Lee d, Jee Hyun Yi c,Thomas B. Friedman e, Dennis Drayna e, Ki Soon Shin c, Un-Kyung Kim a,⁎,1

a Department of Biology, College of Natural Sciences, Kyungpook National University, Daegu, 702-701, South Koreab Soree Ear Clinics, Seoul, South Koreac Department of Biology, Department of Life and Nanopharmaceutical Sciences, Kyunghee University, Seoul, South Koread Department of Otorhinolaryngology-Head and Neck Surgery, College of Medicine, Kyungpook National University, Daegu, South Koreae Laboratory of Molecular Genetics, National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Rockville, MD 20850, USA

⁎ Corresponding author. Tel.: +82 53 950 5353; fax:E-mail address: [email protected] (U.-K. Kim).

1 These authors contributed equally to this work.

0925-4439/$ – see front matter © 2010 Elsevier B.V. Adoi:10.1016/j.bbadis.2010.09.001

a b s t r a c t

Article history:Received 14 April 2010Received in revised form 31 August 2010Accepted 2 September 2010Available online 9 September 2010

Keywords:Hearing lossKCNQ4K+ channelMutationDominant negative effect

Hearing loss is a common communication disorder caused by various environmental and genetic factors.Hereditary hearing loss is very heterogeneous, and most of such cases involve sensorineural defects in theauditory pathway. There are currently 57 known autosomal dominant non-syndromic hearing loss (DFNA)loci, and the causative genes have been identified at 22 of these loci. In the present study, we performed agenome-wide linkage analysis in a Korean family segregating autosomal dominant hearing loss. We observedlinkage on chromosome 1p34, and at this locus, we detected a novel mutation consisting of an 18 nucleotidedeletion in exon 4 of the KCNQ4 gene, which encodes a voltage-gated potassium channel. We carried out afunctional in vitro study to analyze the effects of this mutation (c.664_681del) along with two previouslydescribed KCNQ4 mutations, p.W276S and p.G285C. Although the c.664_681del mutation is located in theintercellular loop and the two previously described mutations, p.W276S and p.G285C, are located in the poreregion, all mutants inhibit normal channel function by a dominant negative effect. Our analysis indicates thatthe intercellular loop is as significant as the pore region as a potential site of pathogenic effects on KCNQ4channel function.

+82 53 953 3066.

ll rights reserved.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

Potassium ions play a key role in sound transduction in the inner ear.In the cochlea, a disparity of K+ concentration ismaintained between theendolymphand the auditory cells in the scalamedia, and the sensory cellshave several types of K+ channels in their plasma membrane for K+

transport. An influx of K+ across the concentration gradient from theendolymph into the hair cells induces the neuronal signaling pathway.Subsequently, K+ efflux from hair cells through other neighboring cellsreturnsK+ to the endolymph. The recycling of K+and themaintenance ofhomeostasis of the ion concentration in the cochlea has a crucial functionin the regulation of the excitability of hair cells, and the subsequentneurotransmission pathway [1].

KCNQ4 is amember of the KCNQ family, which functions as voltage-gated potassium channels in the membrane. Ion flux through thesechannels underlies theM current, and this gene is specifically expressedin the outer hair cells (OHC) of cochlea [2–4]. Several previous studieshave identified mutations in the KCNQ4 gene that are responsible for

autosomal dominant non-syndromic hearing loss (ADNSHL) [5–7], and13 probable pathogenic mutations have been identified in this geneassociated with deafness that maps to the DFNA2 locus [8–16]. Themajority of studies have reported only the detection ofmutations in thisgene, without in vitro functional studies to help understand thephysiological effects of these mutations.

In this study, we describe a novel deletion mutation, c.664_681del inthe KCNQ4 gene as the cause of hearing loss in a Korean family withdeafness mapping to the DFNA2 locus. Furthermore, we performedcomparative in vitro functional studies to better understand thepathogenic effects of c.664_681del mutation along with those of twoverifiedmissensemutations, p.W276S and p.G285C in this gene [17–19].

2. Materials and methods

2.1. Subjects

A four-generation Korean family with ADNSHL (KDF01) wasrecruited from the Department of Otorhinolaryngology-Head and NeckSurgery, Ajou University, Suwon, South Korea (Fig. 1). A total of 24individuals including 11 affected and 13 unaffected members partici-pated in this study. Physical and otoscopic examinations and

http://dx.doi.org/10.1016/j.bbadis.2010.09.001mailto:[email protected]://dx.doi.org/10.1016/j.bbadis.2010.09.001http://www.sciencedirect.com/science/journal/09254439

Fig. 1. Clinical information for KDF01 family with ADNSHL. a. Pedigree of KDF01 family. Squares and circles represent females and males. Affected individuals are denoted by solidsymbols and slash indicates a deceased individual. The affected haplotypes (rectangles) are co-segregated with DFNA2 locus on chromosome 1p34. Microsatellite markers wereselected according to their physical location on the human genome map (National Center for Biotechnology Information: www.ncbi.nlm.nih.gov). b. Audiogram for pure-tonethresholds (PTA) of 11 affected individuals in the family. The average PTA value of right and left ears for each 11 patients was represented by the line graph, and the gray-zoneindicates the extent of normal auditory capacity. The numbers in the brackets are the ages of the members.

537J.-I. Baek et al. / Biochimica et Biophysica Acta 1812 (2011) 536–543

audiological studies were carried out, including pure-tone audiometry(PTA). PTAwas calculatedasanaverageof the thresholdmeasuredat 0.5,1.0, 2.0 and 3.0 KHz, and air-conduction threshold measurements were

performed at 125–8000 Hz. The level of hearing loss is described asfollows depending on PTA: normal hearing, below 20 dB; mild hearingimpairment; 21 to 40 dB; moderate hearing impairment, 41 to 70 dB;

http://www.ncbi.nlm.nih.gov

538 J.-I. Baek et al. / Biochimica et Biophysica Acta 1812 (2011) 536–543

severe hearing impairment, 71 to 95 dB; and profound hearingimpairment above 95 dB. In addition, bone conduction thresholdswere measured at 250–4000 Hz to exclude conductive hearing loss inaffected individuals. One hundred unrelated Koreans with normalhearing were recruited as the normal control from Kyungpook NationalUniversity Hospital, Daegu, South Korea. All participants providedwritten informed consent according to the protocol approved by theEthics Committee of Ajou University Hospital and Kyungpook NationalUniversity Hospital before the study. Genomic DNAs from the familymembers and 100 controls were extracted from peripheral blood usingthe FlexiGene DNA extraction kit (QIAGEN, Hilden, Germany).

2.2. Genetic analysis

A genome-wide linkage scan was carried out using 388 micro-satellite markers from the ABI Prism LinkageMapping Set, version 2.5.These markers have an average spacing of 10 cM across the genome.Six additional markers in the 1p33–1p34.2 region were chosen fromthe NCBI database (www.ncbi.nlm.nih.gov) to narrow down thecandidate region. Marker genotyping was performed by polymerasechain reaction (PCR) using fluorescently labeled primers with theDNA Engine® Thermal cycler (BIO-RAD, Hercules, CA, USA), and theproducts were analyzed on an ABI 3130xl genetic analyzer. Individualgenotypes for each marker were analyzed by using GeneMapper v4.0software (Applied Biosystems Corp., Foster City, CA, USA).

Testing for errors in Mendelian inheritance was done usingPEDCHECK version 1.1 [20]. Two-point linkage analysis was per-formed using the MLINK program within the LINKAGE softwarepackage version 5.2 [21]. LOD scores were computed at differentrecombination frequencies (θ), assuming equal recombination fre-quencies in males and females. Analysis was performed under anautosomal dominant model, with 99% penetrance and a disease allelefrequency of 0.001. In the candidate regions containing markers withhigh LOD scores, haplotypes were constructed by genotypingmicrosatellite markers in patients and other family members.

The entire coding region of KCNQ4was amplified using primer setsdesigned using Primer 3 software (http://www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi). The sequences of the codingregion obtained from direct sequencing were analyzed using an ABI3130xl DNA sequencer (Applied Biosystems Corps., Foster City, CA,,USA). Comparisons between family member DNA sequences and theGenBank reference sequence of KCNQ4 (NM_004700.2) were per-formed to define mutations in the KCNQ4 coding region.

2.3. Plasmid construction

A pCMV6 entry vector containing the KCNQ4 wild type full-lengthcDNA C-terminally tagged with myc and flag (Origene Technologies,Rockville, MD, USA) was used for the functional studies. Mutantplasmids containing the KCNQ4 cDNA mutagenized to c.664_681del,c.827GNC (p.W276S) and c.853GNT (p.G285C) were constructed bysite-directed mutagenesis using pfu Taq polymerase (TaKaRa, Shiga,Japan) and Dpn1 endonuclease (New England Biolabs, Ipswich, MA,USA). Final sequences of the mutagenized cDNA inserted in theplasmid were confirmed by DNA sequencing.

2.4. Cell culture and transient transfection

The human embryonic kidney cell line HEK293 (American TypeCulture Collection, USA) was maintained in minimum essentialmedium alpha (α-MEM) with 10% fetal bovine serum (WelGENE,Daegu, Korea) and 1% penicillin (PAA Laboratories GmbH, Pasching,Austria). Before transient transfection, cells were seeded to a densityof 50–70% confluency on the 12 mm round coverslip for patch-clampand on the 100 mm culture dish for the biotinylation. For eachplasmid to be transfected, plasmid DNA and polyethyleneimine, PEI

(Sigma Aldrich Corp., St. Louis, MO, USA) were mixed in the ratio of1 μg to 3 μg in serum-free α-MEM media and incubated at roomtemperature for 20 min. The plated cells were transfected with themixture and incubated at 37 °C for 24 h. In the case of transfection forpatch-clamp studies, the plasmid DNAwas co-transfected with pEGFPN1 vector in the proportion of 1 μg plasmid to 50 ng pEGFP N1 vectorto discern transfected cells by fluorescence.

2.5. Electrophysiology

One day after transfection, whole-cell voltage-clamp recordingswere performed from a single cell using an EPC10 amplifier (HEKAElektronik, Germany) at room temperature (21–23 °C). Patch-clamppipettes were pulled from borosilicate glass and had a tip resistance of2.5–4 MΩ when filled with pipette solution. The pipette solutioncontained (inmM): 120 KCl, 5.4 CaCl2, 1.75MgCl2, 10 EGTA, 10 HEPES,4 Mg-ATP, and 0.4 Na-GTP; pH was adjusted to 7.2 with KOH. Theexternal solution was constantly perfused (1–1.5 ml/min) with asolution containing (in mM): 140 NaCl, 4 KCl, 2 CaCl2, 1 MgCl2, and 10HEPES; adjusted to pH 7.4, with NaOH. The transfected cells weredetected with green fluorescence via E enhanced green fluorescentprotein (EGFP) expression. KCNQ4 currents were generated withdepolarizing voltage steps in 10 mV increments of 1.5 s duration froma holding potential of −70 mV, followed by a constant potential of−50 mV. Datawere sampled at 5 kHz and filtered at 3 kHzwith Besselfilter. Whole-cell K+ current was measured as the amplitude betweenthe steady state and peak. For calculating current density, the whole-cell K+ current (pA) was divided by the cell capacitance (pF).

2.6. Cell surface biotinylation and western blotting

Expression of KCNQ4 protein on the cell surface was assayed bybiotinylation of cell surface protein. HEK293 cells transfected 24 hpreviously with wild type or mutant type KCNQ4 were washed threetimes with ice-cold PBS, and biotinylated with 0.5 mg/ml Sulfo-NHS-SS-Biotin (Pierce Biotechnology Inc., Rockford, IL, USA) in PBS on icefor 20 min. The reaction was quenched using 100 mM glycine in 1×PBS for 20 min. After washing three times with ice-cold PBS, cellswere collected and suspended with lysis buffer (150 mMNaCl, 1% NP-40, 0.1% SDS, 2 mM EDTA, 6 mM Na2HPO4, and 4 mM NaH2PO4)containing the protease inhibitor PMSF (Benza and Leuptin), andincubated on ice for 30 min. The whole-cell lysate was purified bycentrifugation at 12,000×g for 10 min (4 °C). To isolate biotinylatedproteins, 150 μl of streptavidin-coated agarose beads (Sigma AldrichCorp., St. Louis, MO, USA) was added to the purified lysate, andincubated at 4 °C for 5 h with gentle rotation. The biotin–avidinbinding proteins were collected andwashed three timeswith the lysisbuffer. Sample buffer was added to the final pellet and heated at100 °C. The proteins were separated on 7.5% SDS-PAGE gel, andtransferred to nitrocellulose membranes. The blots were probedovernight with mouse anti-FLAG® M2 antibody (1:1000 dilution;Stratagene Corp., La Jolla, CA, USA) and goat anti-aldolase A antibody(1:1000 dilution; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA)at 4 °C. Detection of flag and aldolase A was performed using goatanti-mouse IgG-HRP and donkey anti-goat IgG-HRP (each 1:1000dilution; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) at roomtemperature for 1 h. Final protein bands were visualized by theenhanced chemiluminescence, ECL solution (Amersham Biosciences,USA) and exposed to X-ray film. These experiments were repeated aminimum of three times. The band intensities from western blottingwere calculated using Image J software (NIH, Bethesda, MD, USA), andthe values were normalized for each experiment. To determine thesignificance of differences between each wild and mutant type, thedata were analyzed by one-way ANOVA using SPSS software version12.0 (SPSS Inc., Chicago, IL, USA). Only the results with a p-value ofless than 0.05 were considered to be statistically significant.

http://www.ncbi.nlm.nih.govhttp://www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgihttp://www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi

Table 1Two-point LOD scores between five additional microsatellite markers on chromosome 1p34.

Marker Genetic maplocation (cM)

Recombination fraction (θ)

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Zmaxa θmax

D1S458 52.70 −5.00 0.28 0.66 0.80 0.84 0.80 0.72 0.60 0.44 0.84 0.20D1S513 60.01 −4.31 −0.52 0.12 0.39 0.51 0.53 0.49 0.40 0.29 0.53 0.25D1S432 69.86 1.28 1.75 1.68 1.50 1.26 0.98 0.69 0.40 0.17 1.75 0.05D1S1188 70.41 0.64 1.24 1.32 1.29 1.20 1.08 0.92 0.73 0.52 1.32 0.10D1S2722 72.59 1.08 2.27 2.42 2.35 2.17 1.91 1.59 1.23 0.82 2.42 0.10D1S3721 72.59 3.59 3.32 3.04 2.74 2.43 2.10 1.74 1.36 0.94 3.59 0.00

a Zmax means a maximum LOD score for the marker in nine different recombination fractions (θ).

539J.-I. Baek et al. / Biochimica et Biophysica Acta 1812 (2011) 536–543

3. Results

3.1. Clinical features of the KDF01 family

The pedigree of the KDF01 family shows a typical autosomaldominant inheritance pattern of hearing loss (Fig. 1a). Clinicalhistories and audiological assays of the affected individuals revealeda symmetrical, bilateral and progressive sensorineural hearing loss.The earliest clinical evidence of hearing loss in this family wasobtained from individual IV-2 at the age of 4 years. The level ofhearing loss (HL) within the family showed variable presentation,frommild to profound, and the severity of HL was proportional to age[22]. Themajority of affected individuals showedmild or moderate HLfor low and mid frequencies, and severe HL for high frequencies(Fig. 1b). The affected individuals had one of two audiogram patterns,flat or slopping, with most of the affected members displaying aslopping audiogram. None of the affected individuals of the familyreported tinnitus, and none displayed clinical features of vestibulardysfunction.

3.2. Linkage analysis and haplotype construction

A genome-wide linkage screen was performed using 388 micro-satellite markers with an average spacing of 10 cM. Linkage analysisrevealed amaximum LOD score of 3.59 at recombination fraction 0.00,obtained with marker D1S3721 on chromosome 1p34. Six additionalmarkers (D1S458, D1S513, D1S432, D1S2706, D1S1188 and D1S2722)spanning 14.89 cM across the chromosome 1p34 regionwere selectedand genotyped to better delineate the candidate region. AlthoughD1S2706 was uninformative for linkage, the five other markers gavepositive LOD scores suggestive of linkage (Table 1).

The chromosome 1p haplotypes of the family were constructedwith seven microsatellite markers. One haplotype showed a signifi-cant LOD score at chromosome 1p34, and all the affected individuals

Fig. 2. Sequence analysis of exon 4 in KCNQ4 by TA cloning. A novel mutation causing deletionfamily including II-1 were heterozygote for the mutation. In contrast, none of 13 normaphenotype in the family.

carried the same haplotype (Fig. 1a). Two informative recombinationevents were identified in six affected individuals (II-1, II-4, III-1, III-9,IV-1, and IV-2) and three unaffected individuals (II-3, II-8, and II-9).The inspection of recombinant haplotypes in these nine subjectsplaced the hearing loss locus between the markers D1S513 andD1S2722. All of eleven affected individuals shared the same diseasehaplotype, while none of the unaffected members carried thishaplotype. Thus, a region of approximately 12 cM on chromosome1p34 between D1S513 and D1S2722 was defined as the candidateregion responsible for hearing loss in this family. This candidateregion was coincident with that of the DFNA2 locus, an autosomaldominant hearing loss locus that has been shown to contain twocausative hearing loss genes, GJB3 and KCNQ4.

3.3. Identification of a novel mutation in KCNQ4 gene

The complete coding regions of the KCNQ4 and GJB3 genes wereanalyzed in 24 KDF01 family members by direct sequencing. While nomutations were identified in the GJB3 gene, we identified a novelmutation in exon 4 of KCNQ4 gene in the affected members of thisfamily. The novel mutation was an 18 nucleotide deletion starting atnucleotide position 664 (c.664_681del). This mutation is predicted tocause the deletion of 6 amino acids in the intra-membrane loopbetween S4 and S5 domains (Fig. 2). All of the 11 affected individualswere heterozygous for this mutation, and none of the 13 normalindividuals in this family and 100 normal controls carried thismutation.

3.4. Effects of mutations on KCNQ4 channel function

To determine the functional effects of mutations in the KCNQ4channel, the channels were transiently expressed in HEK 293 cells andwhole-cell patch-clamp recordings were performed. Cells transfectedwith the wild type KCNQ4 alone expressed well-defined slow-activating

of 6 amino acids is detected in this mutation screening. Eleven affected members in thel members had this mutation. The mutation genotypes were co-segregated with the

image of Fig.�2

Fig. 3. Functional analysis of KCNQ4WT and KCNQ4MT (c.664_681del, p.W276S and p.G285C) in HEK293 cell line. a. Whole-cell membrane currents were recorded from transientlytransfected cells. From a holding potential of −70 mV, currents were evoked by depolarizing voltage steps from −70 mV to 50 mV in 10 mV increments. Scale bar denotes 500 msand 1 nA. b. Comparison of current–voltage relationships. Each point on the curves represents mean values of normalized peak current density at each voltage obtained fromrecorded cells (n=number of cells tested). Curves of three mutant types alone were not represented because current values were measured to zero. c. Comparison of currentdensities measured at +50 mV. †pb0.05, ††pb0.01 versus KCNQ4WT. **pb0.01 versus p.W276S+WT (1:1).

540 J.-I. Baek et al. / Biochimica et Biophysica Acta 1812 (2011) 536–543

outward K+ currents (Fig. 3a). In contrast, three mutant channels(c.664_681del, p.G285C, and p.W276S) failed to display any detectableK+ currents (Fig. 3a).We concluded that all three mutations cause a lossof channel function. Because functional KCNQ channels are composed of4 subunits,we examinedwhether themutant channels showadominantnegative effect as a heteromer containing both mutant and wild typesubunits. Whereas HEK293 cells co-transfected with wild type andc.664_681del or p.G285C exhibited no currents, currents of about 32% ofwild type were detected in p.W276S co-transfected cells (Fig. 3). Themean current density (pA/pF) of KCNQ4W276S:WT=1:1 at a +50 mVvoltage step was significantly different from KCNQ4664_681del:WT=1:1,andKCNQ4G285C:WT=1:1 (KCNQ4W276S:WT: 8.9±2.8, KCNQ4664_681del:WT:2.3±0.7, and KCNQ4G285C:WT=1:1: 2.5±0.5, pb0.01, Newman–Keulspost test). Taken together, these results indicate that heteromericchannels containing c.664_681del or p.G285C subunits are nonfunction-al, while heteromers of the wild type and p.W276S remain at some levelof functionality.

To estimate the effect of c.664_681del, p.W276S and p.G285Cmutations on the trafficking of KCNQ4 to the plasma membrane, wecompared the KCNQ4 expression level in the whole cells with that inplasma membranes by surface protein biotinylation method andwestern blotting in HEK293 cells followed by statistical analysis. Co-transfection of normal and mutant KCNQ4 cDNA was performed usingwild type and 3 mutant types of KCNQ4, to mimic the heterozygouscondition in patients with a dominantly inherited disease. In thewhole-cell lysate (cytoplasmic and surface-expressed), both the wild

type and 3mutant types of KCNQ4 showed no significant difference inquantitative expression, verifying that these three different mutationshave no influence on synthesis of the KCNQ4 protein (Fig. 4a).Furthermore, this expression level was consistent with surface-expressed KCNQ4. The three mutant KCNQ4 proteins showedstatistically equal expression with each other and with KCNQ4WT onthe cell surface as detected by biotinylation. Although there were alittle quantitative difference between the total and surface-expressedprotein in KCNQ4W276S (total 91±6.49, surface 108.25±7.26) andKCNQ4G285C (total 104±16.17, surface 85.33±3.93), it was notsignificant (pN0.05). Upon co-expression of the wild type and eachmutant type, protein expression was not different from wild typeindicating that c.664_681del, p.W276S and p.G285C do not disturbprotein synthesis and polymerization of normal KCNQ4 channel(Fig. 4b). Finally, our results suggest that these threemutations, whichoccur in different regions of the protein, have no effect on the proteinsynthesis and localization to the membrane, and that they disturb theion permeation through the pore by a dominant negative effect.

4. Discussion

In the present study, we used genetic studies of linkage analysisand mutation screening to verify a novel pathogenic gene mutation ina Korean family with dominantly inherited deafness. This mutationwas an 18 nucleotide deletion in the KCNQ4 gene, encoding a subunitof voltage-gated potassium channel. KCNQ4 is normally expressed on

image of Fig.�3

Fig. 4. Quantitative expression analysis of total and surface-expressed KCNQ4 protein. a. Western blot analysis of total and surface-expressed KCNQ4 protein. HEK293 cells weretransiently transfected with wild type (WT), mutant type, or co-transfected with wild type and mutant type (1:1). Whole-cell lysate and biotinylated proteins isolated bystreptavidin beads (surface) were subjected to western blot analysis. The over-expressed KCNQ4 monomers were detected at about 77 kDa, while non-transfected control cellsshowed no signals. Cytoplasmic enzyme, aldolase A (40 kDa) was used as the intracellular control. All of 3 mutant types show quantitatively equal expression with the wild type inthe whole cell and surface. b. Expression levels quantified by densitometry were normalized for each experiment. Each bar indicates the mean values±SEM (standard error of themean) of at least 3 experiments. Black and white bars represent total KCNQ4 expressed in the whole cells and surface-expressed KCNQ4, each. Control indicates non-transfectedcells. All of 7 different types including the wild type and 3 mutants show practically equal expression. Also, statistical analysis using ANOVA demonstrated that no significantdifferences were detected between each type (pN0.05).

541J.-I. Baek et al. / Biochimica et Biophysica Acta 1812 (2011) 536–543

the plasma membrane of outer hair cells in the cochlea. A KCNQ4monomer consists of 2 cytoplasmic domains, 6 trans-membranedomains (S1–S6) including the voltage sensor (S4) domain, and apore domain between the S5 and S6 domains (Fig. 5a) [23]. Thefunctional KCNQ4 channel is composed of 4 subunits, and this channelcontributes to the homeostatic maintenance of K+ ions in the cochlea.Kubisch et al. demonstrated that KCNQ4 is located within the DFNA2locus and that mutations in this gene are causative of ADNSHL [2].Since then, several subsequent studies have identified a total of 13KCNQ4 gene mutations that have been suggested to cause ADNSHL.Most of these 13 mutations are concentrated in the pore region of thechannel. The novel deletion mutation identified in this study,

c.664_681del, is located in the S4–S5 loop region (Fig. 5a). Thissuggests that the Korean population may be useful in expanding themutational spectrum observed in hereditary hearing loss, and thatsuch mutations may lead to a better understanding of the physiologyof KCNQ4 channels.

The roles of intracellular S4–S5 loops in regulation of channelactivation have been substantiated through various functional studiesfor several types of K+ channels. Recently, several structural studiesfor K+ channels, including shaker and hERG channels, demonstratedthat the S4–S5 linker region contributes to the gating of the channelsthrough the interaction with the distal portion of the S6 domain, andthat mutations in this region induce profound dysfunction of the

image of Fig.�4

Fig. 5. Basic representation of the KCNQ4 protein. a. Schematic structure of KCNQ4 and the location of 13 mutations reported in previous studies. The KCNQ4 protein has 6 trans-membrane domains and a pore region. Almost all of the previously described mutations are located in the pore region. The novel deletion mutation, c.664_681del identified in thisstudy is indicated with an asterisk. b. Multi-alignment for amino acid sequences in S4–S5 loop and pore region of the KCNQ family. Their sequences are highly conserved in both theloop and pore regions. C.664_681del is outlined, and two pore mutations, p.W276S and p.G285C are denoted in bold.

542 J.-I. Baek et al. / Biochimica et Biophysica Acta 1812 (2011) 536–543

channels [24–28]. The significance of the S4–S5 loop region in thefunction of the other KCNQ family channels has also been demon-strated. For example, Surti et al. (2005) verified that the activity ofKCNQ2/KCNQ3 channels is regulated by phosphorylation of a specificresidue located in S4–S5 loop [29]. In addition, studies of the KCNQ1(KvLOT1) gene demonstrated that mutations in the S4–S5 loop regioncause a Long-QT syndrome [30–33]. The KCNQ1 mutation causingLong-QT syndrome resides in the same region as the c.664_681delidentified in this study. In various voltage-gated K+ channels, the S4–S5 linker plays a key role in the regulation of channel activity.Moreover, this loop region is remarkably conserved across KCNQchannels KCNQ1–KCNQ5 (Fig. 5b). This suggests that thec.664_681del mutation of KCNQ4 has a profound effect on normalchannel function.

To demonstrate the effect of the c.664_681del mutation in theliving cells, we analyzed the channel expression and electrical activityusing biotinylation and patch-clamp methods. In addition, tounderstand the roles of each domains in the KCNQ4 channel function,comparative analysis for three mutations in the S4–S5 loop(c.664_681del) and pore region (p.W276S and p.G285C) wasperformed. Protein synthesis, trafficking and membrane expressionof the channel were normal in all three mutant types. However, K+

ion flow through membrane expressed KCNQ4 channels was notdetected in these three mutant KCNQ4 channels, indicating thatc.664_681del, p.W276S and p.G285C inhibit normal channel functionby a dominant negative effect. Mencia et al. (2008) reported that p.G296S pore mutation of KCNQ4 impairs cell surface channelexpression. Although p.W276S and p.G285S are also pore mutations,

p.W276S and p.G285S are located in WW motif and GYGD motifwhich play the most important role in the potassium ion selectivity.Especially, Uehara et al. (2008) verified that p.W309R mutation ofKCNQ3 does not impair normal surface expression of the channel, andthe position of p.W309R in KCNQ3 is equivalent to that of the p.W276in KCNQ4 protein [34]. It suggests that the pathogenic effects ofp.W276S and p.G285C are likely due to loss of channel function, notdeficiency in protein synthesis or trafficking.

Interestingly, two mutations, p.W276S and p.G285C showeddifferent magnitudes of the effect, although both mutations arelocated in the pore region. In comparison, c.664_681del and p.G285Cshowed a very strong dominant negative effect compared with p.W276S, although they are located in different domains. However, thedisparity of the mutational effects in vitro did not correspond withtheir clinical phenotypes in vivo. Although p.W276S and p.G285Cmutations showed different severity in the channel dysfunction,patients with deafness carrying p.W276S or p.G285C mutation havethe same clinical symptoms, including onset ages, extension of onsetfrequency, progression rate, and severity of the disease [19,35].Although it is difficult to understand the intricate in vivo mechanismscausing phenotypic characters as well as some differences between invitro and in vivo, we surmise that a slight difference in the degree ofmolecular effect of these three pathogenic mutations is insufficient toprovoke the phenotypic difference.

In conclusion, we identified a novel pathogenic mutation in theKCNQ4 gene, and analyzed its physiological mechanisms for causinghearing loss in a Korean family. This is the first functional study of anon-pore mutation in the KCNQ4 gene, and it verifies that the S4–S5

image of Fig.�5

543J.-I. Baek et al. / Biochimica et Biophysica Acta 1812 (2011) 536–543

loop is as important as the pore region for channel function.Furthermore, we investigated the interrelation between the muta-tional effect at the molecular level and the phenotypic consequencesthrough a comparative study using other KCNQ4 gene mutations. Thisexploration for naturally occurring defects in potassium channelstructure and function can provide valuable information for theclarification of the etiological mechanisms of inherited hearing loss.

Acknowledgements

We are grateful to the family for their collaboration in this study.This work was supported by the Korea Science and EngineeringFoundation (KOSEF) grant funded by the Korea government (MEST)(R01-2008-000-10431-0), a grant of the Korea Healthcare TechnologyR&D Project, Ministry for Health, Welfare and Family Affairs, Republicof Korea, A080588 (UKK).

References

[1] P. Wangemann, K+ cycling and the endocochlear potential, Hear. Res. 165 (2002)1–9.

[2] C. Kubisch, B.C. Schroeder, T. Friedrich, B. Lutjohann, A. El-Amraoui, S. Marlin, C.Petit, T.J. Jentsch, KCNQ4, a novel potassium channel expressed in sensory outerhair cells, is mutated in dominant deafness, Cell 96 (1999) 437–446.

[3] H.S. Wang, Z. Pan, W. Shi, B.S. Brown, R.S. Wymore, I.S. Cohen, J.E. Dixon, D.McKinnon, KCNQ2 and KCNQ3 potassium channel subunits: molecular correlatesof the M-channel, Science 282 (1998) 1890–1893.

[4] T. Kharkovets, J.P. Hardelin, S. Safieddine, M. Schweizer, A. El-Amraoui, C. Petit, T.J.Jentsch, KCNQ4, a K+ channel mutated in a form of dominant deafness, isexpressed in the inner ear and the central auditory pathway, Proc. Natl. Acad. Sci.USA 97 (2000) 4333–4338.

[5] P. Coucke, G. Van Camp, B. Djoyodiharjo, S.D. Smith, R.R. Frants, G.W. Padberg, J.K.Darby, E.H. Huizing, C.W. Cremers, W.J. Kimberling, et al., Linkage of autosomaldominant hearing loss to the short arm of chromosome 1 in two families, N. Engl.J. Med. 331 (1994) 425–431.

[6] T.B. Friedman, A.J. Griffith, Human nonsyndromic sensorineural deafness, Annu.Rev. Genomics Hum. Genet. 4 (2003) 341–402.

[7] L. Nie, KCNQ4mutations associatedwith nonsyndromic progressive sensorineuralhearing loss, Curr. Opin. Otolaryngol. Head Neck Surg. 16 (2008) 441–444.

[8] P.J. Coucke, P. Van Hauwe, P.M. Kelley, H. Kunst, I. Schatteman, D. Van Velzen, J.Meyers, R.J. Ensink, M. Verstreken, F. Declau, H. Marres, K. Kastury, S. Bhasin, W.T.McGuirt, R.J. Smith, C.W. Cremers, P. Van de Heyning, P.J. Willems, S.D. Smith, G.Van Camp, Mutations in the KCNQ4 gene are responsible for autosomal dominantdeafness in four DFNA2 families, Hum. Mol. Genet. 8 (1999) 1321–1328.

[9] P. Van Hauwe, P.J. Coucke, R.J. Ensink, P. Huygen, C.W. Cremers, G. Van Camp,Mutations in the KCNQ4 K+ channel gene, responsible for autosomal dominanthearing loss, cluster in the channel pore region,Am. J.Med. Genet. 93 (2000) 184–187.

[10] P. Van Hauwe, P.J. Coucke, F. Declau, H. Kunst, R.J. Ensink, H.A. Marres, C.W. Cremers,B. Djelantik, S.D. Smith, P. Kelley, P.H. Van deHeyning, G. Van Camp, Deafness linkedto DFNA2: one locus but how many genes? Nat. Genet. 21 (1999) 263.

[11] Z. Talebizadeh, P.M. Kelley, J.W. Askew, K.W. Beisel, S.D. Smith, Novel mutation inthe KCNQ4 gene in a large kindred with dominant progressive hearing loss, Hum.Mutat. 14 (1999) 493–501.

[12] M.S. Hildebrand, D. Tack, S.J. McMordie, A. DeLuca, I.A. Hur, C. Nishimura, P.Huygen, T.L. Casavant, R.J. Smith, Audioprofile-directed screening identifies novelmutations in KCNQ4 causing hearing loss at the DFNA2 locus, Genet. Med. 10(2008) 797–804.

[13] G. Van Camp, P.J. Coucke, J. Akita, E. Fransen, S. Abe, E.M. De Leenheer, P.L. Huygen,C.W. Cremers, S. Usami, A mutational hot spot in the KCNQ4 gene responsible forautosomal dominant hearing impairment, Hum. Mutat. 20 (2002) 15–19.

[14] C.C. Su, J.J. Yang, J.C. Shieh, M.C. Su, S.Y. Li, Identification of novel mutations in theKCNQ4 gene of patients with nonsyndromic deafness from Taiwan, Audiol. Neuro-Otol. 12 (2007) 20–26.

[15] F. Kamada, S. Kure, T. Kudo, Y. Suzuki, T. Oshima, A. Ichinohe, K. Kojima, T. Niihori, J.Kanno, Y. Narumi,A.Narisawa,K.Kato, Y. Aoki, K. Ikeda, T. Kobayashi, Y.Matsubara, Anovel KCNQ4 one-base deletion in a large pedigreewith hearing loss: implication forthe genotype–phenotype correlation, J. Hum. Genet. 51 (2006) 455–460.

[16] A. Mencia, D. Gonzalez-Nieto, S. Modamio-Hoybjor, A. Etxeberria, G. Aranguez, N.Salvador, I. Del Castillo, A. Villarroel, F. Moreno, L. Barrio, M.A. Moreno-Pelayo, Anovel KCNQ4 pore-region mutation (p.G296S) causes deafness by impairing cell-surface channel expression, Hum. Genet. 123 (2008) 41–53.

[17] J. Akita, S. Abe, H. Shinkawa, W.J. Kimberling, S. Usami, Clinical and geneticfeatures of nonsyndromic autosomal dominant sensorineural hearing loss:KCNQ4 is a gene responsible in Japanese, J. Hum. Genet. 46 (2001) 355–361.

[18] V. Topsakal, G. Van Camp, P. Van de Heyning, Genetic testing for hearingimpairment, B-Ent 1 (2005) 125–135.

[19] V. Topsakal, R.J. Pennings, H. te Brinke, B. Hamel, P.L. Huygen, H. Kremer, C.W.Cremers, Phenotype determination guides swift genotyping of a DFNA2/KCNQ4family with a hot spot mutation (W276S), Otol. Neurotol. 26 (2005) 52–58.

[20] J.R. O'Connell, D.E. Weeks, PedCheck: a program for identification of genotypeincompatibilities in linkage analysis, Am. J. Hum. Genet. 63 (1998) 259–266.

[21] G.M. Lathrop, J.M. Lalouel, C. Julier, J. Ott, Strategies for multilocus linkage analysisin humans, Proc. Natl. Acad. Sci. USA 81 (1984) 3443–3446.

[22] X. Liu, L. Xu, Nonsyndromic hearing loss: an analysis of audiograms, Ann. Otol.Rhinol. Laryngol. 103 (1994) 428–433.

[23] R.J. Howard, K.A. Clark, J.M. Holton, D.L. Minor Jr., Structural insight into KCNQ(Kv7) channel assembly and channelopathy, Neuron 53 (2007) 663–675.

[24] S.B. Long, E.B. Campbell, R. Mackinnon, Voltage sensor of Kv1.2: structural basis ofelectromechanical coupling, Science 309 (2005) 903–908.

[25] T. Ferrer, J. Rupp, D.R. Piper, M. Tristani-Firouzi, The S4–S5 linker directly couplesvoltage sensor movement to the activation gate in the human ether-a'-go-go-related gene (hERG) K+ channel, J. Biol. Chem. 281 (2006) 12858–12864.

[26] N.E. Schoppa, F.J. Sigworth, Activation of Shaker potassium channels. III. Anactivation gatingmodel for wild-type and V2mutant channels, J. Gen. Physiol. 111(1998) 313–342.

[27] N.E. Schoppa, F.J. Sigworth, Activation of Shaker potassium channels. II. Kinetics ofthe V2 mutant channel, J. Gen. Physiol. 111 (1998) 295–311.

[28] N.E. Schoppa, F.J. Sigworth, Activation of shaker potassium channels. I. Characteriza-tion of voltage-dependent transitions, J. Gen. Physiol. 111 (1998) 271–294.

[29] T.S. Surti, L. Huang, Y.N. Jan, L.Y. Jan, E.C. Cooper, Identification by massspectrometry and functional characterization of two phosphorylation sites ofKCNQ2/KCNQ3 channels, Proc. Natl. Acad. Sci. USA 102 (2005) 17828–17833.

[30] H. Fodstad, H. Swan, P. Laitinen, K. Piippo, K. Paavonen, M. Viitasalo, L. Toivonen, K.Kontula, Four potassium channel mutations account for 73% of the geneticspectrum underlying long-QT syndrome (LQTS) and provide evidence for a strongfounder effect in Finland, Ann. Med. 36 (Suppl 1) (2004) 53–63.

[31] L. Franqueza, M. Lin, J. Shen, I. Splawski, M.T. Keating, M.C. Sanguinetti, Long QTsyndrome-associated mutations in the S4–S5 linker of KvLQT1 potassiumchannels modify gating and interaction with minK subunits, J. Biol. Chem. 274(1999) 21063–21070.

[32] T. Itoh, T. Tanaka, R. Nagai, K. Kikuchi, S. Ogawa, S. Okada, S. Yamagata, K. Yano, Y.Yazaki, Y. Nakamura, Genomic organization and mutational analysis of KVLQT1, agene responsible for familial long QT syndrome, Hum. Genet. 103 (1998) 290–294.

[33] A.D. Krahn, J. Wang, B. Spindler, A.C. Skanes, R. Yee, G.J. Klein, R.A. Hegele, A novelmutation in KVLQT1, L122P, found in a family with autosomal dominant long QTsyndrome, Am. Heart J. 140 (2000) 146–149.

[34] A. Uehara, Y. Nakamura, T. Shioya, S. Hirose, M. Yasukochi, K. Uehara, AlteredKCNQ3 potassium channel function caused by the W309R pore–helix mutationfound in human epilepsy, J. Membr. Biol. 222 (2008) 55–63.

[35] R.J.H. Smith, M. Hildebrand, DFNA2 Nonsyndromic Hearing Loss, GeneReviews[Internet] (1993).

Pathogenic effects of a novel mutation (c.664_681del) in KCNQ4 channels associated with auditory pathologyIntroductionMaterials and methodsSubjectsGenetic analysisPlasmid constructionCell culture and transient transfectionElectrophysiologyCell surface biotinylation and western blotting

ResultsClinical features of the KDF01 familyLinkage analysis and haplotype constructionIdentification of a novel mutation in KCNQ4 geneEffects of mutations on KCNQ4 channel function

DiscussionAcknowledgementsReferences