Differences in the Pathogenicity of the p.H723R Mutation of the Common Deafness-Associated SLC26A4 Gene in Humans and Mice

Mutations in the SLC26A4 gene are a common cause of human hereditary hearing impairment worldwide. Previous studies have demonstrated that different SLC26A4 mutations have different pathogenetic mechanisms. By using a genotype-driven approach, we established a knock-in mouse model (i.e., Slc26a4tm2Dontuh/tm2Dontuh mice) homozygous for the common p.H723R mutation in the East Asian population. To verify the pathogenicity of the p.H723R allele in mice, we further generated mice with compound heterozygous mutations (i.e., Slc26a4tm1Dontuh/tm2Dontuh) by intercrossing Slc26a4+/tm2Dontuh mice with Slc26a4tm1Dontuh/tm1Dontuh mice, which segregated the c.919-2A>G mutation with an abolished Slc26a4 function. Mice were then subjected to audiologic assessments, a battery of vestibular evaluations, inner ear morphological studies, and noise exposure experiments. The results were unexpected; both Slc26a4tm2Dontuh/tm2Dontuh and Slc26a4tm1Dontuh/tm2Dontuh mice showed normal audiovestibular phenotypes and inner ear morphology, and they did not show significantly higher shifts in hearing thresholds after noise exposure than the wild-type mice. The results indicated not only the p.H723R allele was non-pathogenic in mice, but also a single p.H723R allele was sufficient to maintain normal inner ear physiology in heterozygous compound mice. There might be discrepancies in the pathogenicity of specific SLC26A4 mutations in humans and mice; therefore, precautions should be taken when extrapolating the results of animal studies to humans.


Introduction
Mutations in the SLC26A4 (PDS, GeneID: 5172) gene are the second most frequent cause of human hereditary hearing impairment worldwide, next to mutations in the GJB2 (GeneID: 2706) gene [1]. In some populations, SLC26A4 mutations can be identified in approximately 13-14% of deaf patients [2]. SLC26A4 encodes pendrin, an iodide/chloride/bicarbonate transporter expressed in the inner ear [3], thyroid [4], kidney [5,6], salivary duct [7], and respiratory tract [8]. Recessive SLC26A4 mutations contribute to both Pendred syndrome (PS; MIM #274600) [9] and nonsyndromic hearing loss (DFNB4; MIM #600791) [10], which share the phenotypes of sensorineural hearing impairment (SNHI) accompanied by an enlarged vestibular aqueduct (EVA; MIM 603545) and/or incomplete partition of the cochlea (i.e., Mondini dysplasia), although the phenotype of PS also includes goiter. To date, more than 100 SLC26A4 mutations have been identified (Pendred/BOR Homepage; www.healthcare.uiowa. edu/labs/pendredandbor). Previous reports have described that different SLC26A4 genotypes were correlated with distinct clinical phenotypes, and patients with PS are more likely to have 2 SLC26A4 mutant alleles than those with DFNB4 [11,12]. Many affected patients suffer from progressive or fluctuating hearing loss [13], implying that the natural course can be halted with preventive or therapeutic measures if the pathogenetic mechanisms of SLC26A4 mutations are better elucidated.
In recent years, the understanding of the pathogenesis of DFNB4 and PS has been accelerated by various mouse models with mutations in the Slc26a4 (GeneID: 23985) gene. Certain mouse models revealed congenital profound hearing loss, including the knock-out Slc26a4 2/2 mice [14], the Slc26a4 loop/loop mice with the p.S408F mutation [15], and the Slc26a4 tm1Dontuh/tm1Dontuh mice with the c.919-A.G mutation that we previously reported [16]. The conditional knock-out Tg [E];Tg[R];Slc26a4 D/D mice demonstrated hearing loss of various severity dependent on the time of Slc26a4 expression, with doxycycline initiation at E18.5 resulting in partial hearing loss [17]. These mouse models have provided excellent insight into the pathogenesis; however, 2 basic problems might hurdle the bench-to-bedside translation. First, similar to their human counterparts, mice with different mutations, to some extent, demonstrated different phenotypes, indicating that the pathology associated with each distinct mutation is different. Second, to date, no mouse model that can perfectly simulate the progressive or fluctuating hearing loss in humans has been reported. Investigating mice with other Slc26a4 mutations might tackle these problems. Accordingly, in this study, we generated a knock-in mouse model with the p.H723R (c.2168A.G) mutation, a common SLC26A4 mutation in the East Asian population [18][19][20][21][22], and then, we characterized the associated audiovestibular phenotypes as well as the inner ear pathology.

Construction of Slc26a4 tm2Dontuh/tm2Dontuh Knock-in Mice
The mutation gene-targeting vector was constructed using a recombineering approach previously developed by Dr. Copeland's group [23,24]. From the bMQ323G13 BAC clone (Sanger Institute, Cambridge, UK), we subcloned a 12.8-kb fragment spanning introns 17-21 of Slc26a4 into the PL253 plasmid (Fig. 1A). The subcloned genomic 12.8-kb region was modified in a subsequent targeting round by inserting the neomycin (neo) cassette from the PL452 plasmid and creating the c.2168A.G mutation in exon 19. The targeting vector was then linearized by NotI digestion and electroporated into R1 embryonic stem (ES) cells. G418 (240 mg/mL) and ganciclovir (2 mM) double-resistant clones were analyzed by Southern blot hybridization (Fig. 1B). The loxP-flanked neomycin resistance gene (neo) was used as a selection marker during embryonic stem (ES) cell culture. The c.2168 A.G mutation in exon 19 is labeled with a star. (3)(4) The targeted locus was introduced, and then, the neo cassette was removed. LA, long arm; SA, short arm. (B) Southern blot analysis of ES cell clones. Genomic DNA from 2 targeted and 2 wild-type clones were digested with XbaI and hybridized with the probe to verify the targeting event. (C) DNA sequencing of Slc26a4 +/+ and Slc26a4 tm2Dontuh/tm2Dontuh mice. The electrophoretogram shows the p.H723R mutation. The A to G mutation (arrow) at position 2168 led to the replacement of a histidine (His, H) residue at position 723 with arginine (Arg, R). doi:10.1371/journal.pone.0064906.g001 The retained neo cassette flanked by loxP sites was excised in vivo by transfecting the targeted clone with plasmid transiently expressing the Cre recombinase. Established ES clones were then identified by polymerase chain reaction (PCR) screening and subsequently injected into C57BL/6 blastocysts to produce chimeras. After germline transmission of the targeted mutation allele, we produced the congenic Slc26a4 +/tm2Dontuh mouse line used in this study by repeated backcrossing into the C57BL/6 inbred strain for 6-10 generations, after which mice homozygous for the mutation (i.e., Slc26a4 tm2Dontuh/tm2Dontuh ) were obtained by intercrossing heterozygous mice (i.e., Slc26a4 +/tm2Dontuh ) (Fig. 1C). Reverse transcription-PCR (RT-PCR) of mRNA of inner ear extract followed by direct sequencing also indicated a pure non-chimeric genetic background without unintentionally wild-type Slc26a4 expression in Slc26a4 tm2-Dontuh/tm2Dontuh mice. Corresponding to the human genotypes, mice with compound heterozygous mutations for p.H723R and c.919-2A.G (i.e., Slc26a4 tm1Dontuh/tm2Dontuh ) were also generated by intercrossing heterozygous Slc26a4 +/tm2Dontuh mice with Slc26a4 tm1-Dontuh/tm1Dntuh mice. All animal experiments were carried out in accordance with animal welfare guidelines and approved by the Institutional Animal Care and Use Committee (IACUC) of National Taiwan University College of Medicine (approval no. 20110123).

Audiological and Vestibular Evaluations
For audiological evaluations, the mice were anesthetized with sodium pentobarbital (35 mg/kg) delivered intraperitoneally and placed in a head-holder within an acoustically and electrically insulated and grounded test room. We used an evoked potential detection system (Smart EP 3.90; Intelligent Hearing Systems, Miami, FL, USA) to measure the thresholds of the auditory brainstem response (ABR) in mice. Click sounds, as well as 8, 16, and 32 kHz tone bursts at varying intensity, were generated to evoke ABRs in mice. The response signals were recorded with subcutaneous needle electrodes. The active electrodes were inserted into the vertex and the ipsilateral retro-auricular region with a ground electrode on the back of the mice.
For vestibular evaluations, mice were subjected to a battery of tests, including observation of their circling behavior and head tilting (performed at 3 weeks of age), reaching test, swimming test, gripping test, and a rotarod test (all performed at 8 weeks of age). The methodology of each vestibular test is described in details in our previous study [16].

Inner Ear Morphology Studies
Tissues from the inner ears of mice were subjected to hematoxylin and eosin (H&E) staining, and the morphology of  8,16, and 32 kHz) at 1, 3, 6, and 9 months in mice with different genotypes. Heterozygous mice (i.e., Slc26a4 +/tm2Dontuh ), homozygous mice (i.e., Slc26a4 tm2Dontuh/tm2Dontuh ), and compound heterozygous mice (i.e., Slc26a4 tm1Dontuh/tm2Dontuh ) showed normal hearing as wild type mice (Slc26a4 +/+ ) up to 9 months. doi:10.1371/journal.pone.0064906.g002 each sample was examined with a Leica optical microscope. For both light microscopy and scanning electron microscopy (SEM) studies, inner ears from adult mice were fixed by perilymphatic perfusion with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) through round and oval windows and a small fenestra in the apex of the cochlear bony capsule. Specimens were subsequently rinsed in PBS buffer and decalcified in 4% PFA with 0.35 M EDTA at 4uC for 1 week. For light microscopy studies, the samples were dehydrated and embedded in paraffin. Subsequently, serial sections (7 mm) were stained with H&E. For SEM studies, the samples were dehydrated in ethanol, critical-point dried, gold sputter coated, and then examined in a field emission scanning electron microscope (S-4500; Hitachi, Tokyo, Japan).  Whole-mount studies of mouse inner ear specimens were performed as previously described [25] with some minor modifications. Briefly, after perfusion with 4% PFA, the cochleae were postfixed in the same solution for 2 h at room temperature and washed in PBS. The samples were permeabilized in 1% Triton X-100 for 30 min and washed with PBS, followed by overnight incubation at 4uC in the blocking solution. The tissues were then stained with rhodamine-phalloidin (1:100 dilution; Molecular Probes, Eugene, OR, USA). After washing in PBS, the tissues were mounted using the ProLong Antifade kit (Molecular Probes, Eugene, OR, USA) for 20 min at room temperature. Images of the tissues were obtained using a laser scanning confocal microscope (Zeiss LSM 510; Germany).

Expression of Pendrin
For pendrin expression experiments, we prepared tissue sections from the inner ears of Slc26a4 tm1Dontuh/tm2Dontuh and Slc26a4 tm2Dontuh/ tm2Dontuh mice. Tissue sections mounted on silane-coated glass slides were then deparaffinized in xylene and rehydrated in ethanol. After antigen heat retrieval (500 W microwave oven, in 10 mM citric buffer, pH 6.0, for 20 min), the slides were incubated overnight at 4uC with primary antibodies in PBS and Tween (PBST) (rabbit anti-pendrin, 1:100 [H195]; mouse anti-Myosin VIIa, 1:100 [C-5]; Santa Cruz Biotechnology, Santa Cruz, CA, USA). The slides were then washed and incubated for 1 h at 25uC with appropriate secondary antibodies at a 1:1000 dilution in PBST. After incubation, the slides were washed with PBST and mounted with the ProLong Antifade kit at 25uC. Images were obtained using a laser scanning confocal microscope (Zeiss LSM 510; Germany). Real-time PCR Total RNA was purified using Trizol reagent (Invitrogen, CA, USA) according to the manufacturer's protocol and stored at 280uC until further use. Total RNA was treated with DNase and then reverse transcribed into first-strand cDNA in a 20-mL reaction volume using SuperScript III Reverse Transcriptase (Invitrogen, CA, USA). Quantitative real-time reverse transcription (RT)-PCR assays of Kcnj10 cDNA were performed using a gene-specific double fluorescently labeled TaqMan probe in an ABI Prism 7900 Sequence Detection System (Applied Biosystems) in accordance with the manufacturer's recommendations. The relative mRNA expression level was determined by the 2 2DDCt analysis method, and calculations were performed using the software provided by the manufacturer (Applied Biosystems). Specimens from 5 mice were collected for each experiment, and the experiments were performed in triplicates and averaged. The average mRNA expression level in wild-type mice was normalized to 1 as the control.

Western Blot Analysis
Protein extracts of the inner ear were homogenized in RIPA buffer (Millipore, MO, USA). Equal amounts of proteins were supplemented with dithiothreitol, heated for 5 min at 95uC, separated by SDS gel electrophoresis, and then transferred to a PVDF membrane (Amersham Pharmacia Biotech, Little Chalfont, UK) by semi-dry electroblotting. The PVDF membranes were developed with an enhanced chemiluminescence western blot detection kit (Pierce SuperSignalH West Dura, Rockford, IL, USA) and exposed to Lumi-Film chemiluminescent detection films (Roche Diagnostics, Mannheim, Germany). The specimens were collected from 5 mice for each experiment, and the experiments were performed in triplicates and averaged. The average protein expression level in wild-type mice was normalized to 1 as the control.

Noise Exposure Experiments
For the noise exposure experiments, 10 mice of each genotype were exposed for 3 h to octave-band noise with a peak at 4 kHz, 115 dB SPL. The noise room was fitted with a speaker (model NO. 1700-2002, GSI) driven by a noise generator (GSI-61; Grason-Stadler, Inc.) and a power amplifier (5507-Power; TECHRON). ABR thresholds (dB SPL) in mice at different frequencies (click, 8 kHz, 16 kHz, and 32 kHz) were then recorded at 30 min and at day 1, 2, 3, 7, and 14 after noise exposure.

Thyroid and Renal Serum Biochemistry
Terminal blood samples were obtained from mice at postnatal day 15 to compare the early postnatal T4 surge [26], as well as at 2 and 6 months of age to monitor the long-term thyroid and renal

Immunolocalization and Expression of Pendrin
We then investigated the expression of pendrin in the cochlea of Slc26a4 tm2Dontuh/tm2Dontuh mice and Slc26a4 tm1Dontuh/tm2Dontuh mice (Fig. 5A) by immunolocalization. In both strains of mice, pendrin was normally distributed in the spiral prominence and root cells, indicating that the expression of pendrin was not affected by the p.H723R mutation in mice. Compared with the wild-type mice, no significant difference in the molecular weight (Fig. 5B) or the expression level (Fig. 5C) of pendrin were observed in Slc26a4 tm2-Dontuh/tm2Dontuh mice and Slc26a4 tm1Dontuh/tm2Dontuh mice by western blotting analyses.

Kcnj10 Expression
We then investigated the expression of Kcnj10 (GeneID: 16513) in Slc26a4 +/+ mice, Slc26a4 tm2Dontuh/tm2Dontuh mice, and Slc26a4 tm1-Dontuh/tm2Dontuh mice by real-time PCR (for mRNA expression) and western blotting (for protein expression) analyses. It has been demonstrated that Slc26a4-depleted mice showed decreased Kcnj10 expression, which contributes to the failure of endocochlear potential generation [27]. KCNJ10 (GeneID: 3766) mutations have been observed in EVA patients via digenic inheritance with SLC26A4 mutations [28]. In this study, the stria vascularis of P15 mouse cochleae from Slc26a4 +/+ mice, Slc26a4 tm2Dontuh/tm2Dontuh mice, and Slc26a4 tm1Dontuh/tm2Dontuh mice were isolated by microdissection, and total RNA and protein extracted from these tissue fractions were used for real-time PCR and quantitative immunoblot analyses. Compared with the wild-type mice, no significant differences in the mRNA (Fig. 5D) or protein levels of Kcnj10 (Fig. 5E) were observed in Slc26a4 tm2Dontuh/tm2Dontuh mice and Slc26a4 tm1Dontuh/tm2Dontuh mice.

Noise Exposure Experiments
We then attempted to induce audiologic phenotypes in transgenic mice with noise exposure [29]. Neither Slc26a4 tm2Dontuh/tm2Dontuh mice nor Slc26a4 tm1Dontuh/tm2Dontuh mice showed a significantly higher shift in hearing thresholds at all frequencies 30 min and at day 1, 2, 3, 7, and 14 after noise exposure than Slc26a4 +/+ mice (Fig. 6); this finding confirmed their normal audiologic phenotypes.

Discussion
In this study, we generated a knock-in mouse model, denoted as Slc26a4 tm2Dontuh/tm2Dontuh mice, which segregates the common deafness-associated p.H723R mutation in humans. The p.H723R mutation was found in both nonsyndromic EVA and PS families [18,30]. This missense mutation was the most prevalent SLC26A4 mutation among the Japanese [18] and Koreans [19], and the second most prevalent SLC26A4 mutation among the Han Chinese [20,21]. Recent reports indicated that the p.H723R mutation arose from a common ancestor and was not a mutational hot spot [31]. The pathogenicity of the p.H723R mutation in humans was confirmed by several lines of evidence supporting the fact that p.H723R co-segregated with the phenotypes in the affected families, the prevalence of p.H723R was low in the control populations, and the p.H723 amino acid residue was evolutionally conserved [18]. Moreover, in human cell lines, the p.H723R has been related to trafficking/folding/ glycosylation defects of the pendrin protein. In transfected HEK293 cells and HeLa cells, p.H723R caused protein retention in endoplasmic reticulum and abolished complex glycosylation of pendrin, and the defects in protein processing could be restored considerably by low temperature incubation or treatment with sodium butyrate [32]. Further study revealed the folding defect in the p.H723R-pendrin, and treatment of salicylate, which functioned as a pharmacological chaperone, could restore normal protein localization and anion exchange activity [33].
Interestingly, as shown in the present study, both heterozygous mice (i.e., Slc26a4 +/tm2Dontuh ) and homozygous mice (i.e., Slc26a4 tm2-Dontuh/tm2Dontuh ) with the knock-in p.H723R mutation had normal audiovestibular phenotypes, including normal hearing thresholds up to 9 months, excellent balancing ability, and an intact cochlear and vestibular morphology. There are several possibilities why the p.H723R mutation did not lead to the expected abnormal phenotype in mice. The first possibility is that the abnormal phenotype was present but not yet discovered or that the abnormal phenotype will become evident only under certain environmental conditions [34]. Although this possibility could not be completely ruled out by the results of this study, it was largely weakened after an exhaustive investigation of the audiovestibular phenotypes, including examination of hearing levels up to 9 months, a battery of vestibular tests, comprehensive inner ear morphological studies, and noise exposure experiments. The second possibility is that the phenotypic effects of the p.H723R mutation are influenced by the genetic background of the mice [35]. This was also unlikely; as in our previous study, we generated a knock-in Slc26a4 mouse model with abnormal audiovestibular phenotypes by using the same C57BL/6 strain. Another possibility, which became the most likely scenario for our study after the former 2 possibilities were refuted, is that, there are, in fact, no abnormal phenotypes.
In particular, the absence of abnormal audiovestibular phenotypes in homozygous mice alone is not sufficient to exclude the pathogenicity of p.H723R in mice because the function of the Slc26a4 p.H723R mutation might be only partially defective, and even harboring 2 p.H723R alleles might still maintain adequate Slc26a4 function. To elucidate the pathogenicity of the p.H723R allele, we further generated mice with compound heterozygous mutations (i.e., Slc26a4 tm1Dontuh/tm2Dontuh ) by intercrossing Slc26a4 +/ tm2Dontuh mice with Slc26a4 tm1Dontuh/tm1Dontuh mice, which segregated the c.919-2A.G mutation with abolished function. Mice that were compound heterozygous for p.H723R and c.919-2A.G (i.e., Slc26a4 tm1Dontuh/tm2Dontuh ) had normal audiovestibular phenotypes, indicating that a single p.H723R allele was sufficient to maintain normal inner ear physiology. Moreover, the expression of the downstream Kcnj10 gene, which diminished at P10-P15 following oxidative stress in Slc26a4 2/2 mice [36], was not compromised in Slc26a4 tm2Dontuh/tm2Dontuh mice and Slc26a4 tm1Dontuh/tm2Dontuh mice. These findings indicate that the p.H723R allele is neither pathogenic in mice nor is a mutation with partially ablated function.
As shown in Figure 5, both Slc26a4 tm2Dontuh/tm2Dontuh and Slc26a4 tm1Dontuh/tm2Dontuh mice expressed pendrin of molecular weight and expression level comparable to the wild-type mice, indicating that the trafficking/glycosylation process remained unaffected in the murine p.H723R-pendrin. The difference in the intactness of the trafficking/glycosylation process might contribute to the variation in the pathogenicity of p.H723R between mice and humans. We further analyzed the amino-acid sequences of human and mouse pendrin using ConSeq (http://conseq.tau.ac. il/), a website server predicting biologically important residues in protein [37]. Possibly the intact trafficking/glycosylation process of p.H723R in mice might be attributed to a different alignment of amino acid residues in the vicinity, as well as the embedded location of p.H723 in the pendrin (Fig. S1).
Four other mouse models with Slc26a4 mutations have been reported: knock-out Slc26a4 2/2 mice [14], Slc26a4 loop mice with the p.S408F mutation generated by ENU mutagenesis [15], Slc26a4 tm1Dontuh knock-in mice with the common East Asian c.919-2A.G mutation [16], and conditional knock-out Tg[E];Tg[R];Sl-c26a4 D/D mice [17] (Table 1). Slc26a4 m ice are profoundly deaf with significant vestibular deficits and have EVA and scala media, mimicking the phenotypes in humans [14]. Subsequent research on Slc26a4 loop mice [15] and Slc26a4 tm1Dontuh/tm1Dontuh mice [16] reported similar auditory and vestibular characteristics, complementing the studies on mice. To better simulate the less severe audiological phenotype in humans, Choi et al. generated a binary transgenic mouse line in which Slc26a4 expression could be induced with doxycycline. The authors identified that the E16.5 to P2 was the critical interval in which pendrin was required for acquisition of normal hearing, and demonstrated that doxycycline initiation at E18.5 or discontinuation at E17.5 resulted in partial hearing loss, instead of the profound hearing loss observed in the knock-out Slc26a4 2/2 mice [17]. In contrast, Slc26a4 tm2Dontuh/ tm2Dontuh mice examined in this study had normal audiovestibular phenotypes, representing a distinct and unique mouse strain compared with the other mouse models reported in the literature.
Among humans, patients with PS are more likely to have 2 SLC26A4 mutant alleles than those with nonsyndromic EVA [11,12]. Moreover, the number of SLC26A4 mutant alleles is significantly correlated with the severity of hearing loss in individuals with EVA [38,39]. However, because of its diverse mutation spectrum, it is difficult to delineate the phenotypes associated with a specific SLC26A4 mutation in humans. Consequently, determining the pathogenetic mechanisms of each specific SLC26A4 mutation largely relies on functional studies performed in Xenopus oocytes [40] or cell lines [41,42]. The observation that mice with distinct Slc26a4 mutations have different phenotypes indicates that transgenic mice may serve as appropriate, direct models to investigate corresponding SLC26A4 mutations in humans.
In conclusion, using a genotype-driven approach, we generated a knock-in mouse model segregating the common deafnessassociated SLC26A4 p.H723R mutation in humans. To our surprise, mice with the Slc26a4 p.H723R mutation had a normal audiovestibular phenotype and inner ear morphology. Because there might be differences in the pathogenicity of specific SLC26A4 mutations in humans and mice, precaution should be taken when extrapolating the results of animal studies to humans. Figure S1 Alignment of amino-acid sequences of human and mouse pendrin. The amino-acid sequence (a.a. 651-780) of human pendrin (hum-pendrin) was aligned in relative to the sequence of the mouse pendrin (mse-pendrin) using Conseq. Arrows indicate the p.H723 position. The p.H723 is a highly conserved but buried amino acid residue. Different alignments of amino acid residues in the vicinity of p.H723 and the embedded location of p.H723 in the pendrin might contribute to the variation in the pathogenicity of p.H723R between mice and humans. The first row below the sequence lists the predicted burial status of the site (b, buried; e, exposed). The second row indicates residues predicted to be structurally (s) and functionally (f) important.