Mutations of SLC26A4 cause an enlarged vestibular aqueduct, nonsyndromic deafness, and deafness as part of Pendred syndrome. SLC26A4 encodes pendrin, an anion exchanger located in the cochlea, thyroid, and kidney. The goal of the present study was to determine whether developmental delays, possibly mediated by systemic or local hypothyroidism, contribute to the failure to develop hearing in mice lacking Slc26a4 (Slc26a4−/−). We evaluated thyroid function by voltage and pH measurements, by array-assisted gene expression analysis, and by determination of plasma thyroxine levels. Cochlear development was evaluated for signs of hypothyroidism by microscopy, in situ hybridization, and quantitative RT-PCR. No differences in plasma thyroxine levels were found in Slc26a4−/− and sex-matched Slc26a4+/− littermates between postnatal day 5 (P5) and P90. In adult Slc26a4−/− mice, the transepithelial potential and the pH of thyroid follicles were reduced. No differences in the expression of genes that participate in thyroid hormone synthesis or ion transport were observed at P15, when plasma thyroxine levels peaked. Scala media of the cochlea was 10-fold enlarged, bulging into and thereby displacing fibrocytes, which express Dio2 to generate a cochlear thyroid hormone peak at P7. Cochlear development, including tunnel opening, arrival of efferent innervation at outer hair cells, endochondral and intramembraneous ossification, and developmental changes in the expression of Dio2, Dio3, and Tectb were delayed by 1–4 days. These data suggest that pendrin functions as a HCO3− transporter in the thyroid, that Slc26a4−/− mice are systemically euthyroid, and that delays in cochlear development, possibly due to local hypothyroidism, lead to the failure to develop hearing.
- Pendred syndrome
mutation of the gene SLC26A4, which leads to a loss of function of the protein pendrin, causes an enlargement of the vestibular aqueduct as well as nonsyndromic or syndromic hearing loss including Pendred syndrome (22, 58). Hearing loss can be congenital or progressive during childhood (14, 16). This implies that a reduction or lack of pendrin function is compatible with hearing, but that this deficiency facilitates hearing loss in conjunction with other genetic and/or epigenetic factors.
Hypothyroidism and goiter are two variable manifestations of Pendred syndrome that may depend on nutritional iodine intake (33). Interestingly, the coincidence of goiter and deafness led to the first description of the syndrome (43). Hypothyroidism has variably been observed in patients with Pendred syndrome, which raises the question whether hypothyroidism is an epigenetic factor that contributes to the development of hearing loss in carriers of SLC26A4 mutations (8, 55, 57).
Studies of the consequences of a congenital loss of pendrin have recently been facilitated by the establishment of a mouse model (Slc26a4−/− mice) (21). Slc26a4−/− mice develop prenatally an enlargement of the endolymphatic compartment in the cochlea and the vestibular labyrinth that is consistent with the enlargement of the vestibular aqueduct and cochlear malformations that are seen in human patients (21). Compared with humans, the life span of mice is much compressed and mice are born developmentally immature. Wild-type and Slc26a4+/− mice are born deaf and begin to hear at postnatal day 12 (P12). Slc26a4−/− mice, however, fail to develop hearing (63). The present study was designed to determine whether developmental delays, possibly mediated by systemic or local cochlear hypothyroidism, may be an epigenetic factor that contributes to the failure to develop hearing in Slc26a4−/− mice and possibly in human carriers of mutations that lead to a loss of pendrin function.
Thyroid epithelial cells synthesize and secrete thyroglobulin (Tg) into the lumen of the follicle and use this protein as a matrix for the synthesis of the prohormone thyroxine (T4). Specific tyrosine residues are iodinated and coupled together by thyroperoxidase (Tpo) (5, 17). T4 synthesis requires normal ion transport including the transport of I− across the follicular epithelial cells into the lumen (42). I− is taken up across the basolateral membrane via the Na+/I− cotransporter (NIS), the product of the Slc5a5 gene. The mechanism of I− transport across the apical membrane is not yet fully understood (7, 33). Pendrin, which is an exchanger that transports anions such as Cl−, I−, and HCO3−, is involved in the secretion of I− across the apical membrane of thyroid follicular epithelial cells (24, 50, 51, 53, 56, 66). In addition to pendrin, several other I−-permeable transporters may contribute to I− transport including the anion/proton exchangers, ClC-3 and ClC-5, and the anion channels Cftr and Slc26a7 (6, 19, 40, 61, 65). Expression levels of pendrin in the murine thyroid have been found to be ∼50-fold less than in the kidney (61). In the kidney, cochlea, and in the vestibular labyrinth of the inner ear, pendrin has been found to secrete HCO3− and a lack of pendrin has been shown to reduce the luminal pH, consistent with a reduced rate of HCO3− secretion (37, 48, 63). It is therefore conceivable that pendrin also secretes HCO3− in the thyroid. These findings led to the first goal of the present study, which is to determine whether pendrin has a functional role in the murine thyroid, by measuring pH in thyroid follicles of Slc26a4+/− and Slc26a4−/− mice. In addition, we analyzed expression levels of genes involved in ion transport by gene array.
The thyroid gland releases mainly T4 and, to a much lesser extent, the active hormone tri-iodo-thyronine (T3). Plasma levels of T4 in mice surge and reach a peak at P15 (2, 11) to support the many processes that depend on thyroid hormone during postnatal development. Plasma levels of T4 have been found to be normal in adult Slc26a4−/− mice (21, 30). Adult mice, however, generate much smaller amounts of T4 compared with developing animals. It is conceivable that the importance of pendrin for T4 synthesis has been underestimated. Therefore, the second goal of the present study was to determine whether lack of pendrin compromises the surge in plasma T4 during early postnatal development. This goal was addressed by measurement of total T4 levels in plasma of Slc26a4+/− and Slc26a4−/− mice during early postnatal development.
Thyroid hormone is essential for normal growth and development of the cochlea, brain, and bone, among others. Consequently, hypothyroidism during early development leads to deafness, retarded neuronal development, and reduced bone mineralization and growth. Data from models of severe and prolonged hypothyroidism have established a catalog of events in the developing cochlea, brain, and bone that depend on thyroid hormone. This catalog includes opening of the tunnel in the organ of Corti of the cochlea at P5-P8 (18, 49), shaping of the tectorial membrane in the cochlea through the expression of the tectorial membrane-specific proteins α-tectorin (coded by the gene Tecta) and β-tectorin (coded by the gene Tectb) (31, 45), arrival of efferent innervation at cochlear outer hair cells (12, 60), expression of the neuron-specific gene synaptotagmin Srg1 (20, 44), and the maturation and mineralization of bone-forming chondrocytes and mesenchymal cells (1, 52). The cochlea controls T3 levels locally through the expression of two deiodinases, Dio2 and Dio3, and is therefore somewhat independent of circulating levels of T3 if sufficient amounts of T4 are provided (11, 39). Dio2 forms the active hormone T3 from the prohormone T4. Dio3 inactivates and reduces T3 levels. Dio3 forms the inactive T2 from T3 and furthermore depletes T4 by forming the inactive reverse tri-iodo-thyronine (rT3). In the inner ear, Dio3 gene expression is high during prenatal development and expression is largely lost postnatally. In contrast, Dio2 is not expressed during prenatal development but highly expressed in fibrocytes of the modiolus and the spiral ligament during early postnatal development, with a peak of Dio2 activity being reached at P7 (11, 39). Changes in the local expression of the two deiodinases can be expected to produce a locally confined hypo- or hyperthyroidism. Fibrocyte-containing tissues in the spiral ligament and spiral limbus of Slc26a4−/− mice have been reported to be reduced in thickness, which raises the question whether local T3 concentrations are reduced, thus leading to a locally confined hypothyroidism (62). The third goal of the present study was to determine whether the cochlea of Slc26a4−/− mice bears signs of hypothyroidism. This goal was addressed by localizing Dio2 expression with in situ hybridization, by monitoring the opening of the tunnel in the organ of Corti and the arrival of efferent innervation at outer hair cells by confocal microscopy, and by monitoring tissue mineralization in alizarin-stained tissue sections with laser-scanning microscopy. In addition, the expression of Dio2 and Dio3, which are potential effectors of hypothyroidism, and the expression of the Tectb and Srg1, which are potential reporters of hypothyroidism, were monitored by quantitative RT-PCR in Slc26a4+/− and Slc26a4−/− mice during early postnatal development.
A colony of Slc26a4−/− and Slc26a4+/− mice was established at Kansas State University from breeders kindly provided by Dr. Susan Wall (Emory University). Genetic drift of the colony was limited by back-crossing to the original strain (129SvEvTac obtained from Taconic, Germantown, NY). The colony was freed from helicobacter rodentium through a fostering approach and is maintained free of known and suspected murine pathogens. Serologic tests (Radil, Columbia, MO) that were consistently and repeatedly negative included the following antibodies: EDIM (epizootic diarrhea of infant mice virus–a mouse rotavirus), TMEV (Theiler's murine encephalomyelitis virus–mouse poliovirus, strain GDVII), MHV (mouse hepatitis virus–a mouse coronavirus), MVM (minute virus of mice–a mouse parvovirus), MNV (murine norovirus–a mouse calicivirus), M. pulmonis (mycoplasma pulmonis–the agent of murine mycoplasmosis), MPV (a mouse parvovirus), Parvo NS-1 (a conserved recombinant parvoviral protein, rNS1), and Sendai (Sendai virus–a type 1 paramyxovirus).
Mice used in this study were deeply anesthetized with 4% tri-bromo-ethanol and killed by decapitation or by transcardial perfusion. Preweaning mice received 0.013 ml/g body wt ip and adult mice 0.014 ml/g body wt ip of 4% tri-bromo-ethanol. Anesthesia of neonatal mice (P2) was supported by rapid cooling on an ice slush. All procedures involving animals were approved by the Institutional Animal Care and Use Committee of Kansas State University.
Thyroxine measurements in plasma.
Plasma was collected from sex-matched Slc26a4−/− and Slc26a4+/− littermates. Thyroxine levels were measured at the Diagnostic Laboratory in the College of Veterinary Medicine using the Immulite/Immulite 1000 Total T4 kit (Siemens Healthcare Diagnostics, Deerfield, IL).
Electrophysiological pH measurements.
The interstitial and follicular pH and the transepithelial potential were measured with double-barreled microelectrodes in situ. Thyroid glands were surgically exposed via a ventral approach. The slope of the pH-sensitive electrodes was 56 ± 3 mV/pH unit (n = 12). Procedures for in situ measurements of transepithelial potentials and pH were reported earlier (37, 63). Electrodes were calibrated in situ using three different pH values. Calibration solutions contained (in mM) 130 NaCl, 20 MES for pH 6, 130 NaCl, 20 HEPES for pH 7; and 130 NaCl, 20 tricine for pH 8. Data were recorded analog (Flat Bed Chart recorder, Kipp & Zonen, The Netherlands) for annotation and digital for presentation and data analysis (DIGIDATA 1322A and AxoScope 9, Axon Instruments). Data were analyzed using custom software written by P. W. in LabTalk (Origin 6.0, The Origin, Northhampton, MA).
Double-barreled glass microelectrodes were manufactured from filament-containing glass tubing (World Precision Instruments 1B100F-4, Sarasota, FL) using a micropipette puller (Narishige PD-5, Tokyo, Japan). Before silanization, microelectrodes were baked at 180°C for 2 h to ensure dryness. The longer ion-selective barrel was mounted in the lid of a beaker. The beaker was heated to 210°C and the electrodes were silanized by a 90-s exposure to 0.08 ml di-methyl-di-chloro-silane (Fluka 40136) at room temperature. The shorter reference barrel was protected from silanization by sealing the open end with Parafilm (Alcan Packaging, Chicago, IL). After silanization, microelectrodes were baked at 180°C for 3 h and tips were broken to a final outer diameter of ∼3 μm. The reference barrel was filled with 1 M KCl and the ion-selective barrel was filled at the tip with liquid ion exchanger (Hydrogen ionophore II–Cocktail A, Fluka 95297) and back-filled with buffer solution (500 mM KCl, 20 mM HEPES, pH 7.34). Each barrel was connected to an input of a grounded dual electrometer (World Precision Instruments FD223) via Ag-AgCl wire electrodes. The animal was grounded via a Ag-AgCl wire electrode inserted into the neck musculature. Current pulses (1 nA) were injected via the reference barrel to monitor electrode resistance.
Cochleae were fixed by transcardial and cochlear perfusion and postfixation. The transcardial perfusate was first NaCl solution (6 ml, 1 min) followed by NaCl solution containing 4% paraformaldehyde (24 ml, 4 min). Cochleae were removed and scala tympani and scala vestibuli were individually perfused with NaCl solution containing 4% paraformaldehyde (50 μl, 5 min). For experiments that only required phalloidin and DAPI staining, 0.5% glutaraldehyde was added to the paraformaldehyde solution. NaCl solution contained (in mM) 150 NaCl, 1.6 K2HPO4, 0.4 KH2PO4, 0.7 CaCl2, 1 MgCl2, and 5 glucose, pH 7.4. Postfixation consisted of 1- to 2-h incubation in paraformaldehyde solution at 4°C with rotation.
Temporal bones were decalcified in 10% EDTA, processed through a sucrose gradient, and infiltrated with polyethylene glycol. Midmodiolar cryosections (12 μm, CM3050S, Leica, Nussloch, Germany) were cut such that four cross-sections of the organ of Corti became visible. Sections from Slc26a4+/− and Slc26a4−/− littermates were put on the same slide to aid side-by-side evaluation. Images were taken from equivalent cross-sections. Sections on slides were blocked in 137 mM NaCl, 10.1 mM Na2HPO4, 1.8 mM KH2PO4, 2.7 mM KCl, pH 7.4 with 0.2% Triton X-100 (PBS-TX) and 5% BSA. Slides were incubated overnight at 4°C with primary antibody, rabbit anti-synaptophysin (1:100, Dako, Carpinteria, CA) in PBS-TX with 1–3% BSA. Slides were washed in PBS-TX and incubated for 1 h at room temperature with secondary antibodies, donkey anti-rabbit Alexa 488 (1:1,000, Invitrogen-Molecular Probes, Eugene, OR) in PBS-TX with 1–3% BSA. After incubation, slides were washed with PBS-TX and coverslipped with FluorSave (Calbiochem, La Jolla, CA). Cryosections were viewed by confocal laser-scanning microscopy (LSM 510 Meta, Carl Zeiss, Göttingen, Germany).
Temporal bones were fixed by immersion into Cl−-free solution containing 4% paraformaldehyde (1 h), processed through a sucrose gradient, and infiltrated with polyethylene glycol. Cl−-free solution contained (in mM) 150 Na-gluconate, 1.6 K2HPO4, 0.4 KH2PO4, 4 Ca-(gluconate)2, 1 MgSO4, and 5 glucose, pH 7.4. Midmodiolar cryosections (10 μm, CM3050S, Leica) were cut and stained with alizarin-red (2% in water, 5 s) and washed under running water (20 s). Cryosections were viewed by laser-scanning microscopy using similar intensities of red (633 nm), green (543 nm), and blue (488 nm) laser lights (LSM 510 Meta, Carl Zeiss). Transmitted light intensities were recorded and color images were generated by addition of red, green, and blue images.
Array-assisted gene expression analysis.
Total RNA was isolated from the thyroid of P15 sex-matched Slc26a4−/− and Slc26a4+/− littermates (RNeasy micro kit, Qiagen, Valencia, CA). Total RNA was amplified and converted to cDNA (Ovation RNA Amplification System V2, NuGen Technology, San Carlos, CA). This amplification system has earlier been evaluated in our laboratory (54). The cDNA was fragmented and biotinylated (FL-Ovation cDNA Biotin Module V2, NuGen Technologies) and hybridized to high-density oligonucleotide gene chips (mouse 430 2.0 gene chip, Affymetrix, Santa Clara, CA). RNA was isolated in our lab and processed for gene chip analysis at the Gene Expression Facility at Kansas State University. A total of six chips were run. Three chips each were used to analyze expression in the thyroid of Slc26a4+/− and Slc26a4−/− mice. Raw data were analyzed using commercial (GCOS, Affymetrix) and custom-written software (Excel, Microsoft, Redmond, WA). Quality metrics conformed with MIAME standards (Table 1). Data were submitted to the Geo database (accession number: GSE10589). Average intensities were calculated and the ratio (Slc26a4+/−/Slc26a4−/−) was computed. Ratios >0.7 and <1.3 were considered insignificant.
Quantitative RT-PCR was performed on total RNA isolated from the whole cochlea. The whole cochlea, rather than microdissected fractions, was chosen because it represents a well-defined preparation throughout development from P2 to P15, which includes the transition from a cartilaginous to a bony structure. The associated changes in the mechanical properties would present challenges to the preparation of microdissected fractions that would introduce unnecessary variations in the data. Thus, total RNA was isolated from the whole cochlea (RNeasy micro kit, Qiagen). Quantity and quality of total RNA were evaluated by microfluidic electrophoresis (BioAnalyzer, Agilent, Santa Clara, CA) and by microliter absorption photometry (Nanodrop, Wilmington, DE). RNA samples were accepted for quantitative RT-PCR only when they were free of contamination and had matching electrophoretic and photometric quantifications and excellent RNA quality. Concentrations of RNA samples were adjusted to 10 ng/μl using RNA storage solution (Applied Biosystems/Ambion, Austin, TX) and stored at −80°C before quantitative RT-PCR. Quantitative RT-PCR was carried out in 96-well plates (QuantiTect SYBR Green RT-PCR Kit, Qiagen; iCycler, Bio-Rad, Hercules, CA) using 10 ng of total RNA per well and gene-specific primers (Table 2). RNA samples from sex-matched Slc26a4+/− and Slc26a4−/− littermates were analyzed and referenced to 18S rRNA. Each reaction was carried out in duplicate. The average difference in Ct values between duplicates was 0.36 ± 0.1 SD (average of 21 plates, 96-well that examined high, intermediate, and low expressed genes). Reactions were assembled with the assistance of an automatic pipetting station (Biomek NXp, Beckman Coulter, Fullerton, CA) with hardware modifications and software programming by P. W. RT was performed for 30 min at 50°C and terminated by heating to 95°C for 15 min. PCR consisted of 40 cycles of 30-s annealing at 60°C, 30-s elongation at 72°C, 15-s hot measurement at 78°C, and 15-s denaturation at 94°C. Amplification of a single product of the appropriate size was verified by microfluidic electrophoresis (BioAnalyzer, Agilent). The number of template molecules (T) was estimated according to T = 10 ^log number of molecules at Ct/Ct * log (PCR efficiency). Ct was defined as the cycle at which the fluorescence of the product molecules reached a common threshold chosen in the middle of the log-linear part of the growth curve. The number of molecules at Ct was calibrated by amplifying 18S rRNA present in the samples. PCR efficiency was obtained from the slope of the log-linear phase of the growth curve (46).
In situ hybridization.
Temporal bones were fixed in PBS containing 4% paraformaldehyde (overnight), washed (PBS, 3 × 3 min), cryoprotected (30% sucrose, 2 h, 4°C), and embedded in polyethylene glycol (TissueTek, Torrance, CA). Cryosections (12 μm) were air-dried, stored at −80°C, brought to room temperature, washed (PBS, 3 × 5 min), treated with proteinase K (1 g/ml in PBS, 2 min), postfixed in PBS containing 4% paraformaldehyde (5 min), washed (PBS, 3 × 5 min), acetylated (0.1 M tri-ethanol-amine containing 0.25% acetic anhydride, 10 min), permeabilized (0.1% Triton X-100 in PBS, 10 min), and again washed (PBS, 3 × 5 min). Slides were then overlaid with 200 μl of hybridization buffer, coverslipped, and incubated (20 h, 68°C). Hybridization buffer contained 500 ng/ml digoxygenin-labeled Dio2 antisense or sense probe (39) in 4× Na-citrate buffer (600 mM NaCl, 60 mM Na-citrate, pH 7.0), 50% formamide, 10% dextran sulfate, 1× Denhardt's reagent, and 50 g/ml herring fish sperm DNA. After hybridization, slides were washed at 70°C first with 1× Na-citrate buffer (1×, 15 min) and then with 0.2× Na-citrate buffer (4×, 15 min). Further washes were performed at room temperature. Slides were washed first with 0.2× Na-citrate buffer and then with MABT buffer (2×, 30 min). MABT buffer contained 150 mM NaCl, 100 mM maleic acid, and 0.1% Tween 20, pH 7.5. Slides were blocked (MABT buffer with 2% blocking reagent; Roche, Mannheim, Germany; 2 h, room temperature), incubated with alkaline-phosphatase-conjugated sheep anti-digoxygenin FAB fragments [1:2,000 in MABT buffer with 2% blocking reagent (Roche), overnight at room temperature], and washed with MABT buffer (2×, 30 min). For color development, slides were washed (2×) and incubated (16 h, room temperature) with AP-staining buffer. AP-staining buffer contained 0.38 mg/ml 5-bromo-4-chloro-3-indolyl phosphate, 0.19 mg/ml p-nitroblue tetrazolium chloride, 100 mM NaCl, 5 mM MgCl2, 100 mM Tris·HCl, 0.1% Tween 20, pH 9.5. The color reaction was stopped by washing slides with PBS (2×, 5 min). Slides were then mounted in an aqueous mounting media (Aqua-Poly/Mount, Polysciences, Warrington, PA) and imaged (Nikon 80i microscope, NIS Elements imaging software, Nikon, Melville, NY).
Data are generally presented as average ± SE or, when indicated, as average ± SD with n being the number of animals. Data acquired in paired experiments using sex-matched littermates were evaluated by paired t-test; other data were compared by unpaired t-test. Significance was assumed when P < 0.05.
Lack of pendrin causes acidification of thyroid follicles and a reduced transepithelial potential.
Lack of pendrin reduces the luminal pH in the cochlea and the vestibular labyrinth, consistent with a reduced rate of HCO3− secretion (37, 63). The presence of a similar acidification of the lumen in the thyroid would indicate that pendrin has a physiological role in the murine thyroid. Thus, we measured the follicular pH and the interstitial pH as well as the transepithelial potential in the thyroid of adult Slc26a4+/− (P51 ± 9 SD, n = 8) and Slc26a4−/− (P53 ± 11 SD, n = 6) mice (Fig. 1). Lack of pendrin caused a reduction in the transepithelial potential and an acidification of the follicular pH. No significant difference was found in the pH of the interstitial fluid surrounding the follicles. These observations suggest that pendrin in the murine thyroid is functional and that pendrin in the thyroid, as previously shown for the kidney and the inner ear, is involved in the secretion of HCO3−.
Mice lacking pendrin generate a normal T4 surge during postnatal development.
T4 was measured in plasma collected from Slc26a4−/− and Slc26a4+/− mice at age P5, P10, P15, P30, and P80-P100. A pronounced surge of plasma T4 was observed during early postnatal development (Fig. 2). No significant differences in T4 levels between Slc26a4−/− and Slc26a4+/− mice were found. These observations underscore that pendrin is not essential for T4 production and not necessary for the generation of a normal T4 surge.
Lack of pendrin does not change expression levels of genes involved in T4 synthesis and epithelial ion transport in the thyroid.
Gene expression was evaluated by gene array analysis at P15, which is at the peak of the T4 surge. Total RNA was collected from thyroids of sex-matched P15 Slc26a4−/− and Slc26a4+/− littermates. A selected dataset focused on genes involved in T4 synthesis, ion transport, and epithelial barrier function is given in Fig. 3. Neither the expression of genes involved in T4 synthesis nor the expression of genes involved in epithelial ion transport was changed in the thyroid of Slc26a4−/− mice. No information on the expression levels of pendrin (Slc26a4), Cftr, and Clcn5 was provided since expression of these genes was below detection level in the array. Interestingly, however, the expression of the deiodinase Dio2 was increased. This finding was supported by three probes. It is tempting to speculate whether maintenance of thyroid function in the absence of pendrin requires some level of autocrine compensation. A similar increase in Dio2 expression has recently been observed in a Pendred syndrome patient (41).
Lack of pendrin leads to an enlargement of epithelia-lined scala media and results in a displacement of adjacent fibrocytes and mesenchymal cells.
Endolymphatic spaces in the cochlea and the vestibular labyrinth of Slc26a4−/− mice are known to be significantly enlarged from embryonic day 15.5 onward and this enlargement persists postnatally (21). We quantified the enlargement in the postnatal cochlea by measuring cross-sectional areas of scala media, scala vestibuli, and scala tympani in the lower three cross-sections of the cochlear duct. Data from equivalent cross-sections of Slc26a4+/− and Slc26a4−/− littermates (n = 5) aged P3-P7 were pooled. The cross-sectional area of scala media was 10.4-fold enlarged in Slc26a4−/− mice. In contrast, the cross-sectional area of scala vestibuli was 4.8-fold reduced and of scala tympani was 2.7-fold reduced (Fig. 4). The enlargement of scala media was mainly mediated by a distention of Reissner's membrane that included an increase in the number of cells. In contrast, the organ of Corti and stria vascularis appeared to contain the same number of cells. No enlargements of apical surface areas of cells in the organ of Corti had been noticed (21). Apical surface areas of marginal cells, however, are enlarged by a factor of 2 (27), which is consistent with a distention of stria vascularis roughly by a factor of 2 (Fig. 4).
The enlarged scala media not only bulged in apical direction into scala vestibuli and in basal direction into scala tympani, but also in centripetal direction into the modiolus and in centrifugal direction into the spiral ligament (Fig. 4). Bulging led to a displacement or a distortion of fibrocytes and mesenchymal cells in the spiral ligament and the modiolus, since the otic capsule roughly maintained normal dimensions. The space occupied by the spiral ligament, which is located in the lateral wall behind stria vascularis and reaches from scala vestibuli to scala tympani, was distended by roughly a factor of 2 (Fig. 4). Fibrocytes remained concentrated at the lower portion of the spiral ligament; however, the areas behind stria vascularis and the upper portion of the spiral ligament were depleted of fibrocytes. Displacement and local depletion of mesenchymal cells or fibrocytes may compromise cell signaling mediated by secreted molecules or requiring cell-to-cell contact. It is conceivable that regions of local hypothyroidism occurred due to increased diffusional distances between cells bearing thyroid hormone (TH) receptors and fibrocytes that transiently express Dio2 to generate T3 during early postnatal development. Hypothyroidism has well-known consequences on cochlear development. These consequences, in turn, can be taken as evidence for hypothyroidism.
Lack of pendrin leads to a local depletion of Dio2 expression.
Fibrocytes in the spiral ligament, underneath the spiral limbus, and surrounding the spiral ganglion in the modiolus are known to express Dio2, which encodes Dio2, the enzyme that generates T3 during early postnatal development of the cochlea (39). The observed displacement of fibrocytes implies that regions of local hypothyroidism may arise in the cochlea of Slc26a4−/− mice. To verify this concept, we localized Dio2 expression by in situ hybridization in P7 sex-matched Slc26a4+/− and Slc26a4−/− littermates. Expression of Dio2 was found in fibrocytes of Slc26a4+/− and Slc26a4−/− mice but was notably depleted in Slc26a4−/− mice in the space between stria vascularis and bone (Fig. 5). The observed depletion of Dio2 expression supports the concept of a local cochlear hypothyroidism.
Lack of pendrin delays opening of the tunnel during cochlear development and causes hypertrophy of the tectorial membrane.
Among the most easily observable effects of cochlear hypothyroidism is a delay of the opening of the tunnel in the organ of Corti (18, 49). Opening of the tunnel in the organ of Corti was evaluated in cryosections prepared from Slc26a4+/− and Slc26a4−/− littermates (Fig. 6). Sections were treated with phalloidin to stain actin filaments and with DAPI to stain nuclei. Transmitted light (bright field) images were recorded to visualize unstained structures such as the tectorial membrane. A delay in tunnel opening by ∼1 day and a persistent hypertrophy of the tectorial membrane were observed. Both observations are consistent with hypothyroidism.
Lack of pendrin delays arrival of efferent innervation in the cochlea.
Another easily observable consequence of hypothyroidism is a delay in the arrival of efferent nerve terminals at outer hair cells (12, 60). Arrival of efferents was evaluated in cryosections prepared from Slc26a4+/− and Slc26a4−/− littermates. Efferent nerve terminals were labeled using anti-synaptophysin antibodies (Fig. 7). A delay in the arrival of efferent innervation was observed. This observation provides further evidence for the hypothesis that the cochlea of Slc26a4−/− mice is hypothyroid.
Lack of pendrin alters Dio2 and Dio3 expression in the cochlea.
The cochlea controls T3 levels locally through the expression of two deiodinases, Dio2 and Dio3 (11, 39). Expression levels of Dio2 and Dio3 were monitored by quantitative RT-PCR in total RNA isolated from whole cochleae of sex-matched Slc26a4−/− and Slc26a4+/− littermates (Fig. 8A). Expression of Dio2 was found to peak between P6 and P8, which is consistent with the reported peak of cochlear Dio2 activity at P7 (11). Significant differences between Slc26a4+/− and Slc26a4−/− mice were observed at P6 and P8. Dio2 expression in Slc26a4−/− mice was at P6 lower and at P8 higher than in Slc26a4+/− mice. These differences are consistent with the peak of Dio2 expression in Slc26a4−/− mice being delayed by ∼1 day. Subsequently, Dio2 expression declined and this decline was slower in Slc26a4−/− mice compared with Slc26a4+/− littermates. Expression of Dio3 was found to drop low at P6 and to then to rise again till P10 (Fig. 8B). The low level expression found at P6 coincides with the peak in Dio2 expression and is consistent with peak levels of T3 at this time. Between P6 and P10, Dio3 expression rose again and this rise was delayed in Slc26a4−/− mice by ∼1 day. Taken together, these data suggest that the peak in T3 levels in the cochlea of Slc26a4−/− mice is delayed by ∼1 day.
Lack of pendrin alters Tectb but not Srg1 expression in the cochlea.
A small number of T3-sensitive genes is solely expressed in the organ of Corti or the modiolus of the cochlea and thereby can be used as a marker for thyroid hormone status in these regions. We selected two genes, Tectb and Srg1, to serve as reporters for T3 levels. Tectb (tectorin b) is exclusively expressed in a structure, the greater epithelial ridge, that develops into the spiral limbus and inner sulcus of the organ of Corti (31). Tectb was chosen as a reporter of T3 levels in the organ of Corti. Srg1 (synaptotagmin XII) is a neuron-specific gene (44). Although expression of Srg1 has not yet been localized in the cochlea, it is reasonable to expect that expression is limited to the neurons of the spiral ganglion cells in the modiolus. Thus, Srg1 was chosen as a reporter of T3 levels in the modiolus of the cochlea.
Expression of Tectb has been shown to strongly depend on thyroid hormone (31). In euthyroid rats, expression of Tectb steeply rises from P2 to a peak at P6 and rapidly declines between P8 and P12. In contrast, Tectb expression in severely hypothyroid rats was found to rise slower, reaching a lower level peak in expression at P8 and decline much more slowly thereafter (31). In sex-matched Slc26a4−/− and Slc26a4+/− littermates, we found high expression levels at P2, P6, and P8 (Fig. 8C). The following decline in Tectb expression was significantly slower in Slc26a4−/− mice compared with Slc26a4+/− littermates. These data suggest that the organ of Corti of the cochlea of Slc26a4−/− mice is hypothyroid.
Expression of Srg1 in the cerebellum has been shown to be reduced in hypothyroidism (20, 44). Thus, we used the expression Srg1 as a reporter of hypothyroidism in the spiral ganglion, where Srg1 is suspected to be expressed. No difference in expression was found between Slc26a4+/− mice and sex-matched littermates (Fig. 8D). This observation suggests that spiral ganglion cells in the modiolus are euthyroid.
Lack of pendrin delays tissue mineralization in the cochlea.
Bone formation in the cochlea consists of endochondral and intramembraneous ossification. Both ossification mechanisms, based on data from other bones, are known to depend on thyroid hormone (4, 13). A delayed or reduced tissue mineralization may therefore provide additional evidence for local cochlear hypothyroidism. Mineralization of the otic capsule occurs by endochondral bone formation and mineralization of the modiolus by intramembraneous bone formation. Tissue mineralization was visualized between P1 and P10 in alizarin-stained cochlear cryosections. Zones of resting, proliferative, hypertrophic chondrocytes, and chondrocytes that mineralized the extracellular matrix to form a primary spongiosum, were observed at P1 in Slc26a4+/− and Slc26a4−/− mice at the base of otic capule (Fig. 9A). Ensuing chondrocyte apoptosis compacts the primary spongiosum that is then remodeled into bone. At P1, the otic capsule of Slc26a4+/− and Slc26a4−/− mice consisted of a thick layer of chondrocytes (Fig. 10). At P4, a primary spongiosum was found throughout the otic capsule of Slc26a4+/− and Slc26a4−/− mice, although it appeared thinner in Slc26a4−/− mice compared with Slc26a4+/− mice. Interestingly, between P4 and P8, the primary spongiosum in parts of the otic capsule condensed in Slc26a4+/− mice into two layers of bone that provide in their middle a space for bone marrow (Fig. 10). Only one layer of bone was formed in Slc26a4−/− mice (Fig. 11A). At P6, the inner layer in Slc26a4+/− mice retained the structure of a primary spongiosum, whereas the outer layer appeared to be already more compacted (Fig. 11B). These data demonstrate a compromise in endochondral bone formation that is consistent with hypothyroidism.
Intramembraneous bone formation was observed to begin at P4 in the modiolus of the cochlea of Slc26a4+/− mice where mesenchymal cells surrounding the spiral ganglia of the cochlear nerve began to mineralize the extracellular matrix (Fig. 9B). Intramembraneous bone formation in the modiolus of Slc26a4−/− mice appeared to be delayed by ∼4 days (Fig. 10). The observed delay in intramembraneous bone formation is consistent with hypothyroidism.
Lack of pendrin does not alter tissue mineralization in phalangeal and metatarsal bones.
Bone formation of long bones such as phalangeal and metatarsal bones occurs by endochondral bone formation. A compromise of endochondral bone formation would be expected to lead to a reduced length of the ossified zone between the two growth plates located at opposite ends of the bone. Lengths of the mineralized bones were measured in alizarin-stained phalangeal and metatarsal bones from P8 Slc26a4+/− and Slc26a4−/− littermates (n = 4–5). No differences were found in the lengths of intermediate phalanges 2–5, proximal phalanges 1–5, and metatarsals 1–5 (data not shown). These findings demonstrate that the compromised endochondral bone formation that was observed in the cochlea is not a manifestation of a systemic defect.
The most salient findings of the present study on mice lacking pendrin are 1) that the thyroid has a more acidic follicular pH, generates a lower transepithelial potential but produces a normal plasma T4 surge during early postnatal development, and expresses normal levels of genes involved in T4 synthesis, epithelial ion transport, and barrier function, 2) that the cochlea develops a 10-fold enlargement of scala media which displaces fibrocytes and thereby may distort reciprocal epithelial-mesenchymal interactions that are necessary for normal cochlear development, and 3) that the cochlea shows signs of local hypothyroidism including delayed opening of the tunnel, delayed arrival of efferent innervation, and delayed changes in the expression levels of Dio2, Dio3, and Tectb, as well as reduced endochondral and delayed intramembraneous bone formation.
Pendrin mediates HCO3− secretion into the follicular lumen of the thyroid but is not essential for I− secretion or T4 generation.
The observations that the follicular pH in the thyroid of Slc26a4+/− mice is more alkaline than the surrounding interstitial fluid (Fig. 1) and that loss of pendrin leads to an acidification of the follicular pH suggest that alkaline equivalents such as HCO3− are among the anions transported by pendrin in the thyroid. HCO3− may be the most readily available anion since it is generated from metabolically produced CO2 through carbonic anhydrase-mediated conversion. HCO3− secretion as the significant function of pendrin in the thyroid is consistent with similar functions in the cochlea, the vestibular labyrinth, and the kidney (37, 48, 63). The physiologic relevance of HCO3− secretion in the thyroid may lie in an alkalinization of the apical membrane, which has been shown to facilitate binding of thyroglobulin before endocytosis (25). The observations, however, that Slc26a4−/− mice generated a normal surge in T4 (Fig. 2) and maintained normal expression levels of genes involved in T4 synthesis (Fig. 3) underscore that pendrin is not essential for I− transport or T4 synthesis.
The observations that the thyroid epithelium of mice lacking pendrin generated a smaller lumen-negative transepithelial potential (Fig. 1) but maintained normal expression levels of genes involved in ion transport and barrier function (Fig. 3) suggest that the ion transporters are controlled by translational or posttranslational mechanisms. Ion transport has been extensively studied in porcine primary cultures of the thyroid follicular epithelium, which secrete Cl− and absorb Na+ (3, 9). Primary cultures generate a lumen-negative transepithelial potential that is similar in magnitude to the transepithelial potential that we observe in situ in the murine thyroid. The lumen-negative transepithelial potential is most likely generated by a combination of K+ channels expressed in the basolateral membrane and Na+ and Cl− channels expressed in the apical membrane (9, 10, 36). A possible cause for the reduced transepithelial potential in mice lacking pendrin is an inhibition of the apical Na+ channels or the basolateral K+ channels by an acidic cytosolic pH (47). Whether the more acidic pH in the follicular lumen is associated with a more acidic cytosolic pH, however, is currently unknown.
Cochlear development is compromised by the enlargement of scala media.
Scala media of Slc26a4−/− mice develops an enlargement at embryonic day 15.5, which is shortly after the onset of pendrin expression (21). During early postnatal development, cross-sectional areas of scala media were found to be 10.4-fold enlarged and this enlargement persists into adulthood. It is currently unknown whether this enlargement is due to hypersecretion or hypoabsorption of osmolytes followed by water flux. The finding that the otic capsule roughly maintained normal dimensions is interesting given its formation is induced at embryonic day 12–13 of development, which is immediately before the onset of pendrin expression and also before the onset of the enlargement (35). It appears that the size of the otic capsule is defined at the time of induction of the otic capsule.
The enlargement of scala media within the confines of the otic capsule leads to a displacement and local depletion of mesenchymal cells or fibrocytes that may compromise cell signaling mediated by secreted molecules or requiring cell-to-cell contact. This possible compromise of reciprocal cell-to-cell interactions may be responsible for the delayed development of the cochlear epithelium and the delayed and incomplete ossification of the otic capsule. Although we focused our study on the hypothesis that a local hypothyroidism may contribute to the delayed development, it is conceivable that other cell signaling mechanisms are compromised by the distortion and that these mechanisms contribute to the observed developmental delays that result in the failure to develop hearing.
Cochlear development is possibly compromised by local hypothyroidism.
Development of the cochlea depends on a well-orchestrated sequence of events that include a narrow window during early postnatal development during which T3-mediated cell signaling must take place (32). Invariably, lack of T3 signaling at this critical period leads to a delay in the opening of the tunnel, a thickening of the tectorial membrane, a delay in the arrival of efferent innervation, together resulting in deafness. These hallmarks of hypothyroidism are observed regardless of whether hypothyroidism is due to reduced levels of the prohormone T4, as in Pax8−/− mice (15), Tshrhyt/hyt mice (34), Duox2−/− mice (28), and rats treated with propylthiouracil (59) or methimazole (32), or due to an inability to convert T4 to T3, as in Dio2−/− mice (38), or due to a lack of the T3 receptor, as in Thrb−/− mice (23).
Cochlear development depends on the local production of T3 by Dio2 that is temporarily expressed by fibrocytes located in the spiral ligament and the modiolus (11, 39). The observations that Dio2-expressing fibrocytes are displaced due to the bulging of the epithelial-lined scala media (Fig. 5) suggest that local cochlear hypothyroidism may occur as a consequence of a dilution in fibrocyte density and increased diffusional distances between fibrocytes and cells that bear TH receptors. Evidence for cochlear hypothyroidism in Slc26a4−/− mice comes from the observations of a delayed opening of the tunnel in the organ of Corti (Fig. 6), a hypertrophy of the tectorial membrane (Fig. 6), and a delayed arrival of efferent innervation at outer hair cells (Fig. 7). Delayed changes in the developmentally controlled expression levels of Tectb are also consistent with cochlear hypothyroidism (Fig. 8). A similar delay in the decline of Tectb expression has been observed in methimazole-treated rats (31). In rats, however, the primary finding was a delayed onset of Tectb expression, which occurs in rats at P5-7 and thereby can easily be monitored postnatally. The onset of Tectb expression in mice is apparently earlier such that only the delay in the decline of expression could be observed in the present study. A similar difference between rats and mice in the onset of cochlear genes has been reported for Kcnq4 (64).
The finding of no differences in expression levels of Srg1 appears on first sight to argue against the presence of cochlear hypothyroidism. Srg1 expression has been shown in the cerebellum to be sensitive to hypothyroidism (20, 44). It is conceivable, however, that cerebellar genes such as Srg1 are less sensitive indicators of hypothyroidism, which would be consistent with the general view that cochlear function is more sensitive to hypothyroidism than cerebellar function.
A caveat in the argument for the presence of cochlear hypothyroidism is that the delayed tunnel opening, the hypertrophy of the tectorial membrane, the delayed downregulation of Tectb expression, and the delayed arrival of efferent innervation could all be consequences of alterations in the epithelial cells that, in Slc26a4−/− mice, face a more acidic pH at their apical membrane (63). We therefore looked for markers of hypothyroidism in tissues that are not in contact with scala media and decided to investigate differences in cochlear bone formation. Precursor cells of bone are in contact with perilymph but not with endolymph. Unlike endolymph, no differences in the ionic composition or the pH of perilymph, or blood for this matter, have so far been found (62, 63). Both endochondral and intramembraneous bone formation are dependent on thyroid hormone (4, 13) and both forms of bone formation contribute to the ossification of the cochlea. Cochlear ossification may thus serve as an indicator of hypothyroidism. Endochondral bone formation is a two-step mineralization process that begins with chondrocytes proliferating, differentiating, mineralizing their extracellular matrix, and undergoing apoptosis. Osteoclasts and osteoblasts consequently remodel the mineralized extracellular matrix into bone. In contrast, intramembraneous ossification does not involve chondrocytes but depends on mesenchymal cells congregating and differentiating into osteoblasts that mineralize the extracellular matrix. Endochondral bone formation is the process leading to the ossification of the otic capsule and intramembraneous ossification is the process leading to the ossification of the modiolus. Consistent with cochlear hypothyroidism, endochondral bone formation of the otic capsule was reduced and intramembraneous ossification of the modiolus was delayed (Figs. 10 and 11). The use of these findings as arguments for hypothyroidism rests on the assumption that cochlear bone development resembles the formation of parietal and long bones, where the consequences of hypothyroidism have mainly been studied (1, 4, 52). It is conceivable that factors other than thyroid hormone are secreted by the fibrocytes of the spiral ligament and that the depletion of the fibrocytes leads to the depletion of these unknown factors and thereby to the reduced endochondral bone formation of the otic capsule.
Another argument for the presence of cochlear hypothyroidism comes from the finding that cochlear expression of the deiodinases Dio2 and Dio3 was delayed (Fig. 8). The observed delays in expression are most likely underestimations since expression was measured in the whole cochlea and cochlea development progresses from base to apex. The cause for the delays in Dio2 and Dio3 expression in Slc26a4−/− mice remains undetermined. The observed delays in expression changes of Dio2 and Dio3 can be expected to lead to a delay in T3 generation, which in conjunction with a dilution of Dio2 expressing fibrocytes and increased diffusion distances between these fibrocytes and T3 target cells would contribute to the manifestation of hypothyroidism is specific parts of the cochlea. The observed delay in downregulation of Dio2 expression is conceivably due to the delayed arrival of a negative feedback signal that is returned upon T3 stimulation from TH receptor-bearing target cells.
The pathway for T3 between fibrocytes that transiently express Dio2 and the cells in the organ of Corti that express TH receptors involves most likely mechanisms of T3 transport and two networks of gap junctions. The first network of gap junctions connects fibrocytes among each other. The second network connects epithelial cells of the outer sulcus with epithelial cells in the organ of Corti (29). Epithelial cells of the outer sulcus are also called root cells since they project long basolateral “roots” into the spiral ligament. It is conceivable that the function of these roots is to collect T3 from fibrocytes. Interestingly, prenatal disruption of these gap junction networks leads to a delayed opening of the tunnel, consistent with local hypothyroidism in the organ of Corti, and results in a failure to develop hearing (26). Postnatal disruption of these gap junction networks after P6, however, permits the development of hearing, which is consistent with the essential but transient requirement of T3 for cochlear development (X. Lin, Emory University, Abstract #686, “Mechanisms of deafness caused by connexin 26 and connexin 30 mutations studied in mouse models.” 32nd Midwinter Research meeting of the Association for Research in Otolaryngology, February 2009, http://www.aro.org/abstracts/abstracts.html).
In conclusion, hypothyroidism has repeatedly been observed in patients with Pendred syndrome, which raised the question whether hypothyroidism contributes to the development of hearing loss in people bearing mutations of SLC26A4. Our data suggest that pendrin functions as a HCO3− transporter in the murine thyroid and that Slc26a4−/− mice are systemically euthyroid but develop an enlargement of scala media, which leads to a displacement of fibrocytes and conceivably distorts reciprocal epithelial-mesenchymal interactions. This is possibly leading to a local cochlear hypothyroidism independent of the systemic thyroid status. Systemic hypothyroidism, which can be expected to aggravate local hypothyroidism, is likely a factor that promotes congenital deafness in people bearing mutation of SLC26A4.
We gratefully acknowledge the support by grants from the National Institute on Deafness and Other Communication Disorders National Institutes of Health (NIH)-R01-DC01098 to P. Wangemann and NIH-R01-DC00212 to D. C. Marcus, grants from the National Institute for Research Resources NIH-P20-RR017686-Project 2 to P. Fong, NIH-P20-RR017686-Core B to D. C. Marcus, and NIH-P20-RR017686-Core C to D. C. Marcus, and grants from College of Veterinary Medicine at Kansas State University SFYI to P. Fong and SMILE to P. Wangemann. Furthermore, the support of the intermural program at National Institute of Diabetes and Digestive and Kidney Diseases at the NIH to D. Forrest is gratefully acknowledged.
This work would not have been possible without the dedication of Dr. T. Miesner, S. Rose, J. Dille, and Dr. B. Carter from the Animal Resource Facility at Kansas State University, College of Veterinary Medicine. The authors thank Dr. J. Bai (Gene Expression Facility at Kansas State University), K. Smith (Diagnostic Laboratory in the College of Veterinary Medicine), and C. Linsenmayer (Laboratory Technician) for excellent services.
- Copyright © 2009 American Physiological Society