Thyroid hormone (TH) is essential for the development of hearing. Lack of TH in a critical developmental period from embryonic day 17 to postnatal day 12 (P12) in rats and mice leads to morphological and functional deficits in the organ of Corti and the auditory pathway. We investigated the effects of TH on inner hair cells (IHCs) using patch-clamp recordings, capacitance measurements, and immunocytochemistry in hypothyroid rats and athyroid Pax8−/− mice. Spontaneous and evoked Ca2+ action potentials (APs) were present in control IHCs from P3–P11 rats and vanished in parallel with the expression of a rapidly activating Ca2+- and voltage-activated K+ (BK) conductance. IHCs of hypothyroid rats and athyroid Pax8−/− mice displayed APs until the end of the third postnatal week because of threefold elevated Ca2+ currents and missing expression of BK currents. After the fourth postnatal week, some IHCs showed BK currents whereas adjacent IHCs did not, demonstrated by electrophysiology and immunocytochemistry. To test whether the prolonged spiking activity during TH deficiency may be transmitted at IHC synapses, capacitance measurements were performed in parallel to analysis of otoferlin expression, a protein thought to play an essential role in exocytosis of IHCs. Strikingly, otoferlin was absent from IHCs of hypothyroid rats but not of Pax8−/− mice, although both cell types showed exocytosis with an efficiency typical for immature IHCs. These results demonstrate for the first time a TH-dependent control of IHC spiking activity before the onset of hearing attributable to effects of TH on Ca2+ and BK channels. Moreover, they question an indispensable role of otoferlin for exocytosis in IHCs.
Thyroid hormone (TH) or triiodothyronine (T3) and its receptors are essential for the development of hearing. Congenital thyroid disorders impair hearing, and iodine deficiency causes profound deafness (Dussault and Ruel, 1987; DeLong, 1993; Forrest, 1996). Furthermore, hypothyroidism causes deformities in the organ of Corti and has indicated a critical window of TH requirement during inner ear development preceding the onset of hearing (Deol, 1973; Uziel, 1986; O'Malley et al., 1995; Knipper et al., 2000).
T3 receptors (TRs) are ligand-dependent transcription factors encoded by the related genes TRα and TRβ (Sap et al., 1986; Weinberger et al., 1986). Deletion or mutations of thyroid hormone receptor β (TRβ) cause hearing impairment or deafness in mice (Forrest et al., 1996) and humans (Refetoff et al., 1967; Brucker-Davis et al., 1996), whereas a role of TRα1 in hair cell differentiation has only recently emerged (Winter et al., 2006).
Hearing loss in TRβ-deficient mice was correlated with a severe retardation of a fast-activating potassium conductance, IK,f, in the inner hair cells (IHCs) (Rusch et al., 1998). IK,f transforms the immature IHC into a sensory cell producing graded receptor potentials at the onset of hearing [around postnatal day 12 (P12)] (Kros et al., 1998). Before the upregulation of the rapidly activating Ca2+- and voltage-activated K+ (BK) channel underlying IK,f (Langer et al., 2003; Pyott et al., 2004; Hafidi et al., 2005), IHCs generate spontaneous Ca2+ action potentials (APs) despite lack of sensory input (Kros et al., 1998; Marcotti et al., 1999). This spontaneous activity triggers exocytosis (Beutner and Moser, 2001; Johnson et al., 2005), leads to EPSPs on the afferent nerve fibers (Glowatzki and Fuchs, 2002), and is thought to trigger maturation processes in the auditory pathway. Because Ca2+ APs rely on voltage-activated Ca2+ currents (Beutner and Moser, 2001; Marcotti et al., 2003b; Johnson et al., 2005), the consequences of deleting the predominant IHC Ca2+ channel Cav1.3 (Platzer et al., 2000), lack of Ca2+ APs and lack of BK channel expression, indicate that Ca2+ APs also contribute to the differentiation of IHCs (Brandt et al., 2003).
Here, we investigated whether neonatal Ca2+ APs were affected by TH deficiency using two animal models: (1) Pax8−/− mice that show the most severe hypothyroid phenotype because of absence of thyroid follicular cells (Mansouri et al., 1998) and develop an inner ear phenotype typical for hypothyroidism (Christ et al., 2004), and (2) rat pups treated with methyl-mercapto-imidazol (MMI) during fetal and postnatal development, leading to hypothyroidism and deafness (Deol, 1973; Uziel, 1986; Knipper et al., 2000).
In both animal models, TH deficiency caused a prolongation of spiking activity because of a changed expression of Cav1.3 and BK currents. Capacitance measurements revealed robust exocytosis in TH-deficient IHCs in the entire phase of spiking activity. Strikingly, the IHC protein otoferlin (Yasunaga et al., 1999; Schug et al., 2006) was absent from hypothyroid rat IHCs at the mRNA and protein level, thus questioning the essential role of otoferlin for IHC exocytosis (Roux et al., 2006).
In a parallel study, Sendin et al. (2007) analyzed the effects of TH deficiency on the maturation of ion channels, presynaptic proteins, and exocytosis in Pax8−/− mice.
Materials and Methods
Hypothyroid Wistar rats (Charles River, Sulzfeld, Germany) and athyroid Pax8−/− mice (Mansouri et al., 1998) were used. Hypothyroidism in rats was induced by treating mothers and pups with methyl-mercapto-imidazol (Knipper et al., 2000). Pax8−/− mice were born from heterozygous Pax8+/− parents; their euthyroid littermates Pax8+/− and Pax8+/+ (Mansouri et al., 1998; Christ et al., 2004) served as controls (here designated as Pax8 control). Animals were killed by decapitation (after CO2 anesthesia for ages older than P9) and cochleae were removed according to national ethical guidelines. Animal experiments were approved and undertaken in accordance with the protocol requirements of the University of Tuebingen.
The IHCs of rats or mice were studied in acute preparations of the organ of Corti after the cell somata had been partially exposed with cleaning pipettes and positive or negative pressure. Spiking activity and whole-cell K+ currents in IHCs younger than P12 were recorded in extracellular solution B1 [containing the following (in mm): 153 NaCl, 5.8 KCl, 1.3 CaCl2, 0.9 MgCl2, 0.7 NaH2PO4, 10 HEPES, 5.6 glucose, pH 7.4, 320 mOsm kg−1]. For IHCs P12 or older, a solution with a reduced Cl− concentration was used [B1-mature; containing the following (in mm): 85 NaCl, 70 lactobionate × NaOH, 5.8 KCl, 1.3 CaCl2, 0.9 MgCl2, 0.7 NaH2PO4, 10 HEPES, 5.6 glucose, pH 7.4, 320 mOsm kg−1], which better preserved the organ of Corti and the IHCs (Siegel et al., 2001). The pipette solution I1 consisted of (in mm) 110 K-gluconate, 20 KCl, 10 Na+ phosphocreatine, 5 HEPES, 5 EGTA, 4 MgCl2, 4 Na2ATP, 0.1 CaCl2, and 0.3 GTP, pH 7.35, 305 mOsm kg−1.
To isolate currents through Ca2+ channels from large K+ (and neonatal Na+) currents, 10 mm Ba2+ was used as a charge carrier, as well as a combination of intracellular and extracellular K+-channel blockers. For IHCs younger than P12, the extracellular solution B2 consisted of (in mm) 160 tris-(hydroxy-methyl-amino)-methane × HCl, 10 BaCl2, 5.6 glucose, and 1 MgCl2, pH 7.4, 320 mOsm kg−1 (Michna et al., 2003). For IHCs P12 or older, B2-mature was used [containing the following (in mm): 70 lactobionate × NaOH, 40 NaCl, 35 tetraethylammonium chloride (TEA), 15 4-aminopyridine (4-AP), 10 BaCl2, 10 HEPES, 5.6 glucose, and 1 MgCl2, pH 7.4, 320 mOsm kg−1]. The pipette solution for isolating Ba2+ currents, I2, was composed of (in mm) 120 Cs+ gluconate, 20 CsCl, 10 Na+ phosphocreatine, 5 HEPES, 5 EGTA, 4 MgCl2, 4 Na2ATP, 0.1 CaCl2, and 0.3 GTP, pH 7.35, 305 mOsm kg−1 (Michna et al. 2003). All chemicals were obtained from Sigma (St. Louis, MO) with the exception of ATP, TEA, lactobionate (Fluka, Neu-Ulm, Germany), and iberiotoxin (Alomone Labs, Jerusalem, Israel).
Whole-cell recordings were performed from IHCs of the apical or medial cochlear turn (before P12) or of the apical cochlear turn (P12 or later) at room temperature (22°C) using the whole-cell patch-clamp technique in either current-clamp or voltage-clamp mode with an Axopatch 200B amplifier (Molecular Devices, Union City, CA). Patch pipettes with a resistance of 4–6 MΩ were pulled from quartz glass. Currents were corrected off-line for linear leak currents and voltages were corrected for the uncompensated series resistance and for liquid junction potentials [cf. Neher (1992), B1/I1 = −12.6 mV; B1-mature/I1 = −9.6 mV; B2/I2 = −20 mV; B2-mature/I2 = +2.3 mV, to be added to the respective voltages].
Data were acquired using the software Pulse++ 1.7 (developed by Ulrich Rexhausen, Institute of Physiology II, Tuebingen, Germany, and Klaus Bauer, Max Planck Institute for Medical Research, Heidelberg, Germany) and an ITC16 interface (Instrutech, Greatneck, NY) and were stored on a Macintosh G3 computer (Apple, Cupertino, CA). Voltage traces lasting 10 s were sampled at 5 kHz and filtered at 1.3 kHz; short current traces lasting 30 ms were sampled at 100 kHz and filtered at 10 kHz, and longer current traces (500 ms) at 5 kHz/1.3 kHz.
Data analysis was performed using the Igor software package (Wavemetrics, Lake Oswego, OR) as described previously (Michna et al., 2003); statistical analysis was done with JMP IN 5.1 Software (SAS Institute, Cary, NC). CV denotes coefficient of variation and R2 the coefficient of determination.
A transient voltage change was considered as an action potential when its peak was ≥10 mV positive from the resting potential and its width was <100 ms. How frequently spontaneous and induced spiking occurred at different ages in the two rat groups (control vs hypothyroid) was quantified by logistic regression on covariables age, exp(age), factor MMI treatment, and the group–exp(age) interaction. For Pax8 control and Pax8−/− mice, a simpler model without exponential function and interaction sufficed. The maximum spiking frequency of an IHC was determined as the maximum frequency at which the cell could generate Ca2+ APs, either spontaneously or induced, by increasing injections of current before the membrane potential became arrested in a depolarized state. Resting membrane potentials were determined as average membrane potentials in nonspiking cells. In spontaneously active cells, resting potentials were determined using the voltage of the lower peak of the voltage histogram (bin width, 0.5 mV) in a 10 s recording episode. The membrane potential as a function of postnatal age and MMI treatment was modeled by a four-parameter logistic curve with different ages of half-maximal decrease for the control and the MMI-treated group. Parameters of this nonlinear regression were estimated by the least squares method. Peak current–voltage (I–V) curves of K+ currents (IK) were determined using a peak search routine and by averaging IK in a 2 ms window centered around the time of peak for each voltage (i.e., these currents were determined not necessarily at the same time point of the voltage-induced current responses). In contrast, I–Vs for Ba2+/Ca2+ currents and for the fast component of IK were determined at fixed time points (averaged for the following millisecond) at every voltage, as indicated in the respective figures and legends. Activation time constants of K+ currents were determined by monoexponential fits to the current traces for either 2 ms (when currents activated rapidly) or 7 ms (when only slowly activating currents were present).
The age-dependent rise of peak K+ currents in Pax8 mice was described by analysis of covariance (ANCOVA) of logarithms to obtain constant variance. A group–age interaction allowed for different slopes. Variance of logarithms of exocytosis efficiency was explained (ANOVA) by factors group, age (where applicable), stimulus duration, all of their interactions, and random cell effects.
To analyze exocytosis, we measured increases in the membrane capacitance after voltage steps of different durations, which lead to fusion of synaptic vesicles with the plasma membrane (Neher and Marty, 1982). The “track-in” mode of the Optopatch amplifier (Cairn Research, Faversham, UK) was used where a self-balancing lock-in amplifier is integrated into the patch-clamp circuitry, detecting and compensating changes of membrane resistance and membrane capacitance in real-time (Johnson et al., 2002). IHCs were placed in B1 (before P12) or B1-mature solution (P12 or later) and superfused with a solution containing 5 mm Ca2+ and K+-channel blockers [B3; containing the following (in mm): 110 NaCl, 5 CaCl2, 1 MgCl2, 5.6 glucose, 10 HEPES, 35 TEA, and 15 4-AP, pH 7.4, 320 mOsm kg−1]. Ca2+ dependence of exocytosis was tested with a nominally Ca2+-free superfusion (B4, for which solution B3 was modified by replacing 5 mm CaCl2 with 1 mm MgCl2 and 10 mm NaCl). The pipette solution for measuring changes in membrane capacitance, I3, corresponded to I2 (see above) except that Cs+ methane sulfonate (105 mm) replaced Cs+ gluconate (120 mm).
At a holding potential of −80 mV, a sine wave (frequency, 0.9–1.2 kHz) was applied to the IHC with an amplitude small enough (10–20 mV rms) to avoid significant changes in membrane conductances, which was interrupted by a single voltage step from −80 to 0 mV to trigger Ca2+ influx and exocytosis. The resulting capacitance change (ΔCm) was calculated as the difference of the mean capacitance over 300 ms after depolarization (the first 100 ms were skipped) (Moser and Beutner, 2000) and the mean prepulse capacitance (300 ms). Voltage steps with different durations (5, 10, 20, 40, 80, and 100 ms) were applied in random order.
For whole-mount immunocytochemistry of BKα, the cochlea was injected with 2% paraformaldehyde for 15 min on ice, rinsed with PBS, and the external bone was removed. The modiolus with the organ of Corti attached was stained as described for cochlear cryosections (Knipper et al., 1998, 2000). Antibodies against BKα (Alomone Labs) and neurofilament NF200 (The Binding Site, Heidelberg, Germany) were used. For otoferlin staining, cochlear cryosections were used for which cochleae older than P9 were decalcified for 1–5 min with Rapid Bone Decalcifier (Fisher Scientific, Houston, TX). Otoferlin antibodies were generated by immunizing a rabbit (Pineda, Berlin, Germany) with three synthetic peptides (amino acids 238–263, 1296–1317, 1370–1384; GenBank accession number AF183183) of mouse otoferlin (Schug et al., 2006). The antibody was purified against the amino acids 1370–1384, and specificity was tested by blocking with the specific peptide. Primary antibodies were detected with Cy3-conjugated (Jackson ImmunoResearch, West Grove, PA) or Alexa Fluor 488-conjugated antibodies (Invitrogen, Eugene, OR). Specimens were embedded with Vectashield mounting medium containing 4,6,diamidino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, CA) to stain cell nuclei (blue) and viewed using an Olympus (Tokyo, Japan) AX70 microscope equipped with epifluorescence illumination. Staining was detected in at least three specimens of at least three animals.
Riboprobe synthesis and in situ hybridization.
To generate an otofer-lin-specific riboprobe (Schug et al., 2006), a 634 bp PCR fragment of otoferlin was amplified by reverse transcriptase-PCR using GoTaq DNA polymerase (Promega, Mannheim, Germany) with the primers 5′ -gcttcctgcctaccctcgg-3′ and 5′ -cagcttgtcagcaatgtgatcc-3′ from rat cochlear cDNA. The PCR product was sequenced and otoferlin specificity was confirmed. The fragment was cloned into the pCRII-Topo Vector (Invitrogen) and used for in vitro transcription. Complementary strands for sense and antisense were transcribed from either SP6 or T7 promoter sites in the presence of digoxigenin (DIG)-labeling mix (Roche Diagnostics, Mannheim, Germany).
In situ hybridization (ISH) was performed as described previously (Knipper et al., 1998). Briefly, DIG-labeled antisense or sense probes were diluted in hybridization solution containing 25% microarray hybridization buffer (GE Healthcare, Freiburg, Germany), 25% nuclease free water, and 50% formamide. Sections were stained by an anti-DIG-antibody coupled to alkaline phosphatase for 20 h.
Effects of TH deficiency on spiking activity in rat IHCs
The organs of Corti of postnatal hypothyroid rats were characterized by an immature morphology and an altered “sticky” extracellular matrix. Exposure of IHC somata was more difficult than in controls, leading to lower success rates of whole-cell recordings.
Spontaneous oscillations of the membrane potential occurred in neonatal IHCs, independent of the TH level from P3 onwards shown for control and hypothyroid rats at P7 (Fig. 1A,B). These action potentials (Ca2+ APs) depended on the presence of extracellular Ca2+ (data not shown) and resembled the Ca2+ APs studied in neonatal mouse IHCs (Kros et al., 1998; Marcotti et al., 1999, 2003b, 2004b). Rapidly activating and deactivating Na+ currents were present in <20% of control and hypothyroid IHCs, indicating that Na+ currents were not necessary for the generation of Ca2+ APs. Their amplitude was rather small at P7 (Fig. 1A) when they looked like symmetric voltage oscillations, and increased toward P11, when all IHCs tested displayed spontaneous Ca2+ APs (Fig. 1C,D). The spiking frequency of hypothyroid IHCs exceeded by far the frequencies observed in control IHCs of the same age (Fig. 1C,D). By P13, spontaneous Ca2+ APs had completely vanished in control IHCs that displayed quite negative resting potentials of −70.1 ± 3.3 mV (n = 5). Injection of positive current led to gradual shifts of the membrane potential in the positive direction rather than eliciting any Ca2+ APs (Fig. 1E), indicating the successful transition from a neonatal spiking pacemaker to a mature IHC capable of graded transmitter release (Kros et al., 1998). In contrast, IHCs of hypothyroid rats of the same age continued to generate high-frequency Ca2+ APs, shown here for P15 (Fig. 1F), yet with reduced amplitudes.
At P19, the IHCs of both control and hypothyroid animals did not spike spontaneously anymore. In hypothyroid IHCs, but not in control IHCs, Ca2+ APs could be evoked by injection of small positive currents (2–20 pA), shown for a hypothyroid IHC at P19 (Fig. 2A–D). When comparing the spontaneous Ca2+ APs in IHCs as a function of age (Fig. 2E), relevant differences in the time period in which Ca2+ APs can be induced (Fig. 2F) and the maximal spiking frequency of IHCs (Fig. 2G) in control and hypothyroid animals can be observed. Whereas control IHCs never did spike spontaneously after P12, hypothyroid IHCs showed Ca2+ APs until P15 (Fig. 2E). Logistic regression confirmed later cessation of spontaneous spiking in hypothyroid IHCs. Current injections increased the percentage of spiking control IHCs only from P2 to P5 (Fig. 2F). From P12 onwards, no more Ca2+ APs could be elicited. In hypothyroid IHCs, the percentage of spiking cells could be increased by current injections from P3 to P8 and P15 to P19. Again, logistic regression revealed relevant differences in the termination of induced spiking between both animal groups. It has to be noted that before P7, only negative current injections of −10 to −40 pA could trigger Ca2+ APs, because of the depolarized potential of both control and hypothyroid IHCs (see below).
Another striking difference between IHC properties of control and hypothyroid rats was the maximum spiking frequency achieved in either spontaneous or evoked Ca2+ APs. The average maximum frequencies of Ca2+ APs in control IHCs shortly before the onset of hearing were ∼10 Hz at P7 and P9 and dropped to <5 Hz at P11 (Fig. 2G). At P7, the IHCs of hypothyroid rats displayed average maximum frequencies of Ca2+ APs that were comparable with control IHCs but increased to values as high as 25 Hz at P12. Thereafter, maximum frequencies declined, but were still elevated compared with the highest values of control IHCs achieved at P9. Analysis of the dependence of spike frequencies on the resting membrane potential revealed differences for IHCs from control and hypothyroid rats (Fig. 2H,I) (for determination of the resting potential, see Material and Methods). Spontaneous Ca2+ APs of control IHCs with frequencies up to 7 Hz were generated from resting potentials of −65 to −53 mV (with one exception: −44 mV, 12 Hz) whereas those of hypothyroid IHCs with frequencies from 7 to 24 Hz arose from resting potentials of −60 to −47 mV (Fig. 2H). Current injections were able to induce Ca2+ APs and increased their frequencies as shown in Figure 2I for typical control (filled symbols, asterisk) and hypothyroid IHCs (unfilled symbols) of different ages. They increased the spiking frequency and shifted the membrane potential in the positive direction until IHCs stopped spiking and adopted a depolarized state. Whereas the control IHCs (filled symbols and asterisk) and the P8 hypothyroid IHCs (unfilled rectangle) showed frequency-resting potential relationships with similar steepness, those of P12 and P19 hypothyroid IHCs displayed much steeper curves (unfilled circles and triangles), suggesting a different ratio between depolarizing and repolarizing conductances. Finally, the resting membrane potentials of control and hypothyroid IHCs were compared (Fig. 2K), which shifted from about −40 mV after birth to about −75 mV at the end of the fourth postnatal week in both control and hypothyroid IHCs. Under the assumption of equal upper and lower asymptotes of the logistic curves, the fit resulted in residual SD of 7.2 mV or of R2 = 74% of total variance of 72 observations. The half-maximal membrane potential change from newborn to mature occurred at P16 in hypothyroid rats and P10 in controls (difference, 5.6 d, SE, 0.12 d) (Fig. 2K).
Effects of TH deficiency on Ca2+ currents in rat IHCs
The persistence of high-frequency Ca2+ APs in the IHCs of hypothyroid rats beyond the onset of hearing of control animals raises the question about differences in the underlying conductances. Regenerative Ca2+ APs are based on the interplay of the negatively activating Cav1.3 Ca2+ channel and a slowly activating delayed rectifier-type K+ channel (Muenkner et al., 2002). In normally developing mouse IHCs, Ca2+ APs cease when the BK conductance is expressed (Kros et al., 1998). In parallel, the Cav1.3 conductance peaking at ∼P6 declines (Beutner and Moser, 2001; Marcotti et al., 2003b). We first analyzed IHC Ca2+ currents in control and hypothyroid rats by using 10 mm Ba2+ as the charge carrier.
Figure 3A,B shows representative Ba2+ current (IBa) traces that were evoked by step depolarizations for 8 ms. Both control and hypothyroid IHCs showed rapidly activating and deactivating inward currents that did not inactivate. I–V curves calculated by averaging the current during the last millisecond of the voltage step and plotting it as a function of voltage revealed a larger IBa in the hypothyroid IHC compared with the control IHC at P9, with similar voltage dependence of IBa in the control and the hypothyroid IHC (Fig. 3C). Peak-current amplitudes were determined for both IHC groups for different ages. IBa of control rat IHCs showed a postnatal upregulation and downregulation with a peak amplitude between P9 and P12 and a subsequent downregulation to smaller values (40% of the peak) (Fig. 3D). This effect was comparable with the upregulation and downregulation in mouse IHCs, in which the peak occurs 3–4 d earlier (P6) (Beutner and Moser, 2001; Marcotti et al., 2003b). IHCs of hypothyroid rats at P3 had the same low IBa values, but from P10 onwards adopted values much larger than those of control IHCs, reaching an average of 824.9 ± 231 pA at P19, which corresponds to 180% of the peak values of the control at P9 and P12. At P24, IBa was still higher than the maximum values in control IHCs, and only at P32 IBa showed the same small IBa values as mature control IHCs.
These data demonstrate remarkably increased Ca2+ channel currents in IHCs of hypothyroid rats, strongly suggesting that TH is involved in the developmental downregulation of the Ca2+ conductance. Moreover, the elevated Ba2+/Ca2+ current amplitudes favor the occurrence of Ca2+ APs seen in hypothyroid IHCs as long as the K+ conductances are sufficiently large to repolarize the cell. Otherwise, the membrane potential would become arrested in a depolarized state. We therefore analyzed whole-cell K+ currents as a function of age (Fig. 4).
Effects of TH deficiency on K+ currents in rat IHCs
Whole-cell K+ currents (IK) elicited by step depolarizations were similar in neonatal control and hypothyroid IHCs as shown for P2 (Fig. 4A,B). At P19, the peak IK amplitude of a control IHC had increased sixfold compared with that at P2 and also showed faster activation kinetics (Fig. 4C), which is consistent with the expression of the fast BK current (Kros et al., 1998; Langer et al., 2003; Marcotti et al., 2003a; Pyott et al., 2004). An age-matched IHC from a hypothyroid rat (P19) (Fig. 4D) showed much smaller K+ currents with kinetics similar to those of the P2 hypothyroid IHC, but with a doubled peak amplitude. K+ currents of a P32 hypothyroid IHC showed increased amplitudes and accelerated activation kinetics, but were still smaller than those of the P19 control IHC (Fig. 4E). The differences between IK amplitudes of control and hypothyroid IHCs are further illustrated by the corresponding peak current I–V curves, in which the respective maxima of each current trace were plotted as a function of the membrane potential (Fig. 4F). Figure 4G summarizes peak IK amplitudes as a function of age in control and hypothyroid IHCs: until P11, peak K+ currents of both groups increased moderately and did not differ from each other. From P11 onwards, K+ currents of control IHCs showed a steep rise, consistent with the expression of BK channels around and after the onset of hearing, whereas those of the hypothyroid IHCs increased with age at a much lower rate and stayed smaller compared with controls until P30.
To confirm that hypothyroid IHCs lack expression of BK channels, K+ currents were recorded in response to short (8 ms) step depolarizations, making the differences in activation kinetics more obvious (Fig. 5A–E). Whereas the P2 control IHC and the IHCs of the P2 and P19 hypothyroid animals completely lacked any fast K+ current component (IK,f) (Fig. 5A,B,D), rapid currents that were almost fully activated at 1.2 ms after depolarization (Fig. 5A–D, vertical dashed lines) were recorded in the control P19 IHC (Fig. 5C) and the hypothyroid P32 IHC (Fig. 5E). This is also illustrated by the I–V curves that were calculated between 1.2 and 1.3 ms after the start of depolarization (Fig. 5F).
To directly demonstrate that the BK current underlies the rapidly activating outward K+ current, we used iberiotoxin, a specific peptide blocker for BK channels (for review, see Meir, 2003). However, iberiotoxin did not block all IK,f of a P18 control IHC, even when superfused at concentrations of 100 nm for 10 min, as shown by the black current traces (Fig. 5G, thick trace, before application; thin trace, after 10 min iberiotoxin application). In contrast, application of 10 μm paxilline, another BK channel blocker (Meir, 2003), inhibited IK,f completely (Fig. 5G, gray traces). The differential blocking action of iberiotoxin and paxilline on IK,f in control IHCs is also shown by the I–V curves (Fig. 5H). On average, 100 nm iberiotoxin blocked 87.7 ± 8.5% of IK,f at −10 mV (n = 3), whereas 10 μm paxilline blocked 100.9 ± 1.2% (n = 3). A block of >100% indicates that the complete block of IK,f unmasked the small (100 pA) Ca2+ inward current. Paxilline unmasked the typical delayed rectifier current in control IHCs, which was similar to those shown in Figure 4, A, B, and D, and had an average peak amplitude of 5.7 ± 0.8 nA at 0 mV (n = 3) (data not shown).
Fitting monoexponential functions to the current traces before and after application of iberiotoxin or paxilline, respectively, revealed that (1) IK,f in control IHCs without any blocker activated with time constants of ≤1 ms over the whole voltage range, (2) the iberiotoxin-resistant current was still rapidly activating, albeit slightly slower than the total current, and (3) with paxilline application, only a slow current remained with time constants ranging from ≥6 ms at ∼0 mV to 2–3 ms at ∼25 mV (Fig. 5I). These results confirm the heterogeneity of IHC whole-cell BK currents in the rat that has been described previously for the mouse, where IK,f was composed of an iberiotoxin-sensitive and an iberiotoxin-resistant part, with the latter displaying slightly slower activation kinetics (Marcotti et al., 2004a). They further indicate that the time constant analysis is suitable to test for the presence of BK currents in IHCs. Figure 5K shows activation time constants (τact) of typical IHCs to demonstrate the clear segregation into fast τact values covered by the mature control IHCs (observed in 13 of 13 IHCs) and slow τact values covered by P9 control IHCs (five of five), P9 hypothyroid IHCs (four of four), and hypothyroid IHCs aged P19 to P27 (nine of nine). Hypothyroid IHCs aged P28–P75 adopted a dispersed pattern: part of the IHCs gained a conductance nearly as fast as that of control IHCs at P19 (9 of 34) whereas others did not (25 of 34), indicating a delayed acquisition of the BK conductance in part of the hypothyroid IHCs. Interestingly, within one and the same animal, IHCs displaying the fast K+ current were in direct neighborship of IHCs devoid of the fast current. Figure 5L shows activation time constant curves for selected IHCs of two different hypothyroid organs of Corti. In a P31 organ of Corti, two IHCs possessing the BK current and one lacking it were recorded, whereas in a P38 organ of Corti, vice versa. The dispersed BK current expression pattern within the same organ of Corti up to P75, but also the apparent lack of BK currents in any IHC tested of two hypothyroid litters at ages P38–P48, underlines the heterogeneity of the hypothyroid phenotype.
The mosaic pattern of BK channel expression was further confirmed using whole-mount immunocytochemistry (Fig. 6). The typical BK channel staining around the IHC neck was seen in each IHC of P30 control rats (Fig. 6A). In contrast, in a P35 hypothyroid rat, IHCs were devoid of any BK staining (Fig. 6B). At P53, many IHCs of a hypothyroid rat displayed the BK staining around the neck (Fig. 6C, filled arrowhead), although some of them were BK negative (Fig. 6C, open arrowheads).
Neonatal spiking activity and Ca2+ and K+ currents in athyroid Pax8−/− mice
To clarify whether the delays and deficits in IHC maturation observed in hypothyroid rats were independent of the mode of induction of TH deficiency, we analyzed Ca2+ APs and currents in athyroid Pax8−/− mice and their euthyroid littermates Pax8+/− and Pax8+/+ (Christ et al., 2004), further denoted as Pax8 controls. Within the first postnatal week, IHCs of Pax8 control and Pax8−/− mice generated spontaneous Ca2+ APs of similar amplitude and frequency (data not shown). Pax8 control IHCs had stopped spontaneous spiking at P8, whereas Pax8−/− IHCs did so after P14 (Fig. 7A). Injection of small positive currents prolonged the period of spiking until P12 in Pax8 control IHCs and until P21 in Pax8 −/− IHCs (Fig. 7B). Older ages could not be tested in Pax8−/− mice because of their death around weaning. Logistic regressions of spiking on postnatal age confirmed later cessation ages in Pax8−/− IHCs compared with Pax8 controls for both spontaneous and induced spiking. Maximum spiking frequencies of Pax8 control and Pax8−/− IHCs were in the range of 5–25 Hz and were not different between the two groups (data not shown).
The presence of Ca2+ APs in Pax8−/− IHCs in the third postnatal week most likely was attributable to the same reasons as in hypothyroid rat IHCs: failure of downregulation of the Ca2+ current and failure of upregulation of the BK current. The central tendency of ICa was compared between Pax8 controls and Pax8−/− by van der Waerden test, because normality of the small number of values could not be verified. There was a significant (p < 0.0001) difference between median ICas of Pax8−/− and Pax8 controls (Fig. 7C), indicating that ICa was not downregulated in Pax8−/− IHCs. Peak K+ currents were of similar sizes in neonatal (P2–P4) Pax8 control and Pax8−/− IHCs, indicating that the neonatal delayed rectifier K+ current was not affected by lack of TH (Fig. 7D). Thereafter, average peak K+ currents became significantly larger in Pax8 control compared with Pax8−/− IHCs [e.g., the K+ current of Pax8−/− was estimated to be 43% (CI, 31 to 58%) of the K+ current of controls at P16]. This difference was caused by the lack of IK,f expression in Pax8−/− IHCs at P12 or later, as confirmed by analysis of K+ current activation time constants and immunohistochemistry (data not shown) (Sendin et al., 2007), similar to our findings for IHCs of hypothyroid rats.
Exocytosis in IHCs of hypothyroid rats and athyroid Pax8−/− mice
We demonstrated that IHCs of both animal models of TH deficiency used in this study, hypothyroid rats and Pax8−/− mice, show ongoing spiking activity beyond the onset of hearing of their respective controls, when Ca2+ APs normally have terminated. Spontaneous activity of IHCs will impair hearing as the transducer current will promote Ca2+ APs or lead to an arrest of the IHC in a depolarized state rather than producing graded receptor potentials. We wanted to test whether hypothyroid or athyroid IHCs were able to perform exocytosis at all and, thus, could transmit the aberrant Ca2+ AP signaling after P12. Exocytosis (i.e., fusion of synaptic vesicles with the cell membrane) will increase the membrane capacitance. Therefore, capacitance changes were measured in response to voltage steps of varying durations (5–100 ms) from a holding potential of −80 mV to the potential of maximal Ca2+ influx (0 mV) for two age ranges in hypothyroid and control rats, at P9 ± 2 d and at P19 ± 2 d (Fig. 8). Figure 8, A and B, shows current traces in response to depolarizations for 100 ms (top) and the corresponding capacitance changes for P9 and P19 IHCs (bottom). IHCs from hypothyroid and control rats aged P9 showed Ca2+ inward currents and ΔCm values of comparable sizes. For the control IHC, a trace recorded under brief superfusion with a nominally Ca2+-free solution is added, showing very much reduced ICa and no increase in capacitance (Fig. 8A, light gray traces). Prolonged superfusion of the specimen with Ca2+-free solution caused loss of seals and was avoided. In the mature control IHC (P19), depolarization caused a rather small ICa and smaller ΔCm compared with the very large ICa and larger ΔCm of hypothyroid IHCs (Fig. 8B). ICa inactivation was quite variable and was not different between age-matched hypothyroid and control IHCs (data not shown). We also observed increases of Cm elicited by the high-frequency, 10 mV oscillatory voltage pattern extracted from Fig. 1E in hypothyroid IHCs aged 15–18 d at physiological Ca2+ concentration (1.3 mm) (data not shown), suggesting that their prolonged spiking activity can be transmitted to the primary auditory neurons.
Figure 8, C and D, summarizes ΔCm as a function of duration of the depolarizing stimulus for control and hypothyroid IHCs at P9 (Fig. 8C) and P19 (Fig. 8D). In general, ΔCm showed a monotonic increase with the stimulus length for all four IHC groups. At the age of P9, there was no difference in ΔCm between control and hypothyroid IHCs independent of the stimulus duration (Fig. 8C). However, a reduction of ΔCm was seen in the control IHCs at P19 for depolarizations >20 ms (Fig. 8D). In contrast, ΔCm was markedly elevated in hypothyroid IHCs at P19 compared with P9 (both groups) and to P19 control IHCs (Fig. 8D). To analyze the amount of Ca2+ that triggered ΔCm in the different IHC groups, the net Ca2+ charge was calculated by integration of the Ca2+ current traces for each stimulus duration (Fig. 8E,F). The net Ca2+ influx increased nearly linearly with stimulus duration in control and hypothyroid IHCs of the P9 (Fig. 8E) and the P19 group (Fig. 8F). For the same stimulus duration, P9 control and hypothyroid IHCs (Fig. 8E) and P19 hypothyroid IHCs (Fig. 8E) showed a similar Ca2+ influx, contrasting the markedly reduced Ca2+ influx for stimuli >10 ms in P19 control IHCs (Fig. 8E). This confirms the lack of maturation-dependent reduction of the Ca2+ current (Beutner and Moser, 2001; Johnson et al., 2005) (Fig. 3D).
IHCs normally show an increase in the efficiency of exocytosis during the third postnatal week (i.e., less Ca2+ is needed to trigger the same ΔCm response), which is attributed to the maturation of presynaptic components and a better colocalization of Ca2+ channels and ribbon synapses (Beutner and Moser, 2001; Brandt et al., 2005; Johnson et al., 2005; Sendin et al., 2007). Because ΔCm and ICa were elevated in hypothyroid compared with control IHCs at P19, we analyzed the efficiency of exocytosis as membrane capacitance change per net Ca2+ influx (Fig. 8G,H, fF/pC). ANOVA of exocytosis efficiency explained 77% (adjusted R2) of total variance among the 165 observations on 30 cells. The residual CV was 51% of the respective mean or 37% of random variance. Exocytosis efficiency was a different function of stimulus duration for different combinations of age and genotype (three-way interaction, p = 0.0066). Note that the P19 hypothyroid IHCs were similar to both P9 groups (Fig. 8G,H), suggesting that maturation of presynaptic components or their proper localization was impaired.
The ability to perform exocytosis was tested also in Pax8−/− mice. Figure 9A shows current traces in response to depolarizations for 100 ms (top) and the corresponding capacitance changes (bottom) for a Pax8 control and a Pax8−/− IHC after P12. The Pax8−/− IHC had a threefold elevated Ca2+ current but displayed approximately the same amount of ΔCm compared with the Pax8 control. Averaged capacitance increases as a function of stimulus duration did not reveal differences between Pax8−/− and Pax8 control IHCs (Fig. 9B). The Ca2+ influx that triggered these capacitance changes was larger in Pax8−/− compared with Pax8 control IHCs (Fig. 9C). Regarding exocytosis efficiency, ANOVA explained adjusted R2 = 48% of total variance with a residual CV of 51 or 55% of random variation. The function of exocytosis efficiency on stimulus duration was similar in Pax8−/− and control mice (p = 0.82), but efficiency of Pax8−/− was significantly smaller (59%; p = 0.0003) than that of controls (CI, 37 to 73%). Again, this result is in favor of an impaired maturation of presynaptic components or their correct localization under conditions of TH deficiency.
Differences in otoferlin expression in IHCs of hypothyroid rats and athyroid Pax8−/− mice
We have shown that hypothyroid IHCs of the rat and athyroid IHCs of Pax8−/− mice are able to perform exocytosis in the first three postnatal weeks. A marked difference to control IHCs was, however, the reduced exocytosis efficiency in the third postnatal week, similar to that of neonatal, prehearing IHCs. We therefore analyzed the expression of an IHC protein, otoferlin, which is assumed to play a key role in exocytosis of IHCs (Roux et al., 2006) in hypothyroid rats and Pax8−/− mice using in situ hybridization and immunocytochemistry. IHCs of Pax8 control and Pax8−/− mice showed otoferlin mRNA in early postnatal days (P5) (data not shown) and at P12 (Fig. 10A). Similarly, otoferlin protein was detected in IHCs of both Pax8 control and Pax8−/− mice in the first postnatal week (data not shown) and around the onset of hearing of Pax8 control mice (P12) (Fig. 10B). The same was true for otoferlin mRNA and protein expression in IHCs of control rats (P9 and P19) (Fig. 10C–F, left column), although surprisingly, mRNA and protein were missing in IHCs of hypothyroid rats before (P9) and after (P19) the onset of hearing (Fig. 10C–F, right column). These experiments were repeated two times (ISH) and four times (immunocytochemistry) with cochleae of hypothyroid rats of different litters, all yielding the lack of otoferlin on mRNA and protein level, respectively. This unexpected result demonstrates that exocytosis in hypothyroid IHCs of the rat worked without otoferlin and thus questions the crucial role for otoferlin in hair cell exocytosis.
The present results indicate that the expression of Ca2+ and BK channels in IHCs of mice and rats is TH-dependent. Failure of the downregulation of Cav1.3 channels and upregulation of BK channels causes persistence of Ca2+ APs after P12, the onset of hearing in control animals. Prolonged spiking activity of IHCs could lead to a severe alteration of signals to the brain required for normal functional maturation and for segregation of central neuronal projections.
Altered K+- and Ca2+-channel expression and persistence of spiking
We observed that TH deficiency affected the expression of the fast activating potassium current IK,f carried by BK channels, a feature similar to what has been described for IHCs of mice lacking TRβ (Rusch et al., 1998). However, in contrast to TRβ−/− mice in which IHC BK expression was indistinguishable from TRβ+/+ at P50 (Rusch et al., 1998), the phenotype in hypothyroid rats was more severe showing a striking mosaic expression pattern with BK-positive and BK-negative IHCs present in the same organ of Corti even up to P75. Furthermore, activation time constants of BK-positive hypothyroid rat IHCs were slower than in control IHCs, a feature not observed in TRβ mutants. Slower activation kinetics under hypothyroidism may point to an altered molecular composition of BK channels such as altered splice variants of the α subunit or altered assembly with β subunits (Marcotti et al., 2004a).
In the rat, spontaneous spiking lasts from ∼P5–P11 whereas in mouse it lasts from before or on P1–P7 (Fig. 7) (Marcotti et al., 2003b). Moreover, the resting membrane potential in neonatal rat IHCs was quite depolarized (−40 to −35 mV) compared with the neonatal mouse (around −60 mV) (Marcotti et al., 2003b). In both species, an abrupt termination of any spiking in control IHCs was observed after P12 whereas Ca2+ APs could be evoked in IHCs of hypothyroid rats and Pax8−/− mice until P19–P21. Although Kros et al. (1998) proposed the expression of BK channels around P12 as the sole factor for termination of neonatal spiking, evidence is accumulating that there are multiple reasons for AP termination. In light of the previous findings that BKα−/− mice have normal expression of Ca2+ and other K+ (delayed rectifier and KCNQ4-type) currents, show no spontaneous or evoked spiking around P20 (Oliver et al., 2006), and have normal hearing thresholds up to 8 weeks (Ruttiger et al., 2004), it must be concluded that lack of BK neither prevents the termination of spiking nor induces deafness. Instead, two other processes that counteract the generation of Ca2+ APs (Muenkner et al., 2002) and occur simultaneously to the termination of spiking in the second postnatal week should be taken into consideration as alternative factors: (1) the downregulation of the Ca2+ conductance (Fig. 3) (Beutner and Moser, 2001; Marcotti et al., 2003b) and (2) the expression of a small, very negatively activating K+ (KCNQ4) conductance that sets the resting potential in mature IHCs (Oliver et al., 2003). The retarded downregulation of the Ca2+ conductance in TH-deficient IHCs clearly favors Ca2+ APs, similar to the lack of KCNQ4 expression demonstrated in IHCs of Pax8−/− mice (Sendin et al., 2007), which, for hypothyroid rats, needs to be confirmed.
Deafness of TRβ mutant mice has been explained as a consequence of the delayed expression of BK channels and, hence, retarded functional maturation of the IHCs (Rusch et al., 1998). The phenotype of BKα−/− mice, which have normal ABR thresholds in the first 8 weeks (Ruttiger et al., 2004), requires that additional deficits in the IHCs or the auditory pathway must be taken into consideration for explaining the severe hearing deficits of TRβ mutants. Here, similar to the observations made in the present study, the failure of the decline of Cav1.3 protein leading to increased Ca2+ current amplitudes and persistence of Ca2+ APs may contribute to their hearing phenotype (see below).
Neuronal consequences of the persistence of Ca2+ APs in hypothyroidism
Hair cells mediate the critical function of transducing acoustic stimuli into neural responses. Neurons of the brainstem auditory pathway that extract timing information, such as the octopus or bushy cells of the ventral cochlear nucleus, have strongly differentiating temporal transfer characteristics (Oertel, 1997; Trussell, 1999) and may not develop properly in the third postnatal week if acoustic signal transmission is distorted by the persistent spiking activity of hypothyroid IHCs. Sound-evoked transducer currents would either increase their spiking frequency or lead to an arrest of their membrane potential in a depolarized state. Because the present findings indicate that Ca2+ APs of hypothyroid IHCs can trigger exocytosis, deafness of hypothyroid rats, Pax8−/− and TRβ−/− mice may be considered in the context of a critical period for sensory information flow necessary for development of normal hearing (Rusch et al., 1998). Indeed, at least the central effects caused by hypothyroidism may be partly caused by a distorted activity-dependent synaptogenesis in the phase of persisting Ca2+ APs that cause aberrant input. Increasing evidence indicates a critical temporal window for the activity-dependent maturation of the auditory cortex (Moore, 1985; Kral et al., 2001, 2002) similar to observations in the visual cortex (Katz and Shatz, 1996). Previously, the critical period for spectral tuning of rat auditory cortex was limited to P9–P14 or later (de Villers-Sidani et al., 2007). However, direct, Ca2+ AP-independent effects of TH on the differentiation and functional maturation of neurons in the auditory pathway cannot be excluded (Quignodon et al., 2004). Future studies with a temporally limited suppression of postnatal IHC signaling may give new insights into the role of spontaneous activity for the maturation of the auditory system.
Direct and indirect effects of TH
Here, we present a first indication of a negative regulation of the IHC L-type Ca2+ channel Cav1.3 (Platzer et al., 2000) by TH. The question arises if TH affects IHC conductances indirectly or directly through transcriptional effects. To date, only in outer hair cells a direct transcriptional control has been shown for the K+ channels KCNQ4, BK, and small-conductance Ca2+-activated K+ channel type 2 (SK2), but not for the Ca2+ channel Cav1.3 (Winter et al., 2006) (H. Winter, C. Braig, and M. Knipper, unpublished observation). The effect of TH on IHCs is different from that on outer hair cells and rather suggests the negative regulation of gene and protein expression of Cav1.3 and SK2 (C. Braig and M. Knipper, unpublished observation) by increasing TH levels (Flamant et al., 2006). Studies are ongoing to analyze these aspects of transcriptional control.
TH deficiency and exocytosis
IHC Ca2+ current amplitudes peak during the short period when most of the IHCs display spontaneous spiking in both rat and mouse (Figs. 2, 3, 7) (Beutner and Moser, 2001; Marcotti et al., 2003b). Considering that neonatal IHC presynapses are not as efficient as mature ones (Beutner and Moser, 2001; Johnson et al., 2005), the large Ca2+ amplitude of IHCs around P6 (mouse) or P9–P12 (rat) seems to compensate for the lower efficiency of neonatal exocytosis. The large capacitance increases elicited by voltage pulses of different durations in hypothyroid rat and Pax8−/− IHCs were the result of a low exocytosis efficiency that was (over)compensated by a large Ca2+ influx (Figs. 8C–H, 9B–D). P19 rat hypothyroid IHCs performed exocytosis with the same low efficiency as P9 control or P9 hypothyroid IHCs, indicating a lack of presynaptic maturation normally occurring in the third postnatal week (Beutner and Moser, 2001; Eybalin et al., 2002; Johnson et al., 2005). The reduced exocytosis efficiency (P13–P18) (Fig. 9D) (Sendin et al., 2007) and the lack of maturation of presynaptic proteins in IHCs of Pax8−/− compared with control mice (Sendin et al., 2007) is in line with this conclusion. Given that in hypothyroid rat IHCs the presynaptic machinery stays similarly immature, the reduced exocytosis efficiency would alone impair hearing thresholds at an age of 4 weeks when Ca2+ currents reach the low amplitudes of control IHCs.
Otoferlin is one of the IHC proteins thought to play an essential role in exocytosis (Roux et al., 2006). When mutated, otoferlin causes the human nonsyndromal deafness DFNB9 (Yasunaga et al., 1999, 2000). Unexpectedly, hypothyroid rat IHCs do not express otoferlin mRNA or protein while producing robust capacitance changes, thus, putting into question the indispensable role of otoferlin for IHC exocytosis (Roux et al., 2006).
A lack of otoferlin expression in hypothyroid rat IHCs indicates a regulation of its expression by TH. The striking difference in the phenotype of the athyroid Pax8−/− (otoferlin present) versus that of the hypothyroid rat pups (otoferlin lacking) may result from different TH levels in fetal development (Quignodon et al., 2004). Rat fetuses of MMI-treated mothers cannot receive maternal TH whereas Pax8−/− fetuses are supplied with a certain amount of maternal TH because their Pax8+/− mothers are euthyroid (Christ et al., 2004). Preliminary data indicating that MMI suppresses otoferlin expression in Pax8−/− and Pax8 control mouse pups (data not shown) are compatible with this view. Although studies are currently underway to clarify this issue, the data nevertheless strongly suggest that otoferlin is not required for immature-like exocytosis. This does however not exclude that otoferlin may be essential for more efficient exocytosis typical for mature IHCs.
This work was supported by University of Tüebingen Fortune Grant 1446-0-0 (J.E.) and Deutsche Forschungsgemeinschaft Grants En294/2-4,5 (J.E.) and Kni316/4-1 (M.K.). We thank Sylvia Kasperek and Karin Rohbock for excellent technical assistance, Hans-Peter Zenner for his continuous support, Karl Bauer for providing Pax8+/− mice, and Rama Panford-Walsh for discussion of this manuscript. This work is dedicated to the memory of Alfons Rüsch (1959–2002).
- Correspondence should be addressed to Dr. Jutta Engel, University of Tübingen, Institute of Physiology II and Department of Otolaryngology, Tübingen Hearing Research Centre, Gmelinstrasse 5, D-72076 Tübingen, Germany.