|Home | About | Journals | Submit | Contact Us | Français|
Degeneration of the spiral ganglion neurons (SGNs) of the auditory nerve occurs with age and in response to acoustic injury. Histopathological observations suggest that the neural degeneration often begins with an excitotoxic process affecting the afferent dendrites under the inner hair cells (IHCs), however, little is known about the sequence of cellular or molecular events mediating this excitotoxicity. Nuclear factor κB (NFκB) is a transcription factor involved in regulating inflammatory responses and apoptosis in many cell types. NFκB is also associated with intracellular calcium regulation, an important factor in neuronal excitotoxicity. Here, we provide evidence that NFκB can play a central role in the degeneration of SGNs. Mice lacking the p50 subunit of NFκB (p50−/− mice) showed an accelerated hearing loss with age that was highly associated with an exacerbated excitotoxic-like damage in afferent dendrites under IHCs and an accelerated loss of SGNs. Also, as evidenced by immunostaining intensity, calcium-buffering proteins were significantly elevated in SGNs of the p50−/− mice. Finally, the knock-out mice exhibited an increased sensitivity to low-level noise exposure. The accelerated hearing loss and neural degeneration with age in the p50−/− mice occurred in the absence of concomitant hair cell loss and decline of the endocochlear potential. These results indicate that NFκB activity plays an important role in protecting the primary auditory neurons from excitotoxic damage and age-related degeneration. A possible mechanism underlying this protection is that the NFκB activity may help to maintain calcium homeostasis in SGNs.
Spiral ganglion neurons (SGNs) are the primary carrier of auditory information from the sensory cells of the cochlea to the CNS. Degeneration of SGNs occurs with age and cochlear injuries resulting from noise, ototoxic drugs, and genetic mutations (Kiang et al., 1976; Keithley and Feldman, 1979; Leak and Hradek, 1988; White et al., 2000). The degeneration can be a primary event or occur secondarily as a result of hair cell loss. The process of SGN degeneration appears to involve apoptosis (Dodson, 1997). In vitro studies have shown that the degeneration of SGNs after loss of hair cells involves at least three mechanisms including (1) the cyclic AMP-dependent protein kinase and Ca2+/calmodulin-dependent protein kinase II and IV systems, (2) pathways involving protein kinase C activation, Ca2+ signaling, and mitogen-activated protein kinases, and (3) the c-Jun N-terminal kinase cell-death pathway (Hanson et al., 1998; Green, 2000; Zha et al., 2001; Bodmer et al., 2002; Bok et al., 2003; Hansen et al., 2003; Lallemend et al., 2003). Recent studies have demonstrated that supporting cells in the inner hair cell (IHC) regions as well as neuregulin-erbB receptor signaling and the nicotinic acetylcholine receptor subunit β2 are important for adult SGN survival (Stankovic et al., 2004; Bao et al., 2005; Sugawara et al., 2005). However, the intrinsic mechanisms mediating cell survival and cell death of auditory SGNs are still poorly defined.
The nuclear transcription factor κB (NFκB) is known for its fundamental role in regulating inflammatory responses and apoptosis in response to insults in many cell types. The predominant complex of NFκB in most mammalian cells is p50/p65. Hippocampal pyramidal neurons in mice lacking the p50 subunit of NFκB (p50−/−) exhibit increased damage after exposures to excitotoxins (Yu et al., 1999; Kassed et al., 2002). However, little is known as to the role of NFκB in hearing loss and the degeneration of the auditory nerve. Here, we report that p50−/− mice show an accelerated hearing loss and a progressive degeneration of SGNs with age.
NFκB activity may exact its influence on SGNs by upregulating gene products that regulate Ca2+ levels and modulate apoptosis (Guo et al., 1998; Camandola et al., 2005). The disturbance of Ca2+ homeostasis is a key element associated with excitotoxicity and neuronal degeneration (Mattson and Chan, 2001; Arundine and Tymianski, 2003; Mattson, 2003). Previous studies have shown that noise-induced hearing loss may be caused, in part, by excitotoxicity of SGNs. The excitotoxic damage to SGN is suggested by massive swelling of afferent dendrites under IHCs (Liberman and Mulroy, 1982; Robertson, 1983; Puel et al., 1998; Le Prell et al., 2004). To further test the hypothesis that NFκB activity plays an important role in maintaining Ca2+ homeostasis and protecting SGNs from excitotoxic injury, we examined the expression of calcium-buffering proteins and the profile of afferent dendrites under IHCs in wild-type (WT) and p50−/− mice. In addition, we examined whether NFκB activity effects cochlear susceptibility to noise-induced hearing loss by exposing the WT and p50−/− mice to wideband noise.
The gene-targeting strategy used to generate lines of mice lacking p50 has been described previously (Sha et al., 1995). B6,129PF2 (p50 WT) and B6,129P2-Nfkb1 (p50 knock-out or p50−/−) mice were bred and housed in an Association for Assessment and Accreditation of Laboratory Animal Care-certified animal facility at the Medical University of South Carolina. The homozygous breeding pairs of WT and p50−/− mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Genotyping was performed by PCR of DNA extracted from tail biopsies, as described previously (Sha et al., 1995). All mice received food and water ad libitum and were maintained on a 12 h light/dark cycle. One-, 3-, and 8-month-old mice were used in this study.
Animals were anesthetized with ketamine (100 mg/kg, i.p.) and xylazine (20 mg/kg, i.p.). Body temperature was maintained at 37°C by using a heating pad throughout the experiment. Animals were tracheotomized and placed in a head holder in a heated sound- and vibration-isolated room. The pinna and ear canal were surgically removed, and the round window was carefully exposed via a small hole in the bulla. All experimental protocols were approved by the local Institutional Animal Care and Use Committee and met National Institutes of Health guidelines for animal care and use.
The wideband noise was generated and digitally filtered in the frequency domain with TDT (Alachua, FL) modules and amplified. The sound was delivered with a Beyer DT48 drive (Beyerdynamic, Farmingdale, NY) and monitored with a probe-tube microphone (B&K 4134; Bruel and Kjaer, Norcross, GA). Each anesthetized animal was exposed to a low-level wideband noise at 70 dB sound pressure level (SPL) for 2 h with one animal exposed at a time.
The procedures for recording the distortion production otoacoustic emissions (DPOAEs), compound action potential (CAP) response and endocochlear potential (EP) in the mouse were similar to those described previously, except that the DPOAEs here were obtained from unopened bullae (Lang et al., 2002, 2005). Physiological data were obtained from right ears. The animal was anesthetized as described above and fitted to a head holder in a sound- and vibration-isolated booth.
DPOAEs were measured with an Ariel board (Ariel, Cranbury, NJ) and CUBeDISP software (Etymotic Research, Elk Grove Village, IL) using a B&K 4134 microphone, frequency equalizer, and probe tube. DPOAEs were obtained from unopened bulla after removing the pinna and underlying tissue. The intensity levels of both primaries were fixed at 70 dB SPL. The acoustic assembly, comprising a probe-tube microphone (B&K 4134) and driver (Beyer DT48) was sealed to the ear canal with closed-cell foam. Primary tones were swept from f2 = 4–20 kHz with an f2/f1 ratio of 1.2 and a resolution of 10 points per octave.
The silver-wire CAP electrode was placed inside the round-window niche and referenced to the neck musculature. The tone pips were generated in the frequency domain by TDT equipment and software. CAP thresholds were obtained visually with an oscilloscope at half-octave frequencies from 0.5–20 kHz with tone pips of 1.8 ms total duration with cos2 rise/fall times of 0.55 ms. CAP input/output (I/O) functions in each ear were obtained at 2, 4, 8, and 16 kHz at eight levels from 20 to 90 dB SPL by computer-averaging 24 epochs at each combination of level and frequency.
EPs were recorded in the basal turn of the cochlea. The EP was measured with a micropipette filled with 0.2 M KCl yielding an impedance of ~20–30 MΩ. The output of the micropipette was led to an electrometer (FD 223; World Precision Instruments, Sarasota, FL) for direct recording of the potential. The micropipette was introduced into scala media via 40–60 μm holes drilled through the otic capsule in the basal turn. EP was defined as the voltage difference between scala media and a pool of isotonic saline on the neck muscles.
After the physiological recording, the anesthetized mice were perfused via cardiac catheter with 5 ml of normal saline containing 0.1% sodium nitrite followed by 20 ml of fixative solution consisting of 10% formalin and 0.5% zinc dichromate in 0.9% saline with the pH adjusted to 5.0 just before use. The cochleas were then dissected and immersed in fixative for 45 min. The cochleas used for calcium-buffering protein immunohistochemistry and SGN quantification were decalcified with EDTA, dehydrated, embedded in paraffin, and sectioned at 6 μm thickness.
Deparaffinized and rehydrated sections were immersed in blocking solution for 20 min and then incubated overnight with a primary antibody diluted in PBS at 4°C. The primary antibodies used in this study were against the following: neurofilament 200 (1:200; Sigma, St. Louis, MO), isozyme 3 of plasma membrane Ca-ATPase (PMCA3) (1:500; Affinity BioReagents, Golden, CO), neuronal calcium sensor 1 (NCS1) (1:100; Santa Cruz, CA), calbindin D28K (1:500; Chemicon, Temecula, CA), and synaptophysin (1:100; Novo Laboratories, Newcastle, UK). Secondary antibodies were biotinylated and binding was detected with HRP techniques visualized by labeling with fluorescein (FITC)-conjugated avidin D. Nuclei were counterstained with propidium iodide (PI). Sections were examined on a Zeiss (Jena, Germany) LSM5 Pascal confocal microscope with an argon and HeNe laser. FITC and PI signals were detected by excitation with the 488 nm and 543 nm lines, respectively. Images were scanned at scales of 0.29 μm (x) × 0.29 μm ( y) and a stack size of 146.2 μm (x) × 146.2 μm ( y) with a plan-Apochromat 63×/1.4 oil differential interference contrast objective (Carl Zeiss). The captured images were processed using Zeiss LSM Image Browser version 3,2,0,70 and Adobe (San Jose, CA) Photoshop CS.
For auditory nerve fiber counts, neurofilament 200-positive nerve fibers were examined in every fifth section of the osseous spiral lamina in the basal turn of 1-, 3-, and 8-month-old WT and knock-out mice. The number of nerve fibers passing through 10 habenular openings was counted for each selected section and the average number of fibers per habenula was calculated.
For spiral ganglion cell counts, ganglion cells were examined in every fifth section of Rosenthal’s canal in the basal turns of 1-, 3-, and 8-month-old WT and knock-out mice, and at least three sections per animal were counted. The number of spiral ganglion cells over a cross-sectional area of 146.2 × 146.2 μm was counted and the average cell density was calculated.
Comparisons of immunostaining of calcium-buffering protein in SGNs were made from the images with the same setup for the confocal image acquirements: the pinhole was 106 μm and the wavelength at 488 and 543 nm was 2.0 and 80%, respectively. Care was taken to restrict the sample to the same region of Rosenthal’s canal in the basal turn. Eight-month-old WT and 8-month-old p50−/− mice were examined. Each pair of WT and knock-out sections was prepared for immunostaining together. The investigator was blind with regard to the mouse strains. The pixel intensity of FITC in individual neurons with each antibody was evaluated by using the histogram function in Adobe Photoshop CS. The pixel intensity was measured in 10 neurons randomly selected within the observed field of 146.2 × 146.2 μm and averaged. Relative intensity was quantified by calculating the ratio of the average pixel intensity of the neuron to that of the background in the same field. Three animals per group and three sections per animal were examined. From the mean pixel intensity of neurons obtained for each cochlea, we can calculate the average value for each group.
The procedures for counting hair cells on surface preparations of the basilar membrane have been described previously (Ding et al., 2001). The basilar membrane was carefully dissected from the fixed cochlea, stained with FITC-labeled phalloidin (1 μg/ml in PBS) to label filamentous actin for 20 min and with PI (1 μg/ml in PBS) to label nuclei for 10 min. Hair cells were identified by the presence of the actin-rich hair bundles, the actin belt that rings the apical surface of the cell and a healthy nucleus. Hair cell counts were made from four cochleas of 1-month-old WT mice, four cochleas of 1-month-old knock-out mice, six cochleas of 8-month-old WT mice, and six cochleas of 8-month-old knock-out mice. Separate outer hair cell (OHC) and IHC counts were made over 0.10–0.15 mm intervals of the organ of Corti beginning at the apex. All three rows of OHCs were included in OHC counts. Mean hair-cell densities were calculated by averaging cell density data over successive 20, 30, 30, and 20% segments of the cochlea, referred to as apex, midapex, midbase, and base, respectively.
The anesthetized animals were perfused via cardiac catheter first with 10 ml of normal saline containing 0.1% sodium nitrite followed by 15 ml of a mixture of 4% paraformaldehyde and 2% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4. After removing the stapes and opening the oval and round windows, 0.5 ml of fixative was perfused gently into the scala vestibuli through the oval window. The inner ears were dissected free and immersed in fixative overnight at 4°C. Decalcification was completed by immersion in 400 ml of 120 mM solution of EDTA, pH 7.0, with gentle stirring at room temperature for 2–3 d with daily changes of the EDTA solution. The tissues were postfixed with 1% osmium tetroxide, 1.5% ferrocyanide for 2 h in the dark, dehydrated, and embedded in Epon LX 112 resin. Semithin sections, ~1 μm thick, were cut and stained with toluidine blue. Ultra-thin sections were stained with uranyl acetate and lead citrate and examined by electron microscopy.
To quantify the swollen dendritic terminals under each IHC in the basal turns of WT and p50−/− mice, we used the criteria for swollen dendritic terminals that have been described previously (Hakuba et al., 2000), where vacuole-like spaces replace the afferent terminals of the inner radial nerves. Five 1-month-old WT mice, three 1-month-old p50−/− mice, four 3-month-old WT mice, and three 3-month-old p50−/− mice were examined. At least six IHC regions from each animal in basal turn were examined and the average of the swollen dendritic terminals per IHC region was calculated (Table 1).
Mean CAP thresholds were compared by using a two-way ANOVA (SPSS, Chicago, IL), with genotype as a between-subject variable and frequency as a repeated measure. Mean DPOAE amplitudes, EP values, the results of hair-cell and auditory nerve counts, and the pixel intensity of calcium-buffering proteins were compared by using one-way ANOVA with genotypes as a between-subject variable. Differences between means were considered statistically significant when p < 0.05.
The original p50 mutation was made in the 129/SvEv mouse strain and then back-crossed to C57BL/6J mice (Sha et al., 1995). Both the C57BL/6J and the 129/SvEv strains have hearing losses that begin at high frequencies and progress with age (Willott, 1986; Q. Y. Zheng et al., 1999; White et al., 2000; Lang et al., 2002). CAP thresholds in the wild-type mice show large elevations at 8 months of age. However, CAP threshold shifts in p50−/− mice are accelerated with age compared with WT mice, as shown in Figure 1, A and B. At 1 month of age, CAP thresholds in the knock-outs were elevated by 6–20 dB SPL relative to the wild-type at almost all frequencies. At 3 months of age, CAP thresholds in p50−/− mice were increased by ~20 dB SPL at low frequencies and by ~30 dB at high frequencies compared with the WT. By 8 months of age, p50−/− mice did not respond to high frequencies at 90 dB SPL, whereas the WT mice still had CAP responses at most frequencies tested.
Otoacoustic emissions allow testing of OHC function in vivo (Horner et al., 1985). DPOAEs were recorded from unopened bullae of groups of 3-month-old p50−/− and WT mice (n = 8 for each group) (Fig. 2A). Primaries were presented at 70 dB SPL. In both genotypes, the detectable emissions were found at the higher end of the frequency range (8–20 kHz). There was no significant loss of DPOAEs in knock-out mice compared with the WT mice (ANOVA, p > 0.05).
Our previous study showed that there was no significant loss of EP in C57BL/6J mice up to at least 24 months of age (Lang et al., 2002). Similarly, there was no EP reduction in WT mice at 3 and 8 months of age compared with 1-month-old WT mice (ANOVA, p > 0.05) (Fig. 2B). There was also no reduction of EP with age in p50−/− mice (ANOVA, p > 0.05). At 1, 3, and 8 months of age, mean EP values of p50−/− mice were 110 ± 4.2, 111 ± 3.1, and 107 ± 5.1 mV, respectively. Mean EP values in WT mice were 116 ± 1.1, 120 ± 3.1, and 115 ± 1.9 mV, respectively. There was no significant difference between knock-out and WT mice in any of three groups (ANOVA, p > 0.05).
Synapses under IHCs in the mammalian cochlea are of two types: afferent and efferent endings (Rasmussen, 1953; Engström, 1958). The structural features of afferent and efferent endings make them easily distinguished by transmission electron microscopy (Fig. 3A). The afferent synapses are characterized by a thick postsynaptic membrane density and contain filamentous and granular material. Efferent synaptic specializations are characterized by accumulations of many synaptic vesicles and electron-dense conical spicules on the membrane. It is generally accepted that the efferent fibers associated with IHCs often synapse with the radial afferent dendrites beneath the IHC and do not directly contact the IHC body. However, some studies have also demonstrated that both afferent and efferent fibers contact directly with IHCs in the monkey and mouse cochlea (Kimura, 1984; Sobkowicz et al., 2004).
Damage to radial afferent dendrites beneath IHCs occurs after acoustic stimulation or local application of glutamate agonists (Liberman and Mulroy, 1982; Robertson, 1983; Pujol et al., 1985, 1999; Zheng et al., 1997; Puel et al., 1998; Hakuba et al., 2000). It is believed that the afferent dendritic damage is caused by excessive release of neurotransmitter (most likely glutamate) from IHCs. A characteristic of these excitotoxic pathologies is the presence of massive swelling of afferent terminals under the IHCs. WT and p50−/− mice at ages of 1 and 3 months were processed for transmission electron microscopy. Electron microscopic features of the IHC subcellular synaptic region consist of intermingled afferent inner radial fibers and efferent spiral fibers as shown in a 3-month-old WT mouse (Fig. 3A). In a p50−/− mouse of the same age, vacuole-like spaces replace the afferent terminals of the inner radial nerves, but efferent terminals appear intact (Fig. 3B). The p50−/− mouse also shows membranous structures presumably representing residue from degenerating cell organelles. The cytoplasm in the base of the IHC consists of numerous vesicles infiltrated with normal-looking mitochondria and short profiles of cisternae (Fig. 3B). Figure 3C shows edematous-appearing extracellular spaces between an IHC and supporting cell in another 3-month-old p50−/− mouse. Pathologic lesions are also seen in a 1-month-old knock-out under an IHC (Fig. 3D). In contrast, no major pathologic changes were seen in OHCs or the stria vascularis in 1- and 3-month-old p50−/− and WT mice. Figure 3, E and F, shows the relatively normal appearance of an OHC and the stria vascularis, despite the presence of marked excitotoxic changes in the same cochlea (Fig. 3B).
Excitotoxic pathologies of afferent dendrites under the IHCs were found in cochleas of all six p50−/− mice and in only four of nine WT mice. We counted the number of swollen dendritic terminals per IHC region in the basal turns of cochleas from 1-and 3-month-old WT and p50−/− mice (Table 1). The number of swollen dendritic terminals per IHC region in p50−/− mice was significantly higher than that found in WT mice (ANOVA, p < 0.05).
There are two subpopulations of neurons in the spiral ganglion of the mammalian cochleas. The large type I fibers innervating the IHCs represent ~90–95% of the afferent auditory neurons and are myelinated. The remaining small type II neurons with peripheral processes synapse on IHCs and are unmylinated (Spoendlin, 1969). The swollen afferent terminals were seen only below the IHCs but not the OHCs, indicating that pathologies of the auditory nerve in p50−/− mice involve type I ganglion neurons, not type II neurons.
To define the morphological basis for the accelerated hearing loss with age in the p50−/− mice, especially at higher frequencies (Fig. 1), we examined basal SGN pathology by quantifying the cell density of SGNs and the number of afferent fibers in each habenular opening using light microscopy. There are two populations of afferent neurons in the mammalian cochlea. Large type I neurons with peripheral processes synapsing on IHCs represent ~90–95% of the SGNs. The remaining smaller type II cells synapse on OHCs. For auditory nerve fiber counts, neurofilament 200-positive afferent fibers were examined in sections containing the osseous spiral lamina. The monoclonal anti-neurofilament 200 (Sigma; clone N52, N 0142, phosphorylated and nonphosphorylated) labeled both type I and type II neurons and their processes in WT and p50−/− mice. Our observations are in agreement with the immunohistochemical studies of Mou et al. (1998) and Adamson et al. (2002) in mouse SGNs using the same antibody.
We quantified the cell density of SGNs in the basal cochleas of WT and p50−/− mice at 1, 3, and 8 months of age (Figs. 4A,B,E). The cell density of SGNs in p50−/− mice was relatively constant from 1 to 3 months of age, similar to those in WT mice. At 8 months of age, considerable SGN degeneration was present in both WT and p50−/− mice. The basal cochleas of WT mice at 8 months of age had lost ~28% of their SGNs. These data are in agreement with previous studies in C57BL/6J mice where the basal cochlea showed ~32% loss of SGNs at 12 months of age (Idrizbegovic et al., 2003). In the 8-month-old p50−/− mice, the basal cochlea had a 69% loss of SGNs. The density of SGNs in the knock-outs was about one-half that of the WT mice, and the difference was significant (ANOVA, p < 0.01).
All afferent nerve fibers enter the organ of Corti through the habenulae perforata in the osseous spiral lamina. Neurofilament 200-positive afferent axons were seen in tangential sections through the osseous spiral lamina (Figs. 4C,D). Quantitative analysis in the basal turn showed that the numbers of afferent axons per habenular opening in the 8-month-old p50−/− mice were significantly decreased compared with those of WT mice (Fig. 4F) (ANOVA, p < 0.05).
We also evaluated IHC and OHC survival with surface preparations of the basilar membrane in p50−/− and WT mice (Fig. 5). There was no significant reduction in number of IHC and OHC in 1-month-old p50−/− mice compared with WT mice. By 8 months of age, there was a significant reduction in the number of basal OHCs in both WT and p50−/− mice compared with 1-month-old WT mice (Fig. 5D) (ANOVA, p < 0.05). However, no significant differences in OHC numbers were found between WT and p50−/− mice at either 1 or 8 months of age. Moreover, no significant loss of IHCs was found in either WT or p50−/− mice at 8 months of age (Fig. 5E) (ANOVA, p > 0.05).
Calcium-buffering proteins play an important role in Ca2+ homeostasis. Previous studies in striatal neurons have shown that excitotoxic injury causes a rise in intracellular calcium and an increased immunoreactivity to calcium-buffering proteins (Huang et al., 1995). Numerous animal and human studies have shown the presence of a variety of calcium-buffering proteins in neural structures of the organ of Corti and SGNs including NCS1, PMCA3, calbindin D28K, and synaptophysin (Rehm et al., 1986; Liberman et al., 1990; Nadol and Burgess, 1994; Crouch and Schulte, 1995; Counter et al., 1997; Milosevic and Zecevic, 1998; Coppens et al., 2000; Sage et al., 2000; Imamura and Adams, 2003; Khalifa et al., 2003). To further characterize the cellular abnormalities in the SGNs of p50−/− mice, we compared the expression of NCS1, PMCA3, calbindin D28K, and synaptophysin in three 8-month-old WT and three p50−/− mice using immunohistochemistry (Fig. 6). The SGNs in WT and p50−/− mice were positive for all four examined antibodies. Immunostaining for NCS1 and PMCA3 appeared as a uniform cytoplasm pattern, although the latter staining was weaker. Calbindin D28K was present in both the cytoplasm and nuclei of SGNs. Punctuate immunostaining for synaptophysin was seen in the cytoplasm of SGNs. The immunostaining intensities for all four proteins in SGNs were significantly increased in p50−/− mice compared with WT mice (ANOVA, *p < 0.05; **p < 0.01).
We chose 1-month-old mice for the noise-exposure experiment because CAP thresholds at this age are relatively similar for WT and p50−/− mice (Fig. 1). After 2 h of exposure to a wideband, low-level noise at 70 dB SPL, there were no significant CAP threshold shifts in WT mice (Fig. 7). In contrast, the same exposure caused 7–15 dB SPL threshold shifts across most frequencies tested in p50−/− mice (Fig. 7). These data suggest that p50−/− mice have an increased sensitivity to noise injury.
Numerous studies have shown that NFκB activity is required for neuronal survival under both physiological and pathological conditions (Yu et al., 1999; Mattson et al., 2000; Blondeau et al., 2001; Pennypacker et al., 2001; Bhakar et al., 2002; Kassed et al., 2002; Aleyasin et al., 2004). NFκB activity is greatly increased when brain cells suffer excitotoxic and apoptotic insults. In the cochlea, previous studies have provided direct evidence of activation of NFκB in response to injury, as determined by electrophoretic shift assays. NFκB activity was elevated in chinchilla cochleas after exposure to a 96 dB octave band noise for 6 h, and in mouse cochleas after exposure to lipopolysaccharide and the ototoxic aminoglycoside kanamycin (Wu and Xie, 2002; Ramkumar et al., 2004; Jiang et al., 2005). A recent study showed that selective inhibition NFκB caused apoptosis of immature hair cells in vitro (Nagy et al., 2005). However, there is little information regarding activation of NFκB in the auditory nerve and how NFκB activity may affect cochlear function with age and trauma. Here, mice lacking the p50 subunit of NFκB (p50−/− mice) showed an accelerated hearing loss with age that was highly associated with an exacerbated excitotoxic-like damage in afferent dendrites under IHCs and an accelerated loss of SGNs.
In the cochlea, excitotoxic pathology occurs in response to a variety of injuries. Local application of ouabain and glutamate agonists, kainic acid, or α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid can cause an immediate and massive swelling of the radial afferent dendrites under IHCs (Pujol et al., 1985; Zheng et al., 1997; Puel et al., 1998; Spicer et al., 2002). Here, excitotoxic-like pathologies of afferent dendrites under IHCs were found in 100% of the examined p50−/− mice and only 44% of the WT mice. The average number of swollen afferent dendrites under IHCs in p50−/− mice was significantly higher than that of WT mice. These data strongly suggest that deficiency of the p50 subunit of NFκB increases auditory nerve susceptibility to excitotoxicity.
Although there is not a significant loss of SGNs in 1- and 3-month-old p50−/− mice, the excitotoxic-like pathologies of afferent dendrites under IHCs are the most likely morphological evidence for the cause of the elevated CAP thresholds in the knock-outs. Hearing loss has been associated with the swollen afferent dendrites under IHCs as in mice lacking the glutamate transporter GLAST after sound exposure, in chinchillas after kainic acid excitotoxicity, in guinea pigs after sound exposure, and after intra-cochlear infusion of the aminoglycoside antibiotic amikacin and the glutamate agonist AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) (Robertson, 1983; Zheng et al., 1997; Puel et al., 1998; Duan et al., 2000; Hakuba et al., 2000; Le Prell et al., 2004). Studies have also shown that functional recovery corresponds to structural recovery or reestablishment of the IHC/auditory-nerve synapse after acute excitotoxic-drug exposures (Zheng et al., 1997; Pujol and Puel 1999; X. Y. Zheng et al., 1999; Le Prell et al., 2004).
The disturbance of calcium homeostasis is a major event associated with excitotoxicity in neurons. Excessive synaptic release of glutamate can lead to the disregulation of calcium homeostasis (Arundine and Tymianski, 2003). Disturbances of calcium homeostasis induced by various insults play an important role in pathological processes, culminating in neuronal apoptosis and excitotoxicity (Mattson et al., 2000; Mattson and Chan, 2001). To further investigate the cellular mechanisms underlying the progressive SGN pathology in p50−/− mice, we evaluated changes in the expression of a set of calcium-buffering proteins with immunohistochemistry. These proteins included NCS1, PMCA3, calbindin D28K, and synaptophysin, all of which are involved in calcium-regulating processes in mammalian SGNs (Rehm et al., 1986; Liberman et al., 1990; Nadol and Burgess, 1994; Crouch and Schulte, 1995; Counter et al., 1997; Milosevic and Zecevic 1998; Coppens et al., 2000; Sage et al., 2000; Imamura and Adams, 2003; Khalifa et al., 2003). NCS1, also named as frequenin, is a highly conserved member of the elongation factor-hand calcium-binding protein family expressed mainly in neurons (Braunewell and Gundelfinger, 1999). PMCA is a plasma membrane-associated calcium pump that transports calcium ions out of the cell by using the energy stored in ATP. PMCA is essential for control of cytosolic calcium concentrations (Carafoli, 1991). Calbindin D28K is believed to function as an intracellular calcium buffer and has been localized in several neuronal populations within the CNS as well as SGNs (Celio, 1990; Slepecky and Ulfendahl, 1993). Synaptophysin is a well known marker for synaptic vesicles and is present in auditory nerve terminals in the organ of Corti. It can also act as a calcium-binding protein and is expressed in SGNs of the mammalian cochlea (Anniko et al., 1995; Rask-Andersen et al., 2000; Khalifa et al., 2003).
The increased expression of NCS1, PMCA3, calbindin D28K and synaptophysin in SGNs of knock-out mice, as demonstrated here, strongly suggests a disturbance in their calcium homeostatic mechanisms. The disturbance of calcium homeostasis may be an important cause of the rapid age-related SGN degeneration in NFκB deficient mice compared with the WT. The increased intensity of immunostaining for calcium regulatory proteins in this mutant mouse may be the result of an overall decrement of neuronal viability, or upregulating gene products that regulate Ca2+ homeostasis by the deficiency of NFκB activity. Several studies have documented that activation of NFκB is required to maintain calcium homeostasis and modulate cell death. Cultured striatal neurons from p50−/− mice exhibit perturbed calcium regulation and increased cell death after exposure to mitochondrial toxins (Yu et al., 2000). Inhibition of NFκB activity in PC12 cells by treatment with κB decoy DNA is associated with enhanced elevation of intracellular calcium levels induced by the neurotoxic amyloid β-peptide (Guo et al., 1998). Moreover, fibroblasts lacking the subunit of NFκB p65 exhibit increased inositol 1,4,5-trisphosphate (IP3) receptor-mediated calcium release from the endoplasmic reticulum (ER) and increased sensitivity to apoptosis (Camandola et al., 2005).
However, Ca2+ release from IP3-sensitive ER stores can also trigger activation of NFκB in cultured neurons (Glazner et al., 2001). A recent study showed that synaptic transmission can activate NFκB via local submembranous Ca2+ increases along with a pathway requiring Ca2+/calmodulin-dependent kinase (CaMKII) under physiological conditions (Meffert et al., 2003). Therefore, NFκB may participate in a dynamic transcription-dependent feedback mechanism to control Ca2+ homeostasis in both physiological and pathological conditions.
In noise-induced hearing loss, it is thought that excitotoxic damage to afferent neurons can contribute to acute threshold shifts (Robertson, 1983; Puel et al., 1998). Damage to afferent fibers beneath IHCs after acoustic stimulation has been reported in numerous studies (Spoendlin, 1971; Liberman and Mulroy, 1982; Robertson, 1983; Puel et al., 1996, 1998). We exposed the 1-month-old p50−/− and WT mice to 70 dB SPL wideband noise for 2 h. The low-level noise did not produce a threshold shift in WT mice, consistent with previous studies showing that a temporary threshold shift induced by octave-band noise requires an exposure intensity of at least 94 dB SPL (Wang et al., 2002; Hirose and Liberman, 2003). However, CAP thresholds were shifted ~7–15 dB SPL across most frequencies tested in the p50−/− mice with the same exposure paradigm. These results indicate that p50−/− mice are more sensitive to acoustic exposure than WT mice.
The p50 knock-outs were generated with a C57BL/6 background, a strain that is a well known model for age-related hearing loss. The decline of hearing sensitivity in the C57BL/6 begins at high frequencies and is caused mainly by the loss of hair cells, some loss of SGNs, and a loss of type IV fibrocytes (Spongr et al., 1997; White et al., 2000; Hequembourg and Liberman, 2001; Ohlemiller and Gagnon, 2004). Our results do demonstrate a significant reduction of basal SGNs and OHCs in both WT and p50−/− mice at 8 months of age. However, the degeneration of SGNs and afferent-nerve fibers was significantly greater in the p50−/− mice and occurred without an accompanying increased loss of IHCs or OHCs in the knock-out mice compared with WT mice. Thus, the protective effect of NFκB activation in cochlea seems to be specific to the primary auditory nerve under normal physiological conditions. In the CNS, there is a low-level constitutive activity of NFκB in neurons (Kaltschmidt et al., 1994). It is possible that the activation of NFκB is also involved in the survival of the central auditory nerve system. Additional studies need to be performed using physiological and morphological methods in the central auditory nerve system of p50−/− mice.
The purpose for performing the experiments herein was to gain additional insights into the mechanisms regulating auditory nerve survival and degeneration. The p50−/− knock-out mouse provides a unique model for studying mechanisms of SGN degeneration occurring with noise exposure, ototoxic drugs, or age. A more complete understanding of the functional role of NFκB in regulating the survival and death of auditory SGNs may provide new therapeutic strategies to prevent or ameliorate hearing loss caused by various injuries and to preserve the residual hearing in cochlear implant users.
This work was supported by National Institutes of Health Grants R03DC-07506 (H.L.), R01AG-14748 (R.A.S.), R01DC-00713 (B.A.S.), and R01CA-78688 (D.Z.), MUSC Institution Research Funds (H.L.), and Grant C06 RR014516 from the Extramural Research Facilities Program of the National Center for Research Resources. We thank James Nicholson and Liya Liu for their technical assistance. We also thank three anonymous referees for suggested improvements.