Hidden hearing loss is associated with loss of ribbon synapses of cochlea inner hair cells

Abstract The present study aimed to observe the changes in the cochlea ribbon synapses after repeated exposure to moderate-to-high intensity noise. Guinea pigs received 95 dB SPL white noise exposure 4 h a day for consecutive 7 days (we regarded it a medium-term and moderate-intensity noise, or MTMI noise). Animals were divided into four groups: Control, 1DPN (1-day post noise), 1WPN (1-week post noise), and 1MPN (1-month post noise). Auditory function analysis by auditory brainstem response (ABR) and compound action potential (CAP) recordings, as well as ribbon synapse morphological analyses by immunohistochemistry (Ctbp2 and PSD95 staining) were performed 1 day, 1 week, and 1 month after noise exposure. After MTMI noise exposure, the amplitudes of ABR I and III waves were suppressed. The CAP threshold was elevated, and CAP amplitude was reduced in the 1DPN group. No apparent changes in hair cell shape, arrangement, or number were observed, but the number of ribbon synapse was reduced. The 1WPN and 1MPN groups showed that part of ABR and CAP changes recovered, as well as the synapse number. The defects in cochlea auditory function and synapse changes were observed mainly in the high-frequency region. Together, repeated exposure in MTMI noise can cause hidden hearing loss (HHL), which is partially reversible after leaving the noise environment; and MTMI noise-induced HHL is associated with inner hair cell ribbon synapses.

90 dB SPL, and weakened in 5-dB steps until the ABR response disappeared. The threshold 133 was determined as the lowest level at which a repeatable wave III response could be observed. 134 The representative ABR trace was demonstrated in supplemental figure 1. 135 The detailed acoustic stimulus designed for recording ABR, ABR with masking, CAP and 136 amplitude modulation CAP were as follow. For ABRs stimuli: tone bursts of 10-ms duration 137 with cos2 gating, 0.5-ms rise/fall time. The signal was pre-amplified by 20X, proceeded with 138 a bandpass (100-3000Hz) filter, and superimposed 1000 times. It started with 90 dB SPL and 139 decreased by 5 dB steps. The lowest sound intensity of the wave III was documented as the 140 ABR threshold. Next a combination of 90-dB SPL clicks with uniformly increasing 141 broadband noise (as masking noise, started at 30 dB and increased at 10-dB steps to 80 dB 142 SPL) was given. The amplitude and latency of wave-I and wave-III were documented. The 143 amplitude was defined as the difference between the current peak value and the immediately 144 following valley value. 145 The CAP and amplitude modulation CAP (AM-CAP) were recorded to evaluate auditory 146 temporal processing ability. CAPs were recorded in response to clicks or tone bursts across 147 the same range of SPLs as used in ABR recording. Compared to ABR, the tone bursts were 148 designed at different frequencies (1,2,4,10,20,24 and 32 kHz). Both CAP thresholds and 149 peak amplitudes were measured. The peak-to-peak value was regarded as the CAP peak 150 amplitude. For AM signals, the carrier frequency was designed at the frequencies 2, 10 and 20 151 kHz), and the modulation frequencies varied between 93 and 996 Hz, while the carrier 152 strength was fixed at 80 dB SPL, and the modulation depths varied from 10% to 100% (10%, 153 20%, 40%, 60%, 80% and 100%). Responses were recorded continuously and averaged by 154 Downloaded from http://portlandpress.com/bioscirep/article-pdf/doi/10.1042/BSR20201637/905977/bsr-2020-1637.pdf by guest on 20 March 2021 Bioscience Reports. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BSR20201637 8 looping into a 700 ms window. The response amplitudes at modulation frequencies were 155 calculated and expressed in dB using Fast Fourier Transform (FFT) with in the TDT BioSig 156 software (8192 points). Envelope following responses (EFRs) were considered positive if a 157 clear peak in FFT was identified at 93 or 996 Hz with amplitude 3 dB higher than the troughs 158 above and below the peak. 159 160

Morphology analysis 161
After sacrifice, ears from each animal were used for immunostaining against ribbons and PSD. 162 To observe the ribbon synapse structure, the cochlea was fixed with 4% paraformaldehyde for 163 24 h, the bone wall, the spiral ligament, the vestibular membrane, and the tectorial membrane 164 were removed under the dissecting microscope. Afterwards, the cochlea was permeabilized 165 with 1% Triton X-100 in PBS for 45 min, incubated for 30 min in 10% goat serum in PBS 166 and then incubated in the mixture of two primary antibodies (rabbit anti-CtBP2 167 C-terminal-binding protein 2, Abcam, cat. #ab128871, 1:200, and mouse anti-PSD95, Abcam, 168 cat. # ab2723, 1:200 in the primary antibody diluent) overnight. After 3 times of wash (10 min 169 each), each cochlea was inculcated with secondary antibodies (goat anti rabbit IgG, Abcam, 170 cat. # ab175471, 1:200, and goat anti mouse IgG2a, Invitrogen, 1:200, A21131) for 1 h at 171 room temperature. Samples were washed three times with PBS, and then immersed in 10% 172 EDTA. The basilar membrane was dissected into 4 pieces according to the corresponding 173 cochlea turn. Next, the basilar membrane was mounted on microscope slides. Confocal 174 z-stacks from each ear were obtained using a high-resolution oil-immersion objective (×100) 175 on a confocal laser-scanning microscope (Zeiss LSM 780). Image stacks were then ported to 176 Downloaded from http://portlandpress.com/bioscirep/article-pdf/doi/10.1042/BSR20201637/905977/bsr-2020-1637.pdf by guest on 20 March 2021 Bioscience Reports. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BSR20201637 the image-processing software (Zen 2010 blue and ImageJ), with 0.1 μm layer distance, and 177 the confocal z-stacks were reimaged for 3-D reconstructions. The locations in terms of lengths 178 or distances from the apex were mapped [13][14][15]. A total of 17 regions were analyzed across 179 the basilar membrane, and in each cochlear region, the synapse number was calculated. The 180 final puncta for elements staining were obtained by calculating the average number of pre-or 181 postsynaptic elements for each IHC. 182 183

Statistical analysis 184
Data were analyzed using IBM SPSS 22.0 and GraphPad Prism 6. The quantitative data were 185 expressed as mean ± standard error, and the differences in ABR or CAP responses among 186 groups were analyzed using Tukey's multiple comparisons in the two-way ANOVA method. 187 The number of ribbon synapses among groups was analyzed using Tukey multiple 188 comparison test in one-way ANOVA. A p value less than 0.05 was considered statistically 189

MTMI noise impaired auditory function one-day to one-month post exposure 193
First, ABR examination was conducted in masking noise environment to evaluate the impact 194 of MTMI noise. For ABR wave-I and III latencies, no significant differences were found 195 among groups ( Fig. 1 A and B). The amplitudes of ABR wave-I and wave-III decreased one 196 day after noise exposure and gradually restored during one week/month of recovery (wave-I: 197 masking factor: F=10.20, P＜0.01, group factor: F=306.5, P=0.0001; wave-III: masking 198 Downloaded from http://portlandpress.com/bioscirep/article-pdf/doi/10.1042/BSR20201637/905977/bsr-2020-1637.pdf by guest on 20 March 2021 Bioscience Reports. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BSR20201637 factor: F=19.11, P＜0.01; group factor: F=391.4, P＜0.01) ( Fig. 1 C&D). Collectively, MTMI 199 noise significantly impairs auditory function assessed by ABR one day later, and the residual 200 effect may last for at least one month ( Fig. 1 A&B). 201 The impact of noise exposure on cochlear function was evaluated by the CAP responses to 202 two types of stimuli: click and tone burst. For the click triggered responses, the CAP 203 threshold of the 1DPN group was significantly elevated (p<0.05 vs control), and no 204 statistically significant differences were found in the other two noise exposed groups (Fig.  205 2A). Still, the mean CAP thresholds were slightly higher than control. Similarly, the CAP 206 peak amplitudes were decreased in three noise suffered groups compared to the control group 207 (click sound intensity factor: F=450.6, P＜0.0001; group factor: F=116.1, P＜0.0001) (Fig  208   2B). This finding strongly implied that the auditory function had not acquired a full repair in 209 the following month. On the other side, curves of CAP threshold and amplitude triggered by 210 the tone burst were drawn ( Fig. 2 C & D). The CAP threshold increased with tone burst 211 frequencies, and three noise exposed groups showed significantly shift-up curves compared to 212 the control curve (p<0.05 each noise exposed group vs control) (Fig. 2 C). Similar to the data 213 of click stimuli, CAP amplitudes corresponding to high frequencies were suppressed ( Fig. 2  214 D). However, in the low-frequency range, the CAP peak amplitudes were enhanced. This 215 interesting finding may be due to a sensitized disorder towards low-frequency signals after 216 noise exposure. Next, amplitude modulation CAP responses were recorded as a support. For 217 all animals, the peak amplitude increased with modulation frequencies, and three noise 218 exposed groups had significantly decreased amplitudes in both responses to different 219 frequency modulation depths (Fig. 2 E) and in 60 dB SPL white noise exposure backgrounds 220 Downloaded from http://portlandpress.com/bioscirep/article-pdf/doi/10.1042/BSR20201637/905977/bsr-2020-1637.pdf by guest on 20 March 2021 Bioscience Reports. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BSR20201637 ( Fig. 2 F) (p<0.05 each noise exposed group vs control). No significant differences were 221 found among three noise-exposed groups. These findings strongly highlight an incomplete 222 repair of AM CAP peak amplitude even one month after MTMI noise exposure. 223 224

Noise induced cochlear synaptopathy 225
To reveal the impairment of cochlear synapses induced by noise, immunofluorescence 226 staining was used to observe the dynamic changes in cochlear synapses at different time 227 points after the noise exposure. The presynaptic ribbons were immune-stained with an 228 anti-CtBP2 antibody, and the postsynaptic terminals were stained with an antibody to PSD-95, 229 the specific PDZ-domain protein (Fig. 3A). No apparent changes in hair cell shape, cell 230 arrangement or cell number were observed. At the synaptic level, both presynaptic ribbons 231 and postsynaptic terminals were reduced after noise exposure (Fig. 3B), and this change was 232 most significant in the one-day group (p<0.01). The amount of CtBP2 and PSD-95 showed a 233 partial recovery in the following one week or one month (but still lower than the control 234 group). Moreover, as reported previously, the co-stained (paired) synapses were markedly 235 reduced at 1DPN (F= 56.91, P < 0.01; paired synapse: F=46.34, P < 0.01; ribbon: F=4.075, 236 P=0.011; PSD: F=31.24, P < 0.01). This result confirmed a damage on cochlear synapse from 237 noise. In particular for the 1DPN group, an overall loss of 48.9% paired synapses was found, 238 with 38.8% reduction in presynaptic ribbons and 17.1% reduction in postsynaptic terminals. It 239 is interesting that no difference was found in the PSD amount among three noise exposed 240 groups (p > 0.05), which implied that the postsynaptic injury induced by noise had extremely 241 limited recovery in the following one month. The impairment of cochlear synapse showed a 242 Downloaded from http://portlandpress.com/bioscirep/article-pdf/doi/10.1042/BSR20201637/905977/bsr-2020-1637.pdf by guest on 20 March 2021 Bioscience Reports. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BSR20201637 regional feature, as Fig. 3 C shown. The decrease was mainly in the region of 60% to 80% 243 distance from apex, but not the overall length, which coincided with the changed range of 244 corresponding threshold shift (2000-20000 Hz, Fig. 3 D). Although there is a mismatch 245 between CAP and ABR thresholds, this may be due to different sensitivity of two tests 246 (especially towards the 1DPN group). Also, the scattered distribution of the ribbons was 247 changed after noise exposure. Some ribbons were located on the side of the modiolus, but not 248 close to the outer hair cells, which was consistent with a recent work by Boero et al. The major findings of this study are: (1) repeated exposed in MTMI noise may cause hearing 257 loss within one month, which is reversible after leaving the noise environment; (2) MTMI 258 noise exposure may damage IHC ribbon synapses. Before this study, there are very few works 259 focusing on the effects of MTMI noise on cochlear synaptopathy and function defects. The 260 significance of this study is that we provided further evidence supporting cochlear 261 synaptopathy may be involved in HHL. 262 It has been widely accepted that noise may induced damage on IHCs and SGNs in the cochlea 263 [17], including synaptopathy, which is one of the main reasons of HHL. Increasing evidence 264 In animal studies showing the temporary elevation in hearing thresholds caused by noise 274 exposure, there may be reduced neural responses in an adaptive way, and theoretically the 275 HHL-like state quickly recovers. It has been known that after a single, brief noise exposure, 276 the damaged and the totally destroyed synapses can be partially repaired, but the repaired 277 synapses are functionally abnormal [17]. The mechanism underlying noise induced threshold 278 elevation and hearing recovery is closely related to IHC functions, especially cochlear 279 synapse plasticity. Kujawa and Liberman showed that 2 h of 100 dB SPL noise (8-16 kHz) 280 exposure killed half of IHC/auditory nerve synapses in high-frequency regions permanently 281 but sensitivity to quiet sounds was easy to recover [6]. Although their research focused on the 282 acute noise stress, the conclusion was consistent with our findings. Loss or recovery of both 283 ribbon and synapse plays a crucial role in the final auditory outcomes, but we believe that 284 cochlear synaptopathy is a more direct reason of HHL, under noise exposure or during ageing 285 process [19]. Liberman et al [32] suggested that age-related decline in spiral ganglion cells 286 may contribute to function loss of hearing-in-noise, especially the ganglion cell survival in the 287 high-frequency segments of cochlear location (which was similar to our result Fig. 3D) [19]. 288 Based on the ABR and CAP results in our work, repeated MTMI noise harm responses to the 289 high-frequency stimuli but not low-frequency, which is similar to dysacusis in the elderly. cochlear synapse loss may be a more direct factor that impacts hidden hearing loss and its 297 recovery. In human studies, supportive evidence also exists. Liberman recently showed that in 298 the aging human cochlea there is primary neural degeneration which might explain a loss of 299 speech discrimination ability in the elderly [22]. Moreover, another explain of this finding is 300 the sound-evoked feedback reduction of cochlear amplification [23]. When associated with 301 sustained discharge rate, the threshold shift might be protective during noise exposure. 302 Moreover, we noticed that CAP amplitudes corresponding to high frequencies were 303 suppressed but in the low-frequency range the CAP peak amplitudes were enhanced. This 304 may be due to a sensitized disorder towards low-frequency signals. Although the detailed 305 mechanism is not clear, it is possibly a homeostasis as a protection mechanism. Another 306 explanation is that disordered CAP response is an overall left shift of the CAP peak curve, 307 that the max amplitude in the response to medium frequency slightly turned to the left range 308 Downloaded from http://portlandpress.com/bioscirep/article-pdf/doi/10.1042/BSR20201637/905977/bsr-2020-1637.pdf by guest on 20 March 2021 (low frequency). However, the deep mechanism of this finding requires further study. 309 We observed that the MTMI noise exposure inhibited the responses to high frequencies and 310 caused synaptopathy mainly in the region of 60-80% from the apex. This region is related to 311 the input of high-frequency signals [24,25]. The region near apex had similar CtBP2 and 312 PSD-95 distribution and expression levels (not shown). Collectively, auditory function tests 313 and morphology analysis can mutually support, which both highlighted that MTMI noise 314 induced defects toward high-frequency signals are possibly due to damage in the synapses 315 located at the 60-80% cochlear region from the apex. 316 It is worth to mention that decreased output from the cochlea may trigger a compensatory 317 neural gain in the auditory brainstem (reported in ABR Wave III) [26,27]. Years ago, Hickox 318 and Liberman claimed that noise-exposed mice with cochlear neuropathy show 319 hypersensitivity to sound, which suggests a link between AN damage and hyperacusis [28]. 320 Similarly, we also observed an enhanced CAP amplitude under low frequencies (lower than 4 321 kHz) noise exposure (Fig. 2D). Another explain is that, after HHL, it is easy to cause tinnitus, 322 and the stochastic resonance plays a role during the period of hearing threshold recovery [29]. 323 Together, MTMI noise may lead to complicated disorders in cochlear functions, majorly 324 exhibiting impairment in high frequencies and enhancement in low frequencies. with the synaptic damage/loss (but without permanent threshold shifts), the current functional 335 changes are mainly supported by ABR-wave and CAP-amplitude. 336 In summary, the present study revealed that loss/recovery of the ribbon synapses in the 337 cochlea is associated with ABR-wave and CAP-amplitude changes in NIHHL developed by 338 MTMI noise. These alterations might be associated with cochlear synaptopathy. The datasets generated and/or analyzed during the current study are available from the corresp 351 onding author on reasonable request. 352