|Home | About | Journals | Submit | Contact Us | Français|
Author contributions: M.E.R. designed research; C.C. performed research; C.C., F.M.A., and M.E.R. analyzed data; C.C., F.M.A., and M.E.R. wrote the paper.
Sound deprivation by conductive hearing loss increases hearing thresholds, but little is known about the response of the auditory brainstem during and after conductive hearing loss. Here, we show in young adult rats that 10 d of monaural conductive hearing loss (i.e., earplugging) leads to hearing deficits that persist after sound levels are restored. Hearing thresholds in response to clicks and frequencies higher than 8 kHz remain increased after a 10 d recovery period. Neural output from the cochlear nucleus measured at 10 dB above threshold is reduced and followed by an overcompensation at the level of the lateral lemniscus. We assessed whether structural and molecular substrates at auditory nerve (endbulb of Held) synapses in the cochlear nucleus could explain these long-lasting changes in hearing processing. During earplugging, vGluT1 expression in the presynaptic terminal decreased and synaptic vesicles were smaller. Together, there was an increase in postsynaptic density (PSD) thickness and an upregulation of GluA3 AMPA receptor subunits on bushy cells. After earplug removal and a 10 d recovery period, the density of synaptic vesicles increased, vesicles were also larger, and the PSD of endbulb synapses was larger and thicker. The upregulation of the GluA3 AMPAR subunit observed during earplugging was maintained after the recovery period. This suggests that GluA3 plays a role in plasticity in the cochlear nucleus. Our study demonstrates that sound deprivation has long-lasting alterations on structural and molecular presynaptic and postsynaptic components at the level of the first auditory nerve synapse in the auditory brainstem.
SIGNIFICANCE STATEMENT Despite being the second most prevalent form of hearing loss, conductive hearing loss and its effects on central synapses have received relatively little attention. Here, we show that 10 d of monaural conductive hearing loss leads to an increase in hearing thresholds, to an increased central gain upstream of the cochlear nucleus at the level of the lateral lemniscus, and to long-lasting presynaptic and postsynaptic structural and molecular effects at the endbulb of the Held synapse. Knowledge of the structural and molecular changes associated with decreased sensory experience, along with their potential reversibility, is important for the treatment of hearing deficits, such as hyperacusis and chronic otitis media with effusion, which is prevalent in young children with language acquisition or educational disabilities.
Sensory experience regulates the structure and function of synapses in the central visual, somatosensory, and auditory systems. Previous studies have revealed changes in synaptic morphology and transmission after monocular deprivation or eyelid suture (Maffei et al., 2004; Hofer et al., 2009; Coleman et al., 2010), whisker removal or mechanical whisker manipulation (Allen et al., 2003; Bender et al., 2006; Li et al., 2009), sensorineural hearing loss in congenitally deaf animal models (Redd et al., 2000; Oleskevich and Walmsley, 2002; Lee et al., 2003; Ryugo et al., 2005), cochlear ablations (Gulley et al., 1977; Suneja et al., 1998; Asako et al., 2005; Rubio, 2006), or age-related hearing loss (Wang and Manis, 2005, 2006). More recently, it has become clear that other forms of hearing impairment associated with conductive hearing loss can also alter synaptic transmission in central auditory regions (Xu et al., 2007, 2010). However, despite being the second most prevalent form of hearing loss, conductive hearing loss and its effects on central synapses have received relatively little attention.
Several studies have shown that the effects of conductive hearing loss are age dependent (McGinn and Henry, 1975; Popescu and Polley, 2010; Takesian et al., 2012; Caras and Sanes, 2015). Evidence also exists that in the adult, conductive hearing loss can affect later auditory brainstem response (ABR) waves (Ferguson et al., 1998) and cochlear nucleus activity (Potashner et al., 1997; Tucci et al., 1999, 2001, 2002; Cook et al., 2002; Sumner et al., 2005), as well as upregulate AMPA glutamate receptor (AMPAR) subunits (Whiting et al., 2009; Wang et al., 2011). However, it is unknown whether the effects of conductive hearing loss are reversible. A small number of psychoacoustic studies have suggested plasticity in the adult human auditory system after transient conductive hearing loss, suggesting a gain control mechanism operating at the level of the auditory brainstem (Formby et al., 2003, 2007; Munro and Blount, 2009). These studies led us to investigate whether similar plasticity mechanisms occur in the mature auditory brainstem of the rat after transient conductive hearing loss. We investigated the structural and molecular changes at the auditory nerve synapse on bushy cells of the cochlear nucleus after conductive hearing loss by earplugging. We also monitored functional changes within the auditory brainstem with ABRs.
The auditory nerve ending in the anteroventral cochlear nucleus is a large glutamatergic calyx-type terminal (endbulb of Held) that surrounds the soma of bushy cells (Fig. 1A; Ryugo and Fekete, 1982; Rouiller et al., 1986). Endbulb synapses contain AMPAR subunits with rapid kinetics (GluA3 and GluA4; Wang et al., 1998; Gardner et al., 1999, 2001; Whiting et al., 2009). The slower GluA1 subunit is absent at mature endbulb synapses, and the GluA2 subunit, whose absence is associated with fast synapses, is expressed at very low levels (Wang et al., 1998; Gardner et al., 1999, 2001). Bushy cells are a major cell type in the anteroventral cochlear nucleus, and they enhance the temporal fine structure information that is present in the phase-locked pattern of auditory nerve spikes (Joris et al., 1994). The endbulb of Held is a critical gateway for auditory activity entering the brain; thus, structural and molecular alterations at the synapse could have widely reaching consequences for auditory processing.
In the current study, we show that 10 d of monaural conductive hearing loss by earplugging at postnatal day 30 (P30) leads to increased hearing thresholds as well as to long-lasting presynaptic and postsynaptic structural and molecular effects at the endbulb of the Held synapse. Noteworthy is that P30 past maturation of connectivity and synaptic elements in the cochlear nucleus. These results are unexpected considering the prevalent view that mature endbulb synapses are simple relay points, built for high-fidelity and high-frequency synaptic transmission, and a minimal requirement for synaptic plasticity. Thus, knowledge of the structural and molecular changes associated with decreased sensory experience, along with their potential reversibility, is important for the treatment of hearing deficits, such as hyperacusis and chronic otitis media with effusion.
Twenty P30 male Sprague Dawley rats (Charles River Laboratories) were used in this study. Twelve rats were assigned to the earplugged group, and eight were assigned to the sham group. All the rats were tested for hearing thresholds at P30 before monaural earplugging or sham procedures (see below; Fig. 1B). Rat home cages were positioned in racks in the animal facility room such that both sham and experimental animals were exposed to the same ambient sound. All procedures were approved by the University of Pittsburgh, Division of Laboratory Animal Resources and complied with the NIH requirements for the humane care and use of laboratory rats.
For ABR testing, we used the protocol described by Wang et al. (2011). Sound levels for clicks and pure tones were calibrated using a one-quarter-inch microphone (model PCB-378C01; PCB Piezotronics). The calibration microphone was interfaced with our close-field speaker [MF1-M; Tucker-Davis Technologies (TDT)] using a 2 cm plastic tube, which was placed in a sealed (sound-isolated) syringe tube (the coupler). All connections between the speaker, tube, coupler, and microphone were airtight to avoid inaccurate calibration. Calibration was performed using the TDT program BioSigRZ. All calibrations and ABRs were conducted in a sound-attenuating cubicle (ENV-022SV; Med Associates). All animals were tested under isoflurane anesthesia for hearing thresholds by recording ABRs to clicks and pure tones in a sound-attenuating cubicle. A heating pad was used to maintain a constant body temperature of 37°C. Three register electrodes (an active, a reference, and a ground) were placed subcutaneously on the midline of the scalp (+), on the cheek toward the ear (−), and on the hind leg (ground). The electrodes were connected to a four-channel low-impedance head stage (RA4 LI; TDT). Thresholds were measured for the ipsilateral ear before, during the 10 d of earplugging, and 10 d after earplug removal. ABR testing was conducted for the ipsilateral (plugged) side using an evoked potential workstation (RZ6-A-P1; TDT) and a close-field speaker (MF1-M; TDT). Sound stimuli signals were generated and analyzed using BioSigRZ software (TDT). Click stimuli lasted 0.1 and 5 ms for tone bursts at 4, 16, 24, and 32 kHz, with a linear rise–fall time of 2.5 ms and no plateau. Evoked potentials were averaged over 1024 sweeps, and waveforms were filtered using a bandpass filter of 300–3000 Hz. ABR waveforms were recorded at decreasing intervals of 5 dB beginning with 80 dB for clicks and 100 dB for pure tones in the sham and earplugged rats. Auditory thresholds for clicks and pure tones were determined by blind visual inspection at the lowest intensity at which waves I and II were detectable above noise within a specified time window. The latency and amplitude (peak to peak) of click ABR waves were analyzed at 10 dB SPL above threshold. To determine changes in gain in response to transient earplugging, the average ratios of ABR wave II to I in the earplugged group were normalized by the average ratios from the age-matched sham group, within each age group (P30, P40, and P50; Fig. 1G). Similar calculations were performed to calculate the relative gain between waves IV and I at P30, P40, and P50 (Fig. 1G).
After the initial ABR testing at P30, the rats in the earplugged group (n = 12) were unilaterally earplugged for 10 d until they reached P40, and at that age, the earplug was removed in seven animals, which were then kept alive for another 10 d (P50; Fig. 1B). In the animals of the sham group, a surgical silk was passed around the acoustic meatus but was not tied off. After earplugging, earplug removal or sham procedure (Fig. 1B) rats were returned to their home cage in the animal facility. For the earplugging, earplug removal or sham procedure rats were anesthetized with isoflurane and placed on an isothermal pad.
For the monaural earplugging, a foam plug (Aearo Company) was then carefully placed into the right external canal at P30 (Fig. 1B), and the base of the pinna was closed with stitches. The sham and earplugged groups had ABRs recorded 20 min after the earplugging or sham procedures (data not shown). All animals were inspected daily. After 10 d of earplugging (P40), ABRs were again recorded (Fig. 1B,C), and the earplugs were removed from seven rats. After earplug removal, daily inspections of the external canal and the tympanic membrane were performed with a clinical otoscope to ensure that the ear canal was kept clean and the tympanic membrane was healthy. Any detritus that filled the plugged ear canal was removed. All animals included in the study had a normal-looking external canal and tympanic membrane. Rats were retested at 10 d after earplug removal (P50; Fig. 1B,C). After the appropriate survival time, rats were transcardially perfused.
To prepare the cochlear nucleus for transmission electron microscopy, we followed a protocol described by Deerinck et al. (2010). A total of eight rats (sham P40, n = 2; earplugged P40, n = 2; sham P50, n = 2; earplug-removal P50, n = 2) were used. The animals were paired for all procedures with one sham and one experimental animal in each pair. Animals were deeply anesthetized with a ketamine/xylazine mix (60 and 6.5 mg/kg, respectively) and perfused through the heart with a fixative solution containing 2% paraformaldehyde and 2.5% glutaraldehyde in 0.15 m cacodylate buffer, pH 7.4, with 2 mm calcium chloride at room temperature. Brains were then removed and postfixed for 2 h in the same fixative solution at 4°C. Coronal brainstem slices (80 μm thick) were vibratome sectioned in an ice-cold solution (0.15 m cacodylate buffer and 2 mm calcium chloride). Slices were washed with the same solution and incubated in a solution containing 3% potassium ferrocyanide in 0.3 m cacodylate buffer with 4 mm calcium chloride and 4% aqueous osmium tetroxide. Tissues were washed with doubled distilled H2O (ddH2O) and incubated in a filtered thiocarbohydrazide solution. After this incubation, sections were washed with ddH2O and placed in a 2% osmium tetroxide solution in ddH2O. Sections were washed with ddH2O and placed overnight in 1% uranyl acetate. Afterward, sections were stained en bloc using Walton's lead aspartate solution (Walton, 1979). After several washes in ddH2O, sections were dehydrated in a series of ethanol rinses (50, 70, 85, 95, and 100%), infiltrated with epoxy resin (#EMbed-812; Electron Microscopy Science), and embedded between acetate sheets. Sections were placed in an oven at 60°C for 48 h for polymerization. Ultrathin sections for the analysis of postsynaptic density (PSD) length and thickness (see below) were counterstained with lead citrate to enhance the electron-dense material of the PSD.
After polymerization, the ipsilateral (right side) rostral anterior anteroventral cochlear nucleus of sham and experimental groups were trimmed and block-mounted. All blocks were obtained from similar core regions of the anteroventral cochlear nucleus. A series of 25 serial ultrathin sections (60–70 nm) was cut per block (two blocks per animal) with a Leica EM UC7 ultramicrotome and collected on formvar-coated, single-slot grids. Serial ultrathin sections were observed with a JEOL 1400 transmission electron microscope. Identification of auditory nerve (endbulb) synapses on bushy cell somata was completed following the criteria previously described by Gómez-Nieto and Rubio (2009) and Ryugo et al. (1997). Identified endbulb synapses on bushy cells were photographed at 60,000 × magnification with a side-mounted CCD camera (GATAN). Image brightness and contrast were modified using Adobe Photoshop (Adobe Systems).
We used the Reconstruct software (http://synapses.clm.utexas.edu/tools/reconstruct/reconstruct.stm; Fiala, 2005) to create 3D reconstructions of the PSD of 60 endbulb synapses on bushy cells (n = 15 for each animal group). The electron micrographs for the 3D reconstructions were taken at 60,000×.
In single images (at 60,000 × magnification), in which the presynaptic membrane was parallel to the postsynaptic plasma membrane, the thickness of the PSDs (n = 88, 22 for each animal group; see Fig. 4) was analyzed as described previously by Dosemeci et al. (2001) and Rubio (2006) using ImageJ software (rsbweb.nih.gov/ij/; see Fig. 4). An observer blinded to each animal group performed all measurements.
Using single TEM images, the density of synaptic vesicles (SVs) per endbulb profile was measured in a total of 252 terminals (sham P40, n = 53; earplugged P40, n = 67; sham P50, n = 65; earplug-removal P50, n = 67) at a magnification of 30,000×, after plotting all of the SVs per endbulb terminal and calculating the endbulb ending area. Mitochondria within the axoplasm were not subtracted. The SV size was measured in a total of 9270 vesicles (for all animal groups). We manually delineated the outer profile of all SVs that exhibited both inner and outer membranes using Canvas software; for this analysis, we excluded SV caps or ghosts. To adjust for possible sectioning artifacts (Karunanithi et al., 2002), we made corrections to SV size by applying the correction formula described by Parsons et al. (1995). As shown in Figure 7A in the study by Parsons et al. (1995), we imagined a spherical synaptic vesicle of radio R move through a section of thickness h and plot the largest diameter of the synaptic vesicle profile captured by the section against the position of the synaptic vesicles' center. The average profile radius, r, is equal to the area under the curve divided by the horizontal distance between the extremes of the curve [r = (πR2/2 + Rh)/(2R + h)] as an estimate of the diameter of spherical structures. We also calculated the SV volume, as described by Karunanithi et al. (2002), using both the outer- and inner-diameter measurements (outer diameter minus twice the vesicle membrane thickness).
To determine the expression of vesicular glutamate transporter (vGluT1) in auditory nerve endings and AMPAR subunits (GluA2–4) at the PSDs of endbulb synapses on bushy cells in response to 10 d of earplugging and at 10 d after earplug removal, we used postembedding immunogold after freeze-substitution as described previously (Rubio and Wenthold, 1997, 1999; Whiting et al., 2009). Paired sham (n = 4) and earplugged (n = 6) animals were used for transcardial perfusions, tissue preparation, freeze-substitution, and postembedding immunogold labeling (sham P40, n = 2 rats; earplugged P40, n = 3 rats; sham P50, n = 2 rats; earplug-removal P50, n = 3 rats). Briefly, brains were perfused with a low glutaraldehyde fixative containing 4% paraformaldehyde and 0.5% glutaraldehyde in 0.12 m phosphate buffer, pH 7.2, for freeze-substitution. All brains were postfixed for 2 h in the same fixative solution at 4°C, vibratome sectioned (150 μm thickness), and cryoprotected in glycerol. The rostral anteroventral cochlear nucleus was dissected, frozen by plunging into liquid propane, and processed first for freeze-substitution. For postembedding immunogold labeling for vGluT1 or AMPAR subunits, ultrathin sections (80 nm thickness) were processed using a protocol similar to that described by Rubio and Wenthold (1997) and Rubio et al. (2008). A polyclonal rabbit antibody for vGluT1 (RRID:AB_2617092; Synaptic Systems) was used at a 1:350 dilution. Affinity-purified C-terminus rabbit polyclonal antibodies for GluA2, GluA2/3, and GluA4 were gifts from Dr. Robert Wenthold (NIDCD–NIH, Bethesda, MD) and were used at a dilution of 1:100. Primary antibodies were recognized by anti-rabbit IgG gold-conjugated (10 nm in diameter) secondary antibodies (Ted Pella). Sections were counterstained with uranyl acetate and lead citrate.
We calculated the number of gold particles per PSD and the linear density of gold particles per PSD for GluA2/3, GluA2, and GluA4 in a total of 1804 endbulb synapses on bushy cells as described previously (Rubio and Wenthold, 1997; Rubio, 2006; Whiting et al., 2009). PSD length was measured using Canvas software, and the number of gold particles was manually counted. Additionally, the percentage of labeled and unlabeled PSDs was plotted in relation to the number of gold particles. The number and density of vGluT1 gold labeling was calculated in a total of 202 auditory nerve endings (sham P40, n = 42; earplugged P40, n = 42; sham P50, n = 52; and earplug-removal P50, n = 68). Only gold particles within the axoplasm of the auditory nerve ending were quantified. The number of gold particles per synaptic vesicle was calculated in 20 endbulb profiles per animal group by dividing the total number of gold particles per endbulb profile by the total number of synaptic vesicles per profile. The number and density of gold particles for AMPAR subunits and vGluT1 were not different among the animals analyzed for each of the groups [i.e., sham (P40 and P50), 10 d of earplugging (P40), or 10 d after earplug removal (P50)].
To analyze ABRs across conditions (sham vs earplugged) and age (P30, P40, and P50), two-way repeated-measures ANOVAs (one-factor repetition) were performed (factors, condition × age; repeated factor, age). Data were tested for normality of distribution (Shapiro–Wilk test) and equality of variance (Brown–Forsythe test). If a significant interaction was observed, multiple comparisons (versus control group) were performed (Holm–Sidak method; control group for age, P30; control group for condition, sham). This analysis was performed for each ABR measurement (Amplitudes, Latencies, and Thresholds) and wave (wave I, II, and IV) separately. The purpose of this analysis was to understand whether there was an interaction between condition and age, which would indicate that the effect of different levels of condition (sham vs earplugged) depended on what level of age was present (P30, P40, or P50), i.e., whether there were differences in ABR responses across age depending on the presence or absence of the earplug. Since at P30 both groups are control groups, we do not expect to see a significant difference at this age level. We expect to see significant interactions between condition and age due to differences at P40 and/or P50 between shams and earplugged animals. A significant difference at P50 would indicate that ABRs were not recovered after earplug removal. For the ultrastructural analysis and postembedding immunogold labeling for vGluT1 and AMPA receptors, comparisons were performed using a Mann–Whitney U test. All data are expressed as the means ± SEM, and the level of significance was fixed at 0.05 for all comparisons. Statistical analyses were performed using SigmaPlot 13 (Systat Software) and Prism 5.0d software (GraphPad Software).
To evaluate whether earplugging had an effect on animal hearing thresholds, we measured ABR thresholds in response to clicks and pure tones in the two animal groups (sham vs earplugged) at three different ages (P30, P40, and P50; Fig. 1B). Earplugging significantly increased ABR thresholds for clicks by 43 ± 1.50 dB SPL (two-way repeated-measures ANOVA: main effect of condition, F(1,22) = 88.87, p < 0.001; main effect of age, F(2,22) = 210.99, p < 0.001; interaction, F(2,22) = 144.17, p < 0.001; Fig. 1D). Post hoc comparisons (Holm–Sidak method) showed that the average ABR thresholds of earplugged animals during earplug were significantly higher than ABR thresholds of age-matched sham animals (i.e., P40: t = 17.80, p < 0.001; Fig. 1D). Moreover, the ABR thresholds of earplugged animals after earplug removal were significantly higher than ABR thresholds of age-matched sham animals (i.e., P50: t = 3.42, p < 0.001; Fig. 1D). Therefore, earplugged animals do not recover their ABR thresholds in response to clicks to the level of control animals after 10 d of earplug removal.
We next addressed how ABR thresholds in response to pure tones of different frequencies (4, 8, 16, 24, and 32 kHz) were affected by earplugging. We analyzed the data with a two-way repeated-measures ANOVA for each frequency tested. There was a significant effect of earplugging on ABR thresholds in response to all frequencies (4 kHz: main effect of condition, F(1,22) = 47.86, p < 0.001; main effect of age, F(2,22) = 66.35, p < 0.001; interaction, F(2,22) = 73.84, p < 0.001; 8 kHz: main effect of condition, F(1,22) = 42.87, p < 0.001; main effect of age, F(2,22) = 210.52, p < 0.001; interaction, F(2,22) = 152.25, p < 0.001; 16 kHz: main effect of condition, F(1,22) = 35.81, p < 0.001; main effect of age, F(2,22) = 112.22, p < 0.001; interaction, F(2,22) = 63.56, p < 0.001; 24 kHz: main effect of condition, F(1,22) = 4.30, p < 0.001; main effect of age, F(2,22) = 72.53, p < 0.001; interaction, F(2,22) = 62.29, p < 0.001; 32 kHz: main effect of condition, F(1,22) = 86.52, p < 0.001; main effect of age, F(2,22) = 68.52, p < 0.001; interaction, F(2,22) = 54.12, p < 0.001; Fig. 1E1,E2). Post hoc comparisons (Holm–Sidak method) confirmed that the ABR thresholds for each frequency were significantly higher in earplugged animals during earplugging than in age-matched sham animals (i.e., P40; 4 kHz: t = 13.5, p < 0.001; 8 kHz: t = 14.14, p < 0.001; 16 kHz: t = 11.36, p < 0.001; 24 kHz: t = 13.07, p < 0.001; 32 kHz: t = 13.52, p < 0.001; Fig. 1E1,E2). After earplug removal, the same analysis showed no significant differences in ABR thresholds between earplugged animals and shams at P50 in response to low frequencies (4 kHz: t = 0.84, p = 0.41; 8 kHz: t = 1.04, p = 0.31; Fig. 1E1,E2). However, ABR thresholds in response to the middle and high frequencies were significantly different between earplugged and sham animals at P50 (16 kHz: t = 3.25, p < 0.01; 24 kHz: t = 4.38, p < 0.001; 32 kHz: t = 5.54, p < 0.001; Fig. 1E1,E2). In summary, these data indicate that earplugged animals did not recover their hearing thresholds after earplug removal in response to clicks and frequencies higher than 8 kHz but did recover their hearing thresholds in response to lower frequencies (4 and 8 kHz). After earplug removal, the external canal was clean and the tympanic membrane appeared healthy; nevertheless, the persistence of higher-frequency thresholds suggests either the existence of sensorineural hearing loss or middle ear problems (e.g., increased ossicular chain stiffness) as secondary effects of conductive hearing loss.
To evaluate how earplugging affected ABR amplitudes, we measured wave I, II, and IV amplitudes (peak to peak) at 10 dB SPL above threshold in both animal groups at the three ages (Fig. 1F1–F3). This analysis showed no significant effects of condition and age on ABR amplitudes for all waves, except wave I. Although significant effects were found for wave I amplitudes, no significant interaction was present (two-way repeated-measures ANOVA: main effect of condition, F(1,20) = 5.23, p = 0.045; main effect of age, F(2,20) = 12.72, p < 0.001; interaction, F(2,20) = 0.04, p = 0.96). Wave I amplitudes 10 dB above threshold significantly decreased between P30 and P40 (post hoc comparisons, Holm–Sidak method, t = 4.45, p < 0.001), but this decrease occurs in parallel between the two animal groups and is attributable to the effect of age. A reduction in wave I amplitude has been used as a measure to quantify degeneration in auditory nerve fibers (Shaheen et al., 2015). After earplug removal at P50 wave I amplitude was similar to the age-matched shams, this indicates that earplugging does not result in any measurable cochlear or auditory nerve degeneration. In summary, earplugging had no effect in ABR wave (I, II, and IV) amplitudes measured at 10 dB SPL above threshold.
We finally addressed whether earplugging changed ABR wave (I, II, and IV) latencies at 10 dB SPL above threshold (Table 1). We analyzed the data with a two-way repeated-measures ANOVA for each wave (I, II, and IV). This analysis showed a significant interaction between age and condition for waves I and II except wave IV (wave I: F(2,20) = 30.73, p < 0.001; wave II: F(2,20) = 19.42, p < 0.001; wave IV: F(2,20) = 2.39, p = 0.117). Post hoc comparisons of the interactions showed a significant increase in latency in the earplugged animals at P40 compared with aged-matched shams for waves I and II (Holm–Sidak method; wave I: t = 7.71, p < 0.001; wave II: t = 6.79, p < 0.001). No differences were found between the two animal groups at P50, indicating that after earplug removal, the animals recover their ABR latencies to the level of control animals.
To determine changes in gain (Schaette and McAlpine, 2011) in response to earplugging, the wave ratios of the earplug groups (II/I and IV/I) were normalized by the wave ratios of the age-matched sham group (see Materials and Methods; Fig. 1G). As expected, no changes in gain were observed in the earplug group at P30, since at this age these animals are also controls (relative gain of 1). During earplugging (P40), the relative gain of earplugged animals increased by 86 and 52% in the average wave II/I and IV/I ratios, respectively. After earplugged removal (P50), the relative gain decreased by 29% in the wave II/I ratio while it increased by 61% in the wave IV/I ratio. These results suggest that earplugging leads to a gain increase in both the cochlear nucleus and the lateral lemniscus. On the other hand, once sound levels restore after earplugging removal, the gain increase only persists in the lateral lemniscus that is an auditory brainstem nucleus upstream of the cochlear nucleus.
Next, we investigated the molecular and structural changes at endbulb synapses that could explain the functional deficits we detected with ABRs. vGluT1 is the main glutamate vesicular transported at the endbulb terminals (Zhou et al., 2007; Gómez-Nieto and Rubio, 2009; Fyk-Kolodziej et al., 2011), and vGluT1 expression levels could reflect how much glutamate is released to the synaptic cleft (Erickson et al., 2006; Liguz-Lecznar and Skangiel-Kramska, 2007). In the cochlear nucleus of deafened animals, the expression levels of vGluT1 at endbulb terminals are decreased compared with nondeafened animals (Zeng et al., 2009; Fyk-Kolodziej et al., 2011). We hypothesized that vGluT1 expression at the endbulb would decrease in response to 10 d of monaural conductive hearing loss. Postembedding immunolabeling for vGluT1 showed gold particles distributed within the axoplasm of endbulb terminals of shams, earplugged and after earplugged removal (Fig. 2). As expected, some gold particles for vGluT1 were also observed within the cytoplasm of bushy cells (Ito et al., 2011). Quantitative analysis of vGluT1 labeling within the axoplasm of endbulb endings showed a significant decrease (63%) in the average number and density of vGluT1 gold labeling in the earplugged group compared with age-matched P40 shams (number: U = 40.50, p < 0.0001, Mann–Whitney U test; density: U = 12, p < 0.0001, Mann–Whitney U test; Fig. 2B). The number of vGluT1 gold particles per SV was also significantly decreased in the earplugged animals (n = 0.8 gold particles) compared with age-matched shams (n = 3.5 gold particles; U = 13, p < 0.0001, Mann–Whitney U test; Fig. 2C). The ultrastructural morphometric analysis of the endbulb terminal profiles and SVs showed that the area of endbulb profiles and the average density of SVs within the terminal of earplugged animals were similar to age-matched shams (area: U = 193, p > 0.05, Mann–Whitney U test; Fig. 2B, Table 2; density: U = 1681, p = 0.7, Mann–Whitney U test; Fig. 3B). Interestingly, we found that SVs within the endbulb profiles of the earplugged group were significantly smaller in size (9.26%) and volume (11.81%) compared with the age-matched sham group (size: U = 396e + 006, p < 0.0001, Mann–Whitney U test; volume: U = 3.447e + 006, p < 0.0001, Mann–Whitney U test; Fig. 3C, Table 2). Thus, the smaller SVs and lower levels of vGluT1 labeling at endbulb endings suggest that conductive hearing loss reduces the number of glutamate molecules per SV, and this could lead to a decrease in vesicular glutamate uptake and release.
Ten days after earplug removal, data showed a significant increase of 145% in the average number and density of vGluT1 gold labeling in the earplug-removal group compared with age-matched P50 shams (number: U = 37, p < 0.003, Mann–Whitney U test; density: U = 26, p < 0.0001, Mann–Whitney U test; Fig. 2B). The area of the endbulb terminal profiles was similar between the earplug-removal and age-matched sham animals at P50 (U = 143, p = 0.99, Mann–Whitney U test; Fig. 2B), but we found an increase in SV size and density at endbulb terminals in the earplug-removal group compared with the age-matched sham group (size: U = 57,740; p < 0.0001, Mann–Whitney U test; SV density: U = 1003, p < 0.0001, Mann–Whitney U test; Fig. 3, Table 2). The number of gold particles per SV was similar between the earplug-removal animals (n = 6.4 gold particles) and age-matched shams (n = 4.1 gold particles; U = 139, p = 0.2, Mann–Whitney U test; Fig. 2C). Thus, the higher labeling of vGluT1 at endbulb terminals after earplug removal is explained by the increase in SV number.
The structure and composition of the PSD are modified by external stimuli and by synaptic activity over the time course of seconds to minutes and hours to days (Sheng and Hoogenraad, 2007). Thus, we examined the effect of transient monaural hearing loss on the area and thickness of the PSD of endbulb synapses on bushy cells (Fig. 4). Data from the earplugged animals at P40 compared with age-matched shams showed a 60% increase in PSD thickness (U = 12, p < 0.0001, Mann–Whitney U test). After earplug removal at P50 and when compared with age-matched shams, the PSDs were significantly larger and thicker by 59 and 65%, respectively (area: U = 57, p = 0.006; thickness: U = 6, p < 0.0001; Mann–Whitney U test; Fig. 4). No significant change in PSD area or thickness was observed between P40 and P50 shams (area: U = 113, p = 0.58; thickness: U = 91, p = 0.65; Mann–Whitney U test; Fig. 4). These results indicate that conductive hearing loss induces a long-term enlargement of the PSD at endbulb synapses on bushy cells.
Activity-dependent trafficking of AMPARs is a key mechanism underlying the strengthening and plasticity of glutamatergic synapses (Bredt and Nicoll, 2003; Collingridge et al., 2004). AMPARs are tetrameric complexes composed of homologous or heterologous combinations of GluA1–4 subunits. As different subunit compositions determine the electrophysiological properties of the AMPAR channel (Hollmann and Heinemann, 1994), it is critical to determine the long-term effects of conductive hearing loss on the expression of AMPAR subunits at the endbulb synapse on bushy cells (Fig. 5).
We first analyzed whether conductive hearing loss altered the percentage of the total pool of gold-labeled PSDs of endbulb synapses. In the P40 sham group, we found that 80.6% of the PSDs were gold labeled for GluA2/3, whereas only 43 and 39.3% of the PSDs were labeled for GluA2 or GluA4, respectively. After 10 d of earplugging (P40), the percentage of PSDs labeled for GluA2/3, GluA2, or GluA4 was similar to age-matched shams (p > 0.5, Mann–Whitney U test; Fig. 5B). Second, we determined the number and density of gold particles labeled for GluA2/3, GluA2, and GluA4 at the PSDs of endbulb synapses (Fig. 5C). This analysis revealed that GluA2/3 was upregulated in the earplugged animals compared with age-matched shams (number: U = 9454, p = 0.01, Mann–Whitney U test; density: U = 9423, p = 0.0122, Mann–Whitney U test). On the other hand, the values for GluA2 and GluA4 were not different from those of the age-matched sham group (GluA2, number: U = 6124, p = 0.75, Mann–Whitney U test; density: U = 5406, p = 0.44, Mann–Whitney U test; GluA4, number: p = 0.1690, Mann–Whitney U test; density: p = 0.1603, Mann–Whitney U test). These measurements indicate that the GluA3 subunit, which is an AMPAR subunit with rapid kinetics (Gardner et al., 1999), was increased at endbulb synapses in response to conductive hearing loss. This increase could represent a homeostatic compensatory mechanism in response to sound reduction.
Similar analyses were performed for earplugged animals at P50 (i.e., 10 d after earplug removal) and aged-matched shams (Fig. 6). In the P50 sham group, 70% of the PSDs were gold labeled for GluA2/3, whereas only 45% were labeled for GluA2 or GluA4. In the P50 earplugged group, the percentage of PSDs labeled for GluA2/3, GluA2, or GluA4 was similar to age-matched shams (p > 0.5, Mann–Whitney U test; Fig. 6B). The analysis of the number and density of gold particles per PSD revealed that GluA2/3 gold labeling remained significantly increased (number: U = 12,589; p = 0.0002, Mann–Whitney U test; density: U = 12,301; p < 0.0001, Mann–Whitney U test; Fig. 6C), whereas GluA2 gold labeling decreased compared with the age-matched sham group (number: U = 11,954; p = 0.0008, Mann–Whitney U test; density: U = 11,630; p = 0.0002, Mann–Whitney U test). No differences were observed in the number and density of GluA4 gold particles per PSD between the two animal groups at P50 (number: U = 14,340; p = 0.5455, Mann–Whitney U test; density: U = 14,233; p = 0.4619, Mann–Whitney U test; Fig. 6C). Within the shams group, the average number and density for GluA2/3, GluA2, and GluA4 was similar between P40 and P50 shams (p > 0.5, Mann–Whitney U test). Altogether, these results indicate that AMPAR subunit expression levels at the endbulb synapse do not return to preplugging levels even 10 d after earplug removal. Because of the decrease in GluA2 gold labeling alongside unchanged GluA4 labeling, the observed increase in GluA2/3 indicates that GluA3 was the primary AMPAR subunit at endbulb synapses after earplug removal. The high gold labeling for GluA3 subunits once sound levels are restored after earplug removal is not expected and suggests the absence of homeostatic mechanisms at endbulb synapses after transient sound deprivation.
Conductive hearing loss is the second most prevalent form of hearing loss, but its effects on central synapses have received relatively little attention. Auditory deprivation modifies auditory perception and physiology (Fournier et al., 2014). Psychoacoustics studies in adult humans with conductive hearing loss indicated plastic changes in the auditory brainstem as evidenced by changes in the threshold of an acoustic reflex, suggesting the existence of a gain mechanism operating at the level of the auditory brainstem (Munro and Blount, 2009). Our study shows that monaural conductive hearing loss (an ;40 dB SPL sound attenuation) in young adult rats leads to functional changes that persist after earplug removal and a recovery period of 10 d. Although wave amplitudes measured at 10 dB SPL above thresholds were not affected by earplugging and peak latencies were reversed to age-matched sham levels, the hearing thresholds in response to clicks and middle to high frequencies (16–32 kHz) were not recovered.
This increase in thresholds after a transient period of sound deprivation is consistent with studies performed in humans (Formby et al., 2003, 2007; Munro and Blount, 2009). Transient bilateral sound deprivation in gerbils (earplugging from P23 to P35), followed by a recovery period, does not impair auditory perception (Caras and Sanes, 2015). However, as occurs in the visual system, the effects of bilateral deprivation can be distinct from those because of monaural deprivation (Hutson et al., 2007). Therefore, the increase in thresholds found in our study could be behaviorally relevant during the recovery period or, alternatively, when animals are exposed to a noisy environment during that period. Altogether, our results show long-lasting presynaptic and postsynaptic structural and molecular effects at the endbulb of the Held synapse. Thus, we provide evidence for plasticity at the endbulb of Held synapses in the young adult in response to transient sound deprivation. However, this plasticity leads to an abnormal increase in the central gain within the lower auditory brainstem that may be involved in hyperacusis (for review, see Noreña, 2011).
Monaural earplugging led to a decrease in SV size and a decrease in vGluT1 gold labeling. vGluT1 expression levels determine both the amount of glutamate loaded into vesicles and the amount of glutamate that is released (Wojcik et al., 2004; Wilson et al., 2005). It is still unclear whether the number of vesicular glutamate transporter molecules per SV directly correlates with the size of SVs. However, it was recently shown that vGluT1 function generates the tonicity of glutamatergic SVs, suggesting an effect on SV shape, and influences the stability of SVs in presynaptic terminals (Siksou et al., 2013). Furthermore, an in vivo study has shown that zebrafish glutamate transporter mutants have impaired presynaptic release of glutamate (decreased mEPSC amplitude) attributable to reduced vesicular glutamate uptake (Smear et al., 2007). Therefore, during earplugging, synaptic transmission can be compromised as vGluT1 is the main vesicular glutamate transporter in the endbulb terminals (Zhou et al., 2007; Gómez-Nieto and Rubio, 2009). As a result of the decrease in vGluT1 and smaller SVs, there will be lower levels of luminal glutamate in SVs and, consequently, a decrease in glutamate release at the cleft of the endbulb synapse. In response to reduced cleft glutamate by conductive hearing loss, postsynaptic homeostatic mechanisms, such as increased glutamate receptors and subsequent increased PSD size, may occur at the endbulb synapse. Certainly, we show an increase of the GluA3 AMPAR subunit together with thicker PSDs in animals that were earplugged for 10 d. Based on our finding of an increased expression of AMPAR–GluA3 at the endbulb synapse, it is probable that monaural conductive hearing loss leads to a potentiation of synaptic responses. The potentiation of endbulb synapses through the upregulation of GluA3 might represent a compensatory mechanism to the reduction of auditory nerve activity by conductive hearing loss. A recent study demonstrated that sound-induced trauma slows AMPAR mediated excitatory synaptic current decay times by increasing GluA1 and decreasing GluA4 in the lateral superior olive (Pilati et al., 2016). The increased excitatory charge transfer caused by the slower current decay times boosted the compromised excitatory pathway, presumably enabling audible stimuli to provide a stronger excitatory input than in a normal animal (Pilati et al., 2016). The physiological consequences of an increase of the GluA3 subunit for synaptic transmission in the cochlear nucleus need further investigations, but it could well be that the decay times of excitatory currents are slower in response to earplugging. This can happen, for example, if the ratio of GluA3 flip/flop isoforms is higher at the endbulb synapse in earplugged animals (Mosbacher et al., 1994; Gardner et al., 2001). This could provide a stronger excitatory input that will compensate for the reduction in sensory input. Additionally, our data show a gain increase in the average wave II/I ratio during earplugging in relation to control animals. Independent of whether monaural conductive hearing loss is short (1 d; Whiting et al., 2009) or long (10 d; present study), the GluA3 AMPAR subunit increases at the endbulb synapse in response to a decrease in sound levels. Thus, our results indicate that auditory experience-dependent adult plasticity of endbulb synapses is mediated specifically by regulation of GluA3 subunits.
The functional changes observed in the present study after 10 d of monaural conductive hearing loss in young adult rats and a recovery period of another 10 d were paralleled by structural and molecular presynaptic and postsynaptic changes at the level of the endbulb synapse in the cochlear nucleus.
After earplug removal, we found in the endbulb terminals that vGluT1 gold labeling per SV recovered to sham levels. However, endbulb terminals had more SVs that were also larger. These findings suggest an increase in cleft glutamate because SVs are bigger and should thus have more glutamate molecules (Daniels et al., 2006). Nevertheless, an increase in cleft glutamate after earplug removal does not necessarily indicate more glutamate receptor activation because of the fast deactivation and desensitization of GluA3/4 AMPARs at the endbulb synapse (Gardner et al., 1999; Chanda and Xu-Friedman, 2010). Therefore, the putative increase in cleft glutamate may not translate into larger EPSCs (Vasileva et al., 2012). Hypothetically, an increase in glutamate in the synaptic cleft could have led to a decrease in the number of AMPARs in the PSD, as a way to compensate for the excess of glutamate in the cleft. However, GluA3 subunit labeling was increased in response to conductive hearing loss after the recovery period. The increase in glutamate together with an increase in AMPARs can also, but not necessarily, potentiate synaptic transmission in the endbulb terminal. Additional evidence arguing in favor of an increase in cleft glutamate after earplug removal was the larger and thicker PSDs of endbulb synapses compared with sham-treated animals. Endbulb synapses of congenital deaf cats and mice have hypertrophied PSDs (Ryugo et al., 1997; Lee et al., 2003), but the hypertrophied PSDs resemble those of normal hearing cats after sound stimulation (Ryugo et al., 2005), indicating that auditory nerve synaptic activity influences the size of the synapse. We can speculate that larger PSDs and more GluA3 AMPARs will likely make this synapse less resilient and more vulnerable to changes in environmental sound levels. Thus, our results suggest that normal auditory nerve function, which is important for the transmission of fine timing information that is necessary for sound localization (Yang and Xu-Friedman, 2009), is likely affected after conductive hearing loss. Moreover, our finding that an increase in gain at the level of the lateral lemniscus persists after sound levels are restored suggests long-lasting changes that can compromise auditory processing beyond the cochlear nucleus after conductive hearing loss.
Most of the studies addressing plasticity in the adult (P30 and beyond) have been performed in the barrel or visual cortices. For example, the barrel cortex receptive fields can be altered by experience in adult rats (Polley et al., 2004; Rema et al., 2006); dendritic spines, which are considered to be the morphological correlates of experience-dependent synaptic plasticity, continue to turn over in adults in both the visual and somatosensory cortices (Zuo et al., 2005; Holtmaat et al., 2006). Manipulation of in vivo sensory experience also produces homeostatic changes at cortical synapses. Depriving vision from 2 to up to 7 d scales up excitatory synapses in the visual cortex pyramidal neurons, by incorporating more GluA1 AMPAR subunits, whereas reexposing the visually deprived animals to light scales down the synapses (Goel et al., 2006, 2011; Goel and Lee, 2007). Interestingly, the sensory experience-induced homeostatic synaptic changes are mainly observed as regulation of the amplitude rather than the frequency of AMPA–mEPSCs. This suggests that in vivo sensory experience-dependent changes in the cortex are mainly postsynaptic in nature at the level of AMPAR regulation (Goel and Lee, 2007; Lee, 2012). Overall, our study shows in P30 rats, once past maturation of connectivity and synaptic elements in the cochlear nucleus, that earplugging sets in motion complex changes that may ultimately lead to increases in synaptic strength at endbulb synapses through presynaptic and postsynaptic mechanisms (Fig. 7). At 10 d of earplugging, vGluT1 expression levels at endbulb endings decreased and synaptic vesicles were smaller; postsynaptically, the PSD was thicker and contained more GluA3. After 10 d of earplug removal, vGluT1 expression recovered, but synaptic vesicles were larger compared with both shams and 10 d of earplugging. At the postsynaptic levels, the PSD became enlarged in size and thickness and contained more GluA3 AMPA receptor subunits. It is striking that some of these changes, such as the increase in GluA3 and PSD thickness, continue after earplug removal. Other changes begin after earplug removal and do not reverse, e.g., GluA2 removal and rebound enhancement of SV size and density. The rebound effect is surprising and contrasts with a simpler “homeostatic plasticity” concept (for review, see Nelson and Turrigiano, 2008). Our findings show that sensory-dependent evoked plasticity is more complex than what might be predicted from experiments in reduced systems. Future studies could investigate in the adult long-term hearing sensitivities after conductive hearing loss in chronic otitis media with effusion, which is prevalent in children. Such studies will settle the central mechanisms of deficits in auditory perception, language acquisition, or educational disabilities that occur after inadequate or abnormal sensory experience.
This work was supported by a grant from the NIH (R01DC013048 to M.E.R.) and by internal funding from the Department of Otolaryngology, University of Pittsburgh Medical School (M.E.R.). We authors thank Drs. Karl Kandler and Laurence Trussell for helpful suggestions and critical reading of this manuscript. We also thank Naama Balass for editing this manuscript.
The authors declare no competing financial interests.