Keywords

3.1 Introduction

According to the current standard (ISO1999: 2013(E)), noise-induced hearing loss (NIHL) is defined by sustaining a permanent threshold shift (PTS). However, this definition has been challenged by the fact that noise exposure can cause massive damage to the synapses between inner hair cells (IHCs) and spiral ganglion neurons (SGNs) in the cochleae of laboratory animals without a significant PTS [24, 43, 48, 50, 88, 90, 94]. The synaptic damage and the associated functional deficits in signal coding by auditory nerve fibers (ANFs) have been labelled as noise-induced cochlear synaptopathy. Since coding deficits in the absence of a PTS cannot be detected by routine audiological evaluations that are currently focused on seeking thresholds, they are umbrellaed under the concept of noise-induced hidden hearing loss (NIHHL) [42, 47, 51, 57, 64, 94]. This chapter will review the current knowledge of the potential mechanisms of noise-induced synaptic damage and the following repair processes after a brief review on the special anatomy of the synapse between IHCs and SGNs. The chapter will then shift its focus to the therapeutic methods promoting the regeneration of the synapses.

3.2 Anatomic and Functional Features of Cochlear Ribbon Synapse

The synapse between IHCs and SGNs is characterized by the presence of an electron-dense, ribbon-like structure and therefore called a ribbon synapse. It is mainly found in the retina, the inner ear, and in the pinealocytes. The structural features of the ribbon synapses between IHCs and SGNs are summarized in Fig. 3.1. The synaptic ribbons found in mature hair cells are anchored to the plasma membrane, one ribbon per active zone (AZ). A small number of “floating” ribbons (<5%) were observed and probably reflected the turnover of these subcellular organelles [40, 114]. The synaptic ribbons in the IHC are shaped like an American football, with a tee structure underneath that is formed by a protein named Bassoon. This bar structure anchors the ribbon to the AZ [59].

Fig. 3.1
figure 1

Schematic view of IHC ribbon synapses

Full size image

The main protein forming the framework of the ribbon is called the Ribeye, which consists of two domains: the A-domain is located inside and appears to have a structural role, whereas the B-domain points to the cytoplasmic face of the ribbon, where it interacts with other proteins and tethers synaptic vesicles [2]. The aminoterminal A-domain is not homologous with any other protein in the public databases and therefore specific to the ribbon synapse, whereas the carboxyterminal B-domain is largely identical to the nuclear corepressor protein, C-terminal binding protein 2 (CtBP2). The gene encoding the Ribeye is called the CtBP2 gene, which encodes two proteins: the unique Ribeye(A + B) in the ribbon synapses and the CtBP2, which is also expressed in the cellular nucleus [52, 100, 106, 114]. The Ribeye in photoreceptor cells contains CtBP1 [102], which has not been verified in IHC ribbons.

The scaffold of synaptic ribbons is built up from multiple Ribeyes [52]. The Ribeye A-domain has three interaction docking sites that mediate homotypic interactions with other RIBEYE(A)-domains. In addition, homotypic B-domain interactions can be formed as well as heterotypic interactions between the RIBEYE A- and B-domains, which are regulated (inhibited) by nicotinamide adenine dinucleotide hydride (NADH). In the photoreceptors, ribbon size dynamically changes in response to light: ribbons are disassembled in bright light and reassembled in dark [1, 75, 84, 95, 103]. It is not clear if the ribbons in the IHCs are dynamically disassembled/reassembled.

The most striking functional characteristic of the IHC ribbon synapse is its ability to make fast response to transient signals in the meantime to keep its long-lasting response to persistent stimuli. These features require special mechanisms to enable fast neurotransmitter release (exocytosis) and replenishment, as well as fast recycling of neurotransmitters via endocytosis. It is not entirely clear how these processes are realized. However, they must be related to the special protein compositions and the structure of the cytomatrix of the active zone (CAZ). Several proteins that are important for transmission across conventional synapses are not seen in IHCs. Those include synaptotagmins 1 and 2, synapsins, synaptophysins, synaptogyrin complexins, neuronal SNAREs as well as priming factors of the Munc13 and CAPS families (see reviews [56, 81]). Instead, the function of those proteins seems to be replaced by a single protein, otoferlin, which is located between the ribbon and the presynaptic membrane and strongly interacts with adaptor protein 2 (AP-2) [17, 38, 61].

Bassoon and Piccolo are two big proteins (>400 kDa) that are seen in conventional synapses. Their function in synaptic transmission is not clear. In ribbon synapses, Bassoon is responsible for anchoring the ribbons to the CAZ. The knockout of this protein in the cochlea of mice results in the loss of ribbons and the deterioration of temporal resolution of the auditory nerve fibers (ANFs), without significant change in hearing sensitivity [7, 37]. Piccolo is present in ribbon synapses as a shortened variant, called Piccolino, which is distributed over the entirety of the ribbon. Knockdown of this protein resulted in a lack of dynamic ribbon assembly in the retina of mice [23, 76]. However, it is not clear what role the Piccolino plays in IHC ribbon synapses.

The postsynaptic terminal of the ribbon synapses exhibits similarities with the conventional excitatory synapses. Glutamate has been confirmed as the neurotransmitter in the IHC ribbon synapse [27, 28, 58]. Once the neurotransmitter is released into the synaptic cleft, it activates an α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) residing within a receptor cluster at the postsynaptic density of afferent ANFs [27]. Glutamate receptor subtypes (GluR) 2, 3, and 4 are abundant in IHC ribbon synapses [53]. GluR2 is not expressed until the onset of hearing, while GluR3 and GluR4 are present earlier during development. N-methyl-d-aspartate (NMDA) receptors (NR1, NR2A/B) are also present at the afferent synapses in the cochlea [58, 69, 80]. They are not activated for fast transmission as they are blocked by magnesium at resting membrane potential [31]. However, they modulate the reaction of AMPAR to glutamate at the type I afferent terminals [9, 28].

3.3 Synaptic Damage by Noise

3.3.1 Potential Mechanisms

The finding that noise induces cochlear synaptopathy reveals a novel locus of cochlear damage for NIHL. The damage to the postsynaptic terminal occurs through a similar mechanism as seen in conventional excitatory synapses: the glutamate-induced excitotoxicity. This mechanism is supported by the fact that cochlear infusion with glutamate or agonists mimics noise-induced damage [68, 71, 78]. Calcium influx and accumulation in the postsynaptic terminal is the initial step toward excitotoxicity [98]. Among the ionotropic glutamate receptors (iGluRs), AMPARs, NMDARs, and kainite-Rs, NMDARs have been considered the major contributors to the calcium influx and accumulation in the postsynaptic terminals. Therefore they are mainly responsible for the excitotoxicity in general [65, 98] and in the mammalian cochlea [55, 98]. While the neurotransmission between IHCs and SGNs is mainly mediated by the AMPAR, it is not considered responsible for the calcium influx. However, this opinion has been challenged by a recent report [85]. In this study, three subunits of AMPARs (GluA2, GluA3, and GluA4) were identified, and only GluA2 lacks calcium permeability. Moreover, calcium influx to the postsynaptic neurons was found to occur mainly via the Ca-permeable AMPARs (CP-AMPARs), but not NMDARs as previously recognized. This conclusion is supported by the fact that the CP-AMPAR blocker, IEM1460 (N, N, N-trimethyl-5-(tricyclo [3.3.1.13,7] dec-1-ylmethyl) amino-1-pentanaminium bromide hydrobromide), significantly reduces calcium accumulation in the postsynaptic auditory neurons, whereas the NMDAR blocker, APV (DL-2-amino-5-phosphonovaleric acid), shows no effect.

It is not clear if there is/are presynaptic mechanism(s) for the synaptic damage by noise or other toxic factors other than the Ca2+-mediated glutamate release. The presynaptic ribbons within photoreceptors are dynamic: they are dissembled in bright light and reassembled in dark [1, 75, 84, 95, 103]. This dynamic change serves as a mechanism of adaptation to stimulation and results in the change of neurotransmitters released. However, there is no evidence to date that supports the ribbons in the IHCs being dynamic in their response to acoustic overstimulation. Based upon the immunohistochemistry observation, the noise-induced reduction of the presynaptic ribbons is parallel with the breakdown of the postsynaptic terminals [89, 94]. In the photoreceptors, ribbons are dissembled when the cells are hyperpolarized by light that causes a reduction of [Ca2+]i. In the IHCs, the response to acoustic stimulation is a depolarization of membrane potential (not hyperpolarization) and an increase of [Ca2+]i. Therefore, if there is a disassembly/reassembly process in the IHCs, it must undergo a different mechanism. It is possible that the presynaptic ribbons in the IHCs are broken down after the postsynaptic terminals are destroyed. More research is needed to identify the fate of the ribbon protein after they are broken down by noise.

3.3.2 Selective Damage to Synapses with Low Spontaneous Rate Units

One IHC synapses with more than ten SGNs, and the synapses are distributed around the bottom of the IHC. The susceptibility of the synapses to noise damage appears to be location dependent: the synapses at the modiolar side of the IHC are more easily damaged. Although, the underlying mechanism is not entirely clear, this bias has been linked to the morphological variation around IHCs when identified in immunohistochemistry against CtBP2 and AMPAR (Fig. 3.2). The synapses close to the modiolar side of an IHC have larger ribbons and smaller postsynaptic terminals, whereas the synapses distributed toward the pillar side are the opposite [46]. This difference is functionally important because the synapses located at the modiolar side of IHCs innervate auditory nerve fibers (ANFs) that have lower spontaneous spike rates (SR), higher thresholds, and larger dynamic ranges. These ANFs are considered critical for hearing in noisy backgrounds, where only the high spontaneous SR units are saturated [10, 18, 33, 63, 112].

Fig. 3.2
figure 2

Spatial variations of ribbon synapses around the IHC. The synapses on the modiolar side appear to have a larger ribbon but a smaller postsynaptic terminal. The synapses on this side are also more sensitive to noise-induced damage

Full size image

Several potential mechanisms have been proposed to explain the difference in the noise susceptibility between the synapses around IHCs. Firstly, there is a heterogeneity of Ca2+ channels around IHCs: synapses at modiolar side appear to have more Ca2+ channels per CAZ and display a higher Ca2+ influx and potentially a larger neurotransmitter release [20, 54]. The activation of the Ca2+channels at the modiolar side requires a larger degree of depolarization [54, 109, 110]. This heterogeneity has been linked to the spatial variations in the threshold and dynamic range of ANFs and to synaptic damage by noise around the IHCs [56]. Secondly, the larger amount of neurotransmitter release may be related to the larger ribbon size. The larger ribbons at the modiolar side can harbor more neurotransmitter vesicles close to the CAZ [27, 45, 93]. However, it is not clear if the vesicular priming and replenishment occurs faster for the ribbon synapses at the modiolar side of IHCs. Thirdly, the clearance of the released glutamate likely occurs slower at the modiolar side due to the lower amount of glutamate-aspartate transporters (GLAST) [25, 26, 74]. Fourthly, iGluRs (including AMPARs and NMDARs) are responsible for the glutamate-induced excitotoxic cell death in many neurologic diseases [29, 104]. Previously, NMDARs were thought to play a major role in noise-induced postsynaptic cochlear damage [4, 41], but more recently, this role has been attributed to Ca2+-permeable AMPAR as discussed above [85]. Nevertheless, it is not clear if the NMDARs are selectively distributed to the synapses at the modiolar side of the IHCs [83]. Interestingly, heterogeneity in the relative distribution of both Ca2+-permeable and Ca2+-impermeable subunits of AMPARs has been demonstrated across the IHC-SGN synapses. However, it is not clear how the heterogeneity is related to the synaptic distribution around the IHCs.

The functional significance of the selective damage to ANF synapses with the low-SR units remains to be confirmed. Theoretically, the selective loss of synapses with the low-SR ANFs will impair signal coding in strong background noise, one of the major problems seen in aging subjects [42, 56, 57, 88]. However, this coding deficit remains a speculation and has not been confirmed in single-unit data.

3.3.3 Can the Disrupted Ribbon Synapse Be Rebuilt?

It is currently debated whether the synaptic disruption by noise is reversible. In a pioneering study with CBA mice, no significant recovery of synapse counts was found after the threshold recovery that occurred in 1 week after the noise exposure [43]. Therefore, a single brief noise exposure caused up to 50% loss of synapses, permanently. However, studies from our labs in both Canada and China found that the decrease in synapse count was largely reversible in guinea pigs [50, 89, 94]. This reversibility was also reported in mice other than the CBA strain [90, 91]. It is worthy to notice that the concept of synaptic repair may involve two different phenotypes: (1) the rebuild of the synapses that are destroyed (by synaptogenesis) and (2) the repair of survived synapses that are damaged but not disconnected. The synaptic repair of the second type was reported after the initial noise-induced damage by the group of scientists who first reported noise-induced synaptic damage in the cochlea [67, 70, 71, 77]. In those early reports, the synaptic damage was observed using transmission electron microscopy (TEM). This technique limited the observation on the synapses that were partially damaged, but not destroyed. More importantly, the observation was not quantitative for the counting of total synapses. More recently, the dynamic changes of the number of ribbon synapses were reported in a study using AMPA infusion [79]. In our labs, the rebuild of the disrupted synapses was demonstrated by recovery of synapse counts. Functional data supported the synaptogenesis in that the recovery of the synapse count was matched by the recovery of compound action potential (CAP) measures (Fig. 3.3 a and b) [50, 89, 94]. In addition, the repair is also supported by the morphological changes of the synapses in the noise-damaged cochlea. Shortly after noise-induced damage, some synapses were found to be located up to the level of IHC nuclei and with extremely large ribbons, seen in immunohistochemical observation [89]. The synapse distribution returned to normal several weeks after the noise exposure, suggesting the re-established synapses were formed at a location close to the protein synthesis organelle. Furthermore, the repaired presynaptic ribbons appeared to have uneven sizes, with bigger hollow cores. In addition, many synapses observed weeks after the noise exposure had multiple ribbons to one AZ in TEM observation (Fig. 3.3c) [94]. This feature is seen in naïve ribbon synapses during early development [81] and is consistent with the regeneration of the synapses after they are destroyed by AMPA [79].

Fig. 3.3
figure 3

Evidence for synaptic repair after noise-induced damage. (a) Immunostaining images of pre- and postsynaptic components in control (ctrl) IHC and those 1 day, 1 week, and 1 month post noise (1DPN, 1WPN, and 1MPN). (b) Percentage changes of maximal CAP amplitude and synapse counts after noise. (c) TEM images of IHC ribbon synapses taken at 1MPN, showing hollow cores in some ribbons and double ribbons in some synapses

Full size image

3.4 Synaptic Protection and Regeneration in Noise-Induced Synaptopathy

3.4.1 Synapse Protection

Noise-induced ribbon synapse damage involves the structural breakdown of both presynaptic ribbons and postsynaptic terminals [43, 89]. The mechanisms for the noise-induced damage on the postsynaptic terminal are clear and likely due to the glutamate-mediated excitotoxicity [70,71,72, 79]. Ca2+ overload via GluRs and voltage-gated Ca2+ channels (VGCCs) has been recognized as playing a critical role in noise-induced cochlear damage, both on HCs and postsynaptic terminals [5, 62, 65, 85]. Application of VGCC blockers (both L- and N-types) has shown the ability to protect the cochlear HCs from noise damage, consistent with the distribution of those calcium channels on HCs [34, 39, 49, 87, 101, 113]. However, it is not clear if the application of the blocker can prevent noise-induced synaptic damage.

Since noise causes synaptic damage via GluRs, blockage of these receptors may protect the synapses against noise. HC damage has been seen as part of excitotoxicity in zebrafish larvae, in which iGluRs are found to be expressed in the HCs [86]. Several studies have shown that NMDAR blockers can prevent tinnitus induced by salicylate [12, 66] and noise [4, 32]. Further research is needed to verify potential mechanisms [82, 83]. It is also important to note that blocking of iGluRs may have unforeseen effects. For example, long-term blockage of NMDA has been found to hinder the regeneration of the IHC-SGN synapses after excitotoxic damage [11, 79].

Previously, NMDAR antagonists have been tested for this potential protection [14, 15, 60]. However, the most significant effect of protection was seen on HCs, not on SGNs. Application of the NMDAR antagonists has been reported to reduce the swelling of the afferent dendrites synapsed with the IHCs in guinea pigs [14, 15]. However, the method for synapse quantification in those studies is questionable, since the number of the damaged synapses by noise in those studies was much fewer than that reported more recently using immunohistochemistry staining [43, 50, 89, 94]. Clearer evidence of synaptic protection against noise by the NMDAR antagonist was reported more recently [4]. However, in this study, the antagonist was administered at least 2 days after the noise exposure. Therefore, it is not clear what mechanism is underneath the reduction in synaptic damage. Presumably, the effect of the NMDARs in noise-induced damage to the afferent dendrites is based upon their role as a ligand-gated calcium channel. However, a recent study indicated that the sound-induced calcium entry was not mediated by NMDARs but by Ca2+-permeable AMPARs at the site [85]. This finding has shaken the theoretical basis of using NMDAR antagonists to reduce noise-induced synaptic damage in the cochlea. Previously, one study showed the protective effect of a blocker (caroverine) against both AMPA and NMDA receptors. It reduced the HC loss caused by impulse noise [16]. However, the protective effect on the synapses was not investigated. Therefore, further studies are needed to verify if NMDAR and/or AMPAR antagonists can protect the synapses from noise.

3.4.2 Synapse Regeneration

Since the synaptic damage induced by noise is partially reversible, there exists an endogenous mechanism to maintain the stability of the synaptic connections between SGNs and IHCs. Various studies have indicated the role of neurotrophic factors (NTFs) in synapse formation during development, plasticity, and the maintenance of synaptic stability in the cochlea (see review [21, 22, 73, 111]). Using NTFs appears to be a practical approach to rescue the damaged auditory nerves and their synapses to HCs [3]. Neurotrophins are a subclass of NTFs that are ubiquitously expressed and are very extensively studied. There are four types of neurotrophins in mammals: nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT-3), and neurotrophin-4/5 (NT-4/5). BDNF and NT-3 are the major types of neurotrophins seen in the mammalian cochleae [73]. Within these two, BDNF is highly expressed during early development and declines to undetectable levels in adulthood. NT-3 is the only neurotrophin that exists in the adult cochleae, in addition to glial cell line-derived neurotrophic factor (GDNF) [13, 21, 30]. While the p75 neurotrophin receptor (p75NTR) is the receptor shared by BDNF, NT-3, and GDNF, each of the three factors has its own specific binding site(s): NT-3 binds mainly to the receptor C of the tropomyosin-related kinase (TrkC), BDNF to TrkB, and GDNF to RET-GFRα1 complex [73]. However, TrkA, TrkB, and TrkC are all expressed in the adult cochlea, even though their corresponding ligands (NGF to TrkA and BDNF to TrkB) are not detectable [30]. Therefore, BDNF and NGF can also be used for therapy in addition to NT-3, for the regeneration of the synapses after disruption.

NT-3 has been examined in several studies for its ability to promote the regeneration of the synapse. In the mammalian cochlea, NT-3 is expressed in both HCs and supporting cells [19, 22]. NT-3 overexpression by gene knock-in has been reported to increase the synapse density between IHCs and SGNs and decreased ABR threshold in mice [105]. The study suggested that the supporting cells are a more important source of NT-3 because the selective knock-in of NT-3 in the supporting cells promoted the regeneration of the synapses after disruption by noise exposure. However, the effect of selective knock-in in the HCs on synapses regeneration was not examined in this study.

Two studies showed a rescue effect of exogenous NT-3 to IHC-SGN synapses, when applied through the round window after noise trauma [92, 96]. In the first study done in mice, NT-3 was administered via the round window 24 hours after a 2 h noise exposure at 100 dB SPL, with the synapse count performed days later [96]. The NT-3 was delivered via slow-release gel placed in the round window niche. The protective effect of NT-3 was evaluated in both functional tests of auditory brainstem response (ABR) amplitude and synapse count. However, a large individual variation in the protective outcome made the authors divide the subjects into “effective” versus “ineffective” subgroups. Presumably, the “ineffective” was likely due to the failure in NT-3 application. A weak significance was seen when all of the NT-3 treated subjects were grouped together. In the second study listed above, guinea pigs were used. An equal mixture of NT-3 and BDNF was applied to the round window immediately after the noise exposure, which was given either at 95 or 105 dB SPL for 2 hours [92]. The synapse count observed 2 weeks after the noise exposure showed a significantly greater number in the ear treated with neurotrophins for the subjects receiving the 95 dB noise exposure. Since no data was reported from the subjects receiving the 105 dB noise exposure, and no control subjects were assessed, the interpretation of this data is difficult. Furthermore, the mixture of the two neurotrophins makes it difficult to measure the contribution from each neurotrophin.

Instead of using exogenous neurotrophins, the gene therapy type of approach appears to be more attractive in that it can provide long-term protection against repeated noise exposures. Using the gene knock-in technique, it has been found that overexpression of the NT-3 (but not BDNF) gene in supporting cells could significantly promote the ribbon synapse regeneration after noise-induced damage. However, no such protection was seen if the overexpression was only done in the IHCs [105]. In this study of normal-hearing guinea pigs, the overexpression of NT-3 in supporting cells and IHCs surprisingly increased the synaptic density of IHCs. Furthermore, the increase in synaptic density was accompanied by an increased ABR wave I amplitude and a decreased ABR threshold [105].

Due the ethical considerations, gene knock-in is unlikely to be used on human subjects. Instead, local gene transfection is an approach that can be translated to human clinics. To date, viral vectors appear to be much more effective in gene transfection. Among the viral vectors available, adeno-associated virus (AAV) is the most attractive due to its safety and the ability to cause long-lasting expression of the transfected gene. Several human trials of gene therapy are ongoing using AAV vectors. In a recent report, AAV-mediated NT-3 overexpression was found to cause considerable regeneration of synapses between IHCs and ANFs in guinea pigs that were deafened by aminoglycosides [6]. However, the benefit of NT-3 overexpression in the cochlea has been challenged by a study in which the overexpression was mediated by using (AAV) or adenovirus (Adv) [44]. In this study, subjects receiving the transfection either by AAV or Adv experienced ABR threshold elevations, more with Adv transfection. A decreased synapse count was seen in the subjects receiving Adv, but not AAV. The authors concluded that the elevation of NT-3 levels in the cochlea can disrupt synapses and impair hearing. A comparison between the two studies is impossible because one was done with normal-hearing subjects, while the other was done in subjects deafened by aminoglycosides. Furthermore, neither of the studies provided data for the transfection rate; and the study was done on deafened guinea pigs; no data was reported from a control group [6].

The safety of AAV vectors in cochlear gene transfection has been widely supported in the literature [35, 97, 99] and by our own published work [36, 107]. Recently conducted in our labs, the AAV of serotype 8 that had a surface tyrosine mutation at the residue of 733th amino acid on the capsid (rAAV8-mut773, at the titer of 6.92 × 1013, provided by the Retinal Gene Therapy Group, University of Florida, USA) was applied to transfect NT-3 into the cochlear cells of guinea pigs [8]. Figure 3.4 shows that the transfections of IHCs reached ~100% at the base and spread up to the second turn of the cochlea (10 mm from the apex or 4 kHz region). Therefore, it is a good model to test if the overexpression of NT-3 by AAV could promote synapse regeneration after noise-induced damage. After baseline ABR tests, transfection with AAV-NT-3 was done in one ear of each of the seven guinea pigs, whereas the other ear was given a sham surgery with the injection of the equivalent amount of saline. The ABR was retested 1 week after the transfection surgery. No threshold differences were seen between the ears (Fig. 3.5a). Then the animals were exposed to a high-pass noise with the cutoff at 4 kHz at 105 dB SPL for 2 h to create synaptopathy. A third ABR was administered 2 weeks after the noise exposure, followed by a near-field test of CAP with a round electrode. After the functional tests, the animals were sacrificed, and their cochleae were harvested for morphological evaluation for ribbon synapse counts. Another six animals were recruited as no-noise blank control.

Fig. 3.4
figure 4

AAV-mediated NT-3 transfection. (a) Representative IHCs image from 16 kHz region showing the transfected cells (green) across the cochlea. (b) Transfection cochleogram showing the mean (solid line) and +/-SEM (dashed lines) transfection of IHCs from three cochleae. (This figure is adapted from Chen et al. [8] Gene Therapy 25(4): 251–259)

Full size image
Fig. 3.5
figure 5

Physiological function. (a) ABR threshold of two noise groups tested before surgery and 2 weeks after noise exposure. (b) The click-evoked CAP input-output function. (This figure is adapted from Chen et al. [8] Gene Therapy 25(4): 251–259)

Full size image

Consistent with our previous reports, there was no significant difference in ABR thresholds between the baseline and 2 weeks after the noise exposure (Fig. 3.5a). To evaluate the impact of the synaptic damage on cochlear output, CAP was measured with clicks of different levels (Fig. 3.5b). The ears injected with AAV and saline are labelled as the two noise groups. In both groups, the noise exposure reduced CAP amplitude by more than half, and the input-output (I/O) functions from the noise groups were overlapped. Significant differences among the noise groups were seen in a one-way ANOVA, performed for the maximal CAP amplitude at 90 dB SPL (F = 57.6, P < 0.001). Post hoc tests (Bonferroni method) showed that the differences between the no-noise control group and the two noise groups were significant (control vs noise-NT-3: t = 9.3, p < 0.001; control vs noise-saline t = 9.3, p < 0.001), but not significant between the two noise groups (t = 0.1, p = 1).

To evaluate the synaptic loss induced by the noise, the presynaptic ribbons (CtBP2) and postsynaptic densities (PSDs) were examined in immunohistochemistry (Fig. 3.6). The number of synapses was counted with the puncta of CtBP2s and PSDs that were paired. At each frequency point in each ear, the synapses were counted over eight IHCs to calculate the average synapse density (# of synapses per IHC). The noise-induced synaptic loss was mainly seen in the high-frequency region (>8 kHz, Fig. 3.7a). The effect of the NT-3 overexpression was demonstrated by less synaptic loss in the frequency region between 11.3 and 22.6 kHz (Fig. 3.7a). Over the high-frequency region (>8 kHz), the average synapse densities were 16.4 ± 0.2, 15.2 ± 0.2, and 18.4 ± 0.1 per IHC, for the noise-NT-3 group, noise-saline group, and no-noise control group, respectively. When compared to the no-noise control, this finding resulted in a 17.4% synaptic density reduction in the noise-saline group and a 10.9% reduction in the noise-NT-3 group. Compared between the noise-exposed groups, NT-3 overexpression appeared to reduce the synapse loss by ~38.5% in the high-frequency region. A significant effect of grouping was seen in a one-way ANOVA (F2,477 = 81.3, p < 0.001). The Bonferroni’s post hoc tests revealed the differences between the no-noise control group and the two noise groups (control vs noise-NT-3: t = 7.8, p < 0.001; control vs noise-saline t = 12.6, p < 0.001) and between the two noise groups (t = 4.9, p < 0.001) (Fig. 3.7b).

Fig. 3.6
figure 6

Images of CtBP2 and PSD staining from the three groups at 16 kHz region. (a) (b), and (c) were noise-NT-3, noise-saline, and no-noise control groups, respectively. The dashed lines indicate the outlines of IHCs and their nuclei. Only paired CtBP2 (red) and PSD (green) puncta were counted as synapses

Full size image
Fig. 3.7
figure 7

NT-3 overexpression reduced the noise-induced synaptic loss. (a) The density-frequency curves of paired CtBP2 and PSD puncta. (b) The averaged counts of synapse density in the high-frequency region (between 8 and 32 kHz). One-way ANOVA followed by Bonferroni’s post hoc test revealed the significant differences between the no-noise control group and the two noise groups (not shown) and between the two noise groups. (a) error bar represent mean±SEM, (b) error bar represent mean±SEM. Asterisks indicate a significant level for the comparison between the two noise-exposed ears. ***p < 0.001. (This figure is adapted from Chen et al. [8] Gene Therapy 25(4): 251–259)

Full size image

The nonsignificant difference result for NT-3, seen in the CAP I/O function, is likely due to the frequency range of the click-evoked CAP being biased to low-frequency regions, where no protection in the synapse count was seen. The power spectrum of clicks of 0.1 ms pulses was below 5 kHz. Even with the up-spread of cochlear vibration at a high intensity (90 dB SPL), the auditory nerves with characteristic frequencies higher than 8 kHz are unlikely to be excited. It is interesting to note that a synapse reduction of less than 5% was seen in the low-frequency region, while the CAP amplitude was reduced more than 50%. This suggests that the surviving/repaired synapses are functionally abnormal at this frequency.

In this study, we did not dynamically track the change in synapse counts at different time points after the noise exposure, nor did we compare the change across groups. Therefore, we do not know if the small reduction of synapse counts in the NT-3 overexpressed group resulted from the reduction of the initial loss of the synapses or the promoting effect of NT-3 on the regeneration of the synapses. However, based upon the working mechanism of NT-3 on synapse formation, and the rescue effect of NT-3 observed after noise exposure [92, 96, 108], we hypothesize that the major effect of NT-3 overexpression by AAV transfection in the present study is due to its effect on promoting synapse regeneration.

Based upon the study with the knock-in mouse model, NT-3 from supporting cells appear to be more effective than NT-3 from IHCs for promoting synapse regeneration [105]. In this study, the overexpression of the NT-3 was reportedly not effective at all. However in the present study (accepted), a significant protective effect is seen even though the NT-3 overexpression is limited to only IHCs (Fig. 3.4). While the quantitative comparison is impossible between the two studies due to the use of different species and different techniques for the overexpression, the protective effect in our study may have been limited by the confined transfection mediated by rAAV8-mut773 in the IHCs. We are exploring the use of new AAV that will transfect both the HCs and supporting cells for better protection against noise.

3.5 Conclusion and Future Direction

Gene therapy via cochlear gene transfection is an attractive approach to reduce noise-induced synaptopathy. The significance of this therapy is emphasized by the high probability of exposure to noise that can potentially produce such damage. Since NT-3 in both IHCs and supporting cells contributes to the synaptic regeneration, the AAV vector should be improved to transfect both the IHCs and supporting cells.

More research is needed to understand why synapses to the low-SR ANFs are more sensitive to noise damage. Research is also needed to investigate if there is a dis/reassembly mechanism of ribbons that act adaptively to reduce the traumatic glutamate release in response to intense noise. If this occurs, investigation into how this mechanism is regulated should be pursued. Research on gene therapy should be associated with the mechanisms for the neural transmission across this special synapse. Understanding the mechanisms of noise-induced synaptic damage in association with the working mechanism of ribbon synapses will provide insight toward reducing noise-induced damage and then increasing the amount of repair.