Axon-glia interactions in the Ascending Auditory System

Abstract

The auditory system consists of a peripheral sensory organ (the cochlea) containing specialized mechanosensory cells (hair cells) that convert sound-generated vibrations into action potentials in the auditory nerve. Spike activity in the auditory nerve encodes information regarding the intensity and frequency of sound stimuli, which is transmitted through the ascending neural pathways that reach the auditory cortex. In this review, we focus on the glia and glia-like cells that associate with the neurons in the ascending auditory pathway, and through their interactions with sensory and neuronal cells contribute to the development, maintenance, and modulation of neural circuits and transmission in the auditory system. We also discuss the molecular mechanisms of these interactions, their impact on hearing and on auditory dysfunction associated with pathologies of each cell type.

Introduction

The vertebrate auditory system is the part of the nervous system dedicated to capturing environmental sounds and processing them to extract information about the surroundings and allow for communication between individuals. In mammals, the auditory pathway starts with the organ of Corti located within the cochlea of the inner ear (Fig. 1A). There, inner hair cells (IHCs) transduce sound waves into neural activity through excitatory synapses onto the primary auditory neurons, the type I spiral ganglion neurons (SGNs). SGNs propagate auditory information to the auditory cortex through an ascending pathway containing several relay stations. At every single stage of the pathway, neural cells are in close contact with glia, including the glia-like supporting cells (SuppCs) of the organ of Corti that engulf hair cells (HCs) and their synapses with SGNs, Schwann cells that myelinate the peripheral portion of the auditory nerve (AN), satellite glial cells that engulf SGN cell bodies, oligodendrocytes (OLs) that myelinate the central portion of the AN as well as axons in central auditory pathways, and astrocytes that wrap around synapses in the central nervous system (CNS; (Fig. 1B). In this review, we focus on the interactions these glia have with sensory hair cells and ascending auditory neurons, as well as their roles in the development, maintenance, and modulation of neural circuits and transmission in the pathway. We also discuss our current understanding of the reciprocal neural-glial interactions that influence development and maintenance of hearing and review the auditory dysfunction associated with pathologies of each cell type.

Figure 1. The auditory pathway.

(A) Schematic representation of the auditory periphery, including glial cells associated with SGNs (supporting cells, Schwann cells, satellite glial cells, oligodendrocytes). Oligodendrocyte myelination of SGN axons begins at the CNS–PNS transition zone (arrows). (B) Schematic representation of the central auditory pathway highlighting ITD and IID circuitry at the level of the pons. VCN SBCs (blue) extend excitatory projections to the LSO and MSO in the SOC. VCN GBCs (green) extend excitatory projections to the MNTB, which then send inhibitory projections (red) to the LSO and MSO. Auditory information then passes through several subsequent nuclei including the IC before reaching the MGN and projecting to the auditory cortex. (C) Auditory system function can be measured using auditory brainstem responses (ABR). This method uses electrodes to detect evoked electrical potentials along the early auditory pathway, from the cochlea through the midbrain. The summating potential (SP) reflects activation of inner hair cells, peak I corresponds to evoked potentials in SGNs while later peaks generally correspond to evoked potentials in CN (peak II), SOC (peak III), and IC (peaks IV and V) in the mouse (Henry 1979; Land et al 2016). The magnitude of peak I amplitudes correlates with the number and synchronous firing rate of the SGN fibers. (D) Calyx of held synapse in MNTB is formed by a highly fenestrated calyx ending of the presynaptic GBC axon and the postsynaptic soma of the MNTB principal cell (principal cell projections not shown). Astrocytes are in close proximity to the principal cell soma and extend processes into the spaces between calyceal elements (Ford et al 2009; Reyes-Haro et al 2010). Figure 1B is adapted from Cope et al. (2015). Figure 1C is adapted from Kohrman et al. (2019). AVCN: anteroventral cochlear nucleus; LSO: lateral superior olive; MSO: medial superior olive; GBC: globular bushy cell; SBC: spherical bushy cell; SOC: superior olivary complex; MNTB: medial nucleus of the trapezoid body; MGN: medial geniculate nucleus; SGN: spiral ganglion neuron.

Supporting cells, the glia of the organ of Corti.

The sensory epithelium of the cochlea, the organ of Corti, contains two types of mechanoreceptive HCs - IHC and outer (OHC) hair cells. IHCs are the principal mechanoreceptor cells for sound detection, while OHCs are electromotile cells that amplify the vibrational motion of the organ of Corti and control the gain of auditory responses (Fettiplace 2017). IHC form ribbon-type glutamatergic synapses with the peripheral axons of type I SGNs, bipolar neurons which give rise to approximately ninety percent of myelinated axons in both the central and peripheral parts of the AN. Specialized exocytosis and endocytosis machinery present at the presynaptic ribbon structures support the high temporal fidelity of sound responses at this synapse (Moser, Grabner et al. 2019). The remaining ten percent of AN myelinated axons belong to brainstem efferent neurons (Ehret 1979, Brown 1987, Campbell and Henson 1988). Each type I SGN innervates a single IHC, while each IHC makes synaptic contact with approximately 10 to 30 type I SGN fibers (Coate, Scott et al. 2019). The AN also contains unmyelinated axons that belong to type II SGNs, smaller caliber afferent neurons that innervate OHC. While the roles of type II afferents remain poorly understood, it has been proposed that they signal cochlear damage (Liu, Glowatzki et al. 2015).

SuppCs in the organ of Corti were traditionally considered to be connective tissue providing structural and homeostatic support. However, recent work has shown that these cells exhibit many characteristics of glia and play glia-like roles (Wan, Corfas et al. 2013). The IHCs and their synapses are surrounded by SuppCs known as inner border and inner phalangeal cells, which express molecules that regulate synaptic development and may function in a manner analogous to that of perisynaptic Schwann cells (Wan, Corfas et al. 2013). Below we discuss the diverse roles that SCs play in development and function of the cochlea.

Spontaneous cochlear activity prior to the onset of hearing.

While sound evoked responses are not present in early postnatal cochlea of rodents, type I SGNs display spontaneous electrical activity (Walsh and Romand 1992, Jones, Leake et al. 2007). Although IHC intrinsic firing properties likely contribute to early spontaneous activity of SGNs (Johnson, Eckrich et al. 2011, Johnson, Kennedy et al. 2012), several studies indicate a role for SuppCs in modulating this spontaneous activity via ATP-dependent signaling. Beginning at P0, ATP released from SuppCs within the developing cochlea activates purinergic receptors on IHCs, leading to IHC depolarization and glutamate release at ribbon synapses, which in turn triggers coordinated bursts of action potentials (APs) in the AN (Tritsch, Yi et al. 2007, Tritsch and Bergles 2010, Tritsch, Zhang et al. 2010). The released ATP also activates purinergic Schwann cell autoreceptors, which potentiates IHC depolarization via K+ efflux from SCs (Wang, Lin et al. 2015, Babola, Kersbergen et al. 2020). Later in postnatal development, ATP release declines along with spontaneous activity levels in SGNs (Tritsch and Bergles 2010). Spontaneous activity is important for the maturation of auditory neurons (Seal, Akil et al. 2008, Shrestha, Chia et al. 2018, Sun, Babola et al. 2018) as well as refinement of synaptic connections in central auditory pathways (Erazo-Fischer, Striessnig et al. 2007, McKay and Oleskevich 2007, Clause, Kim et al. 2014).

Neuronal survival and outgrowth.

HCs have traditionally been considered critical for SGN survival, at least in part due to their expression of neurotrophins such as BDNF and NT3 (Green, Bailey et al. 2012). Indeed, mice carrying null mutations in either BDNF or NT3, or their cognate receptors trkB or trkC, result in embryonic loss of SGNs, with NT3 KOs exhibiting virtually complete loss of type I cells (Ernfors, Van De Water et al. 1995, Schimmang, Minichiello et al. 1995, Bianchi, Conover et al. 1996, Fariñas, Jones et al. 2001). Nonetheless, a large body of evidence suggests that cochlear SuppCs play an equally important role in SGN biology. SuppCs are also a major source of BDNF and NT3 in the cochlea (Wiechers, Gestwa et al. 1999, Sugawara, Murtie et al. 2007). Additionally, data from aged or drug-treated animal models indicate a strong association between SGN survival and the number of remaining SuppCs, rather than hair cells (Leake and Hradek 1988, Xu, Shepherd et al. 1993, McFadden, Ding et al. 2004, Sugawara, Corfas et al. 2005). Notably, experimental ablation of IHCs in two independent mouse model systems, both of which left SuppCs intact, demonstrated survival of SGN for at least 3 to 4 months post-ablation, suggesting that the intact SuppCs were capable of providing neurotrophic support in the absence of HCs.(Zilberstein, Liberman et al. 2012, Tong, Strong et al. 2015). Several studies of human temporal bones have also reported significant positive associations between SuppC and SGN survival in the cochlea (Johnsson 1974, Suzuka and Schuknecht 1988). Finally, developmental expression gradients of BDNF and NT3 observed along the length of the cochlea in rodents, including in SuppCs, is likely to influence the extension and refinement of fibers toward target HCs (Wiechers, Gestwa et al. 1999, Huang, Thorne et al. 2007, Sugawara, Murtie et al. 2007). These neurotrophin gradients may also modulate synaptic and ion channel protein expression in SGNs and thereby contribute to differences in firing properties in basal versus more apical SGNs (Adamson, Reid et al. 2002, Zhou, Liu et al. 2005, Flores-Otero, Xue et al. 2007).

The ability of SuppCs to impact SGN survival also appears to depend on Neuregulin-ErbBR signaling as blockade of SuppC ErbBR signaling results in SGN loss in spite of the presence of intact HCs and SuppCs (Stankovic, Rio et al. 2004). Although SGN loss in this model was originally attributed to an associated decrease in cochlear NT3 expression, more recent work using conditional knockout mice demonstrated normal SGN survival following postnatal loss of NT3 or BDNF in either HCs or SCs (Wan, Gómez-Casati et al. 2014). These results suggest that residual neurotrophin expression from either cell type is sufficient for maintenance of SGNs. Other trophic factors that are expressed in the postnatal cochlea, such as insulin-like growth factor-1 (IGF1) or macrophage migration inhibitory factor (MIF), may also contribute to neuronal preservation, as knockout mice for either IGF1 or MIF exhibit SGN loss or altered innervation (Camarero, Avendaño et al. 2001, Bank, Bianchi et al. 2012).

Regulation of synapse formation and regeneration.

While postnatal NT3 expression by SuppCs is not required for SGN survival, it is nevertheless critical for normal IHC-AN synaptogenesis. Neonatal knockout of NT3 in cochlear SuppCs results in a significant decrease in the number of IHC ribbon synapses and impaired hearing sensitivity (Wan, Gómez-Casati et al. 2014). Conversely, overexpression of NT3 in cochlear SuppCs during the neonatal period results in increases in the number of synapses and better hearing sensitivity. In adult mice, SuppC overexpression of NT3 promotes regeneration of ribbon synapses and functional recovery after noise overstimulation (Wan, Gómez-Casati et al. 2014). These results have suggested the therapeutic potential of neurotrophins for repair of IHC-AN synapses following ototoxic damage. Indeed, local delivery of NT3 into the cochlea was able to reduce synapse loss following moderate noise exposure (Suzuki, Corfas et al. 2016). The efficacy of NT3 for synapse protection/repair, however, appears to be sensitive to dosage and/or location of expression. Although overexpression of NT3 in IHCs via delivery of adeno-associated virus prior to noise exposure was able to protect synapses, delivery following noise was unable to reproduce the protective effect of direct NT3 injection (Hashimoto, Hickman et al. 2019).

Regulation of neurotransmission.

Sound evoked synaptic transmission occurs at IHC–SGN ribbon synapses through release of glutamate and activation of AMPA and NMDA receptors on the postsynaptic terminal (Fuchs, Glowatzki et al. 2003). Glutamate accumulation and overactivation of glutamate receptors can degrade the fidelity of glutamatergic signaling (Barbour and Häusser 1997) and potentially lead to excitotoxicity-induced withdrawal of the axon terminal and cell death of SGNs (Pujol and Puel 1999, Jäger, Goiny et al. 2000, Oestreicher, Wolfgang et al. 2002, Duan, Chen et al. 2006, Kostandy 2012). Similar to traditional glial cell-neuronal interactions that act to replenish glutamate neurotransmitter levels, the glutamate-glutamine cycle also occurs in the organ of Corti. SuppCs mediate glutamate removal at the IHC afferent synapse through the activity of glutamate/aspartate transporter (GLAST, also known as EAAT1). Glutamate uptake is then followed by conversion of glutamate to glutamine by glutamine synthetase (Usami, Osen et al. 1992). GLAST expression in SuppCs is closely matched to the amount of glutamate released (Furness and Lawton 2003) and the time course of GLAST expression in SuppCs correlates with the electrophysiological onset and maturation of mouse auditory function (Jin, Kikuchi et al. 2003). Notably, GLAST-deficient mice experience pronounced swelling of afferent terminals and permanent hearing loss after noise exposure (Hakuba, Koga et al. 2000, Glowatzki, Cheng et al. 2006).

Regulation of organ of Corti’s ionic environment.

Recycling mechanisms are required in the cochlea to maintain ion homeostasis, and SuppCs are key players in these pathways. For example, potassium ions (K+) that flow through mechanotransduction channels in HCs during sound stimulation are released through KCNQ4 channels in the HC basolateral membranes and K+ transporter proteins are required for further recycling (Zdebik, Wangemann et al. 2009). Expression of the K/Cl (K-Cl) co-transporter protein KCC4 is restricted to SuppCs in the mature cochlea (Boettger, Hübner et al. 2002). Mice with deletion of the Slc12a7 gene that encodes KCC4 exhibit early progressive degeneration of sensory and SuppCs as well as SGN, leading to profound hearing loss and consistent with a critical role for KCC4 in uptake of K+ into SCs (Boettger, Hübner et al. 2002).

All cochlear SuppCs are connected by gap junctions that permit small molecule exchange, and this is believed to impact potassium ion homeostasis in the inner ear (Zdebik, Wangemann et al. 2009). Gap junctions are formed by connexin protein hemichannel complexes on adjacent cells. Cx26 and Cx30 are the major connexin proteins expressed in cochlear SuppCs and mutations in the genes encoding these isoforms have been associated with hearing loss in mice and in humans (Mammano 2018). Most prominently, mutations in the human connexin 26 gene (GJB2) account for up to 50% of autosomal recessive non-syndromic deafness in many countries (Chan and Chang 2014). Potassium ion homeostasis, however, is unlikely to be the sole critical function for connexins in the cochlea, as several known GJB2 mutations that are associated with hearing loss appear to form channels capable of small ion passage (Wingard and Zhao 2015). Connexins also mediate passage of larger molecules such as ATP and IP3, both of which have been linked to intercellular Ca²⁺ signaling in the organ of Corti (Mammano and Bortolozzi 2018). SuppCs release ATP through apical connexin hemichannels, which results in purinergic receptor activation on SuppCs and leads to both the influx of external Ca²⁺ and release of internal Ca²⁺ stores. ATP release from SuppCs has been implicated in the regulation of spontaneous activity in the early postnatal cochlea (see above). Intercellular passage of the second messenger IP3 also occurs through gap junctions, which similarly activates Ca²⁺ release from internal stores (Mammano and Bortolozzi 2018). The importance of this pathway is supported by the identification of a human deafness-associated variant in Cx30 that retains permeability to small ions but is deficient in intercellular IP3 transfer and the associated Ca²⁺ response (Schütz, Scimemi et al. 2010). In addition, reduction of connexin-dependent Ca²⁺ signaling has been associated with an immature IHC phenotype, consistent with a role for SuppCs in regulating early sensory cell development (Johnson, Ceriani et al. 2017).

Schwann cells

Schwann cells are derived from precursor cells in the neural crest, which migrate to the periphery during embryogenesis, proliferate, and adopt one of two general fates: myelinating or non-myelinating Schwann cells, both of which engage developing axons (Salzer 2015). Myelinating Schwann cells elaborate specialized membranes that wrap single axons and form the myelin sheath. Non-myelinating Schwann cells typically bundle multiple smaller caliber axons, often at or near nerve terminals. A number of the extracellular signaling cues and intrinsic transcriptional pathways have been demonstrated to guide these steps from neural crest progenitor to mature Schwann cell in the PNS (Stolt and Michael 2016). The myelin sheath supports saltatory conduction of neural signals by regulating the assembly and maintenance of neuronal domains that are required for precise localization of voltage gated sodium and potassium channels at nodes of Ranvier (NoR) and at terminal heminodes (Rasband 2016).

Schwann cell development in the cochlea.

Schwann cell development occurs across an extended developmental time period in the vertebrate cochlea. In humans, immature Schwann cells are associated with AN axons by week 9 of gestation (Locher, Groot et al. 2014). Myelin formation, as evidenced by myelin basic protein (MBP) staining, are present by week 22 in the central segments of AN, consistent with earlier electron microscopy studies indicating myelin formation first in central segments that progresses to peripheral segments between weeks 22 and 24 (Lavigne-Rebillard and Pujol 1988, Ray, Roy et al. 2005). In the mouse, Schwann cell progenitors arrive in the developing inner ear by embyronic day 10.5 (E10.5), when neural progenitor cells begin to differentiate into SGNs (Breuskin, Bodson et al. 2010). In rodents, myelination by Schwann cells initiates near birth at the cell body of SGNs and progresses centrally toward the PNS-CNS transition zone and peripherally toward the axon terminals between P8 and P10, just days prior to the onset of hearing (Anniko 1983, Wang, Zhang et al. 2013). Immunolocalization studies in rat indicate that, like in other peripheral nerves, maturation of AN myelin continues throughout the first postnatal month, with progressively tighter clustering of sodium and potassium channels, along with associated proteins, at NoRs and terminal heminodes (Kim and Rutherford 2016).

Regulation of SGN migration, maturation and survival.

Several studies suggest that pre-myelinating Schwann cells produce a ‘stop signal’ required for appropriate SGN migration and fiber extension in the cochlea, and that they might contribute to SGN survival during development. Morris and coworkers found that in mice with loss of ErbB2, a receptor critical for Schwann cell development, SGN cell bodies migrate to unusual positions in the cochlea and their processes overshoot the sensory epithelia during early development (Morris, Maklad et al. 2006). Similarly, in mice that lack Schwann cells due to Sox10 deletion in neural crest cells, SGN fibers extend peripherally to the lateral part of the cochlea, past their normal IHC targets (Mao, Reiprich et al. 2014). Interestingly, while ErbB2 KO mice have significant SGN loss, that is not the case in the Sox10 KO, suggesting that Schwann cells are not essential for SGN survival during this developmental stage, and that the neuronal loss in the ErbB2 KO reflects supporting cell dysfunction (Stankovic, Rio et al. 2004).

There is also evidence that growth factor pathways mediate critical axon-Schwann cell interactions in the AN, including NRG1/ErbB receptor (ErbBR) signaling, which controls myelin thickness in other peripheral nerves (Garratt, Voiculescu et al. 2000, Michailov, Sereda et al. 2004, Chen, Velardez et al. 2006). All SGNs express NRG1 (Morley 1998, Hansen, Vijapurkar et al. 2001, Stankovic, Rio et al. 2004) and AN Schwann cells express ErbBRs (Hansen, Vijapurkar et al. 2001). We have found that loss of ErbBR signaling in myelinating Schwann cells due to expression of a dominant negative ErbBR results in hidden hearing loss (HHL), an auditory neuropathy characterized by normal hearing thresholds but decreases in the amplitude of auditory evoked potentials (Kohrman, Wan et al. 2019, Cassinotti, S. et al. 2020). This dysfunction is associated with deficits in the structure of the terminal heminodes of the AN, consistent with a critical role for the heminodes in ensuring fast, synchronous firing of APs in response to neurotransmitter release from HCs.

Another example is the fibroblast growth factor (FGF) signaling pathway, which has been shown to regulate multiple functions of Schwann cells, including proliferation, differentiation, regeneration of peripheral nerves following injury, and recovery following demyelination (Davis and Stroobant 1990, Grothe and Nikkhah 2001, Hansen, Vijapurkar et al. 2001). SGNs express FGF-1 and FGF-2 (Lin, Koutnouyan et al. 1993, Pirvola, Cao et al. 1995, Silva, Gomide et al. 2005), while Schwann cells and OLs express a number of FGF receptors, including FGFR1, FGFR2, and FGFR3 (Grothe and Nikkhah 2001, Bansal, Miyake et al. 2002). Wang and co-workers showed that inactivation of FGF signaling in myelinating cells resulted in significant loss of SGNs, accompanied by age-related hearing impairment in adulthood, without apparent loss of glia (Wang, Furusho et al. 2009). Although myelination was not quantitatively assessed in this study, apparent decreases in amplitudes of the sound-evoked responses and increases in their latencies are suggestive of subtle myelination defects. However, since the loss of FGF function occurred in both oligodendrocytes and Schwann cells, it remains to be determined which myelinating cell is responsible for the observed phenotypes.

Myelin-related pathologies in the peripheral auditory system.

There is increasing evidence that peripheral myelin disorders can lead to auditory neuropathy, a type of hearing loss where sound detection in the cochlea is normal but transmission of auditory activity to the brain is impaired. This is the case in Guillain-Barre syndrome (GBS), an acute and transient peripheral neuropathy caused by Schwann cell damage due to autoimmune processes (Stathopoulos, Alexopoulos et al. 2015). It has been reported that a portion of GBS patients exhibit abnormal auditory brainstem responses (ABRs), a measure of auditory system function that evaluates synchronous firing along the auditory pathway (Fig. 1C) (Schiff, Cracco et al. 1985, Ropper and Chiappa 1986, Nelson, Gilmore et al. 1988, Ueda and Kuroiwa 2008, Takazawa, Ikeda et al. 2012). Although hearing thresholds gradually recover in some patients, their ABR wave forms show persistent increases in interpeak latencies (Takazawa, Ikeda et al. 2012), consistent with permanent myelin deficits. Using a genetic approach to ablate myelinating Schwann cells in mice, we recently demonstrated that transient AN demyelination leads to permanent HHL that is associated with a disruption of the terminal AN heminodes (Wan and Corfas 2017). These defects are consistent with previous modeling and immunostaining studies that support the heminodes as the initial spike generators in type I SGNs (Rutherford, Chapochnikov et al. 2012, Kim and Rutherford 2016) and indicate that loss of the normal compacted heminode structure significantly impairs synchronized initiation and conduction of APs (Wan and Corfas 2017, Budak, Zochowski et al. 2019). Thus, heminodal disruption may be the underlying cause of the auditory neuropathy observed in some GBS patients.

Another example is demyelinating Charcot-Marie-Tooth (CMT), also known as CMT1, where patients have been reported to present classic auditory neuropathy characterized by normal OHC function but substantial decreases in hearing sensitivity, often with ABR latency alterations (Rance 2005). CMT is a genetically heterogeneous disorder, and the auditory phenotype in CMT patients is quite variable, likely due to the nature of the affected gene, the mutation type, age, and also likely other genetic modifiers (Morelli, Seburn et al. 2017). For example, evaluation of a cohort of children genetically diagnosed with CMT1, mostly with dominant PMP22 duplications (CMT1A), exhibited relatively normal hearing thresholds but decreased speech understanding in background noise (Rance, Ryan et al. 2012). ABR studies demonstrated altered peak latencies consistent with demyelination and prolonged conduction times in the peripheral AN of these patients (Rance, Ryan et al. 2012). Similarly, a large cohort of adult CMT1A patients with normal hearing thresholds was recently found to exhibit HHL, with deficits in speech understanding in noise that were also associated with other deficits in temporal processing of sound (Choi, Seok et al. 2018). On the other hand, a study showed that older CMT1B patients carrying a dominant missense mutation in myelin protein zero (MPZ) exhibited significant threshold increases associated with gross defects in ABR peaks (Starr 2003). Postmortem analysis of the cochlea of one of these patients demonstrated a significant loss of SGNs and demyelination of remaining peripheral fibers, compared with historical age matched controls (Starr 2003).

Our recent studies on mouse models of CMT1 support the notion that peripheral myelin disorders lead to auditory neuropathy with clear specific phenotypes for each mutation (Cassinotti, Ji et al. 2020). CMT1A mice, which carry three copies of the human PMP22 gene (Verhamme, King et al. 2011), exhibit normal hearing thresholds at 4 months of age but decreases in ABR P1 amplitudes, indicating an HHL phenotype. This phenotype is associated with defects in myelin at the terminal heminode of AN. In contrast, Trembler-J mice, which carry a dominant missense mutation in PMP22 and serve as a model of CMT1E, exhibit large threshold elevations at 4 months that are associated with severe demyelination of peripheral AN fibers and loss of SGN. This is consistent with the auditory neuropathy previously reported for Trembler-J mice and with the peripheral demyelination caused by direct ablation of Schwann cells with diphtheria toxin at birth (Zhou, Abbas et al. 1995, Zhou, Assouline et al. 1995). Severe to profound threshold elevations are also observed in adults with similar dominant PMP22 missense mutations (CMT1E)(Boerkoel, Takashima et al. 2002, Kovach, Campbell et al. 2002). These CMT phenotypes highlight the critical role of Schwann cell myelination in neurotransmission in the cochlea and in long term AN survival.

Loss of other Schwann cell proteins, including connexins, also result in peripheral neuropathy and auditory dysfunction. Hearing impairment is observed in a subset of individuals with an X-linked form of demyelinating CMT disease (CMT1X), which is caused by any one of a large number of mutations in the gap junction protein connexin 32 (GJB1)(Kleopa and Scherer 2006). Electrophysiological and pathological studies of peripheral nerves in these patients exhibit evidence of demyelinating neuropathy with prominent axonal degeneration. Cx32 has been implicated in the gap junction-mediated intracellular transfer of small molecules across myelin layers, as well as in hemichannel passage of ATP from activated neurons, and is expressed in both Schwann cells and OLs (Bortolozzi 2018). Alterations in ABR P1 peaks as well as central peaks have been observed in different CMT1X patients, consistent with a role for Cx32 in both the peripheral and central auditory systems (Nicholson and Corbett 1996, Bähr, Andres et al. 1999, Giuliani, Holte et al. 2019). Although not associated with general neuropathy, loss of the gap junction protein connexin 29 in mice causes AN myelin defects associated with a delay in the maturation of auditory thresholds and the prolongation of auditory latencies (Tang, Zhang et al. 2006).

Acquired and age-related hearing loss.

While noise overexposure and some ototoxic drugs have direct damaging effects on HCs, these traumas have also been associated with alterations in Schwann cells and myelin structure. Studies in young rats demonstrated reduced AN myelin thickness and changes in nodal and paranodal dimensions within 4 days of exposure to noise sufficient to cause HC damage and permanent threshold shifts (110 dB SPL at 14.8 kHz for 3hrs/day X 3 days (Tagoe, Barker et al. 2014). Similarly, high intensity noise exposure to young adult mice (8 –16 kHz at 106 or 112 dB SPL for 2 h) induced disruption of AN myelin lamellae and heminodal structures within 1 day of exposure, also suggesting direct effects of noise on myelin structure (Panganiban, Barth et al. 2018). Myelin pathologies in AN have been observed in other animal models following either prolonged low level or acute intense sounds, yet in most cases the effects on myelin were evaluated at later time periods following substantial loss of SGN and thus hampering interpretation (Rossi, Robecchi et al. 1976, Coyat, Cazevieille et al. 2019).

Myelin defects have also been noted in the cochlea within days following exposure to chemotherapeutic compounds such as cisplatin and carboplatin in animal models (Ding, McFadden et al. 2002, van Ruijven, de Groot et al. 2004), consistent with Schwann cell toxicity observed in other peripheral nerves (Carozzi, Canta et al. 2015). In addition to HCs, aminoglycoside antibiotic toxicity may directly target SGNs and peripheral fibers in the cochlea soon after local or systemic delivery, with residual AN fibers often lacking myelin sheaths (Dodson and Mohuiddin 2000, Jiang, Karasawa et al. 2017, Wise, Pujol et al. 2017). Finally, myelin deficits have also been associated with aging in animal models, including thinning and degeneration of myelin surrounding AN fibers in the cochlea, which correlated with decreases in MBP levels and in ABR peak I amplitudes (Xing, Samuvel et al. 2012). Similar age-associated decreases in MBP levels were also observed in AN from human temporal bone samples (Xing, Samuvel et al. 2012), suggesting myelin loss could contribute to the temporal processing abnormalities described in aging humans (Plack, Barker et al. 2014, Harris and Dubno 2017).

Satellite glial cells.

Like most peripheral neurons, SGN cell bodies in the cochlea are enclosed by satellite glial cells (SGCs). Remarkably, and distinct from all other peripheral ganglia, SGCs create a thin myelin sheath around inner ear primary auditory neurons in most vertebrates (Felix 2002). Notably, this is not the case in humans (Tylstedt, Kinnefors et al. 1997, Liu, Edin et al. 2015). The mechanisms that mediate this cell body myelination and its physiological impact remain unknown, but it has been speculated that it might underlie differences in AN conduction properties between humans and other species (Rattay, Lutter et al. 2001, Rattay, Potrusil et al. 2013). Several studies have provided evidence that SGCs are important for proper auditory function. SGCs express glutamine synthetase, which is likely to be important for metabolizing excess glutamate and preventing excitotoxicity (Eybalin, Norenberg et al. 1996). In addition, deletion of the Saposin B (Sap B) gene, which plays an essential role in the regulation of myelin lipid biosynthesis, results in selective loss of cochlear SGCs in mice and is associated with progressive degeneration of SGNs and hearing loss, implicating SGCs in AN maintenance (Akil, Sun et al. 2015). In line with the regenerative ability of SGCs in other sensory ganglia following nerve injury (Elson, Simmons et al. 2004, Donegan, Kernisant et al. 2013, Nascimento, Castro-Lopes et al. 2014), SGCs in the cochlea also regenerate following genetic ablation (Wan and Corfas 2017).

Oligodendrocytes

Oligodendrocytes (OLs) derive from neuroepithelial precursor cells (OPCs) that originate in the ventricular zone of the developing CNS. Intrinsic transcriptional pathways together with extracellular signaling cues guide the migration, proliferation and differentiation of OPCs into mature, myelin-producing OLs (Elbaz and Popko 2019). In contrast to the single nerve fibers wrapped by mature Schwann cells in the periphery, myelinating OLs send out cellular extensions that are capable of wrapping and myelinating multiple CNS axons.

Oligodendrocyte development in the auditory CNS.

Myelination of the central projection of AN and of auditory brainstem axons by OLs initiates near the time of hearing onset in rodents and humans. At birth, OPCs are present in both the AN (Bojrab, Zhang et al. 2017) and the auditory hindbrain (Dinh, Koppel et al. 2014) of mice, with myelination of the central projections of AN first being detectable by P8 near the PNS-CNS transitional zone (Wang, Zhang et al. 2013) (Fig. 1A, B). Myelin formation in the auditory hindbrain and cortex initiates approximately 1 week later (Hackett, Guo et al. 2015, Kolson, Wan et al. 2015). Hearing onset in humans occurs between 24 to 25 weeks of gestation (Birnholz and Benacerraf 1983), with myelination of the central projection of AN as well as the auditory brainstem and midbrain initiating soon after, at 26 to 29 weeks (Moore, Perazzo et al. 1995, Moore and Linthicum 2001). In both species, myelin maturation continues in the auditory CNS, coinciding with further refinement of auditory responses. In the auditory brainstem of mice during the first postnatal month, myelin thickness and axon diameter double in size, coinciding with a doubling of conduction velocity (Sinclair, Fischl et al. 2017). Similarly, in the auditory brainstem and midbrain neurons of humans, increases in myelination are observed up to 1 year of age (Moore, Perazzo et al. 1995). In auditory cortex, myelin signals detected by MRI rapidly increase until 1.5 years of age then progress at a slower rate until adulthood (Su, Kuan et al. 2008).

Oligodendrocyte function in the auditory CNS.

The precise coding and maintenance of timing information of sound stimuli are critical for normal auditory tasks, including spatial hearing and sound localization. Mammals and birds localize sounds in the horizontal plane by comparing the arrival times and intensities through auditory pathways originating at each ear (Grothe and Pecka 2014). Humans are able to discriminate interaural time differences (ITD) as small as 10 to 20 microseconds (Klumpp and Eady 1956) and interaural intensity differences (IID) down to 1 dB SPL (Mills 1960). In mammals, neurons in the superior olivary complex (SOC) integrate sound input from each ear. Medial superior olive (MSO) neurons in the SOC compare the timing of excitatory input from spherical bushy cells in the ipsilateral and contralateral cochlear nuclei (Stotler 1953, Warr 1966, Lindsey 1975, Cant and Casseday 1986) and fire when these binaural inputs arrive within a short time window (Goldberg and Brown 1969, Yin and Chan 1990). In addition to this excitatory input, MSO neurons also receive inhibitory input, most importantly from principal neurons in the medial nucleus of the trapezoid body (MNTB), which fire in response to excitation from single axons of globular bushy cells (GBC) in the contralateral ventral cochlear nucleus (VCN). This contralateral information is transmitted through a large, specialized synapse known as the calyx of Held directly onto the cell bodies of the MNTB principal neurons (Fig. 1D). The fast and precise nature of this synapse is critical for sound processing in the auditory hindbrain (Baydyuk, Xu et al. 2016). Similar integration of excitatory and inhibitory binaural inputs also occurs in lateral superior olive (LSO) neurons in the SOC to detect IID (Grothe and Pecka 2014).

Evidence from a number of model systems suggests that specialized properties of OL-based myelination and of associated axons play a key role in fine tuning the timing of neural transmission in the auditory CNS. Comparisons of intra- and contralateral pathways involved in sound localization suggest that these specializations are likely to influence conduction speeds to compensate for inherent differences in fiber projection lengths and, in the inhibitory pathways, the differences in synaptic connection numbers between pathways. For instance, in the ITD pathway of gerbils, projections from spherical bushy cells to the contralateral MSO exhibit longer internodes and increased axon diameters relative to those in the shorter ipsilateral projections and are expected to elevate conduction velocity and compensate for the longer contralateral projection distance (Seidl and Rubel 2016). Systematic differences in internode lengths and axon diameters that are associated with conduction velocity variation have also been demonstrated in other neural pathways in mammalian sound localization circuits (Ford, Alexandrova et al. 2015). Notably, a recent comparative study between gerbils, which depend upon ITD detection for localization of low frequency sounds, and mice, which rely instead upon IID, demonstrated unique structural specializations of axons and myelin in the localization circuit of each species that correlated with ITD requirements (Stange-Marten, Nabel et al. 2017).

The importance of myelination for temporal processing of sound information in the CNS is highlighted by electrophysiological defects found in Long Evans shaker rats, which carry an insertional mutation of the MBP gene and exhibit little or no condensed myelin in CNS neurons (O’Connor, Goetz et al. 1999). Increased peak I to V latencies and decreased peak amplitudes of ABRs to click sound stimuli in the MBP mutants indicated an increased central conduction time and lack of neural synchrony in the auditory hindbrain (Kim, Renden et al. 2013, Kim, Turkington et al. 2013). In addition, brain slice recordings in the mutants indicated delayed and less reliable AP generation in MNTB principal neurons following high frequency synaptic activation at the calyx of Held. The delays and loss of AP fidelity were associated with altered Na⁺ and K⁺ channel distribution at NoR and the last heminodes of presynaptic GBC axons (Berret, Kim et al. 2016). These results support previous studies indicating the requirement of normal myelin structure for tight clustering of ion channels at nodes and heminodes in the CNS (Rasband, Peles et al. 1999, Susuki, Chang et al. 2013) and demonstrate its critical role for high fidelity transmission along axons and at nerve terminals in the auditory hindbrain.

Myelin plasticity in the auditory CNS.

A large body of evidence supports a role for neuronal activity and experience in modulating myelin structure in the CNS during development and in the mature organism through multiple mechanisms (Almeida and Lyons 2017, Monje 2018). Many studies have demonstrated that such activity-dependent myelin plasticity impacts behaviors and cognitive function. The coincident timing of myelination and development of auditory processing capabilities, including language acquisition in humans, has suggested that auditory experience impacts myelination to refine sensation during auditory system maturation.

Imaging studies in humans support a relationship between sensory-evoked activity and myelin levels in the auditory CNS. MRI studies showed that white matter to gray matter ratios are lower in the Heshl’s gyrus and superior temporal gyrus of congenitally deaf individuals relative to normal hearing controls (Emmorey, Allen et al. 2003, Hribar, Suput et al. 2014). Changes in white matter microstructure in the temporal lobes of deaf subjects relative to hearing controls have also been detected by diffusor tensor imaging (DTI) (Kim, Park et al. 2009, Hribar, Suput et al. 2014). Increasing sensory input, (‘auditory enrichment’), has been associated with enhanced levels of myelination. In rats, age-related decreases in myelin gene expression in the auditory cortex were partially reversed by auditory enrichment in an operant conditioning paradigm (Villers-Sidani, Alzghoul et al. 2010). Similarly, DTI evaluations in humans have demonstrated a positive correlation between more extensive white matter connections in the corpus callosum and both the extent of musical training and the age of training initiation (Bengtsson, Nagy et al. 2005, Steele, Bailey et al. 2013). Using mice, Sinclair et al found that eliciting a hearing loss (50 dB SPL threshold elevation) with ear plugs at the time of normal hearing onset results in thinner myelin wrapping axons projecting to the MNTB of the auditory brainstem (Sinclair, Fischl et al. 2017). Similar deficits were observed following ear plugging of mature mice, suggesting that activity is also important in the maintenance of myelin.

Several mechanisms have been proposed to underlie the impact of neuronal activity and experience on CNS myelination, including impacts on OPC proliferation and differentiation (Foster, Bujalka et al. 2019). Trophic factors expressed by active neurons such as BDNF and NRG1 have been shown to stimulate myelination by OLs (Roy, Murtie et al. 2007, Taveggia, Thaker et al. 2007, Xiao, Wong et al. 2010, Wong, Xiao et al. 2013, Venkatesh, Johung et al. 2015). Likewise, sustained genetic activation in OLs of ERK1/2 kinases, which are part of the intracellular signaling pathway initiated by NRG1- ErbBR engagement (Mei and Nave 2014), results in increased myelin thickness in the CNS and decreased latencies in ABR brainstem peaks (Jeffries, Urbanek et al. 2016). Neurotransmitters such as ATP and glutamate that are released from active neurons may also impact myelination. In vitro studies have indicated an indirect effect of neuronal ATP release via induction of leukemia inhibitory factor in astrocytes, a cytokine with pro-myelinating activity (Ishibashi, Dakin et al. 2006). OPCs express a variety of voltage-gated ion channels and neurotransmitter receptors, including AMPA, NMDA, and GABA receptors (de Biase, Nishiyama et al. 2010). In multiple regions of the CNS, subsets of these progenitors also receive synapse-like input from either glutamatergic excitatory and GABAergic inhibitory neurons (Bergles, Roberts et al. 2000, Lin and Bergles 2003, Ziskin, Nishiyama et al. 2007). Glutamatergic transmission from MNTB neurons to pre-myelinating OL cells (pre-OLs) in the auditory hindbrain has been recently characterized (Berret, Barron et al. 2017). In brain slice recordings from rats during postnatal development, a subset of pre-OLs in the MNTB fired APs that were blocked by glutamatergic antagonists and these APs were dependent upon the voltage-gated sodium channel Nav1.2. Knockdown of Nav1.2 levels using shRNA blocked this sodium current and resulted in altered OL cellular extensions and a decrease in MBP levels, consistent with an impact of activity on pre-OL development and myelination.

In addition to modulating neural transmission through effects on myelination, OL-nerve interactions also appear to impact synaptic plasticity at the calyx of Held synapse in the MNTB via BDNF-trkB signaling (Jang, Gould et al. 2019). Building on prior studies that demonstrated release of BDNF from OLs in response to glutamate in the cortical CNS (Bagayogo and Dreyfus 2009), Jang and co-workers verified BDNF expression by OLs in the MNTB and demonstrated decreased glutamate release from presynaptic neurons in slice preparations from mice that lack BDNF expression in OL precursor cells (Jang, Gould et al. 2019). These conditional Bdnf knockouts also exhibited defects in neural synchrony in the auditory brainstem. The rescue of glutamatergic transmission in the knockouts by local delivery of either BDNF or trkB agonists suggests a mechanism by which BDNF secretion from OLs near the calyx of Held modulates neurotransmitter release from the GBC terminal and supports a modulating role for activity-dependent regulation of presynaptic properties by OLs in the central auditory system.

Oligodendrocyte pathology and hearing impairments.

CNS myelin disorders often affect auditory function, as might be expected given the impact of myelin on conduction velocity and timing in neuronal circuits. Studies of these disorders have provided insights into how myelin impacts processing of auditory signals and the nature of hearing defects that arise due to abnormal CNS myelination. While a number of case reports indicate increases in hearing thresholds in multiple sclerosis (MS) patients that present with lesions in CNS auditory regions (Drulovic, Ribaric-Jankes et al. 1993, Bergamaschi, Romani et al. 1997), larger controlled studies failed to demonstrate a chronic effect of MS on threshold (Doty, Tourbier et al. 2012). Sudden sensorineural hearing loss has also been described in MS patients, but typically resolves over time (Hellmann, Steiner et al. 2011). Although MS does not appear to result in consistent decreases in auditory sensitivity, subtler defects in hearing that involve losses in temporal auditory processing have been associated with the disorder. Common abnormalities in MS patients include increases in ABR I to V peak latencies that are consistent with myelin deficits in the auditory brainstem. MS patients also often exhibit decreased abilities in binaural hearing tasks such as detection of ITDs and IIDs, suggesting impairment of sound localization pathways (Furst and Levine 2015).

Defects in CNS myelin development may also underlie auditory dysfunction. Auditory processing disorder (APD) is a heterogeneous developmental disorder characterized by perceptual problems including deficits in speech comprehension, spatial hearing, and attention. The normal cochlear thresholds often present in individuals diagnosed with APD has suggested that the source of the disorder lies within defects in the central auditory system. Associations between APD and abnormalities in the temporal characteristics of speech-evoked ABRs (Rocha-Muniz, Befi-Lopes et al. 2012, Rocha-Muniz, Befi-Lopes et al. 2014) have led to the hypothesis that decreases in the precision of AP firing and conduction in auditory pathways contribute to perceptual dysfunction (Kopp-Scheinpflug and Tempel 2015). Consistent with this notion, myelin abnormalities have been observed in APD patients in brain regions involved in auditory processing. For example, an increased mean diffusion in auditory cortex was observed in DTI studies of children diagnosed with APD, consistent with a potential disruption of thalamo-cortical communication (Farah, Schmithorst et al. 2014). The same study also found decreased anisotropy in the prefrontal cortex and anterior cingulate, which are areas of the brain closely tied to attention and top–down information processing. This suggests that altered myelination in these structures may underlie difficulties in attending to auditory stimuli in APD.

The transmembrane proteolipid protein 1 (PLP1) is a major constituent of myelin in OL. Loss of function mutations in the human X-linked PLP1 gene generally result in spastic paraplegia, a progressive neuronal degenerative disorder affecting mainly spinocerebellar and lateral corticospinal tracts, while missense mutations with apparent toxic gain of function effects underlie Pelizaeus-Merzbacher Disease (PMD), a more severe leukodystrophy disorder characterized by hypomyelination (Garbern 2007). PMD patients typically have speech difficulties that are associated with normal cochlear function (ABR thresholds and peak 1 amplitudes) but dyssynchrony of later ABR peaks and delayed and/or reduced myelin levels in the auditory brainstem and higher auditory pathways (Kuan, Sano et al. 2009, Coticchia, Roeder et al. 2011, Morlet, Nagao et al. 2018). Defects in CNS myelination may also influence the auditory dysfunction observed in other neurological disorders. For example, many individuals diagnosed with autism spectrum disorder (ASD) exhibit alterations in several aspects of auditory perception including loudness detection and complex speech comprehension (O’Connor 2012). ASD patients also often exhibit reduced detection of temporal features of auditory stimuli such as gaps in sounds (Foss-Feig, Schauder et al. 2017) as well as decreased white matter properties in several brain regions, including the corpus callosum and temporal lobes (Travers, Adluru et al. 2012). Defects in CNS myelin and auditory processing have also been associated with psychiatric disorders such as schizophrenia, bipolar disorder, and chronic depression (Kahkonen, Yamashita et al. 2007, Fields 2008, McLachlan, Phillips et al. 2013, Zenisek, Thaler et al. 2015).

Astrocytes

Astrocytes account for approximately 20-40% of the glial population across different regions in the CNS and play a range of supportive and regulatory roles that contribute to neuronal development, synaptic plasticity and responses to neuronal damage (Bartheld, Bahney et al. 2016). These cells elaborate highly branched processes that interact with neural structures, including synapses, and express ion channels, extracellular matrix proteins, growth factors and cytokines that appear to mediate their influence on neurons (Schiweck, Eickholt et al. 2018). In addition, astrocytes are chemically excitable via expression of a variety of neurotransmitter receptors (Nimmerjahn and Bergles 2015).

Astrocytes development in the central auditory pathway.

Astrocytes have been linked to the postnatal development of the central auditory system (Cramer and Rubel 2016). Immunostaining of astrocyte-associated marker proteins including the intermediate filament glial fibrillary acidic protein (GFAP), the calcium binding protein S100β, and aldehyde dehydrogenase 1 family member L1 (ALDH1L1), revealed the emergence of a diverse pattern of astrocyte cell types in ventral cochlear nucleus (VCN) and MNTB of rodents across the three week postnatal developmental period (Dinh, Koppel et al. 2014, Saliu, Adise et al. 2014). At P14, just after the time of hearing onset and when auditory brainstem pathways are approaching maturity, all astrocyte markers were expressed in the VCN and MNTB, with peak numbers of GFAP-positive processes in both locations while GFAP-positive cell bodies remained outside of these auditory nuclei (Dinh, Koppel et al. 2014). In the MNTB, cell bodies of ALDH1L1-positive and S100β-positive astrocytes were present near principal neurons and calyces of Held, the excitatory terminal of GBC axons on principal neurons, while GFAP-positive processes were often present near the calyces (Dinh, Koppel et al. 2014)(Fig. 1D). This heterogeneous population of astrocytes present during auditory hindbrain maturation has suggested that distinct classes of astrocytes carry out different developmental functions, with early emerging astrocytes participating in neuronal refinement and synaptogenesis and later emerging astrocytes influencing neurotransmission (Dinh, Koppel et al. 2014, Cramer and Rubel 2016).

It has been suggested that interactions between astrocytic processes and the pre- and post-synaptic membranes of the calyx of Held influence the development and function of this critical auditory synapse. Spontaneous calcium transients in astrocytes appear to elicit slow inward currents in post-synaptic principal neurons in early postnatal brain slice recordings in MNTB (Reyes-Haro, Muller et al. 2010). These astrocyte-mediated currents, known as gliotransmission, correlate with astrocytic Ca²⁺ activity and seem to be NMDA receptor-mediated. Glutamate release from the presynaptic GBC neuron mediates fast excitatory synaptic transmission at the calyces of Held. Group II metabotropic glutamate receptors (mGluRs) are expressed in astrocytic processes at the maturing calyces of Held, suggesting that synaptic transmission may also influence astrocyte function during this time (Elezgarai, Bilbao et al. 2001).

Astrocytes play key homeostatic roles in the nervous system, including biosynthetic recycling of the neurotransmitter glutamate at synapses (Verkhratsky, Nedergaard et al. 2015). Studies in the developing postnatal MNTB of rats and gerbils demonstrated that the glutamate transporters GLAST and GLT are exclusively localized in astrocyte processes ensheathing the calyx of Held (Renden, Taschenberger et al. 2005, Ford, Grothe et al. 2009). This is likely to aid in limiting glutamate diffusion between adjacent terminals during this period of structural and functional remodeling of the calyx. Similar to Schwann cells in the cochlea and other glia, glutamate taken up by astrocytes in the MNTB is converted to glutamine; release of glutamine by astrocytes provides a precursor for biosynthetic replenishment of glutamate in adjacent neurons (Uwechue, Marx et al. 2012).

Astrocytes appear to influence neuronal development in the avian auditory hindbrain through release of modulatory factors (Cramer and Rubel 2016). Neurons in nucleus laminaris (NL), which are involved in ITD pathways of sound localization in chicks, display a gradient of dendritic arborization that systematically changes during development of the auditory circuits (Parks and Rubel 1975). Morphological alterations in dendrites coincide with the appearance of GFAP-positive astrocytes, which led researchers to assess the contribution of glial cells to branch order and dendritic morphometric features in NL neurons (Korn, Koppel et al. 2011). Studies in embryonic brainstem slices indicated that conditioned media from cultured embryonic astrocytes promoted the formation of a normal gradient of dendritic morphology in NL neurons, suggesting that astrocytes are an important source of molecules capable of modulating dendritic branching and reorganization during development (Korn, Koppel et al. 2011). Similar brainstem slice experiments also support a role for astrocyte-derived factors in regulating the final numbers of inhibitory synapses on NL neurons, which in turn facilitate more precise binaural coincidence detection (Korn, Koppel et al. 2012).

Astrocyte involvement in auditory system damage.

Glial cells, including astrocytes and microglia, participate in complex adaptive responses to neurological injury and disease, including local inflammation as well as degenerative and neural protective activities (Schiweck, Eickholt et al. 2018). CNS injuries often lead to an increase in reactive astrocytes, which serve to dampen inflammatory responses and increase neuronal viability (Faulkner, Herrmann et al. 2004). Glial responses to injury often produce a 'scar' composed of astrocytes, microglia and associated extracellular matrix material. Such glial scars, while containing the extent of injury, also appear to limit the ability of neural regeneration (Bradbury and Burnside 2019). A number of studies support the involvement of astrocytes and other glia following damage or peripheral sensory loss in the auditory system.

In a model of acute injury by compression of the central portion of the AN in the rat, Sekiya and colleagues observed GFAP-positive reactive astrocytes along with markers of glial scar formation in the vicinity of the injured site, which also extended beyond the PNS-CNS transitional zone boundary into the cochlea (Sekiya, Holley et al. 2015). The injury resulted in partial degeneration of SGNs that was associated with ABR threshold elevations. A regenerative medicine technique based on transplantation of auditory neuroblast cells onto the surface of the injured nerves was used by this group to test for potential re-innervation. The neurons derived from transplanted cells appeared capable of forming glutamatergic synapses with HCs in the organ of Corti and extending processes into the auditory hindbrain (cochlear nucleus). Rats receiving the neuroblasts exhibited modest recovery of central ABR peak thresholds relative to sham-treated controls. The transplanted cells were closely associated with GFAP-positive processes, suggesting that the reactive astrocytes may have provided neural guidance cues, similar to the glia-guided migration of neurons in the developing cortex (Marín and Rubenstein 2003).

Cochlear ablation results in AN degeneration, loss of afferent connections in the cochlear nucleus and additional associated changes in central auditory pathways (Gold and Bajo 2014). Activated glial responses involving astrocytes and microglia have often been observed in the VCN in mammals and birds following loss of AN nerve input, consistent with a damage response (Lurie and Rubel 1994, de Waele, Torres et al. 1996, Janz and Illing 2014). The new calyceal synapses in ipsilateral MNTB following the early postnatal cochlear lesion exhibited closely apposed astrocytes similar to those found during normal development and without apparent increases in cell number, suggesting that pre-existing glial cells are capable of reorganizing at the new calyces (Dinh, Koppel et al. 2014). In contrast to manipulations during development, loss of afferent synapses in VCN following cochlear ablation in the mature animal leads to re-innervation of VCN cells by axon collateral sprouting from VNTB (Kraus and Illing 2004). Molecular studies of astrocyte markers during this reorganization suggest two phases of glial cell responses, one associated with AN and synaptic degeneration and the second associated with reinnervation and new synapse formation (Fredrich, Zeber et al. 2013).

A long-lasting activated glial response has also been observed in the CN of mature rats within days following deafening by loud noise exposure, prior to any evidence of overt AN degeneration (Fuentes-Santamaria, Alvarado et al. 2017). The response was characterized by a large increase in the number of glial processes (astrocytes and microglia) in close apposition with VCN neurons and synapses. Although no overt neuronal loss is likely occurring under these conditions, the glial response may be related to the increase in spontaneous firing rate in CN neurons that is frequently observed following noise trauma (Kaltenbach and Afman 2000, Vogler, Robertson et al. 2011). Glial activation following neonatal deafening also correlated with increased neuronal activity in the inferior colliculus (IC), a central auditory nucleus that receives input from both the periphery and from higher level pathways (Rosskothen-Kuhl, Hildebrandt et al. 2018). The deafened rats exhibited abnormally broad neuronal activity in the IC in response to electrical stimulation, in contrast with the sharp frequency responses observed in normal hearing controls. Glial activation in the IC of the deafened cohort was reflected in substantial hypertrophy of both astrocytes and microglia in the active neuronal regions within 1 day of stimulation and the presence of increased numbers of apposed processes. The response was not associated with any evidence of cell death in the IC and suggests that glial remodeling contributes to auditory system plasticity following sensory loss.

Future directions.

The important role of axon-glia interactions in the auditory system is highlighted by their association with developmental processes and with responses to peripheral hearing loss and neural damage. Attributing functions to particular glial cell types (e.g., SuppCs) has benefited from use of genetic targeting approaches in animal model systems. Use of these approaches, together with related chemo-, opto-genetic and genetically encoded fluorescent indicator strategies, should expand our knowledge of the roles of other glial cell types, such as Schwann cells, astrocytes and microglia, in auditory function (Nimmerjahn and Bergles 2015, Wan and Corfas 2017). As a note of caution, however, a number of studies have underscored the importance of careful consideration of genetic regulatory elements in these approaches, as promoters typically considered specific for particular glial subtypes can be expressed more broadly in both neurons and glia (Wan and Corfas 2017, Brandebura, Morehead et al. 2018). In addition to providing insight into basic processes such as sensory information processing, a more precise understanding of axon-glia interactions in the auditory system is likely to impact development of future regeneration and neural implant treatment strategies for hearing loss.

Abbreviations:

ABR

auditory brainstem response

AN

auditory nerve

AP

action potential

AVCN

anteroventral cochlear nucleus

dB SPL

decibels sound pressure level

ErbBR

ErbB receptor

GBC

globular bushy cell

HC

hair cell

HHL

hidden hearing loss

IHC

inner hair cell

LSO

lateral superior olive

MGN

medial geniculate nucleus

MNTB

medial nucleus of the trapezoid body

MSO

medial superior olive

OHC

outer hair cell

OL

oligodendrocyte

SBC

spherical bushy cell

SuppC

supporting cell

SGC

satellite glial cell

SGN

spiral ganglion neuron

SOC

superior olivary complex