Auditory neuropathy — neural and synaptic mechanisms

Sensorineural hearing impairment is the most common form of hearing loss, and encompasses pathologies of the cochlea and the auditory nerve. Hearing impairment caused by abnormal neural encoding of sound stimuli despite preservation of sensory transduction and amplification by outer hair cells is known as 'auditory neuropathy'. This term was originally coined for a specific type of hearing impairment affecting speech comprehension beyond changes in audibility: patients with this condition report that they “can hear but cannot understand”. This type of hearing impairment can be caused by damage to the sensory inner hair cells (IHCs), IHC ribbon synapses or spiral ganglion neurons. Human genetic and physiological studies, as well as research on animal models, have recently shown that disrupted IHC ribbon synapse function — resulting from genetic alterations that affect presynaptic glutamate loading of synaptic vesicles, Ca2+ influx, or synaptic vesicle exocytosis — leads to hearing impairment termed 'auditory synaptopathy'. Moreover, animal studies have demonstrated that sound overexposure causes excitotoxic loss of IHC ribbon synapses. This mechanism probably contributes to hearing disorders caused by noise exposure or age-related hearing loss. This Review provides an update on recently elucidated sensory, synaptic and neural mechanisms of hearing impairment, their corresponding clinical findings, and discusses current rehabilitation strategies as well as future therapies.

Approximately 360 million people -5% of the world's population -have a disabling hearing impairment 1 .Hearing impairment can lead to social isolation, depression, and a reduction in professional capabilities.Microsurgery -either alone or combined with inno vative implantable hearing aids -successfully tackles conductive hearing loss that arises from disorders of the middle ear.However, sensorineural hearing impairment, which results from disorders of the cochlea and auditory nerve, is more common than conductive hearing loss, and curative pharma cological, gene therapy or stem cell treatments are not yet available for sensorineural hear ing impairments.For such therapies to be developed, an improved under standing of the cellular mechanisms of sensorineural hear ing impairment is essential.Moreover, advanced genetic diagnostics and clinical audiology are the key to identify ing the site of a lesion and to character izing the specific disease mechanism(s) of sensorineural hearing impairment in an individual patient.
Auditory neuropathy is a nosological term that was coined to describe hearing impairment in indivi duals with hereditary motor and sensory neuro pathy 2 .In audi tory neuropathy, the hearing impairment commences down stream from mechano electrical transduction and cochlear amplification of outer hair cells (OHCs).
Recently, human genetic studies and analysis of animal models of auditory neuropathy have elucidated the wide range of disease mechanisms, which include loss of inner hair cells (IHCs) or IHC synapses, impaired synaptic transmission to spiral ganlion neurons (SGNs), and dis rupted propagation of auditory information along the auditory nerve.
In this Review, we provide an update on sensory, syn aptic and neural disease mechanisms in auditory neuro pathy.We summarize recent insights into the function, dysfunction and excitotoxic loss of IHC ribbon synapses and also discuss disease mechanisms primarily affecting spiral ganglion neurons.We then provide a short guide to clinical diagnosis of auditory neuropathy, with a focus on appropriate physiological and psychophysical tests.Finally, we discuss current strategies for hearing rehabil itation in patients with auditory neuropathy and finish by providing an outlook on future therapeutic options to restore hearing.

Impaired sound encoding in auditory neuropathies
For decades, research into hearing loss focused on dys function or loss of sensory hair cells and failure of coch lear ion homeostasis as key mechanisms of sensori neural

Ribbon synapses
Highly specialized synapses between the inner hair cells and spiral ganglion neurons, with an electron-dense structure -the synaptic ribbon -at the presynaptic active zone that mediates neurotransmitter release.

Cochlear microphone potentials
Outer hair cells generate local cochlear potentials that follow the sound stimulus so precisely that they are called 'microphone potentials' .

Otoacoustic emission
Sound generated from within the inner ear that can be measured with a sensitive microphone in the external ear canal to assess outer hair cell function.

Auditory brainstem responses
Evoked potentials in response to repetitive acoustic stimulation that are recorded from scalp EEG electrodes and typically have five peaks, referred to as waves I-V.

Spiral ganglion compound action potential
The first auditory brainstem response peak, wave I, reflects the spiral ganglion compound action potential; this potential can be recorded with better resolution using electrocochleography.

Auditory synaptopathy
Hearing impairment caused by dysfunction or loss of ribbon synapses in the inner hair cells; has been termed auditory synaptopathy and can show clinical findings similar to those described above for auditory neuropathy.
hearing impairment, with less emphasis on neural con duction by the auditory nerve and encoding of sound at the synaptic level.In the 1990s, Arnold Starr and colleagues described sensorineural hearing impairment in indivi duals with spared OHC function, as evidenced by preserved cochlear microphone potentials (CMPs) and otoacoustic emissions (OAEs), but abnormal auditory brain stem responses (ABRs) and spiral ganglion compound action potentials that indicate impaired function of the auditory nerve 2 (FIG.1).On the basis of puretone audio grams, speech recognition was impaired in these patients beyond what would be expected 2 .In addition, the ability of these patients to process temporal cues of acoustic signals was impaired, even though frequency discrimination was preserved 3 ; this temporal encoding deficit impairs sound localization, speech perception and music appreciation.We suggest that the degrada tion of the temporal neural code is the main substrate of the hearing impairment in auditory synaptopathy and neuro pathy (FIG.2).Moreover, individuals with audi tory neuro pathy do not typically derive much benefit from hearing aids owing to impaired function of audi tory nerve, reflecting their reports that making sounds louder does not improve speech comprehension.In the 2000s, the concept of degraded temporal precision in the neural encoding of sound in spiral gan glions became prominent, and the auditory neuropathy phenotype was more widely reported.The wide range of aetiologies that were suggested by genetic and clinical pheno typing led to the introduction of broader clinical classifications, such as 'auditory dyssynchrony' and 'auditory neuro pathy spectrum disorder' .Efforts have been made to reach a consensus on appropriate diagnostics and rehabilitation of hearing [4][5][6] in auditory neuropathy.

Prevalence
Unfortunately, considerable uncertainty prevails regard ing the prevalence of auditory neuropathy.Estimates of how frequently hearing impairment is caused by auditory neuropathy vary substantially, from ≤1% 7,8 to almost 10% 9 of individuals with hearing impairment.Some of this variance is attributed to the inclusion in these studies of populations with different types of hearing impairment.
The reported estimates of the prevalence of genetic auditory neuropathy range from a relatively large prevalence of a single mutation in a specific gene to a single family with one pathogenic mutation in a given gene.For example, the predominant 'Spanish muta tion' (Gln829Ter, resulting in truncation) in OTOF was found in 8% of the congenital nonsyndromic deaf population with autosomal recessive inheritance of Spanish descent, and is the thirdmost common cause for prelingual deafness 10 .By contrast, a mutation in SLC17A8 -which encodes vesicular glutamate trans porter 3 (VGluT3) -has been described in only one large family of Czech descent 11 .Currently, we can only speculate on the extent to which synaptopathic and/or neuropathic disease mechanisms contribute to common forms of hearing impairments, such as noise induced and agerelated hearing loss.However, given the high prevalence of these impairments and the evidence from animal research that synaptopathic mechanisms are involved in noiseinduced and agerelated hear ing loss (discussed below), we should assume auditory neuropathies to have a substantial role.

Sound encoding at the IHC ribbon synapse
Animal research has profoundly advanced our under standing of the disease mechanisms of hearing impair ments, particularly those affecting sound encoding at the hair cell ribbon synapses.Ribbon synapses also exist in the retina 12 ; indeed, human retinal synaptopathies have been shown to cause stationary night blindness [13][14][15] .Moreover, animal models of auditory synaptopathy and neuropathy have shown how disruption of synchro nous afferent signalling in the cochlea can account for clinical observations [16][17][18][19][20][21] .

IHC ribbon synapses have unique properties
The glutamatergic synapse between IHCs and SGNs is highly specialized to achieve indefatigable afferent trans mission at rates of hundreds of Hertz and with sub millisecond temporal precision.Unlike neurons in the CNS, each bipolar SGN receives input from only one pre synaptic active zone of one IHC, which is occupied by an elaborate electrondense specialization, the syn aptic ribbon, that tethers dozens of synaptic vesicles 12,22 (FIG.3).Once the mechanical vibration deflects the hair bundle at the apex of the hair cell, mechanotransducer channels open and the ensuing cation influx generates a depolari zing receptor potential, which triggers Ca 2+ influx through voltagegated Ca 2+ channels at the pre synaptic active zones of the ribbon synapse, driving syn aptic vesicle fusion and glutamate release.Hence, rather than transmitting a binary action potential code, the IHC-SGN ribbon synapse encodes a graded presynaptic signal into a rate code of the SGN.The temporal preci sion of synaptic trans mission is so high that spiketiming in the SGN is locked to the phase of the sound stimulus for frequencies up to approximately 1 kHz 23,24 .
The active zone of the IHC ribbon synapses dif fers from active zones of conventional synapses and even from ribbon synapses of the retina in its molecular composition 25 .Most notably, neither neuronal SNARES 26

Key points
• Auditory neuropathy impairs speech comprehension severely, beyond the extent that would be expected on the basis of increased threshold of audibility • Auditory neuropathy encompasses a range of disease mechanisms that typically disrupt the synaptic encoding and/or neural transmission of auditory information in the cochlea and auditory nerve • Auditory synaptopathy, impaired sound encoding at the synapses between inner hair cells and spiral ganglion neurons, results from genetic defects or insults such as exposure to loud noise • Advanced physiological and psychophysical testing combined with molecular genetic analysis facilitate diagnostics of auditory synaptopathy and neuropathy • Although traditional hearing aids often do not provide substantial benefit for patients with auditory synaptopathy or neuropathy, cochlear implants can provide effective hearing rehabilitation depending on the site(s) of disorder

Organ of Corti
The organ of Corti is the end organ of the sense of hearing that harbours the sensory inner and outer hair cells, as well as afferent and efferent nerve fibres and various types of supporting cells.
(soluble Nethylmaleimidesensitivefactor attachment receptors), SNAREregulators such as complexins 27,28 , nor priming factors of the MUNC13 family 29 seem to operate in mature IHCs.Instead, hair cells employ the multiC 2 domain protein otoferlin 19,29,30 , though oto ferlin is likely to be only the first identified major player in an orchestra of unconventional synaptic proteins reg ulating exocytosis at the IHC synapse.Moreover, unlike in conventional gluta matergic presynapses, vesicular glutamate uptake at the IHC synapse is mediated by VGluT3 31,32 rather than VGluT1 or VGluT2, and Ca 2+ signalling involves the Ca V 1.3 Ltype Ca 2+ channel instead of Ca V 2.1 P/Qtype or Ca V 2.2 Ntype Ca 2+ channels 33,34 .Defects in the genes that encode otoferlin, VGluT3 and the Ca 2+ channel complex cause human auditory synaptopathy (see below).

Properties of SGNs
SGNs are bipolar neurons that are enwrapped by Schwann cells in the cochlea, and by oligodendrocytes where they enter the brainstem.In vivo extracellular recordings of SGN action potentials in response to sound in various model animals have provided important insight into the role of SGNs in sound encoding 35,36 .Each SGN is tuned to a specific frequency that best triggers neural activity.This characteristic frequency corresponds to the tonotopic location of the IHC in the organ of Corti that is innervated by the SGN.Interestingly, even SGNs with near identical characteristic frequencies, which have high likelihood of innervating the same IHC, differ vastly in their rates of spontaneous and soundevoked firing.The origin of this neural diversity is not well understood, but is likely to involve hetero geneous presynaptic and postsynaptic properties among the hair cell ribbon synapses 36 .SGNs exhibit phasic firing, (that is, they fire one or few spikes even in response to longlasting current injections 37,38 ; this feature is probably attributed to hyperpolarizing K + conductances 37 ).SGNs are electri cally 'tight' cells and show large excitatory postsynaptic currents 39 ; single presynaptic release events, possibly even the release of a single vesicle 40 , are sufficient to trigger an action potential 38 .SGN action potentials are most likely generated at the heminode of the peripheral SGN neurite, which is just a few micrometres away from but homogeneous delays of action potential formation, as seen in conductive hearing loss, increase the latency but do not decrease the maximal amplitudes of the SGN population response.Variable delays of action potentials degrade the amplitude of the population response and also increase latency.Loss of activity, for example as a result of loss of SGNs, decreases the amplitude but does not affect the latency and the combination of variable delays and loss of activity produces the worst SGN population responses, if they persist at all. the IHC synapse 38 .Similar to the downstream complete nodes of Ranvier, the heminode that precedes the myelin sheath shows a high density of voltagegated Na + (Na V ) channels 21,41 that, together with passive electrical proper ties of the distal neurite, enable the invading excitatory postsynaptic potentials to readily trigger spike generation in the SGN 38 .This mechanism underlies the high tem poral precision of sound encoding that is required for detecting temporal cues of acoustic stimuli.Impairments of IHC transmitter release, impairment of spike gener ation in or spike propagation along SGNs, and altered synaptic transmission from SGNs to neurons in the cochlear nucleus all degrade the signalling of auditory information to the brain.The resulting loss of tem poral precision (or synchrony) and inaccurate neural representation of the auditory signal (caused by the loss of synapses and/or neurons) are thought to represent key disease mechanisms in auditory synaptopathy and neuropathy (see below).

Disease mechanisms
Here, we review known mechanisms of genetic and acquired auditory synaptopathy and neuropathy, includ ing disruption of IHCs, the SGNs and the ribbon synapse.
We combine information on human genetics and focus on disorders for which analysis of auditory function in animal models has provided insights into the disease mechanism.We will cover genetic disorders related to otoferlin, VGluT3, the Ca 2+ channel complex and OPA1.
Finally, we will review recent findings on how hypoxia, hyperbilirubinaemia, thiamin deficiency, loud noise and ageing affect IHCs, synapses and SGNs, again building on data from animal models and the first observations in humans.

Genetic auditory synaptopathies
Otoferlin genetics, physiology and pathophysiology.OTOF, the gene coding for otoferlin, was among the first genes identified as being affected in autosomal reces sive sensorineural deafness ('deafness, nonsyndromic autosomal recessive 9'; DFNB9) 42 .DFNB9 is a severe to profound prelingual hearing impairment in which none of the symptoms arise from dysfunction of other senses or the brain.It has the audiological signature of auditory neuropathy, and was suspected to reflect a defect of pre synaptic function in inner hair cells 42 soon after its char acterization.During the past decade, rare cases of less severe otoferlinrelated auditory synapt opathies have been reported [43][44][45][46][47] .In some patients with these disorders, an elevation in body temperature (even by only 1 o C) as a result of exercise or fever [43][44][45][46][47] can exacerbate the condi tion into complete deafness accompanied by tinnitus.At normal body temperature, audibility can be close to nor mal in these patients, though their speech recognition is impaired, particularly in background noise.
Otoferlin is expressed in sensory hair cells and belongs to the ferlin family, which also includes the muscle proteins dysferlin and myoferlin.Ferlins have been implicated in membrane fusion events 25,48 and are of great interest to neurologists for reasons beyond their role in otoferlinrelated hearing disorders.Dysferlin, for example, is defective in limbgirdle muscle dystrophy type 2B 49 , which has been linked with impaired mem brane resealing in skeletal muscle cells -a process that requires dysferlindependent fusion of membranous organelles with the plasma membrane 50 .
Like the other ferlins, otoferlin contains a ferlin specific motif of unknown function, a coiledcoil domain that might form a helical bundle with other coiledcoil domains, and six or seven C 2 domains 51 , which generally bind to Ca 2+ and phospholipids and/or interact with other proteins.Several C 2 domains of otoferlin have been implicated in Ca 2+ and phospho lipid binding, though controversy still remains regard ing which of the C 2 domains contribute to binding, and how 19,30,[52][53][54] .The sequence similarity between ferlin C 2 domains and other C 2 domain proteins, such as protein kinase C or synaptotagmin, is low, making it difficult to infer the structural determinants of oto ferlin binding to Ca 2+ and phospholipid from those of better characterized proteins 25 .We are only beginning to obtain insight into the protein structure of ferlins 55,56 .
To date, approximately 90 pathogenic mutations of OTOF have been reported; the majority of these muta tions are assumed to be nonsense or truncation mutations that cause inactivation of otoferlin 57 , result ing in DFNB9type hearing impairment.Otoferlin knockout mice are likely to be an appropriate animal model for DFNB9 (REFS 19,30).Indeed, Ca 2+ triggered exocytosis is nearly abolished in IHCs of these mice 19,30 ; synaptic vesicles were found close to the membrane at the active zone, suggesting that an absence of vesicles was not limiting signal transduction, but a late step of exocytosis was disrupted 30 .This finding, together with biochemical evidence for a role of otoferlin in Ca 2+ and phospholipid binding (see above) support the hypoth esis that otoferlin serves as a Ca 2+ sensor in exocytosis in hair cells, as do synaptotagmins 1 and 2 in conven tional synapses (FIG.4).However, a rescue experiment in which synapto tagmin 1 was expressed in otoferlin deficient IHCs failed to restore IHC exocytosis or hearing 58 .This might reflect the absence of appropri ate interaction partners of synaptotagmin (such as neuronal SNAREs) that are required for its function.Interestingly, otoferlin knockout mice showed a more subtle vestibular dysfunction, both at the hair cell and systems level 59 .This finding is in line with clinical reports of normal vestibular function in patients with DFNB9 (REF.60).
Missense mutations cluster in the regions of DNA that encode the C 2 domains of otoferlin and are likely to impair their function; such mutations can also reduce the otoferlin levels in hair cells 19 .To date, six OTOF mutations have been associated with temperature sensitive deafness [43][44][45][46][47] .The 'pachanga' missense otoferlin mouse mutant, generated in a forward genetic screen 61 , provided important additional insight into the function of otoferlin.In the IHCs of these mutant mice, exocyto sis was normal in response to short stimuli, but much reduced as soon as vesicle replenishment was required at the release sites of the active zone 19 .This defect in vesicle replenishment was shown to nearly abolish sound encoding at the hair cell synapse in vivo.Only the highest sound pressure levels applied at a low stim ulus rate triggered spikes in SGNs at the single neuron level, but population responses of the spiral ganglion (as assessed by ABRs and electrocochleography, which is a powerful diagnostic tool to help identify the site of lesion) remained undetectable.Consistent with a failure in efficient vesicle replenishment for indefati gable transmission, psychophysical and physiological tests revealed progressive failure of afferent signalling (auditory fatigue) in humans with a missense mutation in OTOF 62 .
Although the precise mechanism by which mutation impairs the function of otoferlin at hair cell synapses remains to be elucidated, three interesting hypotheses for the role of otoferlin in vesicle replenishment have been suggested (FIG.4).Otoferlin could prime synaptic vesicles for fusion 19 , potentially via promoting the for mation of short tethers between synaptic vesicles and the membrane of the active zone, as suggested by elec tron tomography 29 , and/or it could facilitate the clear ance of previously exocytosed material from the release site via interaction with endocytic proteins 18,63 .Missense mutations might affect these processes by lower ing the otoferlin levels or by impairing its interactions with other proteins.Indeed, in pachanga mice, the levels of otoferlin were drastically reduced in the IHCs, particu larly at the plasma membrane 19 .This finding could be representative of some, if not all, human otoferlin mis sense mutations.The mechanisms through which ele vated temperature triggers deafness and tinnitus are yet to be explored, but an additional reduction of otoferlin levels might be involved.
Further research on otoferlin is central to under standing the molecular physiology and pathophysiology of synaptic sound encoding.Sitedirected mutagene sis will require more structural information and bio chemistry, as well as efficient mutagenesis.Identification of additional interaction partners of otoferlin will be instrumental in dissecting the molecular machinery at the hair cell ribbon synapse.

Vesicular glutamate transporter 3.
A mutation in locus DFNA25 on 12q21-24 that causes nonsyndro mic autosomal dominant hearing impairment was first described in 2001 (REF.64), and the genetic defect was further characterized 11  A series of studies demonstrated strong expression of and a strict requirement for VGluT3 in afferent synaptic transmission in hair cells of mouse 31,32 and zebrafish 65 (FIG.5).IHCs lacking VGluT3 exhibited normal Ca 2+ influx and synaptic vesicle cycling 32 .However, there was no excitatory afferent synaptic transmission to SGNs 31 , and soundevoked activation of the auditory pathway was absent 31,32 despite normal 32 or only mildly altered 31 presynaptic morphology.The precise disease mechanism through which the mutated VGluT3 protein causes hearing impairment remains to be elucidated.A haploinsufficiency has been postulated, based on reports of individuals with a major deletion in the DFNA25 locus 66 that probably results in a nonfunctional SLC17A8 allele.Alternatively, a gainoffunction mutation could lead to enhanced trans port that might make IHC synapses prone to excito toxic synapse loss (see section on acquired auditory synapt opathies) owing to an increase of the vesicular gluta mate load.Moreover, the question of whether hearing impairment associated with SLC17A8 mutations is non syndromic requires reconsideration, given that hearing impairment along with mental and motor retardation was observed in an individual with a probable SLC17A8 mutation 66 , and mice globally lacking Slc17a8 have seizures, a sign of a generalized CNS disturbance 31 .
Ca 2+ channel complex.A syndromic deafness in humans has been associated with a lossoffunction mutation in the CACNA1D gene, which codes for the poreforming α1 subunit of the Ca V 1.3 Ca 2+ channel 67 that is expressed in hair cells, cardiomyoytes, neurons and neuroendocrine cells.Ca V 1.3 Ca 2+ channels are involved in diastolic depolarization in cardiomyocytes of the sinoatrial node, and mediate more than 90% of the voltagegated Ca 2+ influx in IHCs 33,34 .Accordingly, individuals with the mutation exhibit sinoatrial node dysfunction and deafness (SANND) syndrome 67 .On the basis of animal studies, Ca V 1.3 channels have been implicated in neurotoxicity of dopaminergic substantia nigra neurons 68 and in fear memory 69 .Individuals with SANND did not report neuropsychiatric symptoms, but a detailed assessment is yet to be performed to explore possible CNS effects of Ca V 1.3 abolition 67 and potential endocrine imbalances caused by Ca V 1.3 disruption.
The first mutation to be associated with SANND (insertion of a glycine near the poreencoding region) blocked conduction of the Ca V 1.3 channel, as demon strated in a heterologous expression system 67 .Loss of Ca V 1.3mediated Ca 2+ influx in Ca V 1.3 knockout mice disrupts stimulus-secretion coupling at the hair cell active zone 34 , resulting in deafness 33 .To date, deaf ness associated with mutations in the genes for the auxiliary subunits of the hair cell Ca V 1.3 Ca 2+ channel has not been described.A mouse mutagenesis study revealed that the ß2subunit (Ca V ß2) of this channel is required for hair cell Ca 2+ influx and hearing 70 .The α2δ subunit(s) that are expressed by hair cells remain to be determined.
Interestingly, various genetic retinal disorders result from mutations in genes that code for the subunits of the presynaptic Ca V 1.4 Ca 2+ channel of photo receptors.Such nonsyndromic retinal synaptopathies include con genital stationary night blindness type 2 (REFS 13,14)  and autosomal recessive cone dystrophy 71 .Additional links between hearing impairment and vision disorders resulting from disruption of the Ca 2+ channel complex at ribbon synapses are provided by interaction partners that regulate the abundance and function of the chan nel.In humans, for example, EFhand Ca 2+ binding proteins (CABPs) are affected in congenital stationary night blindness type 2 (attributed to altered CABP4) 15 and autosomal recessive hearing impairment DFNB93 (attributed to altered CABP2) 72 .CABPs are thought to interact and modulate the poreforming α1 subunit of the channel, shifting the voltagedependence of their activation to negative potentials 73 and antagonizing Ca 2+ dependent inactivation 72 .In the retina, lack of CABP4 could impair Ca V 1.4 Ca 2+ channel activation by shifting its operating range outside the physiological receptor potentials of photoreceptors 73 .In DFNB93, a splice site mutation is assumed to induce skipping of exon 6 of CABP2, resulting in a truncated protein that lacks the Cterminal EFhand Ca 2+ binding sites 3 and 4 (REF.72).This truncated protein shows altered Ca 2+ binding and antagonizes Ca 2+ dependent inacti vation of Ca V 1.3 less efficiently.Experiments on Cabp2 mutant mice are needed to fully comprehend the effects of CaBP2 disruption on the hair cell Ca 2+ influx and synaptic sound encoding.

Genetic auditory neuropathies Mutations in optic atrophy 1 (OPA1).
Neuropathic hear ing impairments are common findings among indivi duals with syndromic dominant optic atrophy caused by mutations in the gene optic atrophy 1 (OPA1) 74,75 .Dominant optic atrophy is one of the most frequent genetic optic atrophies, and OPA1 mutations are causal in the majority of the cases 76 .The OPA1 protein is a mitochondria targeted dynaminrelated GTPase 77,78 that is critical for mitochondrial shaping 79 .The majority of the ~200 known human OPA1 mutations (www.mitodyn.org;the MITOchondrial DYNamics variation portal) cause haploinsufficiency via truncation; these mutations are thought to cause nonsyndromic auto somal dominant optic atrophy (DOA) 80 .Missense mutations, proba bly acting via dominant negative action of the mutant protein 75 , are considered to cause syndromic autosomal dominant optic atrophy (DOA+), which often (in 60% of patients) also involves hearing impairment, sensorimotor neuropathy, myopathy and ataxia 75 .DOA+ related hearing impairment often results from the Arg445His missense mutation in OPA1, and typ ically follows the onset of visual impairment with slow progression 75 .Detailed audiological assessment with transtympanic electrocochleography (FIG. 1) indicates normal function of IHCs and OHCs, but a failure of SGN activation in the cochlea, indicative of auditory neuro pathy 74 .On the basis of the specific pattern of remaining neural responses, DOA+ related hearing impairment was postulated to result from degeneration of the peripheral neurite of the SGNs.Indeed, most of the individuals with DOA+ who received a cochlear implant had very good rehabilitation outcomes 74 .To date, no hearing deficits have been reported in Opa1 mutant mice, which carry a splice site mutation that probably causes DOA via haplo insufficiency 81 .Moredetailed investigation of the cellu lar mechanism(s) of OPA1related hearing impairment requires the generation of appropriate mouse models.
Hereditary sensorimotor neuropathies affecting the auditory system.The term auditory neuropathy was originally coined to describe sensorineural hearing impairment in individuals affected by hereditary sensori motor neuro pathies (HSMN) 82 .Rather than providing a compre hensive review of HSMN and hearing impairment, we highlight a few examples for which some mechanis tic information has been obtained from clinical and animal studies.
Several mutations in the MPZ gene, coding for mye lin protein zero in Schwann cells, have been shown to cause auditory neuropathy 83,84 .In one patient, a histo logical analysis of the SGNs in the temporal bone and other sensory nerves demonstrated a consistent demyeli nation and neuronal loss across the nerves 84 .Mutations in the PMP22 gene, including duplication (causing Charcot-Marie-Tooth polyneuropathy type 1a), have been found to cause auditory neuropathy and sensory hearing impairment [85][86][87] . .

DFNB59-related hearing impairments. Mutations in
DFNB59 have been suggested to result in auditory neuro pathy 88 .The protein coded by DFNB59, pejvakin (the name derived from the Persian word for 'echo'), is a mem ber of the gasdermin family and was found in hair cells and neurons of the afferent auditory pathway 61,88 .The interpretation of pevjakin associated hearing impairment as neuropathic was based on pathological synchronized spontaneous otoacoustic emissions (ABRs and OAEs) in both human patients and in a Dfnb59 knockin mouse that reproduces the human Arg183Trp substitution 88 .However, a pejvakin mouse mutant, generated in a ran dom mutagenesis screen, lacked oto acoustic emissions, which together with evidence of pejvakin expression in OHCs and IHCs was interpreted as support for a role of pejvakin in OHCs 61 .Hearing impairment without

Compound action potential
Reflects the synchronized firing of spiral ganglion neurons; assessed by intrameatal or transtympanic electrocochleography.

Glutamate excitotoxicity
Excessive presynaptic glutamate release leading to massive depolarization and subsequent synapse loss.otoacoustic emissions was also observed in human patients with DFNB59 mutations [89][90][91][92] , indicating that DFNB59 mutations primarily or secondarily also affect cochlear amplification.A recent study has identified a role of pev jakin in the proliferation of peroxisomes in hair cells and SGNs in response to loud sound 93 .Peroxisomes proliferate as part of the antioxidant response of these cells during acoustic overstimulation, but deletion of pevjakin in mice abolished this response and was associated with greater oxidative damage in the organ of Corti.

DIAPH3-associated auditory neuropathy.
Autosomal dominant auditory neuropathy 1 (AUNA1) 94,95 is thought to result from overexpression of diaphanous homologue 3 (DIAPH3) 96,97 .Although more work is needed to eluci date the function of DIAPH3 in hair cells and hearing, mouse lines overexpressing Diaph3 in hair cells pro vide some interesting clues about its function: in these mutants, the hair bundle morphology of IHCs, but not of OHCs, was disrupted 97 , underlining the fact that the audiological signature of auditory neuropathy can reveal defects in auditory signalling at a stage as early as IHC transduction.Individuals with AUNA1 have not yet been assessed with electrocochleography.If the hypothesis of defective IHC transduction were correct, electrocochleo graphy would reveal impairment or absence of the sum mating potential, which reflects the summed receptor potential of IHCs.Finally, syndromic neuro pathies, such as Friedreich ataxia, also show the clinical features of an auditory neuropathy 98 .

Acquired auditory synaptopathies
Hearing impairment in preterm infants.Neuropathic hearing impairment is common in preterm infants 99 .Preterm infants are much more likely to have sensori neural hearing loss (1 in 50) than are normalterm neo nates (1 in 1000) 100 .Risk factors for hearing loss include the need for intensive neonatal care 100 , and hyperbili rubinaemia 99,[101][102][103] .Nearly all individuals who did not sur vive in neonatal intensive care units failed ABR screening, and postmortem analysis revealed bilateral hair cell loss in several infants (selectively in IHCs in three of 15), whereas the number of SGNs was not reduced 104 .
The auditory system is particularly sensitive to hyper bilirubinaemia 103 .The precise disease mechanisms of bilirubin neurotoxicity are not yet well understood.The jaundiced Gunn rat that carries a mutation in the gene encoding UDPglucuronosyltransferase 1 is an animal model for human hyperbilirubinaemia 105 .Moreover, application of bilirubin has been used to study acute effects of bilirubinaemia 106 .Hypotheses on bilirubin toxi city in the auditory system include damage to presynap tic terminals via excessive nitric oxide 106 and changes in neuronal Ca 2+ homeostasis 107 .

IHC loss underlies thiamin-deficiency-related auditory neuropathy. Thiamine deficiency has been demonstrated
to cause auditory neuropathy in children 108 .Thiamine deficiencyrelated hearing impairment was characterized in detail in Israeli infants who were fed a thiamine deficient formula.Most of these infants showed auditory neuropathy, which resolved with supplementary thia mine in five of eight children; two of these infants lacked otoacoustic emissions, suggesting OHC dysfunction 108 .
The importance of thiamine for cochlear function was convincingly demonstrated in studies of mice that lacked highaffinity thiamine transporter, which is encoded by Slc19a2 (REFS 109,110).In one of these studies, a massive and rapid loss of IHCs was observed upon dietary thi amine restriction 110 .Although IHC loss associ ated with thiamine deficiency can explain an auditory synapto pathy phenotype, it seems incompatible with the tran sient nature of the hearing loss observed in infants with thiamine deficiency 108 .Moreover, the thiaminerelated hearing impairment was present in the absence of major IHC loss 109 .Therefore, we consider it likely that thiamine deficiency impairs IHC function before causing loss of IHCs.In conclusion, thiamine deficiency can cause audi tory neuropathy, but the disease mechanisms in hearing impairment associated with thiamine deficiency await further analysis in appropriate animal models 110 .

Noise-induced and age-related hearing loss.
In noise induced, druginduced and agerelated hearing loss, sev eral mechanisms lead to dysfunction or loss of both IHCs and OHCs 111 .These types of hearing impairments typi cally affect OHCs (as implicated by the lack of oto acoustic emissions).Damage and loss of afferent IHC synapses has been implicated as a contributor to noiseinduced and agerelated hearing loss, which are the two most common forms of hearing impairment [112][113][114][115][116][117] .Indeed, certain noise exposure protocols can cause a temporary threshold shift with full recovery of otoacoustic emissions (OAEs) and ABR thresholds, but a persistent substantial reduction of SGN compound action potential (CAP) amplitudes.Glutamate excitotoxicity that eventually leads to damage or loss of the postsynaptic terminal of SGNs is likely to underlie the synaptic loss in both disorders 118 (FIG.6).Indeed, agonists of ionotropic glutamate receptors dam age the IHC afferent synapse 113 , and mice lacking the glial glutamate-aspartate transporter GLAST show enhanced susceptibility to noiseinduced hearing loss 119 .
In the past decade, the analysis of excitotoxic synaptic damage has been greatly facilitated by quantitative light microscopy 120 .Counting immunolabeled IHC ribbon synapses in wholemounted mouse organs of Corti 116 , Kujawa and Liberman demonstrated a massive loss of ribbonoccupied IHC synapses in the highfrequency cochlea of rodents, even as a result of sound exposures that do not cause a permanent increase in hearing thresholds 116 (FIG.6).The loss of presynaptic ribbons was paralleled by a reduction in the first wave of ABRs that represents the SGN compound action potential 116 .The fact that the threshold recovered to normal levels was attributed to the recovery of cochlear function upstream of IHC synapses, and the unexpected resistance of syn apses formed with SGNs that have low thresholds and high spontaneous rates to the noise exposure, as shown by in vivo recordings from single SGNs in guinea pigs 121 .Interestingly, a selective loss of low spontaneous rate SGNs was also observed upon ouabain poisoning of the spiral ganglion in gerbils 122 .After noise exposure, the damaged peripheral neu rites do not form new IHC synapses, as described after a pharma cological excitotoxic insult in the guinea pig cochlea 114 , but the reasons for injuryspecific lack of recovery are not clear.Lack of synapse regeneration after noise might reflect impaired neurotrophic signalling in the organ of Corti 123 ; indeed, synaptic loss (deafferentia tion) could be mitigated by virally mediated expression of neurotrophic factors in the mouse cochlea 123 .Synaptic loss also takes place in the cochlea as a result of ageing in mice 115,117,118 and is accelerated if animals were previously subjected to 'nondamaging' noise exposure 118 .

R E V I E W S
Currently, the role of excitotoxic damage to synapses in hearing impairment has only been studied in animal models; hence, we do not yet know with certainty how relevant excitotoxic synaptic loss is to noiseinduced and agerelated hearing impairment in humans.If it is relevant, the selective deafferentiation of the SGNs with low spon taneous firing rates and high sound thresholds, which are most susceptible to excitotoxicity, would be likely to cause problems with speech comprehension under noisy con ditions, whereas normal audibility would be maintained by the most resistant SGNs with high spontaneous rates and low thresholds.Compound action potentials would be expected to remain, albeit reduced in amplitude.The resulting disorder would be assumed to represent a mild form of auditory synaptopathy with normal pure tone thresholds and speech comprehension in quiet environ ments.Clinical studies, including elecrocochleography and assessment of speech comprehension in noisy environ ments, are necessary to clarify the relevance and expres sion of excitotoxic synaptic loss in humans.However, even before such confirmation, the animal studies highlight the importance of limiting occupational and other noise expo sure in humans and indicate that pure tone audiometry that solely tests audibility, as often performed in otolar yngology, is insufficient to reveal such 'hidden hearing loss' (REF.124).

Diagnostics and clinical findings Clinical features and testing strategies
Clinical features of auditory neuropathy differ from fea tures of disorders that primarily affect cochlear ampli fication.Psychophysically, they include poor speech recognition, which is affected beyond the extent expected on the basis of the auditory sensitivity reported by pure tone audiometry.If the patients are able to recognize some speech, recognition is typically severely affected by background noise.Pure tone thresholds vary from near normal (in hearing impairments associated with Ca 2+ binding protein 2 or certain otoferlin missense mutations) to strongly elevated (for example, in prelingual otoferlin related deafness DFNB9).Moreover, pathological loudness adaptation (assessed with Carhart's tone decay test) and poor temporal processing (assessed with the gap detection test or by discrimination of low frequencies) indicate dys function of synaptic or neural auditory processing in the cochlea (TABLE 1).
Different strategies of audiological testing are recom mended according to the feasibility of psychoacoustic test ing at the specific age of a patient (TABLE 1).Physiological testing (primarily ABRs, otoacoustic emissions, such as DPOAEs and transitory evoked OAEs, and electrocochleo graphy (transtympanic recordings are preferable because of a better signaltonoise ratio than with intrameatal record ings)) is most relevant, but results need to be confirmed by psychophysical testing in all age groups.Accurate diagno sis is important, because the decision to carry out cochlear implantation needs sufficient certainty of diagnosis.
The most important physiological findings in auditory neuropathy 2,125,126 are impaired or absent compound action potentials and ABRs despite evidence of OHC function, which can be observed by using cochlear microphonics (intrameatal or transtympanic electrocochleography) and/or measuring otoacoustic emissions.In addition, the middle ear (stapedial) reflex and the contralateral sup pression of otoacoustic emissions are typically lacking.Electrocochleography and tests of neural adaptation can help sitespecific diagnostics beyond discriminating audi tory neuropathy from other forms of sensori neural hearing impairment 87 .For example, the absence of the summating potential in electrocochleography indicates the lack of IHC mechanoelectrical transduction or absence of IHCs, as is seen in thiamine deficiency or perinatal hypoxia.

Distinguishing neuropathy and synaptopathy
Differentiating between synaptic dysfunction and neuro pathy remains challenging, even with advanced ana lysis.For example, electrocochleography results in similar find ings for patients with OPA1-associated auditory neuro pathy and OTOF-associated auditory synapt opathy 74,127 .Moreover, in animal models of auditory synaptopathy and neuropathy, the delayed and broadened compound action potentials can arise from either presynaptic 16,17,128 or post synaptic 21 defects.Assessing the extent of adap tation to pure tones, on the other hand, might facilitate differentiation between synaptopathy and neuropathy.In otoferlin associated auditory synaptopathy, adaptation for both low and high frequency tones has been reported to be strongly enhanced 62 .By contrast, patients with auditory neuro pathy adapted in a normal manner to low frequen cies but showed abnormal adaptation to highfrequency sounds 62 .Overall, individuals with auditory synapto pathy showed faster and more extensive adaptation than those with neuropathy 62 .
Clearly, manifestation of neuropathy or other neuro logical disorders beyond the auditory nerve is suggestive of primary neural dysfunction.In order to assess the scope of the disorder, it is of utmost importance to combine the detailed audiological analysis with vestibular testing, neuro radiological analysis and clinical neurological and ophthal mological investigation.Genetic analysis is of great help, given the heterogeneous aetiology and, in most cases, sporadic nature of the disease; in future, the importance of genetic diagnostic tools will grow.Currently, the ana lysis of individual genes, such as OTOF, MPZ and OPA1, remains essential, but next generation sequencing enables testing for mutations in a large array of relevant genes and will support clinical diagnostics of hearing impairment disorders in near future 129 .Strong bioinformatical analysis methods and validation by Sanger sequencing are required to identify pathogenicity of putative mutations.

Current options for hearing rehabilitation Counselling
Sufficient counselling is of critical importance for indivi duals affected by auditory synaptopathy or neuro pathy, as well as for their families.The common but counterintui tive experience that patients 'can hear but not understand' (REF.2) requires the doctor to explain the underlying patho physiology of sound coding and neural signal propagation.Families need to understand that a synaptic or neural dis order of the peripheral auditory system can cause this type of hearing impairment.Neurological analysis and MRI can comfort patients and their families if providing evidence against a generalized neuropsychiatric disorder.

Cochlear implants versus hearing aids
We recommend fitting of hearing aids as the firstline ther apy, even though they might not provide sufficient rehabili tation (see below and also discussion by Rance and Starr 87 ).In disorders such as auditory synaptopathy, in which spike generation and propagation are maintained, electrical stimulation of SGNs by cochlear implants is expected to provide efficient hearing rehabilitation.Cochlear implants have also proven helpful in auditory neuro pathies arising from SGN disease, such as those caused by mutations in the OPA1 gene, probably because the strong electrical stimuli achieve temporally welldefined SGN activation.

Considerations regarding young infants
Rehabilitation and counselling can be particularly chal lenging in young infants.Balancing diagnostic certainty and best chances for rehabilitation with cochlear implants in this patient group is particularly important, but also particularly challenging: certainty about the disorder increases with more reliable psychophysics as develop ment progresses, repeated physiological testing (poten tially showing maturation of neural responses such as ABRs) and the observed outcome from early hearing aid rehabilitation; on the other hand, the outcome from coch lear implant rehabilitation, which is currently considered as a good option for subjects with auditory synaptopathy or neuropathy 130,131 , is better the earlier the implantation is performed.

Uncertainties in outcomes
Unfortunately, the data on the outcome of auditory neuro pathy rehabilitation are limited 132 .At first glance, solely amplifying sound is not expected to improve auditory signalling, whereas the strong electrical stimulus of a cochlear implant can synchronously activate SGNs even in cases of impaired auditory nerve function 133 .However, although the outcome of auditory neuropathy in children with cochlear implants can indeed be similar to those with hearing impairment owing to other aetiologies, it does not seem to be substantially better than that of auditory neuro pathy fitted with hearing aids 134,135 .Of course, it is quite possible that the patients with auditory neuro pathy who were fitted with cochlear implants were more severely affected and would have had a worse outcome with hearing aids.Moreover, experimental animal stud ies demonstrated a clear relationship between the ampli tude of eABRs and the number of remaining SGNs 136,137 .
In addition, mice with peripheral myelination deficits exhibit prolonged eABR latencies, elevated eABR thresh olds and diminished eABR amplitudes 138 .The uncertain ties in outcome with cochlear implant rehabilitation need to be considered for auditory neuropathy patients with neural aetiologies, whereas cochlear implant rehabili tation can generally be recommended for patients with severe auditory synaptopathy 139 .Therefore, an improved ability to accurately localize the site of lesion, using a combination of genetic diagnostics and transtympanic electrocochleo graphy, would be of great benefit when considering cochlear implant rehabilitation.

Future perspectives for rehabilitation
Progress in hearing aid and cochlear implant technology will, hopefully, improve the rehabilitation of auditory neuro pathy.Specifically, the use of directional micro phones and smart signal processing strategies can partially restore or improve speech comprehension.In the longer run, optogenetic stimulation of the auditorynerve 140 promises improved frequency resolution for coding with cochlear implants 141 .
Recently, preclinical studies have indicated that gene therapy might become a feasible cure of specific forms of auditory neuropathy.Mouse studies have demonstrated successful restoration of hearing by virus mediated gene replacement in experimental genetic auditory synaptopathy 18,142 , and viral vectors are currently being optimized for efficient and specific gene transfer.The size of some of the genes of interest, such as OTOF1 and genes encoding Ca V 1.3 Ca 2+ channels, exceeds the current pack aging capacity of the mostefficient and non pathogenic viral vector, the adenoassociated virus.Procedures for inoculation of the virus into the cochlea have been devel oped in various species and are now already used in a clin ical trial aiming to transdifferentiate supporting cells into hair cells 143 .A recent mouse study demonstrated that viral expression of nerve growth factor in the cochlea restored hair cell synapses following noise trauma, and mitigated noiseinduced hearing loss 123 .Although the results of all these efforts are promising, we anticipate that provi sion of these therapies to human patients with auditory synaptopathy or neuropathy is still a decade away.

Conclusions
Auditory neuropathy is a nosological term that was coined for a hearing impairment that commences down stream from mechanoelectrical transduction and coch lear amplification of OHCs.Human genetics and analysis of animal models of auditory neuro pathy have jointly elucidated disease mechanisms that include loss of IHCs or IHC ribbon synapses, impaired synaptic transmis sion to SGNs and disrupted propagation of auditory information along the auditory nerve.Disorders of IHC ribbon synapses (auditory synapt opathy) arise from genetic defects as well as from sound overexposure.Deep pheno typing involving audiology, neurology, ophthal mology and genetics enables diagnosis of auditory neuro pathy and, in some cases, identification of sensory, synaptic or neural disease mechanisms.Rehabilitation of hearing mostly relies on hearing aids (beneficial in some cases) and cochlear implants, which are particu larly beneficial for indivi duals with severe auditory syn aptopathy.Finally, future therapeutic options to restore hearing include gene replacement therapy for selected genetic disorders, hair cell regeneration, improved coch lear implants and local application of neurotrophins to, for example, improve synaptogenesis following acoustic synaptic insult.

Figure 1 |
Figure 1 | Clinical neurophysiology for definitive diagnosis of auditory neuropathy.a | Transtympanic electrocochleography is straightforward to carry out under local aneasthesia, and provides access to the electrical potentials of the cochlea: microphone potential (MP, primarily reflecting outer hair cell transduction), summating potential (SP, primarily reflecting the summed inner hair cell receptor potential) and compound action potential (CAP, reflecting the synchronized firing of spiral ganglion neurons) of the spiral ganglion (b-e).Example traces obtained in electrocochleography elicited by clicks of increasing volume.b | Clicks elicit MPs in an individual with normal hearing at sound pressure levels as low as 90 dB.c | Clicks elicit SPs and CAPs in an individual with normal hearing.d | In an individual with an auditory synaptopathy caused by OTOF mutation, MPs are intact; e | SPs are also present, but CAPs are not detectable even for 120 dB clicks, indicating intact outer and inner hair cell sound transduction but lack of synchronous SGN activation, most likely caused by failure of synaptic transmission owing to dysfunctional otoferlin protein.b-e adapted with permission from Elsevier Ltd © Santarelli et al.Hear.Res.330, 200-212 (2015).

Figure 2 |
Figure 2 | Disorders affecting inner hair cells (IHCs), their afferent synapses or spiral ganglion neurons (SGNs) degrade the neural representation of sound.a | Each IHC forms afferent synapses with approximately 10-30 SGNs, of which each is thought to get input from only one IHC active zone.b | Population responses, such as compound action potentials and auditory brain stem responses, arise from summation of the synchronized activity of many SGNs.Increasedbut homogeneous delays of action potential formation, as seen in conductive hearing loss, increase the latency but do not decrease the maximal amplitudes of the SGN population response.Variable delays of action potentials degrade the amplitude of the population response and also increase latency.Loss of activity, for example as a result of loss of SGNs, decreases the amplitude but does not affect the latency and the combination of variable delays and loss of activity produces the worst SGN population responses, if they persist at all.

Figure 3 |
Figure 3 | The afferent ribbon synapses between inner hair cells (IHCs) and spiral ganglion neurons (SGNs).The ribbon synapses are highly specialized for indefatigable encoding of sound information with submillisecond temporal transmission.a | Transmission electron micrograph of a ribbon synapse formed by the presynaptic IHC and the postsynaptic SGN.Note the electron-dense synaptic ribbon tethering synaptic vesicle (small black-rimmed spheres).Image courtesy of Dr C. Wichmann, Molecular Architecture of Synapses group, Institute for Auditory Neuroscience, University Medical Center Göttingen, Göttingen, Germany.b | Schematic of an IHC forming several ribbon synapses, each connected to a SGN with distinct spontaneous rate, sound threshold and dynamic range.c | Schematic of the IHC ribbon synapse.The synaptic ribbon is anchored to the presynaptic membrane by large scaffolding proteins.The molecular composition of the IHC ribbon synapse vastly deviates from that of glutamatergic synapses of the CNS.b, c are courtesy of Dr R. Nouvian, Institute for Neurosciences of Montpellier, France.

Figure 4 |
Figure 4 | Otoferlin has an essential role at the inner hair cell (IHC) ribbon synapse.Otoferlin is involved in vesicle fusion with the plasma membrane, and in vesicle replenishment.The role of otoferlin in vesicle replenishment might involve tethering as well as interaction with endocytic proteins, such as adaptor protein 2 (AP2) that enables rapid clearance of exocytosed material from the vesicular release site.AZ, active zone.Adapted with permission from Elsevier Ltd © Pangršic et al.Trends Neurosci.35, 671-680 (2012).

Figure 6 |
Figure 6 | Irreversible loss of inner hair cell (IHC) ribbon synapses during exposure to loud noise in mice.a | Rodent studies have demonstrated that noise-induced synaptic damage, likely resulting from excessive release of glutamate from the IHC presynaptic active zone, leads to irreversible loss of IHC-SGN synapses.b | Confocal micrographs showing the region of organ of Corti show reduced numbers of IHC synaptic ribbons (red spots) in 16-week old mice 2 weeks after 2-hour noise exposure (8-16 Hz, 100 dB; right) compared with mice not exposed to noise (controls).Sections are stained for neurofilament (green, staining afferent SGN nerve fibres) and RIBEYE/CtBP2 (red, stains synaptic ribbons that are seen as small, intensely fluorescent spots) and IHC nuclei (red, large round weakly fluorescent structures).Image courtesy of S. Kujawa, Massachusetts Eye and Ear, Harvard Medical School, Boston, USA.Mouse studies have shown that noise exposures that lead to only temporary increases in hearing threshold (c) can cause an irreversible loss of synaptic ribbons and postsynaptic terminals in the basal cochlea (d, e, f), which mediate high-frequency hearing, which is also reflected in a reduced amplitude of the compound action potential116 .CtBP2, C-terminal binding protein.
). Example traces obtained in electrocochleography elicited by clicks of increasing volume.b | Clicks elicit MPs in an individual with normal hearing at sound pressure levels as low as 90 dB.c | Clicks elicit SPs and CAPs in an individual with normal hearing.d | In an individual with an auditory synaptopathy caused by OTOF mutation, MPs are intact; e | SPs are also present, but CAPs are not detectable even for 120 dB clicks, indicating intact outer and inner hair cell sound transduction but lack of synchronous SGN activation, most likely caused by failure of synaptic transmission owing to dysfunctional otoferlin protein.