Mechanisms driving the functional maturation of the developing mammalian auditory pathway

2.1. Key molecules involved in mammalian cochlear development

The early development of the mammalian cochlea is primarily guided by intrinsic genetic programmes (see also Nelson, Hosamani and Groves Chapter 1 this volume for further details). In brief, the mouse cochlear duct begins to emerge and extend from around embryonic day 11 (E11), a process that requires sonic hedgehog (Shh) signalling. At this early stage, the developing cochlear duct, which will later become the scala media containing the sensory epithelium, is already divided into five distinct regions, each characterised by the expression of specific genes: Otx2 in the prospective Reissner’s membrane, Lmo4 in the stria vascularis, Bmp4 in the lesser epithelial ridge (LER), Sox2 in the pro-sensory domain and Fibroblast growth factor 10 (Fgf10) in Kölliker’s organ. Notably, genes such as Sox2, Fgf10 and Lunatic fringe (Lfng) are also expressed in overlapping patterns between the pro-sensory domain and Kölliker’s organ (reviewed by: Driver & Kelley, 2020; Michalski & Petit, 2019). The cochlear duct continues to extend until about postnatal day 3 (P3) (McKenzie et al., 2004). During this period, most cells in the duct continue to proliferate, except for those in the pro-sensory population, which will give rise to hair cells and supporting cells. These pro-sensory cells exit the cell cycle in a tonotopic gradient, beginning at the apex at E12 and progressing towards the base by E14 (Chen & Segil, 1999; Driver & Kelley, 2020). Following terminal mitosis, pro-sensory cells undergo functional differentiation that progresses in the opposite direction, starting at the base immediately after E14 and progressing towards the apex. This process of differentiation is governed by gradients in the expression of Atoh1 and Shh (Driver & Kelley, 2020).

In addition to tonotopic differentiation, cochlear development is also regulated along the medial-lateral axis, which is essential for establishing the mosaic arrangement of hair cells and supporting cells (Chen et al., 2002; Driver & Kelley, 2020; McKenzie et al., 2004). This process specifies the supporting cells of Kölliker’s organ (which eventually become the inner sulcus cells in the adult), and a single row of IHCs, pillar cells, Deiters’ cells, three rows of OHCs, and outer sulcus cells (Fig. 1C and D). Opposing gradients of Bone morphogenetic protein 4 (Bmp4), Fgfs and Wnt have been shown to play a key role in defining this medial-lateral axis (Driver & Kelley, 2020; Ellis et al., 2019; Geng et al., 2016). The complex pattern of hair cells and supporting cells is mediated by lateral inhibition involving Notch signalling. Atoh1 expression in developing hair cells activates Notch signalling in adjacent cells, leading to the upregulation of Hes1 and Hes5. These factors in turn suppress Atoh1, preventing neighbouring cells from becoming hair cells (Driver & Kelley, 2020; Lanford et al., 2000; Zheng et al., 2000; Zine et al., 2001).

Once hair cell fate is determined, their differentiation into OHCs and IHCs is established embryonically when they begin to express K⁺ and Ca²⁺ channels (Fig. 2). This process is determined by different transcriptional regulators: INSM1 is required to consolidate OHC fate (Wiwatpanit et al., 2018), whereas TBX2 is required for IHC identity (Bi et al., 2022; García-Añoveros et al., 2022; Kaiser et al., 2022) (Fig. 2). Moreover, a recent study has shown that CASZ1 is essential for IHC fate stabilization and OHC survival at postnatal ages (Sun et al., 2025) (Fig. 2). However, the biophysical and morphological differentiation of IHCs and OHCs occurs at around birth in mice, and this also depends on the expression of the non-coding microRNA 96 (miR-96, Kuhn et al., 2011) that is expressed from embryonic stages onwards (Kuhn et al., 2011; Lumayag et al., 2013; Sacheli et al., 2009; Weston et al., 2006) (Fig. 2). miR-96 regulates several downstream developmental genes including Tmc1, Ptprq and Gfi1 (Lewis et al., 2009). Mutations in miR-96 result in a coherent developmental arrest, preventing the normal biophysical and full morphological maturation of IHCs and OHCs (Fig. 2) (Kuhn et al., 2011) and leads to non-syndromic progressive hearing loss in humans and mice (Lewis et al., 2009; Mencía et al., 2009). OHCs are the first to initiate their functional maturation, beginning from around P6-P8 in mice with the acquisition of electromotility (Abe et al., 2007; Marcotti & Kros, 1999) driven by the motor-protein prestin (Zheng et al., 2000). The postnatal expression of prestin requires Helios, a transcriptional regulator encoded by Ikzf2 (Chessum et al., 2018). As OHCs mature around P8, their basolateral membrane also undergoes substantial change, including the expression of KCNQ4 channels (Kubisch et al., 1999; Marcotti & Kros, 1999) and the channels sustaining inhibitory modulation by the efferent synapses (SK2 and nAChRs) (Elgoyhen et al., 2001; Marcotti et al., 2004; Oliver et al., 2000) (Fig. 2). In contrast to OHCs, the functional maturation of IHCs occurs later, around P10-P12 (Kros et al., 1998) (Fig. 2), coinciding with the onset of hearing in most altricial rodents (Mikaelian & Ruben, 1965; Shnerson et al., 1982). While it remains unclear whether IHC functional maturation requires any additional transcription factors, this process is associated with the upregulation of KCNQ4 and BK channels (Kros et al., 1998; Marcotti et al., 2003a) and down-regulation of SK2 channels and α9α10nAChRs (Fig. 2). This downregulation aligns with the loss of axo-somatic efferent contact, which occurs at the same time (Glowatzki & Fuchs, 2000; Katz et al., 2004; Marcotti et al., 2004; Simmons et al., 1996).

The genetic programmes that govern cochlear development and organisation are not solely responsible for determining its final structure and function. Instead, genetic programmes are influenced and guided by spontaneous activity generated by immature hair cells before the onset of sensory-driven function (Friauf & Lohmann, 1999). This spontaneous activity plays a vital role in fine-tuning the regulation of genetic programmes at the cellular level, coordinating the activity among adjacent cells, and shaping cellular and neural circuit function before it becomes fully operational. This means that the developmental processes are not strictly genetically predetermined but can be refined up to their final configuration.

3.1. Modulation of the IHC Ca²⁺-dependent firing activity

Several studies have shown that the frequency and pattern of action potential activity in pre-hearing IHCs from cochlear explants can be modulated by intrinsic and extrinsic mechanisms. This diversity in firing modulation seems to be aimed at regulating not only the maturation of the central auditory pathway, but also that of the IHCs themselves. Previous work in other systems has shown that early spontaneous electrical activity is essential for regulating a variety of cellular responses, including the remodelling of synaptic connections (Zhang & Poo, 2001) and ion channel expression (Moody & Bosma, 2005). In IHCs, mechanisms identified that influence action potential activity include intrinsic factors, such as basolateral membrane and mechano-electrical transducer channels, as well as extrinsic influences such as ATP release from the surrounding supporting cells, and activity from the efferent cholinergic system descending from the brainstem.

3.2. Action potentials in OHCs

Unlike IHCs, much less is known about the origin and modulation of action potentials in OHCs. As discussed in Section 2, OHCs begin acquiring their adult-like biophysical and morphological characteristics by the end of the first postnatal week, approximately a week earlier than IHCs (Fig. 2; Abe et al., 2007; Marcotti & Kros, 1999; Marcotti, 2012; Oliver et al., 1997; Simmons et al., 1996). At this age, the small Ca²⁺ current (Michna et al., 2003) and the presence of the K⁺ current carried by KCNQ4 channels, which are activated at their resting membrane potential (I_K,n: Marcotti & Kros, 1999), prevent OHCs from generating Ca²⁺ action potentials. However, patch-clamp recordings have shown that action potentials can be elicited ex vivo in OHCs during the first postnatal week through membrane depolarisation (Marcotti & Kros, 1999; Oliver et al., 1997; Weisz et al., 2012) or elevated extracellular Ca²⁺ concentrations (Beurg et al., 2008). More recent studies using 2-photon imaging or cell-attached voltage clamp have demonstrated that OHCs can also fire spontaneous action potentials when cochlear explants are maintained under physiological Ca²⁺ concentration, but only during the first 3–4 postnatal days (Ceriani et al., 2019; Jeng et al., 2020c; Yang et al., 2023). This narrow time-window of spontaneous firing activity, compared to IHCs, likely results from the earlier onset of functional differentiation in OHCs. Limited information exists regarding the modulation and functional role of spontaneous Ca²⁺ activity in OHCs. A recent study revealed that ATP released by supporting cells binds to ionotropic P2X₃ receptors on OHCs, increasing their Ca²⁺-induced firing activity (Ceriani et al., 2019). This mechanism was implicated in regulating the maturation of OHCs and their afferent synapses (Ceriani et al., 2019). Furthermore, disruption of the MET channels in OHCs has been shown to impair proper segregation of type II afferent fibres (Sun et al., 2018).

4.1. Calcium signalling in supporting cells

Both immature (Ceriani et al., 2016; Gale et al., 2004; Majumder et al., 2010; Piazza et al., 2007; Rodriguez et al., 2012) and mature (Chan & Rouse, 2016; Dulon et al., 1993; Hool et al., 2023; Lagostena & Mammano, 2001; Sirko et al., 2019) cochlear supporting cells respond to the application of adenosine triphosphate (ATP), and related nucleotides, with elevations in their intracellular Ca²⁺concentration.

During pre-hearing stages of development, Ca²⁺ imaging experiments using ex vivo cochlear preparations have shown that local application of ATP induces Ca²⁺ oscillations in supporting cells (Ceriani et al., 2016; Gale et al., 2004; Piazza et al., 2007; Rodriguez et al., 2012). Pharmacological studies have indicated that these phenomena follow a canonical G-protein-dependent signalling cascade (Anselmi et al., 2008; Beltramello et al., 2005; Berridge et al., 2000; Piazza et al., 2007): ATP binds to G-protein-coupled P2Y receptors on the supporting cell membrane, leading to the activation of phospholipase C (PLC) enzymes (Fig. 5). Activated PLC hydrolyses phosphatidylinositol 4,5-bisphosphate (PIP₂) into diacylglycerol (DAG) and inositol 1,4,5-trisphosphate (IP₃). IP₃ in turn induces the release of Ca²⁺ from the endoplasmic reticulum into the cytosol through IP₃ receptors (IP₃Rs). Calcium oscillations arise from the interplay of Ca²⁺ release and sequestration (e.g., through SERCA pumps), and from the bell-shaped Ca²⁺-dependence of the IP₃R open probability (Bezprozvanny et al., 1991), which cause a biphasic IP₃R response to intracellular Ca²⁺ (Ceriani et al., 2016; De Young & Keizer, 1992; Li & Rinzel, 1994). Reduction of PIP₂ synthesis impairs the acquisition of high frequency hearing indicating an important role for this pathway in development (Rodriguez et al., 2012).

The propagation of Ca²⁺ signals across the supporting cells is due to two interdependent mechanisms (Fig. 5). First, gap junctions formed by connexin proteins couple the cytoplasm of neighbouring supporting cells (Forge et al., 2003) allowing them to share the cytoplasmic environment (Zhang et al., 2005). This allows the intercellular diffusion of IP₃ (Beltramello et al., 2005) propagating the Ca²⁺ signal as an intercellular wave (Anselmi et al., 2008; Beltramello et al., 2005; Majumder et al., 2010; Ortolano et al., 2008). Second, supporting cells can release ATP in the extracellular space either spontaneously or following the extracellular application of ATP onto the hair cells (Anselmi et al., 2008), mechanical stress (Zhao et al., 2005), or damage (Gale et al., 2004; Lahne & Gale, 2008; Lahne & Gale, 2010). Extracellular ATP activates purinergic receptors on neighbouring cells (Anselmi et al., 2008; Zhao et al., 2005), contributing to the intercellular propagation of Ca²⁺ signals (Fig. 5). The spatio-temporal propagation of intra- and intercellular Ca²⁺ signals is further modulated following ATP hydrolysis by ectonucleotidases (Vlajkovic et al., 2002) and mitochondrial buffering (Mann et al., 2009).

The primary candidates for ATP release from supporting cells are connexin hemichannels, which are connexin channels that have no matching hemichannel in the opposing cell membrane (Goodenough & Paul, 2003; Zhao et al., 2005; Anselmi et al., 2008; Majumder et al., 2010). Since hemichannels can open in response to an increase in intracellular Ca²⁺ levels (De Vuyst et al., 2006), it has been proposed that a localised release of ATP by one or a few supporting cells or hair cells can initiate the cascade leading to the long-range propagation of Ca²⁺ waves due to a process termed ATP-induced ATP release (Ceriani et al., 2016). Pannexin channels have also been proposed to underpin the release of ATP in the cochlea (Chen et al., 2015), although other studies have questioned their role (Zorzi et al., 2017; Mazzarda et al., 2020). Indeed, Ca²⁺ signalling due to spontaneous ATP release is unaltered in cochlear preparations from pannexin 1-deficient mice (Zorzi et al., 2017). However, it should be noted that while spontaneous Ca²⁺ signalling was shown to be severely reduced in a Cx30 mutant (Schütz et al., 2010) and a Cx30 global knockout (Mazzarda et al., 2020), a recent study has shown that the activity is not affected in the developing cochlea of Cx26 conditional knockout mice (Kersbergen et al., 2023). Further investigation is needed to completely elucidate the possibly diverse and self-compensating mechanisms responsible for spontaneous ATP release in the developing cochlea.

Ca²⁺ signalling in supporting cells is intimately linked with connexin function in the cochlea. Crucially, mutations in genes encoding connexins that preserve ionic conductance but impair macromolecule permeation through the connexin channel (e.g., the V84L CX26 and T5M CX30 mutations: Beltramello et al., 2005; Schütz et al., 2010) have been linked to hearing loss. Moreover, Ca²⁺ signalling in supporting cells can regulate the expression of both Cx26 and Cx30 (Ortolano et al., 2008). Finally, animal models with connexin deletions have shown that connexins are essential for the functional maturation of IHCs, even though the cells themselves do not express connexins (Johnson et al., 2017). Overall, this evidence highlights the importance of Ca²⁺ homeostatic mechanisms in supporting cells for hearing acquisition and normal cochlear function.

4.2. Spontaneous calcium waves and their relationship with IHC activity in the prehearing cochlea

During developmental stages (up to the second postnatal week in rodents), the cochlear epithelium comprises two distinct domains: the greater epithelial ridge (GER) lying on its medial aspect and containing the IHCs, and the lesser epithelial ridge (LER), which includes the OHCs, in the lateral aspect (Fig. 1) (Dayaratne et al., 2014). During this period, the GER includes Kölliker’s organ, a transient structure that is prominent during the embryonic and early postnatal stages (Dayaratne et al., 2014) and plays a crucial role in early development and maturation of the auditory system (Hinojosa, 1977).

Supporting cells of Kölliker’s organ spontaneously release ATP into the extracellular space, initiating the signalling cascade that leads to the propagation of Ca²⁺ waves along confined regions of the epithelium (Fig. 5) (Babola et al., 2020; Ceriani et al., 2019; De Faveri et al., 2025; Tritsch & Bergles, 2010; Tritsch et al., 2007; Tritsch et al., 2010). Spontaneous Ca²⁺ waves occur more frequently in the proximity of developing IHCs (de Faveri et al., 2025; Eckrich et al., 2018; Tritsch et al., 2007) and have been shown to trigger the simultaneous depolarization of groups of nearby developing IHCs, modulating their firing frequency and resulting in coordinated bursts of activity along the basilar membrane (Babola et al., 2020; de Faveri et al., 2025; Tritsch & Bergles, 2010; Tritsch et al., 2007; Tritsch et al., 2010). This activity ends after ear canal opening and the onset of hearing (Babola et al., 2021; de Faveri et al., 2025), following the regression of Kölliker’s organ and the downregulation of several key molecules involved in the signalling cascade (Hool et al., 2023; Huang et al., 2010; Wang et al., 2015).

Dissection of the signalling cascade leading to supporting cell-induced modulation of IHC activity (see also Section 3, Fig. 4) has been performed through pharmacological methods and through the study of mutant mice. These studies indicate that ATP released by supporting cells in Kölliker’s organ activates P₂Y₁ metabotropic purinergic receptors on neighbouring supporting cells (Babola et al., 2020; Babola et al., 2021). The increased Ca²⁺ in the supporting cells triggers the opening of the calcium-activated chloride channel TMEM16A, which is highly expressed by the supporting cells surrounding the IHCs (Wang et al., 2015), causing efflux of chloride ions. This is accompanied by efflux of water and potassium ions to maintain osmotic balance and electroneutrality (Wang et al., 2015), causing osmotic shrinkage in supporting cells (Tritsch et al., 2007) and depolarisation of neighbouring IHCs (Fig. 4). Finally, the increase in extracellular space due to supporting cell shrinkage quickly dilutes the extracellular potassium concentration, terminating the depolarizing effect on the IHCs and temporally restricting the duration of coordinated burst firing (Babola et al., 2020).

Even though supporting cells of the LER and GER share the same Ca²⁺ signalling cascade (Rodriguez et al., 2012), evidence for spontaneous activity originating in the LER has not been provided. However, Ca²⁺ waves originating in the Kölliker’s organ have been shown to propagate to the LER area in ex vivo cochlear explants and organotypic cultures (Ceriani et al., 2019; Majumder et al., 2010) where they trigger depolarization and Ca²⁺ elevation in Deiters’ cells and have been linked to the coordination of spontaneous activity of developing OHCs in mice during early postnatal ages (P0-P4, Ceriani et al., 2019). Further studies will be required to determine whether this type of activity is propagated centrally through type II spiral ganglion neurons in vivo.

4.3. Other roles for calcium signalling in cochlear non-sensory cells

Besides its role in cochlear development, ATP-dependent Ca²⁺ signalling is also implicated in the sensing of cochlear damage (Gale et al., 2004; Lahne & Gale, 2008; Lahne & Gale, 2010; Piazza et al., 2007). Mechanical damage to hair cells in neonatal cochlear preparations triggers release of ATP from the damaged cells and thus the release of ATP and propagation of Ca²⁺ waves in the supporting cells (Gale et al., 2004; Lahne & Gale, 2010; Piazza et al., 2007). A similar phenomenon can be reproduced by applying a brief puff of ATP to the hair cells (Anselmi et al. 2008; Ceriani et al. 2016). The activation of two distinct intercellular Ca²⁺ waves propagating at different speeds through the supporting-cell and hair-cell regions has been observed (Gale et al., 2004; Lahne & Gale, 2010; Piazza et al., 2007), mediated by both metabotropic (P2Y) purinergic receptors (as described for spontaneous calcium events above) and ionotropic P2X receptors, which drive Ca²⁺ entry from the extracellular space. The propagation of these signals following damage has been linked to the activation of extracellularly regulated kinases 1 and 2 (ERK1/2) (Lahne & Gale, 2008) in cochlear supporting cells, which promote hair-cell death. Although these phenomena have been studied in the neonatal cochlea, the persistence of ATP sensitivity and Ca²⁺ dynamics in the supporting cells of the mature cochlea (Chan & Rouse, 2016; Hool et al., 2023; Park et al., 2024; Sirko et al., 2019) and the role of ATP release during noise exposure (Housley et al., 2013) suggests that they might be important regulatory mechanisms also during normal hearing function.

5.1. Calcium signalling in IHCs

Although Ca²⁺-dependent action potentials in developing IHCs have been extensively investigated, the cellular mechanisms responsible for their initiation and modulation are still not fully understood. As discussed above (Section 3), some studies have shown that IHCs can spontaneously generate Ca²⁺ spikes at least during the first postnatal week (Beutner & Moser, 2001; Johnson et al., 2011; Kros et al., 1998). By contrast, other studies suggested that these spikes are triggered directly or indirectly by the spontaneous release of ATP from supporting cells (Tritsch et al., 2007; Wang et al., 2015). A recent in vivo study has provided new insight, showing that IHCs are able to intrinsically generate spontaneous Ca²⁺ spikes throughout pre-hearing postnatal stages, independent of synchronised activity driven by the Ca²⁺ waves in supporting cells (Fig. 6) (De Faveri et al., 2025). The occurrence of action potentials during the second postnatal week, which are absent in IHCs of ex vivo preparations, is likely attributed to the establishment of the endocochlear potential (EP) (Li, Liu, Zhao, & He, 2020) and the distinct ionic compositions of the endolymph and perilymph (Wangemann and Schacht, 1996). The EP, together with the low-Ca²⁺ concentration in the endolymph (Johnson et al., 2012), facilitates a depolarizing MET current through open MET channels (Waguespack et al., 2007; Lelli et al., 2009). This depolarizing current maintains the resting membrane potential of pre-hearing IHCs near the activation threshold of Ca²⁺ channels (Zampini et al., 2010), thereby promoting spontaneous AP activity.

5.2. Calcium signalling in supporting cells

Consistent with findings from ex vivo studies (see Section 4), supporting cells in the GER of live mice generate spontaneous Ca²⁺ waves throughout pre-hearing stages of cochlear development (Fig. 6). However, the dynamics of these Ca²⁺ waves differ between in vivo and ex vivo conditions. In vivo, Ca²⁺ waves in supporting cells were significantly smaller in size and faster (De Faveri et al., 2025) compared to those observed in cochlear explants (De Faveri et al., 2025; Eckrich et al., 2018; Harrus et al., 2018; Tritsch et al., 2007). Given that ATP is released by supporting cells in response to cochlear damage (Gale et al., 2004; Lahne & Gale, 2008), the large Ca²⁺ waves recorded from cochlear explants, or following the disruption of the in vivo physiological conditions (De Faveri et al., 2025), are likely to result from cochlear damage.

5.3. Calcium signalling in afferent terminals

Our current understanding of how IHCs transmit information to post-synaptic spiral ganglion neurons (SGNs) primarily stems from electrophysiological patch-clamp recordings performed on individual afferent terminals (reviewed by: Johnson et al., 2019; Moser et al., 2023; Reijntjes & Pyott, 2016), which were pioneered by Glowatzki & Fuchs (2002). However, we still have only a very limited understanding of how the different Ca²⁺ signalling dynamics are faithfully transferred from IHC ribbon synapses to SGN afferent terminals in the developing cochlea. In vivo recordings of Ca²⁺ signals reveal that SGN terminals exhibit sporadic Ca²⁺ transients, which are significantly faster than those observed in cochlear explants (De Faveri et al., 2025). Notably, the frequency and amplitude of Ca²⁺ transients recorded in vivo are higher in SGN terminals located on the pillar side compared to the modiolar side of the IHCs of pre-hearing mice. Considering that, in the mature cochlea, low-threshold, high spontaneous-rate SGN fibres are predominantly located on the pillar side of IHCs, while high-threshold, low spontaneous-rate fibres are found on the modiolar side (Liberman, 1978; Siebald et al., 2023), these in vivo findings suggest that this functional segregation is defined during pre-hearing developmental stages.

5.4. Multicellular functional interactions

Although IHCs can intrinsically generate Ca²⁺ spikes in vivo, their frequency is significantly increased by spontaneous Ca²⁺ waves originating from supporting cells. These Ca²⁺ waves also play a crucial role in synchronizing Ca²⁺ activity across multiple IHCs, as previously observed in ex vivo studies (see Section 4). However, under in vivo conditions, Ca²⁺ waves initially synchronize the activity of only a few IHCs, which then spreads longitudinally to recruit additional IHCs (De Faveri et al., 2025). Interestingly, this longitudinal propagation of IHC activity occurs much faster than the Ca²⁺ waves themselves and is independent from Ca²⁺ wave propagation through supporting cells, which was described in cochlear explants (Wang et al., 2015). The mechanism underlying this rapid, longitudinal activation of multiple IHCs remains unclear. One hypothesis could be that K⁺ accumulation in the intercellular space during the repolarisation phase of each AP may lead to the depolarization of the neighbouring IHC.

In live mice, the periodic synchronisation of IHC activity by Ca²⁺ waves has also been shown to increase the proportion of afferent terminals simultaneously activated by a single IHC (De Faveri et al., 2025). The activity among groups of IHCs along the tonotopic axis of the cochlea, which drives synchronous bursts of activity in the developing auditory fibres (Wang et al., 2015; Tritsch et al., 2010), likely contributes to reinforcement and refinement of neuronal projections to discrete areas within the developing auditory pathway (Clause et al., 2014; Müller et al., 2019). Future in vivo studies will be essential to identify the role of the efferent cholinergic system in regulating IHC activity during cochlear development.

6.1. Spiral ganglion neurons require input from inner hair cells for differentiation and maturation

The acoustic signal travelling from the cochlea to the brain can be measured using auditory brainstem responses (ABRs), which provide information about sound processing along the auditory pathway (Fig. 7). The different peaks of the ABR waveform are associated with discrete regions of the ascending pathway from the SGNs up to the auditory cortex (Fig. 7) (Möhrle et al., 2016). The initial stage of sound processing in the ascending auditory pathway is the transfer of acoustic information from the hair cells to the cochlear nucleus (CN) via the SGNs, the primary sensory afferent neurons (Fig. 7). Type I SGNs, constituting approximately 90 % of all SGNs, transmit all of the information from IHCs, while type II SGNs (the remaining ~10 % of SGNs) relay input from the OHCs (Ryugo, 1992). Mature type I SGNs each contact a single IHC, with up to 20 SGNs innervating a single IHC, whereas type II SGNs typically contact on average around 7 OHCs (Weisz et al., 2012). Type I afferent synapses are arranged around the base of the IHC from the modiolar to the pillar side, according to their threshold and spontaneous firing rate. High-threshold low-spontaneous rate SGNs are located towards the modiolar side, whereas low-threshold high-spontaneous rate SGNs are positioned towards the pillar side (reviewed by: Johnson et al., 2019). The differentiation, maturation and maintenance of SGNs require the precise timing and coordination of genetic programmes and physiological activity (Delacroix & Malgrange, 2015).

Immature SGNs display three physiologically defined subtypes based on their firing patterns in response to depolarizing current: a single action potential at stimulus onset, a phasic response (with a few spikes at stimulus onset) or tonic firing patterns. During the first postnatal week, these firing patterns change, leaving only fibres with single or phasic responses (Crozier & Davis, 2014; Davis & Crozier, 2015; Prescott et al., 2008). This developmental progression depends on IHC activity since the knockdown of otoferlin, the Ca²⁺ sensor for exocytosis at IHC ribbon synapses (Roux et al., 2006), delays the maturation of SGN biophysical properties (Conrad et al., 2024).

Single-cell RNA sequencing has identified three different genetic expression profiles among SGNs (type 1a, 1b and 1c), which correspond to the spontaneous firing rates and modiolar-pillar synapse location of mature SGNs (Shrestha et al., 2018; Sun et al., 2018). However, these genetic profiles only partially align with physiologically defined subtypes (Siebald et al., 2023). The maturation of these three genetic SGN subtypes relies on IHC activity during pre-hearing developmental stages (Shrestha et al., 2018). It has been shown that type 1a SGNs develop normally in the absence of IHC-driven activity, whereas types 1b and 1c fail to mature properly and instead adopt a type 1a genetic profile (Shrestha et al., 2018). Additionally, disruption of mechano-electrical transduction impairs the initial segregation of immature SGNs into type I and II, and also the separation of type I SGNs into the three different genetic subclasses (Sun et al., 2018).

In OHCs, the absence of SGN firing, resulting from the knockout of the Ca_V1.3 Ca²⁺ channels, which are essential for generating action potentials and exocytosis, leads to a reduced number of afferent fibres contacting the OHCs (Ceriani et al., 2019). This finding suggests that OHC activity plays a critical role in maturation of type II SGN afferent fibres (Ceriani et al., 2019).

6.2. Brain stem structures require cochlear spontaneous input for maturation

Localized firing of groups of adjacent IHCs along the basilar membrane (De Faveri et al., 2025; Tritsch et al., 2010; Tritsch et al. 2010; Wang et al. 2015; Zhang-Hooks et al., 2016) activates SGNs that encode a restricted range of sound frequencies. As outlined above (see Sections 3 and 4), ATP release from inner supporting cells in Kölliker’s organ induces Ca²⁺ waves that synchronise the activation of IHCs tuned to similar sound frequencies in adult mice, suggesting a role in refining tonotopic maps in the central auditory system. This localized, coordinated spontaneous activity in IHCs is transmitted to SGNs via specialized ribbon synapses, causing SGNs to exhibit stereotypical bursting activity interspersed with quiescent periods (De Faveri et al., 2025; Jones et al., 2007; Sonntag et al., 2009; Tritsch et al., 2010). This burst firing pattern differs from the continuous, stimulus-driven activity seen in mature SGNs in response to acoustic stimulation (Lippe, 1994; Taberner & Liberman, 2005), suggesting it has a developmental role in shaping the central auditory system. Indeed, recordings in pre-hearing rodents from the medial nucleus of the trapezoid body (MNTB) and the inferior colliculus (IC) have shown similar bursting activity (Sonntag et al., 2009; Tritsch et al., 2010), which is abolished following cochlear ablation (Tritsch et al., 2010).

Genetic knockout of otoferlin, which prevents the transfer of spontaneous activity from the IHCs to the SGNs, leads to a consequent loss of spontaneous firing in the medial nucleus of the trapezoid body (MNTB) (Müller et al., 2019). This loss of activity impairs MNTB-lateral superior olive (MNTB-LSO) synaptic refinement and strengthening, reducing the precision of MNTB-LSO topographic connections (Fig. 8A and B) (Müller et al., 2019). Similarly, TMEM16A-conditional knockout mice, which likely have disrupted and unsynchronised IHC firing activity in vivo, show severely diminished burst firing in MNTB neurons before hearing onset, and reduced frequency selectivity shortly after hearing onset (Maul et al., 2022; Kersbergen & Bergles, 2024). These findings underscore the role of cochlear IHC burst firing in refining auditory connections to form the tonotopic axis (Maul et al., 2022). Naturalistic bursting patterns imposed on MNTB-LSO circuit in brain slices induce robust long-term synaptic potentiation, highlighting the importance of the burst firing pattern in tonotopic map refinement (Bach & Kandler, 2020). Further evidence comes from studies on mice with IHCs lacking α9nAChRs, which show normal spontaneous action potential activity but with altered frequency and timing (Johnson et al., 2011; Vetter et al., 1999). This change in spike pattern temporal structure leads to impaired synaptic strengthening, over-retention of MNTB-LSO connections, disrupted functional MNTB-LSO map refinement before hearing onset, and deficits in sound localization (Clause et al., 2014; Clause et al., 2017).

Calcium imaging in awake transgenic mice expressing genetically encoded Ca²⁺ indicators has revealed transient, highly correlated bouts of activity in IC neurons and auditory cortex (AC) along iso-frequency contours of the tonotopic axis prior to hearing onset (Babola et al., 2018). This coordinated activity has been attributed to cochlear-driven spontaneous firing, as bilateral cochlear ablation or pharmacological blockade of P2Y1 purinergic receptors (disrupting IHC burst firing), abolished the coordinated activity in the IC. Additionally, knockout of α9-nAChRs, essential for efferent feedback to the cochlea, reduces symmetrical correlated bilateral activation across the left and right colliculi (Fig. 8C and D), whereas a gain-of-function mutant increases this correlation (Babola et al., 2021).

TMEM16A knockout in the mouse cochlea has been shown to suppress IC neuron burst firing pre-hearing, resulting in broader tuning curves, higher neuronal gain, and compressed tonotopic maps in both the IC and AC (Fig. 8C and D) (Kersbergen et al., 2022). This indicates that spontaneous activity in the IC is required for the refinement of tone responses and connectivity in the tonotopic axis. Despite the AC tonotopic map compression, AC neurons maintain normal response properties, possibly due to alternative sources of spontaneous activity and post-hearing refinement (Babola et al., 2018; Molnár et al., 2020). Overall, spontaneous activity in the brainstem is required for CNS tonotopic organisation and the refinement of brainstem neuron response properties.

6.3. Auditory cortex

The role of peripheral spontaneous activity in shaping the development of the auditory cortex is less well understood. The thalamus and cortico-thalamic loops are established before peripheral inputs reach the cortex, including topographically organized connections within the subplate neurons (Molnár et al., 2020), which are part of the first postmitotic cortical neuronal layer. These neurons respond to peripheral auditory stimulation before responses are seen in the primary recipient layer of the auditory cortex (Wess et al., 2017), potentially explaining the presence of tonotopic maps in individuals with congenital deafness (Striem-Amit et al., 2016). In the absence of peripheral activity, the somatosensory thalamus retains an immature state, with each division resembling one of a ‘higher-order’ (Pouchelon et al., 2014). This suggests that higher-order thalamic-cortical connections may represent the default pattern, implying that peripheral input refines the medial MGB, the primary auditory thalamic subfield (Molnár et al., 2020), and cortical tonotopic maps (Babola et al., 2018).

7.1. Efferent reinnervation of IHC due to trauma or ageing

Age-related changes in IHCs were first suggested by observations of cochlear innervation by the efferent system. In the mature cochlea, MOC efferent fibres directly innervate OHCs, whereas LOC efferent fibres indirectly influence the output of IHCs by synapsing onto their type I afferent nerve fibres (Fig. 1D). Following acoustic trauma or overexposure to AMPA (which mimics a toxically high concentration of the IHC neurotransmitter glutamate), type I nerve fibres endings on guinea-pig IHCs swelled and then disappeared. Within hours of this happening, efferent nerve fibres began to contact the IHCs directly (Ruel et al., 2007), as seen during cochlear development. This direct efferent reinnervation of IHCs has also been observed in ageing mice (8–11 months), associated with loss of type I afferent nerve fibres, OHC loss, and increased ABR thresholds (Lauer et al., 2012). However, recent evidence indicates that in adult IHCs, efferent innervation can occur even in the presence of afferent fibres (Corns et al., 2018; O’Connor et al., 2024). Functionally, these newly formed efferent synapses inhibited IHCs by activating SK channels (Jeng et al., 2021; Zachary & Fuchs, 2015), similar to their role during cochlear development (Marcotti et al., 2004).

7.2. Loss of MET current leads to regression of IHCs to an immature state

Studies on mice with mutations in various hair-bundle proteins have demonstrated that an absence of MET current significantly impacts the maturation of IHCs and OHCs (see also Section 3). For instance, constitutive mutations in the MET channel pore protein TMC1 inhibit the normal development of hair cells, preventing the expression of I_K,f and I_K,n (see Fig. 2) Consequently, IHCs retain the ability to fire action potentials and exhibit immature neurotransmitter release properties (Marcotti et al., 2006). Similar developmental impairments were observed in myosin VI mutant mice, where the hair cells lack MET currents (Roux et al. 2009), as well as in mice with mutations in the Usher-syndrome proteins myosin VIIa, harmonin and protocadherin 15 (IHCs and OHCs: Roberts, 2013; IHCs: Corns et al., 2018). This suggests that the MET current is not only required for IHC maturation, but also for maintaining their mature phenotype. Compelling evidence in support of this hypothesis comes from experiments where myosin VIIa was conditionally knocked out. In these mice, the MET current functioned normally during development but gradually diminished over time, becoming undetectable by P30 (Underhill et al., 2025). Initially, the IHCs acquired the mature potassium currents (I_K,f and I_K,n, see Fig. 2), developed mature presynaptic activity, and normal hearing as measured by ABRs (Corns et al., 2018). However, from P20 onwards, I_K,f and I_K,n declined, disappearing entirely by P60, and the conditional myosin VIIa knockout mice were deaf. During the same time window, IHCs became smaller, re-expressed immature IHC currents (I_K1 and SK2), responded to ACh application with inhibitory SK2 currents, and became re-innervated by efferent fibres.

Further experiments using cochlear cultures from mature gerbils highlighted the crucial role of the resting MET current (which is approximately 10% of the maximal MET current active without hair-bundle stimulation or sound) in preserving the IHC mature physiological characteristics (Corns et al., 2018). In culture, where the hair bundles are not stimulated by sound, I_K,f progressively disappeared over five days in culture. Interestingly, reducing extracellular Ca²⁺ to mimic in vivo endolymphatic levels, which increases the size of the IHC resting MET current, slowed the decay of I_K,f, unless the MET current was pharmacologically blocked.

The findings of Corns et al., (2018) show that mature IHCs regress to an early developmental state within weeks following the loss of their MET current, to a stage where they fire spontaneous action potentials. This regression can result from mutations in MET-associated hair bundle proteins (as observed by Corns et al., 2018), noise-induced damage, or ageing. Although this process is much slower than the reversible stereocilia shortening that occurs within hours when the Ca²⁺ influx via MET channels is reduced (e.g. by MET channel blockers; Vélez-Ortega et al., 2017), it likely stems from a deficit in Ca²⁺ influx through the MET channels (see also Velez-Ortega and Frolenkov Chapter 2 this volume for further details). This Ca²⁺ influx may regulate gene expression in the cells by diffusing into the cytoplasm (Beurg et al., 2010).

7.3. Regression of IHCs and their innervation in the ageing cochlea

A study on the biophysical properties of ageing mice (over one year old; Jeng et al., 2021) revealed changes similar to those reported by Corns et al., (2018) following the gradual reduction of myosin VIIa in hair cells. However, the extent of these changes varied depending on the mouse strain’s susceptibility to age-related hearing loss (Jeng et al., 2021). In C57BL strains with the cadherin 23 mutation (Cdh23^ahl) which leads to early-onset age-related hearing loss, IHCs became smaller, resembling pre-hearing IHCs, had reduced MET currents and showed a significant decrease in I_K,f, although I_K,n was maintained. In mice in which the inherent cadherin 23 mutation was repaired, the progression of age-related hearing loss, as measured by ABR recordings, was slower but still present, and the reductions in I_K,f and cell size were less pronounced. In the mouse strain most resistant to age-related hearing loss (C3H), IHCs displayed minimal change, with only a slight reduction in size with age. Direct efferent reinnervation also varied according to the hearing loss phenotype. However, even in C3H mice, some reinnervation and reappearance of post-synaptic inhibitory SK2 currents in response to ACh were observed (Jeng et al., 2021). This efferent reinnervation likely involved LOC fibres, which typically innervate afferent nerve terminals below IHCs, as opposed to MOC fibres that transiently contact IHCs during cochlear development en route to the OHCs (Simmons et al., 1996). This reinnervation of IHCs appears to replace afferent synapses that are lost during age-related hearing loss, particularly those formed by the high-threshold, low-spontaneous rate fibres on the modiolar side of the IHCs (Jeng et al., 2021).

Interestingly, OHCs also exhibit spontaneous action potentials early in development (Jeng et al., 2020a) and fail to express I_K,n in the absence of the MET current, similar to IHCs. This suggests that OHCs may undergo regression in a similar way to IHCs when MET current input diminishes or ceases. However, OHCs do acquire their characteristic electromotility even without the MET current (Roberts, 2013) or action potential activity (Jeng et al. 2020a). In the ageing cochlea, the primary change observed in OHCs is a reduction in cell size, accompanied by a proportional decreases in I_K,n, electromotility and sparse afferent innervation (Jeng et al., 2020b). These changes occur regardless of the mouse strain’s susceptibility to early age-related hearing loss. Conversely, reductions in MET current size and loss of efferent synaptic connections are associated with early age-related hearing loss (Jeng et al., 2021) due to the cadherin 23 mutation (Cdh23^ahl).