Hearing Restoration through Hair Cell Regeneration: A Review of Recent Advancements and Current Limitations

Abstract

Hearing loss is extremely common, yet limited treatment options are available to restore hearing, and those that are available provide incomplete recovery of hearing detection. For patients who are born with normal hearing, the most common cause of hearing loss is the loss of the sensory hair cells located in the cochlea of the inner ear. Non-mammals, including birds, fish, and amphibians, naturally regenerate new hair cells after damage and this natural process results in functional recovery. While some limited hair cell regeneration also occurs in the immature cochlea of mice, the mature mammalian cochlea does not naturally produce replacement hair cells, and thus hearing loss is permanent. Since the late 1980s, researchers have been investigating mechanisms to convert supporting cells, the cells that remain once hair cells have been killed, into new replacement hair cells. Here we review the current status of hair cell regeneration in the adult cochlea, highlighting recent achievements, as well as challenges that have yet to be resolved.

1. Introduction

Hair cell regeneration was first discovered in the late 1980s in the auditory organ of birds, which is called the basilar papilla (Corwin and Cotanche, 1988; Ryals and Rubel, 1988). Subsequent studies found that other non-mammalian vertebrates (e.g. reptiles, amphibians, and fish) can also regenerate hair cells after damage (Baird et al., 1996; Balak et al., 1990; Taylor and Forge, 2005). Each of these sensory systems share common cell types with the mammalian cochlea: hair cells that detect sound waves or fluid movements through deflection of their stereocilia bundles (Tilney and DeRosier, 1986; Whitfield, 2002) and supporting cells that have glia-like functions including maintenance of structural support, uptake of ions and neurotransmitters, clearance of dying cells or debris through phagocytosis, and the release of growth factors and cytokines (Monzack and Cunningham, 2013; Wan et al., 2013). Hair cells in both mammalian and non-mammalian organs can be killed by ototoxic drugs and loud noise; however, unlike mammals, the neighboring supporting cells in non-mammals respond to that damage by producing new, replacement hair cells, resulting in repaired sensory detection (Marean et al., 1993; Ryals et al., 2013; Smolders, 1999). This process occurs by two cellular mechanisms: Direct transdifferentiation (also called phenotypic conversion) where a supporting cell changes into a hair cell without re-entry into the cell cycle, or mitotic regeneration, where supporting cells first divide and one or more daughter cells takes on a hair cell fate (Adler and Raphael, 1996; Jones and Corwin, 1996; Raphael, 1992; Roberson et al., 2004; Tsue et al., 1994). Unfortunately, no natural regeneration occurs in the mature mammalian cochlea, and, therefore, the hearing loss that results from hair cell death is permanent (Bohne 1976; Oesterle 2008).

While the sensory organs of both mammals and non-mammals contain a mosaic pattern of hair cells and supporting cells, the mammalian cochlea has evolved into a coiled-shaped structure containing two hair cell types with distinct functions and many subtypes of non-sensory cells, including the supporting cells of the organ of Corti (the mammalian auditory sensory organ located within the cochlea). Both hair cells and supporting cells are organized in rows located on top of the basilar membrane, which vibrates in response to acoustic stimulation. Supporting cells are arranged in a regular pattern spanning the medial-lateral axis of the organ of Corti and include (from medial to lateral) subtypes named: inner phalangeal cells, border cells, inner pillar cells, outer pillar cells, Deiters’ cells (in three rows), and Hensen’s cells (Figure 1). Inner and outer hair cells are surrounded by adjacent supporting cells and the inner and outer compartments are separated by the tunnel of Corti, which is a fluid filled channel flanked by inner and outer pillar cells. Many supporting cell subtypes contain a high density of cytoskeletal proteins important for sound detection (Parsa et al., 2012; Szarama et al., 2012). The unique architecture of the cochlea and specialized cell types allow mammals to hear a wider range of frequencies and have better frequency tuning than non-mammals (Fritzsch et al., 2013; Manley, 2000), however, this also likely limits their regenerative plasticity.

Figure 1.

Supporting cell subtypes within the organ of Corti. Inner and outer hair cells are separated by the tunnel of Corti which is flanked by inner pillar cells (blue) on the medial side and outer pillar cells (purple) on the lateral side. Inner hair cells are surrounded by border cells (green) and inner phalangeal cells (pink). Outer hair cells are surrounded by Deiters’ cells (red). There are also two rows of Hensen cells located lateral to the last row of Deiters’ cells.

According to the World Health Organization, hearing loss affects 430 million people worldwide (as of 2024) and is projected to increase to nearly 2.5 billion people by 2050. Hearing loss is also a preventable risk factor for developing dementia with age, and the use of hearing aids can reduce this risk by up to 50% (Lin et al., 2023). In addition to hearing aids, cochlear implants provide hearing improvement to patients with profound hearing loss, however, this requires surgical implantation and replaces the organ of Corti with an electrode. Moreover, recent widely publicized work demonstrated the viability of viral-based gene therapies to treat congenital hearing loss in humans (with first success in repairing a mutation in the Otoferlin gene) (Lv et al., 2024; Qi et al., 2024; Tang et al., 2023; Wang et al., 2024). Therefore, patients faced with the decision of whether to choose a cochlear implant to treat hearing loss often ask their physicians when hair cell regeneration will be an option in the clinic. Because hair cell regeneration uses an existing pool of cells in the ear to convert them into hair cells, this therapeutic strategy would be most appropriate for patients who are born with normal hearing and lose their hearing due to damage from loud noise exposure, ototoxic drugs or aging. However, hair cell regeneration could likely be combined with a hearing aid. In the future, it may also be possible to design a therapeutic regimen that combines hair cell regeneration and gene replacement therapies to both correct the mutated gene that caused hair cell loss in congenital deafness, as well as produce new hair cells from the remaining pool of supporting cells. Here we review the current progress in driving auditory hair cell regeneration in the mature cochlea of animal models and the challenges that remain to translate this work into a clinical therapy for patients.

2. Making Hair Cells in the Mature Cochlea

While non-mammalian species readily regenerate the hair cells of their inner ears following damage, this ability is more complex in mammals. The neonatal cochlea can spontaneously regenerate hair cells following the death of endogenous hair cells, but only when the original hair cells are killed within a few days after birth and only a small number of new hair cells are formed that remain immature (Bramhall et al., 2014; Cox et al., 2014). In addition, regenerated hair cells are found primarily in the apical turn of the cochlea, which is less mature than the middle or basal turn (McGovern et al., 2019). As the cells of the apical turn mature and the mouse cochlea approaches the onset of hearing detection, spontaneous regeneration throughout the cochlea fails and natural regeneration is no longer possible for the remainder of the life of the animal (Cox et al., 2014; Maass et al., 2015; McGovern et al., 2024b; Tao et al., 2021). This indicates that regenerative plasticity declines as the tissue matures and is supported by work showing that repressive changes to the epigenetic landscape occurs in supporting cells during maturation (Nguyen et al., 2023; Stojanova et al., 2016; Tao et al., 2021; Waldhaus et al., 2012). Specifically, over the first postnatal week, the genomic regions of critical genes associated with hair cell identity become inaccessible in supporting cells (Tao et al., 2021) and, therefore, this limitation must be overcome in order to reprogram them into hair cells.

To better understand the loss of regenerative plasticity in the maturing mammalian cochlea, we can look closer at the mechanisms underlying both development and regeneration of the immature mammalian cochlea and how they change during maturation. As with many tissue systems, the cochlea employs cell-cell signaling pathways that control cell fate selection and maturation, such as Notch, Wnt, Bmp, and Fgf; this signaling is critical for the appropriate formation of the hair cell and supporting cell mosaic (Chai et al., 2012; Hayashi et al., 2008; Jacques et al., 2012, 2007; Kiernan, 2005; Kiernan et al., 2006; Lanford et al., 2000, 1999; Mann et al., 2014; Morrison et al., 1999; Ohyama et al., 2010; Pauley et al., 2003; Pirvola et al., 2002; Shi et al., 2014; Zelarayan et al., 2007). While these pathways have been extensively studied during cochlear development and shown to enhance hair cell regeneration in the neonatal cochlea, their role in the mature cochlea is unclear as many of the molecular players that are involved in cell fate selection are downregulated in the first postnatal week.

One signaling pathway that has been extensively studied in cochlear development and hair cell regeneration is the Notch signaling pathway, which controls cell fate selection through a process called Notch-mediated lateral inhibition. In the cochlea, prosensory progenitor cells express the Notch receptor Notch1. Initially, some progenitor cells upregulate the hair cell fate-inducing transcription factor Atoh1, which activates hair cell identity. As they differentiate, hair cells also begin to downregulate Notch1 and upregulate Notch ligands Delta like ligand 1 (DLL1) and Jagged2 (JAG2), which bind to the Notch receptor on their neighboring progenitor cells. When the Notch ligand is bound to the Notch receptor, it induces cleavage of the Notch receptor’s intracellular domain (NICD) that translocates into the nucleus and activates Notch downstream effectors including Hes and Hey genes. One of these genes, Hes5, mediates lateral inhibition by directly inhibiting the expression of Atoh1 to prevent the activation of hair cell identity (Abdolazimi et al., 2016; Kelley, 2006). These Notch signal receiving cells become supporting cells that surround the sensory hair cells. Once hair cells have begun to differentiate around embryonic day 14.5 (Chen et al., 2002) and into the first postnatal week, the inhibition of Notch signaling though either genetic or pharmacological manipulation disrupts Notch-mediated lateral inhibition allowing progenitor cells and supporting cells to switch cell fate and become hair cells (Doetzlhofer et al., 2009; Korrapati et al., 2013; Li et al., 2015; Maass et al., 2015; McGovern et al., 2018; Romero-Carvajal et al., 2015; Yamamoto et al., 2006).

Similar to the natural regeneration that occurs in non-mammalian species, spontaneous hair cell regeneration observed in the neonatal cochlea depends on the loss of Notch signaling from hair cells (Ma et al., 2008; Maass et al., 2015; McGovern et al., 2018; Stone and Cotanche, 2007). When the original hair cells are killed, the adjacent supporting cells lose the expression of Notch effector and hair cell fate suppressor Hes5, suggesting that the loss of Notch-mediated lateral inhibition from endogenous hair cells promotes the conversion of supporting cells into hair cells. If, however, Notch signaling is artificially maintained in supporting cells while hair cells are killed, the supporting cells are no longer able to spontaneously convert into and regenerate hair cells (McGovern et al., 2018). Notch signaling in the neonatal cochlea is also needed to maintain the survival of supporting cells (Chrysostomou et al., 2020; Heffer et al., 2023). Taken together, this indicates that Notch signaling is critical for the maintenance of cell fate in the neonatal cochlea and that loss of Notch signaling permits regenerative competency. Unfortunately, the molecules associated with promoting or maintaining cellular identity in the neonatal cochlea, including the Notch ligands Dll1 and Jag2 and Notch effector genes, are downregulated during the first postnatal week as the cells in the developing cochlea mature (Maass et al., 2015). As a result, the inhibition of Notch signaling, nor the killing of endogenous hair cells, is sufficient to promote the conversion of supporting cells into hair cells one week after birth. This is one mechanism that explains why the mature mammalian cochlea does not naturally regenerate following the loss of endogenous hair cells.

Because the mature cochlea lacks the ability to naturally regenerate lost hair cells, another strategy for hearing restoration is genetic reprogramming of non-sensory cells by the forced expression of hair cell fate-inducing transcription factors, such as Atoh1 (Figure 2A,B,D,F). This strategy bypasses the need for non-sensory cells to activate hair cell networks naturally, and ectopically activates hair cell identity. Because Atoh1 is the first known hair cell program gene to be expressed in developing cochlear hair cells, it was selected for initial genetic reprogramming experiments (Zheng and Gao, 2000). Several studies have demonstrated the ability of Atoh1 alone to reprogram non-sensory cells in the neonatal cochlea, however, as with spontaneous hair cell regeneration, this ability decreases with maturation (Atkinson et al., 2014; Chen et al., 2021a; Iyer et al., 2022; Lee et al., 2020; Liu et al., 2024, 2014, 2012a; McGovern et al., 2024b, 2024a; Walters et al., 2017; Yang et al., 2012). In the mature cochlea, the expression of Atoh1 alone has had mixed results. In one study, Atoh1 expression alone was able to convert non-sensory cells in the mature guinea pig cochlea into hair cells (Atkinson et al., 2014; Izumikawa et al., 2005), however additional attempts using Atoh1 alone in the guinea pig or mouse have failed to reprogram the supporting cells of the organ of Corti (Lee et al., 2020; Liu et al., 2024; McGovern et al., 2024a; Walters et al., 2017). Taken together, these data indicate that, like spontaneous hair cell regeneration, non-sensory cells in the mammalian cochlea become less responsive to Atoh1 expression as they mature.

Figure 2.

Schematic of hair cell regeneration approaches. A, Normal organ of Corti. B,D, after the death of outer hair cells, the remaining supporting cells can be targeted for reprogramming to produce replacement hair cells (F). C,E, when severe damage occurs, both hair cells and supporting cells are killed and a flat epithelium scar fills the organ of Corti space which can also be targeted for reprogramming to produce replacement hair cells (F).

The mechanism underlying the decrease in supporting cell plasticity is likely due to changes in the accessibility of genes associated with hair cell identity among supporting cells. This would mean that, as hair cells and supporting cells mature, their identities become more solidified and less manipulatable. Evidence for this phenomenon comes from an investigation of chromatin accessibility in neonatal cochlear supporting cells. As discussed above, supporting cells at birth are responsive to both the death of endogenous hair cells as well as the expression of the hair cell fate-inducing transcription factor Atoh1. During this time, as well as in the embryonic cochlea, Atoh1 binding loci in the promoter and enhancer regions of ATOH1 target genes are accessible and awaiting activation by Atoh1. However, a week later, the Atoh1 binding sites in the enhancer regions of ATOH1 target genes become inaccessible (Stojanova et al., 2016; Tao et al., 2021). This means that even if a cell expresses Atoh1, Atoh1 cannot bind to and drive the transcription of hair cell network genes. Another transcription factor involved in the development of cochlear hair cells is Pou4f3, which has been identified as a pioneer factor in hair cell development (Yu et al., 2021). Specifically, when Pou4f3 is present in developing hair cells, Atoh1 binding sites are accessible and, therefore, ATOH1 can bind to and activate the expression of hair cell network genes. However, if Pou4f3 is mutated or absent during cochlear development, the chromatin surrounding Atoh1 binding sites is inaccessible to ATOH1, and hair cell development is impaired (Xiang et al., 1998).

The failure of Atoh1 alone to reliably reprogram cells in the mature cochlea and the evidence suggesting that additional factors, such as Pou4f3, are required for Atoh1 to initiate hair cell fate has prompted the exploration of multifactor reprogramming in the mature cochlea. The expression of Pou4f3 along with Atoh1 was sufficient to reprogram some supporting cells into hair cells when expressed at 4 weeks of age (Walters et al., 2017). These results were promising as it demonstrated that genetic reprogramming of mature cochlear cells into hair cells is possible. However, the reprogrammed cells had an abnormal shape and were immature, suggesting that additional factors may be required to overcome genetic and molecular barriers within supporting cells. Because genetic reprogramming attempts to recapitulate the developmental initiation of hair cell fate, understanding the molecular players involved in cochlear hair cell development might hold the key to unlocking hair cell reprogramming.

To investigate this, Menendez et al, analyzed postnatal day 1 mouse cochlear hair cells and identified 16 candidate transcription factors associated with hair cell fate (Menendez et al., 2020). From those 16 factors, they identified 4 that were able to reprogram mouse embryonic fibroblasts into hair cell-like cells in culture. These factors, Atoh1, Gfi1, Pou4f3, and Six1, were further explored in both neonatal and mature mouse models. As with Atoh1 alone, the expression of Atoh1 + Gfi1, or Atoh1 +Gfi1 + Pou4f3, or Atoh1 + Gfi1+ Pou4f3 + Six1 were able to reprogram non-sensory cells in the undamaged immature cochlea into hair cells using either transgenic mouse lines or viral based transduction (Chen et al., 2021b; Iyer et al., 2022; Zhang et al., 2024). However, neither Atoh1 alone nor Atoh1 + Gfi1 were sufficient to reprogram non-sensory cells in the mature cochlea into hair cells (McGovern et al., 2024b, 2024a; Walters et al., 2017). The combination of Atoh1 + Gfi1 + Pou4f3 expression at 3 weeks of age in transgenic mice, however, was able to reprogram numerous non-sensory cells in the inner and outer sulcus and spiral limbus (McGovern et al., 2024b). In addition, combined Atoh1 + Gfi1 + Pou4f3 expression among supporting cells of the organ of Corti was able to activate hair cell identity in a subset of supporting cells that appeared to be limited to the lateral most Deiters’ cells and only in the apical turn of the cochlea (McGovern et al., 2024a).

Interestingly, the death of the original hair cells enhanced transcription factor reprogramming of supporting cells by Atoh1 + Gfi1 + Pou4f3. As described above, when reprogramming factors were expressed among organ of Corti supporting cells, only cells in the apical turn along the lateral most edge of the organ of Corti (likely the third row of Deiters’ cells) were able to respond to reprogramming and convert into hair cells; however, if the original hair cells were killed in addition to transcription factor expression in supporting cells, significantly more regenerated hair cells were observed (McGovern et al., 2024a). Reprogrammed hair cells were no longer limited to the apical turn of the cochlea and instead were observed throughout the length of the cochlea including the middle and basal turns. Moreover, regeneration was now present throughout the medial-lateral axis of the organ of Corti (including all three rows of Deiters’ cells, two rows of pillar cells, and inner phalangeal and border cells) (McGovern et al., 2024a). This suggests that hair cells in the mature cochlea are somehow still controlling cell fate of their neighboring supporting cells, however, the mechanism by which hair cells signal to supporting cells in the mature cochlea is unclear.

Reprogramming of cochlear cells through viral transduction in the mature mammalian inner ear has, unfortunately, been less successful. In the adult mouse cochlea, the expression of Atoh1 +Gfi1 + Pou4f3, or Atoh1 + Gfi1+ Pou4f3 + Six1 through AAV-mediated transduction in undamaged cochleae was insufficient to reprogram non-sensory cells in the inner and outer sulcus or supporting cells of the organ of Corti (Zhang et al., 2024). However, in adult guinea pigs, when both hair cells and supporting cells were destroyed and replaced by an epithelial scar, called the flat epithelium, some reprogrammed hair cells were observed. When Atoh1 + Gfi1 were expressed by adenoviral transduction, a small number of hair cells were observed in the flat epithelium (Lee et al., 2020) (Figure 2C,E,F). In addition, when Atoh1 + Gfi1 + Pou4f3 + Six1 expression was induced, hair cells were observed below the basilar membrane in the scala tympani (Liu et al., 2024). Taken together, the data from these studies suggest that mature supporting cells and other non-sensory cells are responsive to genetic reprogramming, however, more work is required to develop clinically viable targeting strategies, and this will be discussed in section 5 of this review.

There is a small amount of data suggesting that the Notch signaling pathway, which, as described above, controls cell fate in the developing organ of Corti, is still involved in restricting supporting cells from becoming hair cells in the mature ear (Mizutari et al., 2013). Following noise exposure in the adult mouse cochlea, the Notch effector Hes5 was upregulated globally in the inner ear. Hes5 directly inhibits hair cell identity in the neonatal cochlea by negatively regulating the expression of Atoh1 (Abdolazimi et al., 2016). The pharmacological inhibition of Notch signaling (using the gamma secretase inhibitor LY411575) following noise exposure was able to mitigate the increase in Hes5, and this was accompanied by an increase in Atoh1 expression. In addition, Notch inhibition following noise exposure was associated with a decrease in the number of supporting cells, a small increase in outer hair cell numbers, and less severe hearing loss compared to noise exposure alone. Finally, lineage tracing of Sox2-positive supporting cells suggested that supporting cells were converting into a hair cell identity following hair cell loss, but only when noise exposure was accompanied by the Notch inhibitor (Mizutari et al., 2013). Another study using the adult guinea pig found that reduction in the Notch effector gene, Hes1 (using siRNA), after noise exposure resulted in less hearing loss and increased numbers of outer hair cells (Du et al., 2018). However, this work has yet to be replicated and clinical trials investigating the therapeutic efficacy of Notch inhibition to treat hearing loss in human patients with mild to moderate hearing loss showed no changes in pure tone threshold detection compared to baseline (Schilder et al., 2024). Taken together, these data suggest that additional work will be required to better understand how hair cells in the mature cochlea control cell fate of their neighboring supporting cells and whether combined transcription factor reprogramming and inhibition of pathways that promote supporting cell identity (like Notch) will improve hair cell regeneration.

3. Supporting Cells as the Source of Replacement Hair Cells

While cochlear supporting cells have largely been the focus for hearing restoration therapies based on their proximity to lost hair cells, there are limitations to their suitability for hair cell regeneration. One limitation for targeting supporting cells for hair cell reprogramming is that supporting cells are critical for cochlear function, sound detection, and hair cell survival. If supporting cells are converted into hair cells without dividing, the pool of supporting cells will be reduced leading to continued hearing loss (Mellado Lagarde et al., 2013). While deletion of key cell cycle regulators including p27^Kip1 or Rb is effective in driving neonatal supporting cells to proliferate (Liu et al., 2012b; Yu et al., 2010), mature cochlear supporting cells do not divide after these gene manipulations. However, a recent study found that expression of c-Myc in combination with reactivation of the Notch pathway (achieved by expressing the Notch intracellular domain (NICD)) induced cell division among normally quiescent supporting cells in the mature cochlea. It also made them responsive to reprogramming with Atoh1 alone (Shu et al., 2019). This enhanced ability for adult supporting cells to proliferate and convert into hair cells by Atoh1 expression alone also occurred when the cochlea was treated with a cocktail that targeted these two pathways as well as cyclic AMP and Wnt (Quan et al., 2023). Ultimately, forcing supporting cell division prior to hair cell reprogramming by manipulating cell cycle regulators presents additional concerns because these genes are dysregulated in many cancers and, therefore, have the potential to significantly increase the risk to patient health.

Another limitation for targeting supporting cells is that they are no longer present following a severe form of damage that has been reported in cadaveric tissue and has also been replicated in rodent models (Iyer et al., 2021; Wang et al., 2017). This severe damage, called the flat epithelium, occurs when both hair cells and supporting cells are lost and a flat cuboidal epithelium replaces the organ of Corti. Thus, if the supporting cells are no longer present, they cannot be a therapeutic target for hair cell reprogramming. Unfortunately, information surrounding the flat epithelium is limited. We do not know the source for the flat epithelium cells, the transcriptomic profiles of these cells, nor whether they can be targeted for hearing restoration. The two leading hypotheses for the formation of the flat epithelium are that 1) supporting cells have de-differentiated into generic epithelial cells, and 2) the cells of the inner and outer sulci have migrated in and sealed the lesion left behind after hair cells and supporting cells are both killed (He et al., 2020). Recent work by McGovern et al demonstrates that, if the source for the flat epithelium is cells from the floor of the cochlear duct (inner/outer sulcus cells or supporting cells), the origin cells can be converted into hair cells following expression of Atoh1 + Gfi1 + Pou4f3, suggesting that the cells of flat epithelium may be a target for hearing restoration as well. Importantly, reprogramming of non-sensory cells in the inner and outer sulcus not only initiates hair cell identity, but newly formed hair cells could redeploy the Notch signaling pathway and initiate supporting cell identity in their neighboring cells (McGovern et al., 2024b). Moreover, if the flat epithelium is derived from non-sensory cells, they may be able to reactive the cell cycle to populate both hair cells and supporting cells following reprogramming. Not only do non-sensory cells that are adjacent to reprogrammed hair cells activate markers associated with proliferation, but previous work has also demonstrated that, as the flat epithelium forms, cells naturally undergo cell division (Young and Raphael, 2007). Thus, the flat epithelium is a promising target for hearing restoration.

4. Factors Needed for the Functional Recovery of Hearing

For functional recovery to occur after regeneration, not only do we need to create enough new cells to replace what was lost, but they also need to survive, be located in the appropriate location, and mature into inner and outer hair cell fates. Based on the physics of sound wave propagation in the inner ear, hair cells must likely be located in a precise position within the organ of Corti to be functional. Therefore many studies aiming to induce hair cell regeneration have targeted the supporting cell subtypes located in close proximity to endogenous hair cells (e.g. inner phalangeal cells, border cells, pillar cells and Deiters’ cells) (Lee et al., 2020; McGovern et al., 2024a; Sun et al., 2021; Walters et al., 2017; Yamashita et al., 2018). These are also the cells from which spontaneously regenerated hair cells are derived (Bramhall et al., 2014; Cox et al., 2014; McGovern et al., 2019; Xia et al., 2021). However there are many reports showing that new hair cells can be produced from non-sensory cells in ectopic locations that are medial or lateral to the organ of Corti (Doetzlhofer et al., 2009; Kawamoto et al., 2003; McGovern et al., 2024b). It is unknown whether these non-sensory cells will contribute to hearing restoration, since reprogrammed cells located in the inner or outer sulcus are unlikely to be stimulated by sound waves. Yet, as we discussed in section 3, these cells may be the source for the flat epithelium scar that replaces the organ of Corti following severe damage and, therefore, be a potential target for hearing restoration.

Inner and outer hair cells are distinct cell types with unique functions and unique transcriptomes. Outer hair cells increase hearing sensitivity by amplifying basilar membrane motion through a process called electromotility (Dallos, 2008; Hudspeth, 2008). Whereas inner hair cells transduce sound waves into electrochemical signals that activate spiral ganglion neurons for the interpretation of sound. The mouse cochlea contains approximately 700 inner hair cells and 2,000 outer hair cells (Ehret and Frankenreiter, 1977), while the human cochlea contains ~3,500 inner hair cells and 12,000 outer hair cells (Ashmore, 2008). While loss of inner hair cells would produce a more profound hearing loss, outer hair cells are more susceptible to damage caused by multiple insults including exposure to noise, ototoxic drugs and aging making their replacement most relevant to clinical therapy (Li et al., 2002; Oesterle et al., 2008; Rydmarker and Nilsson, 1987; Sha et al., 2001; Stebbins et al., 1979; Wu et al., 2020).

Thus far, hair cells that have been derived via spontaneous regeneration in the neonatal cochlea fail to functionally recover hearing (Cox et al., 2014; Xia et al., 2021). This likely occurs because fewer than 5% of original hair cell numbers are regenerated and they do not survive into adulthood (Cox et al., 2014). Moreover, these new cells appear either immature or express markers of both inner and outer hair cells (Bramhall et al., 2014; Cox et al., 2014; Heuermann et al., 2022). In addition, while the neonatal cochlea can spontaneously regenerate lost hair cells, this developmental period corresponds to the end of the first trimester in human gestation and is, therefore, unlikely to be a useful clinical strategy (Clancy et al., 2001).

As described in Section 2, genetic reprogramming of mature cochlear cells shows promise for hearing restoration, however, functional recovery is yet unrealized and the maturation of reprogrammed hair cells into either an inner or outer hair cell fate is a critical requirement for function. The expression of proteins associated with mature inner vs. outer hair cell identity can be used to better understand the differentiation and maturation of converted hair cells. Two commonly used mature markers are the outer hair cell protein prestin and the inner hair protein VGlut3. In the mature cochlea, activation of Atoh1 + Gfi1 + Pou4f3, in either non-sensory cells or in supporting cells following hair cell damage, produced hair cells in the outer compartment that expressed prestin (McGovern et al., 2024b, 2024a). However, Atoh1, Gfi1, and Pou4f3 are transcription factors involved in the differentiation of both inner and outer hair cells and the addition of co-factors to the reprogramming cocktail specifically designed to promote inner or outer hair cell identity may further enhance maturation. Unfortunately, little is currently known about how the gene networks of inner and outer hair cells diverge, except for three factors (Insm1, Ikzf2 and Tbx2) which have been identified to promote either outer or inner hair cell identity respectively. When Ikzf2 ectopic expression was combined with Atoh1, reprogrammed supporting cells in the adult cochlea converted into hair cells that expressed prestin, with a large increase in the efficiency of this conversion when hair cell damage was induced prior to the expression of Atoh1+ Ikzf2 (Sun et al., 2021). The transcriptome of reprogrammed hair cells revealed that they resembled immature outer hair cells, and unfortunately retained many supporting cell-specific genes suggesting incomplete reprogramming. Thus far, work investigating inner hair cell reprogramming with Tbx2 has only been done in the neonatal cochlea, however transient expression of Atoh1 along with permanent expression of Tbx2 converted supporting cells into hair cells that also expressed VGlut3 (Bi et al., 2022; Li et al., 2023). Taken together these data suggest that reprogramming of inner or outer hair cells is possible, yet additional work to achieve fully reprogrammed and functional cells is required.

Another critical feature of hair cell function is the stereocilia bundle located on the apical surface of each hair cell. Stereocilia bundles have a unique stair-case structure with interconnected links between each individual stereocilium (Goodyear et al., 2005; Nayak et al., 2007). The tallest row of stereocilia on both inner and outer hair cells are attached to the tectorial membrane (Hakizimana and Fridberger, 2021; Verpy et al., 2011). Sound vibrations deflect the bundle towards the tallest stereocilia which causes mechanoelectic transduction (MET) channels to open and depolarization of the hair cell. By contrast, deflection of the bundle in the opposite direction closes MET channels and hair cells become hyperpolarized (Gillespie and Müller, 2009). In addition, stereocilia bundles of outer hair cells have a “V” or “W” shaped which are polarized and aligned with neighboring cells (Tarchini and Lu, 2019). Each of these unique features is critical for hair cell activation by sound and many genetic mutations which cause hearing loss in humans negatively impact the stereocilia bundle and/or its polarity (Petit and Richardson, 2009). Thus, for hair cell regeneration to produce a functional recovery, the stereocilia bundle morphology and polarization is critical. Unfortunately, all attempts to produce hair cells after damage have resulted in cells with stereocilia bundles that resemble developing hair cells in the embryonic cochlea that lack a stair-case shape and polarity (McGovern et al., 2024a; Menendez et al., 2020; Quan et al., 2023; Sun et al., 2021; Walters et al., 2017; Yamashita et al., 2018). Immature bundles are also produced during spontaneous hair cell regeneration in the neonate (Cox et al., 2014; Xia et al., 2021). We are not aware of any studies that investigated whether the bundle is attached to the tectorial membrane in regenerated cells. Thus, further investigation of stereocilia bundle development in regenerated hair cells is required for hearing restoration.

Finally, if regenerated hair cells mature and can be depolarized by sounds waves, their activation must be transduced to spiral ganglion neurons in order for central auditory structures to be activated. Each spiral ganglion neuron only connects with one inner hair cell (Kiang et al., 1982; Liberman, 1982; Spoendlin, 1969). Whereas each inner hair cell make synapses with 10–30 type I spiral ganglion neurons (Bohne et al., 1982; Liberman, 1980; Stamataki et al., 2006) and each outer hair cell synapses with 1–2 spiral ganglion neurons (Thiers et al., 2008). The cochlea is also arranged in a tonotopic map where high frequency sounds are detected in the basal turn of the cochlea and low frequency sounds in the apex. The tonotopic pattern is maintained in spiral ganglion neurons, the auditory brainstem and all other structures throughout the central auditory system. Many studies that produced regenerated hair cells found them to make contact with neurons based on immunostaining with pan-neuronal markers (Kawamoto et al., 2003; Lee et al., 2020; Liu et al., 2024; McGovern et al., 2024a; Walters et al., 2017). The assumption is that these fibers come from the spiral ganglion, but tracing experiments have not been done, and it is also unclear if the tonotopic pattern is conserved. Thus, the achievement of function after hair cell regeneration still remains the biggest hurdle in our field with stereocilia bundle maturation/polarity and innervation by the appropriate spiral ganglion neurons requiring further exploration.

5. Clinical Translation

While numerous studies have demonstrated the feasibility of hair cell reprogramming as a viable approach to hearing restoration in the mature cochlea using rodent models, this approach does not represent a clinically viable therapeutic strategy. These studies utilize genetically modified mice that have permanent and inheritable gene mutations to activate hair cell reprogramming which need to be converted into a gene therapy or drug treatment that transiently alters gene expression and that can be delivered locally to the inner ear of human patients.

One of the biggest challenges in translating these preclinical approaches to human patients is the pharmacokinetics of target compounds or viruses and the complexities of inner ear architecture. Similar to the brain, the inner ear is protected by the blood labyrinth barrier which is composed of specialized tight junctions that prevent molecules and pathogens from entering the ear (Salt and Hirose, 2018). In addition, the length of the cochlear spiral impacts drug diffusion. Thus, small molecules that can pass the blood labyrinth barrier easily, will only reach the basal turn of the cochlea before diffusing out of the ear when given through a local delivery route such as transtympanic or round window injection. However larger molecules such as antibodies, biologics or viruses are better candidates for local delivery and should be retained within the cochlea (Salt and Turner, 2024).

Initial work using adeno-associated viruses (AAV; in mice) or adenovirus (AdV; in guinea pigs) demonstrates that local delivery to cells of the inner ear is possible though surgical approaches. In rodents, drugs and viral particles can be delivered directly to the inner ear using a posterior semi-circular canalostomy (PSCC), a cochleostomy, or though the round window. These routes of administration, while local, have demonstrated some off-target viral transduction in the brain, which likely occurs from the virus passing through the cochlear aqueduct into the cerebrospinal fluid. Therefore, increasing the specificity of viral targeting to cells of interest will be critical for clinical application. Cell type-specificity can be improved through two main components of the viral construct: 1) the viral serotype/capsid, and 2) the promoter that drives gene expression. The viral capsid determines what cell types the virus can transduce and different AAV serotypes have been demonstrated to transduce distinct cell populations in the rodent inner ear. Moreover, AAV serotypes have different levels of off-target labeling outside the inner ear. In addition to the viral serotype, the promoter driving viral gene expression, similar to endogenous gene expression, will control what cell types express the delivered genetic package. Thus, identifying the appropriate viral delivery and gene expression tools will be critical for clinically viable hair cell regeneration therapies.

Initial work targeting reprogramming factors to supporting cells and non-sensory cells has shown promise in neonatal animals using ubiquitous promoters, however, this approach has been less successful in the mature ear, likely due to unsuccessful targeting or expression of the reprogramming factors. Thus far, much work investigating viral gene expression in the inner ear has focused on targeting hair cells for gene replacement therapy in patients with congenital hearing loss. Moreover, the viral tools that target supporting cells or non-sensory cells that have been developed in the immature cochlea often do not exhibit the same transduction patterns when deployed in the mature ear. As discussed in Section 2, when the reprogramming factors Atoh1 + Gfi1 + Pou4f3 + Six1 were expressed in the mature cochlea using AAV, supporting and non-sensory cells did not convert into hair cells. However, based on reporter expression using the same AAV serotype, very few mature cells were transduced and, therefore, it is unsurprising that reprogrammed cells were not detected. Thus, further development of AAV construct, including cell type-specific gene promotors, is required for translating hair cell reprogramming from rodent models to clinical application.

Finally, metrics to ascertain the effectiveness of the therapeutic intervention to drive hair cell regeneration are lacking. Animal studies use the auditory brainstem response (ABR) to measure the minimum sound volume of pure tone stimuli that produces a neural response in a metric termed threshold, where increased thresholds show evidence of hearing loss. However, there is no correlation between the amount of hair cell loss and a change in threshold, and the ABR is measured in 5 decibel steps at 4–5 specific frequencies. Both limitations have a negative impact on our ability to detect small changes in hearing or the impact of adding small numbers of hair cells produced by regeneration. Distortion product otoacoustic emissions (DPOAEs) and cochlear action potentials (CAP) are not much better. Thus, new functional measurements and/or behavioral metrics with increased sensitivity are greatly needed to assess regeneration strategies and inform which would be appropriate for further development and eventual testing in clinical trials.

Highlights:

• Expression of hair cell genes reprograms supporting cells into hair cells

• Combined expression of multiple hair cell genes improves reprogramming

• Cochlear maturation limits the success of many regeneration strategies

• Developing clinical delivery strategies is required