Hearing depends on the transduction of sound to neural signals, a process largely attributed to the sensory epithelium in the cochlea, the organ of Corti, which consists of mechanosensitive hair cells (HCs), surrounding supporting cells (SCs), and neural endings. Within this epithelium, the HCs, arrayed in a single row of inner HCs and 3 rows of outer HCs, fulfil the crucial function of converting sound waves into electrical signals. Because of the lack of regenerative capability for HCs in the mammalian cochlea, loss of HCs leads to lifelong sensorineural hearing loss (SNHL). Therefore, it is necessary to explore effective regenerative strategies for replacing lost HCs. Given that HCs in non-mammalian vertebrates can spontaneously regenerate (Corwin and Cotanche, 1988; Ryals and Rubel, 1988), researchers have tried to explore the possibility of inducing mammalian auditory HC regeneration.

Several research avenues have been used for testing means to induce HC regeneration in the mammalian cochlea. Based on the findings that avian auditory HCs can be replaced by transdifferentiation of SCs within the sensory epithelium (Raphael, 1992; Stone and Rubel, 2000), induced transdifferentiation has been studied using transgenesis or gene transfer approaches. Studies using transgenic mice have demonstrated that manipulation of gene expression of HC genes like Atoh1 can induce formation of new HCs (Iyer and Groves, 2021). Genes identified in the transgenic mouse studies and in developmental studies (Driver and Kelley, 2020) have become useful in gene transfer approaches, which are feasible for clinical use in humans. Gene transfer techniques employing viruses or other transfer vectors to deliver genes to target cells within the inner ear bear a clinical application value (Lee et al., 2020). Relevant to the work presented here, the target cells are the non-sensory cells in deaf ears which degrade to the state of a flat epithelium (FE).

Transcription factors (TFs) are essential in regulating gene expression patterns within tissues, controlling cell fate, and influencing their morphological and functional development (Mulvaney and Dabdoub, 2012; Schimmang, 2013). In the process of inner ear development, Atoh1, a basic helix-loop-helix TF, plays a crucial role in inducing the differentiation of HCs (Bermingham et al., 1999; Woods et al., 2004). Studies using transgenesis have demonstrated that Atoh1 alone is capable of inducing the formation of HCs in vitro (Jen et al., 2019), as well as in neonatal transgenic mice in vivo (Kelly et al., 2012; Liu et al., 2012). Studies using gene transfer methods showed that transdifferentiation of SCs into new HCs can be accomplished using forced expression of the Atoh1 gene in cultures (Shou et al., 2003) and in mature ears in vivo (Kawamoto et al., 2003). However, the work in mature ears in vivo has produced variable results that were partially dependent on the severity of the lesion, with failure to induce regeneration in the severely traumatized organ of Corti (Izumikawa et al., 2008).

A possible way to enhance the ability of TF gene transfer to form new HCs is to use a combination of 2 or more TFs. For instance, Gfi1 and Pou4f3 have garnered attention after being identified as crucial factors for HC differentiation and survival (Costa et al., 2015). Pou4f3 is a downstream target of Atoh1 activation in HCs, while Gfi1 is the target of Pou4f3 (Wallis et al., 2003; Xiang et al., 1997). Moreover, a combination of 4 genes, namely Gfi1, Atoh1, Pou4f3, and Six1 (GAPS), possesses the capacity to convert fibroblast cells into HCs (Menendez et al., 2020). These results motivated our experiments using GAPS expressing viral vectors in an in vivo guinea pig model of severe lesion in the organ of Corti.

We chose to design and develop our viral approach using adenovirus. One reason is the ability of these viruses to accommodate a relatively large set of transgenes, which is necessary for inserting all 4 GAPS genes. The other reason is the need to transduce the SCs in the auditory epithelium. These cells do not efficiently express transgenes after adeno-associated virus infection in mature ears (Kilpatrick et al., 2011) but they do express transgenes more efficiently after adenovirus infection (Excoffon et al., 2006; Ishimoto et al., 2002; Lee et al., 2020). The route of administration of the adenovirus is also critical. Inoculation into the perilymphatic space has demonstrated suboptimal HC transduction when utilizing adenovirus vectors (Dazert et al., 1997; Stover et al., 1999). In contrast, robust transduction was demonstrated after scala media infusion in both mouse and guinea pig models (Excoffon et al., 2006; Ishimoto et al., 2002; Lee et al., 2020). The explanation for the difference remains obscure; nevertheless, we chose the scala media route because of the success of previous studies that used it.

Both mice and guinea pigs commonly serve as animal models for SNHL. The loss of HCs can be induced by an ototoxic drug, by exposure to prolonged or extremely loud auditory signals, or by aging (Hawkins, 1973; Wagner and Shin, 2019). Some lesions involve the direct loss of only HCs, while the SCs are not initially affected and retain their differentiated state. In other cases, such as prolonged post-cochlear implantation periods (Nadol et al., 1994), exposure to ototoxic drugs at certain doses and concentrations (Kim and Raphael, 2007), and certain hereditary cochlear pathologies (Webster, 1992), both HCs and SCs incur damage. This more extensive damage results in the structural transformation of the organ of Corti into a flat or cuboidal simple epithelium, a condition referred to as FE (Wang et al., 2017). To date, no method has been established to induce cochlear FE in the mature mouse cochlea; therefore, the guinea pig emerged as the sole suitable model for inducing FE (Izumikawa et al., 2008; Kim and Raphael, 2007).

In this study, we introduced an adenovirus vector carrying the 4 GAPS genes into the deafened mature guinea pig cochlea through infusion into scala media in vivo. Deafening was achieved by administering neomycin into the scala tympani, leading to the loss of HCs and appearance of a FE. Contrary to our previous research findings, which reported no morphological changes in the FE following the overexpression of Atoh1 alone (Izumikawa et al., 2008), our current investigation demonstrates that the adenovirus-mediated introduction of this combination of 4 TFs can convert the cells within the FE into Myosin VIIa-positive cells which we define here as induced hair cells (iHCs). Our findings offer preliminary evidence suggesting that the delivery of GAPS could potentially constitute an effective strategy for managing hearing loss associated with the formation of a FE in the mature cochlea.