Hearing loss affects hundreds of millions of individuals worldwide, necessitating the development of novel therapeutic interventions to prevent or even restore cellular loss and cochlear function in patients. The loss of cochlear hair cells is the major primary contributor to permanent hearing loss. To effectively develop treatments, it is crucial to have animal models in which hair cell loss can be induced reliably. Hair cells are the most sensitive component of the mammalian cochlea and can be eliminated with acoustic overstimulation or ototoxic drugs such as aminoglycoside antibiotics. Despite this susceptibility, ablating hair cells in the mouse cochlea using loud noise or systemic administration of aminoglycosides is a challenging endeavor (Maraslioglu-Sperber et al., 2024). The shortcomings of many existing drug-based damage models are likely owed to the fact that the effective drug concentration inside the cochlear duct is insufficient to kill the more drug-resistant apical outer hair cells and inner hair cells.
We sought to overcome these shortcomings. First, we replaced the often-used aminoglycoside gentamicin – a natural product and a mixture of different isoforms and impurities – with the chemically defined sisomicin (Benkafadar et al., 2021). Sisomicin is a highly ototoxic impurity of gentamicin (O'Sullivan et al., 2020). Second, we decided to deliver sisomicin directly into the inner ear via the posterior semicircular canal, a direct route into the inner ear (Kawamoto et al., 2001; Suzuki et al., 2017). We previously used this strategy to efficiently ablate auditory hair cells in the chicken inner ear (Benkafadar et al., 2021; Benkafadar et al., 2024; Janesick et al., 2022).
We surmised that using artificial perilymph for drug delivery would replicate a physiological environment that closely resembles the natural conditions in the inner ear, allowing for an accurate assessment of sisomicin's impact on hair cells and associated structures. To ensure consistency and minimize variability, we opted to utilize a commercial preparation of AP for our sisomicin infusion experiments. Our first results of infusing sisomicin in this commercial AP preparation resulted in a rapid and complete loss of outer hair cells and the absence of ABR and DPOAE thresholds. This result was confusing because when we infused sisomicin in freshly prepared AP or saline, we detected noticeable shifts in ABR and DPOAE thresholds without accompanying hair cell loss.
To investigate the factors contributing to this discrepancy, we analyzed the properties of the commercial AP preparation and discovered that it had an extremely high osmolality despite the distributor’s assurance that it was a 1X ready-to-use buffer solution. This prompted us to carefully prepare an iso-osmolar solution of sisomicin in AP, aiming to match the physiological osmolality of extracellular fluids. Remarkably, the iso-osmolar sisomicin solution resulted in minimal shifts in ABR thresholds and no hair cell loss.
Further experiments were conducted to elucidate the essential factors responsible for the depletion of outer hair cells: sisomicin alone, osmolality shift alone, saline versus fAP, or the combined effect of sisomicin and hyperosmotic buffer. Infusion of hyperosmolar saline or sisomicin in saline without adjusting osmolality resulted in partial loss of outer hair cells and noticeable shifts in ABR and DPOAE thresholds. Importantly, when sisomicin was infused in hyperosmolar saline, it replicated the results observed with sisomicin infusion in commercial AP, leading to a nearly complete loss of outer hair cells and the absence of ABR and DPOAE thresholds. These findings suggested a synergistic potentiation of hyperosmolality and sisomicin. This could be attributed to a comparable mechanism seen in mouse damage models using systemic aminoglycoside injection, where coadministration of loop diuretics enhanced the ototoxicity (Taylor et al., 2008).
Analysis of cochlear whole-mounts confirmed the extensive loss of outer hair cells after ho-sisomicin treatment. Intriguingly, inner hair cells and supporting cells remained unaffected within the first 24 hours, suggesting an initial selectivity of ho-sisomicin for outer hair cells. This selective impact aligns with previous studies demonstrating the preferential targeting of outer hair cells by aminoglycosides.
Time course experiments provided insights into the rapid and selective death of outer hair cells through apoptosis. About 40% of outer hair cells were TUNEL-positive within six hours of ho-sisomicin infusion. Because DNA fragmentation is a marker of the latest stage of apoptotic cell death, we assume that induction of apoptosis in outer hair cells happens very fast after the infusion, within the first six hours. The loss of hair cells was sudden and extensive, with evidence suggesting a minimal role for macrophages or supporting cells in immediate debris clearance. This dramatic cell loss contrasts with the gradual decline seen in aging or with milder ototoxic events.
Remarkably, delayed loss of inner hair cells was observed between days 3 and 7 post-ho-sisomicin infusion without corresponding apoptotic indicators. These findings suggest the involvement of different mechanisms in the demise of inner and outer hair cells, warranting further investigation.
Examination of the sensory epithelium seven days after ho-sisomicin infusion revealed the complete loss of inner hair cells. We detected some variability in a small proportion of specimens that manifested in more severe damage, such as supporting cell loss or even epithelial flattening. It is important to note that while variability exists, the method generally is reliable and results in virtually complete ablation of all cochlear hair cells after seven days.
The absence of ABR and DPOAE thresholds across all frequencies indicates an immediate and profound impact on auditory function already three hours post-ho-sisomicin infusion. If only outer hair cells were initially impacted by the infusion, we would observe an absence of DPOAE thresholds but a moderate ABR threshold shift. However, the immediate absence of thresholds on both evaluations was detected before the death of outer or inner hair cells. This observation suggests that the compromise of cochlear function is more complex than simply causing the death of hair cells.
Despite these open questions, we argue that the method has an applied advantage: its synchronicity and the uniformity of its impact across the entire cochlea. As we have previously shown for the avian inner ear (Benkafadar et al., 2021; Benkafadar et al., 2024; Janesick et al., 2022), we expect that the synchronicity, and especially the two distinct waves of outer hair cell loss followed by inner hair cell demise, will be useful to determine gene expression changes in dying hair cells and surviving cochlear floor cells. Moreover, the more complex cochlear pathology after ho-sisomicin presents an useful comparative model when investigated parallel to the more clinical hair cell elimination that can be achieved with diphtheria toxin in the well-established Pou4f3DTR/+ mouse model. Both mouse models are excellent platforms for future single-cell omics experiments to determine a useful baseline for developing cochlear cell reprogramming therapies toward hair cell regeneration or similar endeavors.