ASK1 is a novel molecular target for preventing aminoglycoside-induced hair cell death

Abstract Aminoglycoside antibiotics are lifesaving medicines, crucial for the treatment of chronic or drug resistant infections. However, aminoglycosides are toxic to the sensory hair cells in the inner ear. As a result, aminoglycoside-treated individuals can develop permanent hearing loss and vestibular impairment. There is considerable evidence that reactive oxygen species (ROS) production and the subsequent phosphorylation of c-Jun N-terminal kinase (JNK) and P38 mitogen-activated protein kinase (P38) drives apoptosis in aminoglycoside-treated hair cells. However, treatment strategies that directly inhibit ROS, JNK, or P38 are limited by the importance of these molecules for normal cellular function. Alternatively, the upstream regulator apoptosis signal-regulating kinase 1 (ASK1/MAP3K5) is a key mediator of ROS-induced JNK and P38 activation under pathologic but not homeostatic conditions. We investigated ASK1 as a mediator of drug-induced hair cell death using cochlear explants from Ask1 knockout mice, demonstrating that Ask1 deficiency attenuates neomycin-induced hair cell death. We then evaluated pharmacological inhibition of ASK1 with GS-444217 as a potential otoprotective therapy. GS-444217 significantly attenuated hair cell death in neomycin-treated explants but did not impact aminoglycoside efficacy against P. aeruginosa in the broth dilution test. Overall, we provide significant pre-clinical evidence that ASK1 inhibition represents a novel strategy for preventing aminoglycoside ototoxicity. Key messages ASK1 is an upstream, redox-sensitive regulator of P38 and JNK, which are known mediators of hair cell death. Ask1 knockout does not affect hair cell development in vivo, but significantly reduces aminoglycoside-induced hair cell death in vitro. A small-molecule inhibitor of ASK1 attenuates neomycin-induced hair cell death, and does not impact antibiotic efficacy in vitro. ASK1 may be a novel molecular target for preventing aminoglycoside-induced hearing loss. Supplementary information The online version contains supplementary material available at 10.1007/s00109-022-02188-1.


Introduction
Aminoglycoside antibiotics are used for serious or drug resistant infections of the respiratory tract, urinary tract, central nervous system or intra-abdominal organs, as well as endocarditis or sepsis (in adults and neonates), and prophylactically during and post-surgery. Aminoglycoside therapies preserve life; however, negative side effects are associated with their use. Careful dosing and monitoring regimens limit toxic outcomes, including encephalopathy, neuromuscular blockade, and nephrotoxicity [1]. However, ototoxicity (toxic outcomes relating to the ear) remains a persistent problem.
Aminoglycoside ototoxicity predominantly manifests as irreversible damage to the sensory hair cells (HCs) of the cochlea and vestibular apparatus. Therefore, aminoglycoside-treated individuals can experience permanent damage to their hearing and/or balance. The reported prevalence of aminoglycoside-induced ototoxic outcomes is variable (Table 1). Nevertheless, for those affected, ototoxicity has a profound effect on their quality of life [2,3]. Currently, no treatment exists for the prevention of aminoglycosideinduced ototoxicity. Instead, clinical practice aims to minimise or avoid aminoglycoside use. However, an alternate therapy is not always feasible and a single aminoglycoside treatment can cause permanent damage to the inner ear [4]. Therefore, research efforts have focused on elucidating the mechanisms underlying aminoglycoside-induced HC death, to identify therapeutic targets to mitigate ototoxicity. These studies have established that HCs are susceptible to aminoglycoside toxicity, due to the abundance of HC specific mechano-electro-transduction (MET) channels that facilitate rapid aminoglycoside entry into the cell [5,6]. Once the aminoglycoside has entered the HC, it targets and damages the mitochondrial ribosome, preventing mitochondrial protein synthesis.
Aminoglycoside-induced mitochondrial damage is linked with the drugs' antibiotic mechanism. Specifically, aminoglycosides bind to the A-site on the 16S ribosomal RNA of the 30S ribosome, inhibiting bacterial protein synthesis and killing the bacteria [7]. Shared ancestry between mitochondria and bacteria means that the mitochondrial ribosome is similar to the bacterial ribosome, and therefore susceptible to aminoglycoside binding. Aminoglycoside-induced disruption of mitochondrial protein synthesis is subsequently detected by innate cellular monitoring systems, initiating cell signalling cascades that result in HC death.
Cell death pathways are not well-defined in the inner ear. However, it is known that excessive mitochondrial ROS production and subsequent cellular oxidative stress is a key aspect of the process [8][9][10]. As a result, research aiming to mitigate ototoxicity has predominantly focused on antioxidant compounds, with equivocal and often conflicting outcomes [11]. These studies have not translated to changes in medical care, and in some cases, antioxidant use was associated with serious side effects, such as bleeding, pain, and gastric distress [11]. It is unclear why antioxidant therapies that have been effective in vitro have not achieved particularly good outcomes in vivo. However, normal ROS production is an integral part of cellular and physiological function; thus, antioxidant therapies can induce damaging reductive cell stress [12].
In addition to ROS production, the activation of c-Jun N-terminal kinases (JNK) and p38 mitogen-activated protein kinase (P38) has been reported in aminoglycosidetreated HCs; and various JNK and P38 inhibitors attenuate aminoglycoside-induced HC death [13][14][15][16][17]. However, JNK and P38 have critical homeostatic functions, meaning that inhibition can have toxic effects in other tissues, such as the liver and spiral ganglion neurons [18,19]. Therefore, upstream molecules that drive the apoptosis-specific roles of JNK and P38 might represent better targets for ameliorating aminoglycoside-induced HC death. Recently, Tao et al. demonstrated that apoptosis signal-regulating kinase 1 (ASK1) was the only mitogen-activated protein kinase kinase kinase (MAP3K) to have significantly increased RNA expression after aminoglycoside treatment in HCs [20]. ASK1 is activated by redox imbalance and activates downstream MAP kinases, particularly JNK and P38. In addition, multiple ASK1 inhibitors have been developed, with encouraging results in human clinical trials [21]. Therefore, ASK1 inhibition may be a novel strategy for preventing aminoglycoside-induced HC death.
Ask1 is expressed throughout the murine cochlear epithelium, including IHCs and OHCs [22][23][24] (data may be viewed at https:// umgear. org/ index. html? layout_ id= f64f9 c22& gene_ symbol= map3k 5& gene_ symbol_ exact_ match=0). However, the role of ASK1 as a mediator of aminoglycoside-induced HC death has not been investigated. Furthermore, the role of ASK1 in normal auditory function has not been defined. Therefore, this study aimed to evaluate the importance of ASK1 in the mouse HC by characterising the auditory phenotype of Ask1 −/− mice and examining whether Ask1 −/− HCs are resistant to aminoglycoside toxicity. The utility of pharmacologically-mediated ASK1 inhibition was also tested for preventing aminoglycoside-induced HC death.

Methods
Detailed methodology is provided in Supplementary Material 1. Table 1 The reported prevalence of ototoxic outcomes in humans treated with aminoglycosides. The American Speech Language Hearing Association (ASHA) definition of ototoxicity is described as a hearing threshold shift ≥ 20 dB in one frequency; ≥ 10 dB at any two adjacent frequencies, or loss of response at three consecutive frequencies [25].

Mice
Ask1 −/− mice were generated previously [43] and backcrossed for > 12 generations onto the C57BL/6 background. Wild type (WT) C57BL/6 mice were purchased from the Walter and Eliza Hall Institute of Medical Research (Parkville, Australia) and used as experimental controls. CD-1 mice were purchased from Charles River Laboratories (Raleigh, USA).

Acoustic startle response (ASR) and auditory brainstem response (ABR)
ASRs were measured using the SR-LAB system (San Diego Instruments) as described in [44]. ABRs were measured using an evoked potentials workstation and BioSigRP Stimulate/Record System v4.4.1 (Tucker Davis Technologies). Sound pulses were presented to anaesthetised mice, in short bursts (100 μs duration, repeated 512 times), through a freefield magnetic speaker (model FF1, Tucker Davis Technologies) 10 cm from the mouse's left ear. Hearing thresholds were defined as the lowest sound pressure level capable of eliciting a visible ABR.

Tissue collection and analysis
Adult mice were euthanised by anaesthetic overdose and perfused with 10% NBF. Dissected cochleae were decalcified in 10% EDTA (7 days at 4 °C), oriented in 1% agarose in cryomolds (Sakura Finetek, Torrance, CA, USA) and paraffin-embedded. A microtome cut 2-µm sections parallel to the modiolus. Sections were stained with hematoxylin and eosin (H&E).  Table 1). For HC quantification, images from the explant base to apex were photo-stitched together and de-identified. Fiji software was used to draw two boxes (0.18 mm × 0.09 mm), which were aligned and overlaid either side of the explant mid-point. HCs were counted using the Fiji cell count tool, including cells with > half the cell body inside the rectangle boundary. The average HC number of the two boxes was recorded for each explant.

Western blot analysis
Protein was extracted from three pooled explants per treatment, producing ~ 1 mg/ml protein per sample. Ten micrograms protein was denatured and separated in a precast Gel (Bio-Rad Cat#4,561,093). Protein was transferred onto a 0.45 µm pore PVDF membrane (Immobilon-P, Cat#IPVH00010). Antibodies were incubated as per Supplementary Material Table 1. Relative quantification of steady state protein levels was performed by first normalising to the loading control, and then to a non-treated negative control within each blot.

Statistics
ABR and ASR data was compared between groups for each frequency/volume tested using unpaired T tests (with the Holm Sidak correction for multiple comparisons). Protein levels were compared using a standard two-way ANOVA with Sidak's multiple comparison test. A standard two-way ANOVA was used to identify changes in MICs and post hoc T tests were performed using the two-stage step-up false discovery method of Benjamini, Krieger, and Yekutieli. These analyses were performed using GraphPad prism (version 7.0a for Mac). HC counts were analysed with a three-way analysis of variance, before pairwise Fisher individual tests of the differences of means were performed to ascertain p values for strain/treatment. This analysis was performed in Minitab software (version 17 for Windows) under the guidance of Dr. Sue Finch, Melbourne University Statistical Consulting Platform.

Functional assessment of hearing in Ask1 −/− mice
The acoustic startle response (ASR) of male Ask1 −/− mice was significantly higher than that of WT controls, suggesting that male Ask1 −/− mice are acoustically hypersensitive ( Fig. 2A). A similar trend was present in the data from female Ask1 −/− mice, although this did not reach statistical significance. As factors unrelated to hearing, such as muscle strength, emotional status, or memory function can contribute to the ASR [46,47], the auditory brainstem response (ABR) was performed to assess hearing in the Ask1 −/− strain. ABR testing showed that there was no difference between Ask1 −/− and WT hearing thresholds when measured between 4 and 32 kHz at 4, 10, and 24 weeks of age (Fig. 2B).  (Fig. 3A). These qualitative observations suggested that HC death is delayed in neomycintreated Ask1 −/− explants. This observation also indicated that TUNEL staining was not the most suitable means to assess hair cell death, because hair cells that were lost were not always TUNEL-positive at the time points captured. In contrast, Myosin VIIa immunohistochemistry provided robust and specific hair cell labelling. As a result, the subsequent methodology for quantifying hair cell survival was based on Myosin-VIIa staining.

ASK1 inhibition protects WT HCs from neomycin toxicity
Having demonstrated that a genetic-mediated reduction in ASK1 affords protection against neomycin-induced HC death in vitro, we hypothesised that a small-molecule inhibitor of ASK1 (GS-444217) would also protect WT (C57BL/6) HCs against neomycin toxicity in vitro.
To test if treatment with GS-444217 could compromise HC survival, cochlear explants were cultured for 24 h with GS-444217 doses ranging from 0 to 100 μM. There was no evidence of increased HC death as a result of GS-444217 treatment (one-way ANOVA; OHC counts p = 0.36, IHC counts p = 0.27) (Fig. 4A, B, blue line).
Next, we investigated if treatment with GS-444217 could protect against aminoglycoside-induced HC death in vitro. Cochlear explants were pre-treated for 16 h with GS-444217 (0 μM-100 μM) or vehicle (DMSO), and then incubated for a further 24 h with the addition of 1 mM neomycin (Fig. 4A, B, orange line). As observed previously (Fig. 3), IHCs were highly resistant to neomycin treatment, with less than 1% of IHCs lost after 24 h neomycin treatment (Fig. 4A). Overall, there was no significant difference (p = 0.69) in the number of IHCs in neomycin-treated explants that had been pre-treated with either vehicle (22.7 ± 0.7, n = 7) or GS-444217 (23.1 ± 0.5, n = 9) at the 24 h time point. In contrast, GS-444217 pre-treatment markedly improved OHC survival in neomycin-treated explants, with concentrations over 1 μM GS-444217 significantly attenuating OHC death (Fig. 4B, orange line). For explants pre-treated with vehicle (DMSO) before being exposed to 1 mM neomycin, only 38.1 ± 4.8 Myosin VIIa-positive OHCs remained at the 24 h time point. When this number is normalised to untreated controls, a survival rate of 51% is indicated. OHC number increased to 51 ± 2.5 (68% survival) with a pre-treatment of 0.1 μM GS-444217, and 1 μM GS-444217 treatment resulted in 63 ± 2.5 OHCs surviving (92% survival). At a concentration of 100 μM GS-444217 pre-treatment, 68 ± 1.4 OHCs survived, representing 95% cell survival when compared to the DMSO treated controls (Fig. 4B). Over a 3-day neomycin treatment period, 100 μM GS-444217 afforded significant protection for both IHCs and OHCs (Fig. 4C, D). There was no difference in IHC survival at the 24-and 48 h time points. However, after 72 h of neomycin treatment, an average of 10.2 ± 1.7 IHCs were counted in the field of view for vehicle pre-treated explants, compared to 15.3 ± 0.6 in the GS-444217 pretreated explants (p < 0.001). The effect of GS-444217 pretreatment was more pronounced in the OHCs, protecting against neomycin treatment at every time point tested (Fig. 4D)

Protective effect of GS-444217 in cochlear explants
To investigate how GS-444217 reduces neomycin-induced hair cell death, we performed immunohistochemistry and western blot analysis. As previously observed in our ASK1 genetic knockout experiments (Fig. 3), ASK1 inhibition using GS-444217 limited canonical apoptosis (indicated by TUNEL staining) (Fig. 5A). Immunohistochemistry also showed that GS-444217 pre-treatment limited neomycin-induced ASK1 phosphorylation in cochlear hair cells (Fig. 5B). Western blot analysis of WT cochlear explants that were pre-treated with GS-444217 and then cultured for 24 h with either vehicle (DMSO) or 1 mM neomycin indicated a significant reduction of p-P38 when explants were pre-treated with GS-444217 in DMSO cotreated culture conditions (Fig. 5C). A similar trend of reduced p-P38 in GS-444217 pre-treated explants was observed in explants cultured with neomycin. Likewise, the p-JNK signal tended to be lower in explants that had been pre-treated with GS-444217, for both DMSO and neomycin-treated explants; however, this was not statistically significant. (Fig. 5D). We then performed a time course experiment to elucidate the impact of GS-444217 pre-treatment on JNK and P38 phosphorylation in cochlear explants, from 1 to 24 h post neomycin treatment (Supplementary Figs. 1-3). P38 phosphorylation was most notable in the supporting and epithelial cells of the cochlear explant (Fig. 5E), whereas the strongest p-JNK signal was observed in hair cells 4 h post-neomycin treatment (Fig. 5F). A faint p-JNK signal was also detected in the supporting cells of neomycin-treated cochlear explants ( Supplementary Fig. 3). Notably, a strong p-JNK signal was present in the axons of spiral ganglia in all treatment groups, correlating with previous observations that p-JNK has a homeostatic role for axonal development and maintenance ( Supplementary Fig. 3) [19,48,49].

GS-444217 does not impair antibiotic efficacy against Pseudomonas aeruginosa isolates
To evaluate the possibility that GS-444217 could have an off-target effect upon antibiotic action, the minimum inhibitory concentration (MIC) of amikacin, tobramycin, and neomycin against Pseudomonas aeruginosa (P. aeruginosa) was determined. P. aeruginosa is the most common cause of chronic lung infections in individuals with cystic fibrosis and these individuals are particularly susceptible to aminoglycoside-induced hearing loss, due to long-term aminoglycoside treatment [50]. One reference strain (ATTC 27853) and two clinical isolates (0307 and 0315) were utilised. Tobramycin and amikacin were tested because they are the most clinically utilised aminoglycosides for P. aeruginosa treatment. Neomycin was also tested as an aminoglycoside frequently used in the laboratory. The broth dilution test demonstrated that GS-444217 co-treatment did not significantly change the minimum antibiotic concentration required for any antibiotic tested to inhibit resazurin metabolism in all three P. aeruginosa strains (Fig. 6).

Discussion
This study utilised Ask1 −/− mice and a small-molecule inhibitor of ASK1 (GS-444217) to test if the downregulation of ASK1 attenuates aminoglycoside-induced HC death. Both genetic and pharmacological approaches demonstrated that ASK1 has an important role in the process of HC death and indicated that ASK1 inhibition could limit aminoglycoside ototoxicity.
Prior to this work, the effect of Ask1 knockout on auditory function had not been investigated. We used a combination of histology, immunohistochemistry, and functional ABR testing to demonstrate that Ask1 knockout does not compromise the peripheral auditory structures or hearing thresholds of mice. However, male Ask1 −/− mice presented with a hypersensitive startle response when compared to WT controls, and a similar trend was observed in the female cohort. It is difficult to predict what the underlying cause of the unusually strong ASR might be, because auditory signal integration and efferent control is not well understood [51]. High ASR's generally indicate neural processing dysfunction in an auditory or psychiatric disorder [52], whilst factors such as muscle strength, emotional status, and memory function can also affect the ASR [46,47]. Notably, abnormal behaviour in Ask1 −/− mice has been previously reported, including hyperactivity, enhanced motor coordination, and significantly increased dopaminergic transmission [53]. ASK1 deficiency has also been implicated with impaired sensory gating in individuals with schizophrenia [54] and an ASK1 polymorphism is associated with reduced cognitive empathy in men [55]. Combined, these observations suggest that ASK1 may play a role in neuronal signalling. Whilst this has not yet been directly investigated, Ask1 −/− mice could prove useful for further exploration of neuronal development and signal processing mechanisms.
The major aim of this study was to investigate the potential role of ASK1 in aminoglycoside-induced HC death. Analysis A TUNEL stained, wild type (WT) CD-1 cochlear explants (P3) that had been pre-treated with DMSO (left) or 10 μM GS-444217 (right), and subsequently treated with 1 mM neomycin for 24 h. White = TUNEL (dying cells), green = phalloidin (highlighting hair cells), red = SOX2 (support cells), blue = DAPI (nuclei). Scale bar = 40 μm. B P3 WT explants that had been pre-treated with DMSO or 10 μM GS-444217 before receiving 1 mM neomycin treatment (or saline control) labelled with p-ASK1 (green), phalloidin (red), and SOX2 (yellow). Scale bar = 20 μm. C Representative western blot stained for p-P38 and loading control (GAPDH) followed by protein quantification (n = 3). Note: the faint band above GAPDH is p-P38. D Representative western blot stained for p-JNK1, 2 and 3, and loading control (GAPDH); followed by protein quantification (n = 3). JNK 1, 2 and 3 are present in two isoforms (p54 and p46). Each sample contains protein extracted from three pooled explants that had been pre-treated with either DMSO or 100 μM GS-444217 and subsequently co-treated with either DMSO or 1 mM neomycin (neo) for 24 h. Relative protein levels in DMSO and neomycin-treated samples were calculated by first normalising to the loading control and then to the correlated non-treated control (pre-treatment → none). Error bars = SEM, * = p < 0.05, ** = p < 0.01, as detected using a standard two-way ANOVA. E Support cells in cochlear explants treated with 1 mM neomycin for 12 h are positive for p-P38; however, this signal appears reduced in explants pretreated with 10 μM GS-444217. F Cochlear explants treated with 1 mM neomycin for 4 h and labelled with a p-JNK antibody indicate that 10 μM GS-444217 pre-treatment attenuates pro-apoptosis JNK phosphorylation in cochlear hair cells. For E and F, red = p-P38/p-JNK, green = phalloidin (hair cells), yellow = SOX2 (support cells), blue = DAPI, and scale bar = 20 μm ◂ of Ask1 −/− cochlear explants demonstrated that Ask1 deficiency significantly attenuates neomycin-induced OHC death. Ask1 deficiency also afforded some protection against neomycininduced IHC death; however, this was less pronounced. Subsequent experiments using GS-444217 produced results that correlated with the genetic ablation data. However, HC survival appeared slightly higher in the pharmacological data. The main methodological difference between these independent experiments was the addition of DMSO, used as a vehicle for GS444217. DMSO is a potent antioxidant that has previously been shown to increase hair cell survival [56][57][58]. Therefore, the DMSO used in the GS444217 pre-treatment/control may account for any improvement of HC survival during pharmacological experiments. Overall, GS-444217 pre-treatment resulted in significantly more OHCs surviving 3 days of neomycin treatment, when compared with DMSO pre-treated controls. However, neither genetic knockout nor pharmacological inhibition of ASK1 completely attenuated HC death. This suggests that alternate mechanisms, such as MAPK independent caspase activation, may be removing HCs. Alternatively, damaged HCs can be phagocytosed by cochlear-resident macrophages or supporting cells [59,60]. Western blot analysis indicated that P38 and JNK phosphorylation occurred in GS-444217-treated explants, which correlates with previous studies that utilised Ask1 −/− mice and ASK1 inhibitors in alternate disease models [43,61,62]. Therefore, it is likely that ASK1 independent mechanisms are able to activate the P38 and JNK pathways in cochlear explants. Nevertheless, the increased HC survival observed in GS-444217-neomycin-treated explants was significant, especially given the non-physiological conditions in which cochlear explants are cultured. In particular, the experimental dose of 1 mM neomycin-applied directly to the HCs in vitro-is unlikely to represent the concentration of aminoglycoside found in vivo in cochlear endolymph. This experimental dose is used in vitro to achieve clear results in a limited experimental timeframe, because optimal HC survival is less than one week in vitro [45]. However, a dose achieving cochlear levels of 1 mM neomycin would not be considered safe for use in humans. Likewise, other clinically utilised aminoglycosides do not reach such high concentrations in vivo. For example, a study of 113 individuals receiving amikacin over a 2-week period demonstrated that average peak plasma concentrations did not exceed 17 μM amikacin [63]. Moreover, the stria vascularis acts as an important blood-barrier in vivo, limiting the amount of aminoglycoside entering the endolymph. Kinetic studies in rats have shown that a gentamycin plasma level of 13-14 μM achieves only 3.1 μM in the perilymph and 2.5 μM in the endolymph [64]. Similar results have been observed in the guinea pig, with intravenous gentamycin (300 mg/kg) achieving a peak plasma concentration of 1.1 mM, but only 0.3 mM in the cochlear lymph 3 h after treatment [65]. Therefore, the otoprotective effect of GS-444217 may be stronger in vivo, where HCs are likely to be exposed to significantly lower aminoglycoside concentrations than those used in this in vitro study.
To further investigate the utility of pharmacological inhibition of ASK1 as a protective strategy against aminoglycoside ototoxicity, we tested whether GS-444217 altered the efficacy of aminoglycosides against P. aeruginosa in vitro. Notably, these data indicated that 100 μM GS-444217 did not impair antibiotic efficacy, whereas 1 μM GS-444217 significantly attenuated neomycin-induced HC death. In combination, these results suggest that GS-444217 could be used to prevent aminoglycoside-induced HC death without impacting primary aminoglycoside care. Whilst this must be assessed in vivo, the promising safety of ASK1 inhibitors taken orally in clinical trials, combined with the observation that GS-444217 does not affect aminoglycoside antibiotic efficacy in vitro, implies that ASK1 inhibition could be applied as an otoprotective strategy in the clinic. Further studies are now needed to confirm that GS-444217 and other promising ASK1 inhibitors do not impair antibiotic efficacy against common infections, such as Group B streptococcus, E. coli, Staphylococcus aureus, or Enterobacteriaceae infection. These organisms often cause sepsis in neonates and subsequent aminoglycoside treatment is the greatest risk factor for these newborns developing hearing loss [66][67][68].
Collectively, our data provide compelling evidence that ASK1 inhibition may be a valid strategy for mitigating aminoglycoside-induced hearing loss in humans. Whilst previous studies have demonstrated the safety of ASK1 inhibition in human clinical trials [21], strong in vivo efficacy data is required to provide the evidence base for clinical testing of ASK1 inhibition in aminoglycosidetreated individuals. In particular, the inhibitor dosing schedule required to achieve meaningful auditory protection must be determined. Based on the safety results of human clinical trials to date, we anticipate the minimum dose of an ASK1 inhibitor will have no effect on the innate immune response; however, this needs to be fully evaluated in vivo. In this study, we have demonstrated that 1 μM of GS-444217 significantly attenuates HC death caused by 1 mM neomycin treatment. However, a trend of improved HC survival was also achieved using only 0.1 μM GS-444217. It is difficult to predict the exact GS-444217 dosing regimen that will be required to achieve similar cochlear concentrations in vivo; however, previous studies have achieved GS-444217 plasma concentrations markedly Fig. 6 The effect of GS-444217 on the minimum inhibitory concentration required of tobramycin, amikacin, or neomycin to prevent Pseudomonas aeruginosa metabolism of resazurin. MIC's were calculated using the broth dilution test for A tobramycin, B amikacin and C neomycin. n = three broth dilution plates per isolate for each antibiotic, with each plate containing two replicates for each GS-444217 concentration. Error bars = SEM. Note, no error bars indicate that calculated MIC was the same for all three plates ◂ higher than 1 μM. For example, a single oral dose of 50 mg/kg GS-444217 resulted in plasma levels of 92.8 mM in mice [69]. In rats, a single oral dose of 30 mg/ kg, achieved peak plasma concentrations approaching 100 μM, although GS-444217 plasma levels declined to approximately 0.1 μM after 24 h [70]. This indicates that twice daily GS-444217 dosing may be beneficial. Other ASK1 inhibitors have also been tested in vivo, with 30 mg/kg of GS-459679 achieving plasma concentrations of approximately 6 μM in mice, 6.5 h after administration [71]. Unfortunately, a conspicuous limitation for in vivo testing of ASK1 inhibition as an otoprotective strategy is that mice appear to be intrinsically resistant to aminoglycoside-induced HC death [72]. Therefore, additional models are required to generate the necessary pre-clinical data. The pig may be appropriate, as the porcine metabolic rate, immune system, cochlea, and auditory range closely mimic that of humans [73,74]. In addition, swine models of cystic fibrosis exist that may be valuable for studying the progression of infection and the efficacy of aminoglycoside antibiotics when an adjuvant ASK1 inhibitor is used [75]. A porcine model could also provide information regarding the benefits of ASK1 inhibition in regard to other aminoglycoside toxicities, such as aminoglycoside nephron and neurotoxicity.
This work provides a strong foundation to justify the development of pre-clinical models that may lead to the development of an adjuvant otoprotective therapy. Such a therapy would have far-reaching implications, as aminoglycoside ototoxicity is a significant health issue, not only for those directly affected, but also clinical health care providers and the economy in general. Avoiding the distressing and lifelong co-morbidities of acquired hearing loss, such as loneliness, impaired mobility, mental fatigue, cognitive decline, and increased mortality risk, is critically important for the health of an individual. From an economic perspective, mitigating aminoglycoside ototoxicity will reduce the financial burden of acquired hearing loss within the national health, education, and welfare systems. In the case of critical healthcare, the prevention of aminoglycoside ototoxicity may also improve the utility of aminoglycoside antibiotics. At present, clinicians must consider the risk of permanent hearing loss against the risk to life when prescribing an aminoglycoside treatment. However, reducing ototoxic outcomes would ease the burden experienced by clinicians during this decision-making process and potentially allow for the wider application of aminoglycoside therapy.

Acknowledgements
The authors thank the AREST CF participants, and RCH pathology, animal house staff, Matthew Burton, Dr Sue Finch, and Rosemary Carzino for providing support and/or services used in this project.
Author contribution Jacqueline Ogier: conceptualisation, data curation, formal analysis, funding acquisition, investigation, project administration, methodology, visualisation, writing (original draft, review, and editing). Yujing Gao: conceptualisation, data curation, investigation, methodology, and draft review. Eileen Dunne: conceptualisation, methodology, resources, and draft review. Michael Wilson: conceptualisation and draft review. Gregory H Tesch: conceptualisation, resources, and draft review. David J Nikolic Paterson: conceptualisation, resources, and draft review. Alain Dabdoub: methodology, funding, and resources. Rachel Burt: conceptualisation, methodology, supervision, funding acquisition, resources, and draft review. Bryony Nayagam: conceptualisation, methodology, supervision, and draft review. Paul Lockhart: conceptualisation, methodology, supervision, funding acquisition, resources, and draft review. Data availability Important data generated or analysed during this study is included in this published article. Further information is available from the corresponding author on reasonable request. Written approval for collection, research, and publication using participant sputum samples was given by the parents of participants (signed informed consent forms).

Conflict of interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.