P2X7 receptor is required for the ototoxicity caused by aminoglycoside in developing cochlear hair cells

A


Introduction
Hearing loss is one of the most common sensory disorders in humans, and it is usually caused by aging, noise, ototoxic drugs, or other environmental factors. Hair cells (HCs) are extremely sensitive signal transducers that are capable of responding to hair displacements of only a few nanometers, but this makes them very susceptible to both noise and drug-induced damage Calton et al., 2014). Because mammalian HCs cannot regenerate on their own, loss of HCs leads to permanent hearing impairment.
Aminoglycoside antibiotics (AGAs), including neomycin (Prentki et al., 1986) and gentamicin, are a class of chemokines known to be toxic to HCs (Selimoglu, 2007), and the rapid accumulation of high levels of these drugs in HCs might be the primary reason for the selective toxicity observed in vivo (Hiel et al., 1992;Richardson et al., 1997). AGAs can target and disrupt mitochondria within cells, which causes the release of proapoptotic factors and oxidative enzymes into the cytoplasm thus generating free radicals that damage HCs (Kros, 2019;Jiang et al., 2017;Guo et al., 2019).
There are several potential routes for AGA entry into HCs. Previous studies reported that AGAs can enter both HCs and supporting cells (SCs) via mechano-transduction (MET) channels and ion channels where they induce the synthesis of reactive oxygen species (ROS) (Kitcher et al., 2019;Alharazneh et al., 2011;Choung et al., 2009). In addition, megalin, a low-density lipoprotein receptor transmembrane protein, has also been reported to be an endocytic aminoglycoside receptor in the inner ear (Christensen and Birn, 2002;Chun et al., 2021a;Hosokawa et al., 2018). However, these mechanisms can only partly explain the accumulation of AGAs, and thus further mechanistic studies are needed.
Adenosine triphosphate (ATP), which has long been known as the intracellular energy currency molecule, can be released by cells under stressful conditions or after injury, and this leads to high concentrations of ATP in the extracellular milieu. The transport of extracellular ATP (eATP) and its metabolites is mediated by the evolutionarily conserved purinergic signaling system, among which eATP-P2X7 receptor signaling has become one of the most studied pathways in infectious and inflammatory diseases (Savio et al., 2018). P2X7 receptor is a ligandgated ion channel belonging to the purinergic type 2 (P2) receptor family. The P2 receptor family comprises the P2Y G protein-coupled receptors (including P2Y1, 2, 4, 6, and 11-14) and P2X receptors (P2X1-7) (Ralevic and Burnstock,1998;Franke,2011). P2X7 receptor is the most extensively studied receptor subtype from an immunological perspective. Sustained stimulation of P2X7 receptor by millimolar concentrations of eATP triggers non-selective pore formation that allows the passage of molecules of up to 900 Da, Na + and Ca 2+ influx, and K + efflux resulting in changes in the ionic homeostasis of the cell (Savio et al., 2018). These changes in ionic homeostasis induce subsequent reactions, including oxidative stress, immune reactions, and apoptosis. However, the role of P2X7 receptor in the survival of inner ear sensory cells is poorly investigated.
In this study, we focused on the effect of P2rx7 deficiency on neomycin-induced HC death. We investigated whether and how P2X7 receptor is involved in neomycin-induced hearing loss and confirmed that it potentiates AGAs to enter the HCs. Our results indicate that P2X7 receptor can be upregulated by initial stimulation of neomycin and then activated as a key component in HC death induced by mitochondriadependent oxidative stress. We further demonstrated that P2X7 receptor is involved in the entry of AGAs into HCs, and this suggests new clinical strategies for ameliorating drug-induced ototoxicity.

Mouse models and treatments
Experiments were performed in both sexes of C57BL/6 J WT mice and in B6.129P2-P2rx7 tm1Gab /J Pfizer homozygous mice (these are the P2rx7− /− mice referred to in the text, Stock No: 015809, P2X(7)R KO) ( (Solle et al., 2001;Albalawi et al., 2017;Yue et al., 2017;Chen et al., 2011), originally from the Jackson Laboratory). Mice were maintained at the Experimental Animal Center, Shanghai Medical College of Fudan University, China, at 22 • C on a 12-h light/dark cycle. Postnatal day (PND)0 was defined as the day of birth, and mice aged between PND2 and PND90 were used in all experiments. For in vivo studies, mice received a daily subcutaneous injection of neomycin (200 mg/kg) or sterile saline for 7 days as indicated.
Standard PCR was used for transgenic mouse genotyping using genomic DNA isolated from mouse tail tips, and quantitative real-time PCR (qRT-PCR) was performed to confirm gene expression levels. All animal procedures were performed according to protocols approved by the Animal Care and Use Committee of Fudan University and were consistent with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (Ethical approval number: 20190301011.3). All efforts were made to minimize the number of animals used and to prevent their suffering. (Protocol: https://www.jax. org/Protocol?stockNumber=015809&protocolID=18812).
For GT594 fluorescein observation, the GT594 was dissolved in culture media with neomycin and added to the cochlear explants. The cochleae were imaged with a Leica THUNDER Imager (Leica Microsystems, Wetzlar, Germany) for 6 h. After the observation, the specimens were fixed with polyformaldehyde solution and harvested for further analysis.

Neomycin injury in vitro and ATP measurements
For the tissue explants, the cochleae of PND2 mice were dissected in Hanks solution (Invitrogen, Carlsbad, CA, USA), and the spiral ganglion, Reissner's membrane, and stria vascularis were carefully removed. The cochlear explants were plated onto 35 mm 2 dishes coated with poly-Llysine (Sigma-Aldrich, St. Louis, MO, USA) and cultured in DMEM/F12 medium with N2/B27 supplement (Invitrogen, Carlsbad, CA, USA). After the cochlear explants were attached to the dishes, the culture medium was removed and replaced with medium containing neomycin (2.0 mM, Sigma-Aldrich). Culture medium without neomycin was used as the control. Exogenous ATP or BzATP was supplied together with GT594 in the culture media for cochlear explants up to 6 h. Different concentrations are indicated in the results section.
For the ATP assay, cochlear sensory epithelium from PND2 mice in PBS was placed onto coverslips pre-coated with Cell-Tak (Sigma-Aldrich) diluted 1:1 with PBS and placed into four-well Petri dishes (Invitrogen). The sensory epithelia tissues were then cultured in 100 μl DMEM/F12 medium (mixed 1:1) supplemented with N2 and B27 (Invitrogen) and allowed to attach for 24 h before the experiments. PBS with neomycin (1 mM) or PBS alone was added for 0.5 h, 1 h, 1.5 h, 2 h, 2.5 h, 3 h, 3.5 h, and 4 h. After the neomycin was removed, the sensory epithelia samples were quickly transferred into 100 μl of ATP assay reagent (G9681, Promega, Madison, WI, USA) in 1.5 ml centrifuge tubes using tweezers, and the contents were mixed vigorously for 5 min to induce cell lysis. The contents were then incubated at room temperature protected from light for an additional 25 min to stabilize the luminescent signal. The bioluminescence of the cochlear epithelia was read in a black 96-well plate using a luminescence reader (SpectraMax Luminometer, Molecular Device, PA, USA).

Auditory brainstem response (ABR)
The hearing thresholds of the mice were examined with the ABR test on PND30, PND60, and PND90. Briefly, mice were anesthetized with ketamine (100 mg/kg) and xylazine (25 mg/kg) and placed in a soundattenuating chamber on a thermostatic heating pad to maintain their body temperature at 38 • C. We recorded changes in the electrical activity of the brain in response to sound via electrodes that were placed on the scalp of the mice. Frequency-specific auditory responses were measured using the Tucker-Davis Technology system (TDT System III, Alachua, FL, USA) as previously described Sun et al., 2014).

In vivo neomycin exposure for RNA-Seq and data analysis
To explore the gene expression changes in the cochlear sensory epithelium after neomycin exposure in vivo, we performed RNA-Seq on both neomycin-treated and control epithelia. Neomycin was injected subcutaneously on PND7 and PND8 at a dose of 200 mg/kg into nine C57BL/6 mice. Equal volumes of PBS were injected subcutaneously into another nine mice as controls. After the injection of neomycin or PBS, sensory epithelia were dissected at PND9 in PBS and dissolved in RNA-Later.
The neomycin-treated or control epithelia were split into three fractions for replicates. RNA-Seq libraries of epithelia were prepared with the SMART-Seq v4 Ultra Low Input RNA Kit for Sequencing and the Illumina mRNA-Seq Sample Prep Kit. SPRI beads (Ampure XP, Beckman Coulter, Brea, CA, USA) were used for size selection in each purification step after RNA fragmentation. All libraries were analyzed for quality and concentration using an Agilent Bioanalyzer and Qubit 2.0 Fluorometer, and 2× 150 bp paired-end sequences were generated on an Illumina HiSeq2500 Platform. Fastq files were trimmed with Trimmomatic (Trapnell and Schatz, 2009). Clean reads were mapped to the mouse reference genome (mm10) using HISAT2, followed by transcript assembly using HTSeq and differential gene expression analysis using DESeq2. Genes with adjusted p-values <0.05 and fold change >1.2 were marked as significantly differentially expressed.

qRT-PCR
The mice were sacrificed and their brain tissues and otic capsules were immediately isolated. To obtain the total RNA and protein extract, 10 basilar membranes were pooled from the cultured explants in icecold buffer or from frozen cochleae, and the brain tissue was processed with the Qiagen AllPrep® DNA/RNA/Protein Mini kit (Qiagen GmbH, Hilden, Germany) following the manufacturer's instructions. The proteins were collected for Western blot analysis, and the RNA was used for qRT-PCR. The RNA concentration was measured with a Bio-Rad spectrophotometer (Bio-Rad, Hercules, CA, USA). Complementary DNA was synthesized from 1 μg total RNA by reverse transcription using random primers (Promega) and Superscript III reverse transcriptase (Life Technologies, Foster City, CA, USA) following the manufacturer's protocols. qRT-PCR was performed using a SYBR Green Master (ROX) kit (Roche Diagnostics, Indianapolis, IN, USA) on a C1000 Touch thermal cycler (Bio-Rad).
Each PCR reaction was carried out in triplicate. With GAPDH as the endogenous reference, the relative quantification of gene expression was analyzed using the ΔΔCT method. The primer pairs for the qRT-PCR are shown in Supplementary Table 1 and were designed with the Primer3 online software.

HC electrophysiology recording
Cochlear explants were extracted and cultured at P2 and P5, as described above. Both inner HCs and outer HCs in the apical and middle turns were identified and used for recording. Patch pipettes were prepared by pulling BF150-86-10 glass (Sutter Instrument) on a PC-100 vertical puller (Narishige). The typical pipette resistance in bath solution was 5-8 MΩ, and recordings were made with an EPC10 USB patch clamp amplifier (HEKA). Resting membrane potential recording was done under whole cell mode at room temperature (22-25 • C) with no current injected. For resting membrane potential recording, the internal solution consisted of 110 mM K MeSO 3 , 20 mM KCl, 10 mM HEPES, 2 mM EGTA, 10 mM Creatine monohydrate, 3 mM ATP-Mg, and 0.5 mM GTP-Na, pH 7.2 adjusted with KOH, 280-290 mOsm/l. The tissue samples were bathed in artificial perilymph containing 132 mM NaCl, 5.8 mM KCl, 10 mM HEPES, 0.7 mM NaH 2 PO 4 , 5.8 mM glucose, 5 mM pyruvate, 1.3 mM CaCl 2 , and 0.9 mM MgCl 2 , pH 7.4 adjusted with NaOH, 300-310 mOsm/l. The recording was filtered at 10 kHz with a low pass Bessel filter and digitalized at 20 kHz. Data were stored and analyzed using Igor 6 (WaveMetrics) and Prizm 6 (GraphPad Software).

Immunohistochemistry, TUNEL staining, and image acquisition
The temporal bones of sacrificed WT or P2rx7− /− mice were collected and stored overnight in 4% paraformaldehyde (Sigma-Aldrich) before decalcification in EDTA (4% in phosphate-buffered saline (PBS), pH = 6.4) for a total of 72 h. The otic capsule was then removed, and the cochlea was carefully isolated from the surrounding bony tissue. The samples of decalcified cochlear explants from PND2 mice were fixed with 4% paraformaldehyde for 30 min at room temperature in 0.1 M phosphate buffer. During the antigen retrieval step, a treatment solution of pepsin at pH 6.4 was added to the slice at room temperature for 25 min. The samples were then blocked with 10% donkey serum in 0.1% PBST (1% Triton-X100 in 10 mM PBS) at pH 7.4 for 1 h at 37 • C.
Mito-SOX Red (Life Technologies, Grand Island, NY, USA) was used for measuring ROS. For Mito-SOX Red staining, the cochlear explants were cultured in a dish with DMEM culture medium. After neomycin exposure for 6 h, the culture medium was removed from the dish and the samples were washed with PBS. Prewarmed (37 • C) solution containing Mito-SOX Red was added, and the explants were incubated with the probe for 20 min. After staining, the samples were washed in prewarmed PBS and imaged under a fluorescent-light microscope (Model Eclipse 80i; Nikon, Tokyo, Japan).
The fixed tissues were rinsed with PBS and then incubated with primary antibodies (diluted with 0.1% PBST) for 1 h at 37 • C and then at 4 • C overnight. The primary antibodies were mouse or rabbit anti-Myo7a (1:500 dilution; Proteus Biosciences, Ramona, CA, USA), goat anti-Sox2 (1:300 dilution, Santa Cruz Biotechnology, Dallas, TX, USA), and rabbit anti-P2X7 receptor (1:500 dilution; Sigma-Aldrich). After washing with PBS on the following day to remove the unbound antibodies, the specimens were incubated with the appropriate Alexaconjugated secondary antibodies (diluted in 0.1% PBST) for detection at 4 • C overnight. Appropriate Alexa-conjugated secondary antibodies were used for detection. DAPI (1:800 dilution; Sigma-Aldrich) was used to label the cell nuclei.
Apoptotic cells in the organ of Corti were detected by a terminal deoxynucleotidyl transferase (TdT)-mediated dUTP-biotin nick endlabeling (TUNEL) assay according to the manufacturer's protocol. Briefly, after washing with PBS the samples were incubated with TUNEL reaction mixture (11,684,817,910, Roche, Switzerland) in a humidified chamber at 37 • C for 60 min and finally labeled with DAPI to visualize the nuclei. For the negative control, TdT was omitted from the above procedure and replaced with PBS. A positive control was performed by adding 20 U DNase I (10 U/ml dissolved in 50 mM Tris-HCl buffer, pH 7.4) for 1 h to induce DNA strand breaks prior to the labeling procedures.
Tissues were mounted in anti-fade fluorescence mounting medium and coverslipped. Cochleae were dissected into the apical, middle, and basal turns, and images were taken using a Zeiss LSM 710 confocal microscope (Zeiss, Germany). All images were processed using Image J and Adobe Photoshop CC.

Cell counts
For HC quantification in the neomycin-treated samples, we imaged the entire cochlea using a 40× objective and counted the Myo7a + HCs that remained. The same procedure was used to quantify TUNEL+/ Myo7a + and Mito-SOX Red+/Myo7a + cells. For all experiments, only one cochlea from each mouse was used for immunofluorescence and quantification. Thus, n represents the number of mice examined.

Statistical analyses
All data are expressed as means ± SEM. ABR thresholds were analyzed by two-way ANOVA followed by a Newman-Keuls post-hoc test. Western-blotting and immunofluorescence analysis was performed with a two-tailed, unpaired Student's t-test when comparing two groups or with one-way ANOVA followed by a Dunnett's multiple comparisons test when comparing more than two groups. p < 0.05 was considered statistically significant.

Neomycin triggers ATP release and the upregulation of P2 receptor in the immature cochlea in vitro
We used luciferase-based reactions to quantify ATP concentrations in the cochlear explants from neonatal mice after neomycin exposure. The luminescence of ATP in the neomycin-treated PND2 mouse cochleae in vitro increased over the first 1.5 h, remained relatively stable until about Fig. 1. Changes in ATP and its receptors after neomycin exposure. A) The luminescence of eATP during the first 4 h with neomycin or vehicle added to the culture media. The luminescence was increased in the first 1.5 h and then kept at a relative high level until 3 h after the addition of neomycin. The luminescence decreased rapidly from 3 h to 4 h. B) Protocol for RNA-Seq after neomycin-induced mouse cochlear HC damage in vivo and C) the log-transformed expression of all P2 family ATP receptors between neomycin (orange) and intact (blue) mice. * indicates a significant difference. D) Protocol for P2rx7 and P2X7 receptor examination after neomycin damage in vitro. Immunostaining of Myo7a and P2X7 receptor in E-E2) controls and F-F2) after neomycin exposure for 6 h. The immunofluorescence signal for P2X7R was increased in the neomycin-treated sample. G-H) P2rx7 mRNA levels were highest at the specified point in time of 3 h (G), while protein levels were highest at 6 h after neomycin exposure (H), and both returned to baseline at 12 h. **p < 0.01 vs. control. I) Quantification of the relative expression level of P2X7 receptor after neomycin exposure. The data are presented as the mean ± standard error of the mean, *p < 0.05. HC: hair cell; Scale bar is 10 μm in E. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 3 h after neomycin exposure, and then decreased back to baseline. In the untreated control sample, the luminescence of ATP remained at a low level the entire time (Fig. 1A). These results indicated that neomycin could induce robust ATP release from the cells in the cochlear explant.
To further understand the effects of neomycin on purinergic receptor expression in vivo, we measured the gene expression of the P2 receptor family, including the P2X and P2Y subfamilies. The mice received two injections of either neomycin or saline on PND7 and PND8 and then were sacrificed to isolate the cochleae for RNA-Seq on PND9 (Fig. 1B). We found that three P2 receptor family genes, P2rx3, P2rx7, and P2ry4, were differentially expressed between these two groups (Fig. 1C), which indicated that P2 receptor families might be important targets of neomycin or might be involved in the process of HC damage. Interestingly, both P2rx3 and P2rx7 were increased while P2ry4 was decreased in the neomycin group.
Previous studies have shown that P2X7 receptor protein can be observed in the cochlear HCs in the rodent neonatal period (Nikolic et al., 2003), while P2X3R expression is specific to developing spiral ganglion neurons and their neurite projections to HCs, not in HCs themselves (Huang et al., 2005;Huang et al., 2006). In addition, the level of P2ry4 was too low to be measured by qRT-PCR. Therefore, we next focused on the P2X7 receptor to investigate its role in neomycintriggered ATP increases in the inner ear. We first observed a similar P2X7 receptor expression pattern in the cochlea as in the brain as shown by western blotting, and qRT-PCR analysis showed similar P2rx7 mRNA expression in the cochlea and brain (Supplementary Figure 1A1-A2). The brain tissue served as the positive control.
When neomycin was added to the culture media for PND2 cochleae in vitro (Fig. 1D), the immunostaining density of P2X7 receptor increased only mildly after neomycin exposure for 6 h when compared WT cochleae marked with antibodies against Myo7a (HC marker) and P2X7 receptor. The P2X7R was expressed mainly in the hair cells in PND2 cochleae. C-C2) Whole mount view of a PND30 cochlea marked with antibodies against Myo7a (C1) and P2X7 receptor (C2). P2X7 receptor was almost absent from the cells in PND30 cochleae. D1) ABR thresholds of PND30, D2) PND60, and D3) PND90 mice in vivo. There was no significant difference between P2rx7− /− mice and WT mice in terms of ABR thresholds. The data are presented as the mean ± standard error of the mean. E1-E3) Apical, middle, and basal turns of the cochlea from PND90 WT mice and F1-F3) P2rx7− /− mice. There was no apparent difference between WT and KO mice in both hearing function and morphology of inner ear cells. G) Quantification of HC numbers in all three turns between WT and P2rx7− /− mice. The data are presented as the mean ± standard error of the mean. Scale bar is 20 μm in A, C, and E1 and 10 μm in B.
with vehicle control (Fig. 1E-E2, 1F-F2). We found that at the specified point in time we observed (including 1, 3, 6 and 12 h after neomycin damage in Fig. 1G), the P2rx7 mRNA expression level in the cochlear explants increased at 3 h and then decrease at 6 and 12 h after neomycin damage (Fig. 1G), while the P2X7 receptor protein as expressed at the highest levels at about 6 h after damage, which was longer than the period over which P2rx7 mRNA increased ( Fig. 1H-I). These results suggest that P2X7 receptor is dynamically expressed according to ototoxicity factors such as aminoglycosides and that there is a time-gap between mRNA and protein expression.

The ABR thresholds and sensory cells in P2rx7− /− mice were indistinguishable from wild-type (WT) mice
To determine the cell types expressing P2rx7, we collected cochleae from PND2 to PND 30 mice and performed whole mount or cryosection immunostaining with antibodies against P2X7 receptor and Myo7a, which is a marker for HCs. Using a pepsin antigen retrieval-based procedure, we found that the P2X7 receptor was co-localized with Myo7a, suggesting that P2rx7 was expressed in HCs in the cochlea in WT PND2 mice and was hardly expressed in WT PND30 mice ( Fig. 2A-C), while in the neonatal P2rx7− /− cochleae there were no P2X7 receptor-positive cells in the organ of Corti, which was in accordance with the genotyping results ( Supplementary Fig. 1B-C) and confirmed that P2rx7 was completely knocked out in the inner ear.
We further measured the Myo7a expression level in the cochleae of PND90 P2rx7− /− mice. There was no apparent HC loss in the P2rx7− /− mice, and there were no differences in the total HC number between P2rx7− /− mice and WT control mice, which indicated that the absence of P2X7 receptor did not affect hearing (t-test, p = 0.267, p = 0.226, and p = 0.208 for the apical, middle, and basal turns, respectively; n = 3-4 mice) ( Figure 2E1-E3, F1-F3, G).

Neomycin had little effect on hearing thresholds or HC numbers in P2rx7− /− mice in vivo in the developing cochlea
Neomycin or saline was subcutaneously injected daily into WT mice and P2rx7− /− mice from PND7 to PND14 (Fig. 3A) to induce hearing loss following previously published procedures (Sun et al., 2014), and we measured hearing function using pure tone ABR thresholds in the neomycin-treated and control WT and P2rx7− /− mice on PND30. ABR thresholds increased more in the neomycin-treated WT mice (90 dB SPL) than WT mice treated with PBS and P2rx7− /− mice treated with neomycin or with PBS. Two-way ANOVA revealed a significant effect of neomycin treatment (F 1,25 = 4579, p < 0.001) and a significant interaction between neomycin treatment and frequency (F 5,2 5 = 26.32, p < 0.001) in WT mice. However, there was no effect in the P2rx7− /− mice either with neomycin or with PBS (neomycin treatment: F 1,43 = 1.427, p = 0.2389; interaction between neomycin treatment and frequency: F 5,43 = 0.2746, p = 0.9245) (Fig. 3B).
All mice were sacrificed on PND30 after ABR testing, and the cochleae were collected for further analysis. We found that the number of remaining HCs in WT mice treated with neomycin was significantly lower in all three turns of the cochlea compared to the other three groups (two-way ANOVA, F 3, 36 = 2129, p < 0.0001, n = 5-6 mice in each group), while there was no significant difference between P2rx7− / − mice treated with neomycin or saline ( Figure 3C1-C3, D1-D3; E1-E3; F1-F3; Tukey's multiple comparisons test, apex: p = 0.9385; middle: p = 0.7028; base: p = 0.7193). The results are quantified in Fig. 3G.
To further analyze the role of P2X7 receptor in neomycin-induced cell apoptosis in vivo, we injected neomycin or saline into the WT mice or P2rx7− /− mice daily from PND7 to PND14 and harvested the cochleae on PND17 (Fig. 3A). We found that there were many TUNELpositive cells in the WT mice treated with neomycin, while there were few TUNEL-positive cells in the other groups (WT control vs. WT neomycin, Mann-Whitney U test, p = 0.0179, Fig. 3H-H2, I-I2, J-J2, K-K2). Negative and positive controls from the auditory sensory epithelium are shown in Supplementary Fig. 2. There was a significant difference between WT (neomycin) mice and the other groups (p < 0.01 when the P2rx7− /− neomycin group was compared with the WT neomycin group by unpaired t-test; n = 4-5 mice) (Fig. 3L).
Similarly, the effect of P2X7 receptor knockout in protecting against neomycin-induced cell death was confirmed in vitro as shown in Supplementary Fig. 3. The effective concentration of neomycin was determined by testing a gradient from 1 mM to 5 mM, and the results are shown in Supplementary Fig. 4.

Fluorescein-conjugated gentamycin was hardly taken up by HCs in neonatal P2rx7− /− mice in vitro
To determine the mechanism through which loss of P2X7 receptor protects HCs from neomycin ototoxicity, we used GT594 to trace HC uptake in vitro. At PND2, the cochleae of neonatal WT, P2rx7+/− , and P2rx7− /− mice were dissected out and cultured. After a 12 h recovery period, the cultured tissues were treated with 2 mM neomycin and GT594 for 6 h to damage the HCs because the protein level of P2X7 receptor peaked at 6 h after culturing with neomycin, and the tissues were then harvested and collected for Myo7a and DAPI labelling. The GT594 molecular probe accumulated in most of the HCs from the apex to the base of the cochlea in WT and P2rx7+/− mice, especially in the middle and basal turns ( Figure 4A1-A3, a, B1-B3 and b), while only a few scattered HCs in the P2rx7− /− mice were positive for GT594 ( Figure 4C1-C3, c).
The number of remaining HCs in WT and P2rx7+/− mice treated with neomycin was significantly lower in the middle and basal turns of the cochlea compared to the P2rx7− /− mice treated with neomycin (two-way ANOVA, the difference between genotype, F 4, 104 = 9.184, p < 0.0001, n = 6-11 samples of each turn per group), while there was no significant difference in the apical turn between groups (Tukey's multiple comparisons test, Apex: p = 0.87, WT vs. Similarly, there was a significant difference in the GT594+ cells between these groups (two-way ANOVA, the difference between genotype, F 4, 87 = 63.35, p < 0.0001, n = 6-11 samples of each turn for each group; Turkey's multiple comparisons test, Apex: p = 0.99, WT vs  Figure 4D1 and D2. In addition, to determine whether the ATP concentration levels are altered by neomycin damage in P2rx7− /− PND2 neonatal mice, we measured the ATP concentration in the transgenic mice. The ATP concentrations in P2rx7− /− mice in vitro were similar to those of WT mice, and two-way ANOVA revealed no significant effect of neomycin treatment in the absence of P2X7 receptor (ATP concentration: F 1,10 = 0.7296, p = 0.4130, P2rx7− /− + neomycin vs WT + neomycin) or interaction between neomycin treatment and time (ATP concentration: F 10,50 = 1.397, p = 0.2092), while a significant effect was seen between neomycin treatment (F 2, 90 = 56.26, p < 0.0001) and the interaction between neomycin treatment and time (ATP concentration: F 10, 90 = 13.32, p < 0.0001). These results indicated that lack of P2X7 receptor did not alter the amount of ATP released but did prevent gentamycin from entering into HCs (Fig. 4E).
We next investigated whether GT594 can enter HCs when P2X7 receptors are activated by high concentrations of ATP or the ATP analog 2,3-O-(4-benzoyl-benzoyl) ATP (BzATP). BzATP is more potent than ATP and evokes a higher maximum current (Young et al., 2007). With higher concentration of BzATP (1 mM or 3 mM), we found many more GT594+ cells colocalized with Myosin7a in neonatal WT mice, while there were nearly no GT594+ cells in the P2rx7− /− mice (Supplementary Fig. 5). These results again demonstrated that activation of P2X7 receptor by high concentrations of ATP allows large molecules to pass through the HC membrane and enter the cells.

Knockout of P2X7 receptor had little impact on the resting membrane potential of both inner HCs and outer HCs in the first neonatal week
To determine if loss of the P2X7 receptor alters the resting membrane potential over the course of development or is involved in the pathophysiology of aminoglycoside ototoxicity, we harvested cochlear explants at PND2 and PND5 to record resting membrane potentials from both inner HCs and outer HCs from the apical and middle turns. For inner HCs, the amplitudes of the resting membrane potentials of WT and P2rx7− /− mice at both PND2 and PND5 did not show any significant differences ( Figure 4F1-F2, G1-G2; WT at PND2-45.64 ± 1.3 mV vs. P2rx7− /− at PND2-46.33 ± 1.2 mV, p = 0.728, n = 11 and 6, respectively; WT at P5-52.39 ± 1.3 mV vs. P2rx7− /− at PND5-55.5 ± 2.2 mV, p = 0.235, n = 13 and 12, respectively). The averaged amplitude of the resting membrane potential for WT outer HCs at PND2 was − 46.15 ± 1.3 mV, n = 13 and for P2rx7− /− was − 49.85 ± 0.7 mV, n = 13, and even though the difference was statistically significant (p = 0.022), it was only 3.7 mV on average. Similar to what was observed in inner HCs at PND5, no significant difference was detected for resting membrane potentials of outer HCs at PND5 for both WT and P2rx7− /− mice (WT at PND5-50.40 ± 1.3 mV vs. P2rx7− /− at PND5-49.85 ± 0.7 mV, p = 0.396, n = 13 and 10, respectively).
MET function was examined using FM1-43× dye. The fluorescence intensity in the WT mouse cochlear samples was similar to samples from P2rx7− /− mice, which indicated that knockout of P2X7 receptor had little effect on MET channel function ( Supplementary Fig. 6).

Loss of P2X7 receptor reduced mitochondria-mediated oxidative stress and inhibited NLRP3-mediated pyroptosis in vitro
We measured ROS levels to explore the mechanism through which lack of P2rx7 protects HCs against neomycin-induced damage in vitro. Mito-SOX Red, which produces red fluorescence when it is oxidized by mitochondrial superoxide, was used to measure ROS production in the mitochondria. The cochleae of neonatal mice were dissected out and cultured at PND2. After a 12 h recovery period, the cultured tissues were treated with 2.0 mM neomycin for 6 h to damage the HCs and then harvested and collected for Myo7a and Mito-SOX Red labelling or for There is no significant difference between WT and KO mice both treated with neomycin. The data are presented as the mean ± standard error of the mean. ***p < 0.001 vs. WT (Control); ###p < 0.001 vs. P2rx7+/− . F-G) Membrane potential of inner HCs and outer HCs in WT and P2rx7− /− at PND2 and PND5. Representative recordings for WT and P2rx7− /− at PND2 and PND5 from inner HCs (F1) and outer HCs (F2). Pooled data for each group are shown in G1 and G2. There was a slight difference between membrane potentials in WT and KO mice at PND2. Data are shown as the mean ± standard error of the mean. *p < 0.05 vs WT (Control). Scale bar is 30 μm in A1, A2, B2, and C2 and 20 μm in a, b, and c.
qRT-PCR of ROS-related genes (Fig. 5A). The Mito-SOX Red level was dramatically increased in WT cochleae cultured with neomycin compared with controls, and very little Mito-SOX Red signal was detected in the P2rx7− /− mice even when cultured with neomycin ( Fig. 5B-B3, C-C3, D-D3, E-E3), and these differences were statistically significant (one-way ANOVA, F 3, 10 = 279.4, p < 0.01, Fig. 5F). In addition, the expression level of ROS-related genes, including Duox2 and Alox15, was significantly increased in the WT samples treated with neomycin but not in the WT control or the P2rx7− /− samples (p < 0.001 vs. WT neomycin group, Fig. 5G). These results indicated that P2X7 receptor is involved in neomycin-induced increases in ROS.
To further explore the mechanisms underlying the role of P2X7 receptor in neomycin-induced HC apoptosis, we analyzed cytokines and related receptors involved in inflammation. The nucleotide-binding and oligomerization domain (NOD)-like receptors (NLRs) are members of a large family of soluble and cellular receptors that are involved in a variety of responses to different stimuli. Recently, it was shown that eATP activates the NLRP3 inflammasome through P2X7 receptor and that P2X7 receptor-dependent IL-1β secretion requires inflammasome activation (Yue et al., 2017;Karmakar et al., 2016). NLRP3 was increased in the both WT and P2rx7− /− neomycin-treated mice, but it was barely detectable in the WT control group. The NLRP3 protein level was increased about 2-fold when comparing the P2rx7− /− neomycin group with the P2rx7− /− control group, but the level in the P2rx7− /− neomycin group was less than that in the WT neomycin group ( Figure 5H1-H2). qRT-PCR showed that the NLRP3 and IL-1β levels were dramatically increased in the neomycin-treated WT group, while they were slightly increased in the neomycin-treated P2rx7− /− group G) The expression level of ROS-related genes as determined by qRT-PCR. Only Duox2 and Alox15 were increased in the neomycin-treated WT mice. The data are presented as the mean ± standard error of the mean, ***p < 0.001, WT (Ctr) vs. WT (Neo); ### p < 0.001, WT (Neo) vs. P2rx7− /− (Neo). H1) NLRP3 levels and H2) the quantification of NLRP3 in WT and P2rx7− /− mice with and without neomycin. The data are presented as the mean ± standard error of the mean, **p < 0.01, WT (Ctr) vs. WT (Neo); # p < 0.05, WT (Neo) vs. P2rx7− /− (Neo). I) Expression levels of NLRP3 and IL-1β were strongly increased in neomycin-treated WT mice, while in the KO mice with neomycin, NLRP3 and IL-1β were only slightly increased compared to the intact group. The data are presented as the mean ± standard error of the mean, ***p < 0.001, WT (Ctr) vs. WT (Neo); ### p < 0.001, WT (Neo) vs. P2rx7− /− (Neo). Scale bar is 20 μm in B. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) (p < 0.05 in Figure 5H2 when comparing the neomycin-treated P2rx7− / − group with the WT control group and p = 0.134 in Fig. 5I when comparing the neomycin-treated P2rx7− /− group with the WT control group).

Discussion
AGAs, including gentamicin, neomycin, kanamycin, and tobramycin, are broad-spectrum antibiotics used for treating suspected infections. They are polycationic molecules with molar masses between 300 and 600 g/mol with a maximal cross-sectional diameter of ~0.8 nm (Supplementary Table 2). In mammals, AGAs are selectively toxic to sensory HCs of the inner ear, especially during fetal development and childhood, and it has been demonstrated that AGA-induced ototoxicity is dose dependent in animal models and in humans (Kros, 2019). Several transmembrane mechanisms, including endocytosis by megalin (Moestrup et al., 1995;Chun et al., 2021b), ion channels, and MET channels, might physiologically modulate the trafficking of AGAs, but the actual mechanisms through which AGAs specifically accumulate only in sensory HCs in the inner ear remain unknown.
Here we demonstrate that P2X7 receptor, a member of the ATP receptor family, might be required for the entry of AGAs into the HCs of the developing cochlea. We used P2rx7− /− mice to confirm our hypothesis that abolishing P2rx7 can prevent neomycin-induced HC loss and hearing loss.

P2X7 receptor is present in the neonatal cochlea, and its expression is mildly increased by neomycin-induced ATP release
P2X7 receptor, which is a member of the ATP receptor family, has been reported to be expressed predominantly in cells of the hematopoietic lineage, including macrophages and microglia, and its function changes due to N-and C-terminal splice variants (Di Virgilio and Vuerich, 2015;Kurashima et al., 2012). In the inner ear, P2X7 receptor is distributed mainly in the spiral ganglion and in the neural projections associated with both inner and outer HCs, but it is also expressed in HCs in the organ of Corti and in the vestibular organ in neonatal rodents (Zhao et al., 2005;Brandle et al., 1999;Munoz et al., 2001). Its subunits assemble to form ATP-gated ion channels (Nikolic et al., 2003), and it has been suggested that P2X7 receptor over-expression might be a damage signaling event because the damage-induced increase in P2X7 receptor depends on the release of extracellular ATP (Linden et al., 2019;Adinolfi et al., 2018).
ATP exists in the cochlear endolymph and perilymph and underlies the mechanically induced Ca 2+ waves in a number of tissues, including the cochlea (Newman, 2001;Gale et al., 2004). Under physiological conditions the cochlear endolymph and perilymph contain nanomolar amounts of extracellular ATP, and it has been found that cochlear ATP is mainly released from cochlear supporting cells via gap junction hemichannels (Zhao et al., 2005). In the local area near the cell surface, the ATP concentrations can be high and can reach micromolar levels (Zhu and Zhao, 2010). Under pathological conditions, such as noise or damage, the endolymphatic ATP levels increase, and these levels can reach as high as 20 nM after only a 15 min exposure to 110 dB SPL (Munoz et al., 2001).
In our study, we found that P2rx7 was slightly increased according to the RNA-Seq results in neomycin-treated young mice, and the upward trend was further confirmed by qRT-PCR. Meanwhile, the P2X7 receptor protein level was also increased by neomycin damage. According to the RNA-Seq results, only three members of the P2 receptor family changed significantly. However, among them only P2X7 receptor was mainly expressed in the cell membrane of HCs in the cochlea, while the others were hardly detected in the cochlea or in HCs. Following neomycin insult, extracellular ATP concentrations were dramatically increased within the first 3 h, followed by P2X7 receptor activation at about 6 h. We hypothesize that high ATP concentration may activate the P2X7 receptor and increase the uptake of aminoglycoside antibiotics by HCs. In in vitro experiments, we further confirmed the role of ATP through the application of exogenous ATP or BzATP. We observed that with low concentrations of BzATP of 3-20 μM there were no GT594+ cells.
However, when the concentrations of BzATP were increased to higher than 100 μM, the numbers of GT594+ cells were increased in a dosedependent manner. However, with higher concentrations of BzATP than 3 mM, most of the HCs died within 6 h. These results indicated that neomycin can trigger inner ear cells to release ATP leading to increased intercellular concentrations, which further activates the P2X7R receptor as the "death receptor".
It was notable that at 12 h after neomycin insult, both P2rx7 mRNA and P2X7 receptor were decreased to levels lower than those at 1 h and in controls. We suppose that this might be due to the death of HCs. P2rx7 mRNA is produced by HCs, and when HC death is induced by neomycin the damaged HCs cannot synthesize P2X7 receptor mRNA or protein, which causes the decrease or even total loss of P2rx7 expression.

P2X7 receptor deficiency has little effect on auditory function
As mentioned previously (Sperlagh and Illes, 2014), P2X7 receptor shows two distinct functions in other systems, and it can act as a lowconductance channel that allows the passage of small ions across the membrane or as a high-conductance channel (large-pore channel) that allows the uptake of high molecular weight organic cations. However, the mechanism through which P2X7 receptor switches roles in different tissues or organs remains unknown. In the present study, we observed that the deletion of P2X7 receptor does not affect the physiological status of the cochlea, including ABR thresholds and cochlear structures. Furthermore, we also noted that the resting membrane potential did not change much in the HCs of the P2rx7− /− mice compared with WT mice. The difference in membrane potential between WT and P2rx7− /− mice was only about 3.7 mV in outer HCs in PND2 mice ( Figure 4G2). To our knowledge, it is known that membrane potential changes in HCs might cause neurotransmitter release and further initiate electrical signals transmit to spiral ganglion neurons (Cunningham and Muller, 2019). It is difficult to judge whether or not the slight difference influenced HC functions. However, the consequence of the change may result in preventing AGAs from entering outer HCs. Therefore, further investigation should be done to determine the relationship between P2X7 receptor absence and membrane potential alteration in the future.
P2rx7− /− mice have previously been reported to be normal for many parameters under physiological conditions, for example, the size of fontanelles at birth (Panupinthu et al., 2008), the hematopoietic parameters in the bone marrow microenvironment (Adamiak et al., 2018), and the colonic mucosa (Hofman et al., 2015). In the brain, there are no differences in microglial activation or in the degree of severe MCAOinduced brain damage (the size of the infarct) between WT and P2rx7− /− mice (Ji et al., 2012). In addition, P2X7 receptor deficiency alone does not alter Type 1 diabetes development in nonobese diabetic background mice (Chen et al., 2011). In our study, there was little difference in the pharmacokinetics of AGAs in either the kidney or the inner ear between WT and P2rx7− /− mice.
In some studies, however, it has been reported that there are some functional changes in P2rx7− /− mice. A recent study showed that either genetic deletion or sub-chronic pharmacological inhibition of the P2X7 receptor leads to a decrease in whole-body energy expenditure and a concurrent increase in carbohydrate oxidation (Coccurello and Volonte, 2020). P2X7 receptor deletion also increases the expression of glutamine synthase and ASCT2 (a glutamine:cysteine exchanger) but diminishes the efficacy of N-acetylcysteine (a glutathione precursor) at the glutathione level (Park and Kim, 2020).
Nevertheless, although P2X7 receptor deficiency does not alter hearing thresholds in adult mice, the molecular mechanisms of the P2X7 receptor are very complex and remain a highly debated issue (Sperlagh and Illes, 2014). Furthermore, we failed to find any differences in FM1-43 uptake capacity between neonatal WT and knockout mice, which indicated that the absence of P2X7R may not affect MET channels in the immature cochleae. Therefore, the function of the P2X7 receptor in response to HC damage from AGA exposure and its role in AGA ototoxicity, such as its potential role to be a gateway for AGAs to enter the HCs, need to be more thoroughly investigated. The functional maturation process of HCs involves the orderly expression of various channels, including TMC2 (in the early postnatal mouse cochlea) (Lelli et al., 2009;Nakanishi et al., 2018), TMC1 (from PND3 and rising to a plateau at PND12) (Zheng and Holt, 2021), and PIZEO2 (in developing HCs) . The expression of P2X7 receptor in developing HCs may indicate a potential role in HC function. In addition, it remains unknown whether P2X7 receptor influences MET channels. Thus, further experiments should be performed to verify the role of P2X7 receptor.

P2X7 receptor may be linked to the entry of AGAs into HCs in the immature cochlea
According to a previous study (Alves et al., 2014), we hypothesized that under normal physiological conditions P2X7 receptor is not functionally engaged and only small molecules and a few ATP molecules can pass through the channel. However, after neomycin exposure high micromolar concentrations of ATP (up to millimolar concentrations) are dumped into the extracellular space. It was confirmed in our experiments that extracellular ATP reaches its peak concentration within 1-2 h after neomycin exposure, and such a high concentration is sufficient to activate ATP receptors, including P2X7 receptor (Pelegrin and Surprenant, 2006;Bond and Naus, 2014;Hu et al., 2015;Lahne and Gale, 2008). The functional consequences of P2X7 receptor activation are more important than changes in receptor expression, and these changes are important drivers of disease processes because the opening of the large pore can result in membrane blebbing and cell death. There are two possible mechanisms through which P2X7 receptor transforms from a channel to a pore. The first is the progressive dilation of the P2X7 receptor-gated channel itself, and the second involves the recruitment of an additional pore-forming protein, most likely the pannexin-1 hemichannel (Panx1). Both mechanisms have substantial experimental support (Sperlagh and Illes, 2014).
Pannexin channels are key mediators of extracellular ATP release (Crespo Yanguas et al., 2017), and caspase and PANX1-dependent lysosomal exocytosis plays an essential role in ATP release such as that triggered by immunogenic chemotherapy (Martins et al., 2014). Thus, ATP release is the result of pannexin channel opening, and the resulting abundant ATP release might be the reason why the P2X7 receptor becomes activated and allows the influx of neomycin or other AGAs.
It has also been reported that damage to cochlear HCs triggers changes in intracellular calcium signaling via Ca 2+ waves (Gale et al., 2004;Piazza et al., 2007). When ATP is released from damaged cells, this triggers the mobilization of Ca 2+ and the subsequent release of ATP from supporting cells (Gale et al., 2004), perhaps via connexin hemichannels (Zhao et al., 2005). Thus, cochlear supporting cells act similarly to glial cells because astrocytes also release ATP in response to mechanical stimulation (Newman, 2001;Lahne and Gale, 2008).
Activation of P2X7 receptor is associated with immune responses. ATP is released from the cytosol of damaged cells into the extracellular space via pannexin or connexin channels, and abundant extracellular ATP, including high concentrations of exogenous BzATP or ATP, can activate P2 receptors, including P2X7 and P2X2 receptors. Also, the P2X7 receptor downstream of caspase 11 plays a critical role in pyroptosis in mice (Yang et al., 2015), and P2X7 receptor is a powerful activator of the NLRP3 inflammasome (Franceschini et al., 2015). In the cochlea, Nlrp3 is expressed in CX3CR1-positive cells (Nakanishi et al., 2017), which are the microglia-like cells that we previously reported are present in the damaged cochlea .
We also noticed that in the P2rx7− /− mice some HCs were missing in situ in both the middle and basal turns. However, GT594 was not detected in the neighboring cells. This suggests that AGAs might first only damage a limited number of HCs without entering into cells. Furthermore, in the WT mouse cochlea it appeared that GT594 was localized in both the cytoplasm and the nucleus, which indicated that AGAs can enter the cell nucleus thus further damaging the HCs.
Although P2X7 receptor is very important in the process of HC death in the immature cochlea, ablation of P2rx7 delays the entrance of GT594, which suggests that there should be other mechanisms through which AGAs enter HCs together with P2X7 receptor activation. However, P2X7 receptor might be a primary barrier to high molecular weight molecules from entering the HCs. In the cochleae of the P2rx7− /− mice, HCs survived better than in the control WT mice, while GT594 accumulated much slower than in controls. In addition, even with higher concentrations of exogenous BzATP, GT594 was not found in any HCs in the P2rx7− /− mouse cochlea. Thus, there might be more complex mechanisms involved, and these should be investigated further.

Conclusion
In summary, our study provides evidence that the P2X7 receptor is required for the penetration of AGAs into HCs in the developing cochlea. Neomycin damage induces ATP release from sensory epithelium cells, and the resulting high concentrations of ATP trigger the expression or activation of P2X7 receptor. Activated P2X7 receptor appears to aggravate AGA-induced hearing loss by damaging cochlear HCs. The potential explanation for this is that HC death might be due to the fact that P2X7 receptor increases the penetration of high-molecular weight moleculesincluding neomycin and GT594 -into HCs. Deletion of P2rx7 can reduce HC loss and protect hearing after neomycin exposure. Therefore, P2X7 receptor might represent a novel therapeutic target for reducing HC damage and preventing hearing loss from the use of AGAs.

Ethical approval and consent to participate
Not applicable

Consent for publication
Not applicable.

Availability of data and materials
The datasets generated during or analyzed during the current study are available from the corresponding author on reasonable request.

Funding
This work was supported by the National Natural Science Foundation of China (Nos. 81970879, 81830029, 81900931, 82192860, 82192862, 82171141), the Shanghai Science and Technology Committee (STCSM) Science and Technology Innovation Program (20MC1920200, 21JC401000), Research Projects of Shanghai Municipal Health Committee (2020YJZX0110 and 2022XD059) and the Outstanding Youth Program of Nanjing Municipal Health Commission (JQX20003).