PARP inhibitor rescues hearing and hair cell impairment in Cx26‐null mice

GJB2 (encoding connexin26, Cx26) mutation is the most common genetic cause of hereditary deafness. Cochlear sensory hair cell (HC) death is the core pathologic phenomenon of GJB2‐related deafness. However, mechanism‐based therapy is still obscure. A targeted‐cell conditional Gjb2 knockout mouse model was established in which the Cx26 in Deiters cells and pillar cells were knocked out at birth. We explored the mechanism of HC death caused by deficiency of GJB2 in supporting cells (SCs), which exhibited moderate deafness, early HC death without SCs, and a decrease in distortion product otoacoustic emission. Here, a DNA damage response was observed in HCs during the onset of hearing loss. Apoptosis, necroptosis, and ferroptosis do not contribute to the death of HCs. We identified and demonstrated that parthanatos, a poly‐(ADP‐ribose) polymerase (PARP)‐dependent and caspase‐independent form of cell death, is a mechanism of HC degeneration. Furthermore, we observed that the use of PARP inhibitors effectively alleviated both deafness and HC loss. Our study reveals the specific mechanism of HC death caused by GJB2 deficiency. These findings demonstrate that targeting parthanatos is an HC protective strategy in the prevention of GJB2‐related hereditary deafness.

of HC death due to GJB2 deficiency has not been fully elucidated.
Protective strategies based on cell death mechanisms can effectively rescue different kinds of deafness.2][13] However, in recent years, several novel death modes such as necroptosis, ferroptosis, and parthanatos have been proposed successively.In the noise-induced deafness model, both apoptosis and necroptosis are implicated in HC death, and inhibition of either death pathway shifts HC death to the other pathway. 14Co-occurrence of apoptosis and necroptosis has also been reported in cisplatin and aminoglycoside ototoxicity models, where the administration of inhibitors of both death pathways alleviated HC degeneration in vivo. 15,16Moreover, it has been demonstrated that ferroptosis plays a key role in cisplatin-induced HC loss and that inhibition of ferroptosis protects HCs from death. 17,18Therefore, to prevent GJB2-related HC death, the specific death mechanism must be investigated.
The absence of a suitable mouse model is the main limitation in exploring the mechanism of HC death caused by GJB2 deficiency.In some phenotypically severe mouse models, such as Otog-Cre; Cx26 loxP/loxP , Sox10-Cre; Cx26 loxP/loxP , and Rosa26-CreER; Cx26 loxP/loxP mouse models, the tunnel of Corti and Nuel space fail to open up during cochlear development, accompanied by HC and SC damage or other multiple pathological manifestations. 9,10,19Although HC death is the main cause of deafness, other pathological phenomena in the above model may interfere with the exploration of the mechanism of HC death.In contrast, other mouse models showed delayed hearing loss and mild histological impairments.In Prox1-CreER; Cx26 loxP/loxP mice, Cx26 is specifically knocked out in Deiters cells (DCs) and outer pillar cells (OPCs), and mice show reduced distortion product otoacoustic emission (DPOAE) without significant HC loss. 20n Lgr5-CreER; Cx26 loxP/loxP mice, Cx26 is specifically knocked out in the third row of DCs (DC3), and mice showed corresponding loss of the third row of outer HCs (OHC3) and delayed deafness at high frequency. 21Furthermore, the deletion of cochlear Cx26 did not disrupt the survival of HCs or spiral ganglion neurons (SGNs) for at least 1 month when a mosaic knockout pattern was induced in mouse cochlea 22 or when cochlear Cx26 was knocked out at the mature stage. 23,24In the above model, HCs were locally dead or not dead, and the deafness was very slight; consequently, this model was not suitable for the study of cell death mechanisms.
Currently, the main treatment for hereditary deafness in clinical practice is wearing hearing aids or cochlear implants.Other treatments, such as gene therapy and stem cell transplantation, have proliferated in recent years.
6][27] In addition, functional Cx26 gap junction-forming cells have also been established from induced pluripotent stem cells, contributing to the creation of optimal treatments for various mutations in Gjb2associated deafness.However, the above development is still at the stage of in vitro study. 280][31] Therefore, exploring the mechanism of HC death caused by GJB2 deficiency may identify new targets for intervention.
Fibroblast growth factor receptor 3 (Fgfr3) transcripts are expressed by radial glial stem cells in the embryonic brain and spinal cord and astrocytes in the postnatal central nervous system.The Fgfr3-iCreERT2 line facilitates further studies of the neurogenesis of olfactory bulb, as well as the astrocytes of the central nervous system.In addition, this line may be used to study tissues that normally express Fgfr3, such as developing cartilage and bone. 32,33In the cochlea, Fgfr3-iCreERT2 is specifically expressed in DCs and pillar cells (PCs) (Figure S2).The induction of tamoxifen (TMX) given at P2-3 in mice resulted in 100% tdTomato positivity in DCs and PCs. 34,35herefore, we achieved targeted DC and PC knockout of Cx26 in the inner ear by Fgfr3-iCreERT2 and performed phenotypic observation.
In the present study, a Cx26-null mouse model with HC damage as the main pathological feature was established.Detailed auditory function examination and cochlear sensory epithelium damage assessment were performed in this mouse model.To reveal the mechanism of HC death, we validated different cell death pathways during the critical period of sensory cell damage in this model.We found DNA damage located in dying HCs and consequently the activation of poly-(ADP-ribose) polymerase-1 (PARP-1)-induced HC death via parthanatos.In addition, PARP inhibitors significantly rescued the Gjb2-deficiencyinduced HC death and hearing loss.Our finding suggests a possible mechanism of HC death caused by Cx26 defects in SCs and provides a potential target for therapeutic strategies to treat GJB2-related hereditary deafness.

Mouse model and genotyping
The Cx26 loxP/loxP mice were provided by Prof. Xi Lin of Emory University, Atlanta, GA, USA, and the Fgfr3-iCreERT2 were provided by Prof. Zhiyong Liu from the Chinese Academy of Sciences, Shanghai, China.ROSA26CAGloxP-stop-loxP-tdTomato (tdTomato) mice were provided by Prof. Ren-Jie Chai at Southeast University.Sexually mature Cx26 loxP/loxP mice were selected and crossed with Fgfr3-iCreERT2 mice to obtain Fgfr3-iCreERT2; Cx26 loxP/WT first-generation mice.These Fgfr3-iCreERT2; Cx26 loxP/WT offspring were then crossed with each other to obtain Fgfr3-iCreERT2; Cx26 loxP/loxP mice.Fgfr3-iCreERT2; Cx26 loxP/loxP mice were used as the experimental group and their Cx26 loxP/loxP littermates were used as the control group.The strain was maintained on a CBA/J/Ola (CBA) genetic background.TMX (total dose of 1.5 mg/10 g body weight) was injected subcutaneously on two consecutive days, once on the day of birth (P0) and again on the first day after birth (P1), to specifically activate Cre recombinase and achieve targeted Cx26-null in early cochlear PCs and DCs.Genotyping of mice was performed after polymerase chain reaction (PCR) amplification of tail tissue genomic DNA, as previously described. 10The primer sequences used are listed in Table S1.Mice were given adequate breeding feed and high-pressure filtered water and maintained under a constant temperature (

In vivo drug administration
Mice were randomly divided into different groups.Males and females were not restricted.Mice of the Cx26-null group were given a daily subcutaneous injection of olaparib (Cat# S1060, Selleck Chemicals; 50 mg/kg), rucaparib (Cat# S4948, Selleck Chemicals, 50 mg/kg), Fer-1 (Cat# S7243, Selleck Chemicals, 5 mg/kg), or vehicle at P10-P21.All inhibitors were dissolved in DMSO as storage solution at maximum solubility and diluted with corn oil.

Auditory brainstem response and DPOAE measurement
Auditory brainstem response (ABR) was measured in mice at P22, P30, P45, and P60.After weighing the mice, a compound anesthetic, ketamine hydrochloride (120 mg/kg) mixed with chlorpromazine hydrochloride (20 mg/kg), was injected intraperitoneally.Measurements were made in a special quiet room while the mice were placed on a heating pad at 38 • C under anesthesia.A closed sound field was used.The electrical response was recorded using three sterilised fine subdermal electrodes: the positive recording electrode (+) was inserted subdermally in the middle of the cranial vault, and the negative (-) and ground electrodes were inserted subdermally in the ear to be tested and in the posterior mastoid of the contralateral ear, respectively.Click or tone burst acoustic stimulation was selected, and the tone burst test frequencies were 4, 8, 16, 32, and 40 kHz.The electrical signal scan time was set to 10 ms, and the acoustic stimulation frequency was 21 times/s, with 1024 electrical signal superimpositions.Reactions were recorded using the Tucker-Davis Technology System (RZ6, Tucker-Davis Technologies).The stimulation sound started at 90 dB and decreased at intervals of 10 dB, and when the stimulation sound intensity approached the hearing threshold of mice, it decreased at intervals of 5 dB.For the measurement of DPOAE, the test frequencies were presented as a geometric mean of f 1 and with a ratio of f 1 :f 2 = 1:1.2.The level of the f 2 tone was set to 10 dB lower than that of the f 1 tone.Distortion products were recorded on average 200 times.At each level, the amplitude of the DPOAE at 2f 1 -f 2 was measured.Mice with otitis media were excluded from the experiment.

Real-time quantitative polymerase chain reaction of the cochlea
Total RNA was isolated from cochleae using an RNAprep pure tissue kit (Tiangen Biotech Co. Ltd.), and RNA concentration and purity were determined using a Nan-oDrop 1000 spectrophotometer (NanoDrop).RNA was reverse-transcribed using a PrimeScript RT reagent kit with gDNA eraser (Takara Bio Inc.).Real-time PCR was performed with SYBR Green PCR mix on a Roche Light-Cycler 480 instrument.The 2 −△△Ct method was applied to determine relative mRNA expression, where the relative abundance of each gene was internally normalised to the geometric mean Ct for the reference gene.The primers are listed in Table S1.

Cochlear tissue preparation and immunofluorescent labelling
Cochlear dissection was performed on euthanised mice at the indicated time-points.The temporal bone was harvested and carefully dissected.A small incision was made at the apical tips for proper fluid exchange during immersion fixation with 0.01 M phosphate-buffered saline (PBS) for 1 h at room temperature and subsequently decalcified in ethylene diamine tetraacetic acid (EDTA) for 48 h.For frozen sections, the decalcified cochlear tissues were placed in a gradient sucrose solution for dehydration by immersion in 10%, 20%, and 30% sucrose solutions for 1 h each and then embedded in optimal cutting temperature compound overnight at 4 • C.These samples were sectioned at 10 μm for subsequent procedures.For flattened cochlear preparations, cochleae were dissected into three turns (apical, middle, and basal) under the microscope.The sections or flattened cochlear preparations were blocked in 10% donkey serum with 0.1% Triton X-100 solution for 1 h at room temperature, then incubated overnight at 4 Semi-quantification of immunocytochemical signals was used to assess the expression of proteins of interest.Each image was captured under the same conditions at 60× magnification, followed by quantitative analysis using ImageJ software (National Institutes of Health).Cochlear surface preparations were all stained with phalloidin (red) or DAPI (blue) to identify the OHCs or OHC cell nuclei in the confocal images.The intensity of the background was subtracted and then the average grayscale intensity was calculated for each cell.A total of 48-60 OHCs from four or five individual mice were counted for each region.Relative fluorescence was quantified by normalising the ratio of average fluorescence of target cells in the Cx26-null group to that in the control group.

Counting of HCs and DCs
To obtain a cochleogram, approximately 2100 OHCs were counted in each stretched cochlear preparation in 13 contiguous areas from the apical to the basal portion (0%-100% distance from the apex).The last part of the cochlear basilar membrane is unstable and difficult to obtain, so the portion counted covered almost 94% of the entire basilar membrane.For DC counts, DCs were labelled with SOX2 and a laser scanning confocal microscope was used to capture 60× magnification images at 25%, 50%, and 75% of the area of the flattened preparation.

Statistical analysis
All data are presented as the means ± SEMs and were graphed using GraphPad Prism (Version 8.2.1, GraphPad Software Inc.).When only one factor was involved, oneway analysis of variance (ANOVA) followed by Dunnett's multiple comparison test was used.When two factors were involved, a two-way ANOVA multiple comparison test was used.A p-value <.05 was considered statistically significant.

Establishment of a mouse model for targeted knockout of Cx26 in DCs and PCs of the cochlea
The Fgfr3-iCreERT2; Cx26 loxP/loxP mice were subjected to targeted knockout of Cx26 in DCs and PCs by TMXinduced Cre recombinase.Our previous results showed that the tdTomato signal was mainly observed in DCs and PCs in this mouse model at P7, demonstrating that Cx26 can be successfully knocked out in these cells. 34Consistent with the validation results, there was no expression of Cx26 in PCs or DCs in the cochleae of target Cx26-null mice, and the knockout pattern of Cx26 in the apical-middle-basal area was uniform and consistent (Figures 1A and S1A).Otherwise, the Cx26 expression in the lateral wall or spiral limbus was not affected (Figure 1A).It was observed that Cx30 was normally expressed in DCs and PCs in the cochleae of the Cx26-null mice (Figures 1B and S1B).The flattened cochlear preparations showed that Cx26 labelling was absent from DCs and PCs, in line with Cre expression in the Cx26-null mice (Figure 1C).The cochlear basilar membrane of P7 mice was harvested for protein quantification, and Western blotting (WB) showed that cochlear Cx26 expression in the Cx26-null group was 71.4 ± 19.0% of that in the control group (p < .05,n = 4, Figure 1D,E).
DPOAE was assessed at P30.The decrease in active cochlear amplification was more pronounced at higher frequencies in the Cx26-null group.At 28 kHz, DPOAEs were reduced by 15.3 ± 8.0 dB, 27.9 ± 5.2 dB, 31.0 ± 4.5 dB, 19.0 ± 2.6 dB, and 24.1 ± 5.1 dB in Cx26-null mice compared with the control group at stimulation levels of 25, 35, 45, 55, and 65 dB SPL, respectively (Figure 2H-J).The reduction in DPOAE at 8 and 16 kHz was mainly reflected in the 65 dB stimulation level, with reductions of 9.8 ± 8.0 dB and 17.6 ± 10.2 dB, respectively.Mice of the Cx26-null group also showed a reduction at 25 and 55 dB stimulation levels at 8 kHz, yet the decrease was not obvious (Figure 2H).

Cell loss patterns of HCs and DCs in Cx26-null mice
Quantitative analysis of OHCs and DCs at P22 and P60 revealed cell loss in the targeted Cx26-null mice starting from the basal turn and progressing upward.At P22, cochleograms showed that 45.8%-77.3% of OHC loss could be seen in the basal turn of the cochlea of the targeted Cx26-null group.The middle turn exhibited a sporadic and lateral loss of OHCs, with a loss proportion of 12%-14% (Figure 3A,B).In contrast, there was no loss of inner HCs (IHCs) or DCs (Figure 3B-D).At P60, cell degeneration in targeted Cx26-null mice was aggravated and progressed towards the middle turn.HC counts revealed a loss of 70.7%-85.9% of the OHCs in the basal turn and plaquelike deficiency in the middle turn, while IHCs were also lost in the basal turn (Figure 3E,F).The percentage of surviving DCs in the basal turn was reduced to 23.4 ± 11.0% (p < .001)(Figure 3G,H).

Reduction in ribbon synapses in targeted Cx26-null mice
SGNs in the middle and basal turns of the cochleae appeared degenerated in the targeted Cx26-null group at P60, which was reported in our previous study. 34tBP2 immunostaining (green) revealed the pattern of immunofluorescent staining of ribbon synapses in mice at P60 (Figure 4A,B).The results showed a decrease in the number of ribbon synapses in the IHCs of the targeted Cx26-null group (apex: 15.2 ± 1.4 vs. 12.1 ± 1.6, p = .019;middle: 14.7 ± 0.7 vs. 12.3 ± 0.9, p = .006,n = 5) (Figure 4C,D).

Apoptosis and necroptosis induced by RIP3 may not be involved in HC death in targeted Cx26-null mice
We next investigated the expression of apoptosis-related indicators in the cochleae at P22.The expression of caspase-3, Bax, Bcl2, and cleaved-caspase-3 at both transcriptional (mRNA) and protein levels in targeted Cx26null mice showed no differences from that in the control group (Figure 5A,C,D).Necroptosis is dependent on RIP1 and RIP3 (also known as RIPK1 and RIPK3); therefore, RIP3 was selected for validation.The WB results showed that there was no significant difference in the band density of RIP3 between the experimental and control groups (Figure 5B,E).Likewise, using the same antibody, immunofluorescence quantification revealed that RIP3 expression in OHCs of the basal turn was also not altered significantly compared to the control group (Figure 5F,G).

Ferroptosis may not be the primary mechanism of HC death in targeted Cx26-null mice
Given the known association between HC death and ferroptosis in other ototoxicity models, 17,18 we investigated whether ferroptosis occurred in OHCs of the targeted Cx26-null group.First, we quantified the mRNA of ferroptosis-related genes (Cox2, Gpx4, Fth1, Fth, Ftl, Tfr1, Tfr2, Slc7a11, Slc7a5, and Asl4).Expression of Cox2, Slc7a11, and Asl4 in the targeted Cx26-null group was decreased compared with that in the control group, while other ferroptosis-related indicators were unchanged (Figure 6A).Second, quantitative WB analysis showed no significant difference in the expression of GPX4 or COX2 in cochlear protein extracts between the control and experimental groups (Figure 6B,C).Immunostaining of cochlear tissue confirmed the GPX4 results of WB analysis (Figure 6D-F).Meanwhile, targeted Cx26-null mice were subcutaneously injected with Fer-1, a ferroptosis inhibitor, between P10 and P21 to observe whether there was any phenotypic improvement (Figure 6G).However, Fer-1 did not provoke any changes in hearing impairment in targeted Cx26-null mice (Figure 6H).Similarly, there was no protection against OHC loss (Figure 6I).

DNA damage response was found in targeted Cx26-null mice
To verify the DNA damage in cochlear OHCs of Cx26-null mice, we labelled and localised two DNA damage repair indicators, γH2AX and 53BP1.The fluorescence quantification level of γH2AX in OHCs was 2.6 times higher in the Cx26-null group than in the control group (p < .001,n = 60, 12 OHCs in each of five mice) (Figure 7A,C).In addition, 53BP1 is a central regulator of double-strand break signals and can be used as a marker of DNA double-strand damage.There was almost no 53BP1 expression in the OHCs in control mice, whereas the relative expression of 53BP1 was elevated 3.6-fold in the cochleae of the Cx26-null mice (p < .001,n = 48, 12 OHCs selected from each of four mice) (Figure 7B,D).Taken together, these results indicated DNA damage in OHCs of targeted Cx26-null mice.

Parthanatos contributes to OHC loss in Cx26-null mice
Activation of PARP-1 is the initiating stage of the pathological response.The expression of PARP-1 was analysed by WB, and the results showed that the relative expression of PARP-1 in the cochleae of targeted Cx26-null mice increased 2.0-fold (p < .001,n = 4) compared with that of the control group, while the expression of cleaved-PARP-1 also increased (Figure 8A,D).PAR, a product of PARP-1, can be used as an indirect marker of PARP-1 activity.WB results showed a 1.7-fold (p < .001,n = 4) increase in PAR levels in the Cx26-null mice (Figure 8B,D).As Apoptosis-inducing factor (AIF) nuclear translocation follows PARP-1 activation, we measured the level of AIF protein in nuclear protein extracts and found a 1.7-fold (p = .047,n = 4) increase in AIF expression in the Cx26-null mice (Figure 8C,E).Concomitantly, anti-PAR immunofluorescence in the cochleae was intense and consistent with PAR staining in the OHCs of the Cx26null mice as well as higher levels of AIF expression in the nuclei (Figure 8F,G).Quantification by apex-middlebase fluorescence revealed that PAR-and AIF-positive OHCs exhibited a significant base-to-apex negative gradient, which was consistent with the progression of OHC degeneration in targeted Cx26-null mice (Figure 8H,I).

PARP inhibitor rescues hearing and hair cell impairment in Cx26-null mice
To further confirm the involvement of parthanatos in the process of OHC death due to Gjb2 deficiency, we investigated whether different PARP inhibitors (olaparib and rucaparib) had a rescue effect on damaged HCs and hearing loss.Targeted Cx26-null mice were injected daily with subcutaneous olaparib or rucaparib from P10 to P21, and ABR and OHCs were counted in all the mice at P21-P22 (Figure 9A).First, to confirm the validity of PARP inhibitors, we examined AIF nuclear translocation expression in the basal turn of the cochlea.The AIF immunofluorescence in general showed a clear reduction in the expression of AIF in the treatment (olaparib or rucaparib) groups compared to the targeted Cx26-null group, confirming the efficacy of the PARP-1 inhibitory treatment (Figure 9B-D).

DISCUSSION
In this study, a targeted Cx26-null model was established, which is a suitable model for exploring HC death.Up to now, most conditional Gjb2 knockout mice have exhibited severe hearing loss and cochlear epithelial degeneration.In Otog-Cre, CAG-Cre, Sox10-Cre, Foxg1-Cre, Pax2-Cre, or Rosa26-Cre mice, extensive cochlear Cx26 deletion resulted in rapid degeneration of the cochlear HC and SC during P13-P14, and subsequent SGN loss at P30. Cochlear HC and SC death is so rapid that part of the cochlear epithelium completely disappears and only a layer of the basilar membrane remains.This suggests that Cx26 plays multiple roles in the survival of different cochlear cells. 9,19,35,36However, our recent study showed that targeted DC3 Cx26 knockout induces OHC3 death without DC3 loss. 21This phenomenon suggests that HC injury may be the earliest and most critical pathological process of Gjb2-related cochlear epithelial injury.Cx26 is not expressed in HCs; therefore, OHC3 death in this line may be caused by cellular dysfunction of Cx26-null DC3s.These findings further suggest that HCs and SCs respond differently to cochlear Cx26 knockout.HCs and SCs may have distinct death mechanisms, and many different death patterns of cochlear cells may occur simultaneously in traditional severe cochlear phenotype models.In this study,  we targeted the knockout of Cx26 between DCs and PCs by establishing an Fgfr3-Cre; Cx26 loxP/loxP knockout mouse model, and the resulting mice exhibited early HC loss with moderate hearing impairment.It is worth mentioning that Prox1-creER mice also knocked out Cx26 in PCs and DCs, but there was no significant HC death. 20The reason for this is that prox1 has a limited initiation efficiency in the inner ear, being expressed in 72% of DCs at E16 but present in only 5%-10% of DCs and 5% of PCs at P0-P1. 37In contrast, Fgfr3-iCreERT2 is expressed in 100% of DCs and PCs at P0-P1 (Table 1).The establishment of this model will facilitate further investigation of the mechanism of how the deletion of Cx26 in SCs leads to the corresponding HC impairment, as well as exploration of the treatment of Gjb2-related deafness.We systematically investigated different death pathways that may be involved in cochlear HC degeneration, and found that Gjb2-related HC death was dominated by parthanatos.Our study suggests that apoptosis, necroptosis, and ferroptosis may not be the main death pathway for HC degeneration in this line.Parthanatos is a PARP-1-dependent and caspase-independent form of cell death. 38In this experiment, no elevation of caspase-3 was found in the targeted Cx26-null group, but increased DNA damage was observed in HCs, which in turn caused activation of PARP-1, accumulation of PAR, and migration of AIF into the nucleus.0][41] In particular, the oxidative stress theory suggests that dysfunction of Cx26 impedes the transport of reactive oxygen species (ROS), leading to an accumulation of excess ROS in HCs and an increase in oxidative stress, which in turn leads to massive loss of HCs and SCs. 41It is widely believed that oxidative stress may lead to irreversible DNA damage. 42,43In the inner ear, the ROS-induced DNA damage response drives cellular aging and contributes to accelerated age-related hearing loss. 44PARP-1 is one of the critical mediators of DNA damage repair.This process has a pivotal role in several pathophysiological processes, such as stroke, trauma, diabetes, and Parkinson's disease. 45,46PARP-1 overactivation has also been associated with neurodegenerative diseases, including currently untreatable blinding retinitis pigmentosa and hereditary retinal photoreceptor degeneration. 47,48In the inner ear, this mode of death has been found to be present in streptomycin-or cisplatin-induced ototoxicity. 49,50Our results demonstrate that in a Gjb2-deficient mouse model, cell death occurs in the presence of DNA damage to HCs, which in turn activates PARP-1 causing them to undergo parthanatos.
To date, no valid treatment is available for Gjb2associated hearing loss.A gene therapy study of Gjb2deficiency-associated hereditary deafness showed that virus-mediated gene therapy reduced HC loss in the middle-base turn of the cochlea by 18%-38% without any improvement in hearing. 25Another study showed that perinatal Gjb2 gene transfer partially restored HCs in the base turn, with PC height restored to about 50% of normal and hearing improved by 20-30 dB at 12 and 24 kHz frequencies. 26Recently, in a phenotypically less severe Gjb2 knockout mouse model, adeno-associated virus-mediated Gjb2 gene transfer did not result in any threshold improvement and even exacerbated the hearing loss, resulting in HC loss. 27In this study, PARP inhibitor restored 10-20 dB at 24−40 kHz frequencies, and HCs at the corresponding site recovered up to 50%, which is also an important step in the treatment of Gjb2-deficiencyassociated deafness.PARP inhibitors are analogues of ADP-ribosyl transferases with β-nicotinamide adenine dinucleotide (NAD+) that competitively inhibit the active site of NAD+ with the catalytic domain of PARP-1, resulting in a failure of recruitment to DNA damage-associated repair proteins and interference with their process of restoration of DNA damage. 51In this study, we chose olaparib and rucaparib administered in vivo to inhibit the parthanatos and found that they had a salvage effect on hearing and HC damage in targeted Cx26-null mice, with rucaparib being more effective than olaparib.We speculate that the reasons for this include the following: first, the molecular weight of rucaparib (323.36) is lower than that of olaparib (434.46), which may make it easier for rucaparib to cross the blood-labyrinth barrier; second, Rudolph et al. 52 found that rucaparib has a significantly higher affinity for the active site than olaparib and has a slower release rate and longer duration of action.In recent years, many PARP inhibitors have been tested in clinical trials, primarily for cancer treatment, and a wealth of human tolerability and efficacy data is now available.The use of the inhibitor resulted in hearing recovery of up to 20 dB in the knockout group of mice, which is a great breakthrough in GJB2-deficiency-associated genetic deafness.It is worth noting that the application of PARP inhibitors in vivo still faces the problem of sustained transport to the inner ear.The effect of PARP inhibitors will no longer be significant after a certain period according to previous studies. 53Therefore, in the future, it is important to identify drug transport vehicles that allow for long-term drug release in the inner ear and enable the lowest possible application.
In conclusion, we report that targeted deletion of Cx26 in DCs and PCs results in early OHC loss, deafness, and decreased DPOAE.We then found that DNA damageinduced parthanatos mediates HC death in this line, while in vivo data demonstrate that selective inhibition of PARP-1 may be a viable therapeutic approach to rescue deafness and HC loss in the targeted Cx26-null mouse model.These results suggest that drug therapy based on the mechanism of HC death can effectively rescue murine Gjb2-related hereditary deafness.From a clinical perspective, these findings suggest that treatment of GJB2-related hereditary deafness by inhibiting PARP-1 may help prevent the deafness phenotype caused by parthanatos.Future studies are needed to test the potential clinical significance of this treatment strategy.

A U T H O R C O N T R I B U T I O N S
Yu Sun and Weijia Kong conceived and designed the experiments.Material preparation, data collection, and analysis were performed by Xiaohui Wang and Sen Chen.The first draft of the manuscript was written by Xiaohui Wang.All authors commented on previous versions of the manuscript.All the authors have read and approved the final manuscript.

A C K N O W L E D G E M E N T S
We are grateful to Prof. Xi Lin of Emory University, Atlanta, GA, USA, and Prof. Zhiyong Liu from the Chinese Academy of Sciences, Shanghai, China, for kindly providing mouse lines.The author would like to thank Dr. Yi Lin for her contribution to this article.This work was supported by the National Natural Science Foundation of China (81470696, 82071058, and 81500793) and the National Key Research and Development Program of China (No. 2021YFF0702303).

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declare they have no conflicts of interest.

D ATA AVA I L A B I L I T Y S TAT E M E N T
All data are available in the main text or the Supporting Information.

F I G U R E 1
Targeted deletion of Cx26 in Deiters cells (DCs) and pillar cells (PCs) in targeted Cx26-null mice.(A) Immunofluorescent staining of Cx26 (green) in the apical, middle, and basal turns of the control group and Cx26-null group.Scale bars = 100 μm.Magnified images of the organ of Corti (OC) in the middle turn from the yellow boxes of the corresponding groups.White asterisks (*) indicate the absence of Cx26 in the DCs and PCs of the Cx26-null group.Scale bars = 20 μm.(B) Immunofluorescent staining of Cx30 (green) in the apical, middle, and basal turns of the control group and Cx26-null group.Magnified images of the OC in the middle turn from the yellow boxes.Cx30 labelling remained in the Cx26-null group.Scale bars = 20 μm.(C) Immunofluorescent labelling of the cochlear sensory epithelium for Cx26 (red) in whole-mount preparations.Cx26 labelling is absent in the three rows of DCs, inner pillar cells (IPCs), and outer pillar cells (OPCs) of the Cx26-null group.Scale bars = 20 μm.(D and E) Western blot and histogram showing Cx26 expression in the cochlear epithelium of the control and Cx26-null groups at P7 (N = 4 in each group).Data are expressed as mean with SEM; *p < .05,significantly different from the control group.C, control group; E, Cx26-null group; LW, lateral wall; OHCs, outer hair cells; SLi, spiral limbus; SV, stria vascularis.F I G U R E 2 Auditory brainstem responses (ABRs) and distortion product otoacoustic emissions (DPOAEs) in the targeted Cx26-null mice at different ages.(A-D) Hearing thresholds of the control and Cx26-null group at P22, P30, P45, and P60.(E-G) Age-related changes in hearing thresholds of the mice at 16, 24, and 32 kHz frequencies, respectively.(H-J) DPOAE levels were measured at P30 and the audiograms are presented at f 0 = 8, 16, and 28 kHz (f 2 /f 1 = 1.2 and the level of f 2 set 10 dB below that of f 1 ).The dashed line indicates the noise floor.Data are expressed as mean with SEM; *p<.05, **p < .01,***p < .001,significantly different from the control group, N = 5 in each group.SPL, sound pressure level.

F I G U R E 3
Hair cell (HC) and Deiters cell (DC) loss patterns of the targeted Cx26-null group at P22 and P60.(A) Representative images of HCs in apical, middle, and basal turns from the control and Cx26-null group at P22. White asterisks indicate the absence of outer hair cells (OHCs) in the middle or basal cochlea of the Cx26-null group.Scale bars = 30 μm. (B) Quantification of HC loss at specific cochlear locations in the control and Cx26-null groups at P22. (C) Representative images of DCs (SOX2, white) in different regions from the control and Cx26-null groups.Scale bars = 30 μm. (D) Quantification of DC survival at specific cochlear locations in the control and Cx26-null groups at P22. (E) Representative images of HCs in apical, middle, and basal turns from the control and Cx26-null group at P60. Scale bars = 30 μm. (F) Quantification of HC loss at specific cochlear locations in the control and Cx26-null groups at P60. (G) Representative images of DCs (SOX2, white) in different regions from the control and Cx26-null groups.Scale bars = 30 μm. (H) Quantification of DC survival at specific cochlear locations in the control and Cx26-null groups at P60.Data are expressed as mean with SEM; ***p < .001,significantly different from the control group, N = 4-6 in each group.F I G U R E 4 The pattern of ribbon synapses in Cx26-null mice at P60. (A and B) The pattern of ribbon synapses (CtBP2, green) in inner hair cells (myosin7a, red) of control and Cx26-null groups at the apical and middle turns.Scale bars = 10 μm.(C and D) Statistics of the number of ribbon synapses per inner hair cell (IHC) in apical and middle turns, respectively.Data are expressed as mean with SEM; ***p < .001,significantly different from the control group, N = 5 in each group.

F I G U R E 5
Apoptosis and necroptosis may not be involved in hair cell (HC) death in Cx26-null mice.(A and D) Western blot and histogram of caspase-3, Bcl-2, Bax, and cleaved-caspase-3 in cochlear tissue of control and Cx26-null mice at P22 (N = 4 in each group).(C) Changes in mRNA expression levels of caspase-3, Bax, Bcl-2, and Bak in the cochlea from control and Cx26-null groups at P22 (N = 6 in each group).(B and E) Western blot and histogram of receptor-interacting protein (RIP)3 in cochleae of control and Cx26-null mice at P22 (N = 4 in each group).(F) Immunofluorescent staining of RIP3 (green) in the basal turn of the control and Cx26-null group.Scale bars = 20 μm.(G) Quantification of immunolabelling for RIP3 in outer hair cells (OHCs) from the control and Cx26-null group (48 OHCs from four mice in each group).Data are expressed as mean with SEM; NS: non-significant.C, control group; E, Cx26-null group.

F I G U R E 6
Ferroptosis may not be the primary mechanism of hair cell (HC) death in Cx26-null mice.(A) Changes in mRNA expression levels of ferroptosis-related genes in the cochleae from control and Cx26-null groups at P22 (N = 6 in each group).(B and C) Western blot and histogram of COX2 and GPX4 in cochleae of control and Cx26-null mice at P22 (N = 4 in each group).(D) Modiolar sections showing immunolabelling of GPX4 (green) of organs of Corti (OCs) from the control and Cx26-null mice at P22. Scale bars = 20 μm.(E) Immunofluorescent labelling of whole-mount preparations showed the presence of GPX4 (green) in the basal turn of the control and Cx26-null mice at P22. Scale bars = 20 μm.(F) Quantification of immunolabelling for GPX4 (green) in outer hair cells (OHCs) from the control and Cx26-null group (36 OHCs from three mice in each group).(G) Diagram showing the administration mode of the ferroptosis inhibitor Fer-1.The Cx26-null mice were administered Fer-1 subcutaneously, daily from P10-P21.(H) Auditory brainstem response (ABR) thresholds were not significantly changed in the Fer-1-treated mice (N = 5 in each group).(I) OHC loss was not significantly rescued in the Fer-1-treated mice (N = 5 in each group).Data are expressed as mean with SEM; *p < .05,significantly different from control group; NS: non-significant.C, control group; DCs, Deiters cells; E, Cx26-null group.

F I G U R E 7
Elevated DNA damage response in Cx26-null mice.(A) Immunofluorescent staining of γH2AX (green) in the basal turn of the control and Cx26-null groups.Scale bars = 10 μm.(B) Immunofluorescent staining of 53BP1 (green) in the basal turn of the control and Cx26-null groups.Scale bars = 10 μm.(C) Quantification of immunolabelling for γH2AX in outer hair cells (OHCs) from the control and Cx26-null groups (60 OHCs from five mice in each group).(D) Quantification of immunolabelling for 53BP1 in OHCs from the control and Cx26-null group (48 OHCs from five mice in each group).Data are expressed as mean with SEM; ***p < .001,significantly different from the control group.at 30 and 40 kHz.More significantly attenuated hearing loss at both 30 and 40 kHz was observed in the rucaparib administration group (Figure 10D,F).

F I G U R E 8
Parthanatos may contribute to outer hair cell (OHC) loss in Cx26-null mice.(A, B, and D) Western blot and histogram of poly-(ADP-ribose) polymerase-1 (PARP-1), cleaved-PARP-1, and PAR in cochlear tissue of control and Cx26-null mice at P22 (N = 4 in each group).(C and E) Western blot and histogram of AIF in cochlear nucleoprotein extract of control and Cx26-null mice at P22. Lambin-1 was used as a nuclear loading control (N = 4 in each group).(F) Immunofluorescent staining of PAR (green) in the basal turn of the control and Cx26-null group.Scale bars = 20 μm.(G) Immunofluorescent staining of AIF (green) migration into the nucleus in the basal turn of the control and Cx26-null group.Scale bars = 20 μm.(H) Quantification of immunolabelling for PAR in different regions from the control and Cx26-null groups (36 OHCs from three mice in each group).(I) Quantification of immunolabelling for AIF in different regions from the control and Cx26-null groups (60 OHCs from five mice in each group).Data are expressed as mean with SEM; *p < .05,***p < .001,significantly different from control group; NS: non-significant.C, control group; E, Cx26-null group.F I G U R E 9 Poly-(ADP-ribose) polymerase (PARP) inhibitor administration decreased the expression of AIF in outer hair cells (OHCs) of the inner ear.(A) Diagram showing the administration mode of the PARP inhibitors olaparib and rucaparib.Cx26-null mice were injected subcutaneously with tamoxifen (TMX) at P0 and P1 and administered with vehicle or PARP inhibitors daily from P10-P21.All of the mice were tested for auditory brainstem response (ABR) and sacrificed at P21-P22.(B) Molecular structure of olaparib and rucaparib.(C) Immunofluorescent staining of AIF (green) migration into the nucleus in the basal turn of the Cx26-null + vehicle and Cx26-null + inhibitors groups of mice.Scale bars = 20 μm.(D) Quantification of immunolabelling for AIF in the basal turn of the Cx26-null + vehicle and Cx26-null + inhibitors administration groups (36 OHCs from three mice in each group).

F I G U R E 1 0
Poly-(ADP-ribose) polymerase (PARP) inhibitors rescue hearing and outer hair cell (OHC) loss in Cx26-null mice.(A and B) Diagram showing the administration mode of the PARP inhibitors olaparib and rucaparib.Cx26-null mice were injected subcutaneously with tamoxifen (TMX) at P0 and P1 and administered with vehicle or PARP inhibitors daily from P10-P21.All of the mice were tested for auditory brainstem response (ABR) and sacrificed at P21-P22.(C and E) Changes in ABR thresholds at different frequencies in Cx26-null + vehicle and Cx26-null + inhibitor groups of mice.Blue and red lines represent olaparib and rucaparib, respectively (N = 8 in the olaparib group and N = 5 in the rucaparib group).(D and F) Comparison of the changes in ABR thresholds at 30 and 40 K with olaparib and rucaparib, respectively (N = 7 in each group).(G and H) Representative images of the middle-basal parts of flattened cochlear preparations in different groups.White arrowheads indicate the regions of OHC loss.Scale bars = 200 μm.(I and J) Quantifications of OHC loss at specific cochlear locations in Cx26-null + vehicle and Cx26-null + inhibitor groups of mice.Blue and red lines represent olaparib and rucaparib, respectively (N = 5 in each group).Data are expressed as mean with SEM; *p < .05,**p < .01,***p < .001;NS: non-significant.
Mouse lines with GJB2 mutation.