The Disorder of Mitochondrial Dynamics Causes Deafness By Promoting Macrophage-Mediated Hair Cell Death

Mitochondrial dynamics are essential for maintaining the physiological function of the mitochondrial network, and the disorder of mitochondrial dynamics leads to neurodegenerative diseases. However, how mitochondrial dynamics affects auditory function in the inner ear remains unclear. FAM73a and FAM73b are mitochondrial outer membrane proteins that mediate mitochondrial fusion. Here, we found that FAM73a or FAM73b deciency resulted in elevated oxidative stress and apoptosis of hair cells. Additionally, mitochondrial ssion also causes an increase expression of IL-12 in basilar membrane macrophages through accumulating IRF1. As a bridge between innate and adaptive immune responses, hyperproduction of IL-12 further promoted the polarization of Th1 and tissue damage. Our data highlighted an important role of mitochondrial dynamics in maintaining cochlear homeostasis and hair cell survival. Mitochondrial dynamics not only disturbed hair cell function, but also induced the disorder of immune responses.


Introduction
Hearing loss is one of the most common human diseases. According to 2018 World Health Organization report, there are about 466 million people with hearing disability globally, as 5% of the world's total population (https://www.who.int/en/). More than 0.1 percent newborns suffer from hearing loss, which seriously affects a child's communication, quality of life, and educational attainment [1] . Among all deafness patients, approximately 90% suffer from sensorineural hearing loss (SNHL), which is mainly caused by the loss or damage of cochlear hair cells (HCs) and the degeneration of spiral ganglion neurons after HCs injury. [2] SNHL can be caused by genetic and environmental factors, and genetic defects are responsible for at least 50% of congenital or childhood hearing loss [3] . The identi cation of the genes associated with hearing loss and their underlying mechanisms remains as an urgent but challenging task.
Mitochondria are essential for the physiological function and survival of HCs, and its dysfunction is involved in the pathogenesis of hearing loss under noise exposure, ototoxic drug treatment, or aging [4] .
The disorders of mitochondrial ssion and fusion can lead to abnormal morphology and dysfunction, which is implicated in neurodegenerative diseases [5] , but the role of mitochondrial dynamics in auditory function has not been extensively investigated. Some evidences have ever shown that dynamin-related protein-1 (Drp-1) is necessary for mitochondrial ssion and plays a central role in mitophagy [6,7] . Reduced Drp-1 expression and mitophagy are involved in age-related hearing loss [8] . Similarly, optic atrophy 1 (OPA1) controls mitochondrial inner membrane fusion, and the R445H mutation in the Opa1 causes SNHL [9] . Studies have shown that OPA1 is expressed in HCs and spiral ganglion neurons, and its mutation-caused deafness is due to a functional change in the unmyelinated auditory nerve endings and not to a pathological change in HCs [10][11][12] . FAM73a and FAM73b are mitochondrial outer membrane proteins that are required for mitochondrial outer membrane fusion [13][14][15] . The de ciency of FAM73a and FAM73b greatly disrupted the mitochondrial morphology, thus it led to higher levels of reactive oxygen species (ROS) and signi cant reductions in ATP. However, it remains unclear whether the absence of FAM73a and FAM73b also contribute to deafness similar to Drp-1 and OPA1.
In recent years, immune response and in ammation have also been recognized as one of important pathophysiological factors in HC injury [16] . As a main executor in innate immune system of the cochlea, macrophages are widely distributed in the basilar membrane, the osseous spiral lamina, the lateral wall of the cochlea, and in spiral ganglions under physiological conditions [17] . Adaptive immunity is also considered to be involved in the cochlear immune response [18] . CD4 + T-cells can in ltrate into the basilar membrane and collaborate with macrophages [19] . The resident macrophages of the basilar membrane are activated and produce proin ammatory cytokines, and the monocytes in peripheral circulation also enter the basilar membrane and transform into macrophages in response to noise exposure, ototoxic drug damage, and age-related degeneration [20][21][22][23] . In the models of cochlear injury, HC injury is considered to be the initiator of the immune response, and is su cient to regulate macrophage recruitment into the basilar membrane through fractalkine signaling [24] . The inhibition of the activation and recruitment of macrophages is protective against HC injury caused by ototoxic drugs [25] . FAM73b is involved in macrophage polarization, and regulate the production of IL-12 in response to damage, which further results in increased production of IFN-γ in T cells [26] . Therefore, we investigated the functional changes in macrophages and T cells in the basilar membrane of the cochlea by using Fam73a and Fam73b knockout (KO) mice, indicated by the expression of in ammatory cytokines and the associated essential signal pathways.
In this study, we found that FAM73a and FAM73b were expressed in the mitochondria of HCs, and Fam73a and Fam73b KO resulted in HC loss and the destruction of stereocilia structures. Deletion of FAM73a and FAM73b increased the oxidative stress level and apoptosis in HCs. Along with the genetic ablation of FAM73a and FAM73b, the numbers of macrophages and CD4 + T cells in the cochlear basilar membrane were increased, and the expression of the in ammatory cytokines IL-12 and IFN-γ were obviously elevated. After activated by endogenous damage, mitochondrial division in macrophages led to increased expression of Parkin, which degraded the mono-ubiquitinated CHIP protein and stabilized the protein level of downstream transcription factor IRF1, thereby promoting the secretion of IL-12. IL-12 further directly promotes the production of IFN-γ in T cells, which is involved in HC injury. Our data highlighted the role of FAM73a and FAM73b-mediated mitochondrial dynamics in auditory function and clari ed the effect of their disorders on HC survival by directly disturbing HC function and indirectly regulating macrophage polarization.

Mice and genotyping
Genotypic identi cation of transgenic mice was carried out according to the method described in published literature [26] . All animal experiments were performed in accordance with the protocols approved by the Animal Care and Use Committee of Southeast University and were consistent with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Auditory brainstem response (ABR)
A TDT System III workstation running SigGen32 software (Tucker-Davis Technologies, USA) was used to record ABRs as previously described [27,28] . The mice were anesthetized by intraperitoneal injection of 0.01 g/ml pentobarbital sodium (100 mg/kg body weight). After deep anesthesia, three ne-needle electrodes were inserted under the skin of the mouse at the vertex of the skull, behind the tested ear, and on the back near the tail. The mice were then put into a soundproof room for the ABR test. The TDT hardware and software (BioSig and SigGen) were used to generate the acoustic signals and to process the responses. The ABRs were elicited with tone bursts at 4, 8, 12, 16, 24, and 32 kHz. The tests were performed at a 5 dB interval from 90 dB to 10 dB at each frequency with a gradually decreasing intensity, and ABR thresholds were recorded as the lowest sound intensity at which a stable wave III could be seen and repeated.

Immunohistochemistry
The basilar membranes of the newborn mouse cochleae were dissected with microsurgery forceps and incubated with 4% paraformaldehyde for 1 h at room temperature (RT), while the basilar membranes from ossi ed cochleae were carefully dissected with a microsurgery scalpel after incubating in 4% paraformaldehyde and 0.5 M EDTA overnight. Whole mounts of the basilar membrane were then blocked with 10% heat-inactivated donkey serum, 1% bovine serum albumin (BSA), and 1% Triton X-100 in PBS (0.1 M phosphate buffer, pH 7.2) for 1 h at RT. The samples were incubated with primary antibodies diluted in 5% heat-inactivated donkey serum, 1% BSA, and 10% Triton X-100 overnight at 4°C. The tissues were washed three times with PBST (PBS and 1% Triton X-100) and further incubated at RT for 1 h with secondary antibodies (Alexa Fluor 647 or 555 or 488, Invitrogen) diluted in 0.1% BSA and 0.1% Triton X-100. Finally, the tissues were again washed with PBST three times and mounted on a slide. A Zeiss LSM700 confocal microscope was used to take images. qRT-PCR Total RNA from the brain, from different parts of the cochlea, and from whole cochleae was extracted with ExTrizol Reagent (Protein Biotechnology, PR910), and the reverse transcription from mRNA to cDNA was carried out using cDNA Synthesis kits (Thermo Fisher Scienti c, K1622) according to the manufacturer's instructions. The qPCR was performed using an Applied Biosystems CFX96 qPCR system (Bio-Rad, Hercules, CA, USA) and the SYBR Green (Rox) qPCR Master Mix (Roche Life Science, 04913850001). Validated primers were designed for targeted DNA or mRNA sequences (Table1). The qPCR protocol was an initial denaturing step of 15 s at 95°C followed by 40 cycles of 15 s denaturation at 95°C, 60 s annealing at 60°C, and 20 s extension at 72°C. The expression of mRNA was normalized using the values of Gapdh, and the results were analyzed using the comparative cycle threshold (ΔΔCt) method.

Western blot
Cochleae from two mice were dissected in cold PBS and lysed with 150 ml RIPA Lysis Buffer (Medium, Hangzhou Fu De Biological Technology) and 3 µl 50× protease inhibitor cocktail (Hangzhou Fu De Biological Technology) at 4°C. The primary antibodies were detected by HRP-conjugated secondary antibodies using the ECL detection system. The western blot bands were semiquanti ed using ImageJ software, and the band densities were normalized to background and the relative optical density ratio was calculated by comparison to the reference protein GAPDH or β-actin.

Immunoprecipitation (IP)
Cochleae from two mice were lysed with 150 ml RIPA lysis buffer (Medium, Hangzhou Fu De Biological Technology). CHIP was isolated with antibodies targeting CHIP (Santa Cruz Biotechnology, sc-133066), and Protein A+G was used to capture the antibodies. The ubiquitinated proteins were detected by western blot using anti-ubiquitin antibodies (Santa Cruz Biotechnology, sc-8017).

Fluorescence intensity measurement
Different groups of cochleae were xed, labeled with the same solution, and processed in parallel. The tissue was photographed with confocal microscope using the same parameters. The immunolabeling intensity of antibodies was measured using ImageJ software in which a region of interest was drawn and the mean gray value intensities were measured from 4 or 5 sections per cochlea.

The number and morphology of basilar membrane macrophages
Macrophages were distributed throughout the basilar membrane, and exhibited dendritic, irregular, amoeboid, and spherical morphology. They were identi ed with surface markers F4/80 and Iba1 as have been used in previous studies [29] . To assess the number of macrophages in the apical, middle, and basal turns of the basilar membrane, F4/80-labeled macrophages were counted under the confocal microscope. Images at 20× magni cation taken from each turn of the cochlear whole mounts were used as representative gures. To measure the size of macrophages, ImageJ software was used to outline the membrane boundaries of each cell and calculated the area contained in the drawn region. Five typical cells were selected from each turn of tissue specimen, and their average area represented the size of apical, middle, and basal turns macrophages in the basilar membrane of each individual cochlea.

Macrophage phagocytosis
PHrodo ® zymosan bioparticles conjugate (Invitrogen, P35365) was used to evaluate the phagocytic activity of macrophages. The uorescence of the pHrodo ® dye was activated when the zymosan bioparticles were ingested and exposed to a more acidic PH within the acidic phagocytic vacuoles. Because the extracellular pH is more alkaline, bioparticle uorescence was absent outside the cell. The cochleae were dissected from the skull and placed in live cell imaging solution (A14291DJ, Invitrogen). The membrane labyrinth was opened from the top of the cochlea to remove the basilar membrane, modiolus, and the lateral wall tissue, thereby exposing the inner surface of lateral wall of scala tympani at the basal turn of the cochlea. Then the basal turn was divided into several pieces so that the cochlear bone wall could be at on the slide. PHrodo ® zymosan bioparticles ® conjugate incubated the collected tissues for 90 minutes at 37°C, and then they were rinsed three times for 5 min each using live cell imaging solution. 4% buffered formalin xed the collected tissues for 4 hours and then EDTA decalci ed at 4°C for 1 day. Subsequently, the primary antibody against F4/80 and the appropriate secondary antibody incubated the collected tissues to visualize macrophages.

Drug administration
Clophosome ® -A -Clodronate Liposomes (LCCA) (FormuMax, F70101C-A) provide superior e ciency of macrophage depletion. We intraperitoneally injected mice with LCCA every other day from P30 to P60 at 70 mg/kg. We collected and dissected the cochleae and measured HC loss and the number of macrophages when the drug administration was completed.

Statistical analysis
Microsoft Excel and GraphPad Prism software were used for statistical analyses. All of the data are presented as mean ± SD, and all experiments were repeated at least three times. Two-tailed, unpaired Student's t-tests were performed. P-values <0.05 were considered signi cant, and the level of signi cance is indicated as *P < 0.05, **P < 0.01, ***P < 0.001. All statistical tests were justi ed as appropriate, and the data met the assumptions of the tests. The variance was similar between the statistically compared groups.

Results
FAM73a and FAM73b are expressed in cochlea To determine whether FAM73a and FAM73b are expressed in the cochlea, qRT-PCR was performed in the brain and cochlear tissue of postnatal day 3 (P3) WT mice. Fam73a and Fam73b were indeed expressed in the basilar membrane, lateral wall, and modiolus of the cochlea, although their expression in these regions are not as high as that in brain tissue (Fig. 1A,D). To study their expression pattern during postnatal development, we further examined the expression of Fam73a and Fam73b by qRT-PCR in different age groups. The levels of Fam73a and Fam73b decreased from P14 but was still presented in P30 mice (Fig. 1B,E). Western blots also veri ed the proteins levels of FAM73a and FAM73b in the cochleae of P14 and P30 WT mice (Fig. 1C,F). We then immunolabeled FAM73a and FAM73b with Myosin7a in the whole-mount basilar membrane. Confocal imaging at P3 revealed that FAM73a and FAM73b were expressed in the cytoplasm of HCs rather than in the nucleus (Fig. 1G-H). In P14 and P30 WT mice, FAM73a and FAM73b were stably expressed in the cytoplasm of HCs (Fig. 1I-J). These results suggest FAM73a and FAM73b were expressed in the cochlea at all stages of development.

FAM73a and FAM73b KO mice show progressive hearing loss
To evaluate whether FAM73a and FAM73b affect auditory function and HC survival, we constructed Fam73a and Fam73b KO mice. We rst con rm the e ciency of FAM73a and FAM73b deletion in the cochlea in P30 KO mice by qRT-PCR ( Supplementary Fig. 1A,C) and immunolabeling ( Supplementary Fig.  1B,D). We then evaluated the hearing levels in these mice using the ABR test. In P30 KO mice, ABR thresholds were signi cantly increased at middle and high frequencies after Fam73a deletion, while Fam73b de ciency led to elevated ABR thresholds at all frequencies ( Fig. 2A,G). A similar high-frequency hearing loss was observed both in these two KO mice, but Fam73b deletion caused an earlier and more severe hearing loss at low and middle frequencies than those in Fam73a deletion. Although inner HC loss was not observed, outer hair cell (OHC) loss was consistent with the trend of hearing loss. Scattered OHC loss was seen in the apical, middle, and basal turns both in these two KO mice, and the most severe damage was found in the basal turn (Fig. 2B,C,H,I). However, Fam73a KO mice showed no signi cant difference of OHC loss in the apical turn (Fig. 2B,C). In P60 KO mice, the absence of FAM73a and FAM73b resulted in severe hearing loss (Fig. 2D,J), and the OHC loss in each turn was more severe than that of P30 KO mice (Fig. 2E,F,K,L). We also observed the ultrastructure of the stereocilia in P30 KO mice with a scanning electron microscope and found that the knockout of Fam73a and Fam73b led to the degeneration of the stereocilia (Fig. 2M,N). It was noteworthy that the OHC loss caused by FAM73b de ciency was stronger than that of FAM73a de ciency, and the degree of OHC loss and hearing loss at P60 were worse than those at P30.

Lack of FAM73a and FAM73b causes apoptosis of HCs by enhancing oxidative stress
Previous evidences revealed that absence of Fam7a and Fam73b results in elevated level of oxidative stress [13] . To clarify how FAM73a and FAM73b cause injury on HCs, we thus evaluated the changes of mitochondrial ROS in the HCs of Fam73a and Fam73b KO mice by detecting the oxidative stress markers 3-nitrotyrosine (3-NT) and 4-hydroxynonenal (4-HNE) . In P21 mice, confocal images showed that the levels of 3-NT and 4-HNE were already increased in Fam73a or Fam73b KO mice (Fig. 3A,C,E,G). Quanti cation of the immuno uorescence intensity con rmed a signi cant increase in these two KO mice compared to WT control (Fig. 3B,D,F,H). Western blots also showed increased expression of 3-NT both in Fam73a and Fam73b KO mice compared to controls (Fig. 3I,J). As known, increased oxidative stress can activate apoptosis signaling pathways and induce cell death, so we performed TUNEL staining to identify apoptotic HCs in P21 Fam73a and Fam73b KO mice. Immuno uorescence staining showed that TUNELpositive cells were found in these two KO mice but not in WT mice (Fig. 4A,B,D,E). The qRT-PCR results showed that the expressions of proapoptotic marker genes, including Bax, Casp3, Casp8, Casp9, and Apaf1 was signi cantly higher in the KO mice than those in WT mice (Fig. 4C,F). These results indicate that lack of FAM73a or FAM73b increases oxidative stress in HCs, and subsequent high ROS levels cause apoptosis in HCs.
FAM73a and FAM73b de ciency provokes innate and adaptive immune responses in the basilar membrane Previous studies have shown that macrophages and CD4 + T cells are recruited into the basilar membrane of the cochlea after HC damage [19,24] . FAM73b de ciency has been reported to be essential for the polarization of type 1 macrophages under Toll-like receptor stimulation by controlling mitochondrial morphology switching from fusion to ssion [26] . Therefore, we investigate whether Fam73a and Fam73b in immune cells are also involved in the development of deafness. We measured the in ltration of macrophages and CD4 + T cells in the basilar membrane in Fam73a and Fam73b KO mice. Strikingly, the numbers of macrophages in the basilar membrane were increased in P30 Fam73a and Fam73b KO mice compared to WT control (Fig. 5A,B,E,F). In P60 KO mice, the numbers of macrophages were signi cantly increased in the apical, middle, and basal turns (Fig.5 C,D,G,H). In P30 WT mice, resident macrophages exhibited a slender body with multiple length dendritic projections. Macrophage in Fam73a KO mice showed a similar morphology (Fig. 5A), but their bodies were enlarged in the Fam73b KO mice (Fig. 5E). In P60 WT mice, the macrophage bodies were larger than those in P30 mice in the middle and basal turns (Fig. 5C,G). Fam73a KO mice also exhibited a similar phenotype of macrophage morphology. In the Fam73b KO mice, most of the macrophages transformed into giant irregular or spherical shapes (Fig.  5C,G). The quantitative results con rmed our above observations that the average size of the macrophages in both KO mice was signi cantly larger than in WT mice (Fig. 5J,M), suggesting an activating morphological transformation in these two KO mice. This difference in morphologic transformation between the two KO mice might be due to the different degree of HC damage, because Fam73b KO mice showed earlier and more severe HC damage. Furthermore, we also found that the number of CD4 + T cells in P45 KO mice was signi cantly increased in the whole length of the basilar membrane compared to WT mice (Fig. 5I,K,L,N), indicating an adaptive immune response occurred in two KO mice lines. These results support the hypothesis that lack of FAM73a and FAM73b provokes innate and adaptive immune responses in the basilar membrane.
Lack of FAM73a and FAM73b promotes the expressions of IL-12 and IFN-γ in basilar mebrane.
Because macrophage is a resident immune cell of the basilar membrane , it can affect pathological changes in HC. The phagocytic activity of macrophages is involved in the removal of unwanted or damaged cochlear tissues under both normal and pathological conditions. We thus examine whether phagocytic function of Fam73a and Fam73b KO macrophages distributed on the lateral wall of the scala tympani at the basal turn changed using a confocal microscope (Fig. 6A-H). Although no signi cant difference in number of macrophages were found between the WT and Fam73a KO mice (Fig.6B), that in Fam73b KO mice was signi cantly decreased (Fig.6F), implying that more macrophages are recruited into the basilar membrane due to a severe HC damage in Fam73b KO mice. However, phagocytic activity of these macrophages displayed no difference compared to WT control, suggested by a similar uorescence uptaking using pHrodo ® uorescent bioparticles (Fig.6C,G). The quantitative results con rmed these observations (Fig.6D,H). Collectively, these ndings demonstrated that FAM73a and FAM73b de ciency did not regulate the phagocytic capacity of cochlear macrophages.
Because the different recruitment of macrophages and T cells were observed between KO and WT mice, we further evaluated whether FAM73a and FAM73b de ciency triggered cochlear in ammatory activity. We examined the transcriptional levels of proin ammatory cytokines (Il12a, Il12b, Ifng, Il6, Il1b, and Tnfa), and anti-in ammatory activity (Il10) in the cochlea by qRT-PCR. In P60 Fam73a KO or Fam73b KO mice, the mRNA levels of Il12a and Ifng were signi cantly increased both in Fam73a and Fam73b KO mice compared with WT group (Fig. 7A), while Il10 was signi cantly decreased (Fig. 7D). No difference in mRNA levels of Il6 and Il1b was observed, suggesting Fam73a and Fam73b de ciency only induced a partial phenotypes of M1 subtype. Consistently, M2-type macrophage marker Cd206 and Arg1 was signi cantly decreased in P60 Fam73a or Fam73b KO mice compared with WT mice (Fig. 7B, 7E). The western blot analysis con rmed the downregulation of ARG1 protein expression both in these two KO mice (Fig. 7C,F). To further verify the enhanced expression of IL-12, we immunolabeled IL-12 in the macrophages of the basilar membrane. We found that the expression of IL-12 in macrophages was upregulated in these two P60 KO mice (Fig. 7G-J). To determine whether the antigen presentation function of macrophages was enhanced by the increased expression of IL-12, we immunolabeled MHCII -an antigen-presenting protein -in macrophages. Interestingly, we also found that the expression of MHCII in Fam73a or Fam73b KO macrophages was increased compared with WT mice (Fig. 7K-N). These ndings suggest that mitochondrial dynamics regulates IL-12 expression and capacity of antigen presentation in macrophages, which further promotes the expression of IFN-γ in CD4 + T cells.
To prove the role of macrophages in HC damage in the cochlea, LCCA was intraperitoneally injected into KO mice to deplete macrophages from P30 to P60. We found that the numbers of macrophages were signi cantly decreased in the basilar membrane of P60 KO mice after LCCA injection (Fig. 8A-B,E-F). After deletion of macrophages by LCCA, HC loss was obviously reduced in P60 Fam73a and Fam73b KO mice compared to the untreated groups ( Fig. 8C-D,G-H). These results suggest macrophages play an important role in HC injury in Fam73a and Fam73b KO mice.
FAM73a and FAM73b control macrophage function via regulating Parkin-CHIP-IRF1 signaling A previous study reported that mitochondrial ssion caused by Fam73b ablation increases the expression of Parkin and IRF1, which further promoting the production of IL-12 [26] . To determine the role of Parkin-IRF1 signal in the cochlea of Fam73a and Fam73b KO mice, we rst performed qRT-PCR to determine the mRNA level of Irf1. The expression of Irf1 was not signi cantly different in P60 Fam73a or Fam73b KO mice compared with WT control (Fig. 9A,D). However, western blots showed a signi cantly upregulated expression of IRF1 in the cochlea of these two KO mice ( Fig. 9B-C,E-F). To further verify the increased expression of IRF1 in macrophages in the basilar membrane, we co-stained IRF1 and Iba1 in the whole mount tissue. Confocal images showed an obviously enhanced expression of IRF1 in KO macrophages compared with WT littermate (Fig. 9G-H,I-J). Monoubiquitinated CHIP promoted the degradation of IRF-1, thus we also found the protein expression level of mono-ubiquitinated CHIP was signi cantly reduced in P60 Fam73a or Fam73b KO mice compared with WT mice (Fig. 10A-D). Lastly, qRT-PCR assay (Fig.  11A,D) and western blots ( Fig. 11B-C,E-F) showed that mRNA and protein levels of Park2 was signi cantly upregulated in P60 these two KO mice. To further con rm that the expression of Parkin was upregulated in macrophages of basilar membrane, we immunolabeled Parkin and Iba1 in macrophages. The increased expression of Parkin was observed by confocal images of macrophages from KO mice ( Fig. 11G-H,I-J). Together these results suggest a similar regulating mechanism between peripheral macrophages and those in the cochlea via Parkin-CHP-IRF1 signal to control the production of IL-12.

Discussion
Mitochondria provide energy for cellular activity and thus play key roles in cell survival, apoptosis, and metabolism. Mitochondrial morphology switching between fusion and ssion affects mitochondrial function and leads to changes in the production of ROS and mitophagy. Previous studies have shown that increased levels of ROS damage HCs in noise-, drug-and age-related hearing loss [30][31][32] . The mitophagy protects against HC injury caused by noise exposure, ototoxic drug treatment, and age-related degeneration [8,33,34] . However, the roles of mitochondrial membrane proteins regulating mitochondrial morphology, including Mitofusin 1 (MFN1) and MFN2, in HCs remains unclear due to embryonic lethality of de cient mice. OPA1, as a protein of regulating mitochondrial inner membrane fusion, causes SHNL through auditory neuropathy but not HC injury. Recent studies showed that Fam73a and Fam73b KO mice are viable and can be used as a suitable model to study the in uence of mitochondrial dynamics on auditory function in the inner ear. Our results indicate that FAM73a and FAM73b de ciency results in HC damage through increased production of ROS and the destruction of stereocilia structures. These results suggest that the absence of FAM73a and FAM73b has a harmful effect on HCs through oxidative stress rather than a protective effect through mitophagy.
Previous studies have shown that HC injury increases the number of macrophages and transforms macrophage morphology into an activated shape [24,25] . Our study shows that the number of macrophages signi cantly increases accompanied by larger in size when HCs are severely damaged. To further explore the role of macrophages in HC death, we examined the phagocytic capacity of macrophages. However, the phagocytic capacity of macrophages localized at the luminal surface of the scale tympani cavity did not change in Fam73a and Fam73b KO mice compared with the controls. Intraperitoneal injection of LCCA not only deplete macrophages, but also signi cantly reduced HC damage. Therefore, we believe that macrophages play a detrimental role in HC injury. These results indicate that HC injury activates macrophages by releasing certain endogenous factors triggering the cochlear immune response, and the next studies should identify these molecules and clarify their inducible mechanism.
However, there is no report on the expression of IL-12 and IFN-γ in these models with cochlear aseptic in ammation. Clinical studies have found that TNF-α might play critical roles in sudden SNHL based on blood samples from patients, while IL-12, IFN-γ and IL-10 have been shown not to participate in the pathophysiology of sudden SNHL [39,40] . We found that the expression of proin ammatory cytokines IL-12 and IFN-γ increased, while the anti-in ammatory cytokine IL-10 decreased in Fam73a and Fam73b KO mice, and this suggests a novel mechanism of in ammation in SNHL occurrence caused by the disorder of mitochondrial morphology.
Previous studies have shown that the production of in ammatory cytokines in the cochlea is attributed to the activation of toll-like receptors on the surface of macrophages [41] under noise exposure or ototoxic drug treatment. TLR4 promotes the production of ROS and the activation of the downstream NF-kB signaling pathway [42][43][44] . However, our results presented a novel signal pathway which is composed by Parkin-CHIP-IRF1 [26] . This signal pathway regulated the production of IL-12 in macrophages is independent of ROS or NF-kB, which is different from previous studies. Our data show that the expression of IRF1 and Parkin in macrophages is increased, while the level of monoubiquitinated CHIP is decreased.
These results indicate that macrophages modulate the production of in ammatory cytokines through a novel signal pathway in the cochlea of Fam73a and Fam73b KO mice.
Although macrophages representing innate immune responses have been well studied in the cochlea, the role of adaptive immune responses remains unclear. Adaptive immune cells have been shown to be involved in noise-induced cochlear damage [18] . CD4 + T cells exist in the modiolus of the cochlea under physiological conditions and can enter the basilar membrane of the cochlea together with macrophages in response to noise-induced damage [19,45] . Additionally, inhibition of nuclear factor of activated T cells (NFAT) protects HCs against aminoglycoside ototoxicity [46] . Although these studies have shown the presence of T cells in the inner ear and their activation can damage HCs, its underlying mechanism affecting HC injury remains unknown. Our results showed that the number of CD4 + T cells in the basilar membrane and the expression of IFN-γ derived from CD4 + T cells is signi cantly elevated in the cochlea of Fam73a and Fam73b KO mice. These results implies that CD4 + T cells induced the damage of HCs by secreting IFN-γ, which is the rst time to report the T cell-mediated mechanism in HCs damage.
In conclusion, our study provides disrupted mitochondrial morphology switching as a new risk factor for SNHL.