Bmi1 regulates auditory hair cell survival by maintaining redox balance

Reactive oxygen species (ROS) accumulation are involved in noise- and ototoxic drug-induced hair cell loss, which is the major cause of hearing loss. Bmi1 is a member of the Polycomb protein family and has been reported to regulate mitochondrial function and ROS level in thymocytes and neurons. In this study, we reported the expression of Bmi1 in mouse cochlea and investigated the role of Bmi1 in hair cell survival. Bmi1 expressed in hair cells and supporting cells in mouse cochlea. Bmi1−/− mice displayed severe hearing loss and patched outer hair cell loss from postnatal day 22. Ototoxic drug-induced hair cells loss dramatically increased in Bmi1−/− mice compared with that in wild-type controls both in vivo and in vitro, indicating Bmi1−/− hair cells were significantly more sensitive to ototoxic drug-induced damage. Cleaved caspase-3 and TUNEL staining demonstrated that apoptosis was involved in the increased hair cell loss of Bmi1−/− mice. Aminophenyl fluorescein and MitoSOX Red staining showed the level of free radicals and mitochondrial ROS increased in Bmi1−/− hair cells due to the aggravated disequilibrium of antioxidant–prooxidant balance. Furthermore, the antioxidant N-acetylcysteine rescued Bmi1−/− hair cells from neomycin injury both in vitro and in vivo, suggesting that ROS accumulation was mainly responsible for the increased aminoglycosides sensitivity in Bmi1−/− hair cells. Our findings demonstrate that Bmi1 has an important role in hair cell survival by controlling redox balance and ROS level, thus suggesting that Bmi1 may work as a new therapeutic target for the prevention of hair cell death.

are not only cellular organelles required for aerobic respiration but also the main source of intracellular ROS, which are thought to be the cause of most oxidative damage. Because of the abundant number of mitochondria and high consumption of oxygen, auditory hair cells are exposed to high level of oxidative stress, especially when challenged by ototoxic drugs. 9,[16][17][18][19] The balance between prooxidant and antioxidant molecules is critical to determine the rate of oxidative damage accumulation. Administration of ROS-scavenging antioxidants, 16,[20][21][22][23] as well as inhibition of oxidase, 18 can reduce the ROS production, thus attenuating the subsequent hair cell death in ototoxic drug-treated cochleae.
Bmi1 is a member of the Polycomb group (PcG) family. PcG proteins form large complexes, including PRC1 (Polycomb repressive complex 1) and PRC2 (Polycomb repressive complex 2), which silence target genes by modifying chromatin organization. 24,25 Bmi1 also can bind to the Runx1/CBFβ transcription factor complex to silence target gene in a PRC2-independent manner. 26 Although many researchers focus on the role of Bmi1 in self-renewal of embryonic, adult and cancer stem cell, [27][28][29][30][31][32][33][34][35][36] several evidences suggest that Bmi1 also regulate cell survival by controlling mitochondrial function and ROS level. Previous report showed that Bmi1-deficient thymocytes have impaired mitochondrial function, which lead to a marked increase of intracellular ROS levels and subsequent engagement of the DDR pathway. 13 In Bmi1-deficient CD34(+) stem cells, the reduced ability of self-renewal is associated with enhanced apoptosis, which coincided with increased levels of intracellular ROS. 37 Bmi1 controls memory CD4 T-cell survival through direct repression of Noxa gene in an Ink4a-and Arf-independent manner. 38 Overexpression of Bmi1 in vivo protects human embryonic stem cells (HSCs) from ROS damage and extends the lifespan of HSCs, 39 whereas Bmi1 transduction in vitro reduced irradiation-induced ROS levels by suppressing the oxidase genes, including lactoperoxidase (Lpo), and increased repair of DNA damage in human keratinocytes. 40 Bmi1 also expressed in terminally differentiated cells, such as neurons, 41 besides stem cells and rapidly dividing cells. It is reported that Bmi1 is required in neurons to suppress p53induced apoptosis via regulating the antioxidant defensive response. 42 High Bmi1 expression level in cortical neurons resulted in the suppression of ROS through activation of antioxidant genes and conferred robust protection against DNA-damage-induced cell death or mitochondrial poisoning. 41 However, the expression of Bmi1 and its function in the inner ear have not been reported. In this study, we investigated Bmi1 expression in mouse cochlea and its role in hair cell survival. We found that Bmi1 is expressed in the hair cells and supporting cells, and can regulate the redox balance and ROS levels, thus having an important role in the survival and sensitivity to ototoxic drug of auditory hair cells in mice cochleae.

Results
Bmi1 expressed in auditory hair cells. To investigate the Bmi1 expression in mouse cochlea, we used immunofluorescence staining with anti-Bmi1 antibody (Millipore, Consett, UK). Myosin 7a and sex-determining region Y)-box 2 (Sox2) were used as hair cell and supporting cell markers, respectively. Bmi1 expressed in both hair cells and supporting cells in the cochlea of neonatal and P30 wild-type (WT) mice (Figures 1a and b). Bmi1 also expressed in spiral ligament and spiral ganglion cells (data not shown).
Morphological study showed there was no obvious change of cochlea structure at all three turns in the neonatal Bmi1 − / − mice. The number and morphology of hair cells in neonatal Bmi1 − / − mice is similar to that in WT littermate (Figure 2a), suggesting that Bmi1 does not affect the development of auditory hair cells. However, at P22, outer hair cell loss started   (Figure 3a) (n = 5, Po0.05).Furthermore, the increased injury sensitivity to aminoglycosides of Bmi1 − / − hair cells was also confirmed by in vivo study. Neomycin (125 mg/kg/day) was administrated to the P7 Bmi1 − / − , Bmi1 +/ − and WT mice for 5 days. Ten days after neomycin injection, hair cell loss in the apical, middle and basal turns of WT mice were 0.44 ± 0.32%, 0.25 ± 0.34% and 5.69 ± 1.67%, respectively, whereas in Bmi1 − / − mice, these percentages significantly increased to 0.72 ± 0.48%, 11.05 ± 0.66% and 43.09 ± 4.04%, respectively (Figure 3b) (n = 5, Po0.05). All these results suggested that endogenous expression of Bmi1 has an important role in the injury sensitivity to ototoxic drugs and the survival of cochlear hair cells.  Quantitative RT-PCR and western blotting results showed that the expression of pro-apoptotic genes p53 and p53 target genes, including Noxa and Puma, significantly increased in Bmi1 − / − hair cells (Figures 4e and g). Taken together, these data indicated that apoptosis involved in the increase of aminoglycosides induced hair cell loss in Bmi1 − / − mice. γH2AX (phosphorylated H2AX at the 139th serine residue), a sensitive indicator of DNA double-strand break, triggers Chk2 signaling pathway and p53 activation. 44 After neomycin treatment, the level of γH2AX significantly increased in Bmi1 − / − hair cells when compared with WT controls (Figure 4c), suggesting that DNA-damage-induced Chk2 signalling pathway might be involved in the increased sensitivity to aminoglycosides of Bmi1 − / − hair cells.
Free radicals and mitochondrial ROS levels increased in Bmi1 − / − hair cells after neomycin insult. It has been reported that aminoglycosides induced accumulation of hROS is closely related to hair cell damage. 45 Aminophenyl fluorescein (APF), which is a green probe, can selectively detect hROS, including free hydroxyl radical (•OH), peroxynitrite (ONOO − ) and hypochlorite (− OCl). In the absence of damage, APF fluorescence in cochlear epithelium could not be detected in both Bmi1 − / − and WT mice (Figure 5a). Two hours after neomycin treatment, APF fluorescence could be detected in hair cells and the fluorescence intensity was significantly higher in Bmi1 − / − mice than that in WT control (Figure 5a), indicating that hROS level significantly increased in Bmi1 − / − mice. Auditory hair cells possess many mitochondria and consume large amount of oxygen, and thus are sensitive to oxidative stress. MitoSOX Red, a redox fluorophore detecting selectively mitochondrial superoxide, 13,46,47 were used to evaluate mitochondrial ROS generation in hair cells. In the absence of damage, no MitoSOX Red fluorescence was detected in both Bmi1 − / − and WT epithelium (Figure 5b). Two hours after neomycin treatment, the fluorescence intensity of MitoSOX Red was significantly higher in Bmi1 − / − mice than that in WT control (Figure 5b), indicating that mitochondrial ROS level significantly increased in Bmi1 − / − mice. These results suggested that the increased levels of ROS might be the reason for the increased injury sensitivity to aminoglycosides of Bmi1 − / − hair cells.
The disequilibrium of antioxidant-prooxidant balance deteriorated in Bmi1 − / − hair cells after neomycin injury. Cellular redox homeostasis depends on the antioxidantprooxidant balance. The increment of ROS levels in Bmi1 − / − hair cells might be caused by the decreased antioxidant levels and/or the increased prooxidant levels. Previous studies have reported the decreased expression levels of antioxidant genes (xCT, Nqo1, Sod1 and Sod2) in Bmi1deficient neurons 42 and the increased expression levels of oxidizing enzymes (Alox5, arachidonate 15-lipoxygenase (Alox15), Duox1, dual oxidase 2 (Duox2), cysteine dioxygenase 1 (Cdo1) and Lpo) in Bmi1-deficient thymocytes. 13 To find out how Bmi1 deficiency leads to elevated ROS levels in Bmi1 − / − cochlea, quantitative RT-PCR was used to investigate the expression levels of several antioxidant (xCT, Nqo1, Cat, Sod1, Sod2, Gsr and Gstm1) and prooxidant genes (Alox15, Lpo, Duox2 and Cdo1). In the absence of damage, the expression of both antioxidant and prooxidant genes showed no significant difference between Bmi1 − / − and WT neonatal mice (Figures 5d and e). Two hours after neomycin treatment, the expression levels of both antioxidant and prooxidant genes dramatically increased in the WT cochlea. However, the increment amplitude of prooxidant gene expression, including Alox15, Cdo1 and Lpo, increased significantly more in Bmi1 − / − cochlea than in WT control (Figure 5d), whereas the increment amplitude of antioxidant gene expression, including xCT (also known as Slc7a11, solute carrier family 7, member 11), Nqo1 (NAD(P) H dehydrogenase, quinone 1) and superoxide dismutase 2, mitochondrial (Sod2), increased significantly more in WT control than in Bmi1 − / − cochlea (Figure 5e). These data demonstrated that the disequilibrium of antioxidant-prooxidant balance was aggravated in Bmi1 − / − cochlear hair cells, which resulted in the much more severe ROS accumulation in Bmi1 − / − hair cells than in WT control.

Antioxidant treatment rescued the hair cells in Bmi1
To further investigate whether the increment of ROS levels contribute to the increased injury sensitivity to aminoglycosides of Bmi1 − / − hair cells, the antioxidant Nacetylcysteine (NAC), which is a reduced glutathione (GSH) provider and a direct scavenger of reactive oxygen intermediates, was used to treat the neomycin-injured hair cells. After NAC (20 μM) treatment, hair cell loss dramatically decreased in both Bmi1 − / − and WT control mice, and the number of survived hair cells has no significant difference between Bmi1 − / − and WT mice (Figures 6a and c). The rescue effect of antioxidant treatment was further confirmed by in vivo study (Figures 6b and d), demonstrating that ROS accumulation was the major cause of the high injury sensitivity of Bmi1 − / − auditory hair cells to aminoglycosides.

Discussion
The Polycomb gene Bmi1 is extensively studied in stem cells and rapid dividing cells, such as cancer cells and lymphocytes. Although the role of Bmi1 in stem cell renewal has been largely understood, several evidences suggest that Bmi1 also regulates cell survival by controlling mitochondrial function and ROS level. However, the expression and function of Bmi1 in the cochlea remains unclear. In this study, we reported the expression of Bmi1 and demonstrated that Bmi1 regulates the survival of hair cells by controlling the accumulation of ROS in postnatal mice cochlea. We reported the severe hearing loss (Figure 2), the daily-increasing hair cell loss from P22 ( Figure 2) and the increased susceptibility of hair cells to ototoxic drugs (Figure 3) in Bmi1 KO mice, which indicated that Bmi1 has an important role in regulating the survival of auditory hair cells. After neomycin treatment, we found loss of Bmi1 led to the significant increased accumulation of free radicals and mitochondrial ROS ( Figure 5), which induced DNA damage and apoptosis of hair cell ( Figure 4); and antioxidant treatment rescued this hair cell loss in Bmi1 KO cochlea ( Figure 6).

Role of Bmi1 in cochlear hair cell survival Y Chen et al
The plenty of mitochondria and high oxygen consumption in mouse auditory hair cells make them sensitive to oxidative stress especially challenged by external stimulation such as noise and ototoxic drugs. 9 It has been reported that Bmi1deficient thymocytes 13 and neurons 42 have an increased level of intracellular ROS due to deregulated expression of genes related to mitochondrial function and ROS generation. In our study, we reported severe ototoxic drug-induced hair cell damage ( Figure 3) accompanied with higher levels of free radicals and mitochondrial ROS accumulation ( Figure 5) in Bmi1 − / − hair cells after neomycin injury, which suggested ROS involved in the increased injury sensitivity to aminoglycosides in Bmi1 − / − cochlea. After antioxidant NAC treatment, the number of survived hair cells in Bmi1 − / − mice had no significant difference compared with WT mice (Figure 6). This finding demonstrated that the high ROS level was the major cause of the high injury sensitivity to ototoxic drugs in Bmi1 − / − cochlea, and indicated that Bmi1 regulates the hair cell survival by controlling the levels of ROS.
Balance between prooxidant and antioxidant molecules is critical to determine the rate of oxidative damage accumulation. When external stimulation brings out accumulation of ROS, the expression of antioxidant response genes also increased. In this study, we first reported the injury-induced dramatic increase of oxidizing enzymes (Alox15 and Lpo) and antioxidant genes (xCT, Nqo1 and Sod2) in mouse cochlea ( Figure 5). Previous studies have reported the increased expression of oxidizing enzymes in Bmi1 − / − thymocytes 13 and decreased expression of antioxidant enzymes in Bmi1 − / − neurons. 42 In Bmi1 − / − cochlea, the increment amplitude of oxidizing enzymes increased, while the increment amplitude of antioxidant genes decreased compared with WT control after neomycin injury ( Figure 6), indicating that aggravated disequilibrium of antioxidant-prooxidant balance is responsible to the increased accumulation of ROS. These results suggested that Bmi1 controls the antioxidant-prooxidant balance to maintain redox homeostasis in mouse auditory hair cells.
Several pathways are involved in Bmi1 deficiency-induced cell death, including the following: (1) Bmi1 deficiency leads to mitochondrial dysfunction resulting in increased levels of intracellular ROS and subsequent engagement of the DDR pathway in which Chk2 and p53 were activated; 13 (2) Bmi1 deficiency directly leads to p53-mediated repression of antioxidant genes, resulting in increased ROS; 41 (3) Bmi1 deficiency deregulates the expression of p19, resulting in phosphorylation of p53, which triggers p53-dependent apoptosis; 28 (4) Bmi1 deficiency also can directly deregulate the expression of Noxa (also known as Pmaip1, phorbol-12myristate-13-acetate-induced protein 1) independent of p19 and p53 (Yamashita et al. 38 ). In the absence of neomycin injury, the expression of p16 and p19 had already significantly increased in cultured neonatal Bmi1 − / − cochlea, although there was no significant hair cell loss in neonatal Bmi1 − / − cochlea (Figure 4). After neomycin treatment, significantly more hair cell loss was observed in Bmi1 − / − cochlea when compared with WT control (Figure 3), and this was accompanied by significantly higher levels of free radicals and mitochondrial ROS ( Figure 5), more DNA damage-induced γH2AX foci (Figure 4), as well as higher expression levels of p16, p19, p53 and p53 target genes (noxa and puma) in Bmi1 − / − cochlea than in WT control (Figure 4). ROS can oxidize cell constituents, such as DNA, lead to DNA damage and activate multiple apoptosis pathways including p53-dependent apoptosis. [48][49][50] γH2AX triggers Chk2 signaling pathway and p53 activation. In cooperation with γH2AX, p53 repairs DNA damage or induces cell death. 51 In our study, we found that the peak of free radicals and mitochondrial ROS level was 2 h after neomycin treatment, whereas the peak of the expression of p53, noxa and puma were 12 h after neomycin treatment; thus, we speculated that Bmi1 deficiency led to the disequilibrium of antioxidantprooxidant balance, resulting in increased accumulation of ROS, which then triggered DNA damage and multiple apoptotic pathways including p53-dependent apoptosis.
NAC, an essential precursor to many endogenous antioxidants involved in the decomposition of peroxides, attenuates oxidative stress by replenishing intracellular GSH stores. 52 In clinical trials, NAC improves lung function in patients with chronic obstructive pulmonary disease, highlighting the potential benefit of ROS-directed therapy. 53 In the current study, both in vivo and in vitro results confirmed that NAC treatment ameliorated neomycin-induced hair cell damage, suggesting that antioxidant treatment might be an effective therapy for hearing loss induced by ototoxic drugs. Furthermore, we demonstrated that the increased sensitivity to ototoxic drugs caused by Bmi-1 deficiency was largely rescued by antioxidant treatment, indicating that Bmi1 regulated hair cell survival and sensitivity to ototoxic drugs by maintaining redox balance.
In summary, we first reported the expression pattern of Bmi1 in mouse cochlea and then demonstrated that Bmi1 has an important role in the hair cell survival and regulates the sensitivity of hair cells to ototoxic drug via controlling redox balance. Lastly, we approved that ROS accumulation was the main reason for increased aminoglycoside sensitivity in Bmi1 − / − hair cells, both in vitro and in vivo, via the antioxidant NAC rescue experiments. Our findings may provide new therapeutic targets for prevention of the hair cell death.

Materials and Methods
Mice and genotyping. Bmi1 +/ − mice (129Ola/FVB/N hybrid background) that were backcrossed 10-12 times onto a C57BL/6J background were mated to generate Bmi1 − / − , Bmi1 +/ − and WT mice (littermates) genotyped by PCR, as described previously. 27 This study was carried out in strict accordance with the 'Guiding Directive for Humane treatment of Laboratory Animals' issued by the Experimental protocol (in vivo studies). To explore the effect of Bmi1 deficiency on injury sensitivity to ototoxic drug in cochlear hair cells, neomycin was injected subcutaneously once daily for 5 days from postnatal day 7 (P7) at a final dose of 125 mg/kg/day. After exposure to neomycin, the hearing threshold was evaluated by ABR measurement at P22. For NAC rescue experiment, NAC (200 mg/kg/day, i.p.) were injected for 7 days at the beginning of neomycin insult. Next, mice of either sex were killed and cochleae were fixed in 4% paraformaldehyde (PFA). After decalcification, cochlear epithelium was prepared for morphological analysis.
ABR and DPOAE measurement. ABR analysis was performed in anesthetized P22 and P30 mice, to measure the hearing threshold, as described previously. 54 The hearing threshold of P22 and P30 mice was assessed at four frequencies (8, 16, 24 and 32 kHz) in a TDT system 3 (Tucker-Davies Technologies, Gainesville, FL, USA). DPOAE signal was measured with a TDT-RZ6 system (Tucker-Davis Technologies). The stimuli for eliciting DPOAEs were two sine wave tones of differing frequencies (F2 = 1.2F1) of 1 s duration with F2 ranging from 4 to 40 kHz. The two tones were presented at identical intensities, which ranged from 20-90 dB SPL in 10 dB increments. The acoustic signal picked up by the microphone in the earbar was digitized at 200 kHz and the magnitude of the 2F1 − F2 distortion product determined by FFT. The surrounding noise floor was also calculated by averaging 20 adjacent frequency bins around the distortion product frequency. DPOAE thresholds were calculated offline by interpolating the data and identifying when the signal was greater than −5 dB SPL and greater than two S.D. above the noise floor. If no DPOAE response was detected even at our equipment limits of 90 dB SPL, we arbitrarily defined the threshold to be 90 dB.
Experimental protocol (in vitro studies). Cochlear sensory epithelium was dissected from P0 mice and cultured as previously reported. 55 Neomycin (0.25 mM, Sigma-Aldrich, St. Louis, MO, USA) or cisplatin (10 μM, Sigma-Aldrich) was added for 24 h to kill hair cells. After neomycin or cisplatin was removed, the tissues were cultured in serum-free medium for additional 48 h. For NAC rescue experiment, NAC (20 mM) were added in medium for 48 h at the beginning of neomycin insult.
Immunofluorescence, phalloidin, TUNEL and APF staining. Mouse monoclonal anti-Bmi1 (Millipore, 1 : 500), rabbit polyclonal anti-cleaved , goat polyclonal anti-SRγ (sex-determining region γ)-box 2 (Sox2) (sc-17320, Santa Cruz Biotechnology, Dallas, TX, USA) and phalloidin-tetramethylrhodamine (TRITC) (Sigma-Aldrich, 1 : 1000) were used. Briefly, nonspecific binding sites were blocked for 1 h in 0.3% Triton X-100 and 10% (v/v) heat-inactivated normal serum in PBS (PBT1). Tissues were then incubated overnight at 4°C in PBT1 with primary antibodies. After unbound antibodies were removed, tissues were incubated with the corresponding secondary antibodies conjugated with TRITC, fluorescein isothiocyanate or Cy5 (Jackson ImmunoResearch, West Grove, PA, USA). Counterstaining with DAPI (Sigma-Aldrich) allowed visualization of the cell nuclei. Specimens were examined by confocal fluorescence microscopy (Leica SP5, Heidelberg, Germany). Negative control experiments were performed as above by omitting the primary antibodies. TUNEL Kit (Roche, Indianapolis, IN, USA) was used to detect apoptotic cells, according to the instructions of the manufacturer. APF staining was performed to detect hROS as previous reported. 45 After being exposed in 0.25 mM neomycin, the explants were rinsed with HBSS and stained with APF (12.5 μM, Cell Technology Inc., Beverly, MA, USA) for 30 min at 37°C in humidified air with 5% CO 2 . APF is minimally fluorescent; however, when reacted with hROS (•OH, ONOO − and − OCl), they are converted to fluorescein and exhibit strong, dose-dependent fluorescence. MitoSOX Red (Invitrogen, Carlsbad, CA, USA) was used to detect mitochondrial ROS. After being exposed in 0.25 mM neomycin, the explants were washed with PBS and stained with MitoSOX Red probe (Invitrogen, 5 μM), incubated for 10 min. After staining, the cells were fixed in PFA.