Dibenzazepine promotes cochlear supporting cell proliferation and hair cell regeneration in neonatal mice

Abstract Objectives To investigate the role of dibenzazepine (DBZ) in promoting supporting cell (SC) proliferation and hair cell (HC) regeneration in the inner ear. Materials and Methods Postnatal day 1 wild‐type or neomycin‐damaged mouse cochleae were cultured with DBZ. Immunohistochemistry and scanning electron microscopy were used to examine the morphology of cochlear cells, and high‐throughput RNA‐sequencing was used to measure gene expression levels. Results We found that DBZ promoted SC proliferation and HC regeneration in a dose‐dependent manner in both normal and damaged cochleae. In addition, most of the newly regenerated HCs induced by DBZ had visible and relatively mature stereocilia bundle structures. Finally, RNA sequencing detected the differentially expressed genes between DBZ treatment and controls, and interaction networks were constructed for the most highly differentially expressed genes. Conclusions Our study demonstrates that DBZ can significantly promote SC proliferation and increase the number of mitotically regenerated HCs with relatively mature stereocilia bundles in the neonatal mouse cochlea by inhibiting Notch signalling and activating Wnt signalling, suggesting the DBZ might be a new therapeutic target for stimulating HC regeneration.


| INTRODUC TI ON
Hair cells (HCs) in the cochlea are a critical part of the auditory system and convert the vibrations of sound waves into electrical signals.
The HCs are vulnerable to various insults such as acoustic trauma, ageing, noise exposure and ototoxic drugs. When mammalian cochlear HCs are damaged, this leads to permanent hearing loss due to the inability to regenerate new HCs spontaneously after the cochlear HCs have differentiated and matured. 1 However, the lower vertebrates, like fish and birds, can regenerate new HCs spontaneously when the cochlear HCs are damaged, and it has been shown that the mammalian utricle has very limited spontaneous HC regeneration ability. [2][3][4] Recent studies have shown that supporting cells (SCs) are reliable sources for regenerating HCs after HC loss in both the cochlea and the utricle. There are two kinds of HC regeneration mechanisms in vertebrates, namely direct transdifferentiation, in which SCs directly differentiate into HCs without cell division, 5,6 and mitotic regeneration, in which SC/progenitor cells re-enter the cell cycle to proliferate first and then, several days later, switch fates to become HCs. 7,8 In the mouse cochlea, HCs are interdigitated by SCs, and previous studies have shown that SC loss also leads to apoptosis of HCs. 9 Thus, the goal of HC regeneration strategies is to regenerate the HCs without exhausting the SCs at the same time, and in our work we aim to induce SC/progenitor cell proliferation first and then induce the proliferated SCs/progenitor cells to differentiate into HCs.
Dibenzazepines (DBZs), which are a class of drugs that includes clozapine, carbamazepine and many others, is mainly used for the treatment of epilepsy, trigeminal neuralgia and some psychiatric disorders. All of these drugs are heterocyclic compounds consisting of two benzene rings fused to an azepine ring. Recently, DBZ (also known as γ-secretase inhibitor XX (GSI XX)), a member of the dibenzazepine family, has shown effects on Notch signalling pathways, 10 and the use of DBZ has been extended to other diseases due to its role in modulating Notch signalling and other pathways. [11][12][13] For example, DBZ is used to treat chronic kidney disease by ameliorating fibroblast proliferation in the tubular epithelial cells via inhibition of the TGF-β/Smad2/3 signalling pathway. 14 Also, during the early stage of acute inflammation, DBZ can moderate the effects of acute colitis by protecting the goblet cells through inhibition of Notch signalling and reversing the down-regulated mRNA of the goblet cell-related genes Math1 and MUC2. 12 Finally, DBZ prevents AngIIinduced abdominal aortic aneurysm formation by inhibiting the accumulation of macrophages and CD4 + T cells, Th2 differentiation and ERK-mediated angiogenesis. 13 However, DBZ has not previously been studied in terms of promoting SC proliferation and HC regeneration in the inner ear.
In this study, we treated cultured neonatal mouse cochleae with DBZ to investigate its role in mammalian cochlear HC regeneration. We found that DBZ induced SC proliferation and promoted the mitotic regeneration of HCs with relatively mature stereocilia bundle structures by inhibiting Notch signalling and activating Wnt signalling. This study suggests that DBZ might be a new and effective therapeutic drug to promote the mitotic regeneration of HCs in the mammalian cochlea.
Math1-GFP transgenic mice were obtained from Dr Jane Johnson (University of Texas Southwestern Medical Center, Dallas). 15 The care and use of animals was approved by the Institutional Animal Care and Use Committee of Fudan University in compliance with the NIH guidelines for the care and use of laboratory animals.

| Organotypic culture of mouse cochleae
The mice were euthanized by carbon dioxide asphyxiation and decapitated, and their heads were placed in 75% ethanol and quickly transferred to chilled Hanks' balanced salt solution (HBSS). The temporal bones were dissected out, the cochlea was isolated from the temporal bone using sterile procedures in ice-cold HBSS, and the stria vascular and spiral ganglion were removed with fine forceps.
Explants of the organ of Corti were placed intact on polylysine-coated cover glasses (Sigma, St. Louis, MO, USA) and maintained in four-well culture dishes (Greiner Bio-One, Frickenhausen, Germany) in culture medium composed of Dulbecco's modified Eagle's medium (DMEM) and F12 medium supplemented with N2 and B27 (Invitrogen/GIBCO/BRL, Carlsbad, CA) and 50 IU/mL penicillin (Sigma). The tissues were incubated at 37℃ in a humidified atmosphere at 95°C and 5% CO 2 .

| Treatment of cultured cochleae
Cultures were switched to a culture medium with 10% FBS for about 2 hours to promote adherence to the wells and then treated with Grand Island, NY), which is efficiently incorporated into newly synthesized DNA, was added to the culture medium at a concentration of 10 μM for the entire culture period.
Another group of explanted cochleae was treated with 1 mM neomycin (Sigma) for 12 hours and then thoroughly rinsed in fresh medium. These cochleae were then treated with 5 μM or 10 μM DBZ along with 10 μM EdU for 7 days.

| Immunohistochemistry
The cultured cochleae were harvested and fixed for 30 minutes at room temperature with 4% paraformaldehyde in 0.1 M phosphate buffer (PBS) and then thoroughly rinsed with 0.01 M PBS.

| Image acquisition and quantification
The fluorescence in the organ of Corti was visualized using a Nikon (Japan) Eclipse 80i microscope. The high-magnification fluorescent images were obtained with a Leica TCS SP5 laser-scanning microscope (Wetzlar, Germany). Cells were counted manually on the stored images using the ImageJ software (Wayne Rasband, NIH, USA). The whole-mount cochleae were split into the apex, middle and base for counting the EdU-labelled SCs and HCs, and the cell counts were obtained per 100 μm length of the cochlea. At least five samples in each group from three independent experiments were collected for statistical analysis. The cell counts for the control and treated groups were compared using Student's t test.

| Scanning electron microscopy
The cultured cochleae in each group were rapidly placed into 2.5% glutaraldehyde (pH 7.4) in a refrigerator at 0-4°C for at least 4 hours.
The samples were washed with 0.1 M phosphate buffer and fixed with 1% osmium tetroxide for 2 hours. After further rinsing with phosphate buffer, the cochleae were dehydrated by gradient alcohol solutions and acetone and then soaked and embedded in acetone and embedding solution. The embedded tissues were put into an oven at 37°C overnight and then at 45°C for 12 hours and finally were polymerized at 60°C for 24 hours. Semi-thin sections were prepared after staining, and then, ultra-thin sections with a thickness of about 50-60 nm were sliced on an LKB-1 ultra-thin slicer.
After double staining with 3% uranyl acetate and lead citrate, the specimens were observed and photographed under a JEM-1200EX transmission electron microscope.

| RNA analysis
Ten cultured cochleae from independent culture groups were dissolved in RNALater for extraction of total RNA. RNA-Seq libraries were generated using the Illumina mRNA-Seq Sample Prep Kit.
The RNA-Seq data sets were processed by Salmon. 16 The reference We also performed quantitative real-time PCR (qRT-PCR) to validate the differentially expressed genes. Three to five cultured cochleae from independent culture groups were pooled, and total RNA was extracted using the RNeasy Micro kit (QIAGEN) according to the manufacturer's protocol. The mRNA was reverse transcribed to synthesize cDNA using the GoScript Reverse Transcription System (Promega, Australia). qRT-PCR was performed using GoTaq qPCR Master Mix (Promega, Australia). GAPDH was used for calibration.  (B-B2) The cochleae were treated with 5 μM DBZ and EdU for 3 days, and there were some EdU + /Sox2 + cells, and these were mostly in the apical turn but also in the middle turn of the cochlea. The number of SCs increased compared to the control group, especially in the apical turn. (C-C2) In the cochleae treated with 10 μM DBZ, there were many more EdU + /Sox2 + cells than in the 5 μM DBZ-treated group. (D-D2) The cochleae were cultured with EdU for 3 days, and there were no obvious EdU + /Myo7a + cells. (E-E2) The cochleae were treated with 5 μM DBZ and EdU for 3 days, and there were some EdU + /Myo7a + cells in the apical and middle turns of the cochlea, and most of the EdU + / Myo7a + cells were in the apical turn. (F-F2) In the cochleae treated with 10 μM DBZ, there were many more EdU + /Myo7a + cells than in the 5 μM DBZ-treated group. Histograms show the numbers of EdU + /Sox2 + cells (G1), Sox2 + cells (H1), EdU + /Myo7a + cells (G2) and Myo7a + cells (H2) in the control and DBZ-treated cochleae. Most of the EdU + /Sox2 + cells and EdU + /Myo7a + cells were in the apical turn of the cochlea (*P < .05, **P < .01, ***P < .001)

| Statistics
All data were analysed with GraphPad Prism 6.0 using a two-tailed, unpaired Student's t tests when comparing two groups or with a one-way ANOVA followed by a Dunnett's multiple comparisons test when comparing more than two groups. All data are expressed as either a percentage or as the mean ± SEM. A P-value < 0.05 was considered statistically significant.

| DBZ induced the proliferation of SCs in neonatal mouse cochleae in vitro
We first investigated the effect of DBZ on promoting the proliferation of SCs in cultured mouse cochleae. Cochleae were harvested from P1 C57/BL6 mice and then cultured in DMEM/F12 media with 5 μM or 10 μM DBZ for 3 days, and 10 μM EdU was added for the entire culture period (Figure 1, Figure S1). No Sox2 + /EdU + cells were observed in the vehicle control group ( Figure 1A, Figure S1A, S2A). There were some proliferating SCs as indicated by Sox2 + /EdU + cells in the cochleae treated with 5 μM DBZ, and these were primarily seen in the apical turn with some in the middle turn of the cochlea, while very few were seen in the basal turn ( Figure 1B, 1G1, Figure S1B, S2B). In the cochleae treated with 10 μM DBZ, there were significantly more Sox2 + /EdU + SCs in the apical and middle turns, with most of the Sox2 + /EdU + SCs still seen in the apical turn ( Figure 1C, G1, Figure S1C, S2C). The total number of SCs increased after the DBZ treatment ( Figure 1H1).

| DBZ generated new HCs and increased the HC number in neonatal mouse cochleae in vitro
To investigate the effect of DBZ in HCs in cultured mouse cochleae, we cultured P1 mouse cochleae with 5 μM or 10 μM DBZ and 10 μM EdU for 3 days (Figure 1, Figure S1, S2). In the control group, there were no obvious Myo7a + /EdU + cells ( Figure 1D, Figure S1D, S2A). The total HC number was increased, and there were some Myo7a + /EdU + cells in the cochleae treated with 5 μM DBZ, and these were mostly in the apical turn but also in the middle turn of the cochlea, while there were no obvious Myo7a + / EdU + cells in the basal turn ( Figure 1E, 1 G2, Figure S1E, S2B). In the cochleae treated with 10 μM DBZ, the total HC number was significantly increased and there were many more Myo7a + /EdU + cells in the apical and middle turns of the cochleae ( Figure 1F, G2, Figure S1F, S2C).
Interestingly, among the Myo7a + /EdU + cells in the DBZ-treated cochleae, most of them were also Myo7a + /EdU + /Sox2 + triple positive, and only a few of the Myo7a + /EdU + cells were not Sox2 + , suggesting that the newly regenerated HCs originated from the proliferating Sox2 + SCs ( Figure 1D,F, Figure S1D, S1F). Together, these results demonstrate that DBZ treatment increases the HC number and promotes supernumerary HCs in the apical and middle turns of cultured cochleae in vitro ( Figure 1H2).

| The newly generated HCs induced by DBZ had normal stereocilia bundle structures
The P1 mouse cochleae were treated with 10 μM DBZ and 10 μM EdU for 3 days and then stained with the antibody against Espin, which is an actin binding and bundling protein known to participate in stereocilia elongation during development. 18 The majority of the DBZ-induced EdU + /Sox2 + cells also showed positive Espin staining.
There were many Espin + /EdU + cells in the apical and middle turns of the cochleae, and the number of Espin + cells was clearly increased

| DBZ treatment reduced the expression of Jagged-1 on the cytomembrane of SCs
Previous reports have shown that Notch inhibition can promote

| DBZ induced the proliferation of SCs and mitotic regeneration of HCs after HC ablation in neonatal mouse cochleae in vitro
Neomycin is widely used to induce the HC damage model. To investigate the detailed effect of DBZ in promoting the proliferation of

| The newly regenerated HCs induced by DBZ treatment had HC stereocilia bundle structures
To determine whether the newly regenerated HCs also had stereocilia bundles, P1 mouse cochleae were treated with 1.0 mM neomycin for 12 hours and then cultured in DMEM/F12 media with 5 μM DBZ or 10 μM DBZ and 10 μM EdU for 7 days (Figure 4, Figure S2).
DBZ induced many Myo7a + /EdU + cells and Sox2 + /EdU + cells in the apical turn of the cochlea ( Figure 4B-B3,C-C3), and the newly regenerated HCs had hair bundles ( Figure 4G,H,I, Figure S3). These results showed that most of the newly regenerated HCs induced by DBZ had HC stereocilia bundle structures.

| DBZ-induced newly regenerated HCs had more mature bundle structures than those induced by DAPT in the apical turn of the cochlea
To investigate the bundle structure of newly regenerated HCs induced by DBZ, we cultured P1 mouse cochleae and treated them with in the control and DBZ-treated cochleae. Most of the EdU + /Myo7a + cells and EdU + /Sox2 + cells were in the apical turn of the cochlea. The number of Myo7a + cells and Sox2 + cells in DBZ-treated group increased compared to the control group, especially in the apical turn (*P < .05, **P < .01, ***P < .001) the 10 μM DBZ-treated group the new HCs had longer and more organized mature bundle structures ( Figure 5C-C1).

| The mechanism of DBZ-induced SC proliferation and HC regeneration
To assess the genome-wide gene expression among the neomycinonly controls, the DBZ-treated group and the DAPT-treated group, we compared the transcripts of the cochleae from each group using mRNA sequencing. We selected the significantly differentially expressed genes between the DBZ-treated groups and the DAPTtreated and control groups. Their expression patterns are shown in Figure 6A with detailed pathway information and related mechanisms shown in Figure 6B. Among the differentially expressed genes between the DBZ and control groups (Table S1), we found that many The qRT-PCR experiment results also showed in the DBZ-or DAPT-treated groups, the Jag1, Hes1, Hes5 and Notch1 genes were down-regulated, and the β-catenin, Wnt1 and Atoh1 genes up-regulated ( Figure 6D). These results were consisted with the RNA-seq, which also indicated that DBZ inhibited the Notch signalling pathway and activating the Wnt signalling pathway.

| D ISCUSS I ON
Mature mammalian cochlear HCs do not spontaneously mount a proliferative response after HC degeneration compared to the avian inner ear and the zebrafish lateral line 23,24 in which new HCs are generated from the surrounding SCs by direct transdifferentiation or by mitotic regeneration. [25][26][27] In this study, adding DBZ to the cul-  We show here that the newly regenerated HCs induced by DBZ treatment had stereocilia bundle structures, which showed that DBZ has a distinct advantage over many other γ-secretase inhibitors reported in a previous study. 18 The stereocilia bundle is the site of mechanoelectrical transduction of HCs and is essential for HCs to maintain their physiological function. 18,34 The fact that DBZ promotes the regeneration of HCs with intact stereocilia bundles is a favourable indication that DBZ might lead to recovery of hearing function.
We performed RNA-seq and analysed the gene expression between the control group, the DBZ-treated group and the DAPTtreated group in order to further explore the mechanism behind the SC proliferation and HC regeneration induced by DBZ treatment. Among the differentially expressed genes between the control and DBZ-treated group, we found that the Notch target genes In summary, DBZ can induce the proliferation of SCs and promote the mitotic regeneration of HCs with mature stereocilia bundles in both the absence and presence of HC injury in the mouse cochlea and in a dose-dependent manner. DBZ treatment can significantly down-regulate Notch signalling genes and activate Wnt signalling and some cilia-related genes in the mouse cochlea, which is likely the mechanism through which DBZ promotes SC proliferation and differentiation into HCs with relatively mature stereocilia. Thus, DBZ might be a new and useful therapeutic drug for HC regeneration in the mammalian cochlea to treat hearing loss.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request. F I G U R E 6 Gene expression and related mechanisms involved in DBZ-induced HC regeneration. (A) Hierarchical clustering of expression patterns of all differentially expressed genes. Red represents up-regulated expression levels, and blue represents down-regulated expression levels. Each row represents one gene, and each column represents one experimental group. (B) Differentially expressed genes in the DBZ vs. control group and the DBZ vs. DAPT group, including genes involved in the cell cycle, cell cilium morphogenesis, HC differentiation and Notch/Wnt signalling pathways. Significantly up-regulated genes are labelled with red, and down-regulated genes are labelled with blue. The significance between DBZ vs. control was set at logFC > 1.25 and p-value < 0.1. The threshold between DBZ vs. DAPT was set at p-value < 0.1. If the number of genes involved in each gene set was more than 60, the most significantly differentiated 60 genes are shown (ordered by logFC value). (C) Protein interaction network analysis of genes involved in cilium-related functions that are significantly upregulated (red) and down-regulated (blue) in the control, DAPT and DBZ-treated groups. The grey lines indicate protein-protein interactions recorded in the STRING database. (D) The results of qRT-PCR showed that the DBZ treatment led to the decreased expression of Jagged-1, Hes1, Hes5, Notch1 and increased expression of β-catenin, Wnt1 and Math1, which was consistent with the RNA-seq results