OHC-TRECK: A Novel System Using a Mouse Model for Investigation of the Molecular Mechanisms Associated with Outer Hair Cell Death in the Inner Ear

Outer hair cells (OHCs) are responsible for the amplification of sound, and the death of these cells leads to hearing loss. Although the mechanisms for sound amplification and OHC death have been well investigated, the effects on the cochlea after OHC death are poorly understood. To study the consequences of OHC death, we established an OHC knockout system using a novel mouse model, Prestin-hDTR, which uses the prestin promoter to express the human diphtheria toxin (DT) receptor gene (hDTR). Administration of DT to adult Prestin-hDTR mice results in the depletion of almost all OHCs without significant damage to other cochlear and vestibular cells, suggesting that this system is an effective tool for the analysis of how other cells in the cochlea and vestibula are affected after OHC death. To evaluate the changes in the cochlea after OHC death, we performed differential gene expression analysis between the untreated and DT-treated groups of wild-type and Prestin-hDTR mice. This analysis revealed that genes associated with inflammatory/immune responses were significantly upregulated. Moreover, we found that several genes linked to hearing loss were strongly downregulated by OHC death. Together, these results suggest that this OHC knockout system is a useful tool to identify biomarkers associated with OHC death.

Sound waves deflect stereocilia bundles on the apical surfaces of hair cells to activate mechanoelectrical transduction channels, which causes depolarization and activation of afferent neurons [1][2][3] . There are two types of hair cells in the mammalian cochlea, namely, inner hair cells (IHCs) and outer hair cells (OHCs). IHCs are conventional sensory receptors that transmit most of the acoustic information to the brain via ribbon synapses and type I spiral ganglion neurons (SGNs), which represent 90-95% of all SGNs, forming auditory afferent fibers [1][2][3] . In contrast, OHCs are mainly innervated by efferent fibers and contact with type II SGNs, which constitute only 5-10% of all SGNs 1-3 . However, OHCs play an important role in enhancing the sensitivity to sound. The main task of OHCs is the amplification of sound-induced vibrations via two mechanical activities, hair bundle motility and somatic motility 1,2,4 .
Genetic and environmental risk factors often damage hair cells and lead to the death of these cells. Previous studies have reported that OHCs are more susceptible to environmental risk factors than IHCs based on experiments on animal models. Noise exposure rapidly induces loss of OHCs compared to IHCs in mice [5][6][7] , guinea pigs 8 , and chinchillas 9 . The susceptibility to ototoxic drugs, antineoplastic agents (such as cisplatin), and aminoglycoside antibiotics (such as kanamycin and gentamicin) also differs between IHCs and OHCs. Although these drugs induce apoptotic cell death of both IHCs and OHCs via abnormal accumulation of reactive oxygen species and oxidative stress 1,10 , OHC loss occurs earlier and more extensively than IHC loss in mice [11][12][13] , rats 13 , guinea pigs 14 , and hamsters 15 . In addition, age-related damage is more extensive in OHCs than in IHCs. Several studies have reported that progressive OHC loss was more severe than IHC loss in aged mice 10,16-18 . Studies to elucidate the molecular mechanisms underlying the protection of OHCs from death and the regeneration of OHCs are important because OHC death is associated with several types of hearing loss. Although noise exposure, administration of ototoxic drugs and aged animals have been used as experimental tools to induce degeneration and loss of OHCs, each of these models has some limitations that restrict its use in OHC studies. OHC death is difficult to induce effectively via noise exposure and administration of ototoxic drugs, and researchers must wait for long periods for OHC loss to occur in aged mice. Moreover, the major and most common problem associated with these experimental models is that damage is caused not only to OHCs but also to other cells and tissues, such as afferent fibers or synapses [19][20][21] , the stria vascularis/lateral wall 6,17 and vestibular hair cells (VHCs) 12 . Therefore, new animal models are needed to selectively deplete OHCs without any damage to the other cochlear and vestibular cells to investigate the effect of OHC death on these other cells.
Here, we report on the OHC-toxin receptor-mediated conditional cell knockout (TRECK) system, which enables selective depletion of OHCs. The TRECK method involves conditional depletion of a target cell in vivo via administration of diphtheria toxin (DT) to transgenic (tg) mice carrying a human DT receptor (hDTR) cDNA transgene under the control of a tissue-specific promoter 22,23 . In this study, we show that the use of the OHC-TRECK system with a novel mouse model, Prestin-hDTR, which carries the prestin promoter and the hDTR gene, induces selective depletion of OHCs upon DT administration. Therefore, this mouse model could be useful in studies pertaining to OHC death, although the mechanisms and processes underlying OHC death here would differ from those in mouse models generated by noise exposure, ototoxic drugs, and aging.

Results
Generation of Prestin-hDTR mice. The goal of our study was to selectively deplete OHCs in vivo using the TRECK system (Fig. 1a). A genomic region including the promoter and noncoding exons 1 and 2 of murine Slc26a5 (solute carrier family 26, member 5 gene, also known as prestin), which is targeted to the lateral walls of OHCs 24 , was used to drive the expression of human heparin-binding EGF-like growth factor (hHB-EGF) as a DT receptor (Fig. 1b). The transgene also contained the polyadenylation signals (pA + ) of rabbit β-globin and simian virus 40 (SV40). hHB-EGF was modified at two amino acids (I117V and L148V) within the EGF-like domain. Furukawa et al. reported that modification of hHB-EGF reduces the side effects of EGF-like growth factor activity, such as phosphorylation of the EGF receptor and transduction of signals to neighboring cells, while maintaining the DT sensitivity of wild-type hHB-EGF 25 . We microinjected the construct into fertilized eggs from C57BL/6J mice and confirmed the integration of the transgene into five mice, and three founders exhibited germline transmission. We administered DT to mice from these three tg lines (#38, #53, and #65) at postnatal days 28 (P28) by intraperitoneal (i.p.) injection at a dose of 50 μg/kg and analyzed the OHC phenotype and gene expression levels by immunohistochemistry and quantitative RT-PCR (qRT-PCR) after 7 days, as shown in Fig. 1c. The #38 tg mice exhibited the most severe phenotypes after DT administration. The OHC-specific protein expression of SLC26A5/prestin in DT-treated Prestin-hDTR #38 mice was nearly abolished (Fig. 1d). The expression levels of the OHC marker genes Slc26a5; Ocm (oncomodulin), which is highly expressed in the cytoplasm of OHCs 26 ; and Strc (stereocilin), which is localized in the horizontal top connectors of OHC hair bundles 27,28 , were dramatically decreased to nearly undetectable levels in the cochlear mRNA of DT-treated Prestin-hDTR #38 mice compared with the expression levels of these genes in untreated Prestin-hDTR mice (Fig. 1e). Moreover, by ligation-mediated PCR, we confirmed that the transgene in the #38 mice was integrated into the E3 region on chromosome 14 ( Supplementary Fig. S1). There are no known coding or noncoding genes in the vicinity of the integration site. Therefore, we selected the #38 line as a suitable Prestin-hDTR mouse model for further experiments.
In vivo oHC depletion by oHC-tReCK. First, we investigated the phenotypes of the OHCs by OHC-TRECK of Prestin-hDTR mice at P28 using immunohistochemical and histopathological analyses (Fig. 2a). By staining for myosin VI (MYO6), a known hair cell marker 18,29 , we confirmed that OHCs from Prestin-hDTR mice were almost all lost from the organ of Corti 7 days after administration of DT. Strong MYO6 signals were detected in the cuticular plate and the cytoplasm in both IHCs and OHCs from the apical, middle, and basal areas of the cochleae of DT-treated wild-type mice (Figs 2b and S2), suggesting that DT administration does not affect the hair cells of wild-type mice. In contrast, the signal disappeared from the OHC areas of the cochleae of DT-treated Prestin-hDTR mice (Figs 2b and S2). To determine the time at which OHC loss had occurred, we counted the number of MYO6-positive OHCs from the apical, middle, and basal areas of the cochlea 0, 3, 4, and 7 days after administration of DT (Fig. 2a). Although there was no distinct loss of OHCs in Prestin-hDTR mice several hours (day 0) after administration of DT, we detected approximately 21, 23, and 40% OHC loss at the apical, middle, and basal turns, respectively, on day 3 (Fig. 2c), suggesting that OHC depletion occurred starting from the basal turn of the cochlea. After 4 days, most of the OHC loss had occurred, with 0 to 6% of the cells remaining in the apical, middle, and basal turns of DT-treated Prestin-hDTR mice (Fig. 2b). The rates of OHC loss in DT-treated Prestin-hDTR mice were similar after 7 days.
Moreover, the abnormal phenotypes of OHCs associated with OHC-TRECK were confirmed by histopathological analysis by means of light microscopy (LM) and scanning electron microscopy (SEM). The three OHC nuclei and the single IHC nucleus could be observed by LM in untreated Prestin-hDTR mice. In contrast, the DT-treated Prestin-hDTR mice had no OHC nuclei and exhibited severe degeneration of the tunnel of Corti (Fig. 2d). According to LM analysis, there were no obvious differences in the phenotypes of spiral ganglion cells (SGCs) and the stria vascularis between the untreated and DT-treated Prestin-hDTR mice (Fig. 2d). SEM analysis revealed the phenotypes of the apical surfaces of OHCs in DT-treated Prestin-hDTR mice. V-shaped stereocilia, (2019) 9:5285 | https://doi.org/10.1038/s41598-019-41711-2 www.nature.com/scientificreports www.nature.com/scientificreports/ which is the typical morphology for OHC stereocilia, were rarely detected in DT-treated Prestin-hDTR mice (Fig. 2e). The surfaces of most OHCs showed loss of stereocilia and the typical scar formed by apical expansion of neighboring supporting cells to close the lesions resulting from OHC loss 12,14 .
Next, we recorded the distortion product otoacoustic emission (DPOAE) and auditory brainstem response (ABR) in wild-type and DT-treated Prestin-hDTR mice to confirm the expected decrease or loss of hearing ability by OHC-TRECK (Fig. 3a). We also evaluated the hearing abilities of DT-treated wild-type and untreated Prestin-hDTR mice to determine whether integration of the transgene and administration of DT had any effect on hearing. Loss of DPOAE, which is a measure of OHC function 30,31 , was detected in the DT-treated Prestin-hDTR mice (Fig. 3b). Although determination of DPOAE amplitudes was difficult at low frequencies (4, 5.7, and 8 kHz), the DPOAEs in wild-type, DT-treated wild-type, and untreated mice showed similar levels at all frequencies (Fig. 3b). The ABR thresholds to sound stimuli at 4, 8, 16, and 32 kHz were measured in Prestin-hDTR mice at several time points after DT administration, as shown in Fig. 3c. The ABR thresholds appeared to slowly increase over 3 days, but the differences were not significant. At day 4, the thresholds increased dramatically, reaching a level that was indicative of severe and profound hearing loss to sound stimuli of all frequencies. Moreover, we confirmed that there were no significant differences in the levels of hearing loss between days 4 and 7 after administration of DT. Figure 3d shows representative ABR waveforms for stimuli at the highest sound pressure level www.nature.com/scientificreports www.nature.com/scientificreports/ (dB SPL) at each frequency recorded for DT-treated and untreated wild-type and Prestin-hDTR mice. Similar and clear ABR wave I-V amplitudes were detected at all frequencies for the wild-type, DT-treated wild-type, and untreated Prestin-hDTR mice. In contrast, the ABR amplitudes for DT-treated Prestin-hDTR mice were remarkably decreased at all frequencies.
evaluation of selective oHC depletion by oHC-tReCK. We next investigated the effects of OHC-TRECK, administration of DT and integration of the transgene in cells neighboring OHCs using qRT-PCR and immunohistochemistry (Fig. 4a). We first quantified the mRNA expression levels of Slc17a8 (solute carrier family 17, member 8, also known as Vglut3) and Sox2 (SRY-box 2), which are markers of IHCs and supporting cells (SCs), respectively 32,33 . There were no significant differences in the mRNA expression levels of Vglut3 and Sox2 in DT-treated Prestin-hDTR mice (Fig. 4b). Next, we investigated the phenotypes of IHCs and SCs, which include Deiters' cells (DCs) and outer pillar cells (OPCs), by performing whole-mount staining for phalloidin, anti-MYO6, and anti-SOX2 antibodies (Figs. 4c,d). The morphology of the stereocilia bundles of IHCs and the staining pattern of MYO6 from DT-treated Prestin-hDTR mice were consistent with those from wild-type mice (Fig. 4d), and there was no change in the number of IHCs in DT-treated Prestin-hDTR mice (Fig. 4e). Although the organization of SOX2-positive nuclei was slightly disrupted in DT-treated Prestin-hDTR mice (Fig. 4c), the number of DCs, which intercalated with the three rows of OHCs in their apical phalangeal processes 33,34 , after DT treatment of Prestin-hDTR mice was similar to that of wild-type mice (Fig. 4e). There were also no significant changes in the number of SOX2-positive nuclei from the OPCs, which are in direct contact with the first row of OHCs 33 , by OHC-TRECK (Fig. 4e).
Moreover, we investigated the effect of OHC-TRECK on VHCs in the utricle, crista, and saccule ( Fig. 5a,b). The morphologies of the VHC stereocilia and the number of VHCs from DT-treated Prestin-hDTR mice were the same as those from wild-type mice in the three vestibular tissues (Fig. 5c,d). We also investigated the vestibular functions of DT-treated Prestin-hDTR mice (Fig. 6a) by open-field behavior tests. The tests confirmed that the behaviors of DT-treated Prestin-hDTR mice were normal, and the mice did not exhibit hyperactivity or circling behavior, which were observed in mice with vestibular dysfunction (Fig. 6b,c).
Changes in gene expression profiles in the cochlea by OHC-TRECK. We next performed differential gene expression analysis of cochlear RNAs isolated from the wild-type, DT-treated wild-type, Prestin-hDTR, and DT-treated Prestin-hDTR mice using a microarray to determine the changes in gene expression in OHCs after OHC-TRECK. We first investigated the effects of transgene integration on gene expression by comparing differentially expressed genes (DEGs) between wild-type and Prestin-hDTR mice. We found that 92 and 232 probes were up-and downregulated, respectively, with a ≥2-fold change ( Supplementary Fig. S3a,c). We also determined the DEGs between wild-type and DT-treated wild-type mice to investigate the effect of DT treatment on gene expression. This analysis revealed that 238 probes were differentially expressed; 173 were upregulated and 65 were downregulated by DT administration ( Supplementary Fig. S3b,d). Thus, these results indicated that transgene integration and DT treatment may influence gene expression profiles. Moreover, we identified the common upand downregulated probes, which could lead to a bias in data analysis due to the effects of the transgene and DT administration, by pairwise comparisons between the "wild-type vs Prestin-hDTR" and "wild-type vs DT-treated Prestin-hDTR" groups ( Supplementary Fig. S3a,c) and between the "wild-type vs DT-treated wild-type" and "wild-type vs DT-treated Prestin-hDTR" groups ( Supplementary Fig. S3b,d). We found 113 common up-and downregulated probes (7 in Supplementary Fig. S3d), and these probes were removed from further analysis in this study.
Gene expression analysis identified 449 upregulated probes (fold change cutoff of 2) (Fig. 7b). Within the probe sets, 98 probes (94 genes) overlapped in pairwise comparisons among the "wild-type vs DT-treated Prestin-hDTR", "DT-treated wild-type vs DT-treated Prestin-hDTR", and "Prestin-hDTR vs DT-treated Prestin-hDTR" groups ( Fig. 7b and Supplementary Table S1), and these probes were further analyzed as candidates of OHC-TRECK-specific upregulated genes. Figure 7c highlights a cluster that contains 6 of the top 10 differentially upregulated genes ( Supplementary Fig. S4) from the 98 probe sets. The most highly upregulated gene (fold change of +18.38) was Vgf (VGF nerve growth factor inducible), which is a neuronal polypeptide that is known to be overexpressed by nerve injury and endoplasmic reticulum stress-induced cell death as a neuroprotective effect 35,36 . Gene ontology (GO) analysis of the 94 genes revealed 79 significant GO terms of the "biological process" category (P < 0.05) (Supplementary Table S2). Within the category of "biological process", the enriched GO terms were "biological regulation" (22.97%) and "response to stimulus" (20.1%) (Fig. 7d). Moreover, the association with the GO term "immune system process" was statistically significant (P = 0.0437). Among the broader GO terms, there was a significant association with "response to stress", "immune response", "leukocyte migration", "regulation of immune system process", "regulation of response to stimulus", and "activation of immune response" (Fig. 7e). Moreover, we observed upregulation (fold change of +4.78) of Wnt2 (wingless-type MMTV integration site family, member 2), the expression of which had previously been reported to be significantly increased 3 days after DT treatment in RNA from iDTR;Gfi1-Cre mice 37 (Supplementary Fig. S4 and Supplementary Table S1). The OHC-TRECK-specific upregulated genes in the microarray analysis were validated using qRT-PCR analysis. This analysis confirmed that eight genes among the top 11 highly upregulated genes in the microarray analysis were significantly upregulated in the cochlear RNA from DT-treated Prestin-hDTR mice, although the expression top panels, are shown in the bottom panels. Asterisks indicate the tunnel of Corti. SV, stria vascularis and SGC, spiral ganglion cell. (e) SEM images of OHCs from the middle area of the cochleae in the DT− and DT+ Prestin-hDTR mice. High-magnification images of the OHCs indicated by dotted boxes in the top panels are shown in the bottom panels. The arrow indicates an OHC that retained stereocilia in DT+ Prestin-hDTR mice.
www.nature.com/scientificreports www.nature.com/scientificreports/ of Ddx43 (DEAD box polypeptide 43) and Gm10433 (uncharacterized noncoding RNA gene) was undetectable by qRT-PCR, and the expression levels of Ccr2 and Wnt2 were not significantly upregulated in DT-treated Prestin-hDTR mice (Fig. 7f). We also validated the expression levels of five immune genes, which were predicted to be highly expressed in the cochlea from microarray data. The expression levels were significantly upregulated in DT-treated Prestin-hDTR mice (Fig. 7f). www.nature.com/scientificreports www.nature.com/scientificreports/ We confirmed that 40 probes (36 genes) were differentially downregulated more than two-fold by pairwise comparison among the "wild-type vs DT-treated Prestin-hDTR", "DT-treated wild-type vs DT-treated Prestin-hDTR", and "Prestin-hDTR vs DT-treated Prestin-hDTR" groups ( Fig. 8a,b and Supplementary Table S3). GO analysis confirmed that 15 of the 36 genes were significantly associated with hearing-related GO terms   Table S3). High and/or specific expression of Strc, Chrna10, Chrna9, and Ocm in OHCs has been demonstrated previously [26][27][28]51,52 . We performed qRT-PCR analysis to validate the downregulation of the genes responsible for hearing loss in humans and mice. The analysis confirmed significant downregulation of these genes in the cochlear RNA from DT-treated Prestin-hDTR mice (Figs 1e and 8c). oHC-tReCK in neonatal mice. Next, we investigated the phenotypes associated with OHC-TRECK in neonatal mice (P1) using immunohistochemical analysis (Fig. 9a). The number of OHCs in neonatal mice was clearly decreased in Prestin-hDTR mice 7 days after administration of DT (Fig. 9b,c). However, a few OHCs survived in the cochleae of DT-treated juvenile Prestin-hDTR mice at P8 compared with those of adult stages. Moreover, obvious damage was observed in the IHCs of DT-treated Prestin-hDTR mice. The number of IHCs decreased significantly in DT-treated Prestin-hDTR mice (Fig. 9b,d). Abnormal stereocilia bundles were observed in many IHCs from DT-treated Prestin-hDTR mice.

Discussion
Several cellular knockout methods for hair cells and SCs using DT have been established and are widely utilized in studies of hair cell regeneration 37,53,54 , the innate immune system in cochlear cells 55 , SGC survival after hair cell death 56 , and the relationship between vocalization and auditory experience 57 . In this study, we established Prestin-hDTR mice, which allow in vivo depletion of almost all OHCs approximately 4 days after i.p. administration of DT (Figs 1d, 2b,c and S2). Although the same effect on OHCs can be achieved by tamoxifen-mediated induction in Prestin CreERT2+/− ;Rosa26 DTA/− mice 58 , Prestin-hDTR mice exhibit OHC depletion by simple DT administration, which will prove useful for understanding the consequences of OHC death.
Moreover, our results suggested that the OHC-TRECK method exhibits selective depletion without any damage to IHCs, SCs (DCs and OPCs), SGCs, the stria vascularis, and VHCs (Figs 2d, 4 and 5) 7 days after administration of DT. This selective depletion is very important because the effects of specifically killing OHCs on other   Fig. S4) of 98 common probes (94 genes) (b). (d,e) Gene ontology (GO) analysis of 98 commonly upregulated genes by OHC-TRECK. A pie chart with 7 subdivisions shows GO terms for the "biological process" category (d). *P = 0.0437. The horizontal bar chart shows the enriched GO terms within "response to stimulus" (orange bars) and "immune system process" (green bars) with statistical significance (P < 0.05). The number of genes within each category are indicated on the right side of each bar (e). (f) Validation of the microarray data. The expression levels of nine genes (Vgf, Apoc2, Cdsn, Sirpb1b, Lgals3, Ccr2, Cldn22, Loxl4, and Wnt2) among the top 11 highly www.nature.com/scientificreports www.nature.com/scientificreports/ cell types within the cochlea will be investigated by morphological and electrophysiological phenotypic analysis and gene expression analysis in short-and long-term studies. In addition, we expect that OHC-TRECK will be very useful for the comparison of models with noise-or ototoxic drug-induced HC death and to determine how these insults impact or damage other cell types.
We also investigated the effect of transgene integration and DT treatment on the auditory phenotypes and gene expression of DT-treated and untreated wild-type and Prestin-hDTR mice. In particular, the dose (i.p., 50 μg/ kg body weight) of DT used for TRECK in this study was extremely high compared with those used in previous studies, although DT administration did not lead to mouse death at this dose 22,23 . Comparison of the cochlear RNAs of untreated Prestin-hDTR and DT-treated wild-type mice showed that several genes were differentially expressed ( Supplementary Fig. S3). Although these results indicate the effects of the transgene and DT in mice, there are no significant GO terms (P < 0.05) associated with specific transgenesis-and DT treatment-related DEGs. Moreover, the hearing levels were normal in untreated Prestin-hDTR and DT-treated wild-type mice (Fig. 3b,d). Therefore, we predict that the effects of the transgene and DT on auditory phenotypes are extremely weak in adult mice.
Differential gene expression analysis using Prestin-hDTR mice may provide interesting information related to OHC death. We performed microarray analysis using RNA isolated from cochleae 7 days after DT treatment of Prestin-hDTR mice as a preliminary experiment and observed the upregulation of several immune-related genes, including cytokine receptors and ligands and CD antigens, which are associated with biological processes such as response to stress, immune response, and leukocyte migration (Figs. 7c-e and S4, Supplementary Tables S1 and S2). Upregulation of immune-related genes probably results from inflammatory/immune responses caused by OHC death. Previous studies have reported the upregulation of genes associated with inflammatory/ immune responses by administration of an ototoxic drug 59 , acoustic trauma 60 , and aging 61 . Immune and inflammatory responses are predicted to be essential responses to injury and loss of OHCs; therefore, we suggest that Prestin-hDTR mice can be used to identify genes associated with these responses. Although we performed cochlear gene expression analysis on day 7 (3 days after the death of almost all OHCs), the use of samples obtained 3 and 4 days after DT administration may be more useful because these are the stages at which OHC death occurs (Fig. 2c).
We also identified specific downregulated genes after selective OHC depletion and predicted that these genes are associated with the maintenance and function of OHCs. As expected, the downregulated genes included several genes responsible for the maintenance and function of OHCs, such as Strc, which is essential not only for the formation of the horizontal top connectors of OHC hair bundles but also for the cohesiveness of the mature OHC hair bundles for tip-link turnover 27 ; Chrna9, and Chrna10, which are required for normal synaptic function and integrity of the olivocochlear system in OHC neurons 38,39 ; and Ocm, which probably plays a role in OHC calcium homeostasis 26 (Fig. 8b and Supplementary Table S3). Moreover, the expression levels of the other nine deafness-related genes (Myo15, Gfi1, Lhx3, Barhl1, Pjvk, Tomt, Pou4f3, Tmc1, and Grxcr2) were downregulated more than 2-fold in DT-treated Prestin-hDTR mice. In OHCs from null mutant mice, depletion of Gfi1, Barhl1, Pjvk, Tomt, and Grxcr2 resulted in early-onset degeneration and severe phenotypes compared with IHCs 41,[43][44][45]48 . While the functions of the other downregulated genes in OHCs are not known, analysis of these genes may provide insight into the function and maintenance of OHCs. For example, Clrn2 (clarin 2), which encodes a transmembrane protein, belongs to the clarin family of genes. Another gene in the same family, CLRN1 (clarin 1), is a causative gene of Usher syndrome type III 50 . The expression of Clrn1 is higher in OHCs than in IHCs, and Clrn1 null mutant mice exhibit disorganization of OHC stereocilia 62 . Tcap (titin-cap) is another candidate gene that is responsible for hearing loss. Although Tcap is known as a causative gene of muscular dystrophy in humans 63 , the transcript is detected in the otic vesicles of mouse embryos 64 . Moreover, we are interested in Ppp1r17, which was one of the ten most downregulated genes. The protein encoded by the Ppp1r17 gene functions as an inhibitor of the Ser/Thr phosphatases, protein phosphatase 2A (PP2A) and protein phosphatase 1 (PP1) 65 . Liu et al. recently reported that aging leads to decreased protein levels of PP1 in the mouse cochlea 66 . Thus, these downregulated genes may contribute to the functions and maintenance of OHCs. In addition, these genes might be candidate causative genes for hearing loss in patients for whom the cause of hearing loss is unknown.
Although we expected that the OHC-TRECK system could be used in OHC studies at early postnatal stages, selective depletion of OHCs was not observed by the administration of DT to neonatal Prestin-hDTR mice. We observed obvious damage of the IHCs in DT-treated Prestin-hDTR mice after administration of DT at P1 (Fig. 9). Konishi et al. reported that postnatal administration of DT at days 7 and 14 led to decreased hearing levels and degeneration of IHCs, OHCs, and SGCs in wild-type C57BL/6 mice 67 , indicating that administration of DT alone at juvenile stages has several adverse effects on the immature cochlea. Thus, our TRECK system using Prestin-hDTR mice should be used for OHC studies at adult stages and is currently a limited model for the analysis of OHCs at juvenile stages, although the system might be able to improve the dose and administration method of DT, as several studies have used a similar system for research on hearing at juvenile stages 37,53,55,56 . Another likely explanation for the Prestin-hDTR model being less effective and selective in the neonatal period is that SLC26A5/prestin expression is low at P1. OHCs were predominantly eliminated by TRECK at P1, but several OHCs survived in almost all the cochleae (Fig. 9). The most prominent increase in prestin expression occurred upregulated genes and five immune-related genes (Ccl2, Ccl4, Ccl7, Cd28, and Mpeg1) were validated in the cochlear RNA of DT ± WT and Prestin-hDTR (P-hDTR) mice. The seven immune-related genes are highlighted in green. Error bars = SDs. Asterisks indicate statistical significance (*P < 0.05, **P < 0.01, and ***P < 0.001: one-way ANOVA with Tukey's multiple comparisons test) compared with DT+ P-hDTR mice. ns, no significant difference. www.nature.com/scientificreports www.nature.com/scientificreports/ after P6 in mice and rats 68,69 . Therefore, the efficiencies of OHC depletion could be improved using Prestin-hDTR mice at later juvenile stages that exhibit raised prestin expression.  www.nature.com/scientificreports www.nature.com/scientificreports/ Genomic DNA Purification Kit (Promega). PCR genotyping was performed for genomic DNA using the following primer set: 5′-ATA TCG ATT CGA AAG TGA CTG GTG CCT CGC-3′ and 5′-AGA CAG ACA GAT GAC AGC ACC ACA G-3′ for the HB-EGF cDNA. The mice were i.p. administered DT at a dose of 50 μg/kg body weight.

Methods
qRt-pCR. Total RNA was isolated from the cochleae of wild-type, DT-treated wild-type, Prestin-hDTR and DT-treated Prestin-hDTR mice at P35 (Figs 1c, 4a and 7a) using the PureLink RNA Mini Kit (Thermo Fisher Scientific, Grand Island, NY, USA) according to the manufacturer's instructions. qRT-PCR was performed using TB Green Premix Ex Taq II (Tli RNaseH Plus) (Takara Bio Inc., Kyoto, Japan) and primer sets for 29 target genes, with Gapdh (glyceraldehyde-3-phosphate dehydrogenase) as an internal control (Supplementary Table S5). qRT-PCR primers were designed for separate exons, except those for Sox2 and Ccr2, which are intron-less genes. The products were analyzed on a LightCycler 480 System II instrument (Roche Molecular Systems, Inc., Pleasanton, CA, USA). The signal values were normalized to the median Gapdh signals, and the geometric mean values of the target signals were calculated in triplicate. The expression levels of the genes in untreated wild-type (Figs. 1e, 4b and 8c) and DT-treated Prestin-hDTR (Fig. 7f) mice were assigned an arbitrary value of 1 for comparison.
Immunohistochemistry. The inner ears were removed from the heads of mice of several ages (Figs 1c, 2a, 4a, 5a and 9a) and fixed with 4% paraformaldehyde (PFA). The cochlear and vestibular tissues were dissected from inner ears that were decalcified in 5% EDTA/PBS at 4 °C for 2 days and from non-decalcified inner ears. The tissues were permeabilized in 0.25% Triton X-100 in PBS for 15-30 min and then subjected to three 5-min washes in PBS. After the samples were washed in PBS, nonspecific binding sites were blocked with 0.5% Blocking Reagent (Roche Molecular Biochemicals, Indianapolis, IN, USA) for 1 hour at RT. Samples were incubated with primary antibody diluted in Can Get Signal Immunostain Solution A (TOYOBO, Osaka, Japan) overnight at 4 °C. The following primary antibodies were used: rabbit polyclonal anti-SLC26A5/prestin (Santa Cruz Biotechnology, Dallas, TX, USA; sc-30163, 2 μg/ml), rabbit polyclonal anti-MYO6 (Proteus Biosciences Inc, Ramona, CA, USA; 25-6791, 5 μg/ml), and goat polyclonal anti-SOX2 (Santa Cruz; sc-17320, 2 μg/ml) antibodies. The samples were then washed three times for 5 min in PBS, and an Alexa Fluor 594-conjugated secondary antibody (Thermo Fisher Scientific) and Alexa Fluor 488-conjugated phalloidin (Thermo Fisher Scientific) were diluted to 20 μg/ ml and 4 units/ml, respectively, in Can Get Signal Immunostain Solution B (TOYOBO) for 1 hour at RT. Finally, the samples were washed three times for 5 min in PBS and then mounted onto glass slides using ProLong Gold antifade reagent (Thermo Fisher Scientific). Fluorescence images were obtained using a Zeiss LSM 710 confocal microscope (Carl Zeiss, Jena, Germany), and images were processed using ZEN2009 software (Carl Zeiss). The MYO6-positive OHCs and IHCs, and SOX2-positive DCs and OPCs per 100 μm were counted from the apical (1.2-1.5 mm from the apex), middle (2.1-2.4 mm from the apex), and basal (2.0-3.4 mm from the base) areas in the cochleae. The VHCs visualized by phalloidin were counted from the 2,500-μm 2 area within white boxes in the utricle, crista, and saccule as shown in Fig. 5b. Histological analysis. For preparation of paraffin-embedded cochlear sections, isoflurane-anesthetized mice were perfused through the heart with a buffer containing 4% PFA. The inner ears of Prestin-hDTR and DT-treated Prestin-hDTR mice at P42 (Fig. 2a) were dissected, fixed with 4% PFA overnight, and then decalcified in 5% EDTA/PBS at 4 °C. After decalcification for 14 days, tissues were dehydrated in an ethanol series, cleared in xylene, embedded in paraffin, sectioned (5 μm), and stained with hematoxylin.
SEM of the apical surfaces of OHCs in the ears of untreated and DT-treated Prestin-hDTR mice at P49 (Fig. 2a) was performed as previously described 70 using a Hitachi S-4800 field emission SEM instrument at an accelerating voltage of 10 kV.
Hearing tests. The hearing ability of the mice was evaluated by recording the DPOAE and ABR. To perform these hearing tests, mice were anesthetized by i.p. injection of a medetomidine-midazolam-butorphanol mixture (0.75 mg/kg medetomidine, 4 mg/kg midazolam, and 5 mg/kg butorphanol). After the experiment, mice were administered the alpha-2 adrenergic antagonist atipamezole (0.75 mg/kg).
DPOAEs were recorded from each ear of the wild-type, DT-treated wild-type, Prestin-hDTR, and DT-treated Prestin-hDTR mice at P35 (Fig. 3a) using the ER10X Extended Bandwidth Research Probe System (Etymotic Research, Elk Grove, IL, USA) and EMAV Plus software (version 3.32) (Etymotic Research). For recording DPOAEs, a small ear plug containing a low-distortion probe microphone (Etymotic Research, ER10X-P) and two probe tubes was set in the outer ear canal. Two test sounds were separately introduced into the ear through two tubes. The primary tone stimulus levels L 1 and L 2 were adjusted as L 1 (65 dB SPL) -L 2 (55 dB SPL) = 10 dB SPL. The frequency ratio f 2 /f 1 was fixed at 1.20, with f 2 = 32, 22.6, 16, 11.3, 8, 5.7, and 4 kHz. The cubic distortion product of 2f 1 − f 2 was measured as DPOAE.
ABRs evoked by tone pip stimuli for each pressure level at 4, 8, 16, and 32 kHz were recorded from each ear of the wild-type, DT-treated wild-type, Prestin-hDTR, and DT-treated Prestin-hDTR mice at various ages, as shown in Fig. 3a, using the ABR Workstation (TDT System III, Alachua, FL, USA) as previously described 70 . ABR thresholds were obtained for each stimulus by reducing the SPL first in 10-dB steps and then fluctuating the SPL in 5-dB steps to identify the lowest level at which an ABR pattern could be recognized.
Behavioral test. The wild-type and DT-treated Prestin-hDTR mice at P42 (Fig. 6a) were placed in a (2019) 9:5285 | https://doi.org/10.1038/s41598-019-41711-2 www.nature.com/scientificreports www.nature.com/scientificreports/ Microarray analysis. The cochleae (n = 8 per mouse) of wild-type, DT-treated wild-type, Prestin-hDTR, and DT-treated Prestin-hDTR mice at P35 (Fig. 7a) were used for microarray analysis. Total RNA samples from each mouse were prepared as mentioned above, and the qualities were assessed with a Bioanalyzer 2100 (Agilent Technologies). Two hundred nanograms of total RNA was amplified and labeled using the Agilent Low Input Quick Amp Labeling Kit (Agilent Technologies) according to the manufacturer's protocol. A total of 1.65 µg of Cy3-labeled cRNA was fragmented using the protocol for the Agilent Gene Expression Hybridization Kit. Hybridizations were performed for 17 hours at 65 °C and 10 rpm in a rotating hybridization oven. Slides were scanned with the SureScan microarray scanner (Agilent Technologies) using a scan resolution of 5 µm and the green dye channel with the PMT set to 100%. Data were obtained using Agilent Feature Extraction software (v11.5.1.1) with default settings for all parameters. Data mining was performed using GeneSpring GX ver. 14.9 software (Agilent Technologies). Normalization of the raw data was performed using the percentile shift (75th percentile) followed by baseline transformation to the medians of all samples. The DEGs were screened for a 2-fold change between samples. To interpret the results of transcriptome analysis, GO analysis using GeneSpring GX was conducted for DEGs between the DT-treated Prestin-hDTR mice and other samples (untreated wild-type, DT-treated wild-type, and untreated Prestin-hDTR mice). Microarray data have been deposited in the Gene Expression Omnibus (GEO) database at NCBI (https://www.ncbi.nlm.nih.gov/geo/) under accession number GSE121442. statistical analysis. All results are presented as the mean ± standard deviation (SD). The differences in mRNA expression levels (Fig. 1e), number of cells (IHCs, DCs, OPCs, and VHCs in adult mice, and OHCs and IHCs in neonatal mice) (Figs 4e, 5d and 9c,d) and behavioral data (Fig. 6c) were analyzed using Student's t-test. The number of OHCs at the adult stage was statistically analyzed by a two-way ANOVA with a Bonferroni post-hoc, including the data for all the days for both DT-treated wild-type and Prestin-hDTR mice. Differences in DPOAE levels (Fig. 3b) and mRNA expression levels (Figs 4b, 7f and 8c) were analyzed by a one-way ANOVA with the Tukey post hoc multiple comparison test. GraphPad Prism6 (GraphPad, San Diego, CA, USA) was used to calculate the column statistics and P values.

Data Availability
All data obtained or analyzed during this study are included in this manuscript. Raw data are available from the corresponding author upon reasonable request.