GATA3 maintains the quiescent state of cochlear supporting cells by regulating p27kip1

Haplo-insufficiency of the GATA3 gene causes hypoparathyroidism, sensorineural hearing loss, and renal disease (HDR) syndrome. Previous studies have shown that Gata3 is required for the development of the prosensory domain and spiral ganglion neurons (SGNs) of the mouse cochlea during embryogenesis. However, its role in supporting cells (SCs) after cell fate specification is largely unknown. In this study, we used tamoxifen-inducible Sox2CreERT2 mice to delete Gata3 in SCs of the neonatal mouse cochlea and showed that loss of Gata3 resulted in the proliferation of SCs, including the inner pillar cells (IPCs), inner border cells (IBCs), and lateral greater epithelium ridge (GER). In addition, loss of Gata3 resulted in the down-regulation of p27kip1, a cell cycle inhibitor, in the SCs of Gata3-CKO neonatal cochleae. Chromatin immunoprecipitation analysis revealed that GATA3 directly binds to p27kip1 promoter and could maintain the quiescent state of cochlear SCs by regulating p27kip1 expression. Furthermore, RNA-seq analysis revealed that loss of Gata3 function resulted in the change in the expression of genes essential for the development and function of cochlear SCs, including Tectb, Cyp26b1, Slitrk6, Ano1, and Aqp4.

GATA-binding protein 3 (GATA3) belongs to the GATA family transcription factors that bind the DNA sequence GATA. Previous studies have shown that GATA3 plays a role in the development of many tissues, including T-cells 1-3 , parathyroid 4,5 , kidney [6][7][8] , and cochlea 9 . In humans, haplo-insufficiency of the GATA3 gene causes hypoparathyroidism, sensorineural hearing loss, and renal disease (HDR) syndrome, also known as Barakat syndrome. Among these three phenotypes, all patients have some form of sensorineural deafness and many suffer from congenital deafness 10,11 , suggesting GATA3's essential role in cochlear development.
The mammalian inner ear is derived from the otic placode that invaginates and then closes to form the otic vesicle. The subsequent morphogenesis of the roughly spherical otic vesicle gives rise to a complex structure containing three semicircular canals, two gravistatic receptors (saccule, utricle), and the cochlea in mammals 12,13 . The sensorineural organ of the cochlea (the organ of Corti) is composed of sensory hair cells (HCs) and surrounding supporting cells (SCs). During cochlear development, GATA3 is continuously expressed from otic placode to mature cochlea [14][15][16] . Studies have shown that disruption of Gata3 results in abnormal cochlear sensory epithelium development and its inaccurate innervation 17,18 . Previously, we have used Pax2-Cre mice, which express Cre in the developing inner ear after E9.5 19 , to conditionally knock out Gata3 and have shown that Gata3 is required for the establishment of the prosensory domain and the survival of spiral ganglion neurons (SGNs) 15 . Recently, a study has shown that Gata3 is crucial for the maturation of inner hair cells (IHCs) 20 . However, the role of Gata3 in SCs after cell fate specification is largely unknown.
In this study, we used Sox2 CreERT2 to conditionally knock out Gata3 to investigate the role of Gata3 in the SCs of the neonatal mouse cochleae. Our results revealed that the deletion of Gata3 in neonatal cochleae led to the downregulation of p27 kip1 expression and cell cycle re-entry of SCs. Furthermore, chromatin immunoprecipitation (ChIP) analysis showed that GATA3 directly binds to the promoter of p27 kip1 . Together, our data demonstrated that GATA3 maintains the quiescent state of SCs by regulating the expression of p27 Kip1 .

Results
Loss of Gata3 in early postnatal cochlear SCs results in the hypertrophy of SCs and the expansion of GER. We first analyzed the spatiotemporal expression pattern of Gata3 in the neonatal mouse cochlea. We co-labeled cochleae of wild type mice with antibodies against GATA3 and SOX2, a SC marker, and revealed that GATA3 was broadly expressed in the greater epithelial ridge (GER), inner border cells (IBCs), inner phalangeal cells (IPhCs), inner pillar cells (IPCs), outer pillar cells (OPCs), Deiters' cells (DCs), Hensen's cells (HeCs), HCs, Claudius' cells (CCs), and cells in the outer sulcus at postnatal day 0 (P0) and P7 ( Fig. 1A-F,A′-F′).
To investigate the role of Gata3 in the SCs of the neonatal cochlea, we used the tamoxifen-inducible Sox2 CreERT2 mice to conditionally delete Gata3 in SCs. To generate tamoxifen-inducible, SC-specific Gata3 conditional knockout mice (Gata3 loxP/loxP ; Sox2 CreERT2/+ ), we first crossed Gata3 loxP/loxP mice with Sox2 CreERT2/+ mice to generate Gata3 loxP/+ ; Sox2 CreERT2/+ mice. Then, crossing Gata3 loxP/loxP and Gata3 loxP/+ ; Sox2 CreERT2/+ mice produced Gata3 conditional knockout mice. Without tamoxifen administration, Gata3 loxP/loxP ; Sox2 CreERT2/+ mice were viable and phenotypically indistinguishable from control littermates. In tamoxifen-treated Gata3 loxP/loxP ; Sox2 CreERT2/+ mice (Gata3-CKO mice hereafter), Sox2-CreERT2 is expected to remove the loxP-flanked exon 4 of Gata3 in SCs, resulting in a reading frame-shift and a premature termination codon in exon 5, which deletes the two zinc finger domains and the rest of GATA3 C-terminal sequences essential for DNA-binding and nuclear localization 15 . We administered two tamoxifen injections to activate CreERT2 with a 24-h interval at P1 and P2, and assessed the efficiency of Gata3 deletion at P3 by immunolabeling. The immunogen of the GATA3 antibody used in this study is encoded by exons 2 and 3. Thus, we expected to detect the truncated GATA3 without the zinc finger domains in the cytoplasm but not the nucleus in Gata3-CKO cochleae. Consistent with our prediction, the specific nuclear expression of GATA3, which was seen in the SCs of the control Gata3 loxP/loxP cochlea ( Fig. 2A-D,A′-D′), was ablated in Gata3-CKO SCs ( Fig. 2E-H,E′-H′), while the expression of GATA3 in HCs was not affected except at the apex (Fig. 2H,H′, arrowheads), likely due to active Sox2 expression in apex HCs at P1 and P2. Quantification of GATA3 nuclear expression cells (SOX2 + ) demonstrates that two tamoxifen dosages at P1 and P2 are sufficient to inactivate Gata3 in nearly all SOX2-expressing cells including cochlear SCs (Fig. 2I).  www.nature.com/scientificreports/ The expression of GATA3 (green) is localized in the nuclei of SOX2 + (red) SCs in control cochleae at P3 after the administration of tamoxifen at P1 and P2. (E-H,E′-H′) In Gata3-CKO cochleae, the specific nuclear expression of GATA3 is lost in the SOX2 + SCs but the nuclear expression of GATA3 in HCs is not affected except at the apex (H,H′, arrowheads). IHCs and OHCs are indicated with pound sign (#) and asterisk (*), respectively. (I) Quantification of the number of SOX2 + SCs with GATA3 nuclear expression in 3 adjacent sections (a total thickness of ~ 54 μm) shows the high rate of Gata3 deletion in Gata3-CKO mice. n = 3; **p < 0.01. (J-M,J′-M′) Immunolabeling of the control cochlea with anti-VGLUT3 (red), anti-MYO7A (green) and anti-SOX2 (magenta) reveals the normal pattern of SCs and the orderly arrangement of three rows of OHCs and one row of IHCs at P7 after the administration of tamoxifen at P1 and P2. (N-Q,N′-Q′) In Gata3-CKO cochleae, the orderly pattern of HCs is disrupted along with an increase in SCs (arrowheads) and an expansion of the GER region (bracket). (R,S) Quantification shows that the numbers of GER cells (R) and SCs (S) in 3 adjacent sections (a total thickness of ~ 54 μm) are significantly increased in Gata3-CKO cochleae compared with control mice. n = 3; *p < 0.05; **p < 0.01. Scale bars are 50 μm. www.nature.com/scientificreports/ To analyze the effects of Gata3 conditional knockout on neonatal cochlear SCs, we administered two tamoxifen injections to the mice at P1 and P2 and collected cochleae at P7. We used antibodies against SOX2, MYO7A, and VGLUT3 to identify SCs, HCs, and IHCs, respectively. The control mice exhibited a normal structure of the organ of Corti with three rows of OHCs and one row of IHCs surrounded by SCs arranged in order ( Fig. 2J-M,J′-M′). In contrast, an increased number of SCs and an expansion in the GER region were observed in Gata3-CKO mice ( Fig. 2N-Q,N′-Q′), as shown in quantification data (Fig. 2R,S). The increased SCs and GER expansion disrupted the organization of HCs, a single row of IHCs was inconsistently seen in the sagittal section ( Fig. 2N-Q,N′-Q′).
Gata3 deficiency causes the abnormal proliferation of IPCs, IBCs, and lateral GER in the neonatal cochlea. The observed increase in SCs and GER expansion in the neonatal cochleae of Gata3-CKO mice suggest that Gata3 could play a role in maintaining SC quiescent state. To test this hypothesis, we treated mice with tamoxifen at P1 and P2, and with 5-ethynyl-2′-deoxyuridine (EdU) once a day from P1 to P6. Cochleae were harvested at P7 for the analysis of cell proliferation by Click-it EdU labeling. Consistent with previous studies that postnatal SCs are mitotically quiescent 21 , we observed no EdU + cells in the SCs of the control Gata-3 loxP/loxP cochleae ( Fig. 3A-D,A′-D′). Previously, Sox2 haplo-insufficiency has been shown to cause an increase in the number of inner HCs and the proliferation of inner pillar cells in the neonatal cochlea 22 . Thus, we investigated the change in cell proliferation in Gata3 loxP/+ ; Sox2 CreERT2/+ cochleae and detected a small number of EdU + cells in the SCs and GER region of the apex turns ( Fig. 3H,H′, arrowheads), but not in the basal and middle turns ( Fig. 3E-G,E′-G′). In contrast, significantly more cells in GER and PC regions of Gata3-CKO mice were labeled with EdU throughout the length of the cochlear duct and the number of EdU + cells increased from the base to the apex of the cochlea ( Fig. 3I-L,I′-L′ arrowheads). Statistical analysis of the number of EdU + cells confirmed that compared to the control Gata3 loxP/loxP and Gata3 loxP/+ ; Sox2 CreERT2/+ cochleae, the number of EdU + cells was greatly increased in Gata3-CKO mice (Fig. 3M). Previous studies have shown that SC proliferation is accompanied by the generation of ectopic HCs 21,23-25 . However, no EdU + /MYO7A + cells were observed in Gata3-CKO mice ( Fig. 3I-L), indicating that no ectopic HC was generated from the proliferative SCs.
We then assessed whether the proliferative Gata3-CKO SCs gave rise to ectopic HCs using wholemount labeling of EdU and antibodies against PRESTIN, MYO7A, and SOX2. Compared to the quiescent state of HCs and SCs in the control mice (Fig. 3N,O), increased SC proliferation was detected in Gata3-CKO cochleae, which consequently displaced HCs (Fig. 3P,Q). Again, we observed no EdU + /MYO7A + cells in Gata3-CKO mice (Fig. 3P). In addition, statistical analysis showed no significant change in the number of IHCs in the basal cochleae and of OHCs of the entire cochleae, the number of IHCs was slightly increased in the apical and middle turns of Gata3 loxP/+ ; Sox2 CreERT2/+ (without tamoxifen) and Gata3-CKO cochleae compared with those in Gata3 loxP/loxP cochleae ( Fig. 3R-V, Table S4), suggesting that Sox2 haplo-insufficiency could be the cause for the small increase of IHCs in apical and middle turns. The lack of significant difference in the number of IHCs and OHCs between Gata3-CKO mice and Gata3 loxP/+ ; Sox2 CreERT2/+ mice (without tamoxifen) also showed that the Gata3 deficiency does not result in the generation of ectopic HCs. Taken together, these data indicate that Gata3 deletion leads to cell cycle re-entry of a large number of SCs throughout the entire length of the cochlea but not cell fate conversion of SCs into HCs.
To determine the onset and origin of SC proliferation, we performed a time course analysis by tamoxifen administration at P1 and P2, and collecting cochlear sections at P3-P9 for immunolabeling with antibodies against SOX2 and Ki67, a proliferative marker. In contrast to the absence of Ki67 expression in the SOX2 + SCs of control mice ( Fig. 4A-D,A′-D′,A″-D″), ectopic Ki67 expression was detected in IPCs, IBCs, and the lateral GER in Gata3-CKO mice at P3 (Fig. 4E-H,E′-H′,E″-H″). At P4-P5, there was an increasing number of Ki67 + proliferative cells restricted in these regions in Gata3-CKO mice compared to those of control mice (Fig. S1A,B). At P7, the Ki67 immunostaining signal began to wane (Fig. S1C) and mostly disappeared at P9, while the significant expansion of GER region remained (Fig. S1D). Thus, loss of Gata3 causes the cell cycle re-entry of IPCs, In Gata3-CKO cochleae, numerous EdU + cells are detected in PCs and the GER region (arrowheads). The number of EdU + cells increases from the base to the apex of the cochlea. (M) Quantification of the EdU + cells in 3 adjacent sections (a total thickness of ~ 54 μm) shows that the number of EdU + cells is significantly increased in Gata3-CKO cochleae compared with Gata3 loxP/+ ; Sox2 CreERT2/+ cochleae. No EdU + cell is detected in middle and basal turns of Gata3 loxP/+ ; Sox2 CreERT2/+ cochleae. The total number of proliferative cells gradually increases from base to apex in Gata3-CKO cochleae. n = 3; **p < 0.01. (N-Q) Whole-mount immunostaining shows the disorganization of HCs in Gata3-CKO cochleae as a result of SC proliferation. Antibodies against PRESTIN, MYO7A, and SOX2 are used to label OHCs, HCs, and SCs, respectively. No EdU + /MYO7A + cell is observed in Gata3-CKO cochleae. (R-T) Whole-mount cochleae of Gata3 loxP/loxP , Gata3 loxP/+ ; Sox2 CreERT2/+ (without tamoxifen), and Gata3-CKO mice are labeled with antibody against MYO7A. The middle turn of the cochlea is shown. Arrows indicate extra IHCs. (U,V) Quantification of IHCs (U) and OHCs (V) reveals the number of IHCs is slightly increased in apical and middle turns of Gata3-CKO cochleae compared with Gata3 loxP/loxP cochleae. However, there is no significant change in the number of HCs in Gata3-CKO cochleae compared with Gata3 loxP/+ ; Sox2 CreERT2/+ cochleae (without tamoxifen). n = 3 for Gata3 loxP/+ ; Sox2 CreERT2/+ cochleae and n = 5 for Gata3 loxP/loxP cochleae as well as Gata3-CKO cochleae; *p < 0.05; n.s. not significant. Scale bar 50 μm. www.nature.com/scientificreports/ IBCs, and the lateral GER, and the proliferation capability appears to be limited to these cells before P7. We also analyzed cochlear morphology at P12 to investigate the cell fate of extra SCs. We gave the mice tamoxifen injections at P1 and P2, and collected cochleae at P12. Compared with control mice, the Gata3-CKO mice did not have a typical inner sulcus (IS) structure but retained the expanded GER region (Fig. S2A-H).
GATA3 maintains the quiescent state of cochlear SCs by directly regulating p27 kip1 expression. Previous studies have revealed that the expression of p27 kip1 a cyclin-dependent kinase inhibitor, is essential to maintain the quiescence of SCs and that the absence of p27 kip1 leads to SC proliferation in neonatal www.nature.com/scientificreports/ cochlea [26][27][28] . In our previous study using Pax2-Cre to conditionally ablate Gata3 in the developing inner ear, we have shown that Gata3 is required for the establishment of the prosensory domain and that targeted deletion of Gata3 abolishes p27 kip1 expression in the presumptive prosensory region 15 . We thus investigated whether the observed cell proliferation in Gata3-CKO mice was caused by p27 kip1 down-regulation. Cryosections of control and Gata3-CKO cochleae at P4 were immunolabeled for SOX2 and p27 kip1 . As shown in Fig. 5A,B,A′,B′, while p27 kip1 expression was broadly detected in the SCs and GER region of the control cochlea, it was downregulated specifically in IPCs, IBCs, and the lateral GER of Gata3-CKO mice. Interestingly, these cells retained the expression of SOX2, which has previously been shown to directly modulate p27 kip1 expression 26 . RNAscope in situ hybridization was also used to assess the change in p27 Kip1 expression at mRNA level. The results showed the expression level of p27 kip1 mRNA was decreased specifically in IPCs, IBCs, and the lateral GER while its expression in OPCs, DCs, and HeCs remains high in Gata3-CKO mice cochleae (Fig. 5C,D,C′,D′), consistent with the results of anti-p27 kip1 immunolabeling. In addition, RT-qPCR analysis showed the expression of p27 kip1 was significantly reduced in Gata3-CKO cochleae (Fig. 5E). Our data suggest that the abnormal proliferation of IPCs, IBCs, and lateral GER caused by Gata3 deficiency could be due to the downregulation of p27 kip expression.
To determine whether GATA3 directly regulates p27 kip1 expression, we first analyzed for the binding of GATA3 to p27 kip1 promoter region in the publicly available GATA3 ChIP-seq databases [29][30][31] . GATA3 ChIP-seq data from different tissues or cell lines were downloaded from GEO (GSE20898, GSE92295, and GSE109109) and transferred from mm8/mm9 to mm10 using liftOver from UCSC 32 . All enriched ChIP-seq peaks (bed/bedGraph format) surrounding the p27 kip1 gene were then uploaded to the UCSC genome browser. Four GATA3 motifs were identified in the promoter region of p27 kip1 by using Homer 33 , FIMO 34 and JASPAR 35 . These motifs are located in the enriched peaks and contain the predicted GATA3-binding sites (GBSs) (Fig. 5F, Fig. S3). To confirm these GATA3 motifs in the cochlea, we then immunoprecipitated the chromatin from wild type cochlear ducts at P2 using anti-GATA3 antibody and performed PCR to amplify the GBS-containing sequences. GATA3 binding to GBS1 and GBS2 could be specifically resolved by ChIP-PCR while the binding of GATA3 to GBS3 and GBS4 was assessed together as these two binding sites are tightly associated and could not be effectively separated by shearing (Fig. S4). ChIP-PCR (Fig. 5G, Fig. S5) and ChIP-qPCR ( Fig. 5H) with anti-GATA3 antibody showed that these GBS-containing sequences were significantly enriched in the cochlear duct compared to IgG controls. Additionally, a coding sequence in exon 3 of p27 kip1 was used as a negative control (NC) to show no significant enrichment with anti-GATA3 antibody. Thus, our results indicate that GATA3 directly binds to p27 kip1 in vivo and could positively regulate its expression to maintain the quiescent state of cochlear SCs.

Transcriptomic changes in the cochlear duct of Gata3-CKO mice.
To determine the changes in the transcriptome of Gata3-CKO cochlear ducts, we performed RNA-Seq analysis at P4 after tamoxifen administration at P1 and P2. To visualize transcriptomic differences between the control and Gata3-CKO cochleae, a heatmap was generated showing that the transcriptome pattern of cochlear duct was significantly changed in Gata3-CKO mice (Fig. 6A). We have identified a total of 163 differentially expressed genes (fold-change ≥ 2 and P ≤ 0.05) between control and Gata3-CKO cochleae. Among them, 121 genes were downregulated and 42 genes were upregulated in the Gata3-CKO cochleae (Table S5). Next, we categorized these highly differentially expressed genes between control and Gata3-CKO cochleae into enriched categories according to GO analysis, and revealed that multiple biological processes were disturbed in Gata3-CKO cochleae, including neuronal action potential, cGMP biosynthetic process, positive regulation of potassium ion transmembrane transport, second-messenger-mediated signaling, small GTPase mediated signal transduction, and inorganic cation import across plasma membrane. (Fig. 6B). Based on their relatively high sequencing depth (read count) and their known roles in SC function (Fig. 6C), we selected five significantly down-regulated genes in Gata3-CKO cochleae, Tectb, Cyp26b1, Slitrk6, Ano1, and Aqp4, for further analysis. As shown in Fig. 6D, the RNA-seq values of these five selected genes were verified by RT-qPCR. In addition, we performed in situ hybridization experiments to analyze the spatial expression of Tectb, Cyp26b1, and Ano1, and confirmed that consistent with the data obtained from the RNA-Seq analysis, the expression of Tectb, Cyp26b1, and Ano1 were downregulated in the GER of Gata3-CKO mice (Fig. 6E arrow). Thus, disruption of Gata3 alters the expression of genes involved in cochlear development and diseases.

Discussion
Gata3 deficiency causes human HDR syndrome. Previous studies have shown that Gata3 is needed for the formation of the cochlear prosensory domain during embryonic development and for the survival and axonal projection of SGNs 15,17,18 . Moreover, a recent study has shown that Gata3 is also involved in IHC maturation 20 . In the present study, we used an inducible Sox2 CreERT2 line to delete Gata3 in the neonatal mouse cochlea. Our results have revealed that Gata3 is required to maintain the quiescent state of cochlear SCs and that the absence of Gata3 in SCs leads to the proliferation of IPCs, IBCs, and the lateral GER. In this study, we have shown that Sox2-CreERT2-mediated deletion of Gata3 is mostly limited to the SCs of the cochlea (Fig. 2). However, the expression of GATA3 in HCs was not affected except at the apex, which is likely due to active Sox2 expression in HCs of the apex at P1 and P2. Our results are largely in agreement with the previous reporter expression study showing that Sox2-CreERT2 is expressed in ~ 20% HCs in the basal turns to ~ 84% in the apical turns with tamoxifen given at P0/P1 and in ~ 60% HCs in the apical turns with tamoxifen given at P1 36 . Since the expression of SOX2 is rapidly down-regulated in HCs between P0 and P2 37 , the small difference between our results and those by Walters et al. 36 could be due to the slightly different timing of tamoxifen administration or due to the difference between the half-life of GFP and tdTomato reporter proteins and that of GATA3 or both.
During cochlear development, the terminal mitosis of sensory precursors occurs around E14.5, when the basal turn of the cochlea is the last region to exit the cell cycle 38 www.nature.com/scientificreports/ www.nature.com/scientificreports/ experimental conditions, SCs of neonatal mouse cochlea also exhibit a limited proliferation capacity in vivo. For example, the disruption of cell cycle genes such as p27  and Rb 41 results in the cell cycle reentry of SCs. In addition, Sox2, which is critical to the development of the cochlea 42,43 , is required for the expression of p27 kip1 in the developing cochlea and ablation of Sox2 expression leads to the loss of p27 kip1 expression in the prosensory domain at E14 44 . In neonatal mice, targeted deletion of Sox2 results in the proliferation of IPCs and the absence of p27 kip1 expression in IPCs, while other SCs remain quiescent. Conversely, p27 kip1 -null IPCs proliferate but retain SOX2 expression, consisting with the notion that SOX2 maintains the quiescent state of SCs by positively regulating p27 kip1 expression 26 . Previously, we have shown that Gata3 is required for the establishment of the prosensory domain. Targeted deletion of Gata3 results in the ablation of p27 kip1 expression and the significant downregulation of SOX2 expression in the developing cochlea, suggesting that Gata3 should act upstream of Sox2 and p27 kip115 . In this study, we have observed abnormal proliferation and the downregulation of p27 kip1 expression in the neonatal Gata3-CKO SCs, specifically in IPCs, IBCs, and the lateral GER. However, loss of Gata3 function does not affect the expression of SOX2 in these cells. Our results indicate that while Gata3 is essential for SOX2 expression during the establishment of prosensory domain in the embryonic cochlea, it is dispensable for maintaining SOX2 expression in neonatal SCs. Moreover, the decoupling of the expression of SOX2 and p27 kip1 in Gata3-CKO SCs suggests that SOX2 might not act directly or not be sufficient by itself to regulate the expression of p27 kip1 . The reliance of p27 kip1 expression on GATA3 also suggests that SOX2 could function in parallel with GATA3 to maintain p27 kip1 expression in neonatal SCs.
Published studies have shown that SCs can spontaneously proliferate and form new HCs upon HC damage in the neonatal cochlea 21,23 . Downregulation of Notch signaling pathway and activation of Wnt signaling pathway also lead to the proliferation of SCs and the mitotic regeneration of HCs in the neonatal mouse cochlea 24,25,45 . In the present study, the observed SC proliferation in the neonatal cochleae of Gata3-CKO mice does not result in the de novo generation of ectopic HCs. The effect of Gata3-CKO mutation is consistent with those of Sox2null or p27 kip1 -null mutations that cause SC proliferation but not a cell fate change because other factors such as ATOH1 are required in HC generation 26 . Our observation of the increased number of proliferating SCs from the base to the apex is also consistent with previous studies 25,45 . The gradual increase in SC proliferation from the base to the apex might be due to that cells in the apical turn are less mature. Previous studies have shown that the apical turn of the cochlea is the last region to acquire HC or SC fate during embryonic development 38,46,47 . p27 kip1 , a member of the Cip/Kip family of cyclin-dependent kinase inhibitors, functions to prevent the activation of cyclin-CDK complexes and thus, controls cell cycle progression at G1 phase. In the cochlea, p27 kip1 ablation stimulates the proliferation of SCs [26][27][28] . In this study, we have shown that p27 kip1 expression is absent in the proliferating Gata3-CKO SCs, suggesting that Gata3 could be an upstream regulator of p27 kip1 to maintain the quiescent state of SCs. GATA3 ChIP-seq studies in other tissues have shown the presence of putative GBSs in the promoter region of p27 kip129-31 . Our analysis of neonatal cochlear ducts has also demonstrated that these putative GBS-containing regions in the p27 kip1 promoter are significantly enriched by anti-GATA3 antibody ChIP-PCR, arguing for the direct role of GATA3 in regulating p27 kip1 expression to maintain the quiescent state of cochlear SCs. Nevertheless, we cannot rule out the possibility that GATA3 may also regulate p27 kip1 in an indirect manner.
In addition, a recent study has shown that Gata3 is negatively regulated by p27 kip1 and that p27 kip1 deletion results in the upregulation of GATA3 expression in SCs of mature cochleae. Ectopic GATA3 expression in conjunction with ATOH1 converts more SCs into HCs than ectopic ATOH1 expression alone, consistent with the result that combining p27 kip1 deletion and ectopic ATOH1 expression results in an increased number of SCs converting to HCs in adult mice 48 . Collectively, these data and our results showing the dependence of p27 kip1 expression on GATA3 indicate a mutual negative feedback relationship between GATA3 and p27 kip1 as well as the essential role of GATA3 in HCs formation.
Due to the fact that Gata3 is mainly removed from SCs, RNA-seq analysis on entire cochlear ducts fails to capture all relevant genes. Nevertheless, our RNA-seq analysis has identified a total of 163 genes that are differentially expressed in Gata3-CKO cochlear ducts. Among them are Tectb, Cyp26b1, Slitrk6, Ano1, and Aqp4, which have been implicated in cochlear development and diseases. Tectb encodes β-tectorin, a glycoprotein that is localized to the tectorial membrane (TM). Previous studies showed that Tectb knockout mice exhibit abnormal hearing with decreased sensitivity and sharper frequency selectivity caused by changes in TM wave properties 49,50 . CYP26B1, a member of the cytochrome P450 superfamily, is an enzyme involved in the metabolism of retinoic acid. Previous studies have reported that Cyp26b1 is expressed in mesenchymal cells surrounding the otic epithelium at early embryonic stages. At later stages, Cyp26b1 is confined to supporting cells in the cochlear and vestibular epithelia 51,52 . SLITRK6 is a member of the SLITRK family that consists of neuronal transmembrane proteins controlling neurite outgrowth. Slitrk6 is strongly expressed in mouse cochlear SCs and Slitrk6 deficiency results in a reduction in cochlear innervation as well as the loss of SGNs 53 . ANO1 is a Ca 2+ -activated Clchannel. It has been reported that the periodic excitation of IHCs in the prehearing cochlea is initiated by the buildup of K + around IHCs due to ANO1 associated Clefflux from nearby GER cells. The spontaneous and periodic activities of GER cells are mediated by the ATP/ purinergic autoreceptor/ Ca 2+ /ANO1 pathway 54 . AQP4 is a water channel that is involved in endolymph volume homeostasis and K + homeostasis in the cochlea 55,56 . Taken together, these findings indicate that these genes may act as downstream genes of Gata3 in cochlear SCs and that the altered expression levels of these genes may contribute to Gata3 deficiency-induced sensorineural deafness in HDR syndrome.
Compared to a small reduction in p27 kip1 expression detected by RT-qPCR (p = 0.020, Fig. 5E), our RNA-seq results revealed a small but insignificant reduction in the overall expression level of p27 kip1 in Gata3-CKO mice compared with the control group (p = 0.358). In addition, no other factor involved in regulating cell cycle was identified by RNA-seq. p27 kip1 is broadly expressed in the cochlear duct and Gata3 deficiency leads to the downregulation of p27 kip1 expression in only three cell types-IPCs, IBCs, and the lateral GER (Fig. 5). The failure to identify p27 kip1 and other cell cycle-associated genes by RNA-seq could be due to an insufficient detection www.nature.com/scientificreports/ sensitivity of the RNA-seq approach on cochlear ducts. Future scRNA-seq approach could help confirm the cell type specific change in p27 kip1 expression in Gata3-CKO SCs. In summary, our study has revealed that Gata3 deficiency induces the proliferation of IPCs, IBCs, and the lateral GER but no de novo HC generation in the neonatal mouse cochlea. In addition, the expression of p27 kip1 is downregulated in Gata3-CKO mice compared with that in control mice. ChIP-PCR assays demonstrated that GATA3 directly binds to the promoter region of p27 kip1 . These data indicated that GATA3 regulates p27 kip1 expression to maintain a quiescent state of IPC, IBC, and the lateral GER. Furthermore, GATA3 is required for the maturation and maintenance of SCs by regulating genes essential for SC function.

Materials and methods
Animals and treatment. Sox2 CreERT2 (The Jackson Laboratory, Stock Number: 17593) and Gata3 loxP (previously generated in our lab) mouse strains are previously described 15,57 and were maintained in C57BL/6J background. For Cre activation, tamoxifen (Sigma-Aldrich) in corn oil at a dosage of 40 μg/g body weight was administered into control and Gata3-CKO mice via intraperitoneal injection at P1 and P2. To perform the EdU assay, mice were intraperitoneally injected with EdU (75 mg/kg body weight, Invitrogen) once a day from P1 to P6. All animal procedures used in this study were approved by the University Committee of Animal Resources (UCAR) at the University of Rochester and Institutional Animal Care and Use Committee (IACUC) at Augusta University. All experiments were carried out in compliance with the ARRIVE guidelines, and all methods were carried out in accordance with relevant guidelines and regulations.
For whole-mount immunostaining, whole inner ears were isolated under a dissecting microscope and were fixed in 4% paraformaldehyde at 4 °C overnight with a small hole in the apex. The inner ears were decalcified in 0.2 M EDTA in PBS for 1-5 days according to the developmental stages and cochleae were isolated under a dissecting microscope. After permeabilization and blocking in 0.1% PBST plus 0.03% saponin and 10% horse serum, cochleae were incubated with primary antibodies overnight at 4 °C. On the next day, cochleae were washed with PBS and were incubated with the fluorescently labeled secondary antibodies at 4 °C overnight.
Sections and whole-mounts were analyzed using a Zeiss LSM 510 META confocal microscope.
For in situ hybridization (ISH), cryosections were prepared as described above. The primers listed in Table S1 were used to amplify the last exons of genes of interest for subsequent RNA-probe labeling. Digoxigenin-labeled RNA probes were transcribed by T7 RNA polymerase. ISH was performed as previously described 58 .
For RNAscope in situ hybridization, RNAscope Multiplex Fluorescent Reagent Kit v2 (Advanced Cell Diagnostics (ACD)) was used according to the manufacturer's instructions.
Chromatin immunoprecipitation (ChIP). For ChIP analysis, cochlear ducts were dissected from wild type C57BL/6J mice at P2. ChIP was performed using the Magna ChIP A/G kit (Millipore) according to the manufacturer's instructions with minor modifications. Briefly, the cochlear ducts were crosslinked with 1% formaldehyde for 15 min, and the reaction was stopped by the addition of glycine. The samples were homogenized using a Dounce homogenizer. The nuclei were isolated using cell lysis buffer and resuspended in NEBuffer 3.1 (New England Biolabs Inc.). The samples were digested with 200 U of the restriction enzymes FokI and StyI (New England Biolabs Inc.) for 4 h at 37 °C and subsequently with 100 U of each enzyme for an additional 16 h at 37 °C. The nuclei were then incubated with an additional 200 U aliquot of each enzyme for 2 h at 37 °C 59 . Thereafter, the samples were spun down and resuspended in nuclear lysis buffer. A Covaris E220 sonicator (Covaris) was used to further shear and release chromatin (PIP: 105, duty factor: 2%, CPB: 200, treatment time: 8 min). The supernatant of the sonicated solution was used for immunoprecipitation with rabbit anti-GATA3 (ab199428, Abcam) via an established protocol. GATA3-binding site (GBS)-containing sequences were detected in the precipitated material by standard PCR and qPCR. The primers used are listed in Table S2. A coding sequence in exon 3 of p27 kip1 was used as a negative control (NC).

RNA-sequencing (RNA-seq).
Cochlear ducts at P4 were removed under a dissecting microscope. RNA extraction was carried out using Direct-zol RNA kit (Zymo Research) following the manufacturer's protocol. The RNA-seq library was constructed using a KAPA Stranded mRNA-seq Kit (KK8420). Sequencing was performed on Novaseq (Illumina). The strand marked with dUTP was not amplified, allowing strand-specific sequencing. RNA-seq was performed in three biological replicates. The FASTX-Toolkit (http:// hanno nlab. cshl. edu/ fastx_ toolk it/ index. html) was utilized for quality control of the RNA-seq data. Adaptors were first removed, and short www.nature.com/scientificreports/ reads were filtered out if they contained more than 10% ambiguous bases or more than 50% low-quality bases (Q < 5). Clean reads were mapped to the mouse genome (mm10) with TopHat (v2.1.0) 60 . The read counts for each gene were calculated with HTseq, and the differentially expressed genes were identified by using DESeq2 (cutoff fold-change ≥ 2 and P value ≤ 0.05) 61,62 . A heatmap was constructed using the heatmap.2 function of the R package gplots (v3.0.1.1). Pathway and process enrichment analyses were performed using Enrichr (http:// amp. pharm. mssm. edu/ Enric hr/).

Reverse transcription-quantitative polymerase chain reaction (RT-qPCR).
The cochlear ducts of 4 mice at P4 were isolated under a dissecting microscope. Total RNA was extracted with the RNeasy Mini Kit (Qiagen) in accordance with the manufacturer's protocol. An Iscript cDNA synthesis kit (Bio-Rad Laboratories) was used to generate cDNA libraries. qPCR was performed using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad Laboratories) on CFX96 Real-Time PCR system (Bio-Rad Laboratories) according to the manufacturer's protocol. The housekeeping gene β-actin was used as the control. The primers used in these assays are listed in Table S3.
Data analysis. Quantifications throughout the paper were performed blind to the genotype. All values are represented as the mean ± SEM and SPSS (v19.0) was used for data processing. Statistical analysis was performed using two-tailed Student's t test or ANOVA with post hoc test. P values of less than 0.05 were considered significant. For all experiments, n values represent biological replicates. For quantification of GER cells, GER region was defined between the inner side of IHC and the modiolar edge of the cochlear duct.

Data availability
Data available from the authors upon request. www.nature.com/scientificreports/