The microRNA-183/96/182 Cluster is Essential for Stereociliary Bundle Formation and Function of Cochlear Sensory Hair Cells

The microRNA (miR)-183/96/182 cluster plays important roles in the development and functions of sensory organs, including the inner ear. Point-mutations in the seed sequence of miR-96 result in non-syndromic hearing loss in both mice and humans. However, the lack of a functionally null mutant has hampered the evaluation of the cluster’s physiological functions. Here we have characterized a loss-of-function mutant mouse model (miR-183CGT/GT), in which the miR-183/96/182 cluster gene is inactivated by a gene-trap (GT) construct. The homozygous mutant mice show profound congenital hearing loss with severe defects in cochlear hair cell (HC) maturation, alignment, hair bundle formation and the checkboard-like pattern of the cochlear sensory epithelia. The stereociliary bundles retain an immature appearance throughout the cochlea at postnatal day (P) 3 and degenerate soon after. The organ of Corti of mutant newborn mice has no functional mechanoelectrical transduction. Several predicted target genes of the miR-183/96/182 cluster that are known to play important roles in HC development and function, including Clic5, Rdx, Ezr, Rac1, Myo1c, Pvrl3 and Sox2, are upregulated in the cochlea. These results suggest that the miR-183/96/182 cluster is essential for stereociliary bundle formation, morphogenesis and function of the cochlear HCs.

SCientifiC RepoRts | (2018) 8:18022 | DOI: 10.1038/s41598-018-36894-z lead to progressive non-syndromic hearing loss in mice and humans [21][22][23][24] . In spite of the high sequence homology between miR-96 and miR-182/miR-183, the latter two are unable to compensate for the functional defects caused by the mutations in the seed sequences of miR-96. The phenotypes caused by these point mutations of miR-96 are considered a result of both loss-and gain-of-function effects [21][22][23][24] , because they not only cause loss of regulation of normal target genes of wild-type miR-96, but also create new downstream target genes of the mutant miR-96. Further examination has shown that one of the point mutations causes arrest of HC development and maturation 23 . These data suggest that miR-96 plays a predominant role among members of the miR-183/96/182 cluster in inner ear function. It is uncertain whether miR-182 and miR-183 also play a significant role in mammalian inner ear development and function.
To understand the physiological functions of the miR-183/96/182 cluster, we generated a knockout (KO) mouse model using a gene trap (GT) embryonic stem cell clone, miR-183C GT/GT , in which the miR-183/96/182 cluster gene was inactivated by a GT construct 17 . Previously, we showed that inactivation of the cluster results in syndromic retinal dystrophy with multisensory defects including a circling phenotype, typical of vestibular defect 17 . This result suggests that loss of function of the miR-183/96/182 cluster leads to significant functional defects of the inner ear. Here, we characterize the impact of inactivation of the miR-183/96/182 cluster on inner-ear development and hearing functions. Our results demonstrate that loss of function of the miR-183/96/182 cluster causes severe defects in HC differentiation, morphogenesis and function, leading to profound congenital syndromic deafness in mice.

Results
miR-183/96/182 cluster KO mice have severe balance defects and profound deafness. Previously, we showed that the miR-183/96/182 KO mice have a circling behavior as early as the first week after birth 17 . To further test the defects in balancing, we performed a swimming test on young adult (3 weeks old) KO mice, and their wild-type (WT) littermates. The result revealed that, while the WT control mice swam well, kept calm and never lost balance during the test, the KO mice lost their balance and were unable to keep their heads above the water within seconds after being put in the water bath, confirming a vestibular defect (Fig. 1a).
To test whether inactivation of the cluster affects hearing, we performed auditory-brain stem response (ABR) test in 3-week-old mice. The results showed that the KO mice had no response to acoustic stimuli, whereas the heterozygotes and WT controls similarly showed normal hearing thresholds (Fig. 1b). These data suggest that inactivation of the miR-183/96/182 cluster in the cochlea resulted in profound hearing loss.
The miR-183/96/182 cluster KO mice show severe defects in HC morphogenesis. At postnatal day (P) 3, the organ of Corti, and in particular the HCs, of the KO mice showed several abnormalities compared with WT controls. Phalloidin staining for actin filaments revealed the normal checkboard-like pattern of the HCs and supporting cells (SCs) in the WT reticular lamina, whilst tubulin-labelled ciliary extrusions were identifiable on both HCs, where they were adjacent to the bundle (known as the kinocilium), and SCs (Fig. 2a,c,e). However, in the KO mice, hair cell apices were abnormal, appearing circular rather than heart-shaped; and the hair bundle was malformed. Most HCs of the KO mice lack kinocilia, although the ciliary protrusion was retained on the SCs (Fig. 2b,d,f). These abnormalities were detected along the entire length of the cochlea, as indicated by comparable images from apical, middle and basal regions (Fig. 2b,d,f).
These abnormalities were further confirmed by electron microscopy (EM) (Figs 3 and 4). In the KO mice, the rows of HCs and SCs were disorganized to a small degree (Fig. 3b,c), with the apical region ( Fig. 3b) more affected; the circular apices of the HCs were protruded (Fig. 3b,c compared to Fig. 3a; Fig. 4b compared to Fig. 4a). The hair bundles consisted of thin microvilli-like stereocilia of approximately equal length, reminiscent of immature stereociliary bundles from earlier embryonic stages (Figs 3e and 4d), in contrast to the control, where the staircase pattern of stereocilia was formed, the stereocilia thicker than adjacent microvilli on SCs (Figs 3d and 4c). The peripheral, stereocilia-free zone of the apical surface of HCs, which is visible in the WT (Fig. 3d), was substantially narrower, with the hair bundle covering virtually the entire apical surface in the KO mice (Fig. 3e). These abnormalities were evident in both apical and basal regions (Fig. 3b,c respectively).
At higher magnification, the stereocilia could be seen to be linked by fine filaments in the KO mice (Fig. 3f), but the lack of staircase organization meant that tip links could not be identified. The kinocilium basal body was observed by transmission electron microscopy (TEM), even where a complete kinocilium was absent (Fig. 4d).
The TEM observations also showed that whereas the cytoplasm of the HC bodies was apparently lighter than in the SCs in WT (Fig. 4a), the density of HC cytoplasm was comparable to that of SCs in the KO mice. It was difficult to distinguish the cytoplasm of HCs from the SCs (Fig. 4b). The HC nuclei of the KO mice were distorted (Fig. 4b), when compared with the WT control (Fig. 4a). The SCs between the HCs also showed evidence of distortion in the KO cochlea (Fig. 4a,b). Furthermore, the apical filamentous cuticular plate, clearly visible in the WT (Fig. 4a,c), was not detectable in the KO (Fig. 4b,d).
By P18, scanning electron microscopy (SEM) showed that the HC bundles were completely lost, and HC apices could not be detected in the KO mice (Fig. 3h). The epithelium was composed of non-specified cell types, perhaps expanded SCs (Fig. 3h).
Additional phalloidin staining in P6 cochleae showed that, although most HCs in the KO mice still existed, they appeared, qualitatively, to be more disorganized and have some cell loss (Supplemental Fig. 1), suggesting that HC degeneration starts between P3 and P6. Sox2 plays an essential role in the formation of the prosensory domain and HC differentiation during inner ear development via transient expression in HCs before P2 but retained expression in SCs subsequently [41][42][43][44][45] . To examine the cellular localization and protein expression level of Sox2, we performed co-immunofluorescence  of Sox2 and a HC-specific marker, Myo7a, in P1 cochlear sensory epithelia. Similar results were observed in the cochleae of 3 KO and 2 WT mice. Representative qualitative images are illustrated in Fig. 7. Our result showed that Sox2 staining in the nuclei of HCs appeared qualitatively more intense in the KO (Fig. 7f) vs. WT mice, especially in IHC (Fig. 7e), but there is no apparent qualitative difference in the SCs. This is consistent with increased expression of Sox2 in the cochlear epithelia of KO mice at mRNA level by qRT-PCR analysis (Fig. 6g) and suggests that the miR-183/96/182 cluster targets Sox2 in HCs in vivo. The staining of Myo7a appeared qualitatively much weaker in KO HCs, compared to WT controls, indicating impaired HC differentiation or maturation in KO mice. Intriguingly, in contrast to WT controls, the SCs surrounding the HCs in the greater and lesser epithelial ridges (GER and LER) appeared to be positive to Myo7a staining (arrowheads in Fig. 7b,d).

Discussion
miRNAs are quantitative regulators of gene expression. One miRNA often targets multiple protein-coding genes in the same pathway or functional network, and imposes modest regulation on each of them. This contributes to the maintenance of the homeostasis of the pathway/network under physiological conditions. Defects in a miRNA can thus result in simultaneous dysregulation of multiple genes, leading to significant functional consequences when the composite impact passes a threshold 17,46-50 . Here we provide evidence that the miR-183/96/182 cluster (ii) the cuticular plate is not formed; the HC apex is protruded; and the stereociliary bundle fails to develop to maturity; (iii) there is loss of MET function; and (iv) by P18, the HCs have completely degenerated; the KO mice have a profound hearing loss. These data confirm that the miR-183/96/182 cluster plays important roles in the differentiation, morphogenesis and function of inner ear HCs, and extend previous findings of Weston et al. 7,51 and Fan et al. 52 by providing more structural and expression data relating to the effects of loss of the cluster specifically in the inner ear.
A point mutation in the seed sequence of miR-96 has been reported to cause defects in the maturation of stereociliary bundles in the Dmdo/Dmdo mice 22 . In these mice, the stereocilia form a partially differentiated As we were preparing this manuscript, Fan et al. reported an independent conventional knockout mouse of the miR-183/96/182 cluster by homologous recombination 52 . Similar to what we described here in the miR-183C GT/GT KO mice both temporally and morphologically, the conventional KO mice also showed severe defects in the stereociliary bundle development and HC maturation, misalignment of HCs and profound hearing loss 52 . These striking similarities between our miR-183C GT/GT KO and the independently-created conventional miR-183/96/182 cluster KO mice further strengthen the conclusion that miR-183/96/182 cluster plays an essential role in HC development and function. This also confirms that the miR-183C GT/GT model is a functional KO of the miR-183/96/182 cluster, arguing against the suggestion that the miR-183C GT allele could be hypomorphic 52 . This is further supported by our initial characterization of the expression of miR-183/96/182 cluster in the sensory organs of the miR-183C GT/GT mice 17 . We demonstrated that, resembling the complete inactivation of this cluster in the retina, expression of miR-183, −96 and −182 in WT cochlea was at least 30-, 25-and 21-fold higher than in the cochlea of miR-183C GT/GT mice by qRT-PCR analysis 17 .
Coupled with the morphological studies, our target prediction and gene expression analysis suggest that the miR-183/96/182 cluster imposes a global regulation on the differentiation and function of the organ of Corti through simultaneous regulation on multiple genes involved in various aspects of the development of HCs. Hair cells have complex cytoskeletal networks, especially in the apical part of the cell. The ezrin-radixin-moesin (ERM) family plays important roles in the morphogenesis of the actin cytoskeleton in various cell types [53][54][55] , including HCs 31,56 . In HCs, they interact with Clic5 to form a complex at the ankle of stereocilia that may stabilize the linkage between plasma membrane and actin cytoskeleton 30   Rac1 is a member of the Rho family small GTPases, and has been shown to regulate planar cell polarity (PCP) and morphogenesis of the stereociliary bundle of HCs 35 . Deletion of Rac1 in the otic epithelium resulted in a reduced number of, and severely shortened HCs, with defects in PCP and morphogenesis of the bundle 35 . Our data suggest that loss of the miR-183/96/182 cluster results in dysregulation of Rac1 in P1 cochlear sensory epithelia, which may contribute to the phenotypes in the stereociliary development in the KO mice.
Pvrl1 and 3, also known as Nectin-1 and −3, are immunoglobin-like adhesion membrane molecules. Pvrl1 is expressed in HCs, while Pvrl3 is expressed in SCs. This mutually exclusive expression pattern is involved in the A major aspect of HC function is mechanoelectrical transduction. Our data showed that, in the miR-183/96/182 KO mice, MET no longer functions. The absence of a mature staircase of ranked stereocilia means that the tip-link complex that underlies transduction 57 may not be formed; and indeed, specific tip links have not been identifiable here. Myo1c is one of several unconventional myosins that have been implicated in the slow form of adaptation in HC MET by regulating tip-link tension [58][59][60][61][62][63][64] . Myo1c upregulation in auditory epithelia of the miR-183/96/182 cluster KO mice suggests a possible role of these miRNAs in regulating tip-link function and adaptation.
Sox2 has a dual role in the specification of sensory competence and HC differentiation 43,44 . Its early expression is required to formation of the prosensory domain and HCs [41][42][43] . However, its subsequent downregulation in HCs must occur to allow HCs to become fully differentiated 43,44 . In mice, after P2, its expression is undetectable in HCs but retained in SCs 41,45 . Increased expression of Sox2 in the HCs of P1 KO mice suggests that loss of miR-183/96/182 cluster contributes a delayed downregulation of Sox2 in HCs, and consequently, may affect the functional differentiation of HCs.
We anticipate that many other genes are also regulated by the miR-183/96/182 cluster and contribute to the overall phenotype in HC development and function. A genome wide analysis of gene expression changes in HCs of this mutant model would provide further information and deeper understanding of the roles of miR-183/96/182 cluster in HC development and functions. Phenocopying the mutant phenotype by overexpressing some of these target genes would help confirm their roles in mediating the functions of the miR-183/96/182 cluster in HCs.
The observation of apparent ectopic expression of Myo7a in cells in the GER and LER surrounding the HCs is intriguing. Increasing evidence has shown that a subset of SCs in these regions may act as cochlear progenitor cells with a capacity of regeneration and differentiation into HCs in neonatal mice, especially after damage [65][66][67][68] . In our P1 KO mice, the HCs are morphologically and functionally defective. It is reasonable to hypothesize that, through unknown mechanisms, defective HCs may have signaled and induced progenitor cells to generate more HCs; the ectopic expression of Myo7a has reflected this attempt. However, extensive follow-up studies will be needed to test this hypothesis.
In conclusion, we have demonstrated major effects of the miR-183/96/182 cluster in the development of the organ of Corti and also HCs. Further characterization of HC development and function in mutant mice with single or double knockout of members of the cluster 69,70 is therefore warranted.

Materials and Methods
Mice. Animal care and husbandry were handled in accordance with the National Institute of Health (NIH) guidelines. All protocols were approved by the Institutional Animal Care and Use Committee, Wayne State University. The miR-183/96/182 cluster KO mice, miR-183C GT/GT , are on a 129S2/BL6-mixed background and were originally derived from a gene-trap embryonic stem cell (ESC) clone as described previously 17,71 . Neonatal (P1~6) and 3 weeks old mice KO mice, their heterozygous (miR-183C GT/+ ) and WT littermates (miR-183C +/+ ) were used in this study. At least three mice per genotype were used in each experiment. The age and number of mice used for each experiment are further specified in the method description below and in the text and figure legends.
Swimming test. Young adult (3 weeks old) KO mice and their WT littermates were gently placed in a tank filled with warm tap water and the swimming behavior recorded by a camera of iPhone8 (Apple). They were rescued from the water within 15 seconds (s) after they lost their balance or were unable to keep their heads above the water. The duration from the time when the mice were put in water to when they lost their balance was recorded. All WT mice kept calm and never lost their balance even after 1 minute in water; accordingly, the duration for all WT mice was arbitrarily recorded as 1 minute for quantification purpose. ABR recording. ABR recording was performed as previously described 72 . Briefly, 3 weeks old mice were anesthetized with ketamine (100 mg/kg) and xylazine (20 mg/kg) and placed in a soundproof chamber (Tracoustical, Inc., Austin, Texas) during testing. Their body temperature was maintained by placing them on a homeothermic heating pad. ABR testing was carried out using a TDT system (Tucker-Davis Technologies, Alachua, Florida). Platinum subdermal needle electrodes (MFI Medical Equipment, Inc., San Diego, CA) were inserted at the vertex of the head, and ventrolateral to both ears. The mice were presented with 4-28KHz pure tone stimuli at an intensity starting at 100 decibel sound pressure level (dB SPL) and decrementing in 10-dB steps; this sequence was repeated in 5-dB steps until the lowest intensity that evoked a reproducible ABR waveform (peaks I-IV) was detected.
Immunofluorescence. The inner ears of P1 pups were dissected out from the temporal bone. After removal of the stapes, a small hole was made to open the apex; the cochlea was fixed by perfusion through the oval window and the hole at the apex with 4% paraformaldehyde (PFA) in 1X PBS solutions at 4 °C, and then left in the fixative overnight. After 3 rinses with 1XPBS, the cochlear sensory epithelium was dissected out. For immunostaining of kinocilia, the tissue was soaked in blocking solution (5% normal donkey serum and 0.1% Triton X-100 in 1X PBS) at room temperature for one hour (h), then incubated with anti-acetylated γ-tubulin antibody (Sigma. 1:250) in blocking solution at 4 °C overnight. After three 15-minute (min) washes with PBS containing 0.1% Triton X-100, a Alexa Fluor-568-conjugated secondary antibody (Invitrogen) was applied together with Alexa Fluor-488 Conjugated phalloidin (Invitrogen), which stains actin-rich structure, stereocilia, for 1 h at room temperature. After 3 washes with 1X PBS, the tissue was mounted on slide and images were taken with a confocal microscope (Leica TCS SP8).
For immunofluorescence of cryosections, inner ears of P1 mice (3 KO and 2 WT) were dissected out and fixed with 4% PFA in 1X PBS by perfusion through the oval window and a hole opened at the apex and incubation for 1 h at room temperature on a shaker. After washing with 1X PBS for 15 mins, three times, the inner ears were incubated consecutively in 10%, 15%, 20%, 25% and 30% sucrose in 1X PBS for 30 mins at each concentration at room temperature and then kept in 30% sucrose in 1X PBS at 4 °C overnight. Then the inner ears were transferred to OCT and kept at 4 °C, overnight. After degassed with vacuum, the inner ear was frozen and cryosectioned at 12 μm thickness. After blocking, the cryosections were stained with anti-Myo7A (Proteus Biosciences Inc.. 1:500) and anti-Sox2 (Santa Cruz Biotechnology. 1:250) at room temperature for 2 h. After washing with 1XPBS, the sections were incubated with secondary antibodies (donkey anti-goat Alexa Fluor 546 and donkey anti-rabbit Alexa Fluor 488 secondary antibodies. Invitrogen. 1:1000) for 1 h at room temperature. The images were taken with a confocal microscope (Leica TCS SP8).
Electron microscopy. Cochleae of P3 and P18 mice were excised from the temporal bone and fixed by perfusion as described above using 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer containing 2 mM calcium chloride and incubation for 2 h. They were stored in 1/10th fixative until further processing. For postfixation, the samples were washed in sodium cacodylate buffer and then immersed in 1% osmium tetroxide in sodium cacodylate for 1 h.
For SEM, cochleae were dissected and then prepared using the OTOTO technique 73 . Briefly, segments were incubated in alternating solutions of osmium tetroxide (3X, 2 h) and saturated thiocarbohydrazide (2X, 20 mins) with six washes in water between each change. They were subsequently dehydrated through an ethanol series and critical point dried using a Polaron drier, mounted on platinum stubs using adhesive carbon pads, and examined in a Hitachi S4500 field emission SEM at 5 kV.
For TEM, samples were fixed and dehydrated as for SEM, but then embedded in Spurr's resin 73 , sectioned at 70-100 nm and the sections collected on copper grids. Grids were stained in uranyl acetate and lead citrate and Cochlear organ culture and FM1-43 labeling. The function of the MET channels of mouse cochlea was assessed by FM1-43 labeling as described previously 29 . Briefly, cochlear sensory epithelia of P1 mice were carefully dissected out in cold Hepes-buffered Hanks' balanced salt solution (HBHBSS. Sigma) and cultured in DMEM/ F12 with 7% Fetal calf serum (FCS) (Fisher Scientific) and 5 mg/ml ampicillin (Sigma) on Cell-Tak-coated glass coverslips in Mat-Tk dishes (MatTek Corporation) at 37 °C. After overnight culture, the cochlea was rinsed briefly in HBHBSS and then dipped in HBHBSS containing 3 µM FM1-43 (ThermoFisher Scientific) for 10 seconds (s) and rinsed 3 times in fresh HBHBSS solution. The coverslip was immediately mounted on slide and imaged with fluorescent light microscope (Leica) equipped with epifluorescence optics and FITC filters (488 nm excitation, 520 nm emission).
RNA isolation and qRT-PCR. Cochlear epithelia from P1 mice were dissected in HBHBSS, and quickly transferred to RNAlater (Ambion/ThermoFisher Scientific). Total RNA was prepared using RNeasy micro kit (Qiagen) following manufacturer's instructions. On-column DNase I treatment was performed to avoid traces of DNA contamination. qRT-PCR assays of protein-coding genes were performed using QuantiFast SYBR Green RT-PCR kit and QuantiTect primers (Qiagen) with 18 s rRNA as an endogenous loading control as we described previously 17,[74][75][76] . The relative expression level of a gene of interest was normalized to 18 s rRNA. Briefly, the cycle threshold (Ct) value of the gene of interest (Ct goi ) was normalized to the Ct value of 18 s rRNA (Ct 18s rRNA ) and was calculated as DCt = Ct goi -Ct 18s rRNA . Subsequently, the relative expression level of the gene of interest was calculated as 2 −DCt .
Statistical analysis. All data are shown as mean ± SEM. Two-tailed student t test was performed in the quantitative comparison in the swimming test. Mann-Whitney test (one-tailed) was employed to test the statistical significance between KO mice and WT controls in the qRT-PCR analysis using the GraphPad Prism 6 software (GraphPad Software, Inc., La Jolla, CA). Since this is a simple comparison between KO and WT controls, no multiple test correction was employed in the p-value calculation.