Loss of function of Ywhah in mice induces deafness and cochlear outer hair cells' degeneration

In vertebrates, 14-3-3 proteins form a family of seven highly conserved isoforms with chaperone activity, which bind phosphorylated substrates mostly involved in regulatory and checkpoint pathways. 14-3-3 proteins are the most abundant protein in the brain and are abundantly found in the cerebrospinal fluid in neurodegenerative diseases, suggesting a critical role in neuron physiology and death. Here we show that 14-3-3eta-deficient mice displayed auditory impairment accompanied by cochlear hair cells' degeneration. We show that 14-3-3eta is highly expressed in the outer and inner hair cells, spiral ganglion neurons of cochlea and retinal ganglion cells. Screening of YWHAH, the gene encoding the 14-3-3eta isoform, in non-syndromic and syndromic deafness, revealed seven non-synonymous variants never reported before. Among them, two were predicted to be damaging in families with syndromic deafness. In vitro, variants of YWHAH induce mild mitochondrial fragmentation and severe susceptibility to apoptosis, in agreement with a reduced capacity of mutated 14-3-3eta to bind the pro-apoptotic Bad protein. This study demonstrates that YWHAH variants can have a substantial effect on 14-3-3eta function and that 14-3-3eta could be a critical factor in the survival of outer hair cells.


INTRODUCTION
The 14-3-3 proteins were initially described as tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation proteins (YWHAx) required for the synthesis of neurotransmitters (dopamine, (nor-) adrenaline, serotonin) from catecholamine. 1 Subsequently, studies described the ability of 14-3-3 proteins to form homo-or hetero-dimeric complexes and numerous reports of new binding partners confirmed their chaperone activity. 2,3 This highly conserved family of small 28-33 kDa acidic dimeric proteins consists of seven distinct subunit isoforms (β, ε, γ, η, τ, σ and ζ) encoded by seven genes named YWHAx, (x being either B, E, G, H, Q, S or Z). 4 These proteins mostly bind serine/threonine phosphorylated ligands altering their sub-cellular localization, stability, phosphorylation state, activity or molecular interactions with other targets, thus controlling cell cycle and many signal transduction pathways. 2,3 A primary function of 14-3-3 proteins is the inhibition of apoptosis by retaining pro-apoptotic factors such as Bad or Bax in the cytoplasm. 5,6,7 14-3-3 proteins were originally discovered as abundant molecules in the brain, 8 and follow-up studies confirmed that the highest tissue concentration of 14-3-3 proteins is in the brain, 9 comprising about 1% of total proteins of the brain. In the peripheral nervous system, proteomics experiments showed that 14-3-3 proteins were expressed in the cochlea, 10,11,12 and among the seven 14-3-3 isoforms, 14-3-3eta encoded by the YWHAH gene has been found highly expressed in retinal ganglion cells (RGC). 13 The crucial involvement of 14-3-3 proteins in neuronal physiology led to investigate them in the pathophysiology of neurological diseases as well as considering the YWHAx genes as candidate genes in neurodegenerative conditions. 14,15,16,17,18 Subsequently, 14-3-3 proteins' involvement was confirmed in a number of neurological disorders, including Parkinson disease, amyotrophic lateral sclerosis, Alzheimer disease, epilepsy and Creutzfeldt-Jakob disease, 19 but whether it is involved in sensory organ dysfunction or degeneration remains unknown. Here we have investigated the function of 14-3-3eta in both auditory and visual systems, and we report that the loss of 14-3-3eta protein is associated with cochlear hair cells' degeneration.

Ywhah loss of function induces deafness in mice
To investigate the role of 14-3-3eta in optic and auditory systems, we obtained an embryonic stem cell line harboring a genetrap (GT) cassette in intron 1 of the Ywhah mouse gene and generated the corresponding transgenic mouse. Processing of the mRNA transcribed from this locus resulted in the fusion of the β-galactosidase/neomycin cassette to exon 1, generating a 14-3-3eta truncated protein (Supplementary Figure S1a). Ywhah expression level (Supplementary Figures S1b and c) and 14-3-3eta protein abundance (Supplementary Figure S1d) were decreased by 50% in 14-3-3eta GT/WT mice and were absent in 14-3-3eta GT/GT mice. Mice were morphologically normal and had a normal lifespan but 14-3-3eta GT/GT males were sterile.
We first assessed the visual function of those mice as 14-3-3eta is highly expressed in RGC. 13 Electroretinograms were recorded in animals from 2 to 16 months of age. Both A-and B-wave amplitudes tend to decrease from 12 months of age in 14-3-3eta GT/GT mice in comparison to 14-3-3eta WT/WT and 14-3-3eta GT/WT animals, but this did not reach significance (Supplementary Figure S2a). There was no change in latencies. Measurements of the visual evoked potentials that reflect the integrity of the visual pathway including RGC did not reveal significant difference between 14-3-3eta GT/WT , 14-3-3eta GT/GT mice and 14-3-3eta WT/WT littermate (Supplementary Figure S2b). Accordingly, counting of RGC by Brn3a labeling did not evidence difference between 14-3-3eta GT/GT and 14-3-3eta WT/WT mice (Supplementary Figure S2c).
In addition, some IHCs were missing in B2 in 14-3-3eta GT/GT (Figure 2i). We then examined the tissue expression of 14-3-3eta in the sensory organs. In the cochlea, 14-3-3eta was strongly expressed in the IHC and OHC and in the pillar cells of the organ of Corti (Figures 3a and b). The spiral ganglion neurons were also strongly labeled (Figures 3a and c), whereas no signal was detected in negative controls (Figures 3d-f). In the retina, 14-3-3eta was highly expressed in RGC and slightly in neurons of the inner nuclear layer but not in photoreceptors (Supplementary Figure S3). We then measured the expression level of all 14-3-3 isoform transcripts in adult mice by quantitative PCR. In the cochlea, 14-3-3eta had the highest expression, being at least twofold more abundant than any other 14-3-3 isoform ( Figure 4). In the retina, 14-3-3eta was also highly expressed, with a level similar to that of the theta and epsilon isoforms ( Figure 4). Altogether, 14-3-3 proteins were expressed at a higher level in the cochlea than in the retina, with the highest level for 14-3-3eta.

Ywhah loss of function induces apoptosis in mouse cochlea
To determine whether the inner ear cell loss was due to apoptosis, we performed TUNEL (terminal deoxunucleotidyl transferase biotin-dUTP nick end labeling) staining to examine cochleas from 6-month-old 14-3-3eta WT/WT , 14-3-3eta GT/WT and 14-3-3eta GT/GT mice ( Figure 5). Whereas no TUNEL-positive cells were detected in 14-3-3eta WT/WT cochleas, apoptosis was clearly seen in OHCs and IHCs of the basal part of cochleas from 14-3-3eta GT/WT and 14-3-3eta GT/GT ( Figure 5). These observations support the notion that apoptosis takes place in response to loss of 14-3-3eta and can be the mechanism responsible for cell death in mutant cochleas. These results provide evidence that Ywhah loss of function results in induction of apoptosis in the cochlea.
Identification of novel YWHAH non-synonymous variants in deaf population Knockout mice and tissue expression results suggested that 14-3-3eta could, when mutated or dysregulated, lead to auditory impairment in humans. We thus screened the YWHAH gene (Genbank accession number NM_003405.3) composed of two exons separated by an 8-kb intron 20 in 793 unrelated patients with non-syndromic or syndromic deafness. We found one frameshift, six missense and one intronic heterozygous SNP (single-nucleotide polymorphism), which were absent from the dbSNP, HGVD (Human Genetic Variation Database) databases, Exome Variant Server and 1000 genomes database ( Table 1). Two of them, c.172del25 and c.706G4A, were predicted to be damaging or deleterious. The c.172del25 (p.Ser58Leufs*43) was predicted to produce a truncated 14-3-3eta protein with the loss of 60% of protein sequence, including the substrate-binding domain 4 ( Figure 6a). The c.706G4A (p.Asp236Asn) changes an amino acid involved in target binding and highly conserved across species and among 14-3-3 isoforms 4 ( Figure 6b). The other variants were predicted to be either possibly damaging or benign (Table 1).
YWHAH non-synonymous variants increase apoptosis and abolish Bad interaction As c.172del25 and c.706G4A variants would produce 14-3-3eta proteins with a modified target-binding site, we studied skin fibroblasts carrying these variants. Considering the role of 14-3-3eta in cell death, we analyzed the susceptibility to apoptosis in patient fibroblasts. We found a marked increase in staurosporineinduced cell death (Figure 7a), which was more severe in the c.172del25 (Δ) mutated cells as compared with c.706G4A (Asp236Asn) fibroblasts and a higher fragmentation of the mitochondrial network in patient fibroblasts compared with controls ( Figure 7b).
To confirm these observations, we expressed the wild-type and mutated 14-3-3eta isoforms in HeLa cells to monitor their protective effect on staurosporine-induced apoptosis. Vectors expressing 14-3-3eta, 14-3-3eta Δ189 and 14-3-3eta D236N were transiently transfected in HeLa cells (Figure 7c). Twenty-four hours after transfection, staurosporine was added to HeLa cells. Wildtype 14-3-3eta provided strong inhibition of apoptosis (P = 0.0046), whereas the mutated forms failed to significantly protect from apoptosis (Figure 7d YWHAH altered the antiapoptotic function of 14-3-3eta. Overexpression of wild-type or mutated 14-3-3eta remained without effect on the mitochondrial network and on the mitochondrial membrane potential. Finally, to obtain further insight into the antiapoptotic function of mutated 14-3-3eta, we analyzed the interaction of 14-3-3eta with Bad, a pro-apoptotic member of the Bcl-2 family 21 by immunoprecipitation. 14-3-3 binds and retains phosphorylated Bad in the cytoplasm. When dissociated from 14-3-3, Bad translocates to mitochondria where it binds and inactivates the pro-survival proteins Bcl-2/Bcl-xL, leading to initiation of the apoptotic cell death process. 22 We found that the wild-type 14-3-3eta co-immunoprecipitated with Bad, whereas both variants abolished Bad binding (Figure 8), confirming that 14-3-3eta C-terminal domain was involved in Bad interaction and that consequently alterations in this domain facilitate Bad translocation from the cytoplasm to the mitochondrial outer membrane, thus promoting excessive apoptosis.

DISCUSSION
In this study, we showed that the absence of 14-3-3eta causes a hearing loss with a cochlear OHCs' degeneration in mice, only for high frequencies, that is, from 30 to 90 kHz. We found that 14-3-3  proteins are expressed in both cochlea and retina, at various levels, with the highest expression of 14-3-3eta in the cochlea. In humans, we identified two variants at the heterozygous state in YWHAH, never described before, and we demonstrated that they indeed abolished the protein function.
The auditory exploration of the knockout mouse revealed an early and non-progressing decrease of the ABRs, suggesting a congenital impairment of hearing in 14-3-3eta GT/WT and 14-3-3eta GT/GT mice. However, we have not observed cell loss in the cochlea areas corresponding to the frequencies tested, which represent the median and apical region of the cochlea. Little is known about the role of 14-3-3 proteins in the inner ear. We demonstrated that loss of hair cells in the basal part of 14-3-3eta mutant cochleas are related to apoptosis and highlight the role of 14-3-3eta in OHC survival. These results were consistent with the notion that apoptosis is a major cause of hearing loss in mammals. 23,24 Additionally, Bcl-2 family proteins such as Bax and Bad are critical for the maintenance of hearing function. 25 Further data suggest a critical role of this apoptotic pathway in IHCs' survival, as mutations in the pro-apoptotic protein SMAC/ DIABLO were reported in human dominant hearing loss DFNA64 26 and the pro-apoptotic Bak protein is involved in age-related hearing loss in mice. 27 In the cochlea, death of OHCs after an acoustic trauma has been linked to the activation of pro-apoptotic Bad. 28 Altogether, these studies suggest that Bad is a crucial determinant of cell fate in the cochlea. Therefore, the defects caused by 14-3-3eta dysfunction might result in Bad activation leading to or facilitating the degeneration of cells that specifically rely on high YWHAH expression.
Surprisingly, the absence of 14-3-3eta did not cause electrophysiological and histological alterations of the retina, whereas because of the abundant 14-3-3eta expression in the RGCs, we expected visual impairment. We noted that there was no significant increase in the expression levels of other 14-3-3 in 14-3-3eta-deficient mice (data not shown), therefore there is no clues for a compensatory upregulation of other 14-3-3 isoforms in the retina. Yet, it is still possible that the endogenous levels of other 14-3-3 isoforms were sufficient to compensate for the 14-3-3eta loss, especially in the retina where other 14-3-3 are predominant. In contrast, 14-3-3eta is the predominant isoform in the cochlea, therefore its deficiency may not be sufficiently compensated by other isoforms.
Considering the impact of the loss of 14-3-3eta on the survival of hair cells, we investigated whether this protein could lead to auditory diseases. YWHAH has been extensively screened in large cohorts of patients (42000 cases) and controls (41200 cases) because the YWHAH chromosomal location at 22q12.3 co-localizes with the susceptibility loci identified in schizophrenia, 29,30,31,32 bipolar disorders 33 and Parkinson disease. 34 However, non-synonymous variants were never identified. By screening YWHAH gene in non-syndromic and syndromic deafness, we reported seven nonsynonymous variations never previously found in this gene. We wanted to test whether these variants could impair the 14-3-3eta function. To define the biological function of 14-3-3eta, we have used a combination of two models: downregulation of 14-3-3eta with patient fibroblasts that act for a condition of haploinsufficiency and overexpression of 14-3-3eta mutants in wild-type HeLa cells. We showed that these variants confer an increased susceptibility to apoptosis in fibroblasts, which we were able to reproduce by expressing 14-3-3eta variants in a heterologous system. This deregulated cell death control was related to impair Bad binding to 14-3-3eta. This latter observation further correlates with the alteration of the mitochondrial network fusion process observed in fibroblasts, which is known to be controlled by proapoptotic and antiapoptotic members of the Bcl2 family. As 14-3-3 molecules can interact with a wide variety of cellular proteins, it is possible that genetic variants affect the protection against apoptosis.
This study shows the essential role of 14-3-3eta in OHC survival. In addition, we report for the first time YWHAH variants that can significantly alter the antiapoptotic function of 14-3-3eta by abolishing its interaction with Bad. Altogether, our results underline the fundamental role of 14-3-3eta in auditory physiology.

Examination of mouse electrophysiological audition
Animal surgery. All procedures were performed under general anesthesia achieved by an intra-peritoneal injection of a mixture of Rompun 2% (3 mg/kg) and Zoletyl 50 (40 mg/kg). For DPOAEs and for endocochlear potential recordings, both pinna were severed. Moreover, for endocochlear potential recordings, a retro auricular incision was performed and the bulla tympani was opened to expose the cochlea. Auditory brain stem recordings. The acoustical stimuli were produced by an arbitrary function generator (type 9100 R; Le Croy Corporation, Chesnut Ridge, NY, USA). They consisted in 3-ms tone pips with a 1-ms rise and fall time at a rate of 10/s. Tone pips were passed through a programmable attenuator and delivered to the animal by a JBL 075 loudspeaker (James B. Lansing Sound, Los Angeles, CA, USA) in a calibrated free field. ABRs were differentially derived with subcutaneous needle electrodes placed at the vertex and laterally below the stimulated auditory canal with a reference below the controlateral auditory canal. The signal was amplified (×20 000) (P 511 K, Grass Technologies, West Warwick, RI, USA), digitalized (50 kHz sampling rate, with a 12-bit dynamic range analog to digital converter), averaged 900 times and stored on a computer (Dell Dimention, Austin, TX, USA). Recordings were made at 2, 4, 6, 8, 10, 12, 16, 20, 26 and 32 kHz, and intensity was decreased from 100 to 0 dB sound pressure level (SPL)) (reference 2.10-5 Pa) by 5-dB steps. The threshold was determined as the lowest intensity able to elicit a visible ABR.
Distortion product oto-acoustic emission recordings. DPOAEs were recorded in the external auditory canal with an ER-10C S/N probe (Etymotic Research, Elk Grove Village, IL, USA) consisting of two emitters and one microphone. The two primary tones were generated, and the distortion was processed by the Cubdis system HID 40133DP (Mimosa Acoustics, Champaign, IL, USA). The probe was self-calibrated for the two stimulating tones before each recording. The two tones were presented simultaneously. F 2 was swept from 0.5 to 20 kHz by quarter of octave steps and the f 1 /f 2 ratio was maintained constant at 1.2. The intensities of f 1 and f 2 were set at 60 and 55 dB SPL, respectively. For each frequency, the cubic distortion product (2f 1 − f 2 ) and the neighboring noise magnitude were measured and expressed as a function of f 2 .
Endocochlear potential recordings. After opening the bula tympani, a small hole was made in the otic capsule over the stria vascularis of the basal turn by gently eroding the bone with a custom-designed blade. A glass micropipette filled with 3 M KCl was inserted in the scala media through the stria vascularis with a micromanipulator (Warner Instruments, CT, USA). This electrode, together with a reference needle electrode placed Table 1.   Figure S4a) were designed and optimized to ensure specific amplification of the target sequences.

TUNEL assay on cochlea
Six-month-old mice were euthanized by intracardiac perfusion with 4% PFA. Cochleas were micro-dissected in phosphate-buffered saline (PBS). Cochleas were immunostained for Myosine VIIa (Proteus Biosciences, Ramona, USA) and evaluated for TUNEL labeling using the In Situ Cell Death Detection Kit (Roche Diagnostics) according to the manufacturer's protocol. Cell nuclei were stained with DAPI. Cochleas were then mounted on a slide, coverslipped and imaged using a Zeiss LSM 510 Meta confocal microscope.

Immunoprecipitation
For immunoprecipitation experiments, mutant and wild-type14-3-3eta cDNAs were cloned into pcDNA/V5-His vector and transfected in Hela cells using Lipofectamine 2 000 (Invitrogen). Twenty-four hours after transfection, cells were lysed with 1% Triton lysis buffer (100 mM NaCl, 1% Triton, 0.02 M Tris, 0.005 EDTA) at 4°C for 1 h. Mouse anti-Bad antibody (BD Transduction Laboratories, Franklin Lakes, NJ, USA) was previously coupled to protein-G Sepharose and used for immunoprecipitation. The presence of 14-3-3eta in the Bad complex was revealed by immunoblot with the rabbit anti-V5 antibody (Invitrogen).

YWHAH screening
Genomic DNA was extracted from blood samples by standard methods. A hundred nanograms of DNA was amplified with Taq DNA polymerase (AmpliTaq Gold, Applied BioSystems, Foster City, CA, USA) helped with GC-RICH resolution solution (Roche Diagnostics) for exon 1, using intronic primers representing each exon (Supplementary Figure S4b). Standard cycling conditions were carried out with optimized annealing temperatures: 30 s at 95°C, 1 min at 65°C for exon 1 and 30 s at 60°C for exon 2, and 1 min at 72°C for the 35 cycles, followed by a final elongation at 72°C for 5 min. PCR products were cycled-sequenced using BigDye terminator system (Applied BioSystems), and analyzed on an ABI Prism 3130 capillary sequencer (Applied BioSystems).

Patients
The cohort for YWHAH screening consisted of 793 unrelated patients with syndromic or non-syndromic deafness seen in different Medical Centers (France, Belgium, United Kingdom, Marroco). These patients were referred for genetic deafness. The study was approved by all institutional review boards of the participating institutions, and written informed consent was obtained from all participants or their legal guardians. Patients with nonsyndromic deafness were negative for GJB2 and GJB6 mutations, which are most prevalent genes causing autosomal-recessive non-syndromic deafness.

Cell culture and transfection
HeLa cells were cultured in complete culture medium (Dulbecco's modified Eagle's medium, supplemented with 10% fetal calf serum and 100 units/ml penicillin/streptomycin, Invitrogen) and maintained at 37°C in a humidified 5% CO 2 atmosphere. Cells were seeded at a concentration of a 2 × 10 5 into six-well plates. Transient transfections were performed using Lipofectamine 2000 in Optimem (Invitrogen) according to the manufacturer's protocol. Twenty-four hours after transfection, staurosporine (Sigma, St. Louis, MO, USA; 1 μm for 3 h) was added and then apoptotic nuclei stained with Hoechst (Invitrogen) and counted in EGFP-expressing cells (n = 200) or by immunofluorescence using a cleaved caspase-3 antibody (1/10, Millipore).  Figure 8. Impact of YWHAH variants on Bad interaction. After transfection of empty vector pcDNA-V5, 14-3-3eta-V5, 14-3-3etaΔ189-V5 and 14-3-3D236N-V5 in HeLa cells, the production of the protein was verified by western blotting (Input, middle panel), and immunoprecipitation studies were performed with the Bad antibody. The presence of 14-3-3eta in the Bad complex was shown by immunoblot with the rabbit anti-V5 antibody (IP Bad, top panel). The antibody used for the immunoblot analysis is denoted at the right side. Cell lysates were immunoblotted with the Bad antibody as a control (Input, bottom panel).
Mitochondrial network was evaluated by immunostaining of mitochondria with ATP synthase antibodies (1/500, Millipore). Cells were stained with JC1 (Invitrogen) to analyze mitochondrial membrane potential. 38 Scanning electron microscopy The mice were decapitated under deep anesthesia (Pentobarbital 50 mg/kg), and their cochleae were removed from the temporal bone and fixed in 2.5% glutaraldehyde in 0.1 M-pH 7.3 PBS. The otic capsule was dissected out, and the stria vascularis, tectorial and Reissner's membranes removed. After rinsing in PBS (pH 7.3), the samples were dehydrated in a graded series of ethanol (30-100%), critical point dried in CO 2 , coated with gold palladium and observed using a Hitachi S4000 microscope (Krefeld, Germany).