mTORC2 regulates auditory hair cell structure and function

Summary mTOR broadly controls cell growth, but little is known about the role of mTOR complex 2 (mTORC2) in the inner ear. To investigate the role of mTORC2 in sensory hair cells (HCs), we generated HC-specific Rictor knockout (HC-RicKO) mice. HC-RicKO mice exhibited early-onset, progressive, and profound hearing loss. Increased DPOAE thresholds indicated outer HC dysfunction. HCs are lost, but this occurs after hearing loss. Ultrastructural analysis revealed stunted and absent stereocilia in outer HCs. In inner HCs, the number of synapses was significantly decreased and the remaining synapses displayed a disrupted actin cytoskeleton and disorganized Ca2+ channels. Thus, the mTORC2 signaling pathway plays an important role in regulating auditory HC structure and function via regulation of the actin cytoskeleton. These results provide molecular insights on a central regulator of cochlear HCs and thus hearing.


INTRODUCTION
Sensorineural hearing loss affects a significant and increasing proportion of the population.It is the third most important cause of years lived with disability 1 and the World Health Organization (WHO) estimates that around 5.5% of the global population suffers from disabling hearing loss that requires rehabilitation. 2Due to an increase in life expectancy and other factors such as exposure to environmental noise, this number is expected to increase to over 7% by 2050. 2 However, despite the significant societal burden of hearing loss, there are no adequate treatment options for hearing loss other than prosthetic devices.Most hearing loss, both genetic and non-genetic, has an underlying defect in the peripheral hearing organ, the cochlea. 3,4To date, 124 non-syndromic hearing loss genes (with hearing loss as the only symptom) 5 and around 400 genes for syndromic hearing loss 3 have been identified.It is expected that an additional 350-400 genes are required for hearing. 3The discovery of hearing loss genes contributed significantly to our understanding of the molecular composition and function of the hearing organ, especially the sensory cells, which are few in number and thus biochemically inaccessible.Less is known about signaling pathways that regulate sensory cell function in the inner ear.
Sound is perceived by sensory cells, called hair cells (HCs), in the cochlea of the inner ear.HCs in the vestibule of the inner ear mediate balance.In the mammalian cochlea, there are two types of HCs.Inner hair cells (IHCs), present in one row along the cochlear spiral, are the actual sensory cells.They transmit all the auditory information to the brain via synaptic contacts on afferent fibers of the auditory nerve.Outer hair cells (OHCs), present in three rows, are electromotile to increase sensitivity and frequency selectivity.Depending on their place along the cochlear spiral, HCs respond to different frequencies.High frequencies are perceived at the base of the cochlea whereas lower frequencies at the apex.HCs have microvilli-like protrusions, a so-called hair bundle, on their apical side.These protrusions, also known as stereocilia, are the sound and motion sensors.Deflection of the stereocilia, induced either by sound or head motion, opens mechanoelectrical transduction (MET) channels.This depolarizes HCs leading to electromotility or synaptic transmission onto the vestibulocochlear nerve which relays the information to the central nervous system.
HC loss often correlates with and is thus considered the major cause of hearing loss.However, noise exposure or aging can cause loss of synapses without loss of HCs, a phenomenon termed cochlear synaptopathy. 6,7Loss of HCs and synapses is permanent; adult cochlear HCs do not regenerate in mammals.In addition, there are no specific molecular or cellular therapies for hearing loss.Today, sensorineural hearing loss can be addressed only with hearing aids or cochlear implants.Detailed molecular understanding of HC function and dysfunction is needed to develop better preventive and therapeutic strategies.
Mammalian/mechanistic Target of Rapamycin (mTOR) is a serine/threonine kinase that regulates many cellular processes as part of two structurally and functionally distinct complexes, mTOR complex 1 (mTORC1) and mTORC2.Due to the availability of the specific inhibitor rapamycin, the function of mTORC1 has been investigated extensively.It is a master regulator of cell growth, promoting anabolism and inhibiting catabolic processes such as autophagy.Increasing evidence suggests a damaging role of mTORC1 overactivation in the cochlea, whereas mTORC1 inhibition in the neurosensory epithelium promotes HC survival and hearing protection 8 (and reviewed in 9 ).In contrast

HC-RicKO mice show early-onset, progressive, and profound hearing loss but normal vestibular function
To monitor auditory function, we performed auditory brainstem response (ABR) measurements.HC-RicKO mice showed elevated hearing thresholds already at age 2 weeks, soon after hearing onset.This was strongly progressive over time, with HC-RicKO mice showing an increased hearing threshold of up to 60 dB sound pressure level (SPL) at age 4 weeks, compared with littermate controls.At age 8 weeks, ABR thresholds of HC-RicKO mice were elevated up to 69 dB SPL.At age 12 weeks, HC-RicKO mice were almost completely deaf with no measurable responses for most frequencies and a mean hearing threshold of 89.7 G 6.7 dB SPL at 8 kHz (Figure 2A).Distortion-product otoacoustic emissions (DPOAE), which are a measure of OHC function, showed trends similar to ABRs with an increase in hearing thresholds over time in HC-RicKO mice (Figure 2B).
We then analyzed suprathreshold ABR wave amplitudes and latencies at 90 dB SPL, where responses are mostly independent of OHC amplification and therefore represent mainly IHC function. 33We used an 8 kHz stimulus for the analysis, where we observed measurable responses until 12 weeks of age (Figure 2A).The ABR wave I reflects electrical activity of the auditory nerve, which consists mostly of afferent nerve fibers of spiral ganglion neurons which have synaptic contacts to IHCs.ABR wave I amplitudes were significantly reduced (Figure 2C) and ABR wave I latencies were significantly increased at all time points measured in HC-RicKO mice (Figure 2D).Reduced amplitudes and increased latencies were also propagated to wave IV, which represents electrical activity in the inferior colliculus of the brainstem (Figures 2E and 2F).Most importantly, interwave I-IV latency was unchanged in HC-RicKO mice, suggesting that the auditory signal travels normally along the brainstem and that the cause of the hearing loss is in the peripheral hearing organ (Figure 2G).
While there are known sex differences in hearing with a protective role of estrogen signaling, 34 elevations in ABR and DPOAE hearing thresholds were equally high in male and female HC-RicKO mice (Figures S3A and S3B).These results suggest that Rictor KO had strong effects on hearing independent of sex.Myo15-Cre +/À mice had normal ABR and DPOAE hearing thresholds until 23 weeks of age.After age 36 weeks, they had elevated hearing thresholds as expected for aging, C57BL/6J mice (Figures S4A and S4B).Importantly, at age 12 weeks, ABR and DPOAE hearing thresholds in Myo15-Cre +/À and Rictor fl/fl mice were similar, excluding Cre toxicity as the cause of hearing loss (Figures S4C and S4D).
Interestingly, vestibular function was normal and indistinguishable from control mice in HC-RicKO mice at age 11-12 weeks, when HC-RicKO mice were already deaf (Table S1 and Figure S5).In aging mice, vestibular dysfunction occurs later than auditory dysfunction. 35Nevertheless, even at later timepoints of 28-30, 39-41, and 49-51 weeks of age, vestibular function in HC-RicKO mice was still indistinguishable from control mice (Figure S5).Therefore, Rictor deletion specifically in HCs leads to early-onset, progressive, and profound hearing loss but no observable vestibular dysfunction.

Hearing loss precedes hair cell loss in HC-RicKO mice
HC loss is a frequent hallmark of hearing loss and often considered the main cause of sensorineural hearing loss.Given the profound hearing loss in HC-RicKO mice, we investigated HC loss as the potential underlying cause.Unexpectedly, there was no IHC or OHC loss at age 2 weeks when HC-RicKO mice already show elevated hearing thresholds (Figures 3A and S6A).Notably, at age 4 weeks when there was already profound hearing loss in HC-RicKO mice (Figure 2A), there was no IHC and almost no OHC loss (Figures 3A and S6A).At 8 and 12 weeks, there was increasing IHC and OHC loss at the cochlear base in HC-RicKO mice, and also OHC loss at age 12 weeks in control mice.However, when consulting the place-frequency map, this region mainly included frequencies over 32 kHz and was outside our tested frequency range for ABR and DPOAE.In the tested frequency range (4-32 kHz), there was minor HC loss which was insufficient to explain the profound hearing loss observed in HC-RicKO mice at ages 8 and 12 weeks.For instance, a total (100%) OHC loss cannot account for more than 40-60 dB SPL ABR hearing threshold elevation, as shown by either OHC loss 36 or OHC dysfunction. 37Almost all IHCs and most of the OHCs are present in the medial cochlear turn of HC-RicKO mice at ages 8 (Figure 3B) and 12 weeks (Figure S6A).Therefore, HC loss occurs after hearing loss in HC-RicKO mice and is not the underlying cause for the profound hearing defect.
Apart from HC counts, we examined whether Rictor expression in HCs affected total length of the neurosensory epithelium.Cochlear length along the lateral border of IHCs did not differ between control and HC-RicKO mice at all timepoints examined (Figure S6B).

HC-RicKO mice show abnormal stereocilia but MET channel function is present
Given that HC loss is not the underlying cause of hearing loss, we investigated further cochlear structures relevant for hearing.Since HC-RicKO mice suffer from both OHC (reduced DPOAEs) and IHC (ABR hearing threshold elevation >60 dB SPL) dysfunction, we focused on stereocilia, which are present in both cell types.Stereocilia are the sound sensors containing the MET channels necessary for hearing.In line with no major (B) PCR using genomic DNA from dissected organ of Corti (OC), spiral ganglion (SG) or vestibular organ (vest; macular and cristae ampullaris organs pooled) tissue, separate for left (L) and right (R) inner ears.Genomic DNA from outer ear skin of both genotypes was used as negative control of Cre-mediated recombination.Genomic DNA from heart samples of tamoxifen-induced cardiomyocyte-specific Rictor KO mice was used as positive control (pos ctrl). 85No template control (NTC) contains all PCR reaction components including primers but no DNA.Successful Cre-mediated recombination of the Rictor allele was only found in OCs and vestibular organs of HC-RicKO mice.A5).Interestingly, OHCs of HC-RicKO mice showed stereocilia of different heights in the same row and some missing stereocilia (Figures 4A10 and S7A7-A10).Notably, stereocilia from the middle and small row were significantly shortened in HC-RicKO mice (Figures 4C and S7C), while there was no difference between genotypes in the length of stereocilia from the tallest row (data not shown).In addition, some stereocilia of OHCs from HC-RicKO mice showed an amorphous structure overlaying the tallest and outermost row of stereocilia (Figures 4A8, S7A6, andA8).A similar amorphous structure is transiently seen in wildtype mice at early postnatal ages but absent in stereocilin knockout mice and might represent remnants of the tectorial membrane (TM), which is an acellular roof plate covering the HCs. 38The tallest and outermost row of OHC stereocilia inserts into the TM.These insertions leave imprints on the lower surface of the TM, which are visible via scanning electron microscopy (SEM).The TM of HC-RicKO mice shows visible imprints similar to control mice (Figure S7B), suggesting the outermost OHC stereocilia row was normally contacting the TM.At age 2 weeks, HC-RicKO mice showed normal stereocilia architecture, similar to control mice (Figure S8), indicating that stereocilia developed normally.
To further investigate aberrations in stereocilia and HCs, we performed transmission electron microscopy (TEM) in 12-week-old mice.We did not find any further structural defects in stereocilia in TEM examinations.OHC (Figures 4B1-B2) and IHC (Figures 4B3-B4) stereocilia showed similar thickness and appearance in HC-RicKO mice as in control mice.The paracrystalline organization of actin filaments was also similar in stereocilia of both genotypes.Lastly, the morphology of HCs was also comparable between HC-RicKO and control mice (Figures S7D1-D6).
At the apical end of the middle and small row stereocilia lie the MET channels which are connected via tip-links to the adjacent, taller stereocilium.Any deflection toward the taller stereocilium opens the MET channel which is essential for hearing function.Loss of tip-links and consequently loss of MET channel function leads to middle and small row stereocilia shortening. 28,39We probed MET channel functionality by staining cochleae of 12-week-old mice with the styryl dye FM1-43X, which is known to enter and stain the HCs via MET channel. 40,41However, uptake via endocytosis has also been reported. 42,43IHCs and OHCs from HC-RicKO mice were equally stained with FM1-43X similar to control mice (Figure 4D).BAPTA treatment is known to disrupt the tip-links and therefore MET channel function.BAPTA abolished FM1-43X loading of HCs, thus confirming uptake via the MET channel (Figure 4D).In the apical cochlear turn, HCs from both genotypes were similarly stained with FM1-43X (Figure S9), although weaker than in the medial cochlear turn (Figure 4D) as previously described. 40To summarize, HC-RicKO mice display shortened and missing stereocilia but MET channel function is present, suggesting that mTORC2 is involved in stereocilia maintenance.

HC-RicKO mice show reduced synapse numbers
5][46][47][48] We therefore also investigated synapse numbers and spiral ganglion neuron counts in HC-RicKO and control mice.In IHCs, synapses consist of presynaptic ribbon proteins (Ctbp2) and postsynaptic AMPA-type glutamate receptors (GluR). 49HC-RicKO mice had fewer synapses at age 12 weeks (Figure 5A).Synapses were defined and quantified as juxtaposed CtBP2 and GluR2 immunoreactive puncta.The number of synapses was significantly reduced in all cochlear turns, with the biggest reduction in the medial cochlear turn (Figure 5B).While the number of CtBP2 puncta was reduced in HC-RicKO mice, there was also a significant increase in orphan ribbons (CtBP2 puncta not juxtaposed to GluR2, Figure S10A).Of note, synaptic counts were already reduced at age 4 weeks, indicating that this is an early effect (Figure S10B).The spiral ganglion neurons (SGNs) are the neurons that innervate the IHCs and form the fibers of the cochlear nerve up to the cochlear nuclei in the brainstem.There was no significant difference in SGN counts between HC-RicKO and control mice at age 12 weeks (Figures 5C and 5D) or earlier timepoints (2, 4, or 8 weeks of age, Figures S10C-S10H).SGN loss is frequently observed as a result of HC loss.SGN counts, although not significantly, were reduced in the basal cochlear turn of HC-RicKO mice at age 12 weeks (Figure 5D).However, at this timepoint many IHCs and OHCs were already lost at the base.Hence, HC-RicKO mice show cochlear synaptopathy with loss of synapses but without SGN degeneration.

HC-RicKO mice have a reduced synaptic F-actin network in inner hair cells
IHC synapses are crucial for hearing since they relay all auditory information to the brain.Sound-induced deflection of IHC stereocilia leads to IHC depolarization and Ca 2+ influx at synaptic active zones.Neurotransmitter-loaded vesicles that are tethered to a synaptic ribbon protein can then fuse with the presynaptic membrane, a process dependent on the Ca 2+ sensor otoferlin in IHCs. 501][12] Rictor knockout in different tissues was shown to regulate actin organization. 45,51Notably, it was also shown that a synaptic F-actin network regulates exocytosis at IHC synapses. 52,53isruption of the actin cytoskeleton facilitated exocytosis 52,53 and increased the distance between ribbon synapses and Cav1.3 channels at synaptic active zones. 52Given these findings and the reduced synapse number in HC-RicKO mice, we investigated synaptic active zones containing the ribbon protein (CtBP2), Cav1.3 channels, and the associated actin cytoskeleton.
IHCs from 12-week-old control mice show a dense actin cytoskeleton surrounding the ribbon synapses (Figure 6A).After depolarization of IHCs, Ca 2+ influx via Cav1.3 (also called alpha 1D) channels is necessary for fusion of vesicles loaded on synaptic ribbons with the presynaptic membrane. 54Therefore, Cav1.3 channels are closely localized to synaptic ribbons in IHCs 55 as observed in control samples (Figure 6B).In contrast, HC-RicKO mice showed only a submembranous, cortical actin cytoskeleton (Figure 6C).In addition, the distance between Cav1.3 channels and ribbon synapses was increased in HC-RicKO mice (Figure 6D), as shown by quantification of 255 active zones per genotype (Figure 6E).Notably, a similar phenotype with a reduced synaptic actin network and increased CtBP2-Cav1.3 distance was already observed at age 4 weeks in HC-RicKO mice, again indicating that this is an early effect (Figures S11A-S11E).To summarize, HC-RicKO mice have a reduced F-actin network at synaptic active zones, affecting the spatial organization between Cav1.3 channels and synaptic ribbons.

DISCUSSION
Here, we show that mTORC2 is an important regulator of cochlear IHC and OHC function.HC-RicKO mice show early-onset, progressive, and profound hearing loss.Increased DPOAE thresholds in HC-RicKO mice indicate that OHC function is affected.Importantly, HCs are lost, but this is a late event occurring after deafness.Early events that may account for the hearing loss, included abnormal stereocilia and reduced synapse numbers, disrupted synaptic actin cytoskeleton and increased distances between Cav1.3 channels and ribbon synapses.
Our results are in line with a recent study showing that Atoh1-Cre driven deletion of Rictor in the entire neurosensory epithelium causes progressive and profound hearing loss. 24However, the authors of this previous study attributed the hearing loss to HC loss as a consequence of defective Akt signaling. 24In contrast, we found that hearing loss precedes HC loss in our Rictor knockout mice.Although the reason for this difference is not clear, one possibility is the use of different Cre drivers.Atoh1-Cre mediates deletion in supporting cells and HCs, 56  Myo15-Cre used in our study mediates deletion specifically in HCs. 28In addition, we previously reported that Akt1 and Akt2/Akt3 KO mice show hearing loss without HC loss. 258][59][60] Similarly, in aged mice, a reduction in cochlear HC function precedes HC loss. 61e found abnormal stereocilia in HC-RicKO mice, in particular irregularly shortened and missing small and middle row stereocilia.This phenotype resembles that caused by mutations in actin genes Actb and Actg1. 62,63In adult stereocilia, the actin core is stable and actin turnover occurs at the tips, [64][65][66] a process which is dependent on the MET current. 398][69] However, HC-RicKO mice have normal stereocilia development and MET channel function is present as assessed with styryl dye staining.These results indicate that stereocilia shortening in HC-RicKO mice is not due to MET channel loss and that the MET machinery is functional even after stereocilia shortening.Similarly, stereocilia shortening after MET blockade does not lead to tip-link loss, indicating that tip-links reform or slide down during shortening. 39Knock-down of Rictor in mammalian cell lines leads to a disorganized actin cytoskeleton 11 and reduced cell spreading. 12Furthermore, the F-actin/G-actin ratio is reduced in tissues of Rictor KO mice. 45,51The parallel actin filaments at the core of stereocilia and the stereocilia diameter, as visualized by TEM, appeared normal in HC-RicKO mice.mTORC2 disruption might affect actin cytoskeletal dynamics and turnover rather than stable actin structures.
The actin cytoskeleton that surrounds the ribbon synapses in IHCs was also diminished in HC-RicKO mice.As shown previously, a filamentous actin network populates the basolateral end of IHCs forming a network that surrounds the synapses. 52,53Disruption of this synaptic actin cytoskeleton facilitates exocytosis at the ribbon synapse and confers mechanosensitivity to Cav1.3 channels. 52,53Morphologically, this is mediated by increased distances between Cav1.3 channels and ribbon synapses upon disruption of the actin cytoskeleton. 52Similarly, HC-RicKO mice had increased distances between Cav1.3 channels and ribbon synapses at IHC active zones.The precise mechanism of how the actin cytoskeleton regulates exocytosis remains unclear.However, a mechanism where the actin cytoskeleton regulates the replenishment of vesicles at the IHC synapse has been proposed. 52,53Increased recruitment of vesicles may lead to exhaustion of the vesicle pool after actin cytoskeleton disruption, 53 affecting exocytotic function in the long term.While the functional implications of these effects, measured in vitro, on hearing function are difficult to predict, it is tempting to speculate that the disrupted synaptic actin cytoskeleton and increased Cav1.3-ribbon distance in HC-RicKO mice are contributing to hearing loss.It may lead to disrupted synaptic function and the observed reduction in synaptic numbers.Similarly, it was recently shown that disrupting IHC exocytosis leads to loss of ribbon synapses and subsequently to IHC death. 70n contrast to hearing loss and impaired auditory HC function, we did not observe an obvious vestibular phenotype in HC-RicKO mice.Why is cochlear, but not vestibular HC function affected in HC-RicKO mice?Although cochlear and vestibular HCs share many morphological and molecular features, there are also several important differences.Vestibular HCs retain the kinocilium, a structure which is only transiently present in cochlear HCs, and have a different composition in basolateral K + channels which repolarize HCs. 71Another considerable difference is found in synaptic specializations between the two sensory systems, although they share common molecular components.While in IHCs a single ribbon contacts an afferent fiber in a bouton synapse, vestibular HCs contain calyceal synapses apart from bouton and dimorphic (bouton and calyx) synapses, where multiple ribbons contact a large afferent terminal. 71In addition, single afferent fibers contact multiple vestibular HCs, whereas cochlear afferent fibers exquisitely innervate one IHC. 71Notably, vestibular HCs have a different organization of Cav1.3 channels at the ribbon synapse 72 and even show Ca 2+ independent non-quantal transmission. 71Similarly to HC-RicKO mice, Cav1.3 KO mice are deaf with no vestibular dysfunction, despite greatly reduced Ca 2+ influx. 54,73,74This might be due to normal non-quantal synaptic transmission in Cav1.3 KO mice, which depends on potassium currents that are unaffected in vestibular HCs of Cav1.3 KO mice. 74herefore, a synaptic phenotype might not be apparent in vestibular HCs upon loss of Rictor.Another property found only in the hearing organ of mammals is the somatic electromotility of OHCs. 75We found reduced DPOAEs in HC-RicKO mice indicating that OHC function is disrupted.While this might be explained in part by the abnormal stereocilia, it is possible that a cytoskeletal defect also affects OHC electromotility and amplification function.OHCs have specialized structures in their lateral wall important for electromotility, including a cortical lattice cytoskeletal complex containing actin and spectrin filaments. 76,77It will be interesting to investigate whether potential cortical lattice defects impact OHC electromotility in HC-RicKO mice.
2][13][14] PKC alpha may be involved in the regulation of synaptic processes in IHCs 16 and in regulating OHC electromotility. 17Deletion of RhoA, Rac1 and Cdc42 all lead to abnormal stereocilia [78][79][80][81] and RhoA, Rac1, and Cdc42 small Rho GTPases have been proposed to regulate OHC motility. 82Alternatively, mTORC2 might regulate the actin cytoskeleton via Girdin, 83 which is an Akt substrate and actin binding protein. 84ur results reveal mTORC2 as a critical regulator of auditory HC function by regulating the actin cytoskeleton.mTORC2 is indispensable for both IHC and OHC function, regulating different structures such as stereocilia and synapses.Therefore, mTORC2 is a central regulator of cochlear HCs and thus hearing.It will be of interest to further determine mTORC2 regulated processes in HCs, as well as the precise underlying molecular mechanisms.Of note, Rictor expression decreases in both IHCs and OHCs with aging concomitant with a decrease in HC function. 61Therefore, it will also be interesting to determine the role of mTORC2 in aging HCs, as this might open new perspectives to treat sensorineural hearing loss.detect the wild-type allele (587 bp) and the Myo15-F with the cre-R primer (5 0 -TGGTGCACAGTCAGCAGGTTGG-3 0 ) were used to detect the Myo15-Cre allele (500 bp).The F-Rictor-fl (5 0 -TTATTAACTGTGTGTGGGTTG-3 0 ) and the R-Rictor-fl (5 0 -CGTCTTAGTGTTGCTGTCTAG-3 0 ) primers were used to detect floxed (295 bp) or wild-type Rictor allele (197 bp).F-Rictor-fl primer and R-Exc-Rictor-fl primer (5 0 -CAGATTCAAG CATGTCCTAAGC-3 0 ) were used to amplify a PCR product of 280 bp and confirm the presence of the recombined Rictor allele.No PCR product was detectable in wild-type or unrecombined Rictor allele (Figure 1).Each experimental mouse was genotyped at least twice, once at neonatal ages and once confirmed at adult ages after hearing measurements and/or tissue collection.
The mice were anesthetized using an intraperitoneal injection of ketamine (80 mg/kg), xylazine (12 mg/kg), and acepromazine (2 mg/kg).Body temperature was maintained at 37 C and monitored with a rectal probe.Ophthalmic ointment was applied following anesthesia induction to prevent corneal damage.
ABRs were measured using an RZ6-A-P1 processor with RA4PA preamplifier, RA4LI headstage, and an MF1 speaker in a closed-field setup using the BioSigRZ software (all from TDT, Alachua, FL, USA) and subcutaneous needle electrodes (inserted at the vertex, ipsilateral ear and a grounding electrode at the hind hip) placed in a sound attenuating chamber with a built in Faraday cage.Stimuli were presented either as clicks (duration 0.1 ms, repetition rate 21/s) or tone bursts (gate time 0.2 ms, duration 2.5 ms, repetition rate 21/s) at 4, 8, 16, and 32 kHz.Hearing thresholds were investigated by reducing the sound intensity in 5 dB steps from 90 dB SPL.The hearing threshold was defined as the lowest intensity in dB SPL to generate a visually detectable first and second peak.Acquisition was performed using 100 Hz highpass, 3 kHz lowpass and 50 Hz notch filters, and responses were averaged 512 times.ABR responses were collected and saved offline for later analysis.ABR suprathreshold waves at 8 kHz 90 dB SPL were analyzed for amplitudes and latencies.
DPOAEs were measured using an RZ6-A-P1 processor, two MF1 speakers in a closed-field setup (TDT, Alachua, FL, USA), and a ER10B+ microphone with preamplifier (Etymotic Research, Elk Grove Village, IL, USA).Two continuous primary tones were presented (f1 and f2, each from a separate speaker) with f2/f1 = 1.2 and geometrically centered at test frequencies of 4, 8, 16, and 32 kHz.The levels of the two primary tones were kept equal and decremented for each test frequency from 80 dB SPL to 20 dB SPL in 5 dB steps.Data was collected every 20.971 ms and averaged 128 times.The DPOAE threshold was defined as the lowest intensity in dB SPL to generate a detectable DPOAE at the frequency 2f1-f2.

Vestibular phenotyping
Assessment of vestibular function was performed by using a behavioral vestibular phenotyping pipeline as described in. 89Briefly, a score was calculated by observing behavioral signs (head tossing or circling behavior), and by performing a trunk curl test, a contact righting test, and a swim test.Scores were either no/normal (0) or yes/deficient (1) for the behavioral signs and trunk curl test, respectively; or scored from 0 to 3 in the contact righting and swimming test as shown in Table S1.

FM1-43X styryl dye staining of adult cochleae
12 week old mice were euthanized as described above but without transcardial perfusion and decapitated.The inner ears were extracted and quickly transferred to a sterile plastic Petri dish filled with ice-cold Hanks' Balanced Salt Solution (HBSS, with calcium, 14025092, Thermo Fisher Scientific, Reinach, Switzerland) and 10mM HEPES buffer (15630, Thermo Fisher Scientific, Reinach, Switzerland).Under a stereomicroscope, the stapes was removed from the oval window, and the round and oval window membranes as well as the apex were punctured with a needle to allow HBSS flow through the cochlea.The cochleae were then perfused through the oval and round windows with ice-cold HEPES buffered HBSS for 15 min followed by 30 s perfusion with ice-cold 5mM of FM1-43FX (F35355, Thermo Fisher Scientific, Reinach, Switzerland) diluted in HEPES buffered HBSS.For all samples with 5mM BAPTA (A4926, Sigma Aldrich Chemie GmbH, Steinheim, Germany) treatment, ice-cold HEPES buffered HBSS without calcium was used in all steps (14175095, Thermo Fisher Scientific, Reinach, Switzerland).Negative control samples were perfused with ice-cold HEPES buffered HBSS without FM1-43X.Finally, cochleae were perfused with ice-cold HEPES buffered HBSS to wash out styryl dye through the oval window and washed by quickly placing in HEPES buffered HBSS.The cochleae were then fixed overnight with 4% paraformaldehyde (PFA, FB002, Invitrogen AG, Reinach, Switzerland), decalcified for 8 h as described above with 10% EDTA, permeabilized with 5% Triton X-100 and 10% fetal bovine serum (FBS, 10082147, Thermo Fisher Scientific, Reinach, Switzerland) in 1x PBS for 2 h at room temperature, stained with Phalloidin 568 (1:500, A12380, Thermo Fisher Scientific, Reinach, Switzerland) overnight at 4 C, dissected, and mounted on slides with Vectashield Plus mounting medium.

Image acquisition
Images were acquired either with a Nikon Eclipse Ti microscope, equipped with an A1 point-scanning confocal unit (Nikon AG Instruments, Egg, Switzerland), or with a Nikon Eclipse Ti microscope, equipped with a Yokogawa CSU-W1 spinning disk (pinhole size 25 mm, unless otherwise stated) confocal unit (Nikon AG Instruments, Egg, Switzerland), and a Photometrics Prime 95B camera (cell size: 11 mm 3 11 mm).
For point-scanning confocal, either a 403 air objective (numerical aperture 0.95) or a 1003 oil objective (numerical aperture 1.45) was used.Fluorescence was excited with 402, 488, 560, 647 nm lasers and emission was filtered with 450/50 (PMT detector), 525/50 (GaAsP detector), 595/50 (GaAsP detector), 700/75 (PMT detector) bandpass filters, respectively.The pinhole size was set to 1 AU.The laser intensity and detector gain (offset was set to 0) were adjusted for each channel (on control samples) to prevent over-and undersaturation of the images.The same settings were applied to all samples belonging to the same analysis group.
For the spinning disk confocal, either a 203 air objective (numerical aperture 0.75) or a 403 air objective (numerical aperture 0.95) was used.Fluorescence was excited with 405, 488, 561, 640 nm lasers and emission was filtered with 460/50, 525/50, 630/75, 700/75 bandpass filters, respectively.The laser intensity and camera exposure time were adjusted for each channel to prevent over-and under-saturation of the images.The same settings were applied to all samples belonging to the same analysis group.
For p-Akt staining, images were taken with the spinning disk confocal and the 403 objective as z-stacks (step size of 0.6 mm), pixel size 0.28 mm.
For HC quantification, images were taken with the spinning disk confocal and the 203 objective as z-stacks (step size of 0.6 mm), pixel size 0.55 mm.Representative images for figures were taken with the point-scanning confocal, 403 objective with an additional 1.192 zoom, as z-stacks (step size of 0.50 mm), scans were averaged 2 times per XY section, pixel size 0.26 mm.
For FM1-43X styryl dye staining, images were taken with the spinning disk confocal (pinhole size 50 mm) and the 403 objective as z-stacks (step size of 0.6 mm), pixel size 0.28 mm.
For synapse number quantification, images were taken with the point-scanning confocal, 403 objective with an additional 2.0 zoom, as z-stacks (step size of 0.45 mm), scans were averaged 4 times per XY section, pixel size 0.16 mm.Representative images for figures were taken with the point-scanning confocal, 1003 objective with an additional 1.0 zoom, as z-stacks (step size of 0.15 mm), pixel size 0.12 mm.
For imaging the synaptic actin cytoskeleton, CtBP2/ribbon synapses, and Cav1.33channels, images were taken with the point-scanning confocal, 1003 objective with an additional 1.22 zoom, as z-stacks (step size of 0.125 mm), scans were averaged 4 times per XY section, pixel size 0.1 mm.

Histology and hematoxylin and eosin (H&E) staining
For histological analysis, inner ears were collected, fixed, and decalcified as described for immunofluorescence staining and processed using a 29 h paraffin embedding program.7.5mm sections were cut, stained with H&E using a Gemini autostainer (Thermo Fisher Scientific, Reinach, Switzerland), and imaged using a Hamamatsu NanoZoomer S60 (Hamamatsu Photonics, Hamamatsu City, Japan) slide scanning device using a 403 magnification (numerical aperture 0.75), pixel size 0.23 mm.

Image analysis and quantification
For quantification of Akt-pSer473 signal intensity, Fiji (Fiji is just ImageJ) 91 was used.Briefly, 3D stacks with equal amounts of z-layers (10 layers) were maximum intensity projected.Mean signal intensity for Akt-pSer473 staining in the HC region as well as of the background were measured.The background was subtracted from the signal of the HC region, results were normalized to control (mean = 1) and analyzed.For each cochlear turn, 3-5 regions were measured.
For HC counting and measuring cochlear length, images were maximum intensity projected, all segments from the same cochlea were stitched, and a scale bar was added using the microscope's Nikon NIS software.Then, using Fiji 91 and the Measure_Line plugin from Mass Eye and Ear, Eaton-Peabody Laboratories, Histology Core (https://masseyeandear.org/research/otolaryngology/eaton-peabodylaboratories/histology-core), a line was drawn along the IHC lateral margin to measure the length of the cochlear sensory epithelium (after pixel calibration using the scale bar as described on the plugin site).The same line and plugin were used to generate a mask separating the cochlea into 5% distances from the Apex which was saved as region of interest (ROI).For each 5% distance from the Apex, IHCs, OHCs, and respective missing HCs were counted to plot the results as cochleogram as described in. 92The place-frequency map was calculated using the formula d = (LOG10((f+6.664)/9.8)/LOG10(10))/0.0092(where d is the distance from the Apex in percent and f the frequency in kHz) as described in 86 and derived from. 87,88tereocilia length was analyzed as described in Caberlotto et al. 28 Briefly, the length of measurable stereocilia in scanning electron microscopy (SEM) images from the middle and small stereocilia rows was normalized to the length of stereocilia in the tall row.6-45 cells from the apical and medial cochlear turn of 3 mice were analyzed per genotype.The observer was blinded to genotype.
Synapse numbers were quantified using Imaris 9.9.0 software (Bitplane, Oxford Instruments), where the spot detection tool was used to detect CtBP2 and GluR2 puncta in the IHC region.Synapses were defined as juxtaposed CtBP2 and GluR2 spots (<1 mm).The total amount of synapses was divided by the total number of IHCs in the imaged segment (usually around 20, identified by DAPI-stained nuclei), to display synapse counts per single IHC.3-6 segments per cochlear turn were imaged and examined, averaged for the same ear, and analyzed for comparison between genotypes.Orphan ribbons (CtBP2 puncta not juxtaposed to GluR2) were calculated by subtracting the number of (juxtaposed) synapses from the total number of CtBP2 puncta.
SGN counts were quantified in H&E stained histological sections, on every third slide where three consecutive 7.5mm sections from serial cuts of paraffin blocks were collected along the entire cochlear diameter.Analysis was performed in QuPath version 0.3.0, 93where the region of the Rosenthal's spiral canal was manually annotated and categorized into cochlear region (Apex, Middle, Base).The area, as well as the SGN numbers, in these regions were then measured in QuPath with the Cellpose QuPath extension (https://github.com/BIOP/qupathextension-cellpose,version 0.3.5)using a custom script, where we used a livecell Cellpose 2D model -''L2'' 94 to count the SGNs using parameters: diameter (35) and pixelSize (0.3).Segmented cells were validated based on shape features: SGNs were excluded if they met criteria size (<40 mm 2 ) or circularity (<0.375).The counts were normalized to area (SGN/mm 2 ), averaged for each cochlear turn of the same ear, and analyzed for comparison between genotypes.
The distance between CtBP2 (ribbon) and Cav1.3 channels was measured as described previously. 52,72Briefly, images were deconvolved with Huygens Professional version 21.10 (Scientific Volume Imaging, The Netherlands) using the express deconvolution standard profile.Then, using the Fiji JACoP plugin, 95 the center mass distance was measured using objects based methods.2-3 regions in the medial cochlear turn were imaged each spanning around 13 IHCs.255-275 active zones (CtBP2 and Cav1.3 juxtaposed patches) were measured per genotype, coming from 4 to 5 ears from 2 to 3 animals.To plot intensity profiles the command 'plot profile' was used in Fiji, the data was saved and normalized.

Electron microscopy (EM) studies
For EM studies, 2 and 12 week old mice were euthanized as described for immunofluorescence and the inner ears were quickly isolated.
For scanning electron microscopy (SEM), the inner ears were fixed in 2.5% Glutardialdehyde (4157.2,Carl Roth, Karlsruhe, Germany) overnight at 4 C, decalcified as described for immunofluorescence for 4-7h, washed, and then the cochlear capsule was carefully opened under a stereomicroscope.Afterward, the samples were dehydrated in ascending series of ethanol, critical point dried, mounted, and gold sputtered.Images were acquired using a Philips XL30 ESEM (Philips, Amsterdam, Netherlands) SEM microscope at 5-10kV.
For transmission electron microscopy (TEM), the samples were processed as described in. 96Briefly, inner ears were quickly transferred to Leibovitz's L-15 solution (21083, Thermo Fisher Scientific, Reinach, Switzerland), where the round and oval windows as well as the apical portion of the cochlea were opened.The cochlea was then gently and slowly perfused with 1% glutaraldehyde (16300, all reagents for TEM are from EMS, Lucerna-Chem, Luzern, Switzerland)/4% formaldehyde (15700) in 0.15 M cacodylate buffer (pH 7.2) supplemented with 2 mM CaCl2 and left in the same fixative solution for 1h.Afterward, the samples were fixed with 2.5% glutaraldehyde in cacodylate buffer for 1h.Decalcification was performed as described for immunofluorescence for 3 days and then the cochleae were bisected longitudinally.The cochleae were shortly immersed in 2.5% glutaraldehyde in cacodylate buffer for 15 min and placed for 1h in 2.5% glutaraldehyde in cacodylate buffer with 1% tannic acid to stain the links.Post-fixation was performed in 1% osmium tetroxide/1.5% potassium ferrocyanide in cacodylate buffer for 1h at 4 C in the dark and the samples were then dehydrated on ice in ascending ethanol series.Embedding was performed in a mixture of resin and propylene oxide.Resin polymerization was carried out in an oven at 60 C for 48h.Ultrathin sections were cut with diamond knives, collected with Formvar/carbon-coated copper grids (FCF2010-CU), and impregnated with uranyl acetate and lead citrate.Images were acquired with a FEI Tecnai G2 Spirit (FEI company, Hillsboro, OR, USA) TEM microscope at 80kV and a Veleta camera (EMSIS, Mu ¨nster, Germany) operated by RADIUS software from EMSIS.

QUANTIFICATION AND STATISTICAL ANALYSIS
Results are presented as means G SDs.The statistical analysis was performed and graphs were created with Prism 9 software (GraphPad software, La Jolla, CA, USA).To determine differences between two groups, a student's unpaired t-test was used.Ranked data (balance scores) were analyzed with a Mann-Whitney test.The results were considered statistically significant with a p value <0.05.Asterisks in the figures summarize the degree of significance (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001).Statistical details are indicated in the figure legends.

Figure 1 .
Figure 1.KO confirmation of HC-RicKO mice (A) Hemizygous Myo15-Cre +/À mice were crossed with homozygous Rictor fl/fl mice harboring LoxP sites upstream and downstream of exon 4 and 5, respectively.Forward (F) and reverse primers (R) were used to detect Cre-mediated recombination, resulting in a 280 bp PCR product after Cre-mediated recombination of the Rictor allele.(B) PCR using genomic DNA from dissected organ of Corti (OC), spiral ganglion (SG) or vestibular organ (vest; macular and cristae ampullaris organs pooled) tissue, separate for left (L) and right (R) inner ears.Genomic DNA from outer ear skin of both genotypes was used as negative control of Cre-mediated recombination.Genomic DNA from heart samples of tamoxifen-induced cardiomyocyte-specific Rictor KO mice was used as positive control (pos ctrl). 85No template control (NTC) contains all PCR reaction components including primers but no DNA.Successful Cre-mediated recombination of the Rictor allele was only found in OCs and vestibular organs of HC-RicKO mice.(C) Representative images (maximum intensity projections) of the medial cochlear turn from 4-week-old mice stained with an antibody against Akt-pSer473.Myosin7a antibody, phalloidin and nuclear DAPI staining visualize the hair cells.Scale bar for all figures = 20 mm.(D) Quantification of mean fluorescence intensity (MFI) in arbitrary units (a.u.) of the Akt-pSer473 signal intensity in the medial cochlear turn of 4-week-old mice.n = 3 mice per genotype.Results are presented as means G SDs. Student's t test, *p < 0.05.
Figure 1.KO confirmation of HC-RicKO mice (A) Hemizygous Myo15-Cre +/À mice were crossed with homozygous Rictor fl/fl mice harboring LoxP sites upstream and downstream of exon 4 and 5, respectively.Forward (F) and reverse primers (R) were used to detect Cre-mediated recombination, resulting in a 280 bp PCR product after Cre-mediated recombination of the Rictor allele.(B) PCR using genomic DNA from dissected organ of Corti (OC), spiral ganglion (SG) or vestibular organ (vest; macular and cristae ampullaris organs pooled) tissue, separate for left (L) and right (R) inner ears.Genomic DNA from outer ear skin of both genotypes was used as negative control of Cre-mediated recombination.Genomic DNA from heart samples of tamoxifen-induced cardiomyocyte-specific Rictor KO mice was used as positive control (pos ctrl). 85No template control (NTC) contains all PCR reaction components including primers but no DNA.Successful Cre-mediated recombination of the Rictor allele was only found in OCs and vestibular organs of HC-RicKO mice.(C) Representative images (maximum intensity projections) of the medial cochlear turn from 4-week-old mice stained with an antibody against Akt-pSer473.Myosin7a antibody, phalloidin and nuclear DAPI staining visualize the hair cells.Scale bar for all figures = 20 mm.(D) Quantification of mean fluorescence intensity (MFI) in arbitrary units (a.u.) of the Akt-pSer473 signal intensity in the medial cochlear turn of 4-week-old mice.n = 3 mice per genotype.Results are presented as means G SDs. Student's t test, *p < 0.05.

Figure 2 .
Figure 2. Continued (E) ABR wave IV amplitudes (peak-through) measured for an 8 kHz, 90 dB SPL stimulus in ears/animals specified in (A) at timepoints indicated.Results are presented as means G SDs. Student's t test, ***p < 0.001, ****p < 0.0001.(F) ABR wave IV latencies (onset-peak) measured for an 8 kHz, 90 dB SPL stimulus in ears/animals specified in (A) at timepoints indicated.Results are presented as means G SDs. Student's t test, ns not significant, *p < 0.05, ****p < 0.0001.(G) ABR interwave I-IV latencies (peak-peak) measured for an 8 kHz, 90 dB SPL stimulus in ears/animals specified in (A) at timepoints indicated.Results are presented as means G SDs. Student's t test, ns not significant.(H) Illustrative suprathreshold ABR wave response and schematic representation of ABR waveforms analyzed in panels (C-G).

Figure 4 .
Figure 4. HC-RicKO mice show abnormal stereocilia but MET channel function is present (A) SEM images of control (A1-A5) and HC-RicKO (A6-A10) stereocilia.Panels A1-A2 (control) and A6-A7 (HC-RicKO) show an overview of stereocilia of different hair cell rows.Panels A3, A5 (control) and A8, A10 (HC-RicKO) show outer hair cell stereocilia.Panels A4 (Control) and A9 (HC-RicKO) show inner hair cell stereocilia.Stereocilia from outer hair cells of HC-RicKO mice show an amorphous structure on the outermost row (asterisk), shortened stereocilia (arrowheads), and missing stereocilia (arrows).All images are representative images from the medial cochlear turn of 12-week-old mice.n = 3 mice per genotype.Scale bar sizes are indicated in the corresponding figure panels.(B) TEM images of control (B1, B3) and HC-RicKO (B2, B4) stereocilia.Panels B1 (control) and B2 (HC-RicKO) show outer hair cell stereocilia.Panels B3 (control) and B4 (HC-RicKO) show inner hair cell stereocilia.Magnifications of the insets showing the paracrystalline organization of actin filaments in stereocilia are displayed inside the corresponding image panels.All images are representative images from the medial cochlear turn of 12-week-old mice.n = 3 mice per genotype.Scale bar for all figures = 200 nm.(C) Analysis of outer hair cell stereocilia length in the medial cochlear turn, where the length from the middle (L2) and small (L3) stereocilia rows was normalized to the length of stereocilia in the tall row (L1).45 cells from 3 mice were analyzed per genotype.Results are presented as means G SDs. Student's t test, ****p < 0.0001.(D) FM1-43X staining of adult cochlear hair cells.Images are maximum intensity projections from the medial cochlear turn.Cochleae of 12-week-old mice were either perfused with FM1-43X (green), FM1-43X + BAPTA or HBSS only for the negative control (Neg Control).To identify the hair cells, actin filled stereocilia and cuticular plates were stained with phalloidin (red).Data from n = 3 mice for FM1-43X and 2 mice for FM1-43X + BAPTA staining.Scale bar for all figures = 10 mm.

Figure 5 .
Figure 5. HC-RicKO mice show reduced synapse numbers (A) Representative images (maximum intensity projections) of ribbon synapses (CtBP2) and glutamate receptors (GluR2) of 12-week-old mice from inner hair cells of the medial cochlear turn.Scale bar for all figures = 10 mm.(B) Quantification of synapse counts (juxtaposed CtBP2-GluR2) per inner hair cell (IHC) in 12-week-old mice.Data from n = 6-7 ears from 4 to 5 animals.Results are presented as means G SDs. Student's t test, *p < 0.05, **p < 0.01.(C) Mid-modiolar cochlear sections stained with H&E from 12-week-old mice.The Rosenthal's canal containing the spiral ganglion neurons (SGNs) is outlined with a dashed line.Scale bar for all figures = 250 mm.(D) Quantification of SGN counts in H&E stained histological sections of 12-week-old mice.Data from n = 3-4 mice.Counts were normalized to area (SGN/mm 2 ) and averaged for each cochlear turn of the same ear before analysis.Results are presented as means G SDs. Student's t test, ns not significant.

Figure 6 .
Figure 6.HC-RicKO mice show a reduced synaptic F-actin network in inner hair cells (A) Representative image (maximum intensity projection, 7 z-layers of 0.125 mm each) of an inner hair cell from the medial cochlear turn of a 12-week-old control mouse.Phalloidin labeled F-actin (magenta) forms a network surrounding ribbon synapses (CtBP2) and calcium channels (Cav1.3).Magnification of the inset is displayed in panel (B).Scale bar size = 1 mm.(B) Magnification from the square shown in panel (A).The ribbon synapse (CtBP2, red) lies in proximity of the calcium channel (Cav1.3,green) in control mice.The graph represents the fluorescent intensity profile of the synapse measured along the white dashed line.Scale bar size = 0.5 mm.(C) Representative image (maximum intensity projection, 7 z-layers of 0.125 mm each) of an inner hair cell from the medial cochlear turn of a 12-week-old HC-RicKO mouse.Phalloidin labeled F-actin (magenta) shows a cortical, submembranous localization.Magnification of the inset is displayed in panel (D).Scale bar size = 1 mm.(D) Magnification from the square shown in panel (C).There is a larger distance between the ribbon synapse (CtBP2, red) and calcium channel (Cav1.3,green) in HC-RicKO mice.The graph represents the fluorescent intensity profile of the synapse measured along the white dashed line.Scale bar size = 0.5 mm.(E) Quantification of the center mass distance between ribbon (CtBP2) and Cav1.3.Data from n = 4-5 ears from 2 to 3 animals aged 12 weeks, where 255 active zones per genotype were analyzed.Results are presented as means + SDs.Student's t test, ****p < 0.0001.