The acquisition of mechano‐electrical transducer current adaptation in auditory hair cells requires myosin VI

Key points The transduction of sound into electrical signals occurs at the hair bundles atop sensory hair cells in the cochlea, by means of mechanosensitive ion channels, the mechano‐electrical transducer (MET) channels. The MET currents decline during steady stimuli; this is termed adaptation and ensures they always work within the most sensitive part of their operating range, responding best to rapidly changing (sound) stimuli. In this study we used a mouse model (Snell's waltzer) for hereditary deafness in humans that has a mutation in the gene encoding an unconventional myosin, myosin VI, which is present in the hair bundles. We found that in the absence of myosin VI the MET current fails to acquire its characteristic adaptation as the hair bundles develop. We propose that myosin VI supports the acquisition of adaptation by removing key molecules from the hair bundle that serve a temporary, developmental role. Abstract Mutations in Myo6, the gene encoding the (F‐actin) minus end‐directed unconventional myosin, myosin VI, cause hereditary deafness in mice (Snell's waltzer) and humans. In the sensory hair cells of the cochlea, myosin VI is expressed in the cell bodies and along the stereocilia that project from the cells’ apical surface. It is required for maintaining the structural integrity of the mechanosensitive hair bundles formed by the stereocilia. In this study we investigate whether myosin VI contributes to mechano‐electrical transduction. We report that Ca2+‐dependent adaptation of the mechano‐electrical transducer (MET) current, which serves to keep the transduction apparatus operating within its most sensitive range, is absent in outer and inner hair cells from homozygous Snell's waltzer mutant mice, which fail to express myosin VI. The operating range of the MET channels is also abnormal in the mutants, resulting in the absence of a resting MET current. We found that cadherin 23, a component of the hair bundle's transient lateral links, fails to be downregulated along the length of the stereocilia in maturing Myo6 mutant mice. MET currents of heterozygous littermates appear normal. We propose that myosin VI, by removing key molecules from developing hair bundles, is required for the development of the MET apparatus and its Ca2+‐dependent adaptation.


Introduction
Myo6 was one of the first deafness genes identified (Avraham et al. 1995). Mutations of the gene encoding myosin VI, one of a number of unconventional myosins, are associated with dominant progressive (Melchionda et al. 2001) and recessive congenital (Ahmed et al. 2003) deafness in humans. Mice homozygous for the Snell's waltzer mutation, a 130-bp deletion in the Myo6 gene resulting in a functional null mutation, are deaf and exhibit vestibular dysfunction associated with progressive degeneration of the sensory epithelium in the cochlea and vestibular organs (Avraham et al. 1995). In the cochlea, myosin VI is exclusively expressed in the sensory hair cells and it is abundant in the cuticular plate, the pericuticular necklace and the cytoplasm Heidrych et al. 2009;Roux et al. 2009). Although early studies reported that myosin VI seemed to be absent from the stereociliary bundle of mammalian hair cells (Avraham et al. 1997;Hasson et al. 1997), more recent investigations have shown, using both immunogold and immunofluorescence labelling, that myosin VI is located along the stereocilia (Rzadzinska et al. 2004;Hertzano et al. 2008) between the actin core and the lateral membrane, but absent from the distal tip (Rzadzinska et al. 2004). The hair cells of Snell's waltzer mutant mice have no detectable myosin VI and appear to form normally during embryonic development. However, during the first postnatal week the hair bundles of inner and outer hair cells (IHCs and OHCs) become disorganized, with a partial loss of their normal orientation and a tendency of individual stereocilia to fuse (Self et al. 1999). Similar abnormalities of the hair bundles are seen in mutant Tailchaser mice in which myosin VI is present but, due to a dominant point mutation, unable to move processively (Hertzano et al. 2008;Pylypenko et al. 2015). Thus, motile myosin VI is required for maintaining the organization of the hair bundle during its postnatal maturation.
Mechano-electrical transduction occurs via the opening of highly sensitive mechanically gated ion channels at the tips of hair-cell stereocilia (Beurg et al. 2009). Here we exploit the Snell's waltzer (Myo6 sv , later referred to as sv) mutant mouse to identify a crucial role for myosin VI in the functional development of the hair bundle's mechano-electrical transducer (MET) apparatus during a time when the MET currents are undergoing major biophysical changes (Waguespack et al. 2007;Lelli et al. 2009;Corns et al. 2014).

Ethics statement
The work on mice was licensed by the Home Office under the Animals (Scientific Procedures) Act 1986 and was approved by the University of Sussex, University of Sheffield and Wellcome Trust Sanger Institute Ethical Review Committees.

Electrophysiology
OHCs and IHCs from Snell's waltzer (sv) mutant mice, which have an intragenic deletion in the motor domain of the myosin VI gene (Avraham et al. 1995), and their littermate (heterozygous) controls were studied in acutely dissected organs of Corti from postnatal day 2 (P2) to P9, where the day of birth is P0. Mice were killed by cervical dislocation. All heterozygous and homozygous animals from which organs of Corti were used for experiments were kept frozen at −20°C for subsequent genotyping as previously described (Self et al. 1999). The mice were obtained from Karen P. Steel (Wellcome Trust Sanger Institute, UK) and Fiona Buss (University of Cambridge, UK).
The organs of Corti were dissected and transferred to a microscope chamber and immobilized using a nylon mesh fixed to a stainless steel ring. The chamber was perfused at a flow rate of about 10 ml h −1 , from a peristaltic pump, with extracellular solution composed of (in mM): 135 NaCl, 5.8 KCl, 1.3 CaCl 2 , 0.9 MgCl 2 , 0.7 NaH 2 PO 4 , 2 Na-pyruvate, 5.6 D-glucose, 10 Hepes-NaOH. Amino acids and vitamins for Eagle's minimum essential medium (MEM) were added from concentrates (Invitrogen, UK). The pH was adjusted to 7.5 and the osmolality was about 308 mosmol kg −1 . The organs of Corti were observed with upright microscopes (Zeiss ACM, Oberkochen, Germany; Leica, Wetzlar, Germany) with Nomarski optics.
Hair bundles were mechanically stimulated by a fluid jet with a tip diameter of 7-10 μm, according to experimental designs described before (Kros et al. 1992(Kros et al. , 2002Géléoc et al. 1997;Corns et al. 2014). Mechanical stimuli were either force steps of 50 ms duration or 45 Hz sinusoids. The driver voltage for the fluid jet was low-pass filtered at 1 kHz (8-pole Bessel) with the aim of preventing resonances in the piezo disc. Bundle movements could be recorded simultaneously with the MET currents with a laser differential interferometer (Géléoc et al. 1997). For some of the data shown, bundle displacements were not directly measured together with the MET currents, but inferred from the previously established relationship between bundle displacement and driver voltage using a conversion value of 10 nm V −1 ). We could not reliably measure bundle displacements in the mutant (sv/sv) IHCs because of their particularly disorganized bundle morphology, so for IHC MET current recordings we only quote driving force from the fluid jet rather than bundle displacement. Hair bundle stiffness (measured as translational stiffness at steady state towards the end of the 50 ms force steps to the hair bundle) was calculated from the linear fluid velocity of the jet (calibrated against a carbon fibre of known stiffness) and by modelling the hair bundles as prolate spheroids, as previously described in detail (Géléoc et al. 1997). For OHCs, positive driver voltage and bundle movement indicates fluid flow out of the jet (positioned on the modiolar side), which moves the hair bundle in the excitatory direction towards the kinocilium. For IHCs, excitatory bundle stimulation (again represented by positive driver voltage and bundle movement) was achieved by fluid flow into the jet, because the jet was placed on the strial side. MET currents versus bundle displacement (see Figs 1, 2 and 6) or driver voltage (see Fig. 3) were fitted using a second-order Boltzmann function: where x 1 and x 2 are the set points for the transition between the two closed states and for the opening transition, respectively, in a three-state channel model (Crawford et al. 1989;Géléoc et al. 1997), and a 1 and a 2 are their corresponding sensitivities to displacement or driver voltage.
To test the effects of extracellular Ca 2+ on the MET currents, the hair bundles were perfused with Ca 2+ concentrations of 0.1 and 10 mM (Fig. 5) instead of 1.3 mM, which was used for all the other experiments. The solution containing 0.1 mM Ca 2+ was (in mM): 147 NaCl, 5.8 KCl, 0.1 CaCl 2 , 0.7 NaH 2 PO 4 , 2 Na-pyruvate, 5.6 D-glucose and 10 Hepes-NaOH (pH 7.5; 308 mosmol kg −1 ); that with 10 mM Ca 2+ was (in mM): 132 NaCl, 5.8 KCl, 10 CaCl 2 , 0.7 NaH 2 PO 4 , 2 Na-pyruvate, 5.6 D-glucose and 10 Hepes-NaOH (pH 7.5; 308 mosmol kg −1 ). The solutions were superfused via a pipette with a much larger tip diameter than that of the fluid jet, positioned orthogonally to the axis of mechanical sensitivity of the hair bundle, and the flow did not directly stimulate the stereocilia. For every extracellular solution change, the fluid jet used for stimulating the hair bundles was filled with the new solution by suction through its tip to prevent dilution of the drug concentration. The effects of the Ca 2+ chelator BAPTA (Molecular Probes, The Netherlands) on the MET current were tested by using two different intracellular concentrations (0.1 and 10 mM Na 4 BAPTA), instead of 1 mM EGTA-CsOH, in the L-glutamic acid-based intracellular solution and osmolality was kept constant by adjusting the concentration of the L-glutamic acid.

FM1-43 labelling
Stock solutions of 3 mM FM1-43 (N-(3-triethylammoniumpropyl)-4-(4-(dibutylamino)styryl) pyridiniumdibromide, Molecular Probes) were prepared in water. FM1-43 dye labelling was studied using bath application. After dissection, organs of Corti (aged P3-P6) were held in position at the bottom of a chamber under a nylon mesh and perfused with normal extracellular J Physiol 594.13 solution. All experiments were performed at room temperature (22-25°C), as previously described (Gale et al. 2001). Briefly, the cochleae were bathed with solution containing 3 μM FM1-43 for 10-15 s, and immediately washed several times with normal extracellular solution. The cochleae were then viewed with an upright microscope equipped with epifluorescence optics and FITC filters (excitation 488 nm, emission 520 nm) using a 63× water immersion objective. Images were captured from live cultures at fixed time points after dye application using a 12-bit cooled CCD camera (SPOT-JNR, Diagnostics Inc., USA). For each experiment cochleae from +/sv and sv/sv mutant mice were dissected and processed simultaneously in the same chamber to reduce variability between different experiments. A total number of 12 +/sv and 12 sv/sv cochleae from eight mice of each genotype were used.

Immunofluorescence microscopy
For cadherin 23 (CDH23) labelling, dissected cochleae were treated with 5 mM BAPTA for 15 min to expose CDH23 ectodomain epitopes and then fixed in 3.7% formaldehyde in 0.1 M sodium phosphate buffer for 1 h at room temperature and washed three times in PBS. Cochlear coils were pre-blocked and permeabilized in TBS containing 10% heat-inactivated horse serum and 0.1% TX-100 for 1 h, and incubated overnight in the same solution containing a 1:100 dilution of the rabbit antibody Ela3N (kindly provided by Prof. C. Petit) that is directed against peptide epitopes in the ectodomain of CDH23 (Michel et al. 2005). Samples were washed three times with TBS, and stained with FITC-conjugated swine anti-rabbit Ig (1:100 dilution) and rhodamine-conjugated phalloidin (1:1000 dilution) for 2 h, washed in TBS, mounted in Vectashield and viewed with a Zeiss Axioplan 2 wide-field microscope using a 100× oil immersion lens with a numerical aperture of 1.4. The number of cochleae tested was: 15 +/sv, 13 sv/sv (P2-P6); six homozygous Myo7a 6J (Kros et al. 2002) mutants (6J/6J, P2-P6).

Scanning electron microscopy
Cochleae from homo-and heterozygous mice at P6 were investigated by scanning electron microscopy. Freshly isolated cochleae were locally perfused through oval and round windows with 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) and then fixed for 3 h at room temperature in the same fixative. Samples were then carefully washed in PBS and processed with the OTOTO method adapted from Hunter-Duvar (1978), dehydrated in an ethanol series, critical point dried (CPD 20, BAL-TEC, Balzers, Liechtenstein), mounted on stubs with conductive paint and viewed with a Hitachi FE S-4800 scanning electron microscope operated at 3-5 kV.

Statistical analysis
Averaged data are presented as mean ± SEM and statistical comparisons are based on the paired or unpaired two-tailed Student's t test, in cases where we report how a parameter is affected by one factor. When reporting how two factors (mutant status and postnatal age) affected MET current size we used two-way ANOVA with Bonferroni post hoc tests. P < 0.05 was used as the criterion for statistical significance.

MET currents and adaptation in heterozygous and homozygous mutant OHCs
We first sought to establish the basic properties of the MET currents of cochlear hair cells of Snell's waltzer mice, concentrating on OHCs because IHCs are more difficult to approach for hair bundle stimulation and patch clamp recording during the first postnatal week. In +/sv OHCs bundle movement towards the kinocilium (defined as excitatory and shown as positive displacements in Fig. 1) elicited rapid inward currents of up to −900 pA at the holding potential of −84 mV ( Fig. 1A) that, for intermediate-sized MET currents, declined with two time constants (Fig. 1C): a fast one of 0.34 ± 0.04 ms (contributing 51% of the total decline) and a residual slow one of 12 ± 3 ms (n = 6, P4-P6). This decline is due to adaptation (Eatock et al. 1987;Crawford et al. 1989) and its total extent was 29 ± 4% (n = 6) for small excitatory stimuli. Inhibitory hair bundle stimulation (i.e. away from the kinocilium, shown as negative displacements in Fig. 1) shut off the fraction of the current flowing at rest. At the offset of large inhibitory steps, a transient rebound ( Fig. 1D downward dip: rebound adaptation) was observed. The time course of relaxation following the rebound adaptation proceeded with a fast (0.38 ± 0.04 ms, contributing 70%, n = 6) and a slow time constant (8 ± 2 ms). At +86 mV (i.e. near the Ca 2+ equilibrium potential) adaptation was abolished (Fig. 1B), as previously observed for hair cells from wild-type mice (Kros et al. 2002;Corns et al. 2014) and other vertebrates (Assad et al. 1989;Crawford et al. 1989). Figure 1E shows that at +86 mV the current activated at rest is increased. For eight P4-P6 OHCs the resting open probability of the MET current increased from 5.3 ± 0.8% at −84 mV to 19 ± 2% at +86 mV (P < 0.0001, paired t test). This is consistent with Ca 2+ entry, via the resting MET current at −84 mV, inducing a degree of adaptation, and thus closing some MET channels and inducing a rightward shift and a change in shape of the curve describing the relationship between MET current and bundle displacement (Assad et al. 1989;Crawford et al. 1989;Corns et al. 2014). Using a three-state model of MET channel gating, featuring two closed states and one open state, the shift and shape-change can be explained by a shift in the set point for the transition between the closed states, according to eqn (1), of 49 nm for the P6 cell of Fig. 1E, similar to OHCs (P6-P9) from various wild-type mouse strains (Corns et al. 2014: 61 nm). In cells studied at both potentials, the mean saturating current was −681 ± 32 pA at −84 mV and +635 ± 28 pA at +86 mV (n = 15), with a reversal potential of +3.8 ± 0.2 mV (n = 15), again in the range of values reported for wild-type mouse OHCs (Kros et al. 1992;Géléoc et al. 1997;Corns et al. 2014;Marcotti et al. 2014). In conclusion, MET currents of +/sv OHCs appear no different from those recorded from OHCs of wild-type mice.
Although MET currents could be elicited in sv/sv OHCs up to P7, the gating characteristics of the channel were altered. MET currents from a P7 sv/sv OHC are shown in Fig. 2. Using either hyperpolarized ( Fig. 2A) or depolarized ( Fig. 2B) membrane potentials, currents were not usually detected at rest and could not be elicited for excitatory bundle movements smaller than 60 ± 15 nm (n = 12; only one of these cells had a small resting MET current). There were no signs of an adaptive decline during excitatory force steps ( Fig. 2C) or of a rebound at the end of inhibitory force steps (Fig. 2D). From about P5 some sv/sv OHCs responded to large negative bundle displacements with an inward current ( Fig. 2D) of −34 ± 7 pA (n = 7, P5-P7) instead of the normal reduction in the (inward) resting MET current. This MET current with opposite polarity was different from the anomalous MET current recently reported (Alagramam et al. 2011;Marcotti et al. 2014) because the former occurred in response to smaller bundle displacements and did not show any strong 'hyperadaptation': an unusually strong decline in the anomalous MET currents that does not require influx of extracellular Ca 2+ ). The MET current reversal potential in sv/sv OHCs was +4.6 ± 0.5 mV (n = 15) and not significantly different from +/sv controls. The relationship between MET current size and bundle displacement, both measured just after the onset of activation, is shown in Fig. 2E. In sv/sv OHCs, the operating range of the current-displacement relationship (Fig. 2E) for positive force steps is shifted to the right compared to that of the littermate +/sv controls (Fig. 1E). The resting open probability did not increase upon depolarization in the homozygous mutants and there was no leftward shift or shape change of the current-voltage curves, pointing to a complete absence of Ca 2+ -induced adaptation. The only difference between the fitted Boltzmann curves (eqn 1) was a 10% increase in the set points x 1 and x 2 upon depolarization, possibly due to some artefactual drift in the bundle displacement measurement. This lack of any signs of adaptation was also present in earlier postnatal OHCs (P3-P5), when the hair bundles were less disorganized and larger MET currents (up to −660 pA at −84 mV) could still be recorded, showing that the drastic increase in hair bundle deterioration from about P6 onwards is not the cause of the absence of MET current adaptation. The current-displacement relationship (Fig. 2E) was well fitted with the same absolute Zero current potential (mV) −57 ± 4 (9) −58 ± 4 (12) Membrane capacitance (pF) 5.9 ± 0.4 (9) 6.2 ± 0.6 (12) Linear leak conductance at 1.0 ± 0.5 (9) 1.0 ± 0.4 (12) −84 mV (nS) Maximal I K,neo at 0 mV (nA) 2.7 ± 1.0 (9) 2.4 ± 0.8 (12)  parameter values for positive and negative displacements, except the size of the currents, which was 35% smaller for negative displacements. The fact that we could fit identically shaped current-displacement relationships for positive and negative bundle displacements suggests that, as the hair bundles become progressively more disorganized during postnatal development, some tip links lose their normal orientation. This is likely to be a direct consequence of some stereocilia being misaligned away from the normal direction of bundle sensitivity, which is a feature of sv/sv OHCs from about P3 onwards ( Fig. 4; Self et al. 1999). The biophysical properties of immature sv/sv OHCs, including their zero-current potential, membrane capacitance, as well as the sizes of their delayed-rectifier potassium (I K,neo ), sodium (I Na ) and calcium (I Ca ) currents were otherwise similar to those of +/sv OHCs (Table 1) and indeed wild-type OHCs (Marcotti & Kros 1999).

MET currents of Snell's waltzer IHCs
To establish whether the lack of adaptation was a general feature of sv/sv cochlear hair cells we also conducted some experiments using IHCs. When stimulated with mechanical steps, heterozygous +/sv IHCs at P4 responded to force moving the hair bundles towards the kinocilium with rapidly activating inward currents that, at a holding potential of −81 mV, showed some time-dependent decline indicative of adaptation (Fig. 3A). Fluid force in the opposite direction closed the small fraction of the MET channels that were open when the bundle was in its resting position. When MET currents were elicited during depolarizing voltage steps to +99 mV the adaptive decline was no longer present and the currents instead showed a slow further increase following the initial rapid response (Fig. 3A). Inhibitory force stimuli showed that the current activated at rest was considerably increased compared to that at −81 mV, resulting in a leftward shift and change in the shape of the relationship between MET current and driver voltage to the fluid jet (Fig. 3C). Expressed as a fraction of the maximum, saturating MET current, the resting current increased from 3.4 ± 0.9% at −81 mV to 20 ± 3% at +99 mV (n = 3 IHCs). The maximum saturating MET current was −515 ± 90 pA at −81 mV and 912 ± 161 pA at +99 mV (n = 3). Although fewer data from early postnatal wild-type mouse IHCs are available for comparison than there are for OHCs (Kros et al. 1992;, the MET currents of the +/sv IHCs showed signs of adaptation and appeared normal in size.

. Mechano-electrical transduction by Snell's waltzer IHCs
A and B, MET currents recorded at −81 mV (left panels) and +99 mV (right panels) from control +/sv (A, P4) and sv/sv (B, P4) apical-coil IHCs elicited using 50 ms force-step stimuli (top panels). At −81 mV, +/sv IHCs had a small transducer current at rest and inhibitory bundle displacements turned this off (grey traces). At +99 mV, the resting MET current was increased. In sv/sv IHCs the resting MET current was absent at both membrane potentials (B). The recordings at −81 and +99 mV are from the same IHCs. Resting MET current in the absence of bundle stimulation is indicated by grey dashed lines. C and D, normalized peak MET current recorded from +/sv (C, P4, n = 3) and sv/sv (D, P4, n = 4) IHCs at the holding potential of −81 mV and during a step to +99 mV as a function of driver voltage. Data were fitted with eqn (1). In +/sv (C) at −81 mV a 1 = 0.41 V −1 , a 2 = 0.094 V −1 , x 1 and x 2 = 5.4 V; at +99 mV the parameters were as at −81 mV, except for x 1 = 1.9 V. In sv/sv (D) at positive driver voltages and at −81 mV a 1 = 0.34 V −1 , a 2 = 0.20 V −1 , x 1 = 8.2 V, x 2 = 18.0 V; at +99 mV the parameters were as at −81 mV, except for x 1 = 7.9 V and x 2 = 17.1 V. For negative bundle stimulation a 1,2 and x 1,2 were negative, but had the same absolute values as for positive driver voltage. The average saturating MET current in IHCs was: −515 ± 90 pA (+/sv) and −288 ± 35 pA (sv/sv) at −81 mV; +912 ± 161 pA (+/sv) and +457 ± 52 pA (sv/sv) at +99 mV. For negative driver voltages in sv/sv IHCs, the saturating MET current was about 0.23 of that for positive driver voltages.

J Physiol 594.13
The MET currents of the P4 sv/sv IHCs displayed features similar to those of the OHCs of the homozygous mutants: no time-dependent adaptation or resting MET current was evident at either membrane potential (Fig. 3B). The currents were also about half the size of those of the +/sv IHCs, with maximum currents reaching −288 ± 35 pA at −81 mV and 457 ± 52 pA at +99 mV (n = 4) (P < 0.05 for both potentials), suggesting the loss of some MET channels, while the slow onset of the currents seen in some cases points to less efficient gating of the remaining MET channels. As observed for the OHCs (Fig. 2D and E), inhibitory force steps elicited small MET currents with a similar activation range to those seen in response to excitatory force steps ( Fig. 3B  and D). The relationship between MET current and driver voltage was essentially unchanged upon depolarization (Fig. 3D). Basolateral currents of the sv/sv IHCs are likely to be normal during the first postnatal week, but fail to mature (Roux et al. 2009).

FM1-43 loading is reduced or absent in hair cells from homozygous mutant Snell's waltzer mice
The styryl dye FM1-43, a permeant blocker of the hair-cell MET channel (Gale et al. 2001), has been used to assess the presence of a resting MET current in hair cells. Bath application of FM1-43 resulted in the selective labelling of both IHCs and OHCs from +/sv mice, while dye loading in hair cells of sv/sv mice was strongly reduced or absent (Fig. 4), consistent with hair cells from homozygous mutant mice having little or no resting MET current (Figs 2E and 3D). Despite the disorganized hair bundles, sv/sv OHCs had tip links, just like those of +/sv controls (4E-G), in keeping with the features of their MET currents being distinct from those of the anomalous MET currents observed in the absence of tip links (Alagramam et al. 2011;Marcotti et al. 2014).

Calcium-dependent MET current adaptation is absent in sv/sv OHCs
The absence of the leftward shift of the current-displacement curve at depolarized potentials ( Fig. 2E) suggested that in sv/sv OHCs the MET channel might be insensitive to Ca 2+ modulation. To test this hypothesis directly, MET currents were recorded in both +/sv and sv/sv OHCs during superfusion of different concentrations of extracellular Ca 2+ (Fig. 5). MET current size varied inversely with extracellular Ca 2+ in both genotypes, due to Ca 2+ ions acting as permeant blockers (Howard et al. 1988;Ricci & Fettiplace 1998;Gale et al. 2001;Marcotti et al. 2005). In +/sv OHCs the fraction of the MET current activated at rest varied with extracellular Ca 2+ (Fig. 5A) but this was not seen in sv/sv OHCs (Fig. 5B), again indicative of a lack of Ca 2+ -dependent adaptation. The MET currents of the mutant cell of Fig. 5B did not saturate, hence their 'pointy' appearance, but they were not like the anomalous MET currents (Alagramam et al. 2011;Marcotti et al. 2014) in that they were activated, as normal, in response to force towards the kinocilium. The absence of Ca 2+ -dependent adaptation was confirmed when force steps were applied to the bundle during superfusion of 0.1 or 10 mM Ca 2+ . As shown in Fig. 5C and D, increasing Ca 2+ from 0.1 to 10 mM produced a substantial rightward shift of the current-driver voltage relationship in +/sv but not sv/sv OHCs, indicating that MET channels in the latter lack adaptation driven by Ca 2+ influx. Similar results were obtained in an additional six +/sv (P6-P7) and eight sv/sv (P5-P7) OHCs. We further tested whether Ca 2+ -dependent adaptation of the MET current in sv/sv hair cells was modulated by free intracellular Ca 2+ in the stereociliary bundle. To this end we changed the cell's Ca 2+ buffering capacity using different concentrations of the fast Ca 2+ buffer BAPTA in the intracellular solution and recorded MET currents in response to force steps (Fig. 6). In heterozygous control OHCs (+/sv) and in the presence of a low BAPTA concentration (0.1 mM), MET currents recorded at −81 mV can be seen to adapt during non-saturating bundle displacements (Fig. 6A, left panel). Upon stepping the membrane potential to +99 mV, time-dependent adaptation was no longer present and the resting MET current increased (Fig. 6A, right panel), as also observed in the presence of 1 mM EGTA as the intracellular Ca 2+ buffer (Fig. 1). Increasing the BAPTA concentration to 10 mM, the adaptive decline of the MET current at −81 mV was abolished and the resting open probability of the MET current increased to near 30% of its maximum value in +/sv OHCs (Fig. 6B). By contrast, all forms of Ca 2+ -dependent adaptation were absent in sv/sv OHCs irrespective of membrane potential or BAPTA concentration ( Fig. 6C and D). Membrane depolarization or increasing the intracellular BAPTA concentration from 0.1 to 10 mM produced a leftward shift in the relationship between MET current and bundle displacement in heterozygous control (+/sv: Fig. 6E) but not in sv/sv OHCs (Fig. 6F).
The above findings show that myosin VI is required for the acquisition of MET channel adaptation. Considering that one proposed role for myosin VI is to traffic molecules toward the minus ends of actin filaments (Sweeney & Houdusse, 2010) (i.e. away from the stereocilia and towards the hair cell's basolateral pole), we hypothesized that it may be involved in removing stereociliary components that, whilst necessary for early bundle development, hinder the acquisition of adaptation.

Persistence of transient lateral links correlates with lack of adaptation
CDH23 is a member of the cadherin superfamily of cell-cell adhesion molecules that has been proposed to form, in addition to the upper ends of tip links (Siemens et al. 2004;Kazmierczak et al. 2007), transient lateral links that interconnect stereocilia during early postnatal stages of development (Boëda et al. 2002;Goodyear et al. 2005;Michel et al. 2005). As such, we found that immunoreactivity to the ecto-domain of CDH23 was selectively associated with hair bundles in early postnatal hair cells. The labelling intensity observed with anti-CDH23 in +/sv mice gradually faded over the first 6 days of postnatal development (Fig. 7A). By contrast, the hair bundles of sv/sv mutant mice remained strongly immunoreactive for CDH23 throughout the immature stages studied, up to P6 (Fig. 7B). The hair bundles of mutant OHCs from the same apical region of the cochlea exhibited a significantly greater apparent overall steady-state bundle stiffness (sv/sv: P6, 5.7 ± 0.3 mN m −1 , n = 13) compared to littermate controls (+/sv: P6, 4.7 ± 0.3 mN m −1 , n = 10, P < 0.02), possibly due to the persistence of the lateral links and fusion of stereocilia starting from their bases.
To determine whether this failure to clear CDH23 from the hair bundle during early postnatal development was a specific feature of the Snell's waltzer mouse, we looked at CDH23 immunoreactivity in another mouse mutant with progressively disorganized hair bundles, the Shaker 6J mouse, which has a mutation in Myo7a affecting myosin VIIa expression (Kros et al. 2002). Homozygous Myo7a 6J mutant hair cells (6J/6J) showed a developmental pattern for CDH23 immunoreactivity similar to that of control cells (Fig. 7C), indicating that the persistence of CDH23 was specific to the absence of myosin VI and not due to general hair bundle deterioration.
At P2, when hair bundles express high levels of CDH23, MET current adaptation was not observed in either +/sv or sv/sv hair cells, as reported during normal development (Waguespack et al. 2007). By P4 adaptation was observed in the control +/sv OHCs, but not in the sv/sv mutants ( Fig. 7D and E) that retain CDH23 (Fig. 7B). The progressively deteriorating bundle morphology (Fig. 7B) was correlated with a progressive decline in MET current size (Fig. 7F). The current size of heterozygous control and mutant OHCs was significantly different (P < 0.0001, two-way ANOVA). Bonferroni post hoc tests showed that there was no difference at P2 but at P4, P6 and P7 currents in sv/sv OHCs were significantly smaller (P < 0.0001).

Discussion
We show here that in the absence of myosin VI the MET currents of mouse cochlear hair cells are abnormal. Homozygous Snell's waltzer mutant OHCs and IHCs showed altered hair bundle morphology, an absence of resting MET current and a lack of Ca 2+ -dependent adaptation. The postnatal decline in CDH23 immunoreactivity, which normally takes place concomitant with the loss of transient lateral links from the immature stereocilia, fails to occur in the Snell's waltzer mutants. Thus, our results identify myosin VI as a protein required for the normal maturation of the MET complex. The heterozygous +/sv hair cells, by contrast, had apparently normal MET current properties, including adaptation, size and reversal potential, like those reported in cochlear hair cells from wild-type mice studied under similar conditions in response to stimulation by force from a fluid jet (Kros et al. 1992;Géléoc et al. 1997;Corns et al. 2014;Marcotti et al. 2014). This finding is consistent with the normal pattern of bundle development seen in hair cells of +/sv mice and the normal gross cochlear potentials recorded from these mice (Self et al. 1999), as well as the normal hearing reported in human heterozygous carriers of DFNB37 mutations, which are, like sv, probably functional null alleles (Ahmed et al. 2003).

Cochlear hair cells without myosin VI lack MET current adaptation
Despite the progressively deteriorating hair-bundle structure, MET currents of normal response polarity could be recorded from sv/sv cochlear hair cells up to P7. This indicates that the tip links observed in these mutant hair cells ( Fig. 4F and G) are functional. The significantly smaller MET currents in the mutants compared to those of controls are likely to be related to a progressive reduction in the number of functional stereocilia that have not yet fused (Self et al. 1999). In mutant OHCs, the increased overall hair bundle stiffness together with the lack of a resting MET current would render them ill-suited to signal sound-evoked mechanical events in the mature cochlea (Self et al. 1999). An even larger shift in the channel's sensitivity to bundle displacement has been shown in hair cells carrying a mutation for another unconventional myosin, myosin VIIa (Kros et al. 2002;Marcotti et al. 2014). However, myosin VIIa mutant OHCs exhibited an apparent 'hyperadaptation' (Kros et al. 2002), which is also independent of Ca 2+ entry ) but different from myosin VI mutant cells where adaptation is abolished. Moreover, unlike myosin VI mutant hair cells, the MET channels of myosin VIIa mutant hair cells are not gated by tip links and respond predominantly to stimuli in the negative direction that move the bundle away from the kinocilium . Calcium entry through the MET channel exerts a major regulatory role over its adaptation properties (Eatock et al. 1987;Assad et al. 1989;Crawford et al. 1989;Ricci & Fettiplace, 1998;Corns et al. 2014). However, striking findings from our results were the absence in sv/sv hair cells of a decline of the MET currents during excitatory force steps (Figs 2A and 3B) and the lack of a leftward shift in the current-displacement relationship upon depolarization to near the Ca 2+ equilibrium potential (Figs 2E and 3D), dynamic and steady-state manifestations of Ca 2+ -dependent adaptation, respectively (e.g. Assad et al. 1989;Crawford et al. 1989;Corns et al. 2014). Therefore, in the absence of myosin VI, the intracellular Ca 2+ sensor at or near the MET channel that controls adaptation (Ricci & Fettiplace 1998;Corns et al. 2014) appears not to be functioning. This could be either because the sensor is absent or because it does not respond to changes in Ca 2+ concentration. In the presence of a functional Ca 2+ sensor, adaptation is not evident at low extracellular Ca 2+ (<0.1 mM) or strong intracellular Ca 2+ buffering with BAPTA, but different from the findings with the sv/sv hair cells, this results in a large fraction of the MET current being activated at rest in wild-type cochlear hair cells (Corns et al. , 2016 and +/sv OHCs too (Figs 5 and 6). The combination of a lack of MET current decline during stimulation by mechanical steps and no resting MET current has been reported to occur at the onset of mechanosensitivity at P2-P3 for apical-coil OHCs (Waguespack et al. 2007), suggesting that the normal developmental progression of the MET complex is stalled in the sv/sv OHCs. The slow activation of the MET current that we observed in some cells is also consistent with this lack of developmental progression (Waguespack et al. 2007;Chen et al. 2014). The acquisition of the MET channel's adaptation properties thus depends critically on myosin VI. Intriguingly, myosin XVa has been shown to be required for adaptation in IHCs but not OHCs (Stepanyan & Frolenkov 2009). However, the fact that myosin XVa is normally localized at the tips of the bundle (Belyantseva et al. 2003), as well as the abnormal tip links and the lack of effect of reducing extracellular Ca 2+ on MET current size in the IHCs lacking functional myosin XVa, suggest different causes for the lack of adaptation due to this mutation.

Mechanism of action of myosin VI
Myosin VI is unique among the known unconventional myosins as it can serve as an anchor as well as a processive motor with a reverse direction, i.e. towards the minus end of the actin filament (Wells et al. 1999;Sweeney & Houdusse 2010). Class VI myosin molecules are involved in several intracellular processes (Sweeney & Houdusse, 2010). In cochlear hair cells it has been suggested that myosin VI could be involved in anchoring their apical membrane to the underlying actin-rich cuticular plate (Self et al. 1999;Hertzano et al. 2008) and also in intracellular transport of synaptic vesicles and basolateral membrane proteins required for the functional maturation of IHCs at the onset of hearing (Heidrych et al. 2009;Roux et al. 2009). IHC maturation is influenced by Ca 2+ action potentials (Johnson et al. 2013), which in vivo depend on the resting MET current (Johnson et al. 2012). Therefore, the lack of maturation of the IHC basolateral membrane currents and synaptic machinery that has been observed in the absence of myosin VI (Roux et al. 2009) or other molecules that affect the transducer complex (TMC1: Marcotti et al. 2006;Kawashima et al. 2011;EPS8: Zampini et al. 2011) are probably an indirect effect caused by the absence of a resting MET current.
An important question is: why is myosin VI essential for MET current adaptation? CDH23, together with protocadherin 15 (PCDH15), forms the tip links connecting neighbouring stereocilia (Siemens et al. 2004;Ahmed et al. 2006;Kazmierczak et al. 2007) that gate the MET channels. CDH23 is also associated with lateral links that are transiently present in the immature hair bundles (Michel et al. 2005). In the absence of myosin VI, the CDH23 staining that is associated with the transient lateral links was retained during development. This indicates that in Snell's waltzer mutants, transient lateral links are not removed from the bundles during postnatal development as they are in controls, suggesting there are defects in stereociliary membrane remodelling in the mutant cells. The increased bundle stiffness of mutant OHCs is likely to be caused by the persistence of the lateral links which normally disappear towards the end of the first postnatal week , as well as the disorganized hair bundles and the fused stereocilia (Self et al. 1999). An increase in translational stiffness by itself cannot explain the lack of adaptation by the sv/sv mutant hair cells, as adaptation in mammalian cochlear hair cells is likely to be mediated by a Ca 2+ sensor on the MET channel itself modulating the channel's force sensitivity and not a mechanical process extrinsic to the channel involving unconventional myosins Fettiplace & Kim 2014).
The mechanisms that normally lead to the loss of CDH23 and transient lateral links from the maturing hair bundle are not yet known. Whilst these links may simply have a finite half-life, they might also be actively removed from the hair bundle. This could involve translocation of components such as CDH23 to the apical, non-stereociliary surface of the hair cell. As a minus-end directed motor positioned along the length of the stereocilia, myosin VI is ideally suited for removing hair bundle molecules that are no longer required from the stereocilia. Evidence that myosin VI normally does move down the stereocilia comes from the Tailchaser mouse, in which the mutated myosin VI in homozygous mutants is concentrated at the upper third of the stereocilia rather than evenly distributed along them (Hertzano et al. 2008). In the heterozygote Tailchaser, the hair bundles are less severely affected than those of sv/sv mice (Self et al. 1999;Hertzano et al. 2008) and hearing loss develops gradually (Kiernan et al. 1999). This is consistent with partial functionality of myosin VI in this dominant mutation (Pylypenko et al. 2015), sufficient to support at least the emergence of a resting MET current and thus allowing (inner) hair-cell maturation to proceed further than in the sv/sv mice. There is also evidence that myosin VI keeps two other hair-bundle proteins, CLIC5 and radixin, confined to the base of the bundle, because in the absence of functional myosin VI these proteins redistribute along the length of the stereocilia (Salles et al. 2014). The time course of disappearance of CDH23 ecto-domain immunoreactivity matches the gradual acquisition of MET current adaptation, the establishment of a resting MET current and the speeding up of the channels' activation kinetics (Waguespack et al. 2007;Lelli et al. 2009;Michalski et al. 2009;Chen et al. 2014). Even the composition of the MET channel itself is likely to change during this time window, with TMC1, a point mutation which affects the pore properties of the MET channel (Pan et al. 2013;Corns et al. 2016), gradually replacing TMC2 during the first postnatal week (Kawashima et al. 2011). We propose that myosin VI, by removing stereociliary elements such as CDH23 as a component of the transient lateral links (which are probably required for the integrity of the immature hair bundles), allows the hair bundle and its transduction apparatus to progress in their development, so that the MET channels acquire their physiological resting tension and Ca 2+ -dependent adaptation properties.