|Home | About | Journals | Submit | Contact Us | Français|
Mechanoelectrical transduction (MET), the conversion of mechanical stimuli into electrical signals operated by the sensory cells of the inner ear, enables hearing and balance perception. Crucial to this process are the tip-links, oblique fibrous filaments that interconnect the actin-filled stereocilia of different rows within the hair bundle, and mechanically gate MET channels. In a recent study, we observed a complete regression of stereocilia from the short and medium but not the tall row upon the disappearance of the tip-links caused by the loss of one of their components, cadherin-23, or of one of their anchoring proteins, sans, in the auditory organs of engineered mutant mice. This indicates the existence of a coupling between the MET and F-actin polymerization machineries at the tips of the short and medium stereocilia rows in cochlear hair bundles. Here, we first present our findings in the mutant mice, and then discuss the possible effects of the tip-link tension on stereocilia F-actin polymerization, acting either directly or through Ca2+-dependent mechanisms that involve the gating of MET channels.
Our ability to perceive sounds and maintain balance is based on mechanoelectrical transduction (MET), whereby the inner ear sensory cells (the hair cells) convert mechanical stimuli evoked by sounds or head movements into electrical signals that are transmitted to auditory or vestibular afferent neurons. The hair cells achieve this transduction process within the hair bundle, an array of mechanosensitive microvilli called stereocilia that protrude at their apical surface. The hair bundle operates as a mechanical antenna sensitive to vibrations as small as those driven by thermal noise, yet robust enough to withstand a million times larger sound intensities. In the mammalian auditory sensory organ (the cochlea), the hair bundles are organized into three rows of stereocilia of increasing heights, forming a staircase pattern. Each stereocilium is filled by parallel and uniformly polarized actin filaments (F-actin) that are cross-linked in a densely packed paracrystalline structure.1
The stereocilia actin core undergoes a continuous and slow turnover by which the hair bundle renews in approximately 48 hours in rat and mouse cochlear hair cells maintained in organotypic cultures at early postnatal stages.2,3 This renewing process involves the addition of actin monomers (G-actin) and cross-linking proteins that include espin and fimbrin/plastin4,5 at the stereocilia tips where the barbed ends of the filaments are located and the subtraction of actin monomers at their basal pointed ends. These actin polymerization and depolymerization processes result in a treadmill mechanism whereby the stereocilium actin core moves slowly rearwards while its length is maintained constant throughout time.
The tip-link, a single oblique fibrous link6,7 made of the Ca2+-dependent transmembrane proteins cadherin-23 and protocadherin-15 composing its upper and lower parts, respectively,8–10 extends from the tip of one stereocilium to the side of its taller neighbor. This link is continually kept under tension, possibly through the activities of myosin1c11 or myosin VIIa12–15 (Fig. 1). According to the gating spring model, the prevailing model for hair cell mechanotransduction, tip-link tension modulated by stereocilia deflection during sound stimulation controls the open probability of MET channels.16 These relatively nonselective cation channels (one or two per stereocilium17), of still unknown molecular composition, are located at the tip of each stereocilium from the short and medium rows,18 and are supposed to be directly or indirectly tethered to the tip-link lower part. The activation kinetics of the MET channels is extremely fast, in the microsecond range in mammals,19 indicating direct activation by tip-link tension. Two types of cells transduce sound in the cochlea, specifically, the inner hair cells (IHCs), which are the genuine sensory cells that release neurotransmitter and induce action potentials in the main afferent neurons (type I neurons), and the outer hair cells (OHCs), which locally amplify the sound-induced motion of the sensory epithelium.
We have recently shown that the putative scaffold protein sans, encoded by the gene defective in one of the genetic forms of Usher syndrome type 1, USH1G,20,21 and composed of four ankyrin repeats and a sterile α-motif domain,22 is an essential component of the MET machinery.23 These findings were based on the analysis of Ush1gfl/flMyo15-cre+/− mutant mice that undergo a postnatal deletion of the Ush1g gene in the inner ear hair cells. In these mutants, a loss of the tip-links first occurred in the IHCs, then in the OHCs, and reduced MET currents were recorded in cochlear explants (Fig. 2). The MET current kinetics and the adaptation process were, however, unchanged, suggesting that in the absence of sans, some MET channels are not functional while the others work properly. This proposal predicts the existence of mutant hair cells in which only one or a few MET channels persist that have normal kinetic properties. Here, we reanalyzed in that respect the electrophysiological recordings of Ush1gfl/flMyo15-cre+/− IHCs that we had previously obtained.23 Among the 35 mutant IHCs recorded at postnatal day 8 (P8) in the middle cochlear turn, where most of the tip-links were lost, 5 cells showed MET currents of minimal amplitude at stimulus onset. These currents have similar values from one mutant IHC to another whatever the amplitude of the hair bundle deflection (stimulus size), specifically, 5.4 ± 0.9 pA (mean ± standard deviation) at a holding potential of −80 mV in 1.5 mM external Ca2+ concentration. They display normal sub-millisecond activation kinetics, and have widely variable durations that increase with the amplitude of the hair bundle deflection (Fig. 3). During the maintenance of the stimulation, additional MET currents were seen to be activated after the stimulus onset, with amplitudes similar to those activated at stimulus onset. The frequency and duration of these delayed MET currents (related to the channels' open probability) increased, as expected, with the amplitude of hair bundle deflection (except for very large deflections at which one stimulus-locked MET current lasting all or the larger part of the stimulus duration was usually observed). Minimal MET currents thus have the typical characteristics of single MET channel currents.19 On the assumption of a linear current-voltage relationship and a reversal potential of +3 mV in normal perilymph,19 it follows that the unitary MET channel conductance of the mouse IHCs averages 67 ± 11 pS. It is about 2.5 times smaller than the rat IHC unitary MET channel conductance that was measured after destruction of most of the tip-links by a short exposure of the hair bundle to a submicromolar Ca2+ solution (specifically, 170 pS at 1.5 mM external Ca2+, a value that varied little with cochlear tonotopy).19 This suggests possible inter-species differences that call for a comparative analysis performed under the same experimental conditions. Notably, in 6 out of the 35 mutant IHCs analyzed, the average amplitude of the MET currents at stimulus onset was 11.8 ± 1.2 pA, indicating an average conductance of 120 ± 14 pS (for the same holding potential of −80 mV and in 1.5 mM external Ca2+). The simplest interpretation is that these currents were due to the activation of two persistent MET channels in these cells. Consistent with this view, the additional MET currents recorded during the maintenance of the mechanical stimulation displayed switches between amplitude levels of about 5.5 pA or 12 pA, likely reflecting the stochastic opening of one or two MET channels (Fig. 4).
We also studied Ush1g−/− mutant mice that lack sans at the early stages of hair bundle development and later. In these mutants, despite the disorganization of hair bundles, MET currents could be recorded, albeit of reduced amplitude. Therefore, we concluded that sans is not required for the formation of the tip-link,23 whereas it is thereafter necessary for its maintenance or renewal.
We found that sans is recruited to the plasma membrane in transfected COS-7 cells when either cadherin-23 or protocadherin-15 is co-expressed, implying that it can directly or indirectly interact with the cytoplasmic regions of the two proteins that make up the tip-link. Immunolabeling experiments in the mouse, using optical and electron microscopy, showed the presence of sans mainly at the tip-link lower insertion point up to P8.23 Based on the anomalies of the MET currents and the disappearance of the tip-links in Ush1gfl/flMyo15-cre+/− mice, as well as the immunolocalization of sans, we concluded that this protein is a critical component of the auditory MET machinery. Sans could anchor the lower end of the tip-links to the stereocilia tips. Another immunofluorescence study carried out in the rat, however showed the presence of sans at the the tip-link upper insertion point from P11 onwards,15 which suggests that the protein could undergo a switch of its location during the maturation of the MET machinery, and sequentially play similar anchoring roles at either end of the tip-link.
The cylindrical paracrystal F-actin core of the stereocilia contains a few hundreds of continuous filaments densely packed together1,24 and tightly enclosed by the cell membrane.25 The number of filaments decreases in the basal region of the stereocilium, where it tapers off, leaving just a few tens of filaments that penetrate into the underlying cuticular plate and form the stereocilia rootlets.1,26 The precision with which the staircase pattern of the hair bundle is maintained throughout the lifespan suggests that the dynamic renewal of the stereocilia actin cores is finely regulated.2 Remarkably, in vitro observations indicated that the actin treadmill velocity in stereocilia increases roughly in proportion with their height. As a result, all stereocilia in a given hair bundle are renewed approximately over the same time period.3 A surprising additional finding of our study was that the loss of tip-links in Ush1gfl/flMyo15-cre+/− mice is followed by a regression of stereocilia from the two shorter rows, leading to their complete disappearance in 24–48 h, while the stereocilia of the tall row remain unaltered. The same phenotype was observed in Cdh23fl/flMyo15-cre+/− mutant mice that undergo a postnatal deletion of the cadherin-23 gene in the hair cells. The regression of stereocilia only in the short and middle rows in Ush1gfl/flMyo15-cre+/− and Cdh23fl/flMyo15-cre+/− mice indicates the existence of a F-actin polymerization control mechanism which is common to all stereocilia except the tallest ones. We concluded that in the mutant mice, the F-actin renewal in the stereocilia from the short and middle rows of stereocilia is impeded by the loss of the tip-links and/or the resulting absence of gating of the MET channels. This indicates that the MET machinery plays a hitherto unsuspected role, direct or indirect, on the actin polymerization processes in these stereocilia rows, which is essential for the development and dynamic maintenance of the hair bundle's staircase pattern.
The rate of addition of G-actin monomers at the stereocilia tips (1–10 monomer per second per stereocilia in cochlear hair cells in vitro3) is more than 100 times slower than the rates (in the kHz range) at which MET operates during sound stimulation, and >104 times slower than the MET channel activation speed. It is therefore unlikely that coupling of the two machineries could involve the instantaneous activation kinetics of the MET channels. On the other hand the resting tension of the tip-links, which could be modulated in vivo over much longer time-scales,11 could provide the adequate control parameter for such a coupling. Based on our and earlier findings, we can suggest two distinct, non-exclusive mechanisms. One possibility is that changes in the tip-link tension, by modulating the resting open probability of the MET channels, impact on the F-actin polymerization rate in hair cell stereocilia by changing the local Ca2+ concentration at the tips of the middle and small rows. Indeed, Ca2+ ions are known to regulate the activities of several actin-binding proteins involved in the F-actin treadmill process, including gelsolin, myosin IIIa, espin, plastin/fimbrin located at the stereocilia tips and tropomyosin located at the stereocilia rootlets. Among them, myosin IIIa and espin-1 (both implicated in inherited deafness in humans27)28,29 seem to be the best candidates according to their biochemical properties. In vitro experiments suggest that espin-1 is transported by myosin IIIa at the stereocilia tips, where it controls the F-actin polymerization machinery.30 Since the activity of myosin IIIa (which has nine putative calmodulin-binding motives) is likely to be strongly dependent on Ca2+ ions, it is conceivable that the loss of the MET machinery would result in decreased amounts of espin at stereocilia tips, hence in a reduction of the stereocilia length. It is, however, unclear how such a mechanism could induce the complete disappearance of the short and medium stereocilia rows seen in Ush1gfl/fl Myo15-cre+/− mice, since in the espin-deficient jerker mice, the stereocilia from all three rows are shorter than in wild-type mice, but do not completely disappear.28
Alternatively, the tip-link tension could exert a direct mechanical control on the F-actin polymerization process in hair cells' stereocilia. Earlier observations have shown that the tip shape of stereocilia is influenced by the presence of the tip-link anchored at their apical plasma membrane. The tips of the stereocilia of the short and medium rows have typical prolate (pointed) shapes, and become rounded after disruption of the tip-links by exposure to BAPTA.3 Our study confirms that the presence and absence of an intact tip-link correlates with a pointed and a rounded tip shape, respectively. These observations and modeling work31 suggest that the “pulling effect” exerted by the tip-link tension on the stereocilia membrane affects actin polymerization and results in a prolate tip shape. In their model, Prost et al.31 estimate the force exerted by the polymerizing front of the stereocilium actin core on the opposing membrane to be ≈ 500 pN, or 1–2 pN per actin filament. Each actin monomer being ≈ 6 nm in diameter, the minimum energy required to counterbalance the work of membrane tension upon addition of one monomer is just above the thermal energy kBT. The tip-link tension itself is estimated in the 20–40 pN range,11,31 corresponding to an available work of 20–40kBT during the addition of an actin monomer. This makes a control by the tip-link tension on the actin polymerization rate at stereocilia tips plausible from a purely energetic point of view. One possible mechanism for such a control involves the Brownian ratchet model, which describes the passive growth of a bundle of actin filaments against an opposing membrane.32,33 In this model, the addition of actin monomers to the filaments' barbed ends is allowed by thermal fluctuations of the membrane that occasionally produce gaps larger than the monomer size between the tips of the filaments and the membrane. The growing F-actin bundle exerts a force on the opposing membrane that increases with the rate of actin monomer addition, and reciprocally membrane tension affects the polymerization rate at the bundle tip. Brownian-ratchet mechanisms are thought to drive the growth of many F-actin protrusions, including lamellipodia, filopodia and microvilli.25,33 If such a mechanism plays a role in stereocilia, the tip-link tension would effectively reduce membrane tension at the stereocilia tips, and one would indeed predict a decrease of membrane fluctuations, hence of the F-actin polymerization rate, upon decrease of the tip-link tension. Additional knowledge about the F-actin polymerization molecular machinery of stereocilia needs to be uncovered, and explicit quantitative models developed, to clarify the extent to which such a mechanical influence can be exerted by the tip-link tension on the stereocilia tip complex, and whether this could account for our observations. It is worthy of note that the key molecular players of F-actin polymerization and assembly in stereocilia still escape characterization with the exception of some actin capping and actin-bundling proteins.4,5,25,27,34–36
From the study summarized here, we conclude that the tip-link, which is central to the MET process, also plays a key role in the staircase organization of the hair bundle. When the F-actin polymerization machinery of stereocilia becomes coupled to the MET machinery during hair bundle development, and how this coupling is achieved, remain to be determined. Our previous data suggest that the tip-link may operate this function as soon as it forms.37 Reversing our argument, some proteins previously reported to be involved in the control of the stereocilia length38 may also be components of the MET machinery. It is even tempting to speculate that the MET process in the sensory organs of the vertebrate octavolateralis system (the neuromasts of the lateral line, the auditory and balance organs of the inner ear) has emerged during evolution by exploiting a pre-existing mechanosensitive F-actin polymerization machinery. The hair bundle maintained in tension via the tip-links would then appear as a specialized example of a “tensegrity” structure39,40 evolved to operate mechanotransduction with unique speed and effectiveness.
We thank Jacqueline Levilliers and Jean-Pierre Hardelin for their help in the preparation of the manuscript, Andrea Lelli for comments, Thibault Lagache for discussions, and Serge Picaud for useful suggestions in the analysis of our electrophysiological data. This work has been supported by FAUN Stiftung (Suchert Foundation), Fondation Raymonde and Guy Strittmatter, LHW-Stiftung, the Conny Maeva Charitable Foundation, Fondation Orange, and the Louis-Jeantet Foundation.