Stereocilin in top-connectors is a key element ensuring waveform distortion and suppressive masking, necessary for speech intelligibility and hearing in noise
Notes de la rédaction
This work has been published in the following article:
Verpy E, Weil D, Leibovici M, Goodyear RJ, Hamard G, Houdon C, Lefèvre GM, Hardelin J-P, Richardson GP, Avan P & Petit C (2008) Stereocilin-deficient mice reveal the origin of cochlear waveform distortions Nature, 456, 255-8.
1Several key properties of sound perception rest upon the pre-processing of sound by outer hair cells (OHC) in the mammalian inner ear, that is, one stage ahead of the mechanoelectrical transduction eventually achieved by inner hair cells. It has been acknowledged for two decades that OHCs mechanically amplify sound stimuli, which explains why the auditory system of mammals is sensitive enough to detect sound power levels hardly an order of magnitude above thermal noise. The fine tonotopy observed in the cochlea and reflected in the remarkable ability to discriminate two sounds with slightly different pitches, also results from OHC operation as it acts in a frequency selective manner.
2Natural sounds pose an additional challenge: several frequency components are presented simultaneously instead of sequentially. For example vowels are characterized by a spectrum displaying a few specific harmonics. Spectral complexity increases in the presence of competing sound sources or background acoustic noise. In such cases, if applied indiscriminately to all spectral lines, gain would be inadequate because, acting equally on signal and noise, it would leave the latter swamp neural messages. Because the gain produced by OHCs is accompanied by filtering, but also because the nonlinearities it entails generate suppressive masking interactions, acoustic messages can be cleaned up.
3The place of cochlear nonlinearities in the analysis of frequency mixtures deserves to be specifically examined. The concept of nonlinearity is very general, applying to any system whose response to two simultaneously presented signals is not the arithmetic sum of its responses to either signal when presented alone: instead, when mixed up, some components increase at the expense of others. Masking is a typically nonlinear psychophysical event defined by the fact that the loudness of one sound decreases or even vanishes when another sound interferes. Its cochlear correlate is suppressive masking whereby the mechanical or electrical response to a test tone decreases in the presence of a masking tone. This phenomenon, felt as a nuisance when it is the signal of interest that gets masked, globally turns as an advantage in that it allows the dominant frequency component at one place in the cochlea to become even more dominant by exerting a masking effect on competing, weaker signals. Therefore, suppressive masking can enhance contrasts.
4There is now no doubt that cochlear mechanics is far from linear and it can express its nonlinearities in several ways. Besides suppressive masking, another example is that contrary to high-fidelity devices, OHC operation introduces conspicuous waveform distortions. These distortions are large enough to be heard although not being present in the initial sound stimulus (e.g., Tartini, 1754; Goldstein, 1968). In response to bitonal stimuli at frequencies f1 and f2, distortion of their waveforms generates combination tones at arithmetic combinations of f1 and f2 -hence the best known cubic difference tone at 2f1 – f2, assuming f2 > f1. Not only does the cochlea produce audible sound distortion, but it also reemits them in the external auditory meatus as one category of otoacoustic emissions, namely distortion-product otoacoustic emissions (DPOAEs – Kim et al., 1980). Otoacoustic emissions have become a prominent tool for screening hearing in neonates by way of a small microphone placed in the ear canal: when their absence signals OHC dysfunction, inner hair cells also happen to be impaired in most cases owing to the structural and functional kinship of the two types of sensory cells. Sensorineural deafness is then a likely diagnosis.
5In summary, there is strong evidence that high sensitivity and sharp frequency discrimination are two closely associated properties of the cochlea ensured by OHCs, and that their way of operating induces strong waveform distortions coming out as non-invasively detectable DPOAEs. Last, the very mechanism that leads to instantaneous distortion of sound waveforms is likely strong enough to also be able to contribute to suppressive masking. Although there is no doubt that OHCs and their nonlinear behaviour are at the heart of many perceptive phenomena, this does not prove that all nonlinearities produced by OHCs must have the same origin, that could be identified, for example, as some molecule or substructure of the stereocilia bundle or cell body. Indeed, although strongly nonlinear, the compressive circuits in current use in hearing aids are designed so as to never generate instantaneous waveform distortion. Conversely, other types of nonlinear electroacoustic devices generate waveform clipping even if the gain of their amplifier is turned off.
6Until now the currently accepted picture was a holistic one, positing that at the core of OHC ability to produce gain, filtering, waveform distortion and masking was a common source, the intrinsic properties of the mechanotransduction channels in the membrane of stereocilia on top of OHCs and inner hair cells. These ion channels are opened and shut by stereocilia deflections due to sound vibrations, and when open, let K+ et Ca2+ ions enter into the cells. As a result, the modulations of the membrane potential of OHCs provide an instantaneous image of sound pressure changes. Electromotility, regardless of its exact mechanism, allows OHCs to mechanically react to their fluctuating membrane potential. By adding their motion to that of the acoustic wave at an appropriate time of the sound period, OHCs amplify the stimulation of their own stereocilia bundle according to Gold’s proposal of a regenerative amplifier (1948). Since such an amplifier acts in a frequency-selective manner, electromotility in response to mechanotransduction channel activation provides a common background to gain and filtering properties afforded to the cochlea by OHCs.
7A common explanation to the remaining characteristics of masking and waveform distortions might be inherent to the mandatory nonlinearity associated with the thermodynamics of the mechanotransduction channel. Acting as a gate, this channel exists in at least two states, open and closed. Its opening probability relates to stereocilia deflection according to Boltzmann’s law accounting for the different energies associated with the opened and closed states. Boltzmann’s law is a sigmoid instead of a straight line. Thus when stereocilia bundles are deflected by the sinusoidal pressure wave of a pure tone arriving from the ear canal, the current through mechanotransduction channels, proportional to the opening probability, is a flattened, distorted sinusoid: the ensuing electromechanical feedback injects distortion into the initially sinusoidal sound wave. It was thought that waveform distortion, Tartini tones and DPOAEs were produced in this manner by OHCs. Suppressive masking comes with the presence of waveform distortion regardless of their origin, since when mechanical responses of OHCs to a first sound saturate and tend to get clipped, the responses to a second, superimposed tone are all the more distorted that the responses to the first tone alone were already saturated. The presence of the first tone thus negatively influences –suppresses- the response to the second tone. Masking is reciprocal, yet the stronger of the two tones is bound to win the competition.
8This view of mechanotransduction channel properties as a central player in all aspects of sound pre-processing by OHCs suggested that OHCs ensured, in a remarkably parsimonious manner, a whole set of functions sharing a common origin. As a counterpart, failure of this intrinsic property of channels should also result in the concomitant loss of all beneficial aspects of cochlear pre-processing of sound.
9A recent study of a mutant strain of mice in which the gene coding for stereocilin is inactivated has shown that this holistic view is not valid. When these mutant mice are young enough (around 14-15 postnatal days, P14-15), their cochlear sensitivity is normal, despite some degree of neural immaturity also seen in wild-type littermates. Cochlear filtering is also normal in mutant mice. Mechanoelectrical transduction currents derived from indirect in vivo measurements of OHC activity are normal as well. These characteristics indicate the presence of a full supply of normally functioning mechanotransduction channels. Their thermodynamics thus obeys a normal Boltzmann law and the curve relating the transduction current to stereocilia deflection must be the same sigmoid as in normal ears. Yet in the absence of stereocilin, mice no longer distort waveforms, and for example their mechanotransduction currents in response to loud tones remain sinusoidal up to 100 dB SPL. Likewise, DPOAEs are totally absent, whereas at P14, they are already adult-like in wild-type mice. Furthermore, with even more significant perceptive consequences, when these mutant mice are exposed to a mixture of sounds, suppressive masking is absent or strongly diminished. The level of a masking tone must be about 20 dB louder than in a normal ear to produce a measurable effect. Although some masking persists at the neural level, it is weak. Therefore, in the presence of a mixture of sounds, the mutant cochlea is no longer able to significantly act on the contrasts among components.
10Stereocilin enters in the composition of fibrous links, the top connectors, bonding the apexes of stereocilia inside the hair bundle. In mutant mice, top connectors are absent and the tips of stereocilia in OHCs tend to splay apart, so that the stereocilia bundle has lost some of its normal cohesion.
11So, suppressive masking and waveform distortion come with each other and can vanish even though OHC mechanotransduction channels provide normal amplification and filtering. This unusual experimental situation leads to conclude that the top connectors, and possibly the stereocilin-dependent connectors anchoring the stereocilia bundle to the tectorial membrane contribute to a major cause of distortion, larger than that in relation to the Boltzmann statistics of mechanotransduction channels. Stereocilin-dependent connectors could distort either as a result of an intrinsic property or indirectly by a constraint they might exert on the displacement of the stereocilia bundle or on the response to sound of some of its components.
12In the model that we propose, in mutant mice as well as in normal ones, the operation of OHC mechanotransduction channels follows a displacement-to-current characteristics exhibiting a normal sigmoid shape because its becoming straighter would make the channel strongly insensitive, which was not observed. Likely, this nonlinearity, on its own, is not large enough to generate measurable distortion. In normal mice, it is the presence of top connectors that enables waveform distortions, DPOAEs and suppressive masking to show up in standard measurements. In mutant mice, the same measurements fail to detect these features even though mechanotransduction channels do exhibit their usual Boltzmann-related nonlinearity.
13The next step of our research will be to directly measure the operating functions of the mechanotransduction channels of OHCs at mature stages of the organ of Corti, which requires technological development. It would be important to establish whether and how top connectors might regulate the way OHC operate and contribute to modulating the intensity of their distortions and of suppressive masking. Nonetheless, the dissociation between normal auditory thresholds and missing DPOAEs should lead to reassess the significance of the clinical test of DPOAE recording.
14Perception of complex sounds and hearing in noise are likely perturbed in mutant mice. If one extrapolates this situation to man, one can anticipate that speech intelligibility be more affected than what is predicted from the elevation of hearing thresholds in patients whose deafness is due to lack of stereocilin (deafness DFNB16). The study of its murine model has stressed the importance of waveform distortion and suppressive masking, two sides of a nonlinearity involved in contrast enhancement, that the lack of stereocilin specifically targets. Such contrasts are an essential feature of sensory perception, for vision as well as audition, since in natural environments sensory messages are almost always mixed with competing signals.
Pour citer cet article
The Letter of the Collège de France, Collège de France, Paris, juin 2009, p. 45-47. ISSN 1958-1408
Christine Petit et Paul Avan, « Stereocilin in top-connectors is a key element ensuring waveform distortion and suppressive masking, necessary for speech intelligibility and hearing in noise », La lettre du Collège de France [En ligne], 4 | 2008-2009, mis en ligne le 15 novembre 2010, consulté le 24 avril 2017. URL : http://lettre-cdf.revues.org/775Haut de page
© Collège de FranceHaut de page