More whistles and bells for fly hearing

Amplification of sensory input is a common theme for almost all sensory modalities. Once a sensory signal is received, be it a photon or a molecule of saccharin, the receptor cell will append an amplifying second-messenger cascade onto the initial transduction event. Amplification of sound by hair cells within the cochlea also follows this rule, but with an important caveat: the hair cells produce no second-messenger cascade for the amplification (1); instead, they pump kinetic energy back into the hearing organ to magnify the sound (2).

How this energy is put back into the system is somewhat controversial. Considerable evidence indicates that the outer rows of hair cells contract and expand their somas in response to plasma membrane voltage changes, thereby contributing mechanical energy to the already vibrating basilar membrane on which they reside (3–5). But there is also mounting evidence that another portion of the hair cell is not passively along for the ride on a bucking basilar membrane. The organelle that receives the incoming stimulus, the hair bundle, also seems to play an active role. Vestibular hair bundles can amplify mechanical stimuli (6) by harnessing the energy released during opening and closing of the mechanically activated transduction channel (7). It remains to be seen whether the hair bundles in the cochlea also contribute to the motions of the basilar membrane, but it is clear that these organelles can serve as a source of amplification. Hearing organs appear to be unique among sensory organs in their ability to generate signals that they were meant to detect. In a surprising report in a recent issue of PNAS (8), the humble fruit fly's hearing apparatus appears to be no exception to this tenet: the sensory neurons that detect sound also are capable of moving the insect's hearing organ.

Using laser Doppler vibrometry to monitor the sound-induced movement of the fly's third antennal segment (an auditory receptor structure, functionally equivalent to an eardrum), Göpfert and Robert (8) have shown that Drosophila hearing organs exhibit a nonlinear response to sound of varying intensities, a trait characteristic of vertebrate hearing organs (9). In particular, the fly's organ becomes more compliant in response to lower intensity sound, but it does this in an interesting way. As the stimulus intensity declined, the resonant frequency of the organ also declined. These shifts in the response behavior suggest the presence of a component in the hearing organ that can change its stiffness based on stimulus intensity. In vertebrate hearing organs this behavior is termed compressive nonlinearity, in which the amplification by hair cells of basilar membrane movement decreases as the sound pressure level increases.

A clue about the origin of this nonlinear behavior came when the flies were treated with a pulse of CO₂. The intensity-dependent shift in resonant frequency was abolished within a few seconds, then slowly returned over about a minute. This treatment revealed the requirement of a metabolically active component, likely the mechanosensory neuron, underlying the compliance. This decrease in stiffness of the system could result from a passive change in the mechanical properties of some linkage in the system or from energy being pumped back into the hearing organ.

One indication of energy input is the presence of “active” phenomena like spontaneous movements. When the authors simply monitored the antennae in the absence of sound, they observed spontaneous excursions. The waveform of the movement was not simply sinusoidal, but showed a twitching behavior in which quick displacements (in both directions) were interspersed with relatively quiescent periods. Because the antennae moved quickly in both directions, something in the system must generate force in either orientation. Spontaneously oscillating bullfrog hair bundles show a similar movement (6). This superficial resemblance may belie a more fundamental similarity in the behavior of the transduction channels. In oscillating hair bundles the excursions occur when the 100 or so transduction channels are synchronously transiting between open and closed states, pulling the bundle in opposite directions with each swing of the channel's gate (10). It is compelling to imagine that the transduction channels in the fly's hearing organ might be similarly affecting the mechanical behavior of the antenna.

These spontaneous oscillations are rendered more extreme and more complex by treatment of the animals with dimethyl sulfoxide (DMSO), an anesthetic that can change the excitability of invertebrate neurons (11). Although it is unclear how DMSO alters the mechanosensory organs, the antennae simultaneously oscillated at several distinct increasing frequencies; almost as if individual “tuned” components of the hearing organ were gradually unmasked. Frequency-specific elements (yet another characteristic of vertebrate hearing) allow the deconvolution of complex waveforms into their component frequencies.

Hearing organs are unique in their ability to generate signals that they were meant to detect.

From where do these nonlinear shenanigans arise? The authors addressed this question by recording from mutants flies that have hearing deficits (12, 13). One mutant, nompA, is defective in a protein that attaches sensory neurons to their receptive cuticular structures, causing a conductive hearing loss in these flies (14). These flies showed none of the nonlinear behaviors and showed spontaneous oscillations similar to “dead” organs even after injection with DMSO. In accordance with a loss of mechanical linkages, the relative stiffness of the organ declined relative to the controls; that is, the antenna became sloppier. By disconnecting the neuron from the transduction pathway, the authors were able to infer that the neuron played a critical role in the both the nonlinear behavior of the system and its active oscillations.

The authors shored up the notion that the neurons were behind the nonlinear behavior by examining three additional mutants that have defects in their mechanosensory neurons per se. A transduction channel mutant, nompC, which hears about half as well as control flies (12), showed a concordant loss of nonlinear behavior and reduced oscillatory movements, although DMSO was able to change the resonant frequencies. Loss of NompC channels attenuated these interesting behaviors but did not abolish them altogether, implying the presence of an additional transduction channel partner or another active process altogether. This additional channel might also be conducting the residual transduction current seen in mechanosensory bristle recordings from nompC null flies (15). The NompC channel clearly contributes to the nonlinear and oscillatory behavior of the mechanosensory neuron, but NompC's gating cannot be the sole driving force behind the active processes observed in these hearing organs.

Two other deaf mechanosensory mutants, btv and tilB (12), also showed a complete loss of nonlinear phenomena and DMSO-insensitive, small oscillations. Both of these genes affect ciliary structures that lie at the heart of the mechanosensory dendrites. These two genes further winnow the location of the hearing organ's mechanical activity to the ciliated dendrites of the chordotonal organs that line the second antennal segment and serve as the fly's inner ear. Because tilB sperm are immotile (12), an intriguing association between ciliary motility and active processes in mechanosensory neurons arises. Like hair cells, perhaps flies have seen fit to couple their transduction channels to a molecular motor, in this case a microtubule-based motor, with which to regulate the tension the channels experience. Although the cilia in these mechanosensory neurons have no central microtubules and are therefore thought to be immotile, they could certainly serve as a substrate for a microtubule motor.

The list of improbable similarities between vertebrate hair cells and Drosophila mechanoreceptors continues to expand. In addition to several developmental paradigms and physiological properties that hint at a conserved relationship (16), a nonlinear active process to amplify incoming sound as well as spontaneous acoustic emissions are now characteristics that appear to be intrinsic to the process of sensing a mechanical stimulus like sound. As scientists, we should not necessarily be shocked by the similarities in the hearing of such disparate creatures as dipterans and mammals, but we should certainly be awed.