Hair Cells: Sensory Transduction 1015

Hair Cells: Sensory Transduction

G S G Geleoc and J R Holt, University of Virginia,Charlottesville, VA, USA

ã 2009 Elsevier Ltd. All rights reserved.

Structure of Sensory Epithelia

In each of the organs of the vertebrate inner ear, haircells occur in specialized patches or strips of sensoryepithelium, which generally contain just hair cells andsupporting cells. Because the organs are formed dur-ing development by the infolding of an otic cyst,derived from surface ectoderm, the apical surfacesof hair cells face a closed compartment. Specializedepithelial cells supply this chamber with a uniquefluid, termed ‘endolymph,’ which has a high potassiumconcentration and low sodium and calcium concentra-tions. Tight junctions ringing the apical surfaces of haircells and supporting cells separate the endolymph fromperilymph, a fluid similar to normal saline that bathesthe basolateral surfaces of hair cells. In the mammaliancochlea, proton pumps in epithelial cells of the striavascularis raise the potential of endolymph to aboutþ 80mV; elsewhere, however, the endolymph poten-tial is within a few millivolts of zero.

Structure of the Hair Bundle

The mechanosensitive hair bundle comprises a fewtens to a few hundreds of modified microvilli, termed‘stereocilia,’ arranged in multiple rows of increasingheight (Figures 1(a) and 1(b)). Adjacent stereociliaadhere to each other by means of interciliary links(Figures 1(c)–1(e)), so that when the bundles aredeflected the stereocilia do not separate, but insteadslide past one another: ankle links connect the stereo-cilia bases and lateral links couple the shafts of thestereocilia (Figure 1(c)), and tip links extend from theapical tip of a stereocilium to the side of its tallestneighbor (Figures 1(d) and 1(e)). The stereocilia havean internal structure similar to microvilli: they arebundles of several hundred actin filaments, tightlycross-linked by actin-bundling proteins espin andfimbrin. While ranging in length from about 1 mm(in mammalian cochlea) to over 60mm (in semicircu-lar canals), stereocilia have such stiff actin cores thatthey do not normally bend. The cores taper at theirbases to a few actin filaments, which anchor thestereocilia into a cuticular plate just beneath the api-cal surface (Figure 1(f)). Since stereocilia have rigidcores and taper at their bases, they do not bend butinstead pivot around their insertion point. Many hair

bundles also contain a single, eccentrically placed truecilium, the kinocilium, which, in mature cells, has the9þ 2 arrangement of microtubules characteristic ofmotile cilia (Figure 1(g)). The kinociliummay serve toorganize the bundle during development (it is lost inmature hair bundles of the mammalian cochlea), andit couples the stereocilia to accessory structures inmost other organs.

Axis of Polarity

The staircaselike variation in heights of stereocilia(Figures 1(a) and 1(b)) and the eccentric placement ofthe kinocilium at one edge of the bundle (Figure 1(g))confer a morphological axis of polarity to the bundle.The positive direction is defined as toward the kinoci-lium or the tallest stereocilia (Figure 2(a)); negative istoward the shortest stereocilia. This also describes thephysiological axis of the bundle: a positive deflectionof the stereocilia evokes an excitatory response. Itincreases the membrane conductance by triggeringthe opening of nonselective cation channels (transduc-tion channels), allowing positively charged ions to flowinto the cell and thereby depolarize the cell. At thebundle’s resting position, �10% of the transductionconductance is active (Figure 2(b)). Thus, a negativedeflection decreases the membrane conductance(closes channels open at rest) and hyperpolarizes thecell. This feature allows hair bundles to signal stimuliof either polarity.

The sensitivity of the hair bundle to mechanicalstimuli is remarkable: a deflection of a few tenths ofa micrometer in the positive direction is sufficient toactivate the conductance completely, and deflectionsat the auditory threshold are estimated to be a fewtenths of a nanometer. Side-to-side deflections of thebundle, even several micrometers, cause no conduc-tance change, so that the cell has a vector sensitivityand only senses the component of the stimulus alongthe bundle’s axis of polarity.

The Mechanism of Transduction

The past quarter century has provided a good under-standing of the mechanoelectrical transduction pro-cess, whereby deflections of the stereocilia cause theopening of transduction channels. The process isextremely fast: at mammalian temperatures, channelsbegin to open within about 10ms after deflection ofthe bundle. This can explain a sensitivity for frequen-cies as high as 100 kHz in some animals, but the speedalso rules out transduction mechanisms involving an

Figure 1 The structure of the vertebrate hair bundle. (a) Scanning electron micrograph of a hair bundle protruding from the apical

surface of a mouse utricle hair cell. Utricular hair bundles are composed of�65 stereocilia of graded heights and a single taller kinocilium.

(b) Scanning electronmicrograph of a guinea pig cochlea hair bundle.While a kinocilium is present in immature cochlear hair cells, it is lost

after �postnatal day 3. (c) Transmission electron micrograph of a chick utricle hair bundle. Just above the cuticular plate, the stereocilia

are interconnected by ankle links (large arrowheads). Shaft connectors (small arrowheads) link the lateral aspects of the stereocilia. (d) At

the tip of each stereocilium are tip links (long arrows) and horizontal top connectors (short arrows). (e) High-resolution freeze-etch image

of a tip link from a guinea pig cochlea hair bundle. (f) Transmission electron micrograph of a longitudinal section of the apex of a hair cell

from the saccule of a goldfish. The stereocilia taper at their bases and contain a large number of tightly packed actin filaments that

converge into a dense bundle where they insert into the cuticular plate. (g) Thin cross-section of a hair bundle from the bullfrog saccule,

which reveals the hexagonal arrangement of stereocilia with cores of actin and the single, microtubule-based kinocilium at the right.

Scale bar ¼ 5 mm (a), 1 mm (b), 200 nm (c, d, f), 100 nm (e), 500 nm (g). (a) Reproduced from Holt JR, Gillespie SKH, Provance DW, et al.

(2002) A chemical strategy implicates myosin-k in adaptation by hair cells.Cell 108: 371–381, with permission of Elsevier. (b) Reproduced

from Furness DN and Hackney CM (2006) Springer Handbook of Auditory Research: Vertebrate Hair Cells. Heidelberg: Springer, with

permission of Springer-Verlag, Heidelberg. (c, d) Reproduced from Bashtanov ME, Goodyear RJ, Richardson GP, et al. (2004) The

mechanical properties of chick (Gallus domesticus) sensory hair bundles: Relative contributions of structures to calcium chelation and

subtilisn treatment. Journal of Physiology 559(1): 287–299, with permission of Blackwell Publishing. (e) Adapted from Kachar B,

Parakkal M, Kurc M, Zhao Y, and Gillespie PG (2000) High-resolution structure of hair-cell tip links. Proceedings of the National

Academy of Sciences of the United States of America 97: 13336–13341, with permission from PNAS. Copyright (2000) National Academy

of Science, USA. (f) Reproduced from Fawcett DW (1981) The Cell, Philadelphia: WB Saunders, with permission of Elsevier.

(g) Reproduced from Hudspeth AJ (1983) The hair cells of the inner ear. Scientific American 248(January 1983): 54–64, with permission

of Dr. AJ Hudspeth and Scientific American.

1016 Hair Cells: Sensory Transduction

enzyme cascade or second messenger molecules. Itwas proposed, instead, that deflections exert a forcedirectly on transduction channels to open them, muchas transmembrane voltage exerts a force directly oncharged domains of voltage-gated channels. Severalexperiments have succeeded in measuring smallmovements of hair bundles that occur with channelopening. These experiments provide strong supportfor the idea of directly gated channels, as well asestimates of some of the mechanical properties ofthe transduction apparatus. The mechanical forcerequired to open a single transduction channel isestimated to be about 2 pN and upon opening, thechannel protein undergoes a conformational changeof �2 nm. Mechanosensitive transduction channels

represent a third class of ion channel, in addi-tion to those gated by voltage and ligand binding.A variety of mechanically gated channels have beenidentified in other systems. Properties of the hair celltransduction channel resemble those of ion chan-nels of the transient receptor potential (TRP) fam-ily, inparticular TRPN1 (also known as NOMPC,absent from the mammalian genome) present in theDrosophila bristle. Recently, TRPA1 was presentedas a candidate for the mechanosensitive channel ofvertebrate hair cells, but lack of an auditory or vesti-bular deficit in TRPA1 knockout mice has called thatsuggestion into question. Thus, conclusive evidenceof the identity of the hair cell transduction channelawaits.

1.0

1.0

0

P open

Bundle displacement (x)a b

c

Popen

0.5 µm

1.0

0 ms100

0

Reclosuremodel

Negativemovement

Releasemodel

Positivemovement

Excitatorystimulus

Tip link

Transductionchannel

Adaptation motor

Fast adaptation

Slow adaptation

+−

0.1

Figure 2 Models for hair cell transduction and adaptation. (a) Deflection of the hair bundle toward the tallest stereocilia opens

mechanosensitive transduction channels. (b) A 0.5mm step deflection rapidly opens channels. Over the subsequent 100ms the open

probability declines or adapts to a new steady-state level. The stimulus response relationship (right) shifts in the direction of the applied

stimulus. In the example of a positive stimulus the curve shifts to the right (dashed line). The relationship is best described by a second-

order Boltzmann function. (c) The prevailing model for adaptation posits that following an excitatory stimulus calcium enters open

transduction channels and promotes a decline in open probability on two timescales. Fast adaptation occurs within the first few

milliseconds and may result in change in the relationship between tension and open probability such that the channels are forced closed

(reclosure model). This force is predicted to generate a negative movement in bundle position. Alternatively, fast adaptation may result

from a release of tension within the transduction complex, which allows channels to close and the bundle to move farther in the positive

direction (release model). Slow adaptation (not shown) follows fast adaptation and results from slipping of the adaptation motor down the

side of the taller stereocilium, which also decreases tension and allows the channels to close. (b) Adapted from Pickles JO and Corey DP

(1992) Mechanoelectrical transduction by hair cells. Trends in Neuroscience 15: 254–259.

Hair Cells: Sensory Transduction 1017

Hair cell transduction channels are not particularlyselective, passing all of the alkali cations, many diva-lent cations (including calcium), and small organiccations up to about 0.7 nm in diameter. Thus, thereversal potential is close to 0mV. Taken togetherwith a single-channel conductance of 90–100 pS andestimates of up to 200 channels per cell (1 or 5 perstereocilium), deflections of the hair bundle can evoke

transduction currents of up to �1 nA. Several inde-pendent experiments, including measurement ofextracellular current and focal application of channelblockers, have localized transduction channels to thetips of stereocilia. This was confirmed by imagingcalcium that flows into the cell through the transduc-tion channels. Intracellular calcium was seen to risefirst at the tips of the stereocilia and then more slowly

1018 Hair Cells: Sensory Transduction

along their lengths. Furthermore, these experimentsconcluded that the channels could be found at eitherend and probably at both ends of tip links.Hair cell transduction channels are opened by force

applied to elastic elements, termed ‘gating-springs,’that are stretched by positive deflections (toward thetallest stereocilia) of the hair bundles. Tip links seemideally positioned to be gating-springs because theyrun parallel to the staircase arrangement of the hairbundle and join the top of one stereocilium to the sideof its adjacent taller neighbor along the axis of polarity.The discovery that breaking tip links, by treatmentwith low concentrations of extracellular calcium,abolishes transduction provides compelling evidencethat tip links are an essential element. However, theconcept that tip links act as the gating-springs hasrecently been challenged. High-resolution images haverevealed that tip links are rigid extracellular filaments,�10nm in diameter and 150–200nm long (Figure 1(e)).They appear to consist of two to three coiled fiberswith a helical period of�25 nm and multiple globulardomains �4nm in diameter. These measurementsare consistent with the molecular dimensions of cad-herin 23 (CDH23), which has 27 cadherin domains.CDH23 has been immunolocalized to the hair bundleand mutations in the gene that encodes CDH23 causedeafness and balance disorders in mice and humans.However, the notion that CDH23 is a component ofthe tip link remains controversial because hair bundleimmunolocalization is lost in adult mice despite thepersistence of tip links, which suggests that CDH23may play a role in hair bundle development instead.Nonetheless, the rigid structure of the tip link suggeststhat the search for the molecular identity of biophysi-cally defined gating-spring should be shifted to otherelements, mechanically in series with the transductionchannel.

The Process of Adaptation

Steady deflection of the bundle in the excitatory direc-tion causes an initial activation of transduction chan-nels, followed by a decline in the response toward theresting level as transduction channels close (Figure 2(b)).The instantaneous activation curve measured after100ms or so is shifted to the right, relative to itsposition at rest (Figure 2(b)). Tension on transductionchannels relaxes during the deflection so that addi-tional deflection can reactivate the channels. Similarly,a negative deflection that closes channels is followedby adaptation that increases the tension and shifts theactivation curve to the left. This process, commonlyreferred to as adaptation, allows hair cells to retainhigh sensitivity over a broad operating range and filtersout slower stimuli in favor of rapid stimuli. The

current decline to constant stimuli typically exhibits abimodal decay, which reveals two rates of adaptation:fast adaptation occurs within a few milliseconds or lesswhile slow adaptation is about 10 times slower. Thetwo forms of adaptation, although both dependent oncalcium, reflect distinct mechanisms.

Circumstantial evidence has suggested that slowadaptation requires myosin molecules, perhaps in acluster of a few dozen molecules. Members of themyosin family are the only molecules known tomove on actin, and actin in the cores of stereocilia ispolarized so that myosin would climb up to resettension. The rate at which the climbing phase ofadaptation (for negative stimuli) occurs and the rateat which muscle myosin-coated beads climb up theactin core are 1–2mms�1, similar to the rate of myosinin other systems. Compounds that block myosin–actininteraction abolish adaptation.

Like myosin–actin interactions in other systems,the hair cell adaptation mechanism is regulated bycalcium and the calcium-binding protein, calmodulin.The increase in intracellular calcium concentrationthat results from open transduction channels increasesthe rate of adaptation. When calcium levels rise, themyosin motors dissociate from actin and slip down theside of the stereocilium, which in turn relieves tip-linktension and allows the channels to close.

Out of �40 members of the myosin superfamily,immunolocalization data focused the search for theadaptation motor on myosin 1c (formerly known asmyosin 1-beta). A role for myosin 1c in adaptationwas confirmed physiologically, using a mouse modelin which myosin 1c was mutated, sensitizing it to achemical inhibitor. When the inhibitor was added,both slow and fast adaptations were abolished inmutant but not wild-type vestibular hair cells.

Several other members of the myosin superfamily,including myosin 7a and myosin 15, appear to becritical for normal hair bundle development and func-tion. Naturally occurring mutations in myosin 7acause altered hair bundle morphology and, as a result,auditory and vestibular deficits in mice and humans.Taken together with localization data, this suggeststhat myosin 7a may be involved in holding the stereo-cilia together and maintaining the hair bundle withinthe properly erect mechanosensitive range.

It should be noted that slow (>10ms)myosin-basedadaptation has been most fully characterized in ves-tibular hair cells of bullfrogs and mice. Fast adap-tation has been well characterized in hair cells ofthe turtle auditory organ and recently in the ratcochlea. In these cells, fast adaptation is also calciumdependent and causes a similar shift of the activationcurve. It has been suggested that calcium enteringthrough an open transduction channel can bind to

Hair Cells: Sensory Transduction 1019

and force reclosure of the channel (reclosure model:Figure 2(c)). A second model for fast adaptationhas been proposed that suggests that entering cal-cium triggers the release of a mechanical elementin series with the transduction channel (releasemodel: Figure 2(c)). The release reduces tension tothe gating-spring and allows the transduction chan-nels to close. Although bothmodels explain the rapiddecay in transduction current, they make oppositepredictions on the movement of the hair bundle thatoccurs during fast adaptation. The reclosure modelpredicts that channel reclosure will increase tensionin the gating-spring, which will pull the bundle inthe negative direction (to the left in Figure 2(c)).The release model predicts that there will be a reduc-tion in gating-spring tension which will allow thebundle to move further in response to a constantforce (to the right in Figure 2(c)). Recent data frommice and rats are consistent with the release model,whereas data from turtle and some from frog haircells support the reclosure model. The models are notmutually exclusive and both mechanisms may coexistin the same cell or to varying degrees in hair cells ofdifferent organs and species.

Cochlear Amplification

Mechanical amplification of sound stimuli in mam-mals requires the activity of outer hair cells whichreside in three rows along the entire length of themammalian cochlea. The process is known as co-chlear amplification and functions to enhance sensi-tivity to faint sounds, and to tune or sharpen finepitch perception. Two mechanisms have been pro-posed to underlie cochlear amplification: somaticmotility, driven by the hair cell receptor potential,and active movements of the hair bundle, associatedwith fast adaptation.Somatic motility requires the voltage-sensitive mem-

brane protein, prestin, which is abundant in the lateralmembranes ofmammalianouter hair cells but absent inhair cells of lower vertebrates. The somatic motilitytheory for cochlear amplification posits that soundstimuli generate receptor potentials that evoke volt-age-dependent conformational changes in prestinsuch that the protein contracts with depolarizationand expands with hyperpolarization. Because thecollective motion of the densely packed moleculesoccurs parallel to the lateral membranes, outer haircells shorten and elongate in time with changes inmembrane potential. The force produced by somaticmotility in isolated outer hair cells is about 100 pNmV�1. Forces of this amplitude are predicted to gen-erate synchronous motion of outer hair cells; this,in turn, amplifies the overall motion of the basilar

membrane in phase with the sound stimulus. How-ever, because somatic motility depends on the haircell receptor potential, the membrane time constantwill likely limit force production at the high stimulusfrequencies to which auditory organs are knownto be sensitive. In addition, because cochlear ampli-fication also occurs in lower vertebrates that lackouter hair cells and prestin, a second mechanismhas been invoked.

The other model suggests that the forces generatedby the hair bundle during adaptation are sufficient todrive amplification of sound stimuli in the auditoryorgans of many different species. Although the mech-anism of fast adaptation is controversial, there isagreement that fast adaptation can drive active hairbundle movements. The active bundle movements areon the order of tens of nanometers. If the hair bundlemovements are in phase with the sound stimulus,their collective motion is also predicted to driveamplification of basilar membrane oscillations. Totease apart the relative contributions of somaticmotility and hair bundle motility will require theirselective inactivation in live animals.

The Receptor Potential

Mature hair cells lack voltage-gated sodium chan-nels, therefore they do not generate action poten-tials. Rather, changes in transduction channel activityevoke graded receptor potentials of up to 40mV.The receptor potential is modified by the activity ofvoltage- and ion-sensitive conductances in the baso-lateral membrane. In some hair cells depolarizingreceptor potentials elicit oscillations of the membranepotential that may serve to tune the cell to a particularstimulus frequency. These oscillations result from aninterplay between activation of a-1D calcium chan-nels and BK calcium-dependent potassium channels.Variation in the kinetics of BK channels results fromboth alternative splicing and co-assembly with b sub-units and gives rise to membrane potential oscilla-tions that vary in frequency among hair cells of thesame organ.

The ability of hair cells to signal stimuli of eitherpolarity is preserved at the afferent synapse. At thecell’s resting potential (�50 to �70mV), voltage-sensitive calcium channels are tonically active. Thisresults in a steady release of neurotransmitter, whichin turn results in a steady background firing in thefibers of the eighth nerve. Thus, receptor potentialsof either polarity modulate calcium channel activityand hence transmitter release. A presynaptic electron-dense body (�0.5 mm in diameter) at each release site,around which vesicles cluster, is reminiscent of thesynaptic ribbon in photoreceptors and may serve a

1020 Hair Cells: Sensory Transduction

similar role in mediating continuous release. Neuro-transmitter release is modulated with submillisecondprecision to encode the frequency and intensity ofsound and head movements. The excitatory aminoacid glutamate has been shown to be released by haircells, but it is not clear whether it is the primary trans-mitter or whether other transmitters are co-released.The efferent synapse, carrying feedback control

from the central nervous system (CNS) via the olivo-cochlear fibers, uses acetylcholine as the transmitter.The synapse may be excitatory in some hair cells, butin most organs efferent stimulation causes a slowinhibitory hyperpolarization. The inhibition is causedby activation of calcium-dependent potassium chan-nels (SK). SK channels are voltage insensitive and canbe activated by micromolar concentrations of cal-cium. The a9 and a10 nicotinic cholinergic subunitshave been shown to form the receptor that mediatessynaptic transmission between the efferent fiber andhair cells. The high degree of calcium permeability ofthis receptor may permit sufficient calcium entry toactivate SK calcium-dependent potassium channels,and this, in turn, inhibits the cell.

See also: Cochlear Mechanics; Cochlear Development;

Deafness; Hair Cell Regeneration; Hair Cell

Differentiation; Sensory Aging: Hearing.

Further Reading

Bashtanov ME, Goodyear RJ, Richardson GP, et al. (2004) The

mechanical properties of chick (Gallus domesticus) sensory

hair bundles: Relative contributions of structures to calciumchelation and subtilisn treatment. Journal of Physiology559(1): 287–299.

Eatock RA (2000) Adaptation in hair cells. Annual Review inNeuroscience 23: 285–314.

Eatock RA, Corey DP, and Hudspeth AJ (1987) Adaptation of

mechanoelectrical transduction in hair cells of the bullfrog’s

sacculus. Journal of Neuroscience 9: 2821–2836.Eatock RA, Fay RR, and Popper AN (eds.) (2005) Springer Hand-

book of Auditory Research, Volume 27: Vertebrate Hair Cells.New York: Springer.

Fawcett DW (1981) The Cell. Philadelphia, PA: WB Saunders.

Fekete DM (1999) Development of the vertebrate ear: Insights

from knockouts and mutants. Trends in Neurosciences 6:

263–269.Fettiplace R and Fuchs PA (1999) Mechanisms of hair cell tuning.

Annual Review of Physiology 61: 809–834.

Fettiplace R and Hackney CM (2006) The sensory and motorroles of auditory hair cells. Nature Reviews Neuroscience 7:

19–29.

Fettiplace R, Ricci AJ, and Hackney CM (2001) Clues to the

cochlear amplifier from the turtle ear. Trends in Neurosciences24: 169–175.

Furness DN and Hackney CM (2006) Springer Handbookof Auditory Research: Vertebrate Hair Cells. Heidelberg:

Springer.Geleoc GS and Holt JR (2003) Auditory amplification: Outer hair

cells press the issue. Trends in Neuroscience 26: 115–117.Holt JR and Corey DP (2000) Two mechanisms for transducer

adaptation in vertebrate hair cells. Proceedings of the NationalAcademy of Sciences of the United States of America 97:

11730–11735.

Holt JR, Gillespie SKH, Provance DW, et al. (2002) A chemicalstrategy implicates myosin-K in adaptation by hair cells. Cell108: 371–381.

Hudspeth AJ (1983) The hair cells of the inner ear. ScientificAmerican 248(January 1983): 54–64.

Hudspeth AJ (2005) How the ear’s works work: Mechanoelectrical

transduction and amplification by hair cells. Comptes RendusBiologies 328: 155–162.

Hudspeth AJ, Choe Y, Mehta AD, et al. (2000) Putting ion chan-nels to work: Mechanoelectrical transduction, adaptation,

and amplification by hair cells. Proceedings of the NationalAcademy of Sciences of the United States of America 97:11765–11772.

Kachar B, Parakkal M, Kurc M, Zhao Y, and Gillespie PG (2000)

High-resolution structure of hair-cell tip links. Proceedingsof the National Academy of Sciences of the United States ofAmerica 97: 13336–13341.

LeMasurier M and Gillespie PG (2005) Hair-cell mechanotrans-

duction and cochlear amplification. Neuron 48: 403–415.

Parsons TD and Sterling P (2003) Synaptic ribbon. Conveyor beltor safety belt? Neuron 37: 379–382.

Pickles JO and Corey DP (1992) Mechanoelectrical transduction

by hair cells. Trends in Neuroscience 15: 254–259.Strassmaier M and Gillespie PG (2002) The hair cell’s transduction

channel. Current Opinion in Neurobiology 12: 380–386.

Wu YC, Ricci AJ, and Fettiplace R (1999) Two components

of transducer adaptation in auditory hair cells. Journal ofNeurophysiology 82: 2171–2181.