reviews
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
and IL-1 can be elevated in other viral infections causing meningitis or encephalitis (K. Frei, University Hospital Zfidch, Switzerland). Thus, the lack of specificity of these responses indicates that we do not yet understand their exact influence on the injury to neurons and possibly other CNS cell types that has been observed in AIDS, and highlights the need for additional work in this area.
Another cytokine that is reportedly elevated in the CSF of at least some neu rologically symptomatic HIV-infected patients is tumor necrosis factor oc (TNFo0, which may be produced by microgila (Griffin). Moreover, glutamate can cause oscillations in [Ca2+] and possibly induce TNF~ and interleukin 1 (IL-1) in astrocytes (Merrill). In addition, neopterin (presumably released from cells of the macrophage or microglial lineage, as mentioned by Fuchs et al.) and 132-microglobulin are increased in the CSF of neurologically symptomatic HIV-positive patients. However, their significance, except as possible markers of disease, remains unknown 4-6. Despite these and additional tantalizing cytokine findings in AIDS, other infections affecting the CNS besides HIV-1 can lead to the intrathecal production of various cytokines. For example, TNFoc and IL-1 levels can increase in bacterial meningitis, and INF-y
compared with HIV-negative controis (D. Griffin, Johns Hopkins University School of Medicine, Baltimore, MD, USA), and can lead to microgliosis (J. Merrill, UCLA, CA, USA). Nevertheless, these levels of INF-y actually fall in HIV-infected persons with dementia compared with HIVinfected patients without neurologic disease (Griffin). In contrast, the elevated CSF concentration of quinolinate is closely associated with the degree of cognitive impairment in patients with AIDS 3. Thus, there does not appear to be a tight temporal correlation between INF-y levels and quinolinate levels in the CSF of patients with HIV-associated cognitivemotor complex (of which the AIDS dementia complex is a severe form). This finding may raise questions about the possible relationship between INF-y and quinolinate concentrations in vivo, but perhaps the requirement for such a temporal association is unwarranted.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Stuart A. Lipton
Harvard-Longwood Neurology Program, 300 LongwoodAve, Boston,MA 02115, USA.
References '1 Merrill, J. E. and Chen, I. S. Y. ('1991) FASEBJ. 5, 2391-2397 2 Lipton,S.A. (1992) TrendsNeuros¢i. 15, 75--79 3 Heyes,M. P. etal. ('1991)Ann. Neurol. 29, 202-209 4 Griffin, D. E., McArthur, J. A. and Cornblath, D. R. (1991) Neurology 4'1, 69-74 5 McArthur, J. C. et al. Neurology (in press) 6 Tyor, W. R. et al. (1992) Ann. Neurol. 31,349-360
.
.
.
Mechanoeleddcaltransdudionby hair cells James O. Pickles and David P. Corey JamesO, Picklesis at the Vision, Touchand HeatingResearch Centre, Deptof Physiologyand Pharmacology, Universilyof Queensland, 4072 Queensland, Australia,and David P. Coreyis at the HowardHughes MedicalInstituteand the Dept of Neurology, Massachuse~
Hair cells of the inner ear are one of nature's great success stories, appearing early in vertebrate evolution and having a similar form in all vertebrate classes. They are specialized columnar epithelial cells, with an array of modified microviUi or stereocilia on their apical surface, interconnected by a series of linkages. The mechanical stimulus causes deflection of the stereocilia, stretching linkages between them, and opening the mechanotransducer channels. On a slower time.scale, hair cells adapt in order to maintain optimum sensitivity, with an adaptation motor within the stereocilia acting to keep the resting tension on channels constant.
Mechanosensitive hair cells are found in the auditory and vestibular organs of all vertebrates, as well as in Genera/Hospital, Boston,/HA02114, the lateral line system of fishes and amphibia. While USA: the hair cells are all variations on one basic cell type, each system is given stimulus specificity by the way that the mechanical stimulus is coupled to the hairs or stereoc'llia. In all these systems, deflection of the stereocilia opens mechanically sensitive ion channels to depolarize the cell 1'2. Figures 1 and 2 show the structure of the transducing apical pole of the hair cell. This basic form is found in all mechanosensitive hair cells, whether the inner or outer hair cells of the cochlea, the hair cells of the vestibular system or the hair cells of the lateral line. The stereocllia are arranged hexagonally on the surface of the hair cell, and are composed of actin
254
© 1992.ElsevieSci r encePublishersLtd.(UK)
filaments packed in a dense paracrystal. This gives them a stiff, rigid structure so that they bend only at their base when deflected 3'4. The stereocilia are graded in height, being tallest on one side of the bundle and shortest on the opposite side. Adjacent to the tallest row of stereocilia is a single true cilium called the kinoeilium (which in the mammalian cochlea disappears soon after birth). A line drawn from the kinocilium through the center of the bundle runs parallel to the steepest gradient in stereocilia heights, and forms an axis around which the whole bundle is bilaterally symmetrical. The stereocilia are also extensively crosslinked. Some links run parallel to the apical surface of the hair cell, joining adjacent stereocilia in all directions. These links are thought to hold the stereocilia together, so that their tips slide along one another when the bundle is deflected. There are also specialized connections known as the tip links; these emerge from the tips of the stereoc'flia and run almost vertically along the hair cell axis, to join the side-wall of each adjacent taller stereocilium 5'6 (Figs 1 and 2).
What does the hair cell respond to? Micromanipulation experiments by Hudspeth and Corey 1'7 have provided the most direct evidence that mechanotransduction depends on deflection of the stereocilia. In these experiments, the sensory epithelium was isolated from the sacculus of the bullfrog TINS, VoI. 15, NO. 7, 1992
What is the electrophysiological mechanism of transduction~ The ion channels are relatively non-selective among the alkali cations, with Li +, Na*, K*, Rb* and Cs* being nearly equally permeant TM. Since the apical surface of hair cells in vivo is faced with endolymphatic s of nmetry fluid, which contains a high [K*], we expect that K + will normally carry much of the transducer current. / Ca 2* also passes through the transducer channel but f/ with a lower efficacy. However, Ca z+ is a necessary cofactor in transduction, since transducfion ceases if the concentration falls below about 10 ~ML2n2. One remarkable finding is that there seem to be very few active transducer channels on each cell: in the order of 50-200 depending on the system, or one or two per stereocilium 13. This figure, which is approximate, has been derived independently, from the analysis of mechanotransducer noise TM, from the total transducer current knowing the conductance of an individual channel 15, and indirectly from changes in bundle stiffness associated with transducer channel opening 16 (see below). If the same numbers apply in humans, then all our auditory perception depends on a total of only 6 x 10 s mechanosensitive channels, found on the hair cells from the inner ear. This small number might be understood on energetic grounds; if there is only enough energy in an acoustic stimulus to open a certain number of channels, then having more channels available on the hair cells will give relatively little Fig. 1. Three rows of stereocilia are seen in cross section increase in sensitivity. To permit high-frequency perception in some in a cutaway diagram of the apical transducing pole of an acousticolateral hair cell. The broken line shows the axis animals (up to 200 kHz in some bats17), the transaround which the hair bundle is bilaterally symmetric. The ducer channels must be fast. Direct measurements of stereocilia are graded by height in the direction of this fine, current responses to step-deflections in the bullfrog and the tip links also run parallel to this axis. The single kinocilium in the bundle is situated on the axis of symmetry, adjacent to the tallest row of stereocilia. 5ubmembranous densities underlie the attachment points of the tip links. In addition to the tip links, the stereocilia are also joined just below their tips by lateral bands of links (not shown here) that run parallel to the apical surface of the hair cell. The rootlets of the stereocilia are embedded in a rigid actin matrix known as the cuticular plate. ium
(Rana catesbeiana); the otolithic membrane - which normally couples the stimulus to the stereocilia - was removed from over the stereocilia; and the stereocilia were manipulated directly while recording the intracellular potentials. Mechanoreceptor potentials were shown to depend on deflection of only the stereocilia; separate deflection of the kinocilium had no effect8. Deflection of the bundle towards the tallest stereocilia opens transduction channels, as measured by a conductance increase; this causes a depolarizing receptor potential and, ultimately, the excitation of afferent fibers that form synapses with the hair cell. Deflection in the opposite direction shuts the few channels that are open at rest, causing intracellular hyperpolarization and neuronal inhibition. Deflection at right angles to this axis has no effect9' lo. The axis for maximum activation of the mechanotransducer channels therefore runs parallel to the axis of bilateral symmetry of the hair bundle, deflections in other directions having an effect depending on. the vector component of the stimulus along the best axis 9.
TINS, Vol. 15, No. 7, 1992
Fig. 2. A hair bundle of the chick basilar papilla, showing graded ranks of stereocilia, joined by tip links (arrows). In this preparation, the 6 nm thick central filament of each tip link is made thicker by some of the adhering glycocalyx, as well as by a 25 nm metal coating. The stereocilia to the left of the bundle have parted, allowing one to see between them. 255
X /l~ ~ ~1
threshold of hearing. Although the background noise of thermal energy is the equivalent of 30-50 nm of channels open and close so rapidly that the I /I deflection, cell membrane potential reflects an average open The actual definition of the threshold ~ /~ ~ tI iI probability. becomes a statistical one, dependent on the amount of |\ / ~ / t averaging employed. Comparison of behavioral sensitivity and measured basilar membrane motion2° suggests that the behavioral threshold corresponds to a deflection of the stereocilia of about 0.4 nm, or shear between their rows of approximately 0.04 nm '),I \/,\ ,\ // (Ref. 21). This extraordinary sensitivity is a reflection of the molecular scale over which deflection-induced channel opening occurs. The kinetic model has a further implication. At a certain displacement of the stereocilia, when the D channel moves from its closed to open state, the / / \ gating spring should shorten by an amount equal to the movement of the channel gate. This should add an Energy / \ openjrl/ extra compliance to the system, and should be measurable as a drop in the stiffness of the bundle at the point at which the channels are undergoing their closed open o 1',o Bundle Displacement (/~m) maximum number of transitions between the open and closed states. Such changes in the stiffness of the bundle have been found by Howard and Hudspeth TM, Fig. 3. The tip-rink model for transduction. (A) Mechanical stimuri act directly and have been correlated with the probability of on ion channels in the stereocilia, by increasing tension in a 'gating spring' opening the channel. Together, these experiments attached to the channel gate. (B) At any one time channels are distributed between open and closed states, but an increase in tension (dashed line) makes indicate that the transduction channels are directly the open state energetically more favorable, and shifts the distribution towards gated by mechanical tension. A
C
..~
I\/\,,/
/
l,\/\;
/
'° T /
the open state. (C) A simple structural model comes from the suggestion that each tip link is a gating spring. Displacing the bundle towards the tallest stereocilia (a positive stimulus) stretches the tip links to increase tension. The channels (somewhere in the distal ends of the stereocilia) are shown in this figure to be at the lower end of each tip link, but equally well could be at the upper end, or, indeed, both. (D) The sensitivity curve, relating displacement of the bundle (measured at the tips) to the probability of the channel being open, shows the very narrow operating range of hair cells. A displacement of about a third of a micrometre - approximately the diameter of one stereocilium - is a saturating stimulus. The curve also shows that some channels are open in the resting position, suggesting some resting tension in the tip links.
sacculus, an organ which might not be particularly adapted to high-frequency operation, suggest a latency of 40 ~s at 22°C, which can be expected to reduce to 13 ~s at 37°CTM. This very short latency severely constrains models of channel operation. For instance, it rules out models in which the current is modulated in response to a second messenger inside the ceil. The most likely hypothesis is that the channel is simply and directly opened mechanically by the stereociliary movement. A biophysical model of transduction which successfully explains the channel kinetics was put forward by Corey and Hudspeth 19. They suggested that each channel was pulled open by an elastic link, known as the 'gating spring'. Each channel with its associated spring has an energy profile with energy wells representing open- and closed-state conformations. The Boltzmann distribution governs the probability of channels being in the open state, and the energy levels are such that when no tension acts on the gating spring the closed state is favored. Tension on the gating spring changes the energy profile of the system, favoring the open state (Fig. 3). Since channel opening is a stochastic process, the stimulus merely causes a redistribution between the open and closed states and there is no ultimate 256
What are the structural e l e m e n t s involved in transduction? Still unclear in this model are the location of the channel and what pulls on it. In order to determine the location, Hudspeth explored around a hair bundle with an extracellular electrode while deflecting the stereocilia with a probe; the extracellular potential changes were greatest around the top of the bundle, along the tips of the stereocilia 22. This suggests that the transducer currents were flowing into the cell near or at the tips of the stereocilia. Although challenged by Ohmori23, this site has been confirmed by calcium imaging24 and by spatially localized iontophoresis of the channel blocker, gentamicin 25. Important structural evidence came with the discovery of the tip links by Pickles, Osborne and Corals5'21'26, which has been confirmed by others 27'28. The tip link has a fine central filament (Fig. 4), presumably protein, that is surrounded by glycocalyx29'3°. This filament would be ideal for transferring the movement-induced forces onto single mechanotransducer channels in the membrane. There is also a report that, in some cases at least, the tip link is not a single strand but branches at its upper end (Lim, D. J. and Kachar, B., unpublished observations). The upper end attaches to a density which bridges the plasmalemma to the central actin paracrystal of the stereocilium (Fig. 4). At the lower end of the tip link the stereociliary membrane is often seen to be pulled into a point. This is possibly due to differential shrinkage in fixation; however it does suggest that the link here is not directly attached to the actin paracrystal, but rather attaches indirectly via distensible connectors within or underlying the membrane31. The tip link is in an ideal position to detect the shear or sliding movement between the different rows of TINS, VoL 15, No. 7, 1992
stereocilia. Because the stereocilia are graded in height and the tip links are oriented nearly vertically, the links would be stretched by a deflection of the stereocilia in the direction of the tallest, that is, in the excitatory direction. Deflection in the opposite direction would be expected to take the tension off the tip links. A model for transduction The mechanical gating of channels, their localization and the function of tip links were put together in an appealingly simple model for transduction by Pickles, Osborne and Comis5'21'26. In this model (Fig. 3) the tip link is the gating spring and pulls directly on the ion channels. It is not clear whether the transducer channels are associated with the upper or lower ends of the tip link, or both. Either end of the links would be suitable for transferring forces to membrane-bound channels. The close correlation between the anatomical and the electrophysiological evidence supports this model; for instance, the tip links run almost exclusively along the axis of bilateral symmetry of the hair cell, that is, only along the anatomical axis along which movement of the stereocilia affects the transducer channels since the stereocilia are hexagonally packed, other directions might be equally possible6'32. Tip links seem to be present in all mechanotransducing hair cell systems, having been found in vestibular, cochlear and lateral line systems, and in representatives of all vertebrate classes 6'13'27'33'34. A gradation in heights of the stereocilia is required to accommodate the tip links, and such a gradation is seen in all mechanotransducing acousticolateral hair cells, although the gradation is not seen in ciliated electroreceptors, which are thought also to be of acousticolateral origin 35. Recently, this model has been directly tested by cutting the tip links2o Reducing the calcium concentration around the bundle with the fast calcium chelator BAPTA [1,2-bis(o-aminophenoxy)ethaneN,N,N',N'-tetraacetic acid] irreversibly destroyed transduction in a few tenths of a second. A similar treatment with BAPTA for a few seconds eliminated the tip links, viewed with either scanning or transmission electron microscopy2. When freestanding bundles were treated with BAPTA, they relaxed forward by about a tenth of a micrometre 2, in quantitative agreement with a model for tension regulation in the gating springs 36. These results indicate that tip links indeed convey tension to the transduction channels.
An adaptation mechanism regulates tension in the tip links The extraordinary sensitivity of the transduction mechanism poses a problem for the hair cell, in that larger stimuli easily saturate the transduction. This is a problem for hair cells in the vestibular system that sense acceleration, because the constant acceleration of gravity imposes a stimulus a thousandfold larger than their threshold sensitivity. On a different timescale, this high sensitivity creates a developmental problem for all hair cells, because tip links and channels must be positioned within the hair bundle with an accuracy of 10-20 nm. As a consequence, hair cells have developed an adaptation mechanism that continuously adjusts the TINS, Vol. 15, No. 7, 1992
Fig. 4. Transmission electron micrograph of the tip of a stereocilium in the chick basilar papilla, showing the central filament of the tip link (arrow), densities underlying the upper and lower points of attachment (arrowheads), and the band of lateral links between the stereocilia (double arrowhead). The glycocalyx which surrounds the central filament of the tip link has not been preserved here.
tension in the tip links. In vitro experiments initially showed that steady displacements of a hair bundle caused the sensitivity curve to move along the displacement axis to match the new position of the bundle without appreciably changing shape (Fig. 5) 16. For saturating displacement stimuli, this both reduces the response and repositions the maximum sensitivity at the steady bundle position. The mechanism is apparently active in vivo, at least in the bullfrog saccule where it is thought to underlie an adaptation of the saccular microphonic potential aT. This shift of the sensitivity curve can be reconciled with the tip-links model for transduction by supposing that adaptation involves an adjustment of tension in the tip link, specifically by sliding the upper attachment point of the tip link along the side of the stereocilium (Fig. 5)38. More recently, a quantitative model for adaptation has been developed, based on an active motor element 36. This model can account for a variety of phenomena associated with adaptation, and also predicts active movements of the hair bundle under certain conditions. Such movements of the whole bundle have been observed, which tends to confirm the mechanical basis of adaptation 36. Interestingly, the rate of tension adjustment depends on the concentration of Ca2÷ inside the tips of 257
A /~m0.~[ _~
C
1
hypothesis, it may also be that the antibody recognizes latent channels and not those mechanically connected to tip links, or that the epitope recognized might not be the transduction channel at all. Towards a molecular basis
I
I 100
0
B
lOT//-
ms
--
I /--/ 0
110
Bundle Displacement
{pro)
Fig. 5. Adaptation by adjustment of tip-link tension. (A) A displacement of the bundle opens channels, but they close spontaneously in tens of milliseconds. (B) The closure is due to a shift of the sensitivity curve in the direction of the displacement, suggesting an adjustment of mechanical tension on channels. (C) A way to reconcile a variable tension with the tip-links model is to suppose that the upper tip-link attachment can move along the side of the stereocilium, slipping down to loosen when stretched or climbing up to tighten when relaxed.
the stereocilia, suggesting that the adjustment motor is in the tips very near the transduction machinery. Indeed, displacement of the upper tip-link insertion site has been observed in hair cells fixed in various adaptation states, which is in agreement with the model and confirms the involvement of the tip links 39.
The next level of understanding will entail a molecular description of the elements in the transduction chain. The excitement of tracing the basis of hearing to a few molecules will be tempered by the great difficulty in purifying the relevant proteins, particularly since we expect them to be present in amounts of the order of an attomole per animal45'46. The channel, the tip link and the motor remain biochemically unidentified at present. Of these, the motor molecule responsible for adaptation may be the most tractable, since it resembles the family of myosins in its properties. The actin core of stereocilia provides the first clue, since the only molecules known to move on actin are of the myosin family. The maximum rate of adaptation that tightens the tip links is in the range of 1-2 ~m~/s, the same rate at which myosin moves on actin, and myosin-coated glass beads have been found to climb towards the tips of stereocilia cores at a similar velocity47. However, immunological and biochemical approaches to identifying the motor are still inconclusive. Perhaps the best hope for identifying the channel is by molecular probing for similarity to other ion channels 4a. In 1962 Georg von Bekesy concluded several decades of work on the cochlea by suggesting that 'it does not seem impossible that the final mechanical transformer is of molecular dimensions'. Now, thirty years later, we think we know where that mechanical transformer is, and how it works. Hair cell research is presently at a turning point, as efforts are directed towards identifying and understanding the 'transformer' in molecular terms.
Alternative models for transduction
No alternative models for hair-cell mechanotransduction, beyond that presented here, have been put Selected references 1 Corey, D. P. and Hudspeth, A. J. (1979) Nature 281, forward in the past decade. Earlier, it was suggested 675-677 that transduction might depend on distortion of the 2 Assad, J. A., Shepherd, G. M. G. and Corey, D. P. (1991) whole hair-cell body4° (which has been ruled out by Neuron 7,985-994 3 Flock, A., Flock, B. and Murray, E. (1977) Acta Otolaryngol. experiments in which mechanotransduction was as83, 85-91 sociated with micromanipulation of the stereocilial'8), Tilney, L. G., Egelman, E. H., DeRosier, D. J. and Saunders, or of the kinocilium alone41 (which has been ruled out 4 J. C. (1983) J. Cell Biol. 96, 822-834 by experiments in which the stereocilia were manipu5 Pickles, J. O., Comis, S. D. and Osborne, M. P. (1984) Hear. lated separately from the kinocilium, showing that Res. 15, 103-112 transduction depends only on the formerS). Sugges- 6 Pickles, J. O. etaL (1989) Hear. Res. 41, 31-42 7 Hudspeth, A. J. and Corey, D. P. (1977) Proc. NatlAcad. 5ci. tions that mechanotransduction depended on memUSA 74, 2407-2411 brane distortion at the sites at which the stereocilia 8 Hudspeth, A. J. and Jacobs, R. (1979) Proc. Natl Acad. 5ci. flexed in response to deflection, that is, at their bases USA 76, 1506-1509 where they enter the cell body42, were countered by 9 Shotwell, S. L., Jacobs, R. and Hudspeth, A. J. (1981) Ann. NY Acad. 5ci. 374, 1-10 the experiments described above suggesting that the 10 Lowenstein, O. and Wers&ll, J. (1959) Nature 184, 1807-1808 channels were instead located near the tips of the 11 Ohmori, H. (1985) J. PhysioL 359, 189-217 stereocilia22, 24, 25. 12 Crawford, A. C., Evans, M. G. and Fettiplace, R. (1991) It was found recently that antibodies to an J. Physiol. 434, 369-398 amiloride-sensitive ion channel from renal epithelium 13 Hudspeth, A. J. (1989) Nature 341,397-404 bind near the tips of the stereociliaa3. Since amiloride 14 Holton, T. and Hudspeth, A. J. (1986) J. PhysioL 375, 195-227 also blocks transduction channels a4, although at a 15 Russell, I. J. (1983) Nature 301,334-336 higher concentration than that required to block the 16 Howard, J. and Hudspeth, A. J. (1988) Neuron 1, 189-199 renal channel, the antibody might recognize the hair- 17 Dannhof, B. J. and Bruns, V. (1991) Hear. Res. 53,253-268 cell transduction channels. Antibody binding occurred 18 Corey, D. P. and Hudspeth, A. J. (1979) Biophys. J. 26, mainly at the points of contact between stereocilia and 19 499-506 Corey, D. P. and Hudspeth, A. J. (1983) J. Neurosci. 3, not at either end of the tip links. Although this finding 962-976 is obviously inconsistent with the simple tip-links 20 Sellick, P. M., Patuzzi, R. and Johnstone, B. M. (1982) 258
TINS, VoL 15, No. 7, 1992
J. Acoust. Soc. Am. 72, 131-141 Pickles, J. O. (1985) Prog. NeurobioL 24, 1-42 Hudspeth, A. J. (1982) J. Neurosci. 2, 1-10 Ohmori, H. (1988)J. Physiol. 399, 115-137 Huang, P. L. and Corey, D. P. (1990) Biophys. J. 57, 530a Jaramillo, F. and Hudspeth, A. J. (1991) Neuron 7, 409-420 Osborne, M. P., Comis, S. D. and Pickles, J. O. (1984) Cell Tissue Res. 237, 43-48 27 Little, K. F. and Neugebauer, D-Ch. (1985) Cell Tissue Res.
21 22 23 24 25 26
242,427-432 28 Furness, D. N. and Hackney, C. M. (1985) Hear. Res. 18, 177-188 29 Osborne, M. P., Comis, S. D. and Pickles, J. O. (1988) Hear. Res. 35, 99-108 30 Neugebauer, D-Ch. and Thurm, U. (1987) Cell Tissue Res. 249, 199-207 31 Pickles, J. O., Rouse, G. W. and von Perger, M. (1991) Scanning Microsc. 5, 111 5-1128 32 Comis, S. D., Pickles, J. O. and Osborne, M. P. (1985) J. Neurocytol. 14, 113-130 33 Rhys-Evans, P. H., Comis, S. D., Osborne, M. P., Pickles, J. O. and Jeffries, D. J. R. (1985) J. Laryngol. Otol. 99, 11-19 34 Rouse, G. W. and Pickles, J. O. (1991) J. NlorphoL 209,
111-120 35 Szabo, T. (1974) in Handbook of Sensory Physiology 3 (Fessard, A., ed.), pp. 13-58, Springer 36 Assad, J. A. and Corey, D. P. J. Neurosci. (in press) 37 Eatock, R. A., Corey, D. P. and Hudspeth, A. J. (1987) J. Neurosci. 7, 2821-2836 38 Howard, J. and Hudspeth, A. J. (1987) Proc. NatlAcad. Sci. USA 84, 3064-3068 39 Shepherd, G. M. G. et al. J. Gen. Physiol. (in press) 40 Stevens, S. S. and Davis, H. (1983) Hearing: Its Psychology and Physiology, Wiley 41 Hillman, D. E. (1969) Brain Res. 13, 407-412 42 Dallos, P. (1973) The Auditory Periphery, Academic Press 43 Hackney, C. M., Furness, D. N. and Benos, D. J. (1991) Scanning Microsc. 5, 741-746 44 Jorgensen, F. and Ohmori, H. (1988) J. Physiol. 403, 577-588 45 Shepherd, G. M. G., Barres, B. A. and Corey, D. P. (1989) Proc. Natl Acad. Sci. USA 86, 4973-4977 46 Gillespie, P. G. and Hudspeth, A. J. (1991)J. Cell Biol. 112, 625-640 47 Shepherd, G. M. G., Corey, D. P. and Block, S. M. (1990) Proc. Natl Acad. Sci. USA 87, 8627-8631
Acknowledgements Work in the authors' laboratories was supported by the MRC of the UK and the Australian ResearchCouncil (J.O.P.)and by the Nab'onallnstitutes of Health, the Office of Naval Research, and the Howard Hughes Medical Institute (D.P.C.).
Changesin Caz +-bindingproteinsin human neurodepenerativedisorders Claus W. Heizmann
The cellular distribution of Cae+-binding proteins has been extensively studied during the past decade. These proteins have proved to be useful neuronal markers for a variety of functional brain systems and their circuitries. Their major roles are assumed to be Ca e+ buffering and transport, and regulation of various enzyme systems. Since cellular degeneration is accompanied by impaired Ca e+ homeostasis, a protective role for cae+-binding proteins in certain neuron populations has been postulated. As massive neuronal degeneration takes place in several brain diseases of humans, such as Alzheimer's disease, Parkinson's disease and epilepsy, changes in the expression of Ca 2+binding proteins have therefore been studied during the course of these diseases. Although the data from these studies are inconsistent, the detection and quantification of Cae+-binding proteins and the neuron populations in which they occur may nevertheless be useful to estimate, for example, the location and extent of brain damage in the various neurological disorders. I f future studies advance our knowledge about the physiological functions of these proteins, the neuronal systems in which they are expressed may become important therapeutical targets for preventing neuronal death in an array of neurodegenerative diseases. More than 7000 articles on calcium were published in 1991 (a figure found using the Medline database, with 'calcium' as the search word). This emphasizes the tremendous interest and progress in Ca2+-related research. In nerve cells, Ca 2+ ions activate and regulate a number of key processes, including fast axonal transport of substances, synthesis and release of some neurotransmitters, and membrane excitability. It is also thought that in the CA1 hippocampal region of the brain, Caz+ might induce long-term potentiation and memory storage mechanisms. The Ca2+ message is converted into an intracellular response - in many cases by CaZ+-binding proteins TINS, Vol. 15, No. 7, 1992
and Katharina Braun
that are involved in a wide variety of activities, such as cytoskeletal organization, cell motility and differentiation, cell-cycle regulation, and Ca2+ buffeting and transport. It might therefore be possible that altered levels of some Ca2+-binding proteins (e.g. due to deletion or mutation of the corresponding genes) could lead to an impaired Ca2+ homeostasis in nerve cells and to pathological conditions. Using mostly immunohistochemical techniques, several research groups have now started to search for altered expression of the CaZ+-binding proteins parvalbumin, calbindin-D28K and S100 in affected brain regions of patients suffering from acute insults, such as stroke and epileptic seizures, and from chronic neurodegenerative disorders, such as Alzheimer's, Huntington's, Parkinson's and Pick's diseases. In addition, altered Ca2+ levels have been found in platelets of patients with bipolar affective disorders 1, and Ca 2+ antagonists have been suggested for treatment of psychotic depression. The few proteins that have been investigated in these pathological states belong to a large family of more than 200 members. These proteins are characterized by a common structural motif, the EF-hand, which is present in multiple copies and binds Ca2+ selectively and with high affinity. Each of these domains consists of a loop of 12 amino acids that is flanked by two ochelices. This structural principle was first identified as a result of studying the crystal structure of the Ca2+-binding protein parvalbumin isolated from carp, and is designated the EF-hand because of the arrangement of the E and F helices of parvalbumin. A consensus amino acid sequence for this motif has aided the identification of many new members of this protein family2'3. Ca2+-binding proteins with known functions, such as calmodulin, troponin C, myosin light chains, calpain, calcineurin and recoverin, are far outnumbered by those with unidentified roles. Most Ca2+-
© 1992, ElsevierSciencePublishersLtd,(UK)
Claus W. Heizrnannis at the Dept of Pediatrics, Division of Clinical Chemistry, Universityof Zurich, 5teinwiesstr. 75, CH-8032 Zurich, Switzerland,and
Katharina Braun is at the Institute for Neurobiolosy, Brenneckestr. 6, 0-3090 Magdeburg, FRG.
259
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
and IL-1 can be elevated in other viral infections causing meningitis or encephalitis (K. Frei, University Hospital Zfidch, Switzerland). Thus, the lack of specificity of these responses indicates that we do not yet understand their exact influence on the injury to neurons and possibly other CNS cell types that has been observed in AIDS, and highlights the need for additional work in this area.
Another cytokine that is reportedly elevated in the CSF of at least some neu rologically symptomatic HIV-infected patients is tumor necrosis factor oc (TNFo0, which may be produced by microgila (Griffin). Moreover, glutamate can cause oscillations in [Ca2+] and possibly induce TNF~ and interleukin 1 (IL-1) in astrocytes (Merrill). In addition, neopterin (presumably released from cells of the macrophage or microglial lineage, as mentioned by Fuchs et al.) and 132-microglobulin are increased in the CSF of neurologically symptomatic HIV-positive patients. However, their significance, except as possible markers of disease, remains unknown 4-6. Despite these and additional tantalizing cytokine findings in AIDS, other infections affecting the CNS besides HIV-1 can lead to the intrathecal production of various cytokines. For example, TNFoc and IL-1 levels can increase in bacterial meningitis, and INF-y
compared with HIV-negative controis (D. Griffin, Johns Hopkins University School of Medicine, Baltimore, MD, USA), and can lead to microgliosis (J. Merrill, UCLA, CA, USA). Nevertheless, these levels of INF-y actually fall in HIV-infected persons with dementia compared with HIVinfected patients without neurologic disease (Griffin). In contrast, the elevated CSF concentration of quinolinate is closely associated with the degree of cognitive impairment in patients with AIDS 3. Thus, there does not appear to be a tight temporal correlation between INF-y levels and quinolinate levels in the CSF of patients with HIV-associated cognitivemotor complex (of which the AIDS dementia complex is a severe form). This finding may raise questions about the possible relationship between INF-y and quinolinate concentrations in vivo, but perhaps the requirement for such a temporal association is unwarranted.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Stuart A. Lipton
Harvard-Longwood Neurology Program, 300 LongwoodAve, Boston,MA 02115, USA.
References '1 Merrill, J. E. and Chen, I. S. Y. ('1991) FASEBJ. 5, 2391-2397 2 Lipton,S.A. (1992) TrendsNeuros¢i. 15, 75--79 3 Heyes,M. P. etal. ('1991)Ann. Neurol. 29, 202-209 4 Griffin, D. E., McArthur, J. A. and Cornblath, D. R. (1991) Neurology 4'1, 69-74 5 McArthur, J. C. et al. Neurology (in press) 6 Tyor, W. R. et al. (1992) Ann. Neurol. 31,349-360
.
.
.
Mechanoeleddcaltransdudionby hair cells James O. Pickles and David P. Corey JamesO, Picklesis at the Vision, Touchand HeatingResearch Centre, Deptof Physiologyand Pharmacology, Universilyof Queensland, 4072 Queensland, Australia,and David P. Coreyis at the HowardHughes MedicalInstituteand the Dept of Neurology, Massachuse~
Hair cells of the inner ear are one of nature's great success stories, appearing early in vertebrate evolution and having a similar form in all vertebrate classes. They are specialized columnar epithelial cells, with an array of modified microviUi or stereocilia on their apical surface, interconnected by a series of linkages. The mechanical stimulus causes deflection of the stereocilia, stretching linkages between them, and opening the mechanotransducer channels. On a slower time.scale, hair cells adapt in order to maintain optimum sensitivity, with an adaptation motor within the stereocilia acting to keep the resting tension on channels constant.
Mechanosensitive hair cells are found in the auditory and vestibular organs of all vertebrates, as well as in Genera/Hospital, Boston,/HA02114, the lateral line system of fishes and amphibia. While USA: the hair cells are all variations on one basic cell type, each system is given stimulus specificity by the way that the mechanical stimulus is coupled to the hairs or stereoc'llia. In all these systems, deflection of the stereocilia opens mechanically sensitive ion channels to depolarize the cell 1'2. Figures 1 and 2 show the structure of the transducing apical pole of the hair cell. This basic form is found in all mechanosensitive hair cells, whether the inner or outer hair cells of the cochlea, the hair cells of the vestibular system or the hair cells of the lateral line. The stereocllia are arranged hexagonally on the surface of the hair cell, and are composed of actin
254
© 1992.ElsevieSci r encePublishersLtd.(UK)
filaments packed in a dense paracrystal. This gives them a stiff, rigid structure so that they bend only at their base when deflected 3'4. The stereocilia are graded in height, being tallest on one side of the bundle and shortest on the opposite side. Adjacent to the tallest row of stereocilia is a single true cilium called the kinoeilium (which in the mammalian cochlea disappears soon after birth). A line drawn from the kinocilium through the center of the bundle runs parallel to the steepest gradient in stereocilia heights, and forms an axis around which the whole bundle is bilaterally symmetrical. The stereocilia are also extensively crosslinked. Some links run parallel to the apical surface of the hair cell, joining adjacent stereocilia in all directions. These links are thought to hold the stereocilia together, so that their tips slide along one another when the bundle is deflected. There are also specialized connections known as the tip links; these emerge from the tips of the stereoc'flia and run almost vertically along the hair cell axis, to join the side-wall of each adjacent taller stereocilium 5'6 (Figs 1 and 2).
What does the hair cell respond to? Micromanipulation experiments by Hudspeth and Corey 1'7 have provided the most direct evidence that mechanotransduction depends on deflection of the stereocilia. In these experiments, the sensory epithelium was isolated from the sacculus of the bullfrog TINS, VoI. 15, NO. 7, 1992
What is the electrophysiological mechanism of transduction~ The ion channels are relatively non-selective among the alkali cations, with Li +, Na*, K*, Rb* and Cs* being nearly equally permeant TM. Since the apical surface of hair cells in vivo is faced with endolymphatic s of nmetry fluid, which contains a high [K*], we expect that K + will normally carry much of the transducer current. / Ca 2* also passes through the transducer channel but f/ with a lower efficacy. However, Ca z+ is a necessary cofactor in transduction, since transducfion ceases if the concentration falls below about 10 ~ML2n2. One remarkable finding is that there seem to be very few active transducer channels on each cell: in the order of 50-200 depending on the system, or one or two per stereocilium 13. This figure, which is approximate, has been derived independently, from the analysis of mechanotransducer noise TM, from the total transducer current knowing the conductance of an individual channel 15, and indirectly from changes in bundle stiffness associated with transducer channel opening 16 (see below). If the same numbers apply in humans, then all our auditory perception depends on a total of only 6 x 10 s mechanosensitive channels, found on the hair cells from the inner ear. This small number might be understood on energetic grounds; if there is only enough energy in an acoustic stimulus to open a certain number of channels, then having more channels available on the hair cells will give relatively little Fig. 1. Three rows of stereocilia are seen in cross section increase in sensitivity. To permit high-frequency perception in some in a cutaway diagram of the apical transducing pole of an acousticolateral hair cell. The broken line shows the axis animals (up to 200 kHz in some bats17), the transaround which the hair bundle is bilaterally symmetric. The ducer channels must be fast. Direct measurements of stereocilia are graded by height in the direction of this fine, current responses to step-deflections in the bullfrog and the tip links also run parallel to this axis. The single kinocilium in the bundle is situated on the axis of symmetry, adjacent to the tallest row of stereocilia. 5ubmembranous densities underlie the attachment points of the tip links. In addition to the tip links, the stereocilia are also joined just below their tips by lateral bands of links (not shown here) that run parallel to the apical surface of the hair cell. The rootlets of the stereocilia are embedded in a rigid actin matrix known as the cuticular plate. ium
(Rana catesbeiana); the otolithic membrane - which normally couples the stimulus to the stereocilia - was removed from over the stereocilia; and the stereocilia were manipulated directly while recording the intracellular potentials. Mechanoreceptor potentials were shown to depend on deflection of only the stereocilia; separate deflection of the kinocilium had no effect8. Deflection of the bundle towards the tallest stereocilia opens transduction channels, as measured by a conductance increase; this causes a depolarizing receptor potential and, ultimately, the excitation of afferent fibers that form synapses with the hair cell. Deflection in the opposite direction shuts the few channels that are open at rest, causing intracellular hyperpolarization and neuronal inhibition. Deflection at right angles to this axis has no effect9' lo. The axis for maximum activation of the mechanotransducer channels therefore runs parallel to the axis of bilateral symmetry of the hair bundle, deflections in other directions having an effect depending on. the vector component of the stimulus along the best axis 9.
TINS, Vol. 15, No. 7, 1992
Fig. 2. A hair bundle of the chick basilar papilla, showing graded ranks of stereocilia, joined by tip links (arrows). In this preparation, the 6 nm thick central filament of each tip link is made thicker by some of the adhering glycocalyx, as well as by a 25 nm metal coating. The stereocilia to the left of the bundle have parted, allowing one to see between them. 255
X /l~ ~ ~1
threshold of hearing. Although the background noise of thermal energy is the equivalent of 30-50 nm of channels open and close so rapidly that the I /I deflection, cell membrane potential reflects an average open The actual definition of the threshold ~ /~ ~ tI iI probability. becomes a statistical one, dependent on the amount of |\ / ~ / t averaging employed. Comparison of behavioral sensitivity and measured basilar membrane motion2° suggests that the behavioral threshold corresponds to a deflection of the stereocilia of about 0.4 nm, or shear between their rows of approximately 0.04 nm '),I \/,\ ,\ // (Ref. 21). This extraordinary sensitivity is a reflection of the molecular scale over which deflection-induced channel opening occurs. The kinetic model has a further implication. At a certain displacement of the stereocilia, when the D channel moves from its closed to open state, the / / \ gating spring should shorten by an amount equal to the movement of the channel gate. This should add an Energy / \ openjrl/ extra compliance to the system, and should be measurable as a drop in the stiffness of the bundle at the point at which the channels are undergoing their closed open o 1',o Bundle Displacement (/~m) maximum number of transitions between the open and closed states. Such changes in the stiffness of the bundle have been found by Howard and Hudspeth TM, Fig. 3. The tip-rink model for transduction. (A) Mechanical stimuri act directly and have been correlated with the probability of on ion channels in the stereocilia, by increasing tension in a 'gating spring' opening the channel. Together, these experiments attached to the channel gate. (B) At any one time channels are distributed between open and closed states, but an increase in tension (dashed line) makes indicate that the transduction channels are directly the open state energetically more favorable, and shifts the distribution towards gated by mechanical tension. A
C
..~
I\/\,,/
/
l,\/\;
/
'° T /
the open state. (C) A simple structural model comes from the suggestion that each tip link is a gating spring. Displacing the bundle towards the tallest stereocilia (a positive stimulus) stretches the tip links to increase tension. The channels (somewhere in the distal ends of the stereocilia) are shown in this figure to be at the lower end of each tip link, but equally well could be at the upper end, or, indeed, both. (D) The sensitivity curve, relating displacement of the bundle (measured at the tips) to the probability of the channel being open, shows the very narrow operating range of hair cells. A displacement of about a third of a micrometre - approximately the diameter of one stereocilium - is a saturating stimulus. The curve also shows that some channels are open in the resting position, suggesting some resting tension in the tip links.
sacculus, an organ which might not be particularly adapted to high-frequency operation, suggest a latency of 40 ~s at 22°C, which can be expected to reduce to 13 ~s at 37°CTM. This very short latency severely constrains models of channel operation. For instance, it rules out models in which the current is modulated in response to a second messenger inside the ceil. The most likely hypothesis is that the channel is simply and directly opened mechanically by the stereociliary movement. A biophysical model of transduction which successfully explains the channel kinetics was put forward by Corey and Hudspeth 19. They suggested that each channel was pulled open by an elastic link, known as the 'gating spring'. Each channel with its associated spring has an energy profile with energy wells representing open- and closed-state conformations. The Boltzmann distribution governs the probability of channels being in the open state, and the energy levels are such that when no tension acts on the gating spring the closed state is favored. Tension on the gating spring changes the energy profile of the system, favoring the open state (Fig. 3). Since channel opening is a stochastic process, the stimulus merely causes a redistribution between the open and closed states and there is no ultimate 256
What are the structural e l e m e n t s involved in transduction? Still unclear in this model are the location of the channel and what pulls on it. In order to determine the location, Hudspeth explored around a hair bundle with an extracellular electrode while deflecting the stereocilia with a probe; the extracellular potential changes were greatest around the top of the bundle, along the tips of the stereocilia 22. This suggests that the transducer currents were flowing into the cell near or at the tips of the stereocilia. Although challenged by Ohmori23, this site has been confirmed by calcium imaging24 and by spatially localized iontophoresis of the channel blocker, gentamicin 25. Important structural evidence came with the discovery of the tip links by Pickles, Osborne and Corals5'21'26, which has been confirmed by others 27'28. The tip link has a fine central filament (Fig. 4), presumably protein, that is surrounded by glycocalyx29'3°. This filament would be ideal for transferring the movement-induced forces onto single mechanotransducer channels in the membrane. There is also a report that, in some cases at least, the tip link is not a single strand but branches at its upper end (Lim, D. J. and Kachar, B., unpublished observations). The upper end attaches to a density which bridges the plasmalemma to the central actin paracrystal of the stereocilium (Fig. 4). At the lower end of the tip link the stereociliary membrane is often seen to be pulled into a point. This is possibly due to differential shrinkage in fixation; however it does suggest that the link here is not directly attached to the actin paracrystal, but rather attaches indirectly via distensible connectors within or underlying the membrane31. The tip link is in an ideal position to detect the shear or sliding movement between the different rows of TINS, VoL 15, No. 7, 1992
stereocilia. Because the stereocilia are graded in height and the tip links are oriented nearly vertically, the links would be stretched by a deflection of the stereocilia in the direction of the tallest, that is, in the excitatory direction. Deflection in the opposite direction would be expected to take the tension off the tip links. A model for transduction The mechanical gating of channels, their localization and the function of tip links were put together in an appealingly simple model for transduction by Pickles, Osborne and Comis5'21'26. In this model (Fig. 3) the tip link is the gating spring and pulls directly on the ion channels. It is not clear whether the transducer channels are associated with the upper or lower ends of the tip link, or both. Either end of the links would be suitable for transferring forces to membrane-bound channels. The close correlation between the anatomical and the electrophysiological evidence supports this model; for instance, the tip links run almost exclusively along the axis of bilateral symmetry of the hair cell, that is, only along the anatomical axis along which movement of the stereocilia affects the transducer channels since the stereocilia are hexagonally packed, other directions might be equally possible6'32. Tip links seem to be present in all mechanotransducing hair cell systems, having been found in vestibular, cochlear and lateral line systems, and in representatives of all vertebrate classes 6'13'27'33'34. A gradation in heights of the stereocilia is required to accommodate the tip links, and such a gradation is seen in all mechanotransducing acousticolateral hair cells, although the gradation is not seen in ciliated electroreceptors, which are thought also to be of acousticolateral origin 35. Recently, this model has been directly tested by cutting the tip links2o Reducing the calcium concentration around the bundle with the fast calcium chelator BAPTA [1,2-bis(o-aminophenoxy)ethaneN,N,N',N'-tetraacetic acid] irreversibly destroyed transduction in a few tenths of a second. A similar treatment with BAPTA for a few seconds eliminated the tip links, viewed with either scanning or transmission electron microscopy2. When freestanding bundles were treated with BAPTA, they relaxed forward by about a tenth of a micrometre 2, in quantitative agreement with a model for tension regulation in the gating springs 36. These results indicate that tip links indeed convey tension to the transduction channels.
An adaptation mechanism regulates tension in the tip links The extraordinary sensitivity of the transduction mechanism poses a problem for the hair cell, in that larger stimuli easily saturate the transduction. This is a problem for hair cells in the vestibular system that sense acceleration, because the constant acceleration of gravity imposes a stimulus a thousandfold larger than their threshold sensitivity. On a different timescale, this high sensitivity creates a developmental problem for all hair cells, because tip links and channels must be positioned within the hair bundle with an accuracy of 10-20 nm. As a consequence, hair cells have developed an adaptation mechanism that continuously adjusts the TINS, Vol. 15, No. 7, 1992
Fig. 4. Transmission electron micrograph of the tip of a stereocilium in the chick basilar papilla, showing the central filament of the tip link (arrow), densities underlying the upper and lower points of attachment (arrowheads), and the band of lateral links between the stereocilia (double arrowhead). The glycocalyx which surrounds the central filament of the tip link has not been preserved here.
tension in the tip links. In vitro experiments initially showed that steady displacements of a hair bundle caused the sensitivity curve to move along the displacement axis to match the new position of the bundle without appreciably changing shape (Fig. 5) 16. For saturating displacement stimuli, this both reduces the response and repositions the maximum sensitivity at the steady bundle position. The mechanism is apparently active in vivo, at least in the bullfrog saccule where it is thought to underlie an adaptation of the saccular microphonic potential aT. This shift of the sensitivity curve can be reconciled with the tip-links model for transduction by supposing that adaptation involves an adjustment of tension in the tip link, specifically by sliding the upper attachment point of the tip link along the side of the stereocilium (Fig. 5)38. More recently, a quantitative model for adaptation has been developed, based on an active motor element 36. This model can account for a variety of phenomena associated with adaptation, and also predicts active movements of the hair bundle under certain conditions. Such movements of the whole bundle have been observed, which tends to confirm the mechanical basis of adaptation 36. Interestingly, the rate of tension adjustment depends on the concentration of Ca2÷ inside the tips of 257
A /~m0.~[ _~
C
1
hypothesis, it may also be that the antibody recognizes latent channels and not those mechanically connected to tip links, or that the epitope recognized might not be the transduction channel at all. Towards a molecular basis
I
I 100
0
B
lOT//-
ms
--
I /--/ 0
110
Bundle Displacement
{pro)
Fig. 5. Adaptation by adjustment of tip-link tension. (A) A displacement of the bundle opens channels, but they close spontaneously in tens of milliseconds. (B) The closure is due to a shift of the sensitivity curve in the direction of the displacement, suggesting an adjustment of mechanical tension on channels. (C) A way to reconcile a variable tension with the tip-links model is to suppose that the upper tip-link attachment can move along the side of the stereocilium, slipping down to loosen when stretched or climbing up to tighten when relaxed.
the stereocilia, suggesting that the adjustment motor is in the tips very near the transduction machinery. Indeed, displacement of the upper tip-link insertion site has been observed in hair cells fixed in various adaptation states, which is in agreement with the model and confirms the involvement of the tip links 39.
The next level of understanding will entail a molecular description of the elements in the transduction chain. The excitement of tracing the basis of hearing to a few molecules will be tempered by the great difficulty in purifying the relevant proteins, particularly since we expect them to be present in amounts of the order of an attomole per animal45'46. The channel, the tip link and the motor remain biochemically unidentified at present. Of these, the motor molecule responsible for adaptation may be the most tractable, since it resembles the family of myosins in its properties. The actin core of stereocilia provides the first clue, since the only molecules known to move on actin are of the myosin family. The maximum rate of adaptation that tightens the tip links is in the range of 1-2 ~m~/s, the same rate at which myosin moves on actin, and myosin-coated glass beads have been found to climb towards the tips of stereocilia cores at a similar velocity47. However, immunological and biochemical approaches to identifying the motor are still inconclusive. Perhaps the best hope for identifying the channel is by molecular probing for similarity to other ion channels 4a. In 1962 Georg von Bekesy concluded several decades of work on the cochlea by suggesting that 'it does not seem impossible that the final mechanical transformer is of molecular dimensions'. Now, thirty years later, we think we know where that mechanical transformer is, and how it works. Hair cell research is presently at a turning point, as efforts are directed towards identifying and understanding the 'transformer' in molecular terms.
Alternative models for transduction
No alternative models for hair-cell mechanotransduction, beyond that presented here, have been put Selected references 1 Corey, D. P. and Hudspeth, A. J. (1979) Nature 281, forward in the past decade. Earlier, it was suggested 675-677 that transduction might depend on distortion of the 2 Assad, J. A., Shepherd, G. M. G. and Corey, D. P. (1991) whole hair-cell body4° (which has been ruled out by Neuron 7,985-994 3 Flock, A., Flock, B. and Murray, E. (1977) Acta Otolaryngol. experiments in which mechanotransduction was as83, 85-91 sociated with micromanipulation of the stereocilial'8), Tilney, L. G., Egelman, E. H., DeRosier, D. J. and Saunders, or of the kinocilium alone41 (which has been ruled out 4 J. C. (1983) J. Cell Biol. 96, 822-834 by experiments in which the stereocilia were manipu5 Pickles, J. O., Comis, S. D. and Osborne, M. P. (1984) Hear. lated separately from the kinocilium, showing that Res. 15, 103-112 transduction depends only on the formerS). Sugges- 6 Pickles, J. O. etaL (1989) Hear. Res. 41, 31-42 7 Hudspeth, A. J. and Corey, D. P. (1977) Proc. NatlAcad. 5ci. tions that mechanotransduction depended on memUSA 74, 2407-2411 brane distortion at the sites at which the stereocilia 8 Hudspeth, A. J. and Jacobs, R. (1979) Proc. Natl Acad. 5ci. flexed in response to deflection, that is, at their bases USA 76, 1506-1509 where they enter the cell body42, were countered by 9 Shotwell, S. L., Jacobs, R. and Hudspeth, A. J. (1981) Ann. NY Acad. 5ci. 374, 1-10 the experiments described above suggesting that the 10 Lowenstein, O. and Wers&ll, J. (1959) Nature 184, 1807-1808 channels were instead located near the tips of the 11 Ohmori, H. (1985) J. PhysioL 359, 189-217 stereocilia22, 24, 25. 12 Crawford, A. C., Evans, M. G. and Fettiplace, R. (1991) It was found recently that antibodies to an J. Physiol. 434, 369-398 amiloride-sensitive ion channel from renal epithelium 13 Hudspeth, A. J. (1989) Nature 341,397-404 bind near the tips of the stereociliaa3. Since amiloride 14 Holton, T. and Hudspeth, A. J. (1986) J. PhysioL 375, 195-227 also blocks transduction channels a4, although at a 15 Russell, I. J. (1983) Nature 301,334-336 higher concentration than that required to block the 16 Howard, J. and Hudspeth, A. J. (1988) Neuron 1, 189-199 renal channel, the antibody might recognize the hair- 17 Dannhof, B. J. and Bruns, V. (1991) Hear. Res. 53,253-268 cell transduction channels. Antibody binding occurred 18 Corey, D. P. and Hudspeth, A. J. (1979) Biophys. J. 26, mainly at the points of contact between stereocilia and 19 499-506 Corey, D. P. and Hudspeth, A. J. (1983) J. Neurosci. 3, not at either end of the tip links. Although this finding 962-976 is obviously inconsistent with the simple tip-links 20 Sellick, P. M., Patuzzi, R. and Johnstone, B. M. (1982) 258
TINS, VoL 15, No. 7, 1992
J. Acoust. Soc. Am. 72, 131-141 Pickles, J. O. (1985) Prog. NeurobioL 24, 1-42 Hudspeth, A. J. (1982) J. Neurosci. 2, 1-10 Ohmori, H. (1988)J. Physiol. 399, 115-137 Huang, P. L. and Corey, D. P. (1990) Biophys. J. 57, 530a Jaramillo, F. and Hudspeth, A. J. (1991) Neuron 7, 409-420 Osborne, M. P., Comis, S. D. and Pickles, J. O. (1984) Cell Tissue Res. 237, 43-48 27 Little, K. F. and Neugebauer, D-Ch. (1985) Cell Tissue Res.
21 22 23 24 25 26
242,427-432 28 Furness, D. N. and Hackney, C. M. (1985) Hear. Res. 18, 177-188 29 Osborne, M. P., Comis, S. D. and Pickles, J. O. (1988) Hear. Res. 35, 99-108 30 Neugebauer, D-Ch. and Thurm, U. (1987) Cell Tissue Res. 249, 199-207 31 Pickles, J. O., Rouse, G. W. and von Perger, M. (1991) Scanning Microsc. 5, 111 5-1128 32 Comis, S. D., Pickles, J. O. and Osborne, M. P. (1985) J. Neurocytol. 14, 113-130 33 Rhys-Evans, P. H., Comis, S. D., Osborne, M. P., Pickles, J. O. and Jeffries, D. J. R. (1985) J. Laryngol. Otol. 99, 11-19 34 Rouse, G. W. and Pickles, J. O. (1991) J. NlorphoL 209,
111-120 35 Szabo, T. (1974) in Handbook of Sensory Physiology 3 (Fessard, A., ed.), pp. 13-58, Springer 36 Assad, J. A. and Corey, D. P. J. Neurosci. (in press) 37 Eatock, R. A., Corey, D. P. and Hudspeth, A. J. (1987) J. Neurosci. 7, 2821-2836 38 Howard, J. and Hudspeth, A. J. (1987) Proc. NatlAcad. Sci. USA 84, 3064-3068 39 Shepherd, G. M. G. et al. J. Gen. Physiol. (in press) 40 Stevens, S. S. and Davis, H. (1983) Hearing: Its Psychology and Physiology, Wiley 41 Hillman, D. E. (1969) Brain Res. 13, 407-412 42 Dallos, P. (1973) The Auditory Periphery, Academic Press 43 Hackney, C. M., Furness, D. N. and Benos, D. J. (1991) Scanning Microsc. 5, 741-746 44 Jorgensen, F. and Ohmori, H. (1988) J. Physiol. 403, 577-588 45 Shepherd, G. M. G., Barres, B. A. and Corey, D. P. (1989) Proc. Natl Acad. Sci. USA 86, 4973-4977 46 Gillespie, P. G. and Hudspeth, A. J. (1991)J. Cell Biol. 112, 625-640 47 Shepherd, G. M. G., Corey, D. P. and Block, S. M. (1990) Proc. Natl Acad. Sci. USA 87, 8627-8631
Acknowledgements Work in the authors' laboratories was supported by the MRC of the UK and the Australian ResearchCouncil (J.O.P.)and by the Nab'onallnstitutes of Health, the Office of Naval Research, and the Howard Hughes Medical Institute (D.P.C.).
Changesin Caz +-bindingproteinsin human neurodepenerativedisorders Claus W. Heizmann
The cellular distribution of Cae+-binding proteins has been extensively studied during the past decade. These proteins have proved to be useful neuronal markers for a variety of functional brain systems and their circuitries. Their major roles are assumed to be Ca e+ buffering and transport, and regulation of various enzyme systems. Since cellular degeneration is accompanied by impaired Ca e+ homeostasis, a protective role for cae+-binding proteins in certain neuron populations has been postulated. As massive neuronal degeneration takes place in several brain diseases of humans, such as Alzheimer's disease, Parkinson's disease and epilepsy, changes in the expression of Ca 2+binding proteins have therefore been studied during the course of these diseases. Although the data from these studies are inconsistent, the detection and quantification of Cae+-binding proteins and the neuron populations in which they occur may nevertheless be useful to estimate, for example, the location and extent of brain damage in the various neurological disorders. I f future studies advance our knowledge about the physiological functions of these proteins, the neuronal systems in which they are expressed may become important therapeutical targets for preventing neuronal death in an array of neurodegenerative diseases. More than 7000 articles on calcium were published in 1991 (a figure found using the Medline database, with 'calcium' as the search word). This emphasizes the tremendous interest and progress in Ca2+-related research. In nerve cells, Ca 2+ ions activate and regulate a number of key processes, including fast axonal transport of substances, synthesis and release of some neurotransmitters, and membrane excitability. It is also thought that in the CA1 hippocampal region of the brain, Caz+ might induce long-term potentiation and memory storage mechanisms. The Ca2+ message is converted into an intracellular response - in many cases by CaZ+-binding proteins TINS, Vol. 15, No. 7, 1992
and Katharina Braun
that are involved in a wide variety of activities, such as cytoskeletal organization, cell motility and differentiation, cell-cycle regulation, and Ca2+ buffeting and transport. It might therefore be possible that altered levels of some Ca2+-binding proteins (e.g. due to deletion or mutation of the corresponding genes) could lead to an impaired Ca2+ homeostasis in nerve cells and to pathological conditions. Using mostly immunohistochemical techniques, several research groups have now started to search for altered expression of the CaZ+-binding proteins parvalbumin, calbindin-D28K and S100 in affected brain regions of patients suffering from acute insults, such as stroke and epileptic seizures, and from chronic neurodegenerative disorders, such as Alzheimer's, Huntington's, Parkinson's and Pick's diseases. In addition, altered Ca2+ levels have been found in platelets of patients with bipolar affective disorders 1, and Ca 2+ antagonists have been suggested for treatment of psychotic depression. The few proteins that have been investigated in these pathological states belong to a large family of more than 200 members. These proteins are characterized by a common structural motif, the EF-hand, which is present in multiple copies and binds Ca2+ selectively and with high affinity. Each of these domains consists of a loop of 12 amino acids that is flanked by two ochelices. This structural principle was first identified as a result of studying the crystal structure of the Ca2+-binding protein parvalbumin isolated from carp, and is designated the EF-hand because of the arrangement of the E and F helices of parvalbumin. A consensus amino acid sequence for this motif has aided the identification of many new members of this protein family2'3. Ca2+-binding proteins with known functions, such as calmodulin, troponin C, myosin light chains, calpain, calcineurin and recoverin, are far outnumbered by those with unidentified roles. Most Ca2+-
© 1992, ElsevierSciencePublishersLtd,(UK)
Claus W. Heizrnannis at the Dept of Pediatrics, Division of Clinical Chemistry, Universityof Zurich, 5teinwiesstr. 75, CH-8032 Zurich, Switzerland,and
Katharina Braun is at the Institute for Neurobiolosy, Brenneckestr. 6, 0-3090 Magdeburg, FRG.
259
