JOURNALOFNEUROPHYSIOLOGY Vol. 68, No. 3, September 1992. Printed
in U.S.A.
Mechanical Response of Frog Saccular Hair Bundles to the Aminoglycoside Block of Mechanoelectrical Transduction WINFRIED DENK, ROBERT M. KEOLIAN, AND WATT W. WEBB School of Applied and Engineering Physics and Department of Physics, Cornell University, Ithaca, New York 14853 SUMMARY
AND
CONCLUSIONS
I. Deflections of the mechanosensory hair bundles on frog saccular hair cells were measured interferometrically, with submillisecond temporal and submicrometer spatial resolution, and with subnanometer displacement sensitivity. 2. The direction of the initial bundle deflection (toward the taller stereocilia) in response to a sudden application of aminoglycoside antibiotics shows that the mechanosensory channels are blocked in their mechanically open state. 3. The magnitude of the initial deflection is consistent with published data on the gating swing as derived from the gating compliance. 4. A delayed relaxation and frequently a reversal of the initial deflection were observed and are attributed to the previously reported mechanical adaptation mechanism, which is at least partially controlled by the influx of Ca2+ through the transduction channels. 5. Increases of low-frequency spontaneous motion were found at intermediate blocker concentrations. They can be well accounted for by the fluctuating force exerted on the bundle by the random binding and unbinding of blocker molecules. 6. The mechanical response of the hair bundle to aminoglycosides may be related to their acute and specific ototoxicity. INTRODUCTION
mechanical stimulation normally used to open the channels but also if the channels change configuration spontaneously or in response to some other influence. Such a configuration change, or more precisely a trapping in one configuration, is expected to occur on the binding of a chemical blocker molecule such as streptomycin, because streptomycin does occupy, at least partly, the aqueous pore of the transducer channel ( Kroese et al. 1989). The interaction of aminoglycosides with the hair bundle may, furthermore, be related to the ototoxicity of these clinically important antibiotics (Rybak 1986; Sande and Mandell 1985). In this paper we report the results of experiments showing the mechanical effects that aminoglycosides have both on a hair bundle’s average position and on its position fluctuations. METHODS
Displacements and displacement fluctuations were measured at the tips of hair bundles over a wide range frequency range (up to 100 kHz) with a differential laser interferometer (Denk et al. 1989; Denk and Webb 1990). This interferometer uses a lowpower HeNe laser beam that is focused to a diffraction limited spot through a high-power microscope equipped with Nomarski differential interference contrast optics. Over the whole frequency range the displacement sensitivity far exceeds the Brownian motion of a hair bundle. Streptomycin, dihydrostreptomycin, and gentamicin were applied iontophoretically from glass micropipettes that were filled with a 0.5-M solution and their tips placed -5 pm from the bundle. We estimated that an iontophoresis current of 100 pA should lead to a concentration of - 10 PM at the bundle, assuming a diffusion constant of 10-l’ m2 s-’ and a transport number of 10% of that of Cl- (Kroese et al. 1989). The hair cells we used were part of the saccular sensory epithelia from frogs Rana catesbeiana or Rana pipiens, from which monocellular sheets were isolated microsurgically after enzymatic digestion (Corey and Hudspeth 1983; Denk 1989; Denk and Webb 1992). All experimentswere performed in bath solutionscontaining (in mM) 110Na+, 2 K+, 4 Ca2’, 118Cl +, and 3 N-2-hydroxyethylpiperazine-N’-2-ethanesulfonicacid (HEPES) at a pH of 7.25. All displacementmeasurements and spectrawerecalibratedwith the useof an oscillatory stagemotion (Denk and Webb 1990).
Mechanoelectrical transduction in the inner ear is universally performed by sensory hair cells, which convert a mechanical deflection of their sensory organelle, the hair bundle, into a change of intracellular electrical potential (cf. Howard et al. 1988). Hair cells hereby play a pivotal role in the perception of linear and angular acceleration and of sound. Mechanoelectrical transduction involves a straininduced conformational change in the transducer molecule and a subsequent change of membrane conductance; both are presumably performed by a directly strain-activated transmembrane ion channel (Corey and Hudspeth 1979, 1983; Guharay and Sachs 1984; Holton and Hudspeth 1986). Direct strain activation implies, by virtue of reciprocity (Howard and Hudspeth 1988), that channel openings and closings exert a force on the hair bundle thus moving it. RESULTS This gating displacement (Howard and Hudspeth 1988) is The change in the hair-bundle position in response to a a close physical analogue of the electrical charge displacement (gating current) that accompanies the opening and sudden exposure to streptomycin is shown for a typical case in Fig. 1. Within 5-20 ms after the beginning of the iontoclosing of voltage-gated channels (Armstrong and Bezanilla 1973 ) ; it has been indirectly observed through its effect on phoresis pulse, the bundle begins to move in the positive the hair bundles’ compliance (Howard and Hudspeth direction (toward the taller stereocilia) with a speed that is 1988). The gating displacement, and the corresponding roughly proportional to the iontophoresis current ampliforce on the hair bundle, occurs not only in response to the tude. The magnitude of the peak positive excursion, on the l
0022-3077/92 $2.00 Copyright0 1992TheAmericanPhysiological Society
l
927
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
928
W. DENK,
R. M.
KEOLIAN,
AND
W. W. WEBB
1979), then the channel seems to have a continuous distribution of states between completely closed and completely 1nA open, which can be parametrized by the open probability (p,) and an accordingly varying conductance and length, 3nA both proportional to po. In this view, which is depicted in Fig. 2, it would indeed appear as if the binding of a blocker 5nA molecule forces the channel from a configuration with, say, PO= 0.3 into a configuration with p. = 1. Measurements at larger iontophoresis current amplitudes and for a streptomycin pulse of long enough duration (> 100 ms) show that the initial displacement is followed by 6 a motion in the negative direction, which often took the 1 4 bundle past its resting (prepulse) position, sometimes as far w as 200 nm. In every case the bundle eventually returned to 2 2 its resting position, always within several seconds after the 2 0 end of the pulse. We attribute this displacement reversal to -2 the mechanical adaptation mechanism, which changes the transduction setpoint (Eatock et al. 1987) and the un0.0 0.5 1.0 1.5 2.0 2.5 loaded bundle position (Howard and Hudspeth 1987) in Time (s) response to static stimuli. Adaptation in saccular hair cells FIG. 1. Reaction of saccular hair-bundle deflection (A ) to concentration pulses of streptomycin delivered iontophoretically by current pulses, has been shown to be controlled in part by the influx of channels (Hacohen et al. which are shown in B. Displacement traces were offset from each other to Ca2+ through the transduction avoid overlap. Positive displacement is toward the taller stereocilia. 1989). Because the block of the transduction channel prevents this influx of Ca2+, mimicking a closed channel, it other hand, does not increase further as the current exceeds induces the adaptation mechanism to tighten the gating 1 nA. The response to short iontophoresis pulses or the string in a “futile attempt” to reopen the channels initial reaction to longer pulse was toward the taller stereo- (Fig. 2C). cilia in all but 1 out of 27 cells with an mean excursion of We confirmed that the response direction was indepen12.6 t 10 (SD) nm, averaged over the largest observed dis- dent of the location of the pipette relative to the hair bunplacements for each cell. The largest observed excursion for dle, ruling out the possible artifact that the motion might be caused by electroosmotic flow out of the pipette tip. No any cell was 40 nm. A value 6x = d( 1 - pz) K~Y /( K,, + 7 ‘K~) is expected; 6x = 19 nm, assuming a swing of the gate of 4 response could be elicited from cells that had been shown to react to streptomycin iontophoresis if current was instead nm and a gating spring stiffness Kg of 500 pN/ m, as Howard and Hudspeth ( 1988 ) deduced from their measurements of passed out of pipettes containing only 3 M NaCl or K acethe gating compliance, a geometric gain y of 0.14, a stiffness tate. Measurements of the displacement of a compliant glass fiber attached to the bundle closely resembled what K,, of 4.9 ~,LN/ m per stereocilium as measured by Howard and Ashmore ( 1986), a resting open probability pz of 0.3, was observed by focusing on the hair bundle directly, ruling and a single transduction channel per stereocilium. Our out spurious displacement signals due to changes of the sample of cells was not necessarily representative because bundles’ optical properties that might have been induced in bundles were selected for good visibility and were rejected if the bundle by streptomycin. Alternating use of gentamicin and dihydrostreptomycin, out of a double-barreled microno response at all was seen to an iontophoretic current pulse. Motion in the positive direction is expected, as can be Ca++ seen from Fig. 2, A and B, if the channel is blocked in the mechanically open position by the streptomycin molecule. The experimental results thus strongly support the notion that streptomycin, most likely by occupying part of the lumen of the channel pore, prevents the channels from switching into a mechanically closed configuration. The channel is still electrically closed because the passage of ions through the channel is prevented by the presence of the streptomycin molecule. It should be pointed out that the act of binding of the blocker molecule does not itself have to exert a force on the C A B channel. On the contrary, we have assumed that the blocker FIG. 2. Schematic illustration of the interaction of blocker molecules molecule binds only to the open state, without changing the with the transduction channel. How the binding of a blocker molecule (B) channel’s conformation. What moves the bundle then is 1st leads to a positive displacement is shown in B. After the adaptation the energy that is stored as gating spring tension, which now mechanism, represented by the rack and pinion inside the taller stereocilcannot be relaxed by a closing and shortening of the trans- ium (S), has had time to react, the hair bundle is pulled ( C) in a negative past its original resting position (A ). For the unperturbed state duction channel. On the other hand, if one looks at the direction, (A), the gate, representing the strain-activated channel, is depicted as transduction channel with a time resolution much slower about halfway open, which is intended to represent its averaged position, than its flicker frequency of - 10 5 s-l (Corey and Hudspeth rather than an actual state.
1F’s 11
,I,,
,I,,
,,,I
I,,,
,,I,
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
HAIR-BUNDLE
MOTION
DUE TO TRANSDUCTION
electrode, produced almost identical responses, which also were very similar to those typically elicited by streptomycin. The response to dihydrostreptomycin pulses was reversibly abolished when 100 PM streptomycin was added to the bath. It had been shown previously by Howard and Hudspeth ( 1988 ) that gentamicin (and presumably other aminoglycosides as well) abolishes the gating compliance. This result is an indication that gentamicin blocks the transition between the states of the channel molecule but, by itself, does not answer the question into which state the channel is locked. As a consequence of this loss of the gating compliance on aminoglycoside exposure, the Brownian motion of the bundle was expected to be decreased. We found this to
BLOCK
929
be the case. In addition we observed, as is shown in Fig. 3A, that the spontaneous motion decreased only in the stiffnesscontrolled region (below the roll-off) but was unchanged in the damping-controlled region (well above the roll-off). The range of reduction factors that was found in response to the saturating iontophoretic application of streptomycin (up to ~2 as shown in Fig. 3) is consistent with an elimination of the gating compliance, but the interaction of streptomycin with the adaptation mechanism may play an additional role in changing the bundle stiffness. Very similar results were obtained when streptomycin was added to the bath in a few cases instead of applying it iontophoretically. For a blocker concentration below the saturation regime, when only part of the channels are blocked on average, the B 10
lo2
0.5
lo3
Freq.
r
(Hz)
1.0
P~dstI-,p/P=4v/o -1
-kbCB=OS
\
... ... .... ... I
-
400 pA
IIIIII
I
I
lIll
lo1
-0 24
Td \
r
-
I
1111111
l
. 30Hz x 500Hz
I
Ill11
I
IlIll
11111111
lo2
Freq.
(Hz)
0
E
I
lo1
102
Freq. 5
-kbC!B=looS;: -kbCB=%oS
103
(Hz)
l”tF-
K3-
e2 s mm ml-r P(
0 x 0
I -200
8 0
200
Iiont.
400
lo2
lo1
Freq.
lo3
(Hz)
FIG. 3. Suppression of the gating compliance leads to a decrease in the Brownian motion of a saccular hair bundle as is shown in A with the use of spectra take in the unperturbed state ( 1), during exposure (2) to a saturating (2 100 PM) concentration of dihydrostreptomycin, and 10 min ( 3) after washing. Note the merging of all 3 curves at high frequencies, indicating that bundle damping is unaffected. The distribution of the factors by which the power spectral density below the roll-off was decreased for different cells is shown in B. At intermediate blocker concentrations, a Brownian motion increase is observed up to a roll-off frequency of -50 Hz, which is determined by the blocker’s binding kinetics. Spectra are shown in C for iontophoresis currents of 0, 150, and 400 pA. Corresponding model calculations are shown in D. Relative changes inthe power spectral density (psd) for 2 different frequencies (30 and 500 Hz) are plotted as a function of the iontophoresis current in D together with the result of model calculations as described in the text. Observed (ragged curves) and modeled (smooth curves) spectral changes are compared in F.
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
930
W. DENK,
R. M.
KEOLIAN,
random chemical binding and unbinding of blocker to the channels causes a fluctuating force on the hair bundle. This is the same force that, when exerted simultaneously by all transduction elements, is thought to cause the positive displacement step at the beginning of a streptomycin pulse in Fig. 1. The frequency spectrum of this fluctuating force is determined by the chemical kinetics of binding and release of streptomycin. Figure 3C (middle trace) shows the power spectrum for the spontaneous bundle deflections at a dihydrostreptomycin concentration that was estimated to be - 15 PM. Given the uncertainty in this estimate, this is in reasonable agreement with the experimentally determined (Kroese et al. 1989) binding constant ( -8 ,uM) of dihydrostreptomycin to the transduction channel. The increase in the spontaneous low-frequency motion is combined with a reduction at intermediate frequencies ( 100 Hz to 1 kHz), which is about one-half as large as the ultimate reduction observed at a saturating concentration. In Fig. 3 E the change in spontaneous motion’s spectral density as a function of iontophoresis current is plotted both for 30 and 500 Hz. The horizontal scale was adjusted by eye for the best possible agreement with the data such by supposing that 150 pA corresponds to a forward binding rate of khcR = 100 s-’ (with k, and cRthe rate constant and the blocker concentration, respectively). A mathematical model for the mechanical displacement noise due to aminoglycoside binding fluctuations, which is detailed below, describes reasonably well the magnitude and the frequency distribution of the changes occurring in spectral shape (see the smooth lines in Fig. 3 F). Several factors are likely to contribute to the discrepancy of the absolute magnitudes (compare Fig. 3, C and 0): 1) this bundle was smaller than average, and the amplitude of the power spectrum scales linearly with the number of stereocilia; 2) the motion was measured at the tip of the tallest stereocilium, which is - 1.3 times taller than the kinocilium, to which most of the published values are referenced, and the amplitude of the power spectrum scales quadratically with this factor; 3) hair-bundle displacement spectra often rise toward lower frequencies even far below the rolloff, as reported in an earlier paper (Denk et al. 1989). The behavior of the bundle position’s fluctuation power spectrum can be understood intuitively by attributing the increase of the amplitude at low frequency to fluctuating forces caused by blocker binding and unbinding and the decrease at intermediate frequencies to a partial abolition of the gating compliance. A more quantiative test of the notions about how the hair-cell mechanics work requires the analysis of a more detailed physical model. To obtain the spontaneous fluctuations we first calculated the bundle compliance as a function of frequency; its dissipative part then provides the fluctuation spectrum via the fluctuation-dissipation theorem (Callen and Welton 195 1; Denk et al. 1989). The hair bundle was modeled as described by Howard and Hudspeth ( 1988). The total compliance (which is the reciprocal of the total stiffness KT) is given by 1 /K7,
=
iWV
+
K,
+
r2NK
(1)
with o the frequency, Y the damping ( Denk et al 1989), K, the bundle stiffness wi thout the gating elements, Y the ratio
AND
W. W. WEBB
between tip link extension and bundle tip displacement, N the number of stereocilia in a single bundle, Kg the stiffness of the gating spring, and d the change in effective channel length. Next we consider L&/al, the dependence of the closed probability pc on the total length I, to which the transduction element (gating spring and channel combined) is stretched. The channel blocker, present with a concentration c,, is assumed to bind only to the open state with binding and unbinding rate constants of k, and k,, respectively, yielding the following rate diagram k4 0
closed
k of the length of the transduction element from its average value ((I)) are considered. For dk,/dl, which represents the strain sensitivity of the channel, we use ak,/al = k,Kgd/k,T (using Eq. 5 of Howard and Hudspeth 1988, with k, and T as the Boltzmann constant and the absolute temperature, respectively). We then go to the frequency domain and use Eq. 2c to eliminate pC from Eq. 2b, which then becomes iu+kbcB+k,--
k,kbcB = kg, io + k,
1
+ Cl-
(I))(P,)
%
(3)
We then use Eq. 3 to eliminate pn from Eq. 2a and take the limit of large k, and k, reflecting almost instantaneous equilibration between open and closed states, an assumption that is warranted by transition rates of - lo5 s-l (Corey and Hudspeth 1983). To simplify the notation we also introduce the quantity px = ( 1 + k,/ k,)-’ , which becomes equal to the open probability p. = [ khcB/ k, + pi1 1-l without any blocker present. We finally get dPC -=-
al
Kgkupx( 1 - px)(io + k, + kbcB)d kgT( k, + pxkhcB)( iw + k,, + pxkhcR)
(4)
To formalize the assumption that the adaptation mechanism, by tightening the gating strings, changes the balance between the channel opening (k,) and closing (k,) time constants in such a way as to keep the open probability (p,) constant, we use the relation px = [max(
I/ P: - kbq&),
I]-’
(5)
where p: is the unperturbed open probability. As a result the closed probability pCgradually decreases with increasing blocker concentration (c,). When pCLas thus been reduced to zero, p. cannot be kept constant any longer and suddenly starts to fall if cB continues to rise, resulting in a kink in the theoretical curve. This kink, which also marks the complete
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
HAIR-BUNDLE
MOTION
DUE TO TRANSDUCTION
BLOCK
931
elimination of the gating compliance, occurs at forward binding rate (k,c,) of - 190 s-l, which corresponds to 285 pA on the experimental scale, after the horizontal scale for the theoretical curve had been adjusted to agree as well as possible with the data. Our assumption about the control of the adaptation mechanism is probably not entirely precise; for example, Assad et al. ( 199 1) find the adaptation mechanism controlled by both, Ca2+ concentration and tension, whereas we only consider the effect of Ca2+. A subtle but important point concerning the mechanical equivalent of the transduction element should be mentioned here. If a single transduction element were held under “force (F) -clamp” conditions, its slope stiffness (K :I), defined as the ratio of a small applied force change to the resulting change in the average length of the transduction element, would be given by
overall hair-bundle mechanics. There can, furthermore, remain little doubt that aminoglycosides bind to the channel in a way that prevents it from closing and from shortening, as visualized with the use of the picture developed by Hudspeth and co-workers (Corey and Hudspeth 1979, 1983; Howard and Hudspeth 1987, 1988) (see Fig. 2). The slow response to steps in aminoglycoside concentration, including its time course, direction, and magnitude, can be explained by assuming that aminoglycosides interfere with the mechanical adaptation mechanism by blocking the entry of Ca2+ ions through the transduction channels. Assad et al. ( 1989) have similarly attributed the bundle deflection they observed in response to a stepping of the hair cell’s membrane voltage to +80 mV to prevent the influx of Ca2+ ions, in their case due to the removal of the electrochemical driving force. The recovery we observed at the end of a large streptomycin pulse appears to be too slow h’;: = [ 1/KR + d (dy&YF)]-’ (6) to be accounted for by either the unbinding rate constant of 100 s-l, which we did find to provide the best fit of the withdp,ldF= d*p,( 1 -p,)l(ksT). Thisisclearlydifferent from the slope stiffness under “length (/)-clamp” condifluctuation data (Fig. 3C), or by the intrinsic speed of the tions (K$), defined as the ratio of the change in the average adaptation mechanism (Assad et al. 1989), or by simple diffusion. The local streptomycin concentration may, howforce exerted by the transduction element on its attachment ever, decay slower because of nonspecific binding to mempoints in response to a small imposed change in the length branes (A. J. Hudspeth, personal communication). It canof the transduction element not be ruled out completely that aminoglycosides do, in K!/= KJ I - Kgd2p,(1 - /I,)/( kJ)] (7) addition, interact with mechanical adaptation in ways which we find by combining the last term of Eq. I with Eq. other than by blocking Ca2’ entry through the transduction 4 while setting cn = 0. Length-clamp conditions were used channels. Further clarification and more sensitive tests of our interby Howard and Hudspeth ( 1988) to calculate the transducpretation could be obtained from experiments combining tion elements’ stiffness. The apparent stiffness of the transduction element thus whole-cell voltage clamping with the sensitive displacement experiments depends on the mechanical impedance it is coupled to. The measurements described here. Preliminary showing the effect of streptomycin on the correlation bemodel we used makes, like Howard and Hudspeth ( 1988), the implicit but incidentally correct assumption that the tween electrical and mechanical noise in hair cells (Denk 1989) have been encouraging in this respect. individual transduction element operates in a “stiff” enviOne could finally ask whether the strong mechanical reronment. This assumption does, however, become invalid sponse of the hair bundle might provide a mechanism for if the number of stereocilia in hair bundle becomes too the specific ototoxicity of the aminoglycosides (Ryback small. Although for the “soft” environment (force clamp) 1986). Because extended aminoglycoside exposure leads to the apparent gating-element stiffness KF would behave like conspicuous morphological changes in the hair bundle, the gating spring in series with another spring representing such as the fusion of stereocilia, ototoxicity might, literally, the channel, we find for the stiff case (length clamp) that K$can even become negative, for perfectly reasonable values be due to mechanical self destruction of the hair bundle as a of Q, d, and p. (such as 500 pN/m, 6 nm, and 0.5). As a result of the adaptation motors pulling incessantly at the consequence, the total bundle stiffness might be decreased gating strings. The same mechanism might mediate the destruction of transduction (Crawford et al. 199 1) and the to a value below what it would be after cutting the transduction elements. For this to be valid, there is no need to as- snapping of the tip links (Assad et al. 199 1) seen in low sume that the gating spring can sustain a force under com- external Ca 2+ concentrations. pression if the free energies of the channel states are such We thank Dr. David Corey for comments on the manuscript. that the channel is shut for zero tension in the gating spring. This work was supported by grants from, AT&T Bell Laboratories, OfTo optimize the agreement between data and theory for fice of Naval Research, National Science Foundation, and National Instithe spectral changes induced by aminoglycosides, we ad- tutes of Health (NIH). W. Denk was an IBM predoctoral fellow and W. W. justed some of the parameters somewhat from their pub- Webb a NIH Fogarty Scholar-in-Residence during some of this research. Address for reprint requests: W Denk, AT&T Bell Labs, Murray Hill, lished best values, while staying well within their error margins. For the curves in Fig. 3 we used Kg = 550 pN/m, d = 5 NJ 07974. nm, K~ = 200 pN/m, Y = 60 nN s/m2,pz = 0.35, N= 50, Received 2 February 1992; accepted in final form 1 May 1992. and y = 0.14. Setting k, = 100 s-l best accounted for the REFERENCES position of the roll-off in the low-frequency fluctuations. l
C. M. AND BEZANILLA, F. Currents related to movement of the gating particles of the sodium channels. Nutzrrc Land. 242: 459-46 1, 1973. ASSAD, J. A., HACOHEN, N., AND COREY, D. P. Voltage dependence of adaptation and active bundle movement in bullfrog saccular hair cells. Proc. Nat/. Acad. Sci. USA 86: 29 18-2922, 1989. ARMSTRONG,
DISCUSSION
The observed mechanical responses of hair bundles to aminoglycosides further strengthen the case for a directly gated transduction channel that is tightly coupled to the
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
932
W. DENK, R. M. KEOLIAN,
G. M. G., AND COREY, D. P. Tip-link integrity and mechanical transduction in vertebrate hair cells. Neuron 7: 985-994, 1991. CALLEN, H. B. AND WELTON, T. A. Irreversibility and generalized noise. Physiol. Rev. 83: 34-40, 195 1. COREY, D. P. AND HUDSPETH, A. J. Response latency of vertebrate hair cells. Biophls. J. 26: 499-506, 1979. COREY, D. P. AND HUDSPETH, A. J. Kinetics of the receptor current in bullfrog saccular hair cells. J. Neurosci. 3: 962-976, 1983. CRAWFORD, A. C., EVAN, M. G., AND FETTIPLACE, R. The actions of calcium in the mechano-electrical transducer current of turtle hair cells. .J. Physiol. Lond. 434: 369-398, 199 1. DENK. W. Biophysical Studies c!f‘ Mechano- Electrical Transduction in Hair Cells ( PhD. thesis). Ithaca New York: Cornell Univ., 1989. DENK, W. AND WEBB, W. W. Simultaneous recording of fluctuations of hair-bundle deflection and intracellular voltage in saccular hair cells. In: Cbchlear Mechanics, Structure, Function and Models, edited by .I. P. Wilson and D. T. Kemp. New York: Plenum, 1989, p. 125-l 34. DENK, W. AND WEBB, W. W. Optical measurements of picometer displacements of transparent microscopic objects. Appl. Optics 29: 2382-239 1, 1990. DENK, W. AND WEBB, W. W. Forward and reverse transduction at the limit of sensitivity studied by correlating electrical and mechanical fluctuations in hair cells. Hear. Res. 60: 89- 102, 1992. DENK, W., WEBB, W. W., AND HUDSPETH, A. J. The mechanical properties of sensory hair bundles are reflected in their Brownian motion measured with a laser differential interferometer. Proc. Natl. Acad. Sci. USA 86: 5371-5376, 1989. EATOCK, R. A., COREY, D. P., AND HUDSPETH, A. J. Adaptation of mechanoelectrical transduction in hair cells of the bullfrog’s sacculus. J. Neurosci. 7: 282 l-2836, 1987. GUHARAY, F. AND SACHS, F. Stretch-activated single ion channel currents in tissue-cultured embryonic chick skeletal muscle. J. Physiol. Lond. 352: 685-701, 1984. ASSAD, J. A., SHEPARD,
AND W. W. WEBB
N., ASSAD, J. D., SMITH, W. J., AND COREY, D. P. Regulation of tension on hair-cell transduction channels: displacement and calcium dependence. J. Neurosci. 9: 3988-3997, 1989. HOLTON, T. AND HUDSPETH, A. J. The transduction channel of hair cells from the bull-frog characterized by noise analysis. J. Physiol. Lond. 75: 195-227, 1980. HOWARD, J. AND ASHMORE, J. F. Stiffness of sensory hairbundles in the sacculus of the frog. Hear. Res. 23: 93-104, 1986. HOWARD, J. AND HUDSPETH, A. J. Mechanical relaxation of the hair bundle mediates adaptation in mechanoelectrical transduction by the bullfrog’s saccular hair cell. Proc. Natl. Acad. Sci. USA 84: 3064-3068, 1987. HOWARD, J. AND HUDSPETH, A. J. Compliance of the hair bundle associated with gating of mechanoelectrical transduction channels in the bullfrog’s saccular hair cell. Neuron 1: 189- 199, 1988. HOWARD, J., ROBERTS, W. M., AND HUDSPETH, A. J. Mechanoelectrical transduction by hair cells. Annu. Rev. Biophys. Biophys. Chem. 17: 99124, 1988. HUDSPETH, A. J. AND KROESE, A. B. A. Voltage-dependent interaction of dihydrostreptomycin with the transduction channels in bullfrog sensory hair cells (Abstract). J. Physiol. Land. 345: 66P, 1983. KROESE, A. B. A., DAS, A., AND HUDSPETH, A. J. Blockage ofthe transduction channels of hair cells in the bullfrog’s sacculus by aminiglycoside antibiotics. Hear. Res. 37: 203-2 18, 1989. PICKELS, J. O., COMB, S. D., AND OSBORNE, M. P. The effect of chronic application of kanamycin on stereocilia and their tip links in hair cells of the guinea pig cochlea. Hear. Res. 25: 173- 183, 1987. ROBERTS, W. M., HOWARD, J., AND HUDSPETH, A. J. Hair cells: transduction, tuning, and transmission in the inner ear. Annu. Rev. Cell Biol. 4: 63-92, 1988. RYBAK, L. P. Drug ototoxicity. Annu. Rev. Pharmacol. Toxicol. 26: 79-99, 1986. SANDE, M. A. AND MANDELL, G. L. Antimicrobial agents. In: The Pharmacological Basis qf Drug Action, edited by A. G. Gilman and L. S. Goodman. New York: Macmillan, 1985, p. 1150- 1169. HACOHEN,
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
in U.S.A.
Mechanical Response of Frog Saccular Hair Bundles to the Aminoglycoside Block of Mechanoelectrical Transduction WINFRIED DENK, ROBERT M. KEOLIAN, AND WATT W. WEBB School of Applied and Engineering Physics and Department of Physics, Cornell University, Ithaca, New York 14853 SUMMARY
AND
CONCLUSIONS
I. Deflections of the mechanosensory hair bundles on frog saccular hair cells were measured interferometrically, with submillisecond temporal and submicrometer spatial resolution, and with subnanometer displacement sensitivity. 2. The direction of the initial bundle deflection (toward the taller stereocilia) in response to a sudden application of aminoglycoside antibiotics shows that the mechanosensory channels are blocked in their mechanically open state. 3. The magnitude of the initial deflection is consistent with published data on the gating swing as derived from the gating compliance. 4. A delayed relaxation and frequently a reversal of the initial deflection were observed and are attributed to the previously reported mechanical adaptation mechanism, which is at least partially controlled by the influx of Ca2+ through the transduction channels. 5. Increases of low-frequency spontaneous motion were found at intermediate blocker concentrations. They can be well accounted for by the fluctuating force exerted on the bundle by the random binding and unbinding of blocker molecules. 6. The mechanical response of the hair bundle to aminoglycosides may be related to their acute and specific ototoxicity. INTRODUCTION
mechanical stimulation normally used to open the channels but also if the channels change configuration spontaneously or in response to some other influence. Such a configuration change, or more precisely a trapping in one configuration, is expected to occur on the binding of a chemical blocker molecule such as streptomycin, because streptomycin does occupy, at least partly, the aqueous pore of the transducer channel ( Kroese et al. 1989). The interaction of aminoglycosides with the hair bundle may, furthermore, be related to the ototoxicity of these clinically important antibiotics (Rybak 1986; Sande and Mandell 1985). In this paper we report the results of experiments showing the mechanical effects that aminoglycosides have both on a hair bundle’s average position and on its position fluctuations. METHODS
Displacements and displacement fluctuations were measured at the tips of hair bundles over a wide range frequency range (up to 100 kHz) with a differential laser interferometer (Denk et al. 1989; Denk and Webb 1990). This interferometer uses a lowpower HeNe laser beam that is focused to a diffraction limited spot through a high-power microscope equipped with Nomarski differential interference contrast optics. Over the whole frequency range the displacement sensitivity far exceeds the Brownian motion of a hair bundle. Streptomycin, dihydrostreptomycin, and gentamicin were applied iontophoretically from glass micropipettes that were filled with a 0.5-M solution and their tips placed -5 pm from the bundle. We estimated that an iontophoresis current of 100 pA should lead to a concentration of - 10 PM at the bundle, assuming a diffusion constant of 10-l’ m2 s-’ and a transport number of 10% of that of Cl- (Kroese et al. 1989). The hair cells we used were part of the saccular sensory epithelia from frogs Rana catesbeiana or Rana pipiens, from which monocellular sheets were isolated microsurgically after enzymatic digestion (Corey and Hudspeth 1983; Denk 1989; Denk and Webb 1992). All experimentswere performed in bath solutionscontaining (in mM) 110Na+, 2 K+, 4 Ca2’, 118Cl +, and 3 N-2-hydroxyethylpiperazine-N’-2-ethanesulfonicacid (HEPES) at a pH of 7.25. All displacementmeasurements and spectrawerecalibratedwith the useof an oscillatory stagemotion (Denk and Webb 1990).
Mechanoelectrical transduction in the inner ear is universally performed by sensory hair cells, which convert a mechanical deflection of their sensory organelle, the hair bundle, into a change of intracellular electrical potential (cf. Howard et al. 1988). Hair cells hereby play a pivotal role in the perception of linear and angular acceleration and of sound. Mechanoelectrical transduction involves a straininduced conformational change in the transducer molecule and a subsequent change of membrane conductance; both are presumably performed by a directly strain-activated transmembrane ion channel (Corey and Hudspeth 1979, 1983; Guharay and Sachs 1984; Holton and Hudspeth 1986). Direct strain activation implies, by virtue of reciprocity (Howard and Hudspeth 1988), that channel openings and closings exert a force on the hair bundle thus moving it. RESULTS This gating displacement (Howard and Hudspeth 1988) is The change in the hair-bundle position in response to a a close physical analogue of the electrical charge displacement (gating current) that accompanies the opening and sudden exposure to streptomycin is shown for a typical case in Fig. 1. Within 5-20 ms after the beginning of the iontoclosing of voltage-gated channels (Armstrong and Bezanilla 1973 ) ; it has been indirectly observed through its effect on phoresis pulse, the bundle begins to move in the positive the hair bundles’ compliance (Howard and Hudspeth direction (toward the taller stereocilia) with a speed that is 1988). The gating displacement, and the corresponding roughly proportional to the iontophoresis current ampliforce on the hair bundle, occurs not only in response to the tude. The magnitude of the peak positive excursion, on the l
0022-3077/92 $2.00 Copyright0 1992TheAmericanPhysiological Society
l
927
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
928
W. DENK,
R. M.
KEOLIAN,
AND
W. W. WEBB
1979), then the channel seems to have a continuous distribution of states between completely closed and completely 1nA open, which can be parametrized by the open probability (p,) and an accordingly varying conductance and length, 3nA both proportional to po. In this view, which is depicted in Fig. 2, it would indeed appear as if the binding of a blocker 5nA molecule forces the channel from a configuration with, say, PO= 0.3 into a configuration with p. = 1. Measurements at larger iontophoresis current amplitudes and for a streptomycin pulse of long enough duration (> 100 ms) show that the initial displacement is followed by 6 a motion in the negative direction, which often took the 1 4 bundle past its resting (prepulse) position, sometimes as far w as 200 nm. In every case the bundle eventually returned to 2 2 its resting position, always within several seconds after the 2 0 end of the pulse. We attribute this displacement reversal to -2 the mechanical adaptation mechanism, which changes the transduction setpoint (Eatock et al. 1987) and the un0.0 0.5 1.0 1.5 2.0 2.5 loaded bundle position (Howard and Hudspeth 1987) in Time (s) response to static stimuli. Adaptation in saccular hair cells FIG. 1. Reaction of saccular hair-bundle deflection (A ) to concentration pulses of streptomycin delivered iontophoretically by current pulses, has been shown to be controlled in part by the influx of channels (Hacohen et al. which are shown in B. Displacement traces were offset from each other to Ca2+ through the transduction avoid overlap. Positive displacement is toward the taller stereocilia. 1989). Because the block of the transduction channel prevents this influx of Ca2+, mimicking a closed channel, it other hand, does not increase further as the current exceeds induces the adaptation mechanism to tighten the gating 1 nA. The response to short iontophoresis pulses or the string in a “futile attempt” to reopen the channels initial reaction to longer pulse was toward the taller stereo- (Fig. 2C). cilia in all but 1 out of 27 cells with an mean excursion of We confirmed that the response direction was indepen12.6 t 10 (SD) nm, averaged over the largest observed dis- dent of the location of the pipette relative to the hair bunplacements for each cell. The largest observed excursion for dle, ruling out the possible artifact that the motion might be caused by electroosmotic flow out of the pipette tip. No any cell was 40 nm. A value 6x = d( 1 - pz) K~Y /( K,, + 7 ‘K~) is expected; 6x = 19 nm, assuming a swing of the gate of 4 response could be elicited from cells that had been shown to react to streptomycin iontophoresis if current was instead nm and a gating spring stiffness Kg of 500 pN/ m, as Howard and Hudspeth ( 1988 ) deduced from their measurements of passed out of pipettes containing only 3 M NaCl or K acethe gating compliance, a geometric gain y of 0.14, a stiffness tate. Measurements of the displacement of a compliant glass fiber attached to the bundle closely resembled what K,, of 4.9 ~,LN/ m per stereocilium as measured by Howard and Ashmore ( 1986), a resting open probability pz of 0.3, was observed by focusing on the hair bundle directly, ruling and a single transduction channel per stereocilium. Our out spurious displacement signals due to changes of the sample of cells was not necessarily representative because bundles’ optical properties that might have been induced in bundles were selected for good visibility and were rejected if the bundle by streptomycin. Alternating use of gentamicin and dihydrostreptomycin, out of a double-barreled microno response at all was seen to an iontophoretic current pulse. Motion in the positive direction is expected, as can be Ca++ seen from Fig. 2, A and B, if the channel is blocked in the mechanically open position by the streptomycin molecule. The experimental results thus strongly support the notion that streptomycin, most likely by occupying part of the lumen of the channel pore, prevents the channels from switching into a mechanically closed configuration. The channel is still electrically closed because the passage of ions through the channel is prevented by the presence of the streptomycin molecule. It should be pointed out that the act of binding of the blocker molecule does not itself have to exert a force on the C A B channel. On the contrary, we have assumed that the blocker FIG. 2. Schematic illustration of the interaction of blocker molecules molecule binds only to the open state, without changing the with the transduction channel. How the binding of a blocker molecule (B) channel’s conformation. What moves the bundle then is 1st leads to a positive displacement is shown in B. After the adaptation the energy that is stored as gating spring tension, which now mechanism, represented by the rack and pinion inside the taller stereocilcannot be relaxed by a closing and shortening of the trans- ium (S), has had time to react, the hair bundle is pulled ( C) in a negative past its original resting position (A ). For the unperturbed state duction channel. On the other hand, if one looks at the direction, (A), the gate, representing the strain-activated channel, is depicted as transduction channel with a time resolution much slower about halfway open, which is intended to represent its averaged position, than its flicker frequency of - 10 5 s-l (Corey and Hudspeth rather than an actual state.
1F’s 11
,I,,
,I,,
,,,I
I,,,
,,I,
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
HAIR-BUNDLE
MOTION
DUE TO TRANSDUCTION
electrode, produced almost identical responses, which also were very similar to those typically elicited by streptomycin. The response to dihydrostreptomycin pulses was reversibly abolished when 100 PM streptomycin was added to the bath. It had been shown previously by Howard and Hudspeth ( 1988 ) that gentamicin (and presumably other aminoglycosides as well) abolishes the gating compliance. This result is an indication that gentamicin blocks the transition between the states of the channel molecule but, by itself, does not answer the question into which state the channel is locked. As a consequence of this loss of the gating compliance on aminoglycoside exposure, the Brownian motion of the bundle was expected to be decreased. We found this to
BLOCK
929
be the case. In addition we observed, as is shown in Fig. 3A, that the spontaneous motion decreased only in the stiffnesscontrolled region (below the roll-off) but was unchanged in the damping-controlled region (well above the roll-off). The range of reduction factors that was found in response to the saturating iontophoretic application of streptomycin (up to ~2 as shown in Fig. 3) is consistent with an elimination of the gating compliance, but the interaction of streptomycin with the adaptation mechanism may play an additional role in changing the bundle stiffness. Very similar results were obtained when streptomycin was added to the bath in a few cases instead of applying it iontophoretically. For a blocker concentration below the saturation regime, when only part of the channels are blocked on average, the B 10
lo2
0.5
lo3
Freq.
r
(Hz)
1.0
P~dstI-,p/P=4v/o -1
-kbCB=OS
\
... ... .... ... I
-
400 pA
IIIIII
I
I
lIll
lo1
-0 24
Td \
r
-
I
1111111
l
. 30Hz x 500Hz
I
Ill11
I
IlIll
11111111
lo2
Freq.
(Hz)
0
E
I
lo1
102
Freq. 5
-kbC!B=looS;: -kbCB=%oS
103
(Hz)
l”tF-
K3-
e2 s mm ml-r P(
0 x 0
I -200
8 0
200
Iiont.
400
lo2
lo1
Freq.
lo3
(Hz)
FIG. 3. Suppression of the gating compliance leads to a decrease in the Brownian motion of a saccular hair bundle as is shown in A with the use of spectra take in the unperturbed state ( 1), during exposure (2) to a saturating (2 100 PM) concentration of dihydrostreptomycin, and 10 min ( 3) after washing. Note the merging of all 3 curves at high frequencies, indicating that bundle damping is unaffected. The distribution of the factors by which the power spectral density below the roll-off was decreased for different cells is shown in B. At intermediate blocker concentrations, a Brownian motion increase is observed up to a roll-off frequency of -50 Hz, which is determined by the blocker’s binding kinetics. Spectra are shown in C for iontophoresis currents of 0, 150, and 400 pA. Corresponding model calculations are shown in D. Relative changes inthe power spectral density (psd) for 2 different frequencies (30 and 500 Hz) are plotted as a function of the iontophoresis current in D together with the result of model calculations as described in the text. Observed (ragged curves) and modeled (smooth curves) spectral changes are compared in F.
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
930
W. DENK,
R. M.
KEOLIAN,
random chemical binding and unbinding of blocker to the channels causes a fluctuating force on the hair bundle. This is the same force that, when exerted simultaneously by all transduction elements, is thought to cause the positive displacement step at the beginning of a streptomycin pulse in Fig. 1. The frequency spectrum of this fluctuating force is determined by the chemical kinetics of binding and release of streptomycin. Figure 3C (middle trace) shows the power spectrum for the spontaneous bundle deflections at a dihydrostreptomycin concentration that was estimated to be - 15 PM. Given the uncertainty in this estimate, this is in reasonable agreement with the experimentally determined (Kroese et al. 1989) binding constant ( -8 ,uM) of dihydrostreptomycin to the transduction channel. The increase in the spontaneous low-frequency motion is combined with a reduction at intermediate frequencies ( 100 Hz to 1 kHz), which is about one-half as large as the ultimate reduction observed at a saturating concentration. In Fig. 3 E the change in spontaneous motion’s spectral density as a function of iontophoresis current is plotted both for 30 and 500 Hz. The horizontal scale was adjusted by eye for the best possible agreement with the data such by supposing that 150 pA corresponds to a forward binding rate of khcR = 100 s-’ (with k, and cRthe rate constant and the blocker concentration, respectively). A mathematical model for the mechanical displacement noise due to aminoglycoside binding fluctuations, which is detailed below, describes reasonably well the magnitude and the frequency distribution of the changes occurring in spectral shape (see the smooth lines in Fig. 3 F). Several factors are likely to contribute to the discrepancy of the absolute magnitudes (compare Fig. 3, C and 0): 1) this bundle was smaller than average, and the amplitude of the power spectrum scales linearly with the number of stereocilia; 2) the motion was measured at the tip of the tallest stereocilium, which is - 1.3 times taller than the kinocilium, to which most of the published values are referenced, and the amplitude of the power spectrum scales quadratically with this factor; 3) hair-bundle displacement spectra often rise toward lower frequencies even far below the rolloff, as reported in an earlier paper (Denk et al. 1989). The behavior of the bundle position’s fluctuation power spectrum can be understood intuitively by attributing the increase of the amplitude at low frequency to fluctuating forces caused by blocker binding and unbinding and the decrease at intermediate frequencies to a partial abolition of the gating compliance. A more quantiative test of the notions about how the hair-cell mechanics work requires the analysis of a more detailed physical model. To obtain the spontaneous fluctuations we first calculated the bundle compliance as a function of frequency; its dissipative part then provides the fluctuation spectrum via the fluctuation-dissipation theorem (Callen and Welton 195 1; Denk et al. 1989). The hair bundle was modeled as described by Howard and Hudspeth ( 1988). The total compliance (which is the reciprocal of the total stiffness KT) is given by 1 /K7,
=
iWV
+
K,
+
r2NK
(1)
with o the frequency, Y the damping ( Denk et al 1989), K, the bundle stiffness wi thout the gating elements, Y the ratio
AND
W. W. WEBB
between tip link extension and bundle tip displacement, N the number of stereocilia in a single bundle, Kg the stiffness of the gating spring, and d the change in effective channel length. Next we consider L&/al, the dependence of the closed probability pc on the total length I, to which the transduction element (gating spring and channel combined) is stretched. The channel blocker, present with a concentration c,, is assumed to bind only to the open state with binding and unbinding rate constants of k, and k,, respectively, yielding the following rate diagram k4 0
closed
k of the length of the transduction element from its average value ((I)) are considered. For dk,/dl, which represents the strain sensitivity of the channel, we use ak,/al = k,Kgd/k,T (using Eq. 5 of Howard and Hudspeth 1988, with k, and T as the Boltzmann constant and the absolute temperature, respectively). We then go to the frequency domain and use Eq. 2c to eliminate pC from Eq. 2b, which then becomes iu+kbcB+k,--
k,kbcB = kg, io + k,
1
+ Cl-
(I))(P,)
%
(3)
We then use Eq. 3 to eliminate pn from Eq. 2a and take the limit of large k, and k, reflecting almost instantaneous equilibration between open and closed states, an assumption that is warranted by transition rates of - lo5 s-l (Corey and Hudspeth 1983). To simplify the notation we also introduce the quantity px = ( 1 + k,/ k,)-’ , which becomes equal to the open probability p. = [ khcB/ k, + pi1 1-l without any blocker present. We finally get dPC -=-
al
Kgkupx( 1 - px)(io + k, + kbcB)d kgT( k, + pxkhcB)( iw + k,, + pxkhcR)
(4)
To formalize the assumption that the adaptation mechanism, by tightening the gating strings, changes the balance between the channel opening (k,) and closing (k,) time constants in such a way as to keep the open probability (p,) constant, we use the relation px = [max(
I/ P: - kbq&),
I]-’
(5)
where p: is the unperturbed open probability. As a result the closed probability pCgradually decreases with increasing blocker concentration (c,). When pCLas thus been reduced to zero, p. cannot be kept constant any longer and suddenly starts to fall if cB continues to rise, resulting in a kink in the theoretical curve. This kink, which also marks the complete
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
HAIR-BUNDLE
MOTION
DUE TO TRANSDUCTION
BLOCK
931
elimination of the gating compliance, occurs at forward binding rate (k,c,) of - 190 s-l, which corresponds to 285 pA on the experimental scale, after the horizontal scale for the theoretical curve had been adjusted to agree as well as possible with the data. Our assumption about the control of the adaptation mechanism is probably not entirely precise; for example, Assad et al. ( 199 1) find the adaptation mechanism controlled by both, Ca2+ concentration and tension, whereas we only consider the effect of Ca2+. A subtle but important point concerning the mechanical equivalent of the transduction element should be mentioned here. If a single transduction element were held under “force (F) -clamp” conditions, its slope stiffness (K :I), defined as the ratio of a small applied force change to the resulting change in the average length of the transduction element, would be given by
overall hair-bundle mechanics. There can, furthermore, remain little doubt that aminoglycosides bind to the channel in a way that prevents it from closing and from shortening, as visualized with the use of the picture developed by Hudspeth and co-workers (Corey and Hudspeth 1979, 1983; Howard and Hudspeth 1987, 1988) (see Fig. 2). The slow response to steps in aminoglycoside concentration, including its time course, direction, and magnitude, can be explained by assuming that aminoglycosides interfere with the mechanical adaptation mechanism by blocking the entry of Ca2+ ions through the transduction channels. Assad et al. ( 1989) have similarly attributed the bundle deflection they observed in response to a stepping of the hair cell’s membrane voltage to +80 mV to prevent the influx of Ca2+ ions, in their case due to the removal of the electrochemical driving force. The recovery we observed at the end of a large streptomycin pulse appears to be too slow h’;: = [ 1/KR + d (dy&YF)]-’ (6) to be accounted for by either the unbinding rate constant of 100 s-l, which we did find to provide the best fit of the withdp,ldF= d*p,( 1 -p,)l(ksT). Thisisclearlydifferent from the slope stiffness under “length (/)-clamp” condifluctuation data (Fig. 3C), or by the intrinsic speed of the tions (K$), defined as the ratio of the change in the average adaptation mechanism (Assad et al. 1989), or by simple diffusion. The local streptomycin concentration may, howforce exerted by the transduction element on its attachment ever, decay slower because of nonspecific binding to mempoints in response to a small imposed change in the length branes (A. J. Hudspeth, personal communication). It canof the transduction element not be ruled out completely that aminoglycosides do, in K!/= KJ I - Kgd2p,(1 - /I,)/( kJ)] (7) addition, interact with mechanical adaptation in ways which we find by combining the last term of Eq. I with Eq. other than by blocking Ca2’ entry through the transduction 4 while setting cn = 0. Length-clamp conditions were used channels. Further clarification and more sensitive tests of our interby Howard and Hudspeth ( 1988) to calculate the transducpretation could be obtained from experiments combining tion elements’ stiffness. The apparent stiffness of the transduction element thus whole-cell voltage clamping with the sensitive displacement experiments depends on the mechanical impedance it is coupled to. The measurements described here. Preliminary showing the effect of streptomycin on the correlation bemodel we used makes, like Howard and Hudspeth ( 1988), the implicit but incidentally correct assumption that the tween electrical and mechanical noise in hair cells (Denk 1989) have been encouraging in this respect. individual transduction element operates in a “stiff” enviOne could finally ask whether the strong mechanical reronment. This assumption does, however, become invalid sponse of the hair bundle might provide a mechanism for if the number of stereocilia in hair bundle becomes too the specific ototoxicity of the aminoglycosides (Ryback small. Although for the “soft” environment (force clamp) 1986). Because extended aminoglycoside exposure leads to the apparent gating-element stiffness KF would behave like conspicuous morphological changes in the hair bundle, the gating spring in series with another spring representing such as the fusion of stereocilia, ototoxicity might, literally, the channel, we find for the stiff case (length clamp) that K$can even become negative, for perfectly reasonable values be due to mechanical self destruction of the hair bundle as a of Q, d, and p. (such as 500 pN/m, 6 nm, and 0.5). As a result of the adaptation motors pulling incessantly at the consequence, the total bundle stiffness might be decreased gating strings. The same mechanism might mediate the destruction of transduction (Crawford et al. 199 1) and the to a value below what it would be after cutting the transduction elements. For this to be valid, there is no need to as- snapping of the tip links (Assad et al. 199 1) seen in low sume that the gating spring can sustain a force under com- external Ca 2+ concentrations. pression if the free energies of the channel states are such We thank Dr. David Corey for comments on the manuscript. that the channel is shut for zero tension in the gating spring. This work was supported by grants from, AT&T Bell Laboratories, OfTo optimize the agreement between data and theory for fice of Naval Research, National Science Foundation, and National Instithe spectral changes induced by aminoglycosides, we ad- tutes of Health (NIH). W. Denk was an IBM predoctoral fellow and W. W. justed some of the parameters somewhat from their pub- Webb a NIH Fogarty Scholar-in-Residence during some of this research. Address for reprint requests: W Denk, AT&T Bell Labs, Murray Hill, lished best values, while staying well within their error margins. For the curves in Fig. 3 we used Kg = 550 pN/m, d = 5 NJ 07974. nm, K~ = 200 pN/m, Y = 60 nN s/m2,pz = 0.35, N= 50, Received 2 February 1992; accepted in final form 1 May 1992. and y = 0.14. Setting k, = 100 s-l best accounted for the REFERENCES position of the roll-off in the low-frequency fluctuations. l
C. M. AND BEZANILLA, F. Currents related to movement of the gating particles of the sodium channels. Nutzrrc Land. 242: 459-46 1, 1973. ASSAD, J. A., HACOHEN, N., AND COREY, D. P. Voltage dependence of adaptation and active bundle movement in bullfrog saccular hair cells. Proc. Nat/. Acad. Sci. USA 86: 29 18-2922, 1989. ARMSTRONG,
DISCUSSION
The observed mechanical responses of hair bundles to aminoglycosides further strengthen the case for a directly gated transduction channel that is tightly coupled to the
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
932
W. DENK, R. M. KEOLIAN,
G. M. G., AND COREY, D. P. Tip-link integrity and mechanical transduction in vertebrate hair cells. Neuron 7: 985-994, 1991. CALLEN, H. B. AND WELTON, T. A. Irreversibility and generalized noise. Physiol. Rev. 83: 34-40, 195 1. COREY, D. P. AND HUDSPETH, A. J. Response latency of vertebrate hair cells. Biophls. J. 26: 499-506, 1979. COREY, D. P. AND HUDSPETH, A. J. Kinetics of the receptor current in bullfrog saccular hair cells. J. Neurosci. 3: 962-976, 1983. CRAWFORD, A. C., EVAN, M. G., AND FETTIPLACE, R. The actions of calcium in the mechano-electrical transducer current of turtle hair cells. .J. Physiol. Lond. 434: 369-398, 199 1. DENK. W. Biophysical Studies c!f‘ Mechano- Electrical Transduction in Hair Cells ( PhD. thesis). Ithaca New York: Cornell Univ., 1989. DENK, W. AND WEBB, W. W. Simultaneous recording of fluctuations of hair-bundle deflection and intracellular voltage in saccular hair cells. In: Cbchlear Mechanics, Structure, Function and Models, edited by .I. P. Wilson and D. T. Kemp. New York: Plenum, 1989, p. 125-l 34. DENK, W. AND WEBB, W. W. Optical measurements of picometer displacements of transparent microscopic objects. Appl. Optics 29: 2382-239 1, 1990. DENK, W. AND WEBB, W. W. Forward and reverse transduction at the limit of sensitivity studied by correlating electrical and mechanical fluctuations in hair cells. Hear. Res. 60: 89- 102, 1992. DENK, W., WEBB, W. W., AND HUDSPETH, A. J. The mechanical properties of sensory hair bundles are reflected in their Brownian motion measured with a laser differential interferometer. Proc. Natl. Acad. Sci. USA 86: 5371-5376, 1989. EATOCK, R. A., COREY, D. P., AND HUDSPETH, A. J. Adaptation of mechanoelectrical transduction in hair cells of the bullfrog’s sacculus. J. Neurosci. 7: 282 l-2836, 1987. GUHARAY, F. AND SACHS, F. Stretch-activated single ion channel currents in tissue-cultured embryonic chick skeletal muscle. J. Physiol. Lond. 352: 685-701, 1984. ASSAD, J. A., SHEPARD,
AND W. W. WEBB
N., ASSAD, J. D., SMITH, W. J., AND COREY, D. P. Regulation of tension on hair-cell transduction channels: displacement and calcium dependence. J. Neurosci. 9: 3988-3997, 1989. HOLTON, T. AND HUDSPETH, A. J. The transduction channel of hair cells from the bull-frog characterized by noise analysis. J. Physiol. Lond. 75: 195-227, 1980. HOWARD, J. AND ASHMORE, J. F. Stiffness of sensory hairbundles in the sacculus of the frog. Hear. Res. 23: 93-104, 1986. HOWARD, J. AND HUDSPETH, A. J. Mechanical relaxation of the hair bundle mediates adaptation in mechanoelectrical transduction by the bullfrog’s saccular hair cell. Proc. Natl. Acad. Sci. USA 84: 3064-3068, 1987. HOWARD, J. AND HUDSPETH, A. J. Compliance of the hair bundle associated with gating of mechanoelectrical transduction channels in the bullfrog’s saccular hair cell. Neuron 1: 189- 199, 1988. HOWARD, J., ROBERTS, W. M., AND HUDSPETH, A. J. Mechanoelectrical transduction by hair cells. Annu. Rev. Biophys. Biophys. Chem. 17: 99124, 1988. HUDSPETH, A. J. AND KROESE, A. B. A. Voltage-dependent interaction of dihydrostreptomycin with the transduction channels in bullfrog sensory hair cells (Abstract). J. Physiol. Land. 345: 66P, 1983. KROESE, A. B. A., DAS, A., AND HUDSPETH, A. J. Blockage ofthe transduction channels of hair cells in the bullfrog’s sacculus by aminiglycoside antibiotics. Hear. Res. 37: 203-2 18, 1989. PICKELS, J. O., COMB, S. D., AND OSBORNE, M. P. The effect of chronic application of kanamycin on stereocilia and their tip links in hair cells of the guinea pig cochlea. Hear. Res. 25: 173- 183, 1987. ROBERTS, W. M., HOWARD, J., AND HUDSPETH, A. J. Hair cells: transduction, tuning, and transmission in the inner ear. Annu. Rev. Cell Biol. 4: 63-92, 1988. RYBAK, L. P. Drug ototoxicity. Annu. Rev. Pharmacol. Toxicol. 26: 79-99, 1986. SANDE, M. A. AND MANDELL, G. L. Antimicrobial agents. In: The Pharmacological Basis qf Drug Action, edited by A. G. Gilman and L. S. Goodman. New York: Macmillan, 1985, p. 1150- 1169. HACOHEN,
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 16, 2019.
