Fast recovery of disrupted tip links induced by mechanical displacement of hair bundles

Significance Each of the sensory receptors responsible for hearing or balance—a hair cell—has a mechanosensitive hair bundle. Mechanical stimuli pull upon molecular filaments—the tip links—that open ionic channels in the hair bundle. Loud sounds can damage hearing by breaking the tip links; recovery by replacement of the constituent proteins then requires several hours. We disrupted the tip links in vitro by removing the calcium ions that stabilize them, and then monitored the electrical response or stiffness of hair bundles to determine whether the links could recover. We found that tip links recovered within seconds if their ends were brought back into contact. This form of repair might occur in normal ears to restore sensitivity after damage.


Preparation of the rat's cochlea
The procedures were conducted at the Institut Curie and were approved by the Ethics Experiments were performed on excised cochleae of Sprague-Dawley rats (Rattus norvegicus, Janvier Labs) of both sexes and 7-10 days of age. The dissection and isolation of the cochleae followed a published procedure (1,2). After a rat had been euthanized and decapitated, the inner ears were extracted from the head. Each cochlear bone was carefully opened and the membranous cochlear duct uncoiled from the modiolus. After excision of the cochlear partition, the stria vascularis was removed and the tectorial membrane gently peeled away. An apical or middle turn of the organ of Corti

Preparation of the bullfrog's sacculus
The procedures were conducted at The Rockefeller University and at the Institut Curie with the approval of the respective Institutional Animal Care and Use Committees.
Experiments were performed on hair cells from adult bullfrogs (Rana catesbeiana) of both sexes. After an animal had been euthanized, the sacculi were carefully removed by a standard protocol (3). Each saccular macula was sealed with tissue adhesive (Vetbond, 3M) across a 1 mm hole centered on a 10 mm square of aluminum foil. The foil was situated in a two-compartment chamber with the macular side of the sacculus facing upward. The lower compartment was filled with oxygenated artificial perilymph (114 mM Na + , 2 mM K + , 2 mM Ca 2+ , 118 mM Cl -, 5 mM Hepes, and 3 mM D-glucose; pH 7.4; 230 mOsmol·kg -1 ). The apical surface of the hair cells was exposed for 35 min at room temperature to 67 mg·L -1 of protease (type XXIV; Sigma) to loosen the otolithic membrane, which was carefully removed with an eyelash. The upper compartment was then filled with oxygenated artificial endolymph (2 mM Na + , 118 mM K + , 250 μM Ca 2+ , 118 mM Cl -, 5 mM Hepes, and 3 mM D-glucose; pH 7.4; 230 mOsmol·kg -1 ).

Measurement of hair-bundle position
Experiments on both preparations were conducted with similar apparatus. Each preparation was placed on an upright microscope (BX51WI, Olympus) and the hair cells were visualized with a 60X, water-immersed objective lens of numerical aperture 0.9 and differential-interference-contrast optics. Rat hair cells were observed during experiments with a charge-coupled-device camera (LCL-902K, Watec). Video observations of the bullfrog's sacculus videos were conducted after an additional 4X magnification with a CMOS camera (DCC3240M, Thorlabs) or a high-speed video camera (ZYLA-5.5-CL10-W, Andor).
To record a hair bundle's position, the preparation was illuminated with a 630 nm light-emitting diode (UHP-T-SR, Prizmatix) and the resultant shadow was projected onto a dual photodiode at a magnification of 1300X. The output of the photodiode was lowpass filtered at 2 kHz with an eight-pole anti-aliasing filter (Benchmaster 8.13, Kemo).

Mechanical stimulation with fluid jets
Because they are complexly shaped and poorly cohesive, hair bundles from outer hair cells of the rat's cochlea are difficult to stimulate with glass fibers. We therefore deflected each bundle with a fluid jet driven by a piezoelectric disk, which recruited all the stereocilia (1). When viewed under the objective lens of the microscope in the plane of the sensory epithelium, the tip of each pipette was positioned along the axis of mirror symmetry of each hair bundle at a 8 µm distance on the bundle's abneural side. Liquid exiting the pipette therefore displaced the stereocilia towards their shortest row.

Mechanical stimulation with flexible fibers
Owing to the strong attachments among the stereocilia of a hair bundle from the bullfrog's sacculus, force applied to the kinocilium uniformly displaces all the stereocilia (4). We accordingly used a flexible glass fiber attached to the kinociliary bulb to mechanically stimulate the hair bundle. Each fiber's stiffness and drag coefficient were estimated by measuring the Brownian motion of its tip in water. We then obtained parameter values by fitting the power spectrum of the displacement to a Lorentzian relation (5). The fibers in this study had stiffnesses of 160-380 µN·m -1 and drag coefficients of 150-290 nN·s·m -1 ; they behaved as first-order, low-pass filters with cut-off frequencies near 200 Hz.

Displacement clamping
We used negative feedback to control the position of a hair bundle according to a computer-generated external command (1,6,7). By doing so, we were able to monitor the force required to hold a hair bundle stationary or to deflect it to a desired position. The computer's sampling interval of 200 µs set an upper limit on the potential frequency response of the system, but a eight-pole, low-pass Bessel filter (Benchmaster 8.07, Kemo) imposed a cutoff at 2 kHz between the computer's output and the stimulator's input to ensure stability.
Use of the displacement-clamp system and sinusoidal stimulation allowed us to measure the decrease and subsequent recovery of hair-bundle stiffness with good temporal resolution. However, this approach confronted an inevitable problem: because the response time of the clamp system is finite, responses of progressively higher frequency become progressively less well clamped. The clamp's settling time constant was generally about 2 ms, which corresponded to a corner frequency near 80 Hz. By selecting a stimulus frequency of 40-50 Hz, we accepted some non-ideality in clamping in the interest of improved frequency resolution in stiffness measurements.
The force FSF exerted by the stimulus fiber against a hair bundle was estimated from the positions of the fiber measured at its base and at its tip (8): in which KSF and "# represent respectively the stiffness and hydrodynamic friction coefficient of the stimulus fiber, Y the displacement of its base, and X the displacement of its tip. ̇ and ̇ are the time derivatives of the corresponding variables. Because the stimulus frequencies were well below the cut-off frequency of the fiber, this low-frequency approximation of the periodic force applied by the fiber is expected to be accurate (8).
Positive movements and forces were those directed toward a hair bundle's tall edge.
The stiffness KHB of each hair bundle was estimated by measuring the average force FSF and displacement XHB for 21 successive periods of sinusoidal stimulation. The stiffness was then computed for each sinusoidal train as

Voltage-clamp recording
We recorded mechanoelectrical-transduction currents of outer hair cells of the rat cochlea with whole-cell, tight-seal electrodes. Each micropipette was pulled (P -97, Sutter Instruments) from a thick-walled capillary (1B150F-4, WPI) and fire-polished to obtain a The voltage offset was corrected before forming a gigaohm seal with a cell and the pipette's capacitance was compensated to achieve a cut-off frequency of 1-9 kHz.
Current signals were low-pass filtered at 1.25-12.5 kHz and sampled at intervals of 40-400 µs.

Iontophoresis
We used iontophoretic pulses to deliver Ca 2+ chelators in the vicinity of the hair bundles.
Microelectrodes were fabricated from borosilicate glass capillaries (TW 120-F, World Precision Instruments) with an electrode puller (P-80/PC, Sutter Instruments) and filled with 500 mM EDTA in 1 M NaOH. We used a current amplifier (Axoclamp 2B, Axon Instruments) to control the release of EDTA. A holding current of 10 nA kept EDTA from diffusing into the endolymphatic solution, and pulses of -100 nA released the chelator.
The electrodes' tips were directed at and situated about 2 µm from tops of the hair bundles.

SI Notes
Note S1. Negative hair-bundle movement during exposure to Ca 2+ chelators The sequence of hair-bundle forces associated with the breaking and regeneration of tip links reveals unexpected complexity in recordings from bullfrog hair cells. In six of the seven cells, there was a sustained positive offset of 20.1 ± 7.0 pN (mean ± SEM) at the end of the stimulation protocol with respect to the value before EDTA exposure ( Fig. 3A; SI Appendix, Fig. S4). This force offset was absent when tip links were not broken.
In principle, this tensioning of the hair bundle would be compatible with increased activity of the adaptation machinery (7,9). A decrease in the cytoplasmic Ca 2+ concentration after tip-link rupture would cause the adaptation motors to ascend in the stereocilia and thus generate a negative offset in the position of the hair bundle after tiplink recovery. Nevertheless, this effect was probably masked by the presence of another, more intriguing phenomenon: a negative movement of the hair bundle that occurred seconds after tip-link breakage.
Upon exposure to Ca 2+ chelator there was a sudden increase in the force that reflected a rise in tip-link tension, followed by the abrupt decrease that resulted from tiplink rupture. Although these observations accorded with previous studies (8,10,14,15), the traces also revealed a subsequent rebound in the force (Fig. 3A). The force exerted by the fiber indicated that the displacement clamp acted to counter a negative movement of the hair bundle (SI Appendix, Fig. S4).
Although never observed in outer hair cells from the rat's cochlea, this unexpected effect was present to a certain degree in most recordings from the two-compartment conditions-remains to be explained. One possibility is that the cuticular plate deforms in such a way as to alter the forces within the stereociliary cluster. For hair cells of the bullfrog's sacculus, the cuticular plate is concave upward, a configuration that pushes the stereociliary tips together (12,13). If the curvature of the cuticular plate were to increase after tip-link breakage, the stereocilia of the longest rank would be expected to undergo a negative displacement.

Note S2. Lack of contribution of the kinocilium to negative movements
Aside from those associated with tip links, what other forces might act on a hair bundle?
Each bundle possesses a single kinocilium that bears an axoneme with dynein motors (14). Because the kinocilium can be motile (15,16), it might exert a force that affects the hair bundle's position. To test that possibility, we separated the kinocilium from an oscillating hair bundle and usually held its tip several micrometers away from the stereociliary cluster with a glass microelectrode (17). Because the optical contrast of a bundle with a detached kinocilium was too low to allow the use of a photodiode, we used video microscopy to record the position of the hair bundle and a tracking algorithm (18) to trace independently the positions of the tallest stereociliary row and of the kinocilium Fig. S5A). We were then able to measure both displacements before, during, and after breaking the tip links with EDTA. Even with the kinocilium separated from and moving independently of the stereociliary cluster, four of the five hair bundles tested displayed a negative movement following EDTA exposure (SI Appendix, Fig. 5B).
In some instances, the negative motion proceeded in rapid steps of irregular size, a phenomenon that occurred even when the dissociated kinocilium was immobilized against the epithelial surface by a microelectrode (SI Appendix, Fig. 5C). The negative hair-bundle movements thus stem from a source other than the kinocilium. Fig. S1.

SI Figures and Legends
Step protocol for facilitating tip-link recovery in bullfrog saccular hair cells.

SI Video Caption
Video S1. Recovery of oscillations after iontophoresis of a Ca 2+ chelator. Viewed from above, a hair bundle from the bullfrog's sacculus displays low-frequency spontaneous oscillations. When EDTA is expelled from the pipette at the upper left, the bundle jumps in the positive direction, to the right, and ceases to move. After the metal-coated stimulus fiber at the upper right applies force in the negative direction and is then withdrawn, the bundle resumes oscillations indicative of an intact transduction process.