Conceived and designed the experiments: MCM AJH. Performed the experiments: MCM SHI AJH. Analyzed the data: MCM SHI AJH. Wrote the paper: MCM AJH.
The authors have declared that no competing interests exist.
Hair cells in the auditory, vestibular, and lateral-line systems respond to mechanical stimulation and transmit information to afferent nerve fibers. The sensitivity of mechanoelectrical transduction is modulated by the efferent pathway, whose activity usually reduces the responsiveness of hair cells. The basis of this effect remains unknown.
We employed immunocytological, electrophysiological, and micromechanical approaches to characterize the anatomy of efferent innervation and the effect of efferent activity on the electrical and mechanical properties of hair cells in the bullfrog's sacculus. We found that efferent fibers form extensive synaptic terminals on all macular and extramacular hair cells. Macular hair cells expressing the Ca2+-buffering protein calretinin contain half as many synaptic ribbons and are innervated by twice as many efferent terminals as calretinin-negative hair cells. Efferent activity elicits inhibitory postsynaptic potentials in hair cells and thus inhibits their electrical resonance. In hair cells that exhibit spiking activity, efferent stimulation suppresses the generation of action potentials. Finally, efferent activity triggers a displacement of the hair bundle's resting position.
The hair cells of the bullfrog's sacculus receive a rich efferent innervation with the heaviest projection to calretinin-containing cells. Stimulation of efferent axons desensitizes the hair cells and suppresses their spiking activity. Although efferent activation influences mechanoelectrical transduction, the mechanical effects on hair bundles are inconsistent.
Hair cells in the acousticolateralis system are responsible for the detection of sound, acceleration, vibration, and water flow. Both the high sensitivity and the sharp frequency selectivity in these cells rely upon an active process that supplies energy to the hair bundle, the hair cell's mechanosensitive organelle. Mechanical stimuli detected by the mechanoelectrical-transduction channels in the bundle evoke receptor potentials that trigger the release of neurotransmitter at the hair cell's ribbon synapses and thus stimulate afferent fibers.
The sacculus is a vestibular organ that in various species is sensitive to head tilt, seismic vibrations, and airborne sound. Saccular hair cells are extensively innervated by afferent fibers whose cell bodies are localized in peripheral ganglia
Auditory and vestibular efferents play a role in mechanotransduction by regulating the membrane potential of hair cells and afferent terminals. In addition, mammalian cochlear efferents modulate the membrane potential of outer hair cells to regulate the active process by controlling electromotility
Using the bullfrog's sacculus as a model system, we studied the effect of efferent stimulation on the activity of hair cells. We used immunocytochemistry to characterize the distribution of efferent terminals and ribbon synapses. By electrophysiological recordings, we investigated the effect of efferent activity on the electrical properties of hair cells. Finally, we used photomicrometry to measure the mechanical effect of efferent activation on hair-bundle displacement.
After approval of research protocol 04044 by the Institutional Animal Care and Use Committee of The Rockefeller University, bullfrogs (
A conserved sequence in the N-terminal region of synapsin I (
Isolated sacculi were incubated for 1 h at 4°C in phosphate buffered saline (PBS) solution containing 4% formaldehyde and 0.1% Triton X-100. After washing four times for 10 min each at room temperature in PBS containing 0.1% Triton X-100, we incubated the sacculi for 30 min in the same solution supplemented with 5% bovine serum albumin (BSA). The specimens were then incubated with primary antisera at a dilution of 1∶100-1∶1000 in PBS containing 0.1% Triton X-100 and 5% BSA for 2 h at room temperature or overnight at 4°C. The antisera included those against calretinin (polyclonal, AB1550, Chemicon), CtBP2 (polyclonal, #1869), parvalbumin 3 (polyclonal, #799), SV2 (monoclonal, DSHB), alpha-tubulin (monoclonal, 6G7, Sigma), and beta-actin (monoclonal, A5441, Sigma). After washing five times for 10 min each at room temperature, we incubated the samples with Alexa 488/568-conjugated secondary antibodies (Invitrogen, Carlsbad, CA) at a dilution of 1/500. Where mentioned, DAPI or Alexa 568-conjugated phalloidin was added during the incubation with the secondary antibodies. After washing five times for 10 min each at room temperature, we mounted the specimens between two coverslips with Vectashield (Vector Laboratories, Burlingame, CA).
Tissues were homogenized in PBS containing 0.1% SDS, 1% Triton X-100, 2 mM EDTA, 1 mM PMSF, and protease inhibitors (Roche, Hoffmann-La Roche, NJ), centrifugued at 10,000×g at 4°C for 5 min in a QIAshredder Spin Column (Qiagen Inc., Valencia, CA), diluted with NuPAGE LDS sample buffer and NuPAGE reducing agent (Invitrogen Corp., Carlsbad, CA), incubated at 100°C for 5 min, loaded on NuPAGE Novex 10% Bis-Tris gels, and transferred onto PVDF membranes. Blots were incubated with primary antibodies overnight at 4°C and probed by immunoblotting with horseradish peroxidase-labeled secondary antibodies and enhanced chemiluminescence detection (ECL Plus, GE Healthcare, Chalfont St. Giles, UK). In the preadsorption experiments, antibody binding was blocked by preincubation for 5 h at 4°C with a 20-fold molar excess of the peptide.
Specimens were prepared for transmission electron microscopy by published techniques
Sharp electrodes were produced on a horizontal electrode puller (P-2000; Sutter Instrument, Novato, CA) from borosilicate glass capillaries (1B120F-3, World Precision Instruments, Sarasota, FL). To facilitate penetration of the apical surface of a hair cell, the tapered end of each microelectrode was inserted horizontally into a drop of water and bent 0.5 mm from the tip by heating the drop close to the insertion point
Experiments were conducted under a BX51WI upright microscope (Olympus, Center Valley, PA) equipped with a 200-W mercury lamp (Excite exacte, EXFO, Mississauga, Ontario, Canada), a bandpass filter (HQ600/200m-2p, Chroma, Rockingham, VT), and standard differential-interference-contrast optics. The image formed by a 40X water-immersion objective lens of numerical aperture 0.8 and working distance 3.3 mm was either projected onto a CCD camera (WAT-660D, Watec, Orangeburg, NY) or viewed directly through the eyepieces.
The techniques for mechanical stimulation, imaging, and optical calibration have been described in detail elsewhere
After removal of the differential-interference-contrast components and the microscope's frosted-glass screen, the image of the fiber tip was magnified 958X and projected onto a dual photodiode (SPOT-2D OSI Optoelectronics, Hawthorne, CA). A custom-made circuit attached to the photodiode provided current-to-voltage conversion. The photodiode and circuit were mounted on a two-axis linear stage to allow centering of the shadow of the fiber's tip between the two photosensitive cells. Prior to reaching the photodiode, the magnified image was reflected from a 45° mirror mounted on a piezoelectric actuator (P-840.60, Physik Instrumente, Auburn, MA). To calibrate the photodiode without moving the fiber, a series of displacement steps of alternating polarity was delivered to the piezoelectric actuator, yielding a linear relationship between the probe's displacement and the photomicrometer's output voltage. Hysteresis in the piezoelectric actuator was compensated by delivering a 65%-overshooting prepulse before each calibration step. Recordings used the piezoelectric actuator to deliver a calibration prepulse equivalent to 20 nm at the level of the fiber tip. The photodiode system was in turn calibrated with a heterodyne interferometer (OFV3001/501, Polytec GmbH, Waldbronn, Germany). The fiber was secured at its base to a second stack-type piezoelectric actuator (P835.10; Physik Instrumente, Auburn, MA) driven by a matched power supply (E-663.00, Physik Instrumente, Auburn, MA). The piezoelectric actuator was mounted on an electrophysiological micromanipulator (ROE-200, Sutter Instrument, Novato, CA) and was calibrated with the interferometer.
All devices were controlled by custom-written software (LabVIEW 7.0, National Instruments, Austin, TX). Multifunctional data acquisition boards (PCI-6120, PCI-6733, National Instruments) were used for analog device control and data acquisition. Data analysis was performed with custom-written routines in Matlab (The Mathworks, Natick, MA). Postsynaptic potentials were analyzed with MINIANALYSIS (SYNAPTOSOFT; Jaejin Software, Leonia, NJ). Graphical data were analyzed with ImageJ (NIH).
Although the sacculus receives dense efferent innervation
We used standard epifluorescence and confocal microscopy to analyze the expression pattern of synapsin I in the saccular epithelium of the bullfrog. Efferent terminals were distributed throughout the macula with no obvious spatial pattern (
On the basis of morphology, protein expression, and electrophysiological properties, the hair cells of the frog's sacculus may be classified into at least three classes
Hair-cell group | Number of efferent terminals | Number of synaptic ribbons | ||
Abneural CR- | 7.22±0.44 | (23) | 34.67±1.99* | (24) |
Abneural CR+ | 14.83±0.63 | (48) | 37.35±1.38* | (46) |
Central CR- | 6.09±0.21 | (76) | 20.67±0.51 | (75) |
Central CR+ | 9.61±0.37 | (36) | 8.46±0.85 | (28) |
Neural CR- | 7.22±0.36 | (41) | 32.63±1.31* | (38) |
Neural CR+ | 13.99±0.56 | (87) | 31.56±1.22* | (54) |
Extramacular | 9.03±0.66 | (31) | n.d. |
Values are means ± SEMs. For each group, the number of hair cells tested is given in parentheses. Calretinin-positive hair cells (CR+) and calretinin-negative hair cells (CR–) were statistically analyzed using a two-tailed Student's
To explore the fine structure of efferent presynaptic terminals in hair cells, we performed electron microscopy on the sacculus. In accordance with the typical pattern of efferent synapses
Although saccular hair cells are innervated by both afferent and efferent fibers, little is known about the relative distributions of these two synaptic classes. We assayed the number of synaptic ribbons in the three types of hair cells by labeling with a polyclonal antiserum directed against C-terminal binding protein 2 (CtBP2). This antiserum labeled the synaptic ribbons in all macular hair cells, whereas it labeled the nuclei of epithelial cells outside the macula (
Taken together, these results demonstrate that all hair cells in the bullfrog's sacculus, especially those expressing calretinin, are highly innervated by efferent fibers. This pattern suggests a fundamental role for efferent control in the responsiveness of saccular hair cells.
The electrical properties of hair cells are known to be modulated by the efferent pathway in a variety of sensory receptors
To study the role of efferent activation on saccular hair-cell receptor potential, we first characterized its effect on the resting potential. We used a suction electrode to stimulate the efferent nerve bundle while recording intracellularly from hair cells. Upon efferent stimulation, the majority of hair cells displayed a response with two components, a brief depolarization followed by a strong, long-lasting hyperpolarization (
Inhibitory postsynaptic potentials were evoked by 0.2-mA stimuli 0.05–3 ms in duration. Electrical stimuli of greater amplitude or duration did not produce larger effects. Responses were observed in a total of 52 hair cells, with a maximum hyperpolarization of 27 mV and an average value in 4 mM-Ca2+ saline of 6.4±4.3 mV. In six cells that displayed inhibitory postsynaptic potentials in the presence of 2 mM Ca2+, the hyperpolarization averaged 4.0±2.4 mV. Within the same cells, the responses triggered by efferent stimulation also varied in amplitude and length from trial to trial (
Single electrical stimuli delivered to efferent axons generally produced small, sporadic responses. Rapid, periodic trains of electrical stimuli delivered at a frequency near 200 Hz increased the probability and amplitudes of inhibitory postsynaptic potentials (
Number of pulses | Amplitude (mV) | Rise time (ms) | Decay time (ms) | ||
1 | 64/150 | 0.43±0.04 | 1.6±0.4 | 9±9 | 31±19 |
2 | 115/150 | 0.77±0.03 | 2.0±0.5 | 8±5 | 39±17 |
4 | 146/150 | 0.97±0.01 | 2.2±0.5 | 13±6 | 47±16 |
5 | 148/150 | 0.99±0.01 | 2.6±0.4 | 22±8 | 59±16 |
Each line shows the number of IPSPs (
The synaptic effect additionally depended on the delay between efferent current pulses, with a maximal response evoked by stimuli delivered 3–10 ms apart (
In the turtle's cochlea, efferent activity transforms sharply tuned electrical resonance into overdamped relaxation
In the absence of efferent stimulation, we found 75 cells that displayed electrical resonance that could be observed either as damped oscillations in response to depolarizing current steps (blue trace in
Following efferent stimulation, in addition to hyperpolarization of the resting potential, inhibitory postsynaptic potentials also prevented spontaneous oscillation of the membrane potential in highly tuned hair cells (
Although it is generally accepted that the efferent systems of acousticolateralis organs are inhibitory in nature, there have been reports of efferent enhancement of vestibular afferent responses
In order to distill meaningful mechanical information from background noise, hair cells use an active process that amplifies and tunes their inputs. Ca2+ plays a key role in this active process, both by adjusting the hair bundle to a position of instability through slow adaptation, and by mediating the transduction-channel reclosure, or fast adaptation, that powers amplification
We hypothesized that efferent activity might modulate hair-bundle motion by altering the concentration of Ca2+ in the stereocilia. On a slow timescale, efferent activation might act through the diffusion of Ca2+ from synaptoplasmic cisternae after Ca2+-induced Ca2+ release
To assess these possibilities, we studied the effect of efferent stimulation on hair-bundle motion by recording the displacement of a compliant fiber attached to the hair bundle's tip. To mimic physiological ionic conditions, we filled the upper and lower compartments of the experimental chamber with endolymph and perilymph, respectively. Approximately 100 ms after the onset of a train of efferent stimuli, the resting position of the hair bundle shifted about 3.5 nm toward the kinocilium, then returned to its original position (
To investigate the effect of efferent stimulation during mechanical stimulation, we monitored the motion of hair bundles in response to sinusoidal force. The resting point of the bundle shifted during efferent activity (
To ascertain whether efferent activity modulates transient hair-bundle motion, we compared the steady-state bundle position 100 ms after a force pulse in the presence or absence of efferent stimulation. The mean bundle displacement in the positive direction after efferent activity was 94%±1% (mean ± SEM; ten cells, 76 pulses) of the control value (
In summary, we found that efferent stimulation affects the motion of the hair bundle by shifting the bundle's position. The rapidity of this effect is consistent with a mechanism involving efferent modulation of the hair bundle's Ca2+ concentration through a change in the driving force on Ca2+, rather than diffusion of Ca2+ from intracellular stores to the hair bundle.
We found that the efferent pathway differentially innervates distinct types of hair cells in the bullfrog's sacculus. The flask-shaped cells containing the Ca2+ buffer calretinin and expressing large Ca2+-dependent K+ currents receive a greater number of efferent terminals than the cylindrical cells with large voltage-dependent Ca2+ currents. This pattern accords with the high sensitivity of the flask-shaped cells to apamin
In addition to the robust immunofluorescence signal for synapsin I in efferent terminals, a fainter but consistent signal was localized adjacent to synaptic ribbons when the gain of the confocal image was increased nearly to saturation (
We encountered many saccular hair cells that generated action potentials in response to depolarization. Action potentials have been reported in hair cells from the sacculi of bullfrogs and leopard frogs
In spite of the significant number of reports, the physiological relevance of action potentials in hair cells remains unknown. In principle, hair cells should lose their frequency-tuning properties by leaving a subthreshold regime. However, the generation of action potentials may not interfere with the tuning of those hair cells sensitive to frequencies below roughly 50 Hz, in which a spike could occur during the depolarizating phase of each cycle. In fact, this strategy would increase the synaptic efficiency for low-amplitude stimuli with no detriment to tuning. Alternatively, the sacculus might sacrifice frequency information for the sake of temporal resolution, thus becoming more sensitive to faint, transient stimuli.
We have investigated efferent modulation of the hair cell's receptor potential in resonant as well as spiking saccular hair cells. Electrical stimulation of efferent axons evoked large inhibitory postsynaptic potentials that degraded hair-cell frequency tuning and suppressed firing.
In agreement with previous reports from turtle and mammals
We found that efferent stimulation shifts the resting position of a hair bundle by a few nanometers without a striking change in bundle stiffness. Both positive and negative motions were evoked by efferent stimulation. We observed these effects in the large, cylindrical, calretinin-negative, centrally located hair cells of the sacculus, whose mechanical properties are well established. Given that calretinin-positive cells are more densely innervated by efferent terminals, we cannot rule out a different efferent effect on these cells. Comparable directional variability has been described in the turtle's cochlea following intensely depolarizing current pulses
Characterization of anti-synapsin antiserum. A, In the top panel, an alignment of the N-termini of synapsin I in three different species shows the conserved sequence (underlined in the Xenopus laevis sequence) used for antiserum generation. NCBI accession numbers are in parentheses. The middle panel displays an alignment of the synapsin isoforms found in Xenopus laevis. The bottom panel shows an alignment of all synapsin isoforms found in the mouse Mus musculus. Blue denotes residues conserved among all sequences, red indicates residues present in two of the three sequences, and green shows residues conserved in two of the three isoforms. B, The right panel presents an immunoblot loaded with brain extract from the bullfrog, mouse, and chicken and incubated with the purified Rb1498 antiserum against synapsin I. Immunoreactive bands occur at approximately 80 kDa and 55 kDa. The lower bands might correspond to synapsin isoforms or degradation products. The left panel demonstrates that the presence of the peptide used for immunization eliminates immunoreactivity. C, A confocal section (lower panels) shows the presence of efferent terminals at the level of hair cells in the saccular macula. The characteristic punctate labeling by the anti-synapsin I antiserum is absent (upper panels) after preadsorption of the antiserum with the corresponding antigen. Both samples were assayed in parallel and Z-stack confocal sections were obtained with the identical acquisition settings. D, A maximal-intensity projection of confocal Z-stacks illustrates efferent terminals labeled by an anti-synapsin I purified antiserum (Gp118) (synapsin I, arrowheads) and synaptic ribbons (CtBP2) in the basolateral region of hair cells. Note the anti-synapsin I antibody faintly labels afferent postsynaptic terminals localized adjacent to synaptic ribbons. Scale bars: C, 10 µm; D, 5 µm.
(3.54 MB TIF)
A, Applying a total of 120 consecutive single stimuli to efferent fibers elicited 73 inhibitory postsynaptic potentials of a wide variety of magnitudes from a hair cell maintained in 4 mM Ca2+. B, The amplitude and area of the inhibitory postsynaptic potentials recorded in another cell are plotted against the delay between pairs of efferent shocks (mean ± SEM; number of events: N2 ms = 27, N3 ms = 39, N5 ms = 120, N10 ms = 36, N30 ms = 42, N40 ms = 37). C, Injection of depolarizing current pulses (lower traces) triggered action potentials in a hair cell. The amplitude and frequency of the spikes depended on the level of depolarization. The cell was maintained in a two-compartment chamber with 4 mM Ca2+ endolymph and 2 mM Ca2+ standard saline solution. D, Single action potentials occurred both at the onset of depolarization and as rebound spikes following hyperpolarizing current steps. The cell was exposed to 2 mM Ca2+ standard saline solution. E, Efferent stimulation (asterisk) inhibited spontaneous oscillatory activity at the resting potential. F, An excitatory efferent effect was occasionally obtained after the hyperpolarizing component of the inhibitory postsynaptic potential. The records show the responses to four consecutive efferent shocks from a hair cell bathed in 4 mM Ca2+ standard saline solution. Notice the failure of spike generation on one occasion (asterisk). The inset shows action potentials recorded in the same cell after depolarization as well as rebound spikes following hyperpolarizing current steps (scale bar: 100 ms).
(2.60 MB TIF)
Effect of efferent stimulation on the motion of a bundle. A, A hair bundle was mechanically stimulated by ±50 nm at 60 Hz with a flexible glass fiber (middle trace). Following efferent stimulation (bottom trace), the bundle's movement (upper trace) was significantly reduced. B, When another hair bundle was subjected to the same paradigm with 30 Hz stimulation of ±10 nm, its motion was augmented after efferent stimulation. Each trace is the average of 20–30 repetitions recorded in a two-compartment preparation with 0.25 mM Ca2+ endolymph and 2 mM Ca2+ standard saline solution. Upward deflections denote movement in the positive direction, towards the kinocilium.
(2.35 MB TIF)
A confocal Z-stack animation of the macular periphery illustrates the presence of efferent terminals exclusively at the basolateral region of hair cells. Synapsin I is shown in red, parvalbumin 3 in green, and DAPI-labeled nuclei in blue. The beginning of the movie corresponds to the somata of the supporting cells and the end to the apical surfaces and hair bundles of the hair cells.
(15.52 MB AVI)
A three-dimensional animation of a confocal stack of images of two extramacular hair cells depicts the presence of efferent terminals contacting their basolateral region. Synapsin I is shown in red and calretinin in green.
(0.69 MB AVI)
A confocal Z-stack animation of the macula illustrates the occurrence of synaptic ribbons of different sizes labeled by an antiserum against CtBP2 (green). Nuclei are labeled with DAPI (blue). The movie progresses from the apical surfaces of hair cells, which are faintly marked by anti-CtBP2, to the nuclear layer of supporting cells. Note the presence of a mitotic figure in the lower left corner.
(4.93 MB AVI)
We thank Dr. S. Abeytunge for development of the photomicrometer circuit, Ms. A. Le Boeuf for collaborative assistance with the construction of the photomicrometer, Mr. B. Fabella for assistance with photomicrometer calibration and programming of the experimental software, Drs. D. Andor and S. Lagier for their help with Matlab programming, Drs. A. Pasolli and E. Fuchs for access to the transmission electron microscope, Dr. R. Uthaiah for providing the anti-CtBP2 antiserum, and the Developmental Studies Hybridoma Bank (DSHB) at the University of Iowa for supplying the anti-tubulin and anti-SV2 antibodies. The members of our research group, especially Dr. J. Fisher and Mr. S. Patel, provided helpful comments on the manuscript.