Spikes and membrane potential oscillations in hair cells generate periodic afferent activity in the frog sacculus

Abstract

To look for membrane potential oscillations that may contribute to sensory coding or amplification in the ear, we made whole-cell and perforated-patch recordings from hair cells and postsynaptic afferent neurites in the explanted frog sacculus, with mechanoelectrical transduction (MET) blocked. Small depolarizing holding currents, which may serve to replace the in vivo resting MET current, evoked all-or-none calcium spikes (39-75 mV amplitude) in 37% of hair cells tested, and continuous membrane potential oscillations (14-28 mV; 15-130 Hz) in an additional 14% of cells. Spiking hair cells were on average taller and thinner than nonspiking hair cells, and had smaller outward currents through delayed rectifier channels (I(KV)) and noninactivating calcium-activated potassium channels (I(BK,steady)), and larger inward rectifier currents (I(K1)). Some spiking hair cells fired only a brief train at the onset of a current step, but others could sustain repetitive firing (3-70 Hz). Partial blockade of I(BK) changed the amplitude and frequency of oscillations and spikes, and converted some nonspiking cells into spiking cells. Oscillatory hair cells preferentially amplified sinusoidal stimuli at frequencies near their natural oscillation frequency. Postsynaptic recordings revealed regularly timed bursts of EPSPs in some afferent neurites. EPSP bursts were able to trigger afferent spikes, which may be initiated at the sodium channel cluster located adjacent to the afferent axon's most peripheral myelin segment. These results show that some frog saccular hair cells can generate spontaneous rhythmic activity that may drive periodic background activity in afferent axons.

Figure 1.

Spontaneous activity in hair cells. Current-clamp recordings (I = 0) from four hair cells (two timescales in A–C). A, Repetitive spikes (39 mV amplitude) with occasional failures. B, Variable amplitude oscillations. C, Small voltage fluctuations with no obvious periodicity. D, Oscillations interrupted by a spike followed by a hyperpolarization and gradual return to oscillation. A second hyperpolarization later in the trace was not preceded by a spike. When depolarized by ≥20 pA, this cell fired a repetitive spike train (data not shown). Ruptured-patch recordings without (A–C) or with (D) amiloride in the bath.

Figure 2.

Holding current modulates spikes and oscillations. Each panel shows two current-clamp recordings (red, black) from one cell, with holding current and spike or oscillation frequency shown in the corresponding color. A, Depolarization increased the spike rate and reduced timing variability. B, Depolarization increased oscillation frequency and decreased the amplitude.

Figure 3.

Spike or oscillation amplitude and frequency as a function of holding current, and in relation to cell morphology. A, Spike amplitudes at different holding levels for nine hair cells that fired repetitive spike trains. The abrupt appearance of full-size spikes with increased holding current reveals their all-or-none character. B, Spike frequencies for the same cells as in A. C, Maximum oscillation amplitudes at different holding levels in seven nonspiking hair cells. D, Oscillation frequencies for the same cells as in C. Different colored lines in A and B and in C and D are for corresponding cells. E, Distribution of maximum amplitudes of voltage fluctuations observed in 49 hair cells: spiking (red), oscillating (blue), and nonoscillating (black). F, Positive relationship between hair cell length:apical diameter ratio and maximum amplitude. G, Negative relationship between length:apical diameter ratio and peak-power frequency at the holding current that produced the maximum amplitude. Markers in F and G are colored as in E.

Figure 4.

Responses to sinusoidal currents. A, Spiking hair cell was held at I = 0 before application of a DC + sinusoid stimulus (50 ± 50 pA) beginning at time 0. The frequency was swept from 0 to 34 Hz (top axis) in 15 s (bottom axis). Spikes ceased above ∼20 Hz (red arrow), leaving a smaller oscillatory response to the higher-frequency stimulus. B, Another spiking hair cell ceased firing at ∼30 Hz (red arrow) when stimulated as in A, but without a DC offset (0 ± 50 pA). C, An oscillating hair cell showed bandpass amplification centered near 20 Hz in response to stimulation by 80 ± 20 pA (black) but not 40 ± 20 pA (gray). Note the slow appearance of oscillations during the first 2 s after application of the 80 pA offset. D, In another oscillating hair cell (perforated patch), changing the stimulus from 15 ± 20 pA to 55 ± 20 pA shifted the envelope of amplification to a higher frequency. E, A hair cell (perforated patch) that did not oscillate at any constant holding current amplified oscillatory inputs in a broad band centered near 20 Hz (black). Application of 100 nm iberiotoxin (blue) to block I_BK caused this cell to generate 20 mV oscillations for DC stimuli, and shifted the center frequency for amplification down by approximately one octave.

Figure 5.

Membrane currents vary between spiking and nonspiking hair cells. A, Mean responses to V_cmd steps from −70 to −30 mV for spiking (n = 10), oscillating (n = 4) and nonoscillating (n = 13) hair cells, leak subtracted (P/10 or P/20) with no series resistance compensation. B, Same cells as in A but for V_cmd steps from −70 to −120 mV. C, Scatter plot of steady-state current densities evoked by depolarizing versus hyperpolarizing steps in individual cells. In all but one spiking hair cell, the magnitude of the inward current density at V_cmd = −120 mV exceeded the outward current density at V_cmd = −30 mV. The opposite was true for nonoscillating hair cells, and oscillating hair cells were intermediate. For V_cmd = −120 mV, there was little overlap in the distributions of current densities between spiking and nonoscillating cells. D, Negative correlation between zero-current potential (V_rest) and inward current density at V_cmd = −120 mV. E, F, Amplitude (E) and frequency (F) of the largest spike, oscillation, or small fluctuation in each cell plotted versus current density at V_cmd = −30 mV. Pearson product-moment correlation coefficients (r) were 0.29 for E and −0.52 for F. G, Voltage-clamp data (thick black traces, top and bottom) from an oscillating hair cell were well fit (red traces, top and bottom) by the sum of four depolarization-activated currents (top traces: I_Ca, green; I_BK,transient, light blue; I_BK,steady, dark blue; I_KV, yellow) or two hyperpolarization-activated currents (bottom traces: I_K1, gray; I_h, purple). Note the small I_h and I_K1 at V_cmd = −120 mV in this cell. H, same as G, for a spiking hair cell that had larger I_K1 and smaller I_BK,steady and I_KV. I, Average fitted current densities (mean ± SEM) for spiking, oscillating, and nonoscillating hair cells (*p < 0.05; **p < 0.01).

Figure 6.

Variability within the class of spiking hair cells. A–C, Voltage clamp (left), current–voltage relationships (middle), and current clamp samples from three different spiking hair cells. Voltage-clamp records are individual traces of net current without RC compensation or leak subtraction. From the holding potential of −70 mV, 50 ms steps were delivered in 10 mV increments in the range of −150 to +30 mV. Some traces were removed for clarity. Red traces are for V_cmd = −40 mV. After the capacitive transient crossed zero current (at ∼0.4 ms), the trace was divided into 0.5 ms windows (rainbow colored markers) for calculating the current–voltage relationships, which were corrected for series resistance errors based on the net current in each time window. Note the different x-axis scales for the current-clamp data. The red arrows denote the onset of a current step of +50 pA in A and +100 pA in B and C. A, In current clamp, a “phasic” (completely adapting) hair cell fired a train of large amplitude spikes at step onset and then stopped. In the current–voltage plot, note the small but persistent net inward current for potentials between −45 and −15 mV. B, Repetitively firing hair cell at low frequency. C, Repetitively firing hair cell at higher frequency (perforated-patch recording).

Figure 7.

Influence of I_Ca and I_BK on hair cell spikes in the frog sacculus. A, Leak-subtracted (P/10) voltage-clamp traces in response to depolarizing commands from −70 to −30 mV before (black) and after (blue) adding 100 nm iberiotoxin to the bath to partially block I_BK. The blocked current was calculated by subtraction (red). Note the more obvious inward Ca²⁺ current after removing I_BK by treatment with iberiotoxin. B, Top, Leak-subtracted (P/10) voltage-clamp traces for 20 ms steps to −30 mV for control (black), 100 nm iberiotoxin (blue), and 100 nm iberiotoxin + 50 μm nifedipine (yellow). Bottom, Current-clamp records for 5 s steps from 0 to +100 pA delivered to the cell in B under each pharmacological condition. Spikes were promoted by iberiotoxin, and then abolished by nifedipine. C, An oscillating hair cell in response to 1 s current clamp stimuli to +100 pA (top) and +200 pA (bottom) in normal (1.8 mm, black) or high (10 mm, green) extracellular Ca²⁺.

Figure 8.

Innervation of the frog sacculus. A, Axons visualized by confocal microscopy after backfilling the anterior branch of the VIIIth cranial nerve with fluorescein–dextran. Afferent neuron somata (bottom right) that innervate the sacculus, utriculus, and the anterior and lateral semicircular canals can be seen proximal to the divergence of the saccular branch. The saccular macula (top center) glows with the diffuse fluorescence of many small axons. Most of the brightly stained (large-diameter) axons shed their myelin near the proximal margin of the macula, a few at the lateral and distal margins. Top inset, Small-diameter neurites make multiple synaptic contacts (bright spots) as they weave between hair cells (dark circles). Bottom inset, Numerous terminal and en passant boutons (green; fluorescein–dextran) contact each hair cell (red; FM1-43 stain). For electrical recordings from the semi-intact epithelium, the nerve was cut short (red line). B, Schematic shows placement of the recording electrode (blue) on a fine neuritic branch. C, DIC image of a furrowed epithelium shows a neurite (arrow) making a bouton contact (arrowhead) onto a hair cell (hc). D, At the level of the basement membrane, a myelinated axon (arrows) expands into an nonmyelinated varicosity (*). It then branches into slender nonmyelinated neurites (arrowheads) that extend into the synaptic layer above the basement membrane, becoming accessible to recording electrodes. To the top left is a damaged axon in which the varicosity has been disconnected from the myelinated segment. E, A monoclonal pan-Na_V channel antibody (Rasband et al., 1999) labeled neurites near the basement membrane, often at the first branch point (entire macula shown above, enlargement below). Na channel clusters extended 2–5 μm into each branch (forked structures in bottom; red arrow points to a trifurcation). F, Neurofilament antibody (3A10) strongly labeled nonmyelinated neurites (entire macula shown above, enlargement below). Scale bars (in μm): A, 50 (right), 20 (top inset), 15 (bottom inset); B, 10; C, D, 5; E, F, 50 (top), 10 (bottom).

Figure 9.

Bursts of background synaptic activity in postsynaptic fibers are composed of brief individual events sensitive to CNQX. A, Two lines of continuous data acquired at the onset of periodic synaptic activity show discrete EPSC-events that superimpose into bursts. Large bursts triggered all-or-none action potentials (evident as action currents) in afferent neurites that escaped voltage clamp (V_cmd = −80 mV). B, Examples of subthreshold activity in current clamp (B1) and voltage clamp (B2). Each of the four traces shows one burst. C, EPSC bursts (C1) from a different cell were 75% blocked by adding 10 μm CNQX to the bath (C2). No action potentials occurred in B and C.

Figure 10.

Periodic EPSC burst activity at rest is abolished in high K⁺. A1, EPSP bursts displayed on three different timescales, with 1 μm TTX in the bath to prevent spiking. Bursts were 10–35 ms in duration and contained a variable number of EPSC-events of variable amplitude (top trace). EPSCs were absent between bursts (middle trace). The intervals between bursts (bottom trace) were approximately integer multiples of 150 ms. A2, Autocorrelation (normalized log scale, y-axis) of EPSC-event times for the same cell showed preferred EPSC-event intervals (in milliseconds, x-axis). The large peak near 0 ms represents the short intervals within a burst. The broad peaks centered near 150 ms and multiples thereof represent the periodicity of successive bursts, despite skipped fundamental intervals. B, Data obtained later in the same recording, immediately before (black) and after (blue) a change of the extracellular medium from 2 to 40 mm K⁺. C, Histogram of intervals between EPSC-events before solution exchange (black) shows short intervals that correspond to closely spaced EPSC-events within each burst, and long intervals that correspond to the time between the last event in one burst and the first event in the next. After solution exchange (blue), the interval histogram showed an exponential distribution characteristic of a Poisson process. D, Bath depolarization with high K⁺ resulted in larger EPSC events compared with control. Data from 30 s in each condition. E, EPSC event times were measured at the peaks (red x's). Current amplitudes were measured as the difference from baseline (green x's) to peak.