Frequency selectivity of synaptic exocytosis in frog saccular hair cells

Abstract

The ability to respond selectively to particular frequency components of sensory inputs is fundamental to signal processing in the ear. The frog (Rana pipiens) sacculus, which is used for social communication and escape behaviors, is an exquisitely sensitive detector of sounds and ground-borne vibrations in the 5- to 200-Hz range, with most afferent axons having best frequencies between 40 and 60 Hz. We monitored the synaptic output of saccular sensory receptors (hair cells) by measuring the increase in membrane capacitance (deltaC(m)) that occurs when synaptic vesicles fuse with the plasmalemma. Strong stepwise depolarization evoked an exocytic burst that lasted 10 ms and corresponded to the predicted capacitance of all docked vesicles at synapses, followed by a 20-ms delay before additional vesicle fusion. Experiments using weak stimuli, within the normal physiological range for these cells, revealed a sensitivity to the temporal pattern of membrane potential changes. Interrupting a weak depolarization with a properly timed hyperpolarization increased deltaC(m). Small sinusoidal voltage oscillations (+/-5 mV centered at -60 mV) evoked a deltaC(m) that corresponded to 95 vesicles per s at each synapse at 50 Hz but only 26 vesicles per s at 5 Hz and 27 vesicles per s at 200 Hz (perforated patch recordings). This frequency selectivity was absent for larger sinusoidal oscillations (+/-10 mV centered at -55 mV) and was largest for hair cells with the smallest sinusoidal-stimuli-evoked Ca2+ currents. We conclude that frog saccular hair cells possess an intrinsic synaptic frequency selectivity that is saturated by strong stimuli.

Fig. 1.

SV pools and ΔC_m in whole-cell recordings from frog saccular hair cells. In these and all other whole-cell experiments, the intracellular Ca²⁺ buffer was 1 mM EGTA. (a) The SB is surrounded by docked (green) and nondocked (purple) SVs. “Afferent” labels a postsynaptic terminal. (Scale bar, 200 nm.) (b) Representative C_m traces for depolarizations to −20 mV lasting 10 ms (small response) or 200 ms (large response). We did not attempt to interpret C_m measurements during the depolarization-induced changes in membrane conductance, which have been blanked during the interval from the onset of the depolarization until 30 ms after V_m was returned to −80 mV (dashed lines). (c) Boxed region from b. C_m(t) was averaged in 100-ms windows (red) surrounding the blanked interval, and the difference was used to compute ΔC_m. (d) ΔC_m plotted as a function of step duration. Each point is the average ΔC_m for the first depolarization applied to each cell (mean ± SEM; n shown in parentheses; each cell contributed one ΔC_m value at one duration only; total, n = 66 cells). There was no significant difference for any pairwise comparison of means at 10, 25, 30, and 50 ms (see Table 2, which is published as supporting information on the PNAS web site). Green and purple dashed lines indicate estimated numbers of docked SVs and docked plus undocked SVs, respectively, that are associated with synaptic ribbons. (e) Mean Ca²⁺ influx during the stimuli in d. Data are for whole-cell, voltage-clamp recordings. Evidence for the Ca²⁺ dependence of ΔC_m and the ensemble-averaged C_m(t) traces for the data in d are shown in Figs. 8, 9b, and 10, which are published as supporting information on the PNAS web site.

Fig. 2.

Interposed hyperpolarization can increase ΔC_m. (a) Representative I_Ca during the four stimulus patterns shown in b. Each cell received patterns A and B, plus one or more of the other patterns. (b) Stimuli A–D are steps to −55 mV, except in a few cells in which the ΔC_m responses to stimuli A and B were too small to measure, in which case the depolarization was increased to −50 mV. Step timing: stimulus A, 30 ms; stimulus B, 2 × 10 ms separated by 10 ms; stimulus C, 2 × 14 ms separated by 2 ms; stimulus D, 2 × 2 ms separated by 26 ms. (c) Mean ΔC_m responses (±SEM) to the four stimulus patterns. The response to stimulus B was significantly larger than all others (see text). An alternative analysis in which the ΔC_m responses in each cell were first normalized by the ΔC_m response to stimulus waveform B gave similar results. (d) Individual ΔC_m from the 19 cells receiving stimulus patterns A and B. Responses to the first stimulus presentation are shown for waveforms A (solid gray bars) and B (striped bars).

Fig. 3.

ΔC_m responses to sinusoidal stimulation at 5 (gray), 50 (red), and 200 (blue) Hz. Each cell received either weak (±5 mV centered at −60 mV, solid bars) or strong (±10 mV centered at −55 mV, striped bars) stimulation for 1 s at each frequency with 30 s between stimuli. Presentation order was randomized among cells from the six possibilities. The strong stimuli were delivered around a more depolarized baseline than the weak stimuli to mimic the asymmetric transduction current in these cells (45). To avoid possible rundown effects, only responses to the first presentation of the three frequencies are included in this figure. (a) V_m and leak-subtracted I_m from one representative whole-cell recording. (b) Total Ca²⁺ influx did not differ across frequencies (same cells as d). (c) Ensemble-averaged R_s (Upper) and C_m (Lower) for all weak stimuli (same cells as d). (d) For weak stimuli (whole-cell, n = 11 cells), ΔC_m at 50 Hz was significantly greater (55 ± 12 fF) than at 5 Hz (20 ± 10 fF, P = 0.0001) or 200 Hz (18 ± 12 fF, P = 0.008). No significant differences between frequencies were observed for strong stimuli (whole-cell, n = 5 cells). (e and f) Similar results were obtained from perforated-patch recordings (n = 11 cells; only the weak stimuli were used). ΔC_m at 50-Hz stimuli (76 ± 17 fF) was significantly greater than ΔC_m at 5 Hz (26 ± 14 fF, P = 0.014) or 200 Hz (20 ± 13 fF, P = 0.006).

Fig. 4.

Correlation of I_Ca amplitude during weak sinusoidal stimulation with the frequency selectivity of ΔC_m. Combined whole-cell and perforated-patch data. (a) Representative current traces from one cell that had a large I_Ca and one cell that had a small I_Ca. (b) To quantify the preference for 50 Hz in each cell, we divided the ΔC_m at 50 Hz by the mean ΔC_m at all three frequencies and correlated this measure with the mean integrated Ca²⁺ influx across frequencies for perforated-patch (gray inverted triangle) and whole-cell (black inverted triangle) recordings. The regression line and correlation coefficient are shown.

Fig. 5.

Frequency selectivity in individual cells. ΔC_m responses from the subset of cells in Fig. 3, from which two or more stimulus presentations were delivered at each frequency. Sinusoidal stimuli were delivered for 1 s in a fixed order within each cell (e.g., 5, 50, 200, 5, 50, 200, and so on) and randomized between cells, with a 30-s interstimulus interval. (a) Averaged C_m traces for 10 stimulus presentations at each frequency to one cell (perforated-patch recording) collected over 15 min. (b) Means (±SEM) of within-cell means. Whole cell (n = 11): ΔC_m at 50 Hz (61 ± 9 fF) was significantly greater than at 5 Hz (35 ± 8 fF, P = 0.001) and 200 Hz (27 ± 11 fF, P = 0.0006); perforated patch (n = 11): ΔC_m at 50 Hz (70 ± 17 fF) was significantly greater than at 5 Hz (19 ± 12 fF, P = 0.012) and 200 Hz (20 ± 12 fF, P = 0.006). (c) Within-cell ΔC_m means (±SEM, number of repetitions in parentheses) are plotted across frequencies for 13 cells. Ensemble-averaged C_m(t) traces for two cells are shown in Fig. 10, which is published as supporting information on the PNAS web site.

Fig. 6.

Steep voltage sensitivity of I_m and ΔC_m. Continuous recording of V_m, I_m, and C_m in a cell that was presented with a series of sinusoidal voltage oscillations (50 Hz, ±5 mV, 10-s duration, 10 s between stimuli) centered at −70, −65, −60, and −55 mV (see also Fig. 7). (Inset) Enlargement of stimulus response (V_m) and current response (I_m) around the onset of the −60 ± 5 mV stimulus. The V_m oscillation during the rest period that ends at t = 65 ms appears as a broad black band in V_m and is the result of the 1.5-kHz probe from the lock-in amplifier, which is turned off during the 50-Hz stimulation.