Evidence that fast exocytosis can be ... - Wiley Online Library

439

J Physiol 560.2 (2004) pp 439–450

Evidence that fast exocytosis can be predominantly mediated by vesicles not docked at active zones in frog saccular hair cells Brian W. Edmonds, Frederick D. Gregory and Felix E. Schweizer Department of Neurobiology, David Geffen School of Medicine at UCLA, University of California, Los Angeles, 650 Charles E. Young Drive South, Los Angeles, CA 90095, USA

The prevailing model of neurotransmitter release stipulates that Ca2 + influx triggers the rapid fusion of vesicles that are docked at presynaptic active zones. Under this model, slower tonic release is supported by vesicles clustered nearby that have to translocate to the release sites before fusion. We have examined this hypothesis at the afferent synapse of saccular hair cells of the leopard frog, Rana pipiens. Detailed morphological measurements at this ribbon synapse show that on average 32 vesicles are docked at each active zone. We show that at this ‘graded’ synapse, depolarization produces an exocytotic ‘burst’ that is largely complete within 20 ms after fusion of 280 vesicles per active zone, almost an order of magnitude more than expected. Recovery from paired pulse depression occurs with a time constant of 29 ms, indicating that replenishment of this fast-fusing pool of vesicles is also fast. Our results suggest that non-docked vesicles are capable of fast fusion and that these vesicles constitute the vast majority of the fast-fusing pool. The view that the population of fast-fusing presynaptic vesicles is limited to docked vesicles therefore requires re-evaluation. We propose that compound fusion, i.e. the fusion of vesicles with each other before and/or after they fuse with the membrane can explain multivesicular release at this synapse. (Resubmitted 6 April 2004; accepted after revision 10 August 2004; first published online 12 August 2004) Corresponding author F. E. Schweizer: Department of Neurobiology, David Geffen School of Medicine at UCLA, University of California, Los Angeles, 650 Charles E. Young Drive South, Los Angeles, CA 90095, USA. Email: [email protected]

Exocytotic fusion of neurotransmitter-containing vesicles with the plasma membrane is the basis of most interneuronal communication in the nervous system. Neurotransmitter release usually occurs at specialized presynaptic active zones (Palade & Palay, 1954) where voltage-gated Ca2+ channels and synaptic vesicles are colocalized. A subpopulation of vesicles at the active zone is in contact with or docked at the presynaptic membrane (Couteaux & Pecot-Dechavassine, 1974). The short time delay between Ca2+ entry and neurotransmitter release (Katz & Miledi, 1965; Augustine et al. 1985; Sabatini & Regehr, 1996) does not allow for transport or diffusion of non-docked vesicles to the presynaptic membrane. Therefore, only docked vesicles are thought to be immediately available for fusion (Schikorski & Stevens, 2001) and the time course of release should slow if the pool of docked vesicles is exhausted. In this study we have used frog saccular hair cells to test the hypothesis that at ribbon

B. W. Edmonds and F. D. Gregory contributed equally to this work

C The Physiological Society 2004

synapses the number of presynaptic vesicles available for fast fusion is limited by the number of docked vesicles. Ribbon synapses are characterized by the presence of a presynaptic, electron-dense structure that is either plate-like (synaptic ribbon, e.g. found in photoreceptors (Sjostrand, 1958)) or spherical in shape (synaptic body, e.g. present in auditory hair cells (Smith & Sjostrand, 1961) and fish lateral line hair cells (Hama, 1965)). Club-shaped synaptic bodies have been described in fish electroreceptors (Wachtel & Szamier, 1966) and intriguing changes in the shape of the synaptic body during development from spherical to plate-like have been observed in the mouse cochlea (Sobkowicz et al. 1982). Synaptic bodies are studded with small synaptic vesicles and lie in close proximity to the active zone, two features that led to the proposal of their function as conveyor belts, delivering vesicles to the docking sites prior to release (Bunt, 1971; Gleisner et al. 1973). The conveyor belt model is consistent with the hypothesis that docked vesicles form a pool that is released with short latency following onset of Ca2+ influx. Those vesicles associated DOI: 10.1113/jphysiol.2004.066035

440

B. W. Edmonds and others

with the synaptic body could constitute a second pool that is released with slower kinetics, limited by the time required for vesicle transport to the plasma membrane (Mennerick & Matthews, 1996; von Gersdorff et al. 1996; Neves & Lagnado, 1999). We chose mechano-sensitive hair cells from the sacculus of leopard frogs for our studies since their morphological and physiological properties have been characterized in unsurpassed detail. Also, the lack of processes makes it possible to gain electrical access to the release sites and compare the physiological properties of fast exocytosis with the reported morphology. The 3-dimensional ultrastructure of the afferent synapse between saccular hair cells and primary afferents from the VIIIth nerve has been described (Lenzi et al. 1999, 2002). On average, there are 32 vesicles docked at each active zone, while 380 vesicles are tethered to the synaptic body. An additional ‘cloud’ of outlying, cytoplasmic vesicles that are neither tethered to the synaptic body nor the plasma membrane surrounds the active zone. The rise in intracellular [Ca2+ ] in response to a depolarization is sharply restricted to the hair cell active zone since voltage-gated Ca2+ channels responsible for synaptic signalling are found almost exclusively in this region (Roberts et al. 1990) and a highly mobile Ca2+ buffer (Edmonds et al. 2000) prevents the local increase of free Ca2+ ions from spreading to adjacent regions (Roberts, 1993, 1994; Issa & Hudspeth, 1996; Zenisek et al. 2003). Given these data, it has been proposed that three morphologically defined vesicle pools fulfil physiologically distinct functions: docked vesicles support phasic release, ribbon-associated vesicles support tonic release and the outlying, cytoplasmic vesicles adjacent to the ribbon replenish vesicles leaving the ribbon (see, for example, von Gersdorff & Matthews, 1997; Neves & Lagnado, 1999; Holt et al. 2004). Docked vesicles would thus be expected to fuse with fast kinetics and ribbon-associated vesicles, requiring translocation toward the membrane, would constitute a pool fusing with slower kinetics, limited by the speed of vesicle movement. The outlying pool might not be detectable as a kinetically distinct, third component if replenishment is not rate limiting. In this study, we tested whether the number of fast-fusing vesicles is limited to the number of morphologically docked vesicles. We defined fast fusion as fusion occurring within 20 ms, i.e. fusion prior to the time it would take a molecular motor to move a vesicle by one vesicle diameter (> 20 ms; see Discussion). This pragmatic definition should limit the ‘fast phase’ to vesicles docked at the plasma membrane. By measuring cell membrane capacitance to assay vesicle fusion with the plasma membrane we found two phases of exocytosis: a fast phase that is exhausted in 20 ms and a slower phase that continued for at least 5 s, consistent with the above model. Surprisingly, we find that the fast phase is supported by the fusion of about 9 times more vesicles than those previously observed as docked at the

J Physiol 560.2

presynaptic membrane. Our results therefore indicate that fast exocytosis at this ribbon synapse (< 20 ms) is predominantly mediated by vesicles that are not docked at active zones at the time of stimulus onset. Methods Isolation of hair cells

Normal Ca2+ Ringer solution contained (mm): 112 NaCl, 2 KCl, 1.8 CaCl2 , 0.7 MgCl2 , 3 d-glucose and 5 Hepes (pH with NaOH to 7.25). For low Ca2+ Ringer solution, MgCl2 was omitted and CaCl2 was lowered to 0.1 mm. The osmolality of the external solutions was adjusted to 220–225 mmol kg−1 . Unless stated otherwise, all reagents were obtained from Fluka (Sigma-Aldrich Corp.). We used hair cells from the sacculus of small (5.5–7.5 mm in length) leopard frogs (Rana pipiens, wild-caught male and female; various suppliers). Frogs were chilled on ice and killed using the double pithing method according to protocols approved by the UCLA Institutional Animal Care and Use Committee. Sacculi were removed from the inner ear and incubated for 6 min in low Ca2+ Ringer solution (on ice) to which 1 mm MgCl2 and 1 mm EGTA had been added. Sacculi were then transferred to low Ca2+ Ringer solution containing subtilisin (Protease Type XXIV; Sigma) at 50 µg ml−1 for 12–15 min (at room temperature). Cells were dissociated onto concanavalin-A-coated glass coverslips by gently scraping the macular epithelium with an eyelash attached to the end of a handheld probe. Cells were visualized on an inverted microscope (Zeiss, Axiovert 25 or S100; Carl Zeiss Inc. Oberkochen, Germany) using a 32 × objective. Electrophysiological recordings

Recordings were made at room temperature (∼23◦ C) using the perforated patch configuration of the whole-cell voltage clamp method. Patch pipettes were made with borosilicate glass (TW150F-4; World Precision Instruments, Sarasota, FL, USA) and pulled to tip resistances of 1.5–3.5 M measured in our standard recording solutions. To reduce stray capacitance, tips were coated with HIPEC R6101 (Dow Corning, Midland, MI, USA; pre-incubated for 6 h at 90◦ C). Approximately 0.5 mm of the pipette tip was front-filled with intracellular solution containing (mm): 114 caesium aspartate (l-aspartate from Sigma), 0.08 CaCl2 , 2 MgCl2 , 5 Hepes, 1 EGTA (pH 7.25, 220–225 mmol kg−1 ). The pipette was back-filled with intracellular solution to which nystatin (50 µg ml−1 ; stored as 1000 × stock in DMSO at 20◦ C; Sigma) had been added. Recordings were made with an Axopatch 200 A (Axon Instruments) patch clamp amplifier or an Optopatch (Cairn Research, Faversham, UK) using jCLAMP software (Scisoft Co., New Haven, CT, USA). Membrane voltages were corrected for a C The Physiological Society 2004

Fast fusion without stable docking

J Physiol 560.2

measured liquid junction potential error of −13 mV between the intracellular and extracellular solutions. Series resistance of 14.2 ± 3.2 M (mean ± s.e.m.; n = 49) was uncompensated. Unless indicated otherwise, data were filtered at 30 kHz and sampled at 100 kHz with further filtering imposed in software. Ca2+ currents shown in Figs 2 and 3 were leak subtracted using a P/5 protocol and, for each depolarization, the nominal voltage was corrected for the voltage error due to the series resistance. Peak currents at the nominal voltages were then obtained by cubic spline interpolation and the average I–V plot (Fig. 2B) was constructed by averaging the interpolated current values at voltages between −120 and +20 mV. I–V plots from individual cells were fitted with the equation: I = gmax (V − Vrev )/1 + exp((V 1/2 − V /k))

(1)

where g max is the maximum conductance, V rev the reversal potential, V 1/2 the voltage at which the half-maximal current is achieved, and k is the slope factor indicating the steepness of the voltage dependence. The average values from the fit were: g max = 0.011 ± 0.001 nS, V rev = 39.2 ± 1.5 mV, V 1/2 = −41.9 ± 1.2 mV and k = 7.05 ± 0.14 mV. An I–V curve calculated from these average values is superimposed onto the data (dashed line in Fig. 2B). The conductance versus voltage plot (Fig. 2C) was obtained by measuring the chord conductance between the test voltage and the reversal potential for the current. The chord conductance was calculated at all voltages as the current divided by the driving force and normalized using g max . The data were then averaged and fitted with a Boltzman distribution: g = 1/exp((V 1/2 − V )/k))

(2)

Capacitance measurements

Capacitance measurements were made using a dual sine-wave voltage (DSWV) stimulus protocol when the series resistance was stable and less than 20 M (Donnelly, 1994; Rohlicek & Schmid, 1994; Gillis, 1995). Cells were held at −75 mV, and a DSWV (r.m.s. amplitude: 6 mV; maximal negative and positive excursions: −81.75 mV and −63 mV, respectively) of 390.5 and 781 Hz was superimposed on the applied holding potential. Admittance analysis was then performed with jCLAMP software on the Fast Fourier Transform (FFT) of the current response every 2.56 ms. The analysis yielded estimates for membrane capacitance (C m ), series resistance (Rs ), membrane resistance (Rm ) and the holding current (I) during the course of the experiment (Santos-Sacchi et al. 1998) at every time point. In contrast to some other techniques for measuring capacitance (for a review see Gillis, 1995), this method employs the same algorithm throughout the C The Physiological Society 2004

441

recording, does not require cancellation of baseline Rs and C m , and makes no assumptions about Rm staying constant during the recording (Santos-Sacchi, 2004). To validate the dual sine-wave method, we measured baseline capacitance in 10 cells using both the dual sine-wave approach and the RC cancellation circuitry of the Optopatch. The values obtained with both methods were indistinguishable (2-sine wave: 14.69 ± 0.83 pF; RC cancellation: 14.7 ± 0.74 pF; n = 10). To verify that large changes in capacitance can be accurately measured with this technique, we used a model cell (Axon Instruments; nominal values: C m = 33 pF, Rs = 10 M, Rm = 500 M) and placed a capacitor of nominally 4.7 pF in series with C m . Using the DSWV method we determined a baseline capacitance of 37.14 ± 0.02 pF. Rapid removal of the 4.7 pF capacitor elicited a drop in the measured capacitance of 4.64 ± 0.05 pF, in excellent agreement with its nominal value. Not surprisingly, this large change in capacitance could not be accurately measured using a single sine wave and the built-in lock-in capabilities of the Optopatch amplifier (measured C m change: 0.48 ± 0.05 pF), since the phase angle of the measurement has to be readjusted when large changes in C m occur (Bookman et al. 1991; Gillis, 1995). However, when the automatic track-in feature of the Optopatch was enabled (Johnson et al. 2002) an accurate measurement could be made (measured C m change: 4.61 ± 0.07 pF). A similar conclusion was recently reached by Thoreson et al. (2004) who compared the DSWV method implemented in jCLAMP with the square wave method. Furthermore, our values for total membrane capacitance of frog saccular hair cells are also consistent with values published in the literature (e.g. 8–16 pF (Smotherman & Narins, 1998); 15 ± 3 pF (Lioudyno et al. 2000); 7.1–17.3 pF (Holt et al. 2001); 16 pF used for model (Catacuzzeno et al. 2003)). We conclude that the DSWV method allows for accurate and reliable detection of large capacitance changes and offers the advantage over single sine-wave approaches since C m , Rs and Rm can be continuously monitored during the course of the experiment. Additional analysis on the parameters was performed off-line with Igor Pro 4.0 (WaveMetrics, Inc., Lake Oswego, OR, USA) and Origin 6.0 (MicroCal, Northampton, MA, USA). In some experiments, hair cells were held at −85 mV, enabling us to use a DSWV of larger amplitude (r.m.s. amplitude: 13 mV; maximal negative and positive excursions: −99.8 mV and −59 mV, respectively) without activating voltage-dependent conductances, and thus improve the signal to noise ratio of our estimates. In paired pulse experiments (Fig. 6) we increased the temporal resolution of our parameter estimates to once every 0.64 ms using a higher frequency DSWV (1562.5 and 3125 Hz). The DSWV was turned off during depolarization to avoid interference with Ca2+ channel activation.

442


Estimation of vesicle number

Results

Our interpretations depend, to some extent, on correctly estimating the number of fused vesicles. We used a value of 37 aF per vesicle based on the vesicle diameter of 39 nm measured from three-dimensional reconstructions of electron micrographs of frog saccular hair cells (Lenzi et al. 1999). While shrinkage of the preparation prior to EM analysis might have led to an underestimate of the vesicle diameter (Lenzi et al. 1999) a good correspondence between vesicle diameters determined from electron micrographs and by membrane capacitance measurements has recently been reported (Klyachko & Jackson, 2002). We have assumed the ‘canonical’ membrane capacitance of 1 µF cm−2 , although recent measurements of the specific membrane capacitance for various cell types have indicated that the correct value may be lower (≤ 0.8 µF cm−2 ) (Solsona et al. 1998; Gentet et al. 2000; Roth & Hausser, 2001). We are thus probably underestimating, rather than overestimating, the actual number of vesicles fusing. Our interpretations also depend on a correct estimate of the number of active zones per frog saccular hair cell. This number has been determined by different investigators using diverse methods. Issa and Hudspeth, using the fluorescent dye fluo-3 to label synaptic bodies, estimated the number at 18 ± 4 (Issa & Hudspeth, 1994), a result recently confirmed using evanescent wave microscopy (Zenisek et al. 2003). Roberts and colleagues used the dual approach of focal recordings, to identify ion channel clusters, and electron microscopy to identify active zones. Their results suggested that hair cells contain 20 channel clusters and ∼19 active zones (Roberts et al. 1990). In turtle hair cells, a maximum of six active zones per cell have been reported (Tucker & Fettiplace, 1995). For our interpretations, we have taken the conservative and generally accepted value of 20 active zones per frog saccular hair cell.

Measurement of exocytosis

Statistical methods

We utilized bootstrapping methodology to test the confidence of our conclusions from Fig. 4 by simulating the null hypothesis without making any assumptions about the statistical distribution of the data (Efron & Tibshirani, 1991). For a given duration of depolarization and stimulation strength, data were placed into a vector in Resampling Stats (www.resample.com). These data were randomly sampled N times with replacement (N is the total number of data points), a mean was calculated and this was repeated 15 000 times, allowing the construction of a frequency histogram of the re-sampled means and the estimation of statistical significance (P < 0.05). For other data a standard t test was performed as the test for statistical significance. For averaged data, values are given as means ± s.e.m.

J Physiol 560.2

We measured exocytosis as a change in cell surface area by monitoring cell membrane capacitance (Neher & Marty, 1982). To avoid wash-out of endogenous Ca2+ buffers (Edmonds et al. 2000; Moser & Beutner, 2000) we used the perforated-patch configuration of the patch clamp technique for all experiments. To avoid confounding effects of repetitive stimulation, only responses to the very first depolarization (or pairs of depolarization) for each cell were included in the analysis. The cell membrane capacitance (C m ) of unstimulated hair cells was between 8 and 22 pF (14.9 ± 3.3 pF, n = 49; see Methods) and remained stable unless the cell was stimulated by depolarization. Figure 1A shows a typical example of a recording from a hair cell with an initial cell membrane capacitance of 14 pF. Depolarization from the holding potential of −75 mV to −20 mV for 25 ms activated an inward current (Fig. 1A; I), mostly carried by Ca2+ ions since K+ conductances were blocked by internal caesium ions and these cells do not express voltage-dependent Na+ conductances (Lewis & Hudspeth, 1983). Upon repolarization, the membrane capacitance increased by 400 fF (Fig. 1A; C m ). Using an average capacitance of 37 aF per vesicle (Lenzi et al.), the observed increase corresponds to the release of 540 vesicles at each of the 20 active zones (Roberts et al. 1990; Issa & Hudspeth, 1994; Zenisek et al. 2003). As reported previously for frog saccular hair cells (Parsons et al. 1994), the capacitance signal decayed back to baseline over many seconds (data not shown), most probably reflecting slow endocytosis (see Discussion). Since all recordings were performed using the perforated-patch technique, it seems unlikely that fast endocytosis was inhibited by hydrostatic pressure in the patch pipette as has been reported for retinal bipolar cells (Heidelberger et al. 2002). The observed slow retrieval of the bulk of exocytosed membrane makes it unlikely that our measurements of exocytosis were compromised by endocytosis, at least not for short stimulus durations. As indicated in Fig. 1A, the series resistance (Rs ) was stable during the experiment; however, the input resistance (Rm ) was higher immediately after repolarization and quickly relaxed back to baseline. The transient increase in resistance, which is paralleled by an outward current (I), probably reflects voltage-dependent closing of inward rectifier K+ channels as it was mostly blocked by the addition of 1 mm CsCl to the external solution (Fig. 1B) (Holt & Eatock, 1995). Most experiments were performed without external caesium and since external caesium did not appear to alter capacitance changes (20 ms depolarization in Cs+ : 301 ± 125 fF, n = 5; 25 ms depolarization without Cs+ : 248 ± 67 fF, n = 12; P = 0.7), data with and without caesium were pooled. C The Physiological Society 2004


J Physiol 560.2

443

Kinetics of Ca2 + current

The initial rate of exocytosis following depolarization depends on a number of factors including the amplitude and the rate of activation of the Ca2+ current (I Ca ). Therefore, we examined the voltage dependence of the amplitude and rise time of I Ca . In agreement with previous reports (Armstrong & Roberts, 1998; Rodriguez-Contreras & Yamoah, 2001), the Ca2+ current in hair cells showed little inactivation (Fig. 2A) and began to activate with weak depolarizations to around −65 mV with a half-maximal current at −50 mV and a peak inward current at −20 mV (Fig. 2B). We also estimated channel activation by calculating the conductance at all test voltages (see Methods). Again, full activation was achieved at −20 mV whereas at −50 mV only 25% of the maximal conductance had been activated (Fig. 2C). Figure 3A shows two recordings of I Ca in response to depolarizing steps to −50 and −20 mV. The rising phase of each of these currents was well fitted by a third order exponential: ICa (t) = I¯Ca (1 − e−t/τ )3 ,

(3)

where I¯Ca is the maximum value for I Ca and τ is the time constant associated with the movement of 1 of 3 hypothetical gating charges within the Ca2+ channel (Fig. 3A and B) (Hudspeth & Lewis, 1988). Mean values for τ were 687 ± 156 µs (n = 7) and 316 ± 47 µs (n = 9) for jumps to −50 and −20 mV, respectively. The average 10–90% rise times were 1.5 ms and 0.7 ms for steps to −50 and −20 mV, respectively. Activation of I Ca was very fast at all potentials tested (Fig. 3B) and thus appears too fast to significantly contribute to the time course of exocytosis (see below). Depolarizations to −50 mV and to −20 mV were used in subsequent experiments since they elicit fast activation of sustained I Ca with different amplitudes. Fast phase of exocytosis

To determine the secretory capacity of hair cells, we depolarized the voltage clamped cells to −20 mV for a range of durations. Figure 4A shows examples of the observed capacitance changes. The magnitude of the capacitance change increases with longer depolarizations; however, longer depolarizations trigger less release than might be expected from a linear relationship suggesting saturation of the fast exocytotic process. A plot of the mean capacitance increase for stimuli in the range of 2–30 ms (Fig. 4B, filled triangles; n = 3–13) indeed demonstrates a fast but limited phase of exocytosis. A 20 ms depolarization to −20 mV elicited an average capacitance increase of 253 ± 78 fF (n = 8; range 109–783 fF; 4 ms: mean = 54 ± 19 fF, n = 8, range 2–124.5 fF; 10 ms: mean: 226 ± 87 fF, n = 9, range 5.7–844 fF). These data were well described by a ‘fast’ exponential of order 4.8 that saturates at 209 fF with a time constant of 2.7 ms (Fig. 4B). C The Physiological Society 2004

Figure 1. Depolarization elicits capacitance increases A, voltage command protocol V c indicates the step depolarization for 25 ms from −75 to −20 mV. A 6 mV sine wave is superimposed onto the holding potential but switched off during the depolarization (notice thicker line in baseline). The cell membrane capacitance C m increased from a baseline value of 14 pF by 400 fF in response to the depolarization (arrow). Depolarization elicited inward Ca2+ currents in the presence of Cs+ in the patch pipette (I, −58 pA baseline). Notice the transient reduction in the holding current after the depolarization. The membrane resistance Rm transiently increased after the depolarization. Series resistance Rs remained constant at 11 M. B, similar to A but in the presence of 1 mM Cs+ in the external solution to block inward rectifier potassium channels. Note the absence of a decrease in the holding current I and the absence of an increase in Rm after the depolarization. (1 s depolarization to −50 mV, 12.8 pF baseline capacitance, 397 fF capacitance increase).

444


This amplitude corresponds to the fusion of 282 vesicles at each of the ∼20 active zones (Roberts et al. 1990; Issa & Hudspeth, 1994; Zenisek et al. 2003) about an order of magnitude more than the 32 vesicles that are

Figure 2. Voltage-activated Ca2 + current in saccular hair cells A, representative traces of voltage-activated Ca2+ currents obtained for a range of test potentials from −120 to +20 mV from a holding potential of −85 mV. B, current–voltage relation of the voltageactivated Ca2+ currents. Voltages were corrected for series resistance and data values were interpolated to the voltages indicated using the cubic spline method. Vertical bars indicate S.E.M. (n = 9). The dashed line was calculated according to eqn (1): I = gmax (V − V rev )/1 + exp((V 1/2 − V /k)) where gmax = 0.011 nS, V rev = 39.2 mV, V 1/2 = −41.9 mV and k = 7.05 mV, taken as the averages from fits to I–V curves from individual cells (see Methods). C, conductance–voltage relationship. The average conductance data were fitted with a Boltzman distribution (eqn (2) g/gmax = 1/(1 + exp ((V 1/2 − V )/k)) (dashed line; V 1/2 = −42.0 mV, k = 7.5 mV).

J Physiol 560.2

docked (Lenzi et al. 1999). Even assuming dense hexagonal packing of docked presynaptic vesicles, the active zone could accommodate only 80 vesicles corresponding to 60 fF of potential capacitance increase, at least 3 times less than we observed. Therefore, the amplitude of the fast phase that we detect cannot be accounted for by the fusion of vesicles pre-docked at active zones. At the ribbon synapse in bipolar cell terminals it has been suggested that the number of synaptic vesicles that fuse with short latency depends on stimulus strength, that is, on the amplitude of the Ca2+ current (von Gersdorff & Matthews, 1997). Therefore, in addition to eliciting maximal Ca2+ currents by stepping to −20 mV, we examined capacitance changes resulting from weaker stimuli. Depolarization to −50 mV recruited only half of the maximal Ca2+ current (Fig. 2B), due to the opening of fewer Ca2+ channels (∼75% fewer on average than at −20 mV, Fig. 2C). A 20 ms depolarization to −50 mV

Figure 3. Activation of the Ca2 + current is fast A, the rising phase of representative current traces for steps to −20 and −50 mV are shown. These were fitted with a 3rd order exponential (eqn (3): I Ca (t) = I¯ Ca (1 − exp(−t/τ ))3 (dashed lines; τ = 476 µs at −50 mV and τ = 244 µs at −20 mV). To illustrate the faster rise at −20 mV, the peak value for ICa for the jump to −50 mV was scaled to the peak for −20 mV (dotted line). B, voltage dependence of τ . Voltages were corrected for junction potential and series resistance and linear interpolation was used to obtain values at the voltages indicated. These interpolated values were then averaged (n = 7–9). C The Physiological Society 2004

J Physiol 560.2


445

elicited an average capacitance increase of 220 ± 31 fF (n = 7; range 138–348 fF), not significantly different (P < 0.001) from the increase at −20 mV (4 ms: mean 10 ± 6 fF, n = 3, range 0–19 fF; 10 ms: mean 93 ± 21 fF, n = 3, range 58–132 fF). Similarly, the amplitude of the fast component for depolarization to −50 mV obtained from fitting (213 fF) was virtually identical to the amplitude of the fast component at −20 mV (209 fF) (Fig. 4B, dotted lines). Thus, either stimulus recruited about 9 times more vesicles for fast release than are identified morphologically as docked. In contrast to the correspondence in amplitudes, the rates of release were significantly different: the fast exponential component of release at −50 mV had a time constant of 5.5 ms, about twice the value at −20 mV (2.7 ms). To estimate the maximal release rates at −50 and −20 mV we differentiated the curves fitted to the change in C m as a function of duration of the stimulus (Fig. 5). The maximal release rate at −20 mV was 32 fF ms−1 , corresponding to a peak vesicle fusion rate of 43 vesicles ms−1 at each active zone. The values at −50 mV were 16 fF ms−1 and 22 vesicles ms−1 , respectively (Fig. 5). In summary, while the rate of exocytosis does depend on stimulus strength, the size of the limited, fast pool does not. This pool of fast-fusing vesicles greatly exceeds the reported number of vesicles docked at actives zones (Lenzi et al. 1999). Slow phase of exocytosis

Longer depolarizations to −20 mV of up to 5 s in duration trigger an additional phase of exocytosis that appeared to continue at a constant rate of ∼1.1 pF s−1 , corresponding to the fusion of about 1500 vesicles s−1 at each active zone (Fig. 4C, filled triangles; average increase at 1 s:

Figure 4. Two phases of exocytosis A, superimposed capacitance jumps in response to 4, 25 and 100 ms depolarizations (arrow). B, average capacitance increases ( C m ) are plotted against the time of depolarization to −20 mV () and to −50 mV (•). The data for depolarizations up to 500 ms were fitted (dashed lines) with the equation: C m = A (1 − exp (−t/τ ))n + st, where A, τ and s were dependent variables and n was fixed to be the same for depolarizations to either potential (n = 4.8, A−20 = 209 fF, τ −20 = 2.7 ms and s−20 = 1.1 fF ms−1 ; A−50 = 213 fF, τ −50 = 5.5 ms and s−50 = 0.14 fF ms−1 ). Dotted lines indicate fitted curve without the linear term. The times to reach 90% of maximal amplitude at −20 and −50 mV were 7.7 and 21.1 ms, respectively. C, linear slow component is long lasting. Same data set and fits as in B but shown on an expanded time scale. The fitted curves from B were extrapolated to 5 s.


Figure 5. Maximal fusion rates Rates of capacitance change for depolarizations to −20 mV (continuous line) and to −50 mV (dashed line). The maximal rate of secretion at −20 mV was 33 fF ms−1 and was reached 4.2 ms after the start of depolarization. The maximal rate of secretion at −50 mV was 16 fF ms−1 and was reached 8.6 ms after the start of depolarization.

446


J Physiol 560.2

1.7 ± 0.5 pF; n = 8; range 0.8–4.1 pF). Depolarizations to −50 mV also yielded a slow component of release (Fig. 4C; filled circles; average increase at 1 s: 326 ± 16 fF; n = 3; range 301–357 fF), proceeding at a slower rate of 0.14 pF s−1 , corresponding to the fusion of 190 vesicles s−1 at each active zone. Therefore, the rate of exocytosis from the slow pool also depends on stimulus strength.

amplitude of the response to the second stimulus was only 54% of the initial response (Fig. 6, filled triangle; 71% recovery for 200 ms ISI (data not shown)). This longer lasting depression is consistent with adaptation to a prolonged stimulus (Furukawa & Matsuura, 1978).

Paired pulse plasticity

Using cell membrane capacitance measurements from frog saccular hair cells we have tested the hypothesis that the number of morphologically docked vesicles limits the amplitude of the fast phase of neurotransmitter release. We find that stimuli to −50 mV and to −20 mV both elicit the fast fusion (defined here as fusion within 20 ms of stimulus onset) of 280 vesicles, at each active zone. Based on morphological studies only 32 vesicles are docked at the active zone at rest (Lenzi et al. 1999). Our data suggest that vesicles not docked at active zones are capable of fast fusion and that they can indeed constitute the majority of fast-fusing vesicles. Thus, at the frog saccular hair cell synapse, docking at the active zone is not a prerequisite for fast fusion. While the pool of fast-fusing vesicles is much larger than expected, it is depleted with stimuli of 20 ms and is refilled with a time constant of 29 ms. Finally, consistent with previous experiments (Parsons et al. 1994; Mennerick & Matthews, 1996; Holt et al. 2004), we find that continued depolarization yields a second, slower phase of exocytosis that draws on a pool of vesicles that is not depleted even after 5 s of stimulation. A previous study in frog saccular hair cells (Parsons et al. 1994) reported a slow phase of exocytosis but failed to detect a fast phase. While the nature of this discrepancy is unclear, a relatively large amplitude sine wave (40 mV peak to peak, V h : −70 mV: maximal positive excursion −50 mV, versus maximal positive excursion −63 mV in this study; see Methods; both studies report similar series resistance values) might have triggered the release of the fast pool before data acquisition was initiated. Such depression of the readily releasable pool of vesicles by prolonged depolarization has recently been reported for chick cochlear hair cells (Eisen et al. 2004), but other experimental differences might also contribute. In agreement with data from frog sacculus (Parsons et al. 1994), mouse (Moser & Beutner, 2000) and chick cochlear hair cells (Eisen et al. 2004) we did not observe a fast component of endocytosis following depolarization-induced exocytosis. While a fast component of endocytosis was observed in response to flash photolysis of caged calcium that raised the global intracellular free calcium concentration above 15 µm, no fast endocytosis was observed at lower free calcium concentrations (Beutner et al. 2001). This might indicate that calcium is elevated only locally in response to depolarization (see below) and that calcium does not rise above 15 µm at endocytic sites. However, since capacitance

As indicated in Fig. 4B, a 10 ms depolarization to −20 mV elicits the release of approximately 90% of the fast pool of vesicles. Under physiological conditions these cells respond to stimulation in the range of 20–200 Hz (Koyama et al. 1982; Hudspeth & Lewis, 1988; Lewis, 1988), corresponding to bouts of exocytosis every 5–50 ms. We therefore tested the efficacy of a second 10 ms stimulus delivered after a 10 ms interstimulus interval (ISI). We observed a marked depression of the second response relative to the first for a 10 ms ISI (64% depression; Fig. 6). Recovery from paired pulse depression was then investigated by increasing the ISI from 10 to 100 ms, and the data were fitted with a single exponential with a time constant of 29 ms. The observed recovery time course of the fast component of release is consistent with the functional requirement that hair cells produce bouts of exocytosis at frequencies to which they are maximally sensitive (Glowatzki & Fuchs, 2002). Interestingly, longer depolarizations of 100 ms elicited much more long-lasting depression. For example, with an ISI of 100 ms, the

Figure 6. Recovery from paired pulse depression Two 10 ms depolarizations were separated by varying interstimulus intervals (ISI; x-axis). The percentage of the second capacitance increase relative to the first is plotted; a value of 100% indicates no depression. The data were well fitted by a single exponential with a time constant of 29 ms (continuous line). Dashed lines indicate 90% expectation intervals. The triangle indicates the value for the mean paired pulse depression with two 100 ms depolarizations and an ISI of 100 ms.

Discussion


J Physiol 560.2


measurements detect net changes in surface area, we cannot exclude the existence of endocytosis during the depolarization. In such a case, our measurements would represent an underestimate of the number of vesicles fusing, especially during the slow phase but conceivably also during the fast phase of exocytosis. Fusion of vesicles outside of the active zone

An alternative hypothesis that could account for the near 10-fold discrepancy between the number of fast-fusing vesicles and vesicles docked at active zones is that ‘docked’ extrasynaptic vesicles are also fusing during the stimulus. Several observations suggest that this is not the case. First, voltage-gated Ca2+ channels, responsible for synaptic signalling, are almost exclusively located at the active zone (Roberts et al. 1990; Issa & Hudspeth, 1994; but see Rodriguez-Contreras & Yamoah, 2001) and Ca2+ entry hot spots have been detected and determined to colocalize with active zones (Issa & Hudspeth, 1994; Zenisek et al. 2003). Second, Ca2+ ‘spillover’ from the active zone to vesicles docked outside of the active zone is not likely to raise the Ca2+ concentration high enough to support vesicle fusion due to the strong endogenous calcium buffering capacity of hair cells (Roberts, 1993, 1994; Edmonds et al. 2000; Schneggenburger & Neher, 2000). Third, if the 5000 ‘excess fusing vesicles’ not docked at active zones (300 total fusing vesicles minus 32 docked at each of 20 active zones) were each tightly associated with at least one calcium channel with an amplitude of 0.5 pA (Rodriguez-Contreras & Yamoah, 2003) to support vesicle fusion, then the extrasynaptic calcium current alone would have to be considerably larger (> 2.5 nA) than the measured current (∼600 pA). Fourth, at −50 mV, the open probability of single Ca2+ channels is significantly smaller than at −20 mV (Rodriguez-Contreras & Yamoah, 2003). Indeed, only about 25% of the channels activated at −20 mV are activated at −50 mV (Fig. 2C). Therefore, if extrasynaptic vesicles and thus single channels were significantly contributing to exocytosis, then the amplitude of the fast phase at −50 mV should be smaller than at −20 mV, which is in direct contrast to our experimental observations (Fig. 3B). Fifth, extrasynaptic fusion events would have to outnumber synaptic events almost 10 : 1, drastically reducing or even eliminating the specificity of synaptic transmission. Sixth, the fast pool is refilled very rapidly (Fig. 6), which would be unlikely if it were composed primarily of outlying vesicles. In addition, direct observation of fusion events has indicated the existence of fusion hot-spots (Zenisek et al. 2000, 2003; Holt et al. 2004). Furthermore, active transport is too slow to allow for a significant number of outlying docked vesicles to be recruited into the active zone during the fast phase of exocytosis (see below). Such a recruitment could be responsible for the general depletion of vesicles seen after C The Physiological Society 2004

447

30 min of stimulation (Lenzi et al. 2002), much longer than the 20 ms considered here. It is worth noting that at the frog neuromuscular junction, fusion at the active zone was elicited with stimuli of short duration (‘0 min’) while extrasynaptic fusion events were only observed if the stimulation lasted for more than 1 min (Ceccarelli et al. 1988). Thus, while we can certainly not exclude that a few outlying vesicles contribute to our signal, it is exceedingly unlikely that fusion of vesicles occurs predominantly outside of active zones. The near 10-fold difference between the number of morphologically docked and fast-fusing vesicles can also not be explained by an undercount of docked vesicles. Assuming hexagonal tight packing, only 80 vesicles, approximately one third of the fast-fusing vesicles, could be docked in the area where free Ca2+ can reach concentrations high enough to support fusion (active zone with radius of ∼138 nm plus a rim of 40 nm (Roberts et al. 1990; Lenzi et al. 1999)). Taken together, our data indicate that at least three to nine times more vesicles than those docked at the active zone can fuse with fast kinetics, and points to the unconventional hypothesis that a non-docked source of vesicles contributes to the fast phase of release. Capacitance increases that are larger than expected have also been elicited in cochlear hair cells, but in response to flash photolysis of caged calcium (Beutner et al. 2001). However, this large pool could not be depleted by prior depolarization and the authors argue that fusion of vesicles localized outside of the active zone and not associated with calcium channels contribute a majority of the vesicles fusing in response to flash photolysis. Flash photolysis raises Ca2+ concentrations at active zones with a spatio-temporal profile that is distinct from that induced by depolarization. In addition, Ca2+ concentration is increased at sites away from Ca2+ channels and active zones, which may account for an abundance of extrasynaptic fusion (Beutner et al. 2001). It will therefore be interesting to test whether uncaging of calcium in frog saccular hair cells could release an even larger number of vesicles, namely outlying membrane-bound vesicles in addition to the large depolarization-dependent pool observed here. Fast-fusing vesicles and compound fusion

Morphological data demonstrate the existence of a small pool of docked vesicles at the active zone of hair cells. Our data are consistent with the idea that these vesicles make up an ‘ultrafast’ pool of vesicles (Mennerick & Matthews, 1996) that we did not resolve. Indeed, based on resampling statistics, more vesicles fused in less than 2 ms than the number docked, and more vesicles fused in less than 4 ms than the maximum number theoretically packed at the active zone. This emphasizes the major finding reported here, namely that a large excess of vesicles fuse during a

448


time interval too short to allow for vesicle movement or for diffusion of calcium to distant sites. What then are the anatomical correlates of these fast-fusing vesicles? Electron micrographs demonstrate the existence of a pool of about 380 vesicles that are bound to the synaptic body in frog saccular hair cells (Lenzi et al. 1999). Perhaps the synaptic body-associated vesicles contribute to the fast phase of release, with the synaptic body acting as a conveyor belt to supply vesicles to the release sites (von Gersdorff, 2001). However, molecular motor proteins such as the kinesin KIF3A that is associated with synaptic ribbons of photoreceptors (Muresan et al. 1999) have transport speeds of only 0.05–2 µm s−1 (Hirokawa, 1998). During the 20 ms of the fast phase, such a motor can transport vesicles at most a mere 40 nm, i.e. only about one vesicle diameter. Given a diameter of the synaptic body of 450 nm (Lenzi et al. 1999), those vesicles located on the synaptic body farthest from the active zone would have to move at rates of more than 22 µm s−1 or 10 times faster than the fastest known kinesin (Hirokawa, 1998). In fact, actual movement of synaptic vesicles has been estimated to be much slower (50 nm s−1 (Rizzoli & Betz, 2004); 0.5 µm s−1 (Klopfenstein & Vale, 2004); 0.75 µm s−1 (Rea et al. 2004)). For this reason we consider it unlikely that the ribbon actively delivers vesicles to the presynaptic membrane for fusion during the fast phase of exocytosis. The ribbon more probably serves as a conveyor belt during the slower sustained phase of release and/or to maintain the structural integrity of the synapse during massive membrane turnover (Bunt, 1971; Lenzi et al. 2002; Parsons & Sterling, 2003; Holt et al. 2004). Since there are not enough vesicles within 40 nm of the plasma membrane to account for the amplitude of the fast phase, we instead propose that vesicles further away from the membrane fuse with vesicles closer to or even docked to the presynaptic membrane, a process known as compound fusion (Palade, 1975). Such compound fusion was previously suggested as a mechanism for release at ribbon-type synapses (Heidelberger et al. 1994; Heidelberger, 1998; Parsons & Sterling, 2003) and is consistent with the large invaginations occasionally observed at release sites of frog saccular hair cells (Lenzi et al. 1999, 2002). Such invaginations are also compatible with a lack of visible ‘inflation’ of the hair cells even in response to massive surface membrane increase (Fig. 4C). Compound fusion can also account for the large spontaneous postsynaptic events recorded in single afferent fibres in the frog vestibular system (Rossi et al. 1977) and for the very large postsynaptic events in recordings from single boutons contacting cochlear hair cells whose frequency, but not amplitude, changed during depolarization of the hair cells (Glowatzki & Fuchs, 2002). Such large events could drive the postsynaptic membrane quickly to threshold in order to maintain precise timing information (Glowatzki & Fuchs, 2002)

J Physiol 560.2

and our experiments support the proposal (Parsons & Sterling, 2003) that they might be due to compound fusion. Given the tight spatio-temporal calcium buffering in hair cells (Roberts, 1993, 1994), what could trigger the fusion of more distant vesicles in a compound exocytosis scheme? We propose three entirely speculative possibilities. (1) While calcium cannot diffuse very far in unconstrained space (Roberts, 1994), the space at the active zone of hair cells is rather restricted by the presence of vesicles and the synaptic body. This might allow for a larger and faster calcium increase close to the synaptic body, while still restricting lateral diffusion of calcium. (2) The synaptic body could directly be involved in signal transduction and might serve as a ‘calcium wire’ to deliver the stimulus deep into the terminal (a calcium ion absorbed at one end could ‘push’ another calcium ion out at the other end). (3) The cisternae formed by the fusing vesicles might serve as an additional source for a fusion stimulus away from the plasma membrane. The fast-fusing pool of vesicles is depleted within 20 ms in response to both strong and weak stimuli (Fig. 4B). Consistent with a depletion of the fast pool, we also find strong paired pulse depression at this synapse (Fig. 6). Recovery from paired pulse depression is remarkably fast with a time constant of 29 ms. This is a much faster rate of recovery than has been reported for the ribbon-containing retinal bipolar cells (Mennerick & Matthews, 1996), and faster than was reported in cochlear hair cells (Moser & Beutner, 2000). Fast recovery might allow the synapse to release sufficient amounts of transmitter to trigger postsynaptic action potentials even in response to repetitive stimuli that mimic behaviourally relevant stimuli more closely than single step depolarizations (Koyama et al. 1982; Hudspeth & Lewis, 1988; Lewis, 1988). Our experiments suggest that vesicles not initially docked to the plasma membrane can fuse with fast kinetics. Thus, in contrast to current models, stable docking does not appear to be a strict requirement for fast neurotransmitter release at graded synapses. It remains to be elucidated whether vesicles not docked at active zones can fuse with fast kinetics at other types of synapses or whether the ribbon plays a unique role in the fusion of these vesicles. References Armstrong CE & Roberts WM (1998). Electrical properties of frog saccular hair cells: distortion by enzymatic dissociation. J Neurosci 18, 2962–2973. Augustine GJ, Charlton MP & Smith SJ (1985). Calcium entry and transmitter release at voltage-clamped nerve terminals of squid. J Physiol 367, 163–181. Beutner D, Voets T, Neher E & Moser T (2001). Calcium dependence of exocytosis and endocytosis at the cochlear inner hair cell afferent synapse. Neuron 29, 681–690. C The Physiological Society 2004

J Physiol 560.2


Bookman RJ, Lim NF, Schweizer FE & Nowycky M (1991). Single cell assays of excitation-secretion coupling. Ann N Y Acad Sci 635, 352–364. Bunt AH (1971). Enzymatic digestion of synaptic ribbons in amphibian retinal photoreceptors. Brain Res 25, 571–577. Catacuzzeno L, Fioretti B & Franciolini F (2003). Voltage-gated outward K currents in frog saccular hair cells. J Neurophysiol 90, 3688–3701. Ceccarelli B, Fesce R, Grohovaz F & Haimann C (1988). The effect of potassium on exocytosis of transmitter at the frog neuromuscular junction. J Physiol 401, 163–183. Couteaux R & Pecot-Dechavassine M (1974). Les zones specializees des membranes presynaptiques. (Specialized areas of presynaptic membranes.) C R Acad Sci Hebd Seances Acad Sci D 278, 291–293. Donnelly DF (1994). A novel method for rapid measurement of membrane resistance, capacitance, and access resistance. Biophys J 66, 873–877. Edmonds B, Reyes R, Schwaller B & Roberts WM (2000). Calretinin modifies presynaptic calcium signaling in frog saccular hair cells. Nat Neurosci 3, 786–790. Efron B & Tibshirani R (1991). Statistical data analysis in the computer age. Science 253, 390–395. Eisen MD, Spassova M & Parsons TD (2004). Large releasable pool of synaptic vesicles in chick cochlear hair cells. J Neurophysiol 91, 2422–2428. Furukawa T & Matsuura S (1978). Adaptive rundown of excitatory post-synaptic potentials at synapses between hair cells and eight nerve fibres in the goldfish. J Physiol 276, 193–209. Gentet LJ, Stuart GJ & Clements JD (2000). Direct measurement of specific membrane capacitance in neurons. Biophys J 79, 314–320. Gillis KD (1995). Techniques for membrane capacitance measurements. In Single-Channel Recording, ed. Sakmann B & Neher E, pp. 155–197. Plenum Press, New York. Gleisner L, Flock A & Wersall J (1973). The ultrastructure of the afferent synapse on hair cells in the frog labyrinth. Acta Otolaryngol 76, 199–207. Glowatzki E & Fuchs PA (2002). Transmitter release at the hair cell ribbon synapse. Nat Neurosci 5, 147–154. Hama K (1965). Some observations on the fine structure of the lateral line organ of the Japanese sea eel Lyncozymba Nystromi. J Cell Biol 24, 193–210. Heidelberger R (1998). Adenosine triphosphate and the late steps in calcium-dependent exocytosis at a ribbon synapse. J General Physiol 111, 225–241. Heidelberger R, Heinemann C, Neher E & Matthews G (1994). Calcium dependence of the rate of exocytosis in a synaptic terminal. Nature 371, 513–515. Heidelberger R, Zhou ZY & Matthews G (2002). Multiple components of membrane retrieval in synaptic terminals revealed by changes in hydrostatic pressure. J Neurophysiol 88, 2509–2517. Hirokawa N (1998). Kinesin and dynein superfamily proteins and the mechanism of organelle transport. Science 279, 519–526. Holt M, Cooke A, Neef A & Lagnado L (2004). High mobility of vesicles supports continuous exocytosis at a ribbon synapse. Curr Biol 14, 173–183. C The Physiological Society 2004

449

Holt JR & Eatock RA (1995). Inwardly rectifying currents of saccular hair cells from the leopard frog. J Neurophysiol 73, 1484–1502. Holt JC, Lioudyno M, Athas G, Garcia MM, Perin P & Guth PS (2001). The effect of proteolytic enzymes on the alpha9nicotinic receptor-mediated response in isolated frog vestibular hair cells. Hear Res 152, 25–42. Hudspeth AJ & Lewis RS (1988). Kinetic analysis of voltageand ion-dependent conductances in saccular hair cells of the bull-frog, Rana catesbeiana. J Physiol 400, 237–274. Issa NP & Hudspeth AJ (1994). Clustering of Ca2+ channels and Ca2+ -activated K+ channels at fluorescently labeled presynaptic active zones of hair cells. Proc Natl Acad Sci U S A 91, 7578–7582. Issa NP & Hudspeth AJ (1996). The entry and clearance of Ca2+ at individual presynaptic active zones of hair cells from the bullfrog’s sacculus. Proc Natl Acad Sci U S A 93, 9527–9532. Johnson SL, Thomas MV & Kros CJ (2002). Membrane capacitance measurement using patch clamp with integrated self-balancing lock-in amplifier. Pflugers Arch 443, 653–663. Katz B & Miledi R (1965). The effect of termperature on the synaptic delay at the neuromuscular junction. J Physiol 181, 656–670. Klopfenstein DR & Vale RD (2004). The lipid binding pleckstrin homology domain in UNC-104 kinesin is necessary for synaptic vesicle transport in C. elegans. Mol Biol Cell. Klyachko VA & Jackson MB (2002). Capacitance steps and fusion pores of small and large-dense-core vesicles in nerve terminals. Nature 418, 89–92. Koyama H, Lewis ER, Leverenz EL & Baird RA (1982). Acute seismic sensitivity in the bullfrog ear. Brain Res 250, 168–172. Lenzi D, Crum J, Ellisman MH & Roberts WM (2002). Depolarization redistributes synaptic membrane and creates a gradient of vesicles on the synaptic body at a ribbon synapse. Neuron 36, 649–659. Lenzi D, Runyeon JW, Crum J, Ellisman MH & Roberts WM (1999). Synaptic vesicle populations in saccular hair cells reconstructed by electron tomography. J Neurosci 19, 119–132. Lewis ER (1988). Tuning in the bullfrog ear. Biophys J 53, 441–447. Lewis RS & Hudspeth AJ (1983). Voltage- and ion-dependent conductances in solitary vertebrate hair cells. Nature 304, 538–541. Lioudyno MI, Verbitsky M, Holt JC, Elgoyhen AB & Guth PS (2000). Morphine inhibits an alpha9-acetylcholine nicotinic receptor-mediated response by a mechanism which does not involve opioid receptors. Hear Res 149, 167–177. Mennerick S & Matthews G (1996). Ultrafast exocytosis elicited by calcium current in synaptic terminals of retinal bipolar neurons. Neuron 17, 1241–1249. Moser T & Beutner D (2000). Kinetics of exocytosis and endocytosis at the cochlear inner hair cell afferent synapse of the mouse. Proc Natl Acad Sci U S A 97, 883–888. Muresan V, Lyass A & Schnapp BJ (1999). The kinesin motor KIF3A is a component of the presynaptic ribbon in vertebrate photoreceptors. J Neurosci 19, 1027–1037.

450


Neher E & Marty A (1982). Discrete changes of cell membrane capacitance observed under conditions of enhanced secretion in bovine adrenal chromaffin cells. Proc Natl Acad Sci U S A 79, 6712–6716. Neves G & Lagnado L (1999). The kinetics of exocytosis and endocytosis in the synaptic terminal of goldfish retinal bipolar cells. J Physiol 515, 181–202. Palade G (1975). Intracellular aspects of the process of protein synthesis. Science 189, 347–358. Palade GE & Palay SL (1954). Electron microscope observations of interneuronal and neuromuscular synapses. Anat Rec 118, 335–336. Parsons TD, Lenzi D, Almers W & Roberts WM (1994). Calcium-triggered exocytosis and endocytosis in an isolated presynaptic cell: capacitance measurements in saccular hair cells. Neuron 13, 875–883. Parsons TD & Sterling P (2003). Synaptic ribbon. Conveyor belt or safety belt? Neuron 37, 379–382. Rea R, Li J, Dharia A, Levitan ES, Sterling P & Kramer RH (2004). Streamlined synaptic vesicle cycle in cone photoreceptor terminals. Neuron 41, 755–766. Rizzoli SO & Betz WJ (2004). The structural organization of the readily releasable pool of synaptic vesicles. Science 303, 2037–2039. Roberts WM (1993). Spatial calcium buffering in saccular hair cells. Nature 363, 74–76. Roberts WM (1994). Localization of calcium signals by a mobile calcium buffer in frog saccular hair cells. J Neurosci 14, 3246–3262. Roberts WM, Jacobs RA & Hudspeth AJ (1990). Colocalization of ion channels involved in frequency selectivity and synaptic transmission at presynaptic active zones of hair cells. J Neurosci 10, 3664–3684. Rodriguez-Contreras A & Yamoah EN (2001). Direct measurement of single-channel Ca2+ currents in bullfrog hair cells reveals two distinct channel subtypes. J Physiol 534, 669–689. Rodriguez-Contreras A & Yamoah EN (2003). Effects of permeant ion concentrations on the gating of L-type Ca2+ channels in hair cells. Biophys J 84, 3457–3469. Rohlicek V & Schmid A (1994). Dual-frequency method for synchronous measurement of cell capacitance, membrane conductance and access resistance on single cells. Pflugers Arch 428, 30–38. Rossi ML, Valli P & Casella C (1977). Post-synaptic potentials recorded from afferent nerve fibres of the posterior semicircular canal in the frog. Brain Res 135, 67–75. Roth A & Hausser M (2001). Compartmental models of rat cerebellar Purkinje cells based on simultaneous somatic and dendritic patch-clamp recordings. J Physiol 535, 445–472. Sabatini BL & Regehr WG (1996). Timing of neurotransmission at fast synapses in the mammalian brain. Nature 384, 170–172. Santos-Sacchi J, Kakehata S & Takahashi S (1998). Effects of membrane potential on the voltage dependence of motilityrelated charge in outer hair cells of the guinea-pig. J Physiol 510, 225–235. Santos-Sacchi J (2004). Determination of cell capacitance using the exact empirical solution of dy/dCm and its phase angle. Biophys J 87, 714–727.

J Physiol 560.2

Schikorski T & Stevens CF (2001). Morphological correlates of functionally defined synaptic vesicle populations. Nat Neurosci 4, 391–395. Schneggenburger R & Neher E (2000). Intracellular calcium dependence of transmitter release rates at a fast central synapse. Nature 406, 889–893. Sjostrand FS (1958). Ultrastructure of retinal rod synapses of the guinea pig eye as revealed by three-dimensional reconstructions from serial sections. J Ultrastruct Res 2, 122–170. Smith CA & Sjostrand FS (1961). Structure of the nerve endings on the external hair cells of the guinea pig cochlea as studied by serial sections. J Ultrastruct Res 5, 523–556. Smotherman MS & Narins PM (1998). Effect of temperature on electrical resonance in leopard frog saccular hair cells. J Neurophysiol 79, 312–321. Sobkowicz HM, Rose JE, Scott GE & Slapnick SM (1982). Ribbon synapses in the developing intact and cultured organ of Corti in the mouse. J Neurosci 2, 942–957. Solsona C, Innocenti B & Fernandez JM (1998). Regulation of exocytotic fusion by cell inflation. Biophys J 74, 1061–1073. Thoreson WB, Rabl K, Townes-Anderson E & Heidelberger R (2004). A highly Ca2+ -sensitive pool of vesicles contributes to linearity at the rod photoreceptor ribbon synapse. Neuron 42, 595–605. Tucker T & Fettiplace R (1995). Confocal imaging of calcium microdomains and calcium extrusion in turtle hair cells. Neuron 15, 1323–1335. von Gersdorff H (2001). Synaptic ribbons: versatile signal transducers. Neuron 29, 7–10. von Gersdorff H & Matthews G (1997). Depletion and replenishment of vesicle pools at a ribbon-type synaptic terminal. J Neurosci 17, 1919–1927. von Gersdorff H, Vardi E, Matthews G & Sterling P (1996). Evidence that vesicles on the synaptic ribbon of retinal bipolar neurons can be rapidly released. Neuron 16, 1221–1227. Wachtel A & Szamier R (1966). Special cutaneous receptor organs of fish: the tuberous organs of Eigenmannia. J Morph 119, 51–80. Zenisek D, Davila V, Wan L & Almers W (2003). Imaging calcium entry sites and ribbon structures in two presynaptic cells. J Neurosci 23, 2538–2548. Zenisek D, Steyer JA & Almers W (2000). Transport, capture and exocytosis of single synaptic vesicles at active zones. Nature 406, 849–854. Acknowledgements We thank all members of the Schweizer laboratory for feedback and discussions. We thank Drs White, Otis, Glykys, Grinnell and Narins for discussion and critical reading of the manuscript. We also thank two anonymous reviewers for extensive and constructive comments. This work was supported by a grant from the NINDS to F.E.S. (NS41317). Author’s present address B. W. Edmonds: Department of Molecular Biology and Biochemistry, University of California at Irvine, 540 Steinhaus Hall, Irvine, CA 92697, USA. C The Physiological Society 2004