| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Physiology (2002), 541.3, pp. 811-823
© Copyright 2002 The Physiological Society
DOI: 10.1113/jphysiol.2001.013485
 |
ABSTRACT |
|---|
 TopAbstract IntroductionMethodsResultsDiscussionReferences |
|---|
Hippocampal synapses display a range of release probabilities. This is partially the result of scaling of release probability with the total number of releasable vesicles at each synapse. We have compared synaptic release and vesicle pool sizes across a large number of hippocampal synapses using FM 1-43 and confocal fluorescence microscopy. We found that the relationship between the number of recycling vesicles at a synapse and its release probability is dependent on firing frequency. During firing at 10 Hz, the release probability of each synapse is closely related to the number of recycling vesicles that it contains. In contrast, during firing at 1 Hz, different synapses turn over their recycling vesicle pools at different rates leading to an indirect relationship between recycling vesicle pool size and release probability. Hence two synapses may release vesicles at markedly different rates during low frequency firing, even if they contain similar numbers of vesicles. Both further kinetic analyses and manipulation of the number of vesicles in the readily releasable pool using phorbol ester treatment suggested that this imprecise scaling observed during firing at 1 Hz resulted from synapse-to-synapse differences in the proportion of recycling vesicles partitioned into the readily releasable pool. Hence differential partitioning between vesicle pools affects presynaptic weighting in a frequency-dependent manner. Since hippocampal single unit firing rates shift between 1 Hz and 10 Hz regimes with behavioural state, differential partitioning may be a mechanism for encoding information in hippocampal circuits.
(Received 4 November 2001; accepted after revision 22 March 2002)
Corresponding author J. Waters: Abteilung Zellphysiologie, Max-Planck-Institut für medizinische Forschung, Jahnstrasse 29, 69120 Heidelberg, Germany. Email: jwaters{at}mpimf-heidelberg.mpg.de
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Hippocampal synapses display a wide range of release probabilities, release at many synapses being notably unreliable (Hessler et al. 1993; Rosenmund et al. 1993; Goda & Südhof, 1997). Precisely which parameters determine the efficacy of release (the average number of vesicles released per action potential) at each synapse is unclear, but the number of vesicles in the readily releasable pool may be a critical determinant. Furthermore, it has been suggested that synaptic parameters scale together, a larger active zone accompanying larger numbers of vesicles in recycling and readily releasable pools and resulting in a relatively high release probability (Dobrunz & Stevens, 1997; Murthy et al. 1997; Schikorski & Stevens, 2001). This 'strict scaling' model therefore predicts a close relationship between the number of presynaptic vesicles and synaptic efficacy.
Most functional studies of hippocampal synaptic properties have examined release in response to single stimuli or low-frequency trains (0.5 Hz). Although single unit recordings in vivo indicate that CA1 pyramidal cells fire at approximately 1 Hz at rest, during activity (such as running in a wheel) the firing rate increases to 5-20 Hz and may be sustained throughout the duration of the activity (Wiener et al. 1989; Czurkó et al. 1999; Hirase et al. 1999).
Using the FM series of fluorescent styryl dyes (Ryan et al. 1993; Betz et al. 1996; Murthy, 1999), we have examined the release properties of individually identified hippocampal synapses during stimulation at 1 Hz and at 10 Hz, corresponding to firing frequencies at rest and during sustained activity, respectively. Across a large population of synapses, we observed that different synapses release different proportions of their recycling vesicle pools per stimulus during firing at 1 Hz. In contrast, all synapses released similar proportions during firing at 10 Hz. These data indicate that, contrary to the strict scaling model, efficacy scales only loosely with the number of recycling vesicles at each synapse during firing at low frequencies, but that the relationship becomes more direct during higher-frequency stimulation. Furthermore, the degree to which phorbol ester potentiated release from a synapse was dependent on the proportion of the recycling vesicle pool released per stimulus before phorbol application, suggesting that this coupling might be altered in a synapse-specific manner by the activation of second messenger pathways. One explanation of our data is that different synapses partition different proportions of their recycling vesicles into the readily releasable pool and that this parameter therefore influences release probability in a synapse-specific manner.
This frequency dependence of the relationship between recycling vesicle pool size and release probability may shape the weighting of synapses across a neuronal population in a firing frequency-dependent manner. This effect may therefore be an important mechanism, influencing the output of populations of neurones in a frequency-dependent manner.
|
METHODS |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Preparation of dissociated cultures
Dissociated hippocampal cultures were prepared from post-natal day 2 Sprague-Dawley rats as previously described (Waters & Smith, 2000), in accordance with guidelines laid down by the Stanford University Administrative Panel on Laboratory Animal Care. Pups were killed by decapitation. Hippocampi were dissected and the dentate gyrus removed. After treating for 15 min at room temperature in 10 mg ml-1 trypsin, the tissue was dissociated by trituration through the tip of a fire-polished siliconized glass Pasteur pipette. Dissociated cells were collected by centrifugation at 800 g at 4 °C and plated onto Matrigel-coated coverslips in Neurobasal medium (Life Technologies, Gaithersburg, MD, USA) supplemented with B-27 (Life Technologies), 28 mM glucose, 1.3 µM transferrin (Calbiochem, La Jolla, CA, USA), 2 mM glutamine, 0.7 units ml-1 insulin (Sigma, St Louis, MO, USA) and 1 % fetal calf serum (Hyclone, Logan, UT, USA). Cells were maintained at 37 °C in an atmosphere containing 5 % CO2 until use after 10-16 days in vitro.
FM 1-43 staining and destaining
A coverslip was mounted in a custom-made, low-volume (60 µl) laminar perfusion chamber on the stage of an inverted microscope (Zeiss IM 35). This permitted continuous perfusion at approximately 1 ml min-1 while imaging through the coverslip to which the cells adhered. Images were acquired using either transmitted light and Nomarski optics or an epifluorescence configuration. Cells were perfused with a modified Tyrode solution consisting of (mM): 119 NaCl, 5 KCl, 2 CaCl2, 2 MgCl2, 25 Hepes and 30 glucose and with 10 µM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), 50 µM D-aminophosphonovalerate (APV) and 3 µM bicuculline added to reduce spontaneous activity and prevent recurrent excitation during stimulation. All imaging experiments were performed at room temperature (22-23 °C).
Stock solutions of FM 1-43 and FM 2-10 in water were stored at 4 °C. FM dyes were diluted into Tyrode solution to a final dye concentration of 15 µM and 200 µM, respectively. Following addition of FM dye to the perfusing solution, exocytosis was induced using a 10 Hz train of 100 stimuli delivered through platinum electrodes positioned on opposite sides of the perfusion chamber (stimulus duration 1 ms, field strength 50 V cm-1). Note that a longer staining protocol was used for the data presented in Fig. 2 (see Results). The briefer 100-stimulus protocol was used for experiments in which two rounds of FM staining-destaining were to be performed, since this protocol minimized dye exposure and, therefore, non-specific staining. Control experiments using the intracellular calcium dye fluo 4-AM to detect action potential-induced calcium transients verified that stimulation invariably succeeded in firing neurones at the required frequency.
Following staining, the preparation was washed in dye-free medium for 10 min to reduce non-specific staining prior to image acquisition. Subsequent stimulation (at either 1 Hz or 10 Hz) resulted in destaining of the preparation. Destaining curves were derived from images acquired during this stimulus train. Destaining images were acquired at 3 s intervals during 10 Hz stimulation and at 10 s intervals during 1 Hz stimulation.
All data were adjusted for non-specific staining by subtracting the mean fluorescence intensity of each punctum taken from 10 images collected following 900 or 1200 destaining stimuli. This was sufficient to release all available vesicles, since further stimulation released no additional dye. Although a small decrease in fluorescence was often observed following 1200 stimuli (e.g. Fig. 5A) this was attributable to slight photobleaching of non-specific staining since it also occurred in the absence of stimulation. It seems unlikely that this gradual photobleaching of residual fluorescence could have a strong quantitative influence on our data since residual staining after 1200 stimuli represented only approximately 10 % of the fluorescence intensity in the stained condition.
Where data were normalized to the stained condition (see for example Fig. 2B), data were normalized to the mean fluorescence intensity from 10 images acquired prior to the destaining stimulus train. Where a second round of staining-destaining was performed, a 15 min rest period was inserted between the end of the first destaining stimulus train and the start of the second staining stimulus.
Stock solutions of phorbol 12,13-dibutyrate (PDBu) were made at 1 mM in DMSO and stored at -20 °C until use. PDBu was added to the perfusing Tyrode solution to a final concentration of 1 µM. The resulting concentration of DMSO (0.1 %, v/v) did not influence FM 1-43 staining or destaining in control experiments. Where used, PDBu was applied for 5 min, beginning 5 min after the end of the staining stimulus train and 5 min prior to the start of the destaining stimulus train.
Imaging techniques
The sample was illuminated using the 488 nm line of an air-cooled argon ion laser at the minimal intensity commensurate with acceptable signal-to-noise of the fluorescent dye (60 µW at the back aperture of the objective). Laser light was focused onto the preparation using an oil immersion objective lens (
40, 1.3 NA Dapo UV, Olympus, Tokyo, Japan) and emitted fluorescence was collected through an OG 520 nm longpass filter. A Bio-Rad MRC 500 confocal laser scanning microscope running dedicated software with custom modifications was used to acquire images. Images were stored digitally for off-line analysis using custom software (View, Dr Noam Ziv, Rappaport Institute, Haifa, Israel) or a commercial equivalent (Metamorph, Universal Imaging, West Chester, PA, USA). Fluorescence intensity measurements were taken from regions of interest of approximately 1.5 µm 1.5 µm, corresponding to individual puncta, each visibly separate from its nearest neighbours. Regions of staining larger than 1.5 µm in diameter were excluded from analysis. All fluorescence intensities were corrected for non-specific staining by subtracting the intensities measured following complete destaining of the preparation.
Statistics were performed using commercial software (SigmaStat for Windows 1.0, Jandel Corporation, San Rafael, CA, USA). Though data were frequently displayed as means ± S.E.M. for presentation purposes, non-parametric statistics were employed throughout (see Results). Hence, Wilcoxon's signed rank test was used with paired data and the Mann-Whitney rank sum test with unpaired data.
|
RESULTS |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The FM series of fluorescent dyes has proved extremely useful as tools for the study of presynaptic vesicle recycling properties in hippocampal neurones (Ryan et al. 1993; Betz et al. 1996; Murthy, 1999). We have used FM 1-43 and FM 2-10 and confocal fluorescence microscopy to monitor vesicle release from hippocampal neurones in dissociated cultures, such as that shown in Fig. 1A.
 |
View larger version [in this window] [in a new window] |
|
|
Figure 1. Example of sequential FM 1-43 staining and destaining A, Nomarski image of a mixed neuronal-glial culture after 13 days in vitro. B, the same field after staining with FM 2-10. The staining protocol consisted of 100 stimuli at 10 Hz then removal of extracellular dye 30 s after cessation of the stimulus train. C, fluorescence image following destaining with a 900 stimulus train at 10 Hz. Remaining fluorescence represents non-specific staining, all vesicular staining having been released. D, composite image comparing staining during subsequent trials. Data were corrected for non-specific staining. Red represents the first round of staining (image C subtracted from image B) and green the second round (images E - F), yellow denotes regions where red and green overlap. Note that almost all puncta appear in the same position in the two images indicating that these synapses are positionally stable. E and F, fluorescence images from a second round of staining and destaining using an identical protocol to that for images B and C. Comparison of E with C reveals an increase in non-specific staining. Note also that fluorescence intensities are similar following 1st and 2nd rounds of staining. Scale bar represents 10 µm. | ||
After staining with FM 1-43 or FM 2-10 (see Methods) clusters of presynaptic vesicles were visible as a punctate fluorescence staining pattern (Fig. 1B). The association of FM dyes with vesicular membranes is reversible and dye was released upon subsequent action potential firing (destaining; Fig. 1C). This procedure could be repeated, which permitted sequential measurement of presynaptic properties at the same identified locations since the position of most fluorescent puncta changed little between trials (Fig. 1D). Images from the second round of staining- destaining are included in Fig. 1 (E and F) for comparison with data from the first round (B and C).
Comparison of destaining at 10 Hz and 1 Hz
We used the activity-dependent loss of FM 1-43 staining to monitor rates of vesicle release in response to stimulation. We began by staining the preparation using a prolonged train of 900 stimuli at 10 Hz, leaving FM 1-43 in the perfusing solution for an additional minute after termination of the stimulus train. This staining protocol stains the entire recycling vesicle pool (Ryan & Smith, 1995; Ryan et al. 1996). The fluorescence intensities of individual puncta were then monitored through time by acquiring successive images before and during the destaining stimulus train. Fluorescence intensities recorded from 40 puncta in a single field of view are shown in Fig. 2A. The fluorescence of each punctum was stable prior to delivery of the 10 Hz destaining stimulus train. During stimulation, each punctum destained with an approximately exponential time course (not shown). Destaining was complete after 900 stimuli. (Note that all data have been corrected for non-specific staining by subtracting the fluorescence intensity remaining after 1200 stimuli.)
 |
View larger version [in this window] [in a new window] |
|
|
Figure 2. Pronounced kinetic heterogeneity at 1 Hz, but not 10 Hz A, fluorescence intensities of 40 individual fluorescent puncta followed through time. Traces begin in the stained condition. Staining consisted of 900 stimuli at 10 Hz, followed by an additional minute in FM 1-43 then a 10 min wash period. During stimulation at 10 Hz, each punctum loses fluorescence through time. Note that all puncta are fully destained by 900 stimuli at 10 Hz. Data were corrected for non-specific staining by subtracting the fluorescence intensity after 1200 stimuli. B, the same data following normalization to the fully stained condition. C, similar data derived using the same loading protocol, but using a 1 Hz destaining stimulus train. Note that some synapses destained only very slowly, if at all, necessitating the use of a 10 Hz stimulus train (arrow) to complete destaining. Data from the fully destained condition are visible on the right. | ||
Release efficacy (measured as loss of fluorescence per stimulus) was greater for synapses with larger total vesicle pool sizes (greater initial fluorescence). This is consistent with published data (Murthy et al. 1997). To examine whether or not there were substantial differences in the time constants of release between synapses, we normalized each trace to its initial fluoresence. Figure 2B illustrates the resulting 'destaining curves' for the 40 fluorescent puncta represented in Fig. 2A. Only modest differences in the fraction of initial fluorescence lost per stimulus (fractional destaining rate) were observed between synaptic puncta across numerous preparations.
In contrast, in response to a 1 Hz stimulus train, fractional destaining rates displayed marked heterogeneity across the synaptic population (Fig. 2C). Some synapses destained so slowly in response to a 1 Hz train that a subsequent 10 Hz train was necessary to complete destaining of the preparation (Fig. 2C). Similar heterogeneity of fractional destaining rates was observed in numerous preparations in response to 1 Hz stimulation.
One possible explanation for the heterogeneity observed at 1 Hz might be that failure of action potential initiation and/or propagation influences release from some synapses more than others, perhaps due to the relative position of different synapses with regard to axonal branching patterns. To address this possibility we monitored destaining at 10 Hz after reducing the extracellular calcium concentration to 0.1 mM. This should decrease action potential threshold and favour propagation, but decrease presynaptic calcium influx and therefore presynaptic efficacy. Pronounced heterogeneity was observed under these conditions (data not shown) indicating that heterogeneity is related to synaptic efficacy, not action potential failure.
We have examined this frequency-dependent heterogeneity in further detail, repeating rounds of FM dye staining and destaining to make two measurements of activity at each synapse. A comparison of destaining curves at 1 Hz and 10 Hz is presented in the form of a frequency histogram in Fig. 3A. As anticipated from Figs 2B and C, a greater range was observed at 1 Hz than at 10 Hz. The destaining curves of rapidly destaining puncta followed an exponential time course (not shown). However, it was not possible to reliably fit the decay time course of those puncta which displayed minimal destaining during the 1 Hz train. We have therefore expressed fractional release rates as a percentage of fluorescence released (percentage destaining) after 150 destaining stimuli.
 |
View larger version [in this window] [in a new window] |
|
|
Figure 3. Properties of 1 Hz heterogeneity A, frequency histogram comparing the extent of destaining following 150 stimuli at 1 Hz and 10 Hz. Frequency is displayed as the percentage, rather than absolute number, of puncta in each bin to allow direct comparison of 1 Hz and 10 Hz data, despite different numbers of observations. Data represent 616 puncta at 1 Hz and 375 puncta at 10 Hz. B, scatter plot illustrating the 1 Hz fractional destaining rates of 333 puncta during two subsequent trials. Each point represents the destaining percentages of a single fluorescent punctum after 150 stimuli at 1 Hz in each trial. The data were best fitted with a regression line of slope 1.009 passing through the origin. C, scatter plot comparing 1 Hz fractional destaining rates with initial fluorescence intensity in the stained condition. The staining protocol consisted of 100 stimuli at 10 Hz. Fractional destaining rates are represented as percentage destaining after 150 stimuli at 1 Hz. Small points represent the values of 333 individual puncta. Large circular symbols represent the mean (± S.E.M.) values for the same data binned according to their rates of destaining (bins each represent 10 percentage units). Inset, scatter plot comparing 1 Hz fractional destaining rates with fluorescence intensity in the stained condition for a single experiment (108 puncta). D, plots showing the absolute amount of fluorescence released in response to 150 stimuli at 1 Hz as a function of initial fluorescence staining intensity. Each of the 333 data points represents data from one synapse. | ||
If the fractional rates observed at 1 Hz reflect differences between synapses, rather than fluctuations in release or measurement artifacts, one would expect the rates at each synapse to be reproducible. Comparison of the fractional destaining rates at 1 Hz for each synaptic punctum during sequential trials indicated that fractional destaining rates at 1 Hz were indeed reproducible (Fig. 3B) and were therefore characteristic of each synapse. In addition, no correlation was observed between fractional rates at 1 Hz and total recycling vesicle pool size (Fig. 3C; n = 333 puncta).
Since small differences in measurement error between experiments may obscure subtle correlations, we also examined the relationship between fractional destaining rate and total pool size in each individual experiment. Data from one experiment are displayed in the inset in Fig. 3C. In no single experiment was any relationship between total pool size and fractional destaining rate at 1 Hz evident. To address the specific possibility that synapses with a small total pool size displayed slower fractional release rates at 1 Hz than the population mean (see Discussion), we compared the mean initial fluorescence of the synapses with slowest fractional destaining rates with that of the population of synapses as a whole. For the experiment displayed in the inset to Fig. 3C, the 10 % of synapses with the slowest fractional destaining rates had a mean (±S.E.M.) initial fluorescence of 29.2 ± 3.0 arbitrary fluorescence units. By comparison, the whole population had a mean (± S.E.M.) initial fluorescence of 32.5 ± 1.3 arbitrary fluorescence units. The total pool size of the slowest destaining synapses was not significantly different from that of the whole population (P = 0.55, Mann-Whitney rank sum test).
The influence that these differences in fractional destaining rates have on synaptic efficacy is illustrated in Fig. 3D, which shows the number of vesicles (amount of fluorescence) released in response to 150 stimuli at 1 Hz as a function of recycling pool size (initial staining intensity). There is a clear trend for synapses with larger recycling pool sizes to release more vesicles than those with smaller recycling pool sizes, consistent with published observations (Dobrunz & Stevens, 1997; Murthy et al. 1997). However, recycling pool size is not the only determinant since different synapses may release different amounts of dye despite exhibiting similar initial staining intensities.
These observations indicate that any two synapses may release vesicles at markedly different rates during low-frequency firing, even if they contain similar numbers of vesicles. Release rates are reproducible and so are characteristic of any given synapse. In contrast, heterogeneity across the synaptic population is much less pronounced at 10 Hz.
Heterogeneity at 10 Hz during the initial phase of release
Although pronounced heterogeneity was not observed at 10 Hz, small differences in the fractional rates of destaining between synapses were observed at this frequency (see Fig. 2B). To determine whether these small differences in fractional destaining rates at 10 Hz were related to the more substantial differences observed at 1 Hz, we performed two rounds of staining-destaining once at each frequency (1 Hz and 10 Hz). This permitted a direct comparison of destaining in response to these two frequencies at the same, identified synapses. To maximize the temporal resolution of our measurements, we used FM 2-10 rather than FM 1-43 (Ryan et al. 1996). To compare fractional rates at these two frequencies we ranked 214 puncta according to their fractional rates at 1 Hz, then compared the mean fractional rates at 10 Hz of the 50 fastest and 50 slowest puncta. The data indicate that puncta that destained slowly and rapidly at 1 Hz also destained at different fractional rates at 10 Hz (Fig. 4A).
 |
View larger version [in this window] [in a new window] |
|
|
Figure 4. Direct comparison of 1 Hz and 10 Hz destaining curves A, comparison of the fractional destaining rates at 10 Hz of two groups of puncta stained with FM 2-10: those with rapid and those with slow fractional rates at 1 Hz (filled and open circles, respectively). Data points represent the mean (± S.E.M.) of 50 puncta (of a total population of 214). Inset, mean destaining curves at 1 Hz for the same two groups. The asterisk denotes the data point after 150 stimuli (see below). B, the same 10 Hz curves (presented in A) normalized to the fluorescence intensity after 150 destaining stimuli. Note that the curves converge after the first few images. To analyse this phenomenon in more detail the percentage destaining occurring between each image pair was calculated for each punctum. For instance, the percentage destaining occurring between images labelled A and B was calculated using the following formula: % destaining = 100 | ||
Further examination of the data revealed that the difference in fractional destaining rates between the two curves was related to the initial phase of destaining. Figure 4B shows the two 10 Hz destaining curves from Fig. 4A presented in a different manner: they have each been normalized to their respective fluorescence values attained after 150 destaining stimuli. Clearly, the difference between these two curves is limited to the initial phase of destaining. This indicates that fractional destaining rates at 1 Hz are related to the fractional rates during the initial (but not later) phase at 10 Hz. To determine the precise duration for which 1 Hz and 10 Hz rates correlate, we manipulated the data in another fashion, calculating the percentage destaining occurring between each sequential pair of images (see legend to Fig. 4B). The resulting data are presented in Fig. 4C. A significant difference was observed between the fractional rates of destaining after 30 and after 60 stimuli at 10 Hz, but not thereafter (P < 0.01, Mann-Whitney rank sum test). These data therefore indicate that fractional rates at 1 Hz are correlated with fractional destaining rates during the initial, but not subsequent, phase of destaining at 10 Hz.
Since no correlation was observed between fractional rates at 1 Hz and total pool size (see Fig. 3C) these data relating fractional rates at 1 Hz and at 10 Hz lead to a further prediction: that the initial rate of destaining at 10 Hz should not correlate with total pool size. The data presented in Fig. 4D demonstrate that this is indeed the case.
Effects of phorbol ester
It has been suggested that initial release probability during 10 Hz stimulation is related to the number of vesicles contained within a readily releasable pool (Dobrunz & Stevens, 1997). Furthermore, it has been suggested that phorbol esters may promote release by increasing the number of vesicles in the readily releasable pool (Stevens & Sullivan, 1998; Waters & Smith, 2000). We have therefore examined the effects of phorbol ester on the heterogeneity of destaining rates at both 1 Hz and 10 Hz.
To examine the effects of phorbol ester, one round of staining-destaining was conducted then vesicles were stained a second time before phorbol application. Phorbol-12,13-dibutyrate (PDBu; 1 µM) was applied for 5 min immediately prior to the second round of destaining (Waters & Smith, 2000). Phorbol treatment resulted in an increase in fractional release rate at 10 Hz (Fig. 5A). No change occurred in controls, which were not treated with PDBu (data not shown). Manipulating the data as before (see Fig. 4C), it was evident that the effect of PDBu was limited to the first two data points on the destaining curve (Fig. 5B).
 |
View larger version [in this window] [in a new window] |
|
|
Figure 5. Effect of PDBu at 10 Hz A, 10 Hz destaining kinetics measured with FM 1-43 before and after PDBu treatment. Data points represent mean (± S.E.M.) measurements from 141 puncta. B, same data represented as percentage destaining between each image, as in Fig. 4C. The effect of PDBu was significant after 30 and 60 destaining stimuli (P < 0.01, Wilcoxon's signed rank test) where destaining was increased by 110 % and 25 % respectively. C, 10 Hz destaining kinetics measured with FM 2-10 before and after PDBu treatment. Data points represent mean (± S.E.M.) measurements from 138 puncta. D, percentage destaining between each image for destaining curves derived using FM 2-10. The effect of PDBu was significant after 30 and 60 destaining stimuli (P < 0.01, Wilcoxon's signed rank test). The degree of potentiation was also similar to that with FM 1-43, being 115 % and 50 % after 30 and 60 stimuli, respectively. | ||
Since this effect could potentially be influenced by the slow off-rate of FM 1-43, we repeated these experiments, using FM 2-10 (Fig. 5C and D). Quantitatively similar potentiations were observed with both dyes, indicating that the slower off-rate of FM 1-43 does not substantially influence our measurements. Since fractional release rates at 1 Hz are related to initial rates at 10 Hz, one might also expect an effect of phorbol esters at 1 Hz. Figure 6A illustrates such an effect on FM 1-43 destaining. No change occurred in controls, which were not treated with PDBu (data not shown).
 |
View larger version [in this window] [in a new window] |
|
|
Figure 6. Effect of PDBu at 1 Hz A, 1 Hz destaining curves generated with FM 1-43 before and after PDBu treatment. Data points represent mean (± S.E.M.) measurements from 78 puncta. B, same data represented as percentage destaining between each image, as in Fig. 4C. The effect of PDBu was significant after 10, 20, 30, 40, 50 and 60 destaining stimuli (** P < 0.01, * P < 0.05, Wilcoxon's signed rank test) where destaining was increased by 78 %, 46 %, 54 %, 74 %, 78 % and 74 %, respectively. | ||
These data indicate that phorbol ester treatment changes mean synaptic release properties. It was possible that phorbol ester alters release properties in a synapse-specific manner, thereby influencing heterogeneity. We therefore sought to determine whether or not the initial release rate of a synapse could predict its sensitivity to PDBu. The effect of PDBu on release rates at 10 Hz was most pronounced at the data point taken after 30 stimuli (see Fig. 5B). We therefore compared release induced by a 30-stimulus train at 10 Hz before and after PDBu treatment. Fig. 7A shows these data presented to illustrate the relationship between initial fractional release rate and the effect of PDBu. Only those synapses which released less than approximately 20 % of their recycling vesicles were sensitive to PDBu. After PDBu treatment the percentage of vesicles released was similar across the population of synapses.
 |
View larger version [in this window] [in a new window] |
|
|
Figure 7. Effect of PDBu depends on initial synaptic properties A, plot to show the relationship between the effect of PDBu and the proportion of recycling vesicles in the readily releasable pool before PDBu treatment. Release was induced by 30 stimuli at 10 Hz. Data from 774 puncta were binned according to the percentage released before PDBu treatment (2-unit bins) and are plotted as means (± S.E.M.). B, same data presented to compare release before and after PDBu treatment. The line indicates the relationship expected if PDBu treatment were excluded. C and D, similar data to those presented in A and B, but using a 1 Hz stimulus (30 stimuli). Data represent a total of 417 puncta. | ||
At 1 Hz, synapses displayed a similar differential sensitivity to PDBu. Synapses that released more than approximately 20-30 % of their recycling vesicles in response to 30 stimuli at 1 Hz were insensitive to PDBu (Fig. 7C). As at 10 Hz, PDBu partially compensated for the lower fractional release rates of some synapses at 1 Hz (Fig. 7D). Hence PDBu treatment has an 'equalizing' effect on release at both 1 Hz and 10 Hz; after PDBu treatment, all synapses release an equal proportion of their vesicle pools in response to a brief stimulus train.
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
We have examined the relationship between stimulus frequency and fractional release rates across a large number of hippocampal synapses. Striking heterogeneity was observed upon stimulation at 1 Hz, the population of synapses displaying a wide range of fractional destaining rates. Repeated measurements revealed that fractional destaining rates at 1 Hz were reproducible and therefore characteristic of each synapse. In contrast, at 10 Hz heterogeneity was less pronounced and the duration of heterogeneity was limited to approximately the first 60 stimuli. Similarly, the effect of phorbol ester on destaining at 10 Hz was limited to the first 60 stimuli.
Previous authors have described the release properties of hippocampal synapses in terms of the segregation of vesicles into different pools. Two functional pools have been proposed. The recycling vesicle pool consists of all vesicles that will recycle in response to electrical activity. The readily releasable pool, a subset of the recycling pool, consists of vesicles which are preferentially released, possibly because they are already docked at the active zone and/or primed for release. Correlations have been observed between (i) release probabilty and the number of vesicles in the readily releasable pool (Dobrunz & Stevens, 1997), (ii) the numbers of vesicles in total recycling and readily releasable pools (Murthy & Stevens, 1999), and (iii) total recycling pool size and release probability (Murthy et al. 1997). In addition, morphological studies have revealed correlations between other parameters such as number of docked vesicles and area of the active zone (Schikorski & Stevens, 1997). One interpretation of these studies is that many presynaptic properties scale in a stringent manner, such that two synapses with similar active zone areas will contain similar numbers of recycling vesicles, have similar readily releasable pool sizes and so display similar release probabilities.
Viewed in terms of these multi-pool models, the heterogeneity we observed may be related to the readily releasable pool of vesicles, since (i) the readily releasable pool strongly influences the efficacy of release in response to single stimuli and, therefore, during firing at 1 Hz (Dobrunz & Stevens, 1997), (ii) the readily releasable pool is responsible for release during only the first few tens of stimuli at 10 Hz (Dobrunz & Stevens, 1997; Murthy & Stevens, 1999), and (iii) phorbol esters increase the number of vesicles in the readily releasable pool (Stevens & Sullivan, 1998). Hence, different fractional destaining rates would result if either (i) the release probability per readily releasable vesicle differs from synapse to synapse, or (ii) the proportion of recycling vesicles in the readily releasable pool differs from synapse to synapse. These two explanations are not mutually exclusive, but each might potentially account for the observed heterogeneity without recourse to the other. These two possible explanations are considered below.
Dobrunz & Stevens (1997) found a close correlation between release probability and readily releasable pool size for single hippocampal synapses. This suggests that variation in release probability per readily releasable vesicle is minimal and, therefore, that the heterogeneity that we observed is unlikely to reflect different release probabilities per readily releasable vesicle.
Our alternative hypothesis concerns the proportion of recycling vesicles in the readily releasable pool. As discussed above, previous authors have found a correlation between recycling and readily releasable pool sizes, suggesting that the proportion of vesicles in the readily releasable pool is constant (Murthy & Stevens, 1999). These previous conclusions were derived from measurements at individually identified synapses. However, the conclusions derived from these studies were derived from mean relationships between synaptic parameters, measured across populations of synapses. It is possible, therefore, that these data describe the mean trends across synaptic populations, but that the properties of individual synapses may deviate substantially from these trends.
Our hypothesis is that the proportion of recycling vesicles in the readily releasable pool differs from synapse to synapse and that this accounts for the heterogeneity of our data. This hypothesis may be consistent with published data if the mean proportion of recycling vesicles in the readily releasable pool is equal to that of these earlier studies. The mean fractional destaining observed in our experiments after 60 stimuli at 10 Hz was 31.9 ± 1.1 %. This is remarkably similar to the figure reported by Murthy & Stevens (1999) who calculated that 32 % of recycling vesicles were within the readily releasable pool. The mean recycling vesicle pool size was also similar at all initial fractional destaining rates at 10 Hz (Fig. 4D), again consistent with the observations of Murthy & Stevens (1999). In view of the above discussion, it is important to note that we observed correlations between total recycling pool and readily releasable pool sizes and release probability similar to those reported by previous authors (data not shown).
We therefore consider it likely that heterogeneity in fractional destaining rates at 1 Hz and during the initial phase of destaining at 10 Hz reflects synapse-to-synapse differences in the proportion of vesicles in the readily releasable pool. Although we cannot categorically exclude other possible mechanisms, or a combination of two or more mechanisms, we consider that this hypothesis provides the most likely explanation for our data.
The effects of phorbol ester are readily explained by this hypothesis. At both 10 Hz and 1 Hz, phorbol ester treatment strongly potentiated release only at synapses with low fractional release rates. The lack of sensitivity of some synapses to PDBu may indicate that there is a ceiling on the proportion of the recycling pool which can be maintained in the readily releasable condition and that this ceiling had been reached at some synapses prior to PDBu treatment. Our data indicate that a physiological stimulus acting in a similar manner to phorbol esters, such as activated protein kinase C, may push the proportion of recycling vesicles in the readily releasable pool towards this ceiling value, altering the frequency dependence of transmitter release in a synapse-specific manner.
Our conclusions, detailed above, lead to the following model of synaptic release. The efficacy of release in response to a single stimulus or during low-frequency firing (insufficient to induce facilitation or depression) is determined principally by the number of vesicles in the readily releasable pool. Dobrunz & Stevens (1997) also arrived at this conclusion following detailed examination of the properties of individual synapses using electrophysiological techniques and the hippocampal slice preparation. Our data suggest that the size of the readily releasable pool is determined by two factors: the total recycling vesicle pool size and the proportion of these vesicles in the readily releasable condition. As a result, mean synaptic efficacy scales with total pool size. However, scaling is imprecise since the release properties of each individual synapse may be dominated by the proportion of its recycling vesicles that are partitioned into its readily releasable pool. The stringency of scaling can therefore be increased by forcing the proportion of recycling vesicles in the readily releasable pool towards the same value at all synapses. This can be achieved in either of two ways: by exhausting the readily releasable pool using a sustained high-frequency stimulus (which pushes the proportion in the readily releasable pool towards zero at all synapses) or by phorbol ester treatment (which forces the proportion in the readily releasable pool towards its ceiling value).
This model is also consistent with studies at other CNS synapses indicating that release probability is not determined by vesicle pool size alone. At the calyx of Held, for instance, recycling pool size and release probability change in opposite directions during development from P5 to P14, effectively compensating for each other (Taschenberger & von Gersdorff, 2000; Iwasaki & Takahashi, 2001). Clearly, these two parameters do not scale at the calyx of Held. Similarly, the release probabilities at climbing and parallel fibre synapses in the cerebellum are markedly different despite similar docked pool sizes (Xu-Freedman et al. 2001) indicating that parameters other than docked pool size must influence release probability at one or both of these synapses. That pool size is not the only determinant of release probability has also been suggested for hippocampal synapses, where ultrastructural studies have indicated that the proportion of vesicles released differs substantially between synapses (Harata et al. 2001). Although the underlying mechanisms controlling release probability are not clear at each of these synapses, these data clearly preclude stringent scaling with vesicle pool size.
Although our data suggest that vesicle partitioning between recycling and readily releasable pools is an important determinant of release probability at hippocampal synapses, other factors are also likely to be important and may contribute to the synaptic variability that we have observed. One such factor is presynaptic receptors, which have been shown to modulate release at the calyx of Held (Takahashi et al. 1996), in the locus coeruleus (Dubé & Marshall, 2000) and mossy fibre synapses in the hippocampus (Scanziani et al. 1997). Another possible variable is the extent to which release occurs without fusion and mixing of the vesicle and presynaptic plasma membranes. The latter could permit staining, but not destaining of synaptic vesicles (Henkel & Betz, 1995). If release were to occur by two pathways, one entailing complete fusion of presynaptic vesicles and another incomplete fusion, the balance between these two pathways might also contribute to the synaptic variability that we have observed.
Implicit in our conclusions is the assumption that the intensity of fluorescence staining is linearly related to the recycling vesicle pool size. In order for this assumption to be accurate, all recycling vesicles must be stained with dye and all stained vesicles must emit equal fluorescence. Previous authors have shown that staining with just a few action potentials yields fluorescence intensity distributions consistent with equal staining of each vesicle (Murthy et al. 1997; Ryan et al. 1997). Hence our fluorescence measurements should accurately represent the number of vesicles stained. Whether all released vesicles are stained is less clear and we cannot exclude the possibility that some vesicles recycle without becoming stained with FM dye. However, our estimates of recycling pool size arise from FM dye staining using high-frequency stimulation (10 Hz). FM staining seems to correlate quite well with the size of the docked vesicle pool following high-frequency stimulation (Schikorski & Stevens, 2001). Hence, although it is possible that incomplete FM staining of the recycling pool may lead to an underestimate of the number of vesicles in the recycling pool, any such underestimate is likely to be minimal.
We have shown that, at the level of individual synapses, synaptic efficacy is less directly related to recycling vesicle pool size than was previously thought. Hence recycling vesicle pool size and synaptic efficacy are related by a frequency-dependent variable, these two parameters being tightly linked only during sustained firing at high frequencies, such as 10 Hz. As a result, the relationship between vesicle pool sizes and synaptic efficacy may change through time during firing patterns observed in vivo (Czurkó et al. 1999; Hirase et al. 1999). Activation of phorbol ester-sensitive signalling cascade(s) also tightens the relationship between total pool size and synaptic efficacy. In this manner, sustained firing and second messenger cascades each exert synapse-specific influences on synaptic weighting. This mechanism may therefore have important implications for the encoding of information and output of neural circuits.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
| BETZ, W., MAO, F. & SMITH, C. (1996). Imaging exocytosis and endocytosis. Current Opinion in Neurobiology 6, 365-371 | [Medline] |
| CZURKÓ, A., HIRASE, H., CSICSVARI, J. & BUZSÁKI, G. (1999). Sustained activity of hippocampal pyramidal cells by 'space clamping' in a running wheel. European Journal of Neuroscience 11, 344-352 | [Medline] |
| DOBRUNZ, L. E. & STEVENS, C. F. (1997). Heterogeneity of release probability, facilitation, and depletion at central synapses. Neuron 18, 995-1008 | [Medline] |
| DUBÉ, G. R. & MARSHALL, K. C (2000). Activity-dependent activation of presynaptic metabotropic glutamate receptors in locus coeruleus. Journal of Neurophysiology 83, 1141-1149 | [Abstract/Full Text] |
| GODA, Y. & SÜDHOF, T. C. (1997). Calcium regulation of neurotransmitter release: reliably unreliable? Current Opinion in Cell Biology 9, 513-518 | [Medline] |
| HENKEL, A. W., BETZ, W. J. (1995). Staurosporine blocks evoked release of FM1-43 but not acetylcholine from frog motor nerve terminals. Journal of Neuroscience 15, 8246-58 | [Abstract] |
| HESSLER, N. A., SHIRKE, A. M. & MALINOW, R. (1993). The probability of transmitter release at a mammalian central synapse. Nature 366, 569-572 | [Medline] |
| HIRASE, H., CZURKÓ, A., CSICSVARI, J. & BUZSÁKI, G. (1999). Firing rate and theta-phase coding by hippocampal pyramidal neurons during 'space clamping'. European Journal of Neuroscience 11, 4373-4380 | [Medline] |
| IWASAKI, S. & TAKAHASHI, T. (2001). Developmental regulation of transmitter release at the calyx of Held in rat auditory brainstem. Journal of Physiology 534, 861-871 | [Abstract/Full Text] |
| MURTHY, V. N. (1999). Optical detection of synaptic vesicle exocytosis and endocytosis. Current Opinion in Neurobiology 9, 314-320 | [Medline] |
| MURTHY, V. N., SEJNOWSKI, T. J. & STEVENS, C. F. (1997). Heterogeneous release properties of visualized individual hippocampal synapses. Neuron 18, 599-612 | [Medline] |
| MURTHY, V. N. & STEVENS, C. F. (1999). Reversal of synaptic vesicle docking at central synapses. Nature Neuroscience 2, 503-507 | [Medline] |
| ROSENMUND, C., CLEMENTS, J. D. & WESTBROOK, G. L. (1993). Nonuniform probability of glutamate release at a hippocampal synapse. Science 262, 754-757 | |
| RYAN, T. A., REUTER, H. & SMITH, S. J (1997). Optical detection of a quantal presynaptic membrane turnover. Nature 388, 478-482 | [Medline] |
| RYAN, T. A., REUTER, H., WENDLAND, B., SCHWEIZER, F. E., TSIEN, R. W. & SMITH, S. J (1993). The kinetics of synaptic vesicle recycling measured at single presynaptic boutons. Neuron 11, 713-724 | [Medline] |
| RYAN, T. A. & SMITH, S. J (1995). Vesicle pool mobilization during action potential firing at hippocampal synapses. Neuron 14, 983-989 | [Medline] |
| RYAN, T. A., SMITH, S. J & REUTER, H. (1996). The timing of synaptic vesicle endocytosis. Proceedings of the National Academy of Sciences of the USA 93, 5567-5571 | [Abstract] |
| SCANZIANI, M., SALIN, P. A., VOGT, K. E., MALENKA, R. C. & NICOLL, R. A. (1997). Use-dependent increases in glutamate concentration activate presynaptic metabotropic glutamate receptors. Nature 385, 630-634 | [Medline] |
| SCHIKORSKI, T. & STEVENS, C. F. (1997). Quantitative ultrastructural analysis of hippocampal excitatory synapses. Journal of Neuroscience 17, 5858-5867 | [Abstract/Full Text] |
| SCHIKORSKI, T. & STEVENS, C. F. (2001). Morphological correlates of functionally defined synaptic vesicle populations. Nature Neuroscience 4, 391-395 | [Medline] |
| STEVENS, C. F. & SULLIVAN, J. M. (1998). Regulation of the readily releasable vesicle pool by protein kinase C. Neuron 21, 885-893 | [Medline] |
| TAKAHASHI, T., FORSYTHE, I. D., TSUJIMOTO, T., BARNES-DAVIES, M. & ONODERA, K. (1996). Presynaptic calcium current modulation by a metabotropic glutamate receptor. Science 274, 594-597 | |
| TASCHENBERGER, H. & VON GERSDORFF, H. (2000). Fine-tuning an auditory synapse for speed and fidelity, developmental changes in presynaptic waveform, EPSC kinetics, and synaptic plasticity. Journal of Neuroscience 20, 9162-9173 | [Abstract/Full Text] |
| WATERS, J. & SMITH, S. J (2000). Phorbol esters potentiate evoked and spontaneous release by different presynaptic mechanisms. Journal of Neuroscience 20, 7863-7870 | [Abstract/Full Text] |
| WIENER, S. I., PAUL, C. A. & EICHENBAUM, H. (1989). Spatial and behavioural correlates of hippocampal neuronal activity. Journal of Neuroscience 9, 2737-2763 | [Abstract] |
| XU-FREEDMAN, M. A., HARRIS, K. M. & REGEHR, W. G. (2001). Three-dimensional comparison of ultrastructural characteristics at depressing and facilitating synapses onto cerebellar purkinje cells. Journal of Neuroscience, 21, 6666-6672 | [Abstract/Full Text] |
Acknowledgements
We would like to thank Murali Prakriya and Charles F. Stevens for advice and comments on the manuscript. This work was supported by funds from the NIMH Silvio Conte Center for Neuroscience Research (MH48108) to S.J S.
This article has been cited by other articles:
|
|
 |
|
  K. L. Moulder, X. Jiang, A. A. Taylor, W. Shin, K. D. Gillis, and S. Mennerick Vesicle Pool Heterogeneity at Hippocampal Glutamate and GABA Synapses J. Neurosci., September 12, 2007; 27(37): 9846 - 9854. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Houeland, A. Nakhost, W. S. Sossin, and V. F. Castellucci PKC Modulation of Transmitter Release by SNAP-25 at Sensory-to-Motor Synapses in Aplysia J Neurophysiol, January 1, 2007; 97(1): 134 - 143. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P.-Y. Pan, Q. Cai, L. Lin, P.-H. Lu, S. Duan, and Z.-H. Sheng SNAP-29-mediated Modulation of Synaptic Transmission in Cultured Hippocampal Neurons J. Biol. Chem., July 8, 2005; 280(27): 25769 - 25779. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Wang and M. W. Quick Trafficking of the Plasma Membrane {gamma}-Aminobutyric Acid Transporter GAT1: SIZE AND RATES OF AN ACUTELY RECYCLING POOL J. Biol. Chem., May 13, 2005; 280(19): 18703 - 18709. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
