Stimulation-Induced Formation of the Reserve Pool of Vesicles in Drosophila Motor Boutons (original) (raw)

Abstract

We combined electron microscopy (EM), synaptic vesicle staining by fluorescent marker FM1-43, photoconversion of the dye into an electron dense product, and electrical recordings of synaptic responses to study the distribution of reserve and recycling vesicles and its dependence on stimulation in Drosophila motor boutons. We showed that, at rest, vesicles are distributed over the periphery of the bouton, with the recycling and reserve pools being intermixed and the central core of the bouton being devoid of vesicles. Continuous high-frequency stimulation followed by a resting period mobilized the reserve vesicles into the recycling pool and, most notably, produced an increase in vesicle abundance. Recordings of synaptic activity from the temperature-sensitive endocytosis mutant shibire during continuous stimulation until complete depression provided an independent estimate of the increase in vesicle abundance on intense stimulation. EM analysis demonstrated that continuous stimulation produced an increase in the vesicle density, whereas during a subsequent resting period, vesicles filled empty areas of the bouton, spreading toward its central core. Although the observed structural potentiation did not alter basal transmitter release, it produced an increased synaptic enhancement during high-frequency stimulation. The latter effect was not observed when the boutons were potentiated using high-frequency stimulation without a subsequent resting period. We concluded therefore that the newly formed vesicles replenish the reserve pool during a resting period following intense stimulation.

INTRODUCTION

Three distinct pools of synaptic vesicles are universal among different preparations: the readily releasable pool, the recycling pool, and the reserve pool (Rizzoli and Betz 2005). The readily releasable pool is comprised of vesicles docked to the presynaptic membrane and properly activated for release (Rosenmund and Stevens 1996; Sakaba and Neher 2001; Schikorski and Stevens 1997). The recycling pool maintains exo/endocytosis at moderate stimulation paradigms (Harata et al. 2001; Koenig and Ikeda 2007; Lin et al. 2005; Richards et al. 2003). The reserve pool is a depot of synaptic vesicles that are mobilized into the recycling pathway during intense stimulation (Gaffield et al. 2006; Koenig and Ikeda 1996; Richards et al. 2000; Takei et al. 1996; Voglmaier and Edwards 2007).

The studies using the uptake of the lipophylic dye FM1-43 combined with electron microscopy (EM) and photoconversion of the dye (de Lange et al. 2003; Harata et al. 2001; Rizzoli and Betz 2004) showed that, at vertebrate synapses, the recycling and reserve vesicles are largely intermixed. In contrast, in the Drosophila neuromuscular junction, the recycling vesicle pool was suggested to occupy the periphery of the bouton, whereas the reserve pool was assumed to be spread toward the central core of the bouton (Kidokoro et al. 2004). This concept was based on optical studies that showed that the dye loading at mild stimulation paradigms produced peripherally distributed labeling of the bouton, whereas high-frequency stimulation produced labeling of the entire area of the bouton, including its central part (Akbergenova and Bykhovskaia 2007; Kuromi and Kidokoro 1998, 2000; Verstreken et al. 2005).

However, ultrastructural study of the Drosophila nmj (Roche et al. 2002) showed that, typically, the central region of type Ib boutons is devoid of synaptic vesicles. Consistently, electron micrographs of Drosophila type Ib boutons presented in classical ultrastructural studies (Karunanithi et al. 1997; Kittel et al. 2006; Renger et al. 2000) show vesicles distributed over the periphery of the bouton with a central part of the bouton being devoid of vesicles and partially occupied by mitochondria. Furthermore, visualization of the entire vesicle pool (Poskanzer and Davis 2004) did not produce labeling of central parts of synaptic boutons.

These studies suggest a hypothesis that a substantial redistribution of synaptic vesicles might occur in a bouton on intense stimulation, and this might change the release efficacy of the bouton. To test this hypothesis, we used EM analysis to study the effect of intense stimulation on vesicle distribution and abundance.

METHODS

Preparations and chemicals

Canton S strain of Drosophila melanogaster was used in this study. Where indicated, experiments were performed on shibire ts1 mutant (stock 106278, Bloomington Drosophila Stock Center at Indiana University) or syt-eGFP mutant (stock 6923, same source; Zhang et al. 2002). Experiments were performed on Ib boutons (Dasari and Cooper 2004; Lnenicka and Keshishian 2000) of muscles 6 and 7 of abdominal segments 2, 3, or 4 of third-instar larvae. Preparations were dissected in physiological solution containing (in mM) 130 NaCl, 36 sucrose, 5 KCl, 2 CaCl2, 2 MgCl2, and 5 HEPES, pH 7.3. (Jan and Jan 1976), pinned to sylgard, and cut open along the dorsal midline, and internal organs were removed to expose the nerves and the muscles.

FM1-43 staining

FM1-43 (2.5 μM; Molecular Probes) was loaded in the presynaptic boutons during stimulation. After dye loading, preparations were briefly washed in Ca2+-free solution containing 75 mM Advasep-7 (Biotium) and rinsed in physiological saline without Advasep-7. Application of Advasep-7 (Kay et al. 1999) reduced background fluorescence and sharpened the borders of the boutons.

Confocal imaging

All the stained boutons were imaged with identical settings using a Zeiss FS2 microscope with a ×63 water immersion objective (0.95 NA) connected to a real-time confocal unit (Ultraview, PerkinElmer Life Sciences) equipped with a CCD camera (ORCA ER, Hamamatsu). FM1-43 fluorescence was exited at 488-nm Kr/Ar laser wavelength, and a 500-nm-long-pass emission filter was used for FM1-43 detection. Z series were taken at 0.3-μm steps to image the whole bouton. To ascertain that we collected data from only one type of bouton (Ib), only large boutons (≥3 μm diam) were analyzed.

Quantitative analysis of confocal images

The quantitative analysis of confocal stacks was performed as described in our earlier study (Akbergenova and Bykhovskaia 2007). Briefly, Volocity 6.1 software (Improvision) was used for three-dimensional reconstruction of the boutons, for the measurements of the integral fluorescence intensity, and for the measurements of the stained volume of a bouton. For each series of confocal images, a background area was selected, a histogram of fluorescent intensities of the background area was constructed, and the maximal background intensity was chosen as a threshold. In each confocal plane, a stained portion of a bouton was selected automatically, as all the pixels where fluorescence intensity exceeded the threshold. The stained volume was calculated as a sum of all the pixels where fluorescence intensity exceeded the background threshold over a stack of confocal planes. Integral fluorescence of the stained volume was calculated by subtracting the background threshold value from the fluorescence intensity at each pixel and summation of all the pixel intensities over the stack of confocal planes.

To compare the distribution of FM1-43 fluorescence with the distribution of synaptic vesicles obtained by EM, we calculated the stained area of the bouton. Adobe Photoshop 10.0 was used to create and analyze the image of a bouton that represented the average over a series of confocal planes. Stained area of a bouton was calculated as a sum of all the pixels in this image where the fluorescence intensity exceeded a background threshold. The entire area of the bouton was calculated as a sum of all the pixels within outlined borders of the bouton.

To obtain fluorescent profiles of individual boutons (Akbergenova and Bykhovskaia 2007) (Fig. 1 C), Scion software (Scion) was used to analyze the image of a bouton that represented the average over a series of confocal planes. A fluorescent profile was calculated along the major (longest) cross-line of a bouton. Relative fluorescent intensities of individual pixels were plotted against the relative positions of these pixels along the major cross-line.

FIG. 1.

FIG. 1.

Vesicles occupy the periphery of Drosophila Ib boutons. A: electron micrograph showing synaptic vesicles distributed over the periphery of a bouton with a central core devoid of vesicles and partially occupied by mitochondria. Scale bar: 200 nm. B: the area of a bouton occupied by vesicles was outlined manually for a subsequent quantitative analysis. Scale bar: 200 nm. C: synaptic boutons loaded with the dye FM1-43 during the stimulation at 3-Hz frequency for 15 min (left) and unloaded during the stimulation at 3-Hz frequency for 10 min with no dye added (right). Ring-shaped staining pattern indicates the peripheral distribution of the recycling vesicles. Scale bar: 1 μm. D: stained area of the bouton (left) was outlined automatically (right) as all the pixels with the intensity exceeding the background threshold. Scale bar: 1 μm. E: FM1-43–stained area of the bouton matches the area occupied by synaptic vesicles obtained by the electron microscopy (EM) analysis. Data collected from 21 FM1-43 stained boutons (14 larvae) and 26 electron micrographs (3 larvae).

EM

Preparations were fixed in 1% glutaraldehyde/4% paraformaldehyde in 0.1 M cocodylate buffer for 2 h at room temperature and incubated at 4°C overnight. After washing in 0.1 M cocodylate buffer with 0.1 M sucrose added, samples were postfixed in 1% osmium tetroxide, dehydrated through a graded series of ethanol and acetone, and embedded in Embed 812 epoxy resin (Electron Microscopy Sciences). Thin sections (70–90 nm) were collected on Formvar/carbon-coated copper slot grids, and contrasted with lead citrate. Samples were examined on a Phillips transmission electron microscope at 100 kV. Micrographs were analyzed using Adobe Photoshop 10.0 software. Vesicle density was calculated as the total number of vesicles divided by the area occupied by vesicles. Areas devoid of vesicles or those occupied by mitochondria were not included in the calculation of the vesicle density. To ascertain that only Ib type boutons were analyzed and to exclude the sections that were cut at the edge of a bouton, we selected the micrographs showing the boutons of ≥2 μm diam with small (∼30 nm diam) clear vesicles (Atwood et al. 1993; Jia et al. 1993). In stimulated preparations, we labeled the stimulated segment by a cactus needle, which is always clearly seen in the tissue fixed for EM, and thus the stimulated segment was readily identified in the semi-thin slices and in the micrographs. In some preparations, we labeled the area of interest by polystyrene fluorescent microspheres, as described in Kapitsky et al. (2005).

Photoconversion of the dye FM1-43

The photoconversion procedure was adopted from Harata et al. (2001), Schikorski and Stevens (2001), and Teng and Wilkinson (2000). After dye loading (10 μM FM1-43 added to the bath), preparations were fixed for 15 min in a regular EM fixative, washed for 30 min in physiological saline, preincubated in DAB (1.5 mg/ml in physiological saline) for 10 min, and illuminated for 20 min under a ×63 water immersion objective using a mercury lamp with a 485 ± 10 band-pass excitation filter. At this point, fluorescence staining was completely bleached, and DAB reaction product was visible. After a brief superfusion with physiological saline, the illuminated site was marked by a cactus needle to be identified in a subsequent EM analysis. The preparations were left overnight in regular EM fixative and processed as for conventional EM.

Recordings and quantal analysis of postsynaptic responses

Synaptic responses were recorded from preparations bathed in physiological solution optimized for recordings of synaptic activity (Stewart et al. 1994), containing (in mM) 70 NaCl, 5 KCl, 1 CaCl2, 20 MgCl2, 10 NaHCO3, 5 trehalose, 115 sucrose, and 5 HEPES. The nerve was stimulated electrically via suction electrode of a 10-μm-diam tip. Synaptic responses were recorded focally from the boutons visualized with DIC optics using macropatch electrode of 5- to 10-μm tip diameter. The electrodes were manually bent to enable recordings under a ×63 magnification water immersion objective (Zeiss) with 1.8-mm working distance. Recordings were digitized with Digidata A/D board and Axoscope software (Axon Instruments) and analyzed off-line. Quantal content was calculated as a ratio between the average amplitude of synaptic responses and the amplitude of spontaneous quantal events. Evoked and spontaneous synaptic potentials were detected using in-house software (Bykhovskaia 2008). Spontaneous activity was recorded for 5 min at each recording site.

Statistical analysis

Datasets were compared used two-sided _t_-test and one-way ANOVA test. The difference was tested at the 5% level of significance.

RESULTS

Vesicles have peripheral distribution and the reserve and recycling pools are intermixed

We used conventional EM to examine the distribution of all the vesicles in the bouton (Fig. 1, A and B). The results of the EM analysis were correlated with the staining pattern observed after loading the dye FM1-43 at a mild stimulation paradigm (Fig. 1, C and D). When the dye FM1-43 was loaded during a low-frequency (3 Hz) electrical stimulation of the nerve for 15 min, a ring-shaped staining pattern was observed (Fig. 1_C_, left). More prolonged stimulation did not produce any further increase in the fluorescence (data not shown), indicating that the entire recycling pool was stained. The boutons were almost completely destained during the nerve stimulation at 3-Hz frequency for 10 min with no dye added (Fig. 1_C_, right). To study the dye distribution quantitatively, we measured the area of the stained portion of the bouton (Fig. 1_D_; 57.0 ± 2.6%).

The peripheral distribution of the dye was previously interpreted as a peripheral distribution of the recycling pool of vesicles and more central distribution of the reserve pool (Kuromi and Kidokoro 2000). However, the EM analysis showed that typically all the vesicles are distributed predominantly over the periphery of the bouton, with central parts of the boutons being devoid of vesicles and partially occupied by mitochondria (Fig. 1, A and B). Morphometric analysis showed that only 52 ± 3% of the bouton is occupied by vesicles (Fig. 1_E_). Our data agree with the earlier microscopy and ultrastructural studies (Gramates and Budnik 1999; Marek and Davis 2002; Poskanzer and Davis 2004; Roche et al. 2002) where synaptic vesicles were not generally observed at the center of synaptic boutons.

Thus we showed that the spatial distribution of the recycling pool correlates with the spatial distribution of all the vesicles in the bouton. This result suggests that the reserve and recycling pool might be spatially intermixed. To address this question conclusively, we used photoconversion of the dye into an electron dense product (Harata et al. 2001). To stain selectively the recycling pool, the dye was loaded at 3 Hz for 15 min, the preparations were treated to photoconvert the dye, and EM analysis was performed. Stained vesicles appeared black on electron micrographs (Fig. 2, A–C). We found that the recycling pool (black vesicles) constituted 32.0 ± 3.2% of the total pool of vesicles. Black and white vesicles appeared homogeneously intermixed. To confirm this observation quantitatively, we counted recycling (black) vesicles in the immediate vicinity of the membrane (within 100 nm) and in the layers positioned further away (100–200 and 200–300 nm). Layers with a more central location were not analyzed, because, in the majority of the boutons, these layers were devoid of vesicles. The proportion of recycling vesicles was similar among all the layers (Fig. 2_D_), showing that recycling vesicles do not have a preferential location in the bouton.

FIG. 2.

FIG. 2.

Reserve and recycling vesicles are intermixed in Drosophila motor boutons. A_–_C: micrographs showing intermixed recycling (dark lumen) and reserve (translucent) vesicles in Ib type boutons. Scale bar: 200 nm. D: the density of recycling (dark) vesicles is not affected by their proximity to the synaptic membrane. Data collected from 14 boutons (3 larvae).

Thus we showed that the recycling and the reserve pools are intermixed. This result challenges the interpretation of earlier studies, where dye loading at a high stimulation frequency produced spreading of the dye toward the center of the bouton (Akbergenova and Bykhovskaia 2007; Kuromi and Kidokoro 2000; Verstreken et al. 2005), which was explained as a mobilization of the centrally located reserve vesicle pool (Kuromi and Kidokoro 2000). Our EM analysis suggests an alternative interpretation, namely, that on intense stimulation, extra vesicles are formed and spread toward the central core of the bouton. To test this hypothesis, we studied synaptic ultrastructure and vesicle distribution after electrical stimulation of the nerve at a high (10 Hz) frequency.

High-frequency stimulation produces formation of extrasynaptic vesicles

Central parts of type Ib boutons can be stained when the dye FM1-43 is loaded at a stimulation frequency of 10 Hz (15 min) followed by a 10-min resting period (Akbergenova and Bykhovskaia 2007; Verstreken et al. 2005) (Fig. 3 A, left). To understand whether this loading pattern produces any changes in synaptic vesicle distribution, we performed EM analysis of the preparations that were fixed after the stimulation and a subsequent resting period. When the dye FM1-43 was loaded using this stimulation protocol, a significantly larger part of the bouton was stained (Fig. 3_B_), the labeling was more evenly distributed (Fig. 3_C_), and fluorescence intensity was significantly higher (Fig. 3_D_) than during the dye loading at a 3-Hz stimulation frequency. Subsequent stimulation at a low frequency (3 Hz for 10 min) with no dye added produced only partial destaining (Fig. 3, A and D). These results indicate that a centrally located population of vesicles became involved in the exo/endocytic pathway, and this population of vesicles is not readily involved in exocytosis at mild stimulation paradigms.

FIG. 3.

FIG. 3.

An extra pool of vesicles is formed after intense stimulation of the nerve. A: synaptic boutons loaded with the dye FM1-43 during the stimulation at 10-Hz frequency for 15 min followed by a 10-min resting period (left) and unloaded during the stimulation at 3-Hz frequency for 10 min with no dye added (right). Note that the entire boutons, including their centers, are stained (left) and that the central core of the bouton is not destained (right). Scale bar: 1 μm. B: volume of the FM1-43–stained portion of the bouton after the dye loading at a 10-Hz stimulation frequency significantly exceeds the volume of the stained portion of the bouton obtained after the dye loading at 3 Hz. C: fluorescent profiles of FM1-43–stained boutons. A peripheral staining at a 3-Hz frequency typically produces a saddle-like profile with a distinct minimum (top left), whereas staining of the entire bouton at a 10-Hz frequency typically produces a bell-shaped profile (top right, a different bouton). Average fluorescent profiles (bottom) of the boutons stained at 3- and 10-Hz stimulation frequencies are significantly different, showing a peripheral staining at 3 Hz and a staining of the entire bouton at 10 Hz. D: integral fluorescence (relative units) after the dye loading at 3 and 10 Hz and after destaining at 3 Hz for 10 min (gray portions of the bars). The integral fluorescence of the boutons loaded at 10-Hz frequency was significantly increased, whereas destaining was only partial, indicating that an extra population of vesicles has been involved in the recycling pathways during the stimulation at 10 Hz. E: an electron micrograph showing a typical preparation stimulated at 10 Hz for 15 min with a subsequent 10-min rest. Vesicles are spread largely over the entire bouton with only a small portion of the bouton devoid of vesicles. Scale bar: 200 nm. F: in stimulated preparations, the area occupied by vesicles was significantly increased. G: in stimulated preparations, the total number of vesicles per synaptic bouton was significantly increased. H: in stimulated preparations, a modest but statistically significant increase in the vesicle density was observed. EM data collected from 23 boutons (3 larvae); FM1-43 data collected from 15 boutons (14 larvae). *Significant difference (P < 0.05, 2-sided _t_-test).

To test this hypothesis, we stimulated the nerve (15 min at 10 Hz + 10-min rest) and fixed the preparations for EM. The stimulated segments were labeled (as described in methods) and identified on the micrographs. Morphometric analysis was performed to determine the area occupied by synaptic vesicles, vesicle density, and vesicle abundance. Notably, in the stimulated preparations, synaptic vesicles were spread toward the center of the bouton (Fig. 3_E_) and occupied a significantly larger portion of the bouton than at rest (Fig. 3_F_). The areas devoid of vesicles diminished, and they were largely occupied by mitochondria. Furthermore, overall vesicle abundance (Fig. 3_G_) significantly increased after nerve stimulation. These results suggest that extra vesicles have been formed after the intense stimulation, and these vesicles occupied the parts of the bouton that had been empty at rest. A rather substantial increase in vesicle abundance (by 43%; Fig. 3_G_) was associated with a noticeable increase in the area occupied by vesicles (from 52 ± 3 to 71 ± 4% of a bouton) and with a modest (by 18%; Fig. 3_H_) but statistically significant increase in vesicle density.

Thus we showed that intense stimulation produces a structural potentiation, i.e., an increase in the number of vesicles in synaptic boutons. To obtain an independent assessment of this increase in vesicle abundance, we took advantage of a temperature sensitive endocytosis mutant shibire. This mutant is normal at room temperature, whereas at nonpermissive temperatures (>29°C), exocytosis is intact but endocytosis is completely blocked, and nerve terminals are completely depleted of vesicles after stimulation (Koenig and Ikeda 1983; Poodry and Edgar 1979). We used focal recordings of postsynaptic responses from individual Ib type boutons (Fig. 4, A and B) to evaluate the number of vesicles in potentiated and nonpotentiated boutons. In control experiments, the shibire preparations were stimulated at nonpermissive temperatures (35°C) at the frequency of 10 Hz until complete depletion (Fig. 4, C and D, ▪). Calculating the net quantal content of all the recorded responses allowed us to evaluate the number of synaptic vesicles per bouton (Fig. 4_E_, white bar). To evaluate the number of vesicles per bouton in potentiated preparations, we potentiated shibire preparations at a permissive temperature (19°C) using the 10 Hz for 15 min + 10-min rest stimulation paradigm and increased the temperature to 35°C and used the vesicle depletion stimulation paradigm (10-Hz frequency until complete depletion). We found that potentiated preparations depressed more slowly (Fig. 4_D_, □) and had significantly more vesicles (Fig. 4_E_, black bar). This experiment provides an independent proof of the increase in the number of synaptic vesicles in potentiated boutons. Remarkably, the estimates of the increase in vesicle abundance obtained by the two used methods, EM analysis and shibire depletion, are in excellent agreement (43 vs. 40%).

FIG. 4.

FIG. 4.

Complete depression of potentiated and nonpotentiated shibire boutons showed that potentiation increases the number of vesicles per bouton. A: focal recordings from Ib boutons observed by DIC. Boutons (left) are marked by arrowheads. The recording electrode (right) is positioned against a synaptic bouton. B: examples of the recorded synchronous excitatory postsynaptic potentials (EPSPs; an arrowhead, 3 sweeps are superimposed) and asynchronous spontaneous events (arrows). C: continuous stimulation at nonpermissive temperatures produced depression, caused by the endocytic blockade. D: the depression was less pronounced in potentiated preparations, suggesting the increase in the number of vesicles. E: estimated vesicle number in potentiated (open bar) and nonpotentiated (solid) bar boutons. *Significant difference (P < 0.05, 2-sided _t_-test). Data collected from 31 boutons (31 larvae).

It should be noted, however, that the observed increase in the vesicle abundance (by 40–43%) cannot account completely for a very pronounced (sixfold; Fig. 3_D_) increase in FM1-43 fluorescence produced by the potentiation paradigm. Because the photoconversion experiments showed that, at rest, only 32.0 ± 3.2% of all the vesicles belong to the recycling pool, the observed discrepancy between a drastic increase in fluorescence and a modest increase in vesicle abundance can be ascribed to mobilization of the existing reserve pool in the exo/endocytic pathway. To test this hypothesis and to study how the extra pool of vesicles is formed, we performed the photoconversion experiment after vesicle staining during intense stimulation.

New vesicles are formed by enhanced endocytosis during intense stimulation and redistributed toward the central core of the bouton during a subsequent resting period

To test whether the increase in vesicle abundance on intense stimulation might have been produced by enhanced endocytosis, we stained the vesicles using the 10 Hz for 15 min + 10-min rest stimulation paradigm and performed the photoconversion assay. We found that, on this stimulation paradigm, most of the vesicles became involved the recycling process (Fig. 5, A_–_C). Thus in addition to the increase in vesicle numbers, intense stimulation drastically increased the proportion of vesicles participating in exo/endocytosis (from 32 to 93%; Fig. 5_D_).

FIG. 5.

FIG. 5.

During the potentiation paradigm (15 min at 10 Hz + 10-min rest), most of the vesicles undergo the dye uptake. A_–_C: micrographs showing vesicles that did undergo the dye uptake (dark lumen) and those that did not participate in recycling (translucent) in Ib type boutons. Scale bar: 200 nm. D: proportion of the stained vesicles after the dye loading at a mild stimulation paradigm (3 Hz) or at an intense stimulation parading (10 Hz). The latter paradigm produces staining of a vast majority (93%) of vesicles. Data collected from 14 boutons (3 larvae). *Significant difference between the datasets collected at 3- and 10-Hz stimulation frequencies for the total number of vesicles (black asterisk) and the number of vesicles that have uptaken the dye (gray asterisk).

These results showed that, on intense stimulation, two processes take place. First, most of the vesicles that were not involved in exo/endocytosis during mild stimulation paradigms (the reserve pool, ∼68% of all the vesicles present at rest) become involved in the recycling pathway. Second, formation of extra vesicles takes place. Most likely, these extra vesicles are formed by means of excess membrane retrieval, because Fig. 5_D_ clearly shows that most of the extra vesicles did uptake the dye.

Excess membrane retrieval is likely to produce shrinkage of the presynaptic membrane and a decrease in bouton size. To test whether this is the case, we took advantage of the syt-eGFP line (Zhang et al. 2002), which expressed green fluorescent protein (GFP)-labeled synaptotagmin. We imaged a bouton at syt-eGFP preparations at rest and stimulated the nerve for 15 min at 10-Hz frequency and acquired a stack of confocal images immediately following the stimulation. Finally, we acquired the third stack of images following a 10-min resting period (Fig. 6 A, 2 representative boutons).

FIG. 6.

FIG. 6.

Stimulation of nmjs at the green fluorescent protein (GFP)-tagged synaptotagmin mutant (Syt-eGFP) showed that boutons shrink on the potentiation paradigm (15 min at 10 Hz + 10-min rest), and synaptotagmin redistributes toward the center of the bouton during the resting period. A: 2 representative boutons: overlay of confocal stacks (top) and individual confocal planes. Images are taken before stimulation (left), immediately following stimulation (middle), and after the resting period (right). B: the area of the entire bouton, including its central core, was outlined and measured from the overlay of confocal series. C: the size of the bouton significantly decreased after the stimulation and did dot change after the subsequent resting period. *Significant (P < 0.05, 1-way ANOVA test) difference from nonstimulated preparations. Data collected from 20 boutons (11 larvae).

We found that the stimulation was typically accompanied by a significant shrinkage of the boutons and a change in their shape (Fig. 6_A_, middle vs. left column). During a subsequent resting period, some redistribution of fluorescence toward the center of the bouton was observed (Fig. 6_A_, right vs. middle column). Area measurements (Fig. 6_B_) showed that boutons significantly shrink following stimulation (Fig. 6_C_) but do not change their size after the subsequent resting period.

The observed redistribution of fluorescence toward the central core of the bouton during the resting period (Fig. 6_A_) suggests that vesicles may redistribute toward empty areas of the bouton not during the stimulation but during a subsequent resting period. To test this hypothesis, we analyzed the ultrastructure of the preparations that have been stimulated for 15 min at 10-Hz frequency and were fixed immediately after the stimulation without a delay (Fig. 7 A). Notably, the boutons fixed immediately after the stimulation had an increased number of vesicles, and this number did not significantly increase during the delay (Fig. 7_B_). However, the delay significantly altered the vesicle distribution (Fig. 7, A, C, and D). In the preparations fixed with no delay, the vesicle density was significantly higher (Fig. 5_C_), and vesicles occupied a significantly smaller area (Fig. 7_D_) than in the preparations fixed after the delay. Thus during nerve stimulation, vesicle abundance and vesicle density increases, probably because of an enhanced endocytosis. During a subsequent resting period, vesicles redistribute and spread over a larger area (Fig. 7_D_) toward the center of the bouton, and consequentially, the vesicle density decreases (Fig. 7_C_).

FIG. 7.

FIG. 7.

During the delay following an intense stimulation, vesicles redistribute over a larger area toward the center of the bouton, but their abundance does not change. A: electron micrographs showing a bouton fixed immediately after the stimulation (left; 15 min at 10 Hz) and a bouton fixed after the stimulation with a 10-min delay. Note that vesicles are densely packed over the periphery of the bouton fixed with no delay and spread over the bouton fixed after the delay. Sale bar: 200 nm. B: the number of synaptic vesicles per bouton is not significantly altered by the delay. C: vesicle density is decreased after the delay. D: area occupied by the vesicles is increased after the delay. *Significant difference (P < 0.05, 2-sided _t_-test). Data collected from 41 boutons (8 larvae).

It should be noted that this model does not support the conclusions that were based on an earlier study (Kuromi and Kidokoro 2002), where selective FM1-43 staining of the central core of the bouton (presumably, the reserve pool) was achieved using a “delayed load protocol.” This study showed (Kuromi and Kidokoro 2002) (Fig. 1) that loading the dye during a delay period following high-frequency stimulation produced selective staining of a central part of the bouton. To address the discrepancy, we tested whether the central core of the bouton can be selectively stained during the delay period. We added the dye FM1-43 immediately following the intense stimulation (15 min at 10 Hz) and loaded it for 10 min in the absence of stimulation (Fig. 8, A and B, 2 representative boutons). The observed staining was very faint and rather nonspecific, with no preference for the central core of the bouton (bright-field images of the strings of the boutons are presented in Fig. 8, C and D). To ascertain that the boutons were functional, in the end of each experiment we loaded them with the dye in the presence of high K+ (90 mM) and achieved bright staining (Fig. 8, E and F). Thus in our hands, the central core of the bouton could not be stained selectively using the “delay load protocol.” This result further supports our conclusion that extra vesicles are formed during intense stimulation and redistributed toward the central core of the bouton during a subsequent resting period.

FIG. 8.

FIG. 8.

“Delayed load” paradigm does not produce a selective staining of the centrally located reserve pool. A and B: 2 representative images taken after the delay load paradigm (FM1-43 added to the solution after the 10-Hz, 15-min stimulation and loaded for 10 min) show very faint and rather nonspecific staining. C and D: the same fields of view taken with a bright field illumination show distinct strings of boutons. E and F: subsequent staining during high K+ application (90 mM for 5 min) shows that the boutons were functional.

Newly formed extra vesicles are directed to the reserve pool following intense stimulation

Thus our results suggest a new mechanism for synaptic potentiation, specifically, formation of an extra pool of vesicles following intense stimulation. To test how this structural potentiation affects synaptic function, we recorded spontaneous and evoked synaptic activity after the intense stimulation followed by a resting period (potentiation with rest; Fig. 9 A) or immediately after intense stimulation (potentiation without rest; Fig. 9_B_). Following either potentiation paradigm, synaptic responses (at 3-Hz stimulation frequency) or spontaneous synaptic activity (for 5 min) was recorded. Neither evoked (Fig. 9_C_) nor spontaneous (Fig. 9_D_) activity was affected by the preconditioning stimulation, suggesting that the newly formed vesicles do not participate in basal neuronal transmission. These results also indicate that the fusion machinery was not affected by either potentiation paradigm.

FIG. 9.

FIG. 9.

Vesicles newly formed on intense stimulation do not initially participate in exocytosis but replenish the reserve pool during the resting period following the stimulation. A: stimulation protocol: potentiation with rest. Basal evoked and spontaneous release (1) or synaptic enhancement during a continuous stimulation (2) recorded after the preconditioning intense stimulation followed by a 10-min delay. B: stimulation protocol: potentiation without rest. Basal evoked and spontaneous release (1) or synaptic enhancement during a continuous stimulation (2) recorded immediately after the preconditioning intense stimulation. C: quantal content of evoked synaptic responses was not affected by either potentiation paradigm. D: the rate of spontaneous release was not affected by either potentiation paradigm. E: synaptic enhancement during continuous stimulation was increased on potentiation with rest (○). Potentiation without rest (▵) did not produce any increase in synaptic enhancement (compare with control, ▪), and even produced a significant depression after 5-min stimulation. The 3 traces are significantly different (P < 0.05, regression test). Each data point represents an average ± SE from 100 subsequent EPSPs. Overall, data collected from 73 boutons (73 larvae).

To test whether the newly formed pool of vesicles might become involved in the recycling pathway during intense stimulation, we compared the rate of release during a prolonged continuous stimulation at 10-Hz frequency in potentiated (with or without rest; Fig. 9, A and B) and nonpotentiated preparations. In control, the rate of release increased by ∼25% within the initial 1.5–2 min of stimulation, reached a plateau, and started to gradually decay after 7–10 min of stimulation (Fig. 9_E_, ▪). These kinetics were significantly altered in preparations potentiated with rest (Fig. 9_E_, ○). A very pronounced increase in synaptic amplitude (by ∼50%) was observed within the initial 20 s of stimulation. Synaptic responses continued to increase and reached ∼170% of the initial synaptic amplitude after 1.5–2 min of stimulation. Thus in preparations potentiated with rest, the synaptic enhancement during a prolonged continuous stimulation at 10-Hz frequency was significantly stronger than in nonpotentiated preparations. It is notable that this increase in synaptic enhancement was not a consequence of a lower initial release, because basal evoked release was not affected by either potentiation paradigm (Fig. 9_C_). The observed increase in synaptic enhancement suggests that the newly formed vesicles were directed to the reserve pool and thus became involved in the recycling during subsequent intense stimulation.

This was not the case for the preparations potentiated without rest (Fig. 9_B_). These preparations did not show any significant change in synaptic enhancement (Fig. 9_E_, ▵). In contrast, as the test stimulation continued over 5 min, these preparations showed an increased depression. Thus in the preparations potentiated without rest, no functional synaptic enhancement was observed. Evidently, in these preparations, the newly formed vesicles did not participate in the exo/endocytic process.

Taken together, these results suggest that the vesicles formed during an intense stimulation are initially unable to undergo exocytosis and recycling. However, during a resting period following the stimulation, they are directed to the reserve pool and can undergo exocytosis during a subsequent intense stimulation.

DISCUSSION

The main finding of our study was that, during prolonged high-frequency stimulation, Drosophila motor boutons form an extra pool of vesicles, presumably by means of enhanced endocytosis. The increase in the number of vesicles in synaptic boutons was shown using EM analysis following continuous stimulation, as well as using a depletion paradigm in the temperature-sensitive endocytosis mutant shibire. Thus we showed that a synapse becomes structurally potentiated after continuous stimulation by increasing abundance of synaptic vesicles. Although the basal transmitter release remained unchanged after this structural potentiation, synaptic enhancement during high-frequency stimulation was increased. We concluded therefore that the newly formed vesicles are directed to the reserve pool and thus get involved in the recycling pathway only during prolonged high-frequency stimulation.

We found that potentiation occurs in two stages. First, vesicle density increases as a result of a continuous stimulation, and boutons significantly shrink. At this point, the newly formed vesicles seem to be unavailable for exocytosis and recycling. Second, during a consecutive resting period, vesicles redistribute within synaptic boutons, filling its central core and become a part of the reserve pool, which can be involved in recycling during a subsequent intense stimulation. Thus it could be suggested that the empty central core of type Ib boutons serves as a reserve space, which can be filled with the reserve pool of vesicles on intense stimulation.

In Drosophila type I boutons, vesicles have peripheral distribution and the reserve and recycling pools are intermixed

It has already been noted in an earlier studies (Jia et al. 1993; Roche et al. 2002) that the central core of the Ib type bouton is typically devoid of vesicles and partially occupied by mitochondria and nonvesicular structures. The results of EM analysis presented here agree with these studies. We showed that vesicles occupy ∼50% of the bouton and that the peripheral distribution of vesicles can account for the ring-shaped staining pattern of FM1-43 loading during nerve stimulation at low frequency.

Optical monitoring of the dye FM1-43 performed in Drosophila Ib motor boutons (Kuromi and Kidokoro 2005 for review) suggested peripheral distribution of the recycling pool and more central distribution of the reserve pool. We performed photoconversion of the dye combined with EM analysis and showed that, contrary to this suggestion, the reserve and recycling pool are intermixed. Similar observations have been made at vertebrate synapses (Harata et al. 2001; Rizzoli and Betz 2004; Schikorski and Stevens 2001).

Thus our results compel us to revise a view (reviewed in Kidokoro et al. 2004; Kuromi and Kidokoro 2005) that the reserve pool occupies, predominantly, the central core of the Drosophila Ib boutons. Rather, we showed that, at rest, the reserve and recycling pools are intermixed and that the central core of the bouton is filled with newly formed vesicles on intense stimulation.

It is of interest whether vesicles are packed densely over the periphery of the bouton because of some active mechanism or whether such a distribution results from local recycling in the vicinity of the synaptic membrane. Notably, during continuous stimulation, vesicle density increases while the area occupied by vesicles remains unchanged, and this observation favors the existence of an active mechanism for the vesicle packing and peripheral distribution. Because our previous study (Akbergenova and Bykhovskaia 2007) showed that the peripheral FM1-43 loading pattern is disrupted in synapsin null mutants, it is plausible that the peripheral vesicle distribution is maintained by synapsin/actin interactions. Such a mechanism would be consistent with an established role of synapsin in tethering vesicles to the cytomatrix and maintaining vesicle clustering (Hilfiker et al. 1999). In this case, vesicle redistribution during the resting period following intense stimulation could be attributed to synapsin phosphorylation (Chi et al. 2001, 2003). This hypothesis remains to be tested directly.

Structural potentiation: the vesicle number increases

We showed that a nerve terminal can increase its efficacy by increasing vesicle abundance. It is already established that the nerve stimulation can result in mobilization of reserve vesicles into the exo/endocytic pathway (Richards et al. 2003; Rizzoli et al. 2003), vesicle movement toward active zones (Applegate and Landfield 1988; Leenders et al. 2002) and, on a long-term scale, filling empty varicosities with synaptic vesicles (Kim et al. 2003). Our study provides the first evidence that synaptic potentiation can occur by filling empty areas of an active bouton with extra vesicles.

We favor the hypothesis that extra vesicles are formed by an enhanced endocytosis. In strong support of this hypothesis, we showed that, after the potentiation paradigm, most of the vesicles took up the dye (Fig. 5). One could suggest that the extra vesicles could be formed by other mechanisms, for example, by an axonal transport, and then uptake the dye following subsequent exocytosis and recycling. However, our study of synaptic activity in potentiated boutons (Fig. 9) showed that the newly formed vesicles do not initially participate in exocytosis and recycling (in other words, belong to a resting pool; Poskanzer and Davis 2004), but become directed to the reserve pool following a subsequent resting period. Taken together, these results strongly support the hypothesis that the vesicle numbers are increased during intense stimulation by means of enhanced endocytosis. This hypothesis was further strengthened by our observation that synaptic boutons significantly shrink and reduce their size following continuous stimulation.

This suggestion implies that the presynaptic membrane is sufficiently plastic to supply the material for the formation of an extra vesicle pool without compromising the synaptic architecture. Because it is known that, during massive exocytosis, the entire vesicle pool can be substantially depleted (Applegate and Landfield 1988; Colasante and Pecot-Dechavassine 1996; Heuser and Reese 1981), it seems plausible that the presynaptic membrane is indeed very plastic, so it can accommodate either enhanced exocytosis or enhanced endocytosis. This suggestion is in line with capacitance measurements performed in neurosecretory cells (Engisch and Nowycky 1998), which showed that stimulus-evoked exocytosis can be followed by excess membrane retrieval that surpasses exocytosis.

The potentiation of the bouton by means of enhanced endocytosis also implies that a reservoir of synaptic vesicle proteins exists on the plasma membrane, and these synaptic proteins can be readily taken up again. This suggestion is supported by recent studies (Fernandez-Alfonso et al. 2006; Wienisch and Klingauf 2006) that showed that a large reservoir of synaptic proteins exists on the plasma membrane, and these proteins are taken up into vesicles during endocytosis. At present, it is hard to tell whether this membrane reservoir of synaptic proteins would be sufficient to provide for the excess endocytosis. We cannot rule out the possibility that synaptic vesicle proteins may be delivered by axonal trafficking and inserted into newly endocytosed vesicles with a delay. Such a mechanism would be consistent with a finding that axonal trafficking can be activity dependent (Shakiryanova et al. 2006), and, hypothetically, it could explain the observed inability of the newly exocytosed vesicles to participate in the exocytic process immediately on their uptake. We also cannot completely rule out the possibility that a small portion of the newly formed vesicles is delivered via axonal transport.

Functional potentiation: the reserve pool is involved in the recycling process and newly formed vesicles are directed to the reserve pool

We visualized the reserve and recycling pool of vesicles using the FM1-43 photoconversion technique during a mild and intense stimulation paradigm and showed directly that 1) the reserve pool exists in Drosophila nmj, and 2) it becomes involved in the recycling pathway during intense stimulation paradigms. Furthermore, analysis of synaptic activity following intense stimulation showed that extra vesicles formed during the intense stimulation are initially unable to participate in the recycling pathway and hence they belong to the resting pool. During a resting period following the intense stimulation, these extra vesicles replenish the reserve pool, i.e., the pool of vesicles that becomes involved in recycling only during intense stimulation. Thus our results showed the importance of a resting period following an intense stimulation. We showed that, during this period, vesicles undergo spatial redistribution in the bouton, filling its central core, and gain an ability to undergo exocytosis in response to intense stimulation, being thus directed to the reserve pool.

Thus our study points to the major mechanisms that enable a synapse to maintain sustained activity during intense stimulation: 1) recruitment of reserve vesicles; 2) formation of new vesicles and an increase in vesicle numbers, probably by means of enhanced endocytosis; and 3) replenishment of the reserve pool by the newly formed vesicles, which is associated with their spatial redistribution.

GRANTS

This study was supported by National Institutes of Health Grants R01 MH-61059 and U54 NS-039408.

Acknowledgments

We thank Dr. Thomas Schikorski for help in adjusting the photoconversion procedure to the Drosophila preparation.

REFERENCES