Nerve terminal specializations include mechanisms for maintaining a subpopulation of vesicles in a docked, fusion-ready state. We have investigated the relationship between synaptotagmin and the number of morphologically docked vesicles by an electron microscopic analysis ofDrosophila synaptotagmin (syt) mutants. The overall number of synaptic vesicles in a terminal was reduced, although each active zone continued to have a cluster of vesicles in its vicinity. In addition, there was an increase in the number of large vesicles near synapses. Examining the clusters, we found that the pool of synaptic vesicles immediately adjacent to the presynaptic membrane, the pool that includes the docked population, was reduced to 24 ± 5% (means ± SEM) of control in thesyt null mutation.
To separate contributions of overall vesicle depletion and increased spontaneous release from direct effects of synaptotagmin on morphological docking, we examined syt mutants in an altered genetic background. Recombining syt alleles onto a second chromosome bearing an as yet uncharacterized mutation resulted in the expected decrease in evoked release but suppressed the increase in spontaneous release frequency. Motor nerve terminals in this genotype contained more synaptic vesicles than control, yet the number of vesicles immediately adjacent to the presynaptic membrane near active zones was still reduced (33 ± 4% of control).
Our findings demonstrate that there is a decrease in the number of morphologically docked vesicles seen in syt mutants. The decreases in docking and evoked release are independent of the increase in spontaneous release. These results support the hypothesis that synaptotagmin stabilizes the docked state.
To account for the rapidity of synaptic transmission, it has been proposed that a subset of synaptic vesicles in nerve terminals is docked at active zones where they form a pool that is readily releasable by an action potential. Electron microscopy of terminals has revealed a population of synaptic vesicles immediately adjacent to the presynaptic membrane (Couteaux and Pecot-Dechavassine, 1973). At least a portion of these morphologically docked vesicles is likely to correspond to the physiologically docked, fusion-ready pool.
Biochemical experiments have implicated the vesicle protein synaptotagmin in several aspects of nerve terminal function, including Ca2+ sensing (Brose et al., 1992; Chapman et al., 1995; Sutton et al., 1995; Shao et al., 1996, 1997) and endocytosis (Zhang et al., 1994; Jorgensen et al., 1995). In addition, synaptotagmin may participate in vesicle docking (Petrenko et al., 1991; Bennett et al., 1992; Sollner et al., 1993). Synaptotagmin contains two C2 repeats (Perin et al., 1990; Wendland et al., 1991), and homologous motifs occur in protein kinase C and cytosolic phospholipase A2, where they are thought to mediate a Ca2+-dependent translocation of these enzymes to membranes (Clark et al., 1991). Several presynaptic membrane proteins bind synaptotagmin in vitro, including syntaxin, SNAP-25, neurexins, and the receptors for activated protein kinase C (Petrenko et al., 1991; Bennett et al., 1992; Mochly et al., 1992; Schiavo et al., 1997). These interactions may recruit vesicles to the release site in a manner analogous to the translocation of other C2 domain-containing proteins.
Recent genetic and pharmacological studies provide direct support for an involvement of synaptotagmin in neurosecretion (Bommert et al., 1993; Elferink et al., 1993), but they also suggest that synaptotagmin may not be essential for synaptic transmission. Drosophilalarvae and Caenorhabditis elegans that lack the synaptotagmin (syt) gene are sluggish and uncoordinated yet are able to crawl and feed (DiAntonio et al., 1993b; Littleton et al., 1993b; Nonet et al., 1993). Evoked transmitter release is reduced to ∼10% of control at the neuromuscular junctions of Drosophila syt null mutants while spontaneous vesicle release is increased (Broadie et al., 1994). Similar reductions in Ca2+-stimulated release are seen in hippocampal cultures from mice with altered synaptotagmin I (Geppert et al., 1994).
A decrease in evoked transmitter release could arise from several possible defects individually or in combination: a decrease in the number of docked vesicles, a decrease in the efficacy of Ca2+-sensing or fusion, or an overall decrease in the number of vesicles. Two of these possibilities, an overall decrease in vesicles and a decrease in docked vesicles, can be addressed by a morphological examination. In the present study CNS synapses as well as a defined neuromuscular synapse were analyzed by light and electron microscopy in Drosophila syt mutants. The decrease in the number of morphologically docked vesicles that we observed in the absence of synaptotagmin supports the hypothesis that synaptotagmin stabilizes the docked state of vesicles at release sites.
MATERIALS AND METHODS
Genetics. Four synaptotagmin mutant lines were used for analysis. syt AD4is a null mutation with a stop codon at amino acid 32. This mutant will be referred to as syt null.syt AD3 is a hypomorph with a Y to N mutation at amino acid 364 (DiAntonio and Schwarz, 1994).*,syt AD3 and*,syt null are chromosomes on which the syt alleles were placed in a different genetic background. These chromosomes were generated by homologous recombination between a second chromosome bearing one of thesyt mutations (syt null orsyt AD3 chromosomes) and one bearing a P-element (P[HsGal4]). When the P-element containing portion of these recombined chromosomes is made homozygous, the increase in spontaneous transmitter release normally seen insyt mutants is suppressed (see Results). It is possible that the change in spontaneous release frequency is attributable to a novel mutation caused by the insertion of this P-element. Indeed, when the P-element containing parent chromosome (without any sytmutations) is made homozygous, the flies are uncoordinated (our unpublished observation). However, it is also possible that these changes could be caused by more complicated multigenic factors located on the chromosome bearing the P-element. Therefore, we will refer to this recombinant P-element-bearing second chromosome simply as *. Oregon R (OrR) larvae were used for wild-type controls.
To collect null mutants for fixation, we out-crossed females (syt null/Gla,Bc) toOrR (+/+) males.syt null/+ siblings were collected and crossed, and homozygous mutant (syt null/syt null) first instar larvae were selected on the basis of delayed hatching and sluggish behavior (DiAntonio et al., 1993b).syt AD3/syt nullheterozygous larvae also were examined. These were generated by crossing syt AD3/+ bysyt null/+ and selecting as above. The third and fourth syt mutations also were studied as heterozygotes (*,syt AD3/*,syt nulland*,syt AD3/syt null). These were generated by crossing*,syt AD3/Gla,Bc by*,syt null/Gla,Bc or by syt null/Gla,Bcand selecting third instar larvae lacking Bc. An independent mutant line, In(2)syt D27, which is also a null for syt (DiAntonio et al., 1993b), had a qualitatively similar phenotype.
Immunohistochemistry. Larvae were glued (Nexaband, Burns Veterinary Supply, Farmer’s Branch, TX) to Sylgard-coated dishes containing cold Ca2+-free Fly Ringer’s solution (Jan and Jan, 1976). The cuticle was dissected open with a glass needle, the gut was removed, and the larvae were fixed in 1% formaldehyde in PBS for 15 min. They were rinsed briefly in PBS containing 0.1% Triton X-100 (PBST) and incubated overnight at 4°C in DCSP-1 [a monoclonal antibody directed againstDrosophila cysteine string protein (Zinsmaier et al., 1994)] or a polyclonal rabbit antibody directed against horseradish peroxidase (HRP; ICN Biochemicals, Costa Mesa, CA) diluted 1:100 in dilution medium (PBST containing 10% normal goat serum). They were washed in PBST for 3 hr, incubated for 1 hr in a fluoresceinated secondary antibody (ICN Biochemicals), washed in PBST for 1 hr, and mounted in Citiflur AF-1 (City University, London, UK). For DCSP-1 experiments the CNS in whole mounts of first instar larvae were photographed on a Zeiss Axiophot microscope (Oberkochen, Germany). For anti-HRP experiments the synaptic boutons were counted on muscle fiber number 6 from abdominal segments 2–5 of third instar larvae (38 fibers from five animals for*,syt AD3/*,syt nulland 34 fibers from five animals for wild-type).
Electron microscopy. Dissected larvae were transferred immediately to cold primary fixative (1% acrolein and 2.5% glutaraldehyde in 0.1 m cacodylate buffer, pH 7.1) for 30–60 min. They were post-fixed in 1% osmium tetroxide in 0.1m cacodylate for 1 hr, embedded in 1.5% agar to facilitate handling, dehydrated, and embedded in Embed 812 (Electron Microscopy Sciences, Fort Washington, PA). The 70 nm sections were stained with uranyl acetate and Sato’s lead (Sato, 1967). For first instar larvae a section was cut at a random orientation approximately through the center of a larval brain hemisphere. The entire neuropil of this cross section was photographed at 12,000× magnification in the electron microscope for each of 10 larvae (threesyt null, threesyt AD3/syt null, and four wild-type). For third instar larvae, sections through the region of neuromuscular junctions on muscle fiber number 6 from abdominal segments 2–5 were collected from each of six larvae (three*,syt AD3/*,syt nulland three wild-type). Electron micrographs covering the junctional region were taken at 12,000× magnification.
Image analysis. Control and experimental electron micrographs were printed together at the same magnification (3×) and then were coded and randomized for blind analysis. In each experiment the micrographs from a mutant larva were randomized with micrographs from a wild-type larva. However, both mutants were not always in each experiment; therefore, more wild-type synapses and larvae were used. Random synapses from widespread regions of first instar larval CNS were selected for analysis by using an adaptation of standard morphometric search protocols (Weibel, 1979). Each micrograph was overlaid with a grid pattern and was sampled systematically (in a spiral pattern) for synapses that had clear pre- and postsynaptic membranes. To avoid weighting the measurements in any given brain region, we included no more than five synapses per micrograph in the analysis. The first five synapses encountered in the grid pattern that fit the search criteria (i.e., clear pre- and postsynaptic membranes) were marked, regardless of the number of synapses per terminal or the number of synapses per micrograph. This morphometric search protocol usually resulted in only one synapse per terminal being marked; however, the protocol occasionally yielded two synapses in a given terminal. For analysis of vesicle distributions the individual synapses were magnified (∼3×) through a video camera, and the images were captured onto a Macintosh computer. The total numbers of synapses analyzed were 231 for wild-type (from four larvae), 154 forsyt null (from three larvae), and 122 forsyt AD3/syt null(from three larvae).
The presynaptic membrane, as determined by the extent of the synaptic cleft, was marked in these cross sections (e.g., Fig.2 B) and measured with either an Image 1 analysis system (Universal Imaging, West Chester, PA) or National Institutes of Health Image software (Bethesda, MD). The shortest distance from the presynaptic membrane to the center of each vesicle was measured (e.g., Fig. 2 B). To avoid the inclusion of vesicles located closer to a neighboring synapse, we included only vesicles within 200 nm of the presynaptic membrane in the quantification. Almost all vesicles in the small category were clear and 30 nm in diameter, which is typical of small clear synaptic vesicles inDrosophila (Budnik et al., 1990). The large vesicle category was much more heterogeneous; vesicle diameters ranged from 45 to ∼90 nm, and some were opaque. Microtubules cut transversely occasionally may resemble a vesicular structure; however, their small diameter [∼20 nm (Peters et al., 1991)] permitted unequivocal exclusion from this study.
To assess the distribution of small clear synaptic vesicles with respect to the presynaptic membrane, we sorted CNS distance data into 6 nm bins. The mean number of vesicles per synapse was graphed versus distance from the presynaptic membrane (see Fig. 5 A). Because small clear synaptic vesicles have a radius of 15 nm, we have defined the morphologically docked pool as those vesicles for which the centers are 12–18 nm from the presynaptic membrane.
Similar measurements were made on third instar neuromuscular junctions, with a few modifications. After coding and randomizing mutant and wild-type micrographs, we marked neuromuscular junctions with clear pre- and postsynaptic membranes and at least one presynaptic dense body. Images were imported into National Institutes of Health Image software, as described above. To assess the distribution of vesicles in the vicinity of active zones, we marked 100 nm of presynaptic membrane on either side of a presynaptic dense body. Then the perpendicular distance from the marked presynaptic membrane to the center of each vesicle within 200 nm was measured (see Fig. 8 B). Because the presynaptic membrane of sectioned third instar larval neuromuscular junctions often extends several hundreds of nanometers beyond active zones, the measurements were restricted to vesicles directly above the marked region of the presynaptic membrane (i.e., near active zones). The total number of synapses analyzed was 91 for wild-type (from three larvae) and 106 for*,syt AD3/*,syt null(from three larvae).
Electrophysiology. Intracellular voltage recordings were made from body wall muscle 6 from abdominal segments 4 or 5 of third instar larvae according to the procedures of Stewart and colleagues (Stewart et al., 1994), except that stocks used for the recordings were maintained at 18°C. Larvae were dissected and recorded in HL3 Ringer’s solution [containing (in mm) 70 NaCl, 5 KCl, 1.5 CaCl2, 20 MgCl2, 10 NaHCO3, 5 trehalose, 115 sucrose, and 5 HEPES, pH 7.2]. Four fibers from*,syt AD3/*,syt nulllarvae, 10 fibers from*,syt AD3/syt nulllarvae, and six fibers from wild-type larvae were analyzed; only one fiber per larva was used. The CNS was removed by sectioning the nerves near the ventral ganglion. For evoked potentials the nerve end was stimulated by using a heat-polished suction electrode. Mean miniature excitatory junction potential (mEJP) frequency was calculated from data that were collected for ∼2 min.
Vesicles are targeted to and clustered at synapses insyt mutants
Two Drosophila lines carrying mutations in thesynaptotagmin gene (DiAntonio et al., 1993b; DiAntonio and Schwarz, 1994) were analyzed.syt AD4, which will be referred to simply as syt null, produces no synaptotagmin protein (see Materials and Methods) and is homozygous-lethal. syt AD3 is a point mutant with a single altered amino acid in its second C2 domain. This allele has a less severe phenotype and permits a few mutants to survive to adulthood when it is placed over a deficiency that removessyt(syt AD3/Df (2L)DTD2). In the present study we used larvae that were eithersyt null homozygotes or, a less severe allelic combination,syt AD3/syt nullheterozygotes.
Synapses were examined by electron microscopy in first instar larvae, the last stage at which viable larvae can be collected for the null alleles. Immunocytochemistry has demonstrated that the sytproduct is located both at CNS synapses and neuromuscular junctions (DiAntonio et al., 1993b; Broadie et al., 1994). To obtain a large number of nerve terminals for quantitative analysis, we used random sections through larval brain hemispheres (Fig.1 A,C,E; see Materials and Methods). Neuromuscular junctions also were examined in wild-type (Fig. 1 B) andsyt null (Fig.1 D) first instar larvae. Qualitatively, neuromuscular junctions showed the same phenotype as the central synapses of first instar larvae (see below).
To analyze CNS synapses, we adopted sampling procedures (see Materials and Methods) to insure a diverse population of synapses in each sample so that general and robust changes in vesicle distribution could be assessed. Large numbers of synapses from widespread regions of brain hemisphere neuropil were examined for each genotype. Synaptotagmin is expressed throughout the neuropil and is believed to be important for synaptic transmission at all or most synapses (DiAntonio et al., 1993a;Littleton et al., 1993a; Broadie et al., 1994; DiAntonio and Schwarz, 1994). Thus a role in a fundamental process such as vesicle docking would be expected to be manifest in the general population of synapses. For each mutant specimen that was examined, a wild-type control was sectioned and analyzed in parallel. As shown below, there was some variability in the observed morphological parameters from sample to sample. In part, this may be attributable to differences in the subsets of synaptic types that fell into a given sample. However, this variation proved small in comparison to the differences between genotypes. Thus, whether parameters were compared for each individual mutant specimen and its paired wild-type control (as in Fig. 6) or by pooling all of the data for a genotype (as in Figs. 3, 5, 7, 9), robust and statistically significant changes were detected that were attributable to the mutations.
syt mutants of both genotypes had a generally normal ultrastructure (Budnik et al., 1990): conventional presynaptic membrane specializations were observed, each with clear 30 nm synaptic vesicles clustered nearby, as in wild-type larvae (Figs. 1,2). However, there were some distinct differences in the vesicle populations. Nerve terminals fromsyt null mutants contained fewer 30 nm synaptic vesicles (arrow, compare Fig.1 C,D with A,B; see also Fig.3). This change was most dramatic in regions of the terminal that were distant from synaptic sites; few vesicles in mutant terminals were observed outside tight clusters near synapses (see Fig. 1 C–E). In addition, larger heterogeneous vesicles were more abundant in the mutants (Figs. 1 C–E, arrowhead; 3). The large vesicles were not concentrated near active zones.syt AD3/syt nullhad an intermediate phenotype with a less dramatic decrease in the number of synaptic vesicles (Figs. 1 E, arrow; 3) and an intermediate number of large vesicles (Figs. 1 E, arrowhead; 3).
To quantify the changes at mutant synapses, we counted vesicles within 200 nm of the presynaptic membrane. Because the morphological properties of individual synapses within the CNS were quite variable, a blind morphometric analysis of large numbers of randomly selected synapses was conducted to determine the average properties in each genotype (see Materials and Methods). Vesicles were categorized as either small clear synaptic vesicles (<45 nm in diameter) or large vesicles (≥45 nm in diameter). The number of synaptic vesicles per synapse in syt null homozygotes andsyt AD3/syt nullwas decreased to 48 and 66% of wild-type, respectively (Fig. 3). In contrast, large vesicles were 3.4-fold and 2.6-fold more abundant insyt null andsyt AD3/syt nullmutants, respectively (Fig. 3).
The decrease in synaptic vesicles observed in the mutant terminals raised the possibility that vesicles might not be targeted correctly in the mutants. For example, mutant vesicles might accumulate in the cell body because of inefficient transport down the axon, as seen in kinesin mutants (Hall and Hedgecock, 1991). A brief examination of the surrounding neuronal cell body layer, in sections in which the central neuropil was analyzed extensively, showed no obvious accumulations of 30 nm vesicles (n > 30 somas; data not shown). To investigate the overall distribution of vesicles more thoroughly, we examined the distribution of another synaptic vesicle-associated protein, cysteine string protein (CSP; Mastrogiacomo et al., 1994;Zinsmaier et al., 1994), by immunocytochemistry. As in control animals (data not shown), CSP staining in the nervous system ofsyt null mutants is highly concentrated in the neuropil, with minor staining in the cell body regions (Fig. 4). Thus, synaptic vesicles appear to be targeted correctly to nerve terminals in thesyt mutants.
The number of morphologically docked vesicles is reduced insyt mutants
The nerve terminal may be viewed as containing distinct pools of synaptic vesicles: synaptic vesicles that are not obviously associated with a synapse, synaptic vesicles that are clustered in the vicinity of a synapse, and synaptic vesicles that are immediately adjacent to the presynaptic membrane, which we define as morphologically docked (see Fig. 2 A, arrow; see also Materials and Methods). Although this last pool may include vesicles that are adjacent to the membrane at a synapse by chance, without being functionally docked, this pool also must include the docked vesicles (Couteaux and Pecot-Dechavassine, 1973; Koenig et al., 1993). Indeed, the morphological assessment of the docked state of vesicles has been used in numerous investigations of synaptic mechanisms (Bommert et al., 1993; Hunt et al., 1994; Broadie et al., 1995; O’Connor et al., 1997). By examining the distribution of synaptic vesicles relative to the presynaptic membrane, we have determined that the reduction in vesicle number is not uniform throughout the terminal. In particular, as described below, the pool of morphologically docked vesicles is reduced more sharply than the pool of vesicles clustered in the close vicinity of the synapse.
For this analysis we marked the extent of the presynaptic membrane and measured the distance from the center of each vesicle to the nearest point of the presynaptic membrane (see Fig. 2 B). The data were pooled into 6 nm bins (Fig.5 A). The bin of vesicles for which the centers were 12–18 nm (one radius ± 3 nm) from the presynaptic membrane was defined as the morphologically docked pool. Because the area in each bin generally increases with increasing distance from the presynaptic membrane (see Fig. 2 B), equal numbers of vesicles per bin actually reflect a decrease in vesicle density away from the synapse. To facilitate the comparison between the syt mutant and control terminals, we graphed the data for the mutants as a percentage of control (Fig. 5 B). The observed reduction in synaptic vesicles was not uniform across this region. Although syt mutants had fewer 30 nm synaptic vesicles than controls, the vesicles nevertheless were clustered tightly in the vicinity of synapses. In the mutants the pools of vesicles 18–50 nm from the presynaptic membrane showed the least attenuation. Here, both syt mutants had ∼75% of the control number of vesicles. Despite this relative accumulation of vesicles nearby, in the syt mutants the number of 30 nm vesicles immediately adjacent to the presynaptic membrane was reduced markedly: to 24% of control insyt null homozygotes and to 38% insyt AD3/syt null.
The dramatic decrease in the number of morphologically docked vesicles was a consistent finding that could not have resulted from the chance selection of atypical regions in individual specimens and a consequent distortion of the mean. To illustrate this point, in Figure6 we have graphed the mean number of morphologically docked vesicles per synapse for each individual mutant that was analyzed and paired it with the control specimen from the same blind experiment. In every pair the number of morphologically docked vesicles was reduced significantly in the synaptotagminmutant relative to its wild-type control (p values ranged from p < 0.02 to ≪ 0.001; Student’st test). Although variation naturally was encountered from synapse to synapse and from data set to data set, the trend was consistent and the change in the mean for each genotype was statistically different from control at p ≪ 0.001 (see Fig. 5 B). Thus, our methods for collecting large amounts of data from widespread regions of a single randomly oriented cross section through a brain hemisphere succeeded in identifying a widespread and dramatic change in vesicle docking that far exceeded the normal variation from sample to sample.
The mean cross-sectional length of the presynaptic membrane in these sections was not altered substantially in the mutants: wild-type, 177 ± 5 nm; syt null, 169 ± 5 nm;syt AD3/syt null, 196 ± 6 nm (length ± SEM). To determine whether these slight differences in mean cross-sectional presynaptic length between individual larvae significantly influenced the docking data, we normalized the number of morphologically docked vesicles per synapse to the mean length of presynaptic membrane in each larva. As shown in Table 1, the mean number of morphologically docked vesicles per micrometer of presynaptic membrane was not significantly different between individual larvae within each genotype (wild-type, p < 0.2;syt null, p < 0.6;syt AD3/syt null,p < 0.3; ANOVA), yet the mean number of morphologically docked vesicles per micrometer of presynaptic membrane was reduced to 25% of control insyt null (p≪ 0.0001; ANOVA) and 37% of control insyt AD3/syt null(p ≪ 0.0001; ANOVA). Thus, along with the decrease in morphological docking per synapse, there was a decrease in docking per unit length of presynaptic membrane.
The decrease in docking occurs at neuromuscular junctions and is independent of the increase in spontaneous release
To determine whether the decrease in the number of morphologically docked vesicles in the syt mutants was secondary to the increased frequency of spontaneous release, we examined the mutant genotype*,syt AD3/*,syt null. This genotype suppresses the increased rate of spontaneous release normally seen in syt mutants. The *,sytchromosomes were created by homologous recombination events that changed the genetic background of the second chromosome (see Materials and Methods). The basis of this change has not yet been characterized; it may be attributable to simple disruption of a novel gene by the P-element located on the * chromosome or to a more complex multigenic effect (see Materials and Methods). Because the*,syt AD3/*,syt nullmutants survive to the third instar larval stage, both the physiological measurements and the morphological measurements were done at the third instar neuromuscular junction of muscle fiber number 6.*,syt AD3/*,syt nullexhibited the expected decrease in evoked transmitter release (Fig.7 A). However, the increased rate of spontaneous transmitter release usually seen in sytmutants was suppressed in this genotype; the rate of spontaneous release now remained the same as control (Fig. 7 B;p < 0.8; Student’s t test). On the other hand,*,syt AD3/syt nullexhibited the expected increased rate of spontaneous release (Fig.7 B; p ≪ 0.001; Student’s t test). Thus, this suppression of the increased rate of spontaneous release was seen only in larvae that were homozygous for the uncharacterized portion (*) of the second chromosome (see Materials and Methods).
The decreased rate of spontaneous release in this genotype could arise from a decreased probability of release at the synapse or from a decrease in the number of synapses on the muscle fiber. To address the latter possibility, we counted the number of boutons on muscle fiber number 6 in anti-HRP-stained preparations of*,syt AD3/*,syt nulland wild-type. There was no statistically significant difference in the number of boutons per muscle fiber: mutants had 58.7 ± 4.1 (34 fibers from six animals), whereas wild-type had 53.1 ± 3.2 (38 fibers from six animals; mean number of boutons per fiber ± SEM;p < 0.3; Student’s t test).
To determine the effect of this change in genetic background on vesicle populations and vesicle distribution, we examined the ultrastructure of third instar neuromuscular junctions on muscle fiber number 6 (Fig.8). The distribution of synaptic vesicles with respect to the presynaptic membrane was determined as described above, with a few modifications. The presynaptic membrane at neuromuscular junctions of third instar larvae was often >1 μm in length. To restrict the analysis to the vicinity of active zones, we marked off 100 nm of presynaptic membrane on each side of the dense body of the active zone. Then the perpendicular distance from the center of each synaptic vesicle to this marked region of the presynaptic membrane was measured (Fig. 8 B). All of the vesicles within 200 nm perpendicular to this region of presynaptic membrane were included in the analysis.
We found that the overall number of synaptic vesicles near active zones in the*,syt AD3/*,syt nullmutants is increased slightly as compared with control (Fig.9 A; p < 0.01; Student’s t test). This increase in*,syt AD3/*,syt nullis in direct contrast to the overall reduction in the number of synaptic vesicles near synapses in the two synaptotagminmutants (syt null andsyt AD3/syt null) that exhibit an increased frequency of spontaneous release (Broadie et al., 1994; DiAntonio and Schwarz, 1994). The inverse correlation between vesicle number and spontaneous release frequency suggests that the overall number of vesicles near synapses may be influenced by the rate of spontaneous release.
The number of large vesicles near active zones in the*,syt AD3/*,syt nullmutants also is increased as compared with wild-type (Fig.9 A), similar to the changes in the other two mutants (see Fig. 3). Thus, the changes in the small vesicle population in the immediate vicinity of synapses do not correlate directly to the changes in the large vesicle population in this area.
To determine the number of morphologically docked vesicles in these mutants, we graphed the number of small vesicles per synaptic region versus the distance from the presynaptic membrane (Fig. 9 B). Because only synaptic vesicles directly above the marked region were included, the area in each bin is generally constant regardless of distance from the membrane. In wild-type synapses it is apparent that the pool of vesicles immediately adjacent to the presynaptic membrane is quite enriched. Naturally, because these data were obtained from third instar neuromuscular junctions and not first instar central synapses, the proportion of docked vesicles is not directly comparable to those in Figure 5 A. The data were regraphed as a percentage of control to aid comparison (Fig. 9 C). Despite the slight increase in the overall number of synaptic vesicles near active zones (Fig. 9 A), the number of morphologically docked vesicles is reduced markedly in these syt mutants to 33% of control (Fig. 9 C; p ≪ 0.001; Student’st test).
The results from the*,syt AD3/*,syt nullmutants demonstrate that the decrease in morphological docking (1) occurs at a defined peripheral synapse as well as at randomly sampled CNS synapses, (2) is independent of overall vesicle depletion, and (3) is independent of the increased rate of spontaneous vesicle fusions found in the other syt mutants.
We have conducted a morphometric analysis of the ultrastructural phenotype of synaptotagmin mutants. We are particularly interested in the subset of vesicles that are morphologically docked (i.e., immediately adjacent to the presynaptic membrane), because vesicles that are functionally docked are likely to be included in this pool (Couteaux and Pecot-Dechavassine, 1973; Bommert et al., 1993; Koenig et al., 1993; Hunt et al., 1994; Broadie et al., 1995; O’Connor et al., 1997). In the CNS those 30 nm vesicles for which the centers are located within 18 nm of the presynaptic membrane are defined as morphologically docked. At the neuromuscular junction the analysis was limited further to 100 nm of presynaptic membrane on either side of the dense body. Synapses from all of the syt mutants that were analyzed exhibited a dramatic decrease in morphologically docked vesicles: CNS synapses fromsyt null mutants had 24% of the control, CNS synapses fromsyt AD3/syt nullhad 38% of the control, and neuromuscular junctions on muscle fiber number 6 from*,syt AD3/*,syt nullhad 33% of the control number of morphologically docked vesicles per synapse.
Larval brain neuropil contains a heterogeneous mixture of synapses (Bate and Martinez Arias, 1993). Synaptotagmin is expressed throughout the neuropil and is thought to function at most, perhaps all, synapses (DiAntonio et al., 1993a; Littleton et al., 1993a; DiAntonio and Schwarz, 1994) although, naturally, synaptotagmin function has not been demonstrated at every synapse in the CNS. To minimize any sampling artifact caused by differences between individual CNS synapses or changes in active zone size, we took the following precautions during data collection and analysis: (1) We analyzed large numbers of randomly selected synapses (Weibel, 1979) from widespread regions of larval brain hemispheres and limited the number of synapses per micrograph included in the study to prevent synapses in any one region from unduly influencing the mean (see Materials and Methods). (2) We conducted a pairwise comparison of each mutant larva to its simultaneously analyzed control and found the morphological changes to be universal (e.g., Fig.6). A sampling artifact would require that all six mutant sections over-represent putative “low docking” areas, whereas all four wild-type sections over-represent putative “high docking” areas. Because each sample was independent and from a randomly oriented section, the differences we observed are attributable to the mutations. (3) We analyzed the number of morphologically docked vesicles per synapse (see Fig. 5) as well as the number per unit length of presynaptic membrane (see Table 1). (4) We included analysis of a defined synapse, the neuromuscular junction on muscle fiber number 6 (see Fig. 9). If the decrease in morphological docking at CNS synapses of syt mutants were an artifact of sampling or abnormal CNS development, then a similar decrease would not be expected near active zones of a specific neuromuscular junction.
We found that the mean number of morphologically docked vesicles at synapses in syt mutants was reduced dramatically in every case. Every pairwise comparison of CNS synapses from a sytmutant larva to its simultaneously analyzed wild-type control showed a decreased number of morphologically docked vesicles. When they were normalized per unit length of presynaptic membrane, sytmutants still exhibited a dramatic decrease in the mean number of morphologically docked vesicles as compared with wild-type (p < 0.0001; ANOVA), even though individual wild-type larvae were not significantly different from each other (p < 0.2; ANOVA). Data from the identified neuromuscular synapse fully corroborated the CNS data. Taken together, these data indicate that synaptotagmin function is required to achieve wild-type levels of morphological docking.
The decreased number of morphologically docked vesicles cannot be accounted for by overall vesicle depletion. First, the overall number of synaptic vesicles near synapses insyt null (the mutant with the most severe depletion) is reduced to 48% of control, but morphologically docked vesicles are reduced further to 24% of control. Second, the number of vesicles near but not touching the presynaptic membrane (the 18–50 nm bin; see Fig. 5 B) is reduced to only 73% of control; thus, the neighboring supply of vesicles shows the least depletion. Third, neuromuscular junctions in the*,syt AD3/*,syt nullmutants exhibit a decrease in the docked pool of vesicles even though there was a slight increase in the total number of vesicles nearby.
A similar analysis reveals that the decrease in docking is also independent of the rate of spontaneous release.syt null andsyt AD3/syt nullboth exhibit an increased rate of spontaneous release (Broadie et al., 1994; DiAntonio and Schwarz, 1994), whereas*,syt AD3/*,syt nulldoes not (see Fig. 7 B), yet the number of morphologically docked vesicles is decreased in all three. Thus, the effect of synaptotagmin on morphological docking is not secondary to vesicle depletion or an increased rate of spontaneous release but appears to be a primary function of synaptotagmin.
Synaptotagmin is important for efficient coupling of presynaptic activity to transmitter release. In syt mutants the evoked release is reduced to ∼10% of control, and the rate of spontaneous fusions is increased three- to fivefold (Broadie et al., 1994;DiAntonio and Schwarz, 1994). Evidence that synaptotagmin is a Ca2+-binding protein and therefore may serve as a Ca2+ sensor has overshadowed the potential role of synaptotagmin in vesicle docking (Bennett et al., 1992; Schiavo et al., 1997). However, synaptotagmin has been shown to interact with the presynaptic membrane proteins syntaxin (Bennett et al., 1992; Chapman et al., 1996; Kee and Scheller, 1996) and SNAP-25 (Schiavo et al., 1997), suggesting that synaptotagmin may help to anchor vesicles at release sites. Indeed, the persistence of morphologically docked vesicles in the presence of tetanus toxin, which cleaves some isoforms of VAMP/synaptobrevin (Hunt et al., 1994; Broadie et al., 1995), suggests that another vesicle protein or proteins may function during docking; synaptotagmin is an excellent candidate (Schiavo et al., 1997). Our ultrastructural results support the hypothesis that synaptotagmin plays a direct role in vesicle docking.
The overall depletion in synaptic vesicles and the increased number of large vesicles seen in some syt mutant terminals are consistent with the hypothesis that vesicle recycling or biosynthesis also may be compromised [as suggested for syt mutants ofC. elegans (Jorgensen et al., 1995)]. This depletion is particularly apparent in the null mutant, in which the absence of a major vesicle protein may decrease the efficacy of synaptic vesicle formation. The increased rate of spontaneous release seen in these mutants also may contribute to the overall vesicle depletion. The precise nature of the large irregular vesicles is not yet known; they may represent recently retrieved membrane, excess endosomes, or incorrectly assembled synaptic vesicles. At CNS synapses they may account for much of the membrane that is missing from the reduced small vesicle pool. However, further studies of syt mutants that use vesicle tracers, such as HRP or FM 1–43, will be necessary to determine whether any of these large vesicles are part of the synaptic vesicle cycle.
We propose a model (Fig. 10) in which one function of synaptotagmin is to stabilize vesicles in the docked state. This could be accomplished by one or more of the following mechanisms: (1) by increasing the recruitment of vesicles from the cytoplasm to docking sites (positive recruitment), (2) by preventing the vesicle from dissociating from docking sites (retention), and (3) by preventing these vesicles from fusing before stimulated release (negative regulator of spontaneous release). Because the reduction in morphologically docked vesicles was independent of the rate of spontaneous fusions, synaptotagmin appears to preserve the docked state directly. In addition, the increase in spontaneous release suggests that those vesicles that dock in syt mutants have an increased probability of spontaneously “fusing.” A deficit in docking stability, as proposed in our model, can account for both of the electrophysiological defects seen in syt mutants: decreased EJP size and increased rate of spontaneous release.
The dramatic decrease that we observed in vesicles immediately adjacent to the presynaptic membrane contrasts with the increase seen when synaptotagmin-based peptides were injected into squid terminals (Bommert et al., 1993). This discrepancy may be attributable to several causes. The intracellular actions of the peptides are uncertain at present; they may mimic synaptotagmin and promote docking, whereas the mutations, which disrupt synaptotagmin, reduce docking. Alternatively, the discrepancy may arise from the difference between blocking a portion of the protein and deleting it entirely, particularly if distinct domains of the protein mediate distinct functions in the vesicle cycle. For example, the acutely applied peptide acts on vesicles that have assembled correctly and contain synaptotagmin, whereas the mutations reveal that normal vesicle number and distribution require synaptotagmin.
The recruitment of vesicles to release sites by synaptotagmin may be Ca2+-dependent, consistent with the observed Ca2+-dependent translocation of a synaptotagmin C2 domain to the membrane (Chapman and Jahn, 1994) and the role of C2 domains in translocation of protein kinase C and phospholipase A2 to membranes. Additional effects of Ca2+ binding to synaptotagmin may act downstream of docking and promote membrane fusion. Analyzing more mutations may help to determine which biochemical interactions with synaptic proteins accomplish the stabilization of the docked state.
This work was supported by a Silvio Conte Center for Neuroscience Award from the National Institute of Mental Health (T.L.S.) and two grants from the Muscular Dystrophy Association (T.L.S. and N.E.R.). We are grateful to Drs. M. Ramaswami and E. Buchner for antibodies to CSP and to Ms. Fran Thomas for technical assistance.
Correspondence should be addressed to Dr. Thomas L. Schwarz at the above address.