Synaptotagmins are trafficked to distinct subcellular domains including the postsynaptic compartment

Neurotransmitter release is tightly regulated by intracellular calcium levels and requires SNARE complex assembly and disassembly. The search for calcium receptors that regulate SNARE-dependent fusion has focused on the synaptotagmins, a family of transmembrane proteins containing tandem calcium-binding C2 domains (for review see Jahn et al., 2003; Koh and Bellen, 2003; Yoshihara et al., 2003). Synaptotagmin 1 (Syt 1) was identified as an abundant synaptic vesicle integral membrane protein with calcium-dependent phospholipid binding properties (Perin et al., 1990). Genetic studies in Drosophila melanogaster and mice have demonstrated that loss of Syt 1 specifically eliminates the fast synchronous component of release, without removing the slow asynchronous component (Geppert et al., 1994; Yoshihara and Littleton, 2002). Mutations in syt 1 also disrupt the fourth order calcium dependence of synchronous fusion, suggesting Syt 1 functions as the presynaptic calcium sensor for fast synchronous release (Littleton et al., 1994; Fernàndez-Chacòn et al., 2001; Yoshihara and Littleton, 2002; Stevens and Sullivan, 2003).

Apart from Syt 1, more than a dozen additional synaptotagmins have been identified in mammals (Südhof, 2002), whereas the Caenorhabditis elegans and Drosophila melanogaster genomes encode eight and seven synaptotagmin genes, respectively (Lloyd et al., 2000; Adolfsen and Littleton, 2001). Several observations suggest that different synaptotagmin isoforms might cooperate to regulate the same exocytotic process, including dense core vesicle fusion in PC12 cells (Saegusa et al., 2002; Tucker et al., 2003). Heterooligomerization of distinct synaptotagmins has also been hypothesized to regulate the calcium sensitivity of neurotransmitter release (Littleton et al., 1999; Desai et al., 2000; Wang et al., 2001). Alternatively, each synaptotagmin isoform may participate in distinct membrane trafficking pathways. Supporting this model, several synaptotagmin isoforms do not colocalize with Syt 1 (Butz et al., 1999; Martinez et al., 2000; Ibata et al., 2002). To investigate the possibility that other synaptotagmins are involved in regulating neurotransmitter release, we characterized the seven synaptotagmins encoded in the Drosophila genome. We find that synaptotagmin isoforms localize to nonoverlapping subcellular compartments, suggesting that they participate in the regulation of distinct membrane trafficking steps in vivo.

Taking advantage of the recently completed Drosophila genome, putative synaptotagmin genes have been identified using BLAST analysis with known mammalian synaptotagmin isoforms (Adams et al., 2000; Lloyd et al., 2000; Adolfsen and Littleton, 2001). Seven synaptotagmin isoforms are present in the fly genome and show a conserved domain structure consisting of an NH₂-terminal transmembrane domain followed by tandem C2 domains. A comparison of the amino acid sequence encompassing the negatively charged residues important for calcium coordination within each C2 domain is shown in Fig. 1 A. Only the Syt 1 and Syt 7 isoforms encode all the coordination residues for both C2 domains. Three of the remaining isoforms (Syt 4, Syt α, and Syt β) display at least 60% conservation of these charged residues, while two isoforms (Syt 12 and Syt 14) show significant divergence (Fig. 1 A), suggesting that the function of some synaptotagmins may not require calcium binding.

To determine the relationship between Drosophila and other metazoan synaptotagmin isoforms, we performed a cluster analysis of the predicted synaptotagmin proteins encoded in currently sequenced genomes. Synaptotagmin sequences were collected from the C. elegans (C. elegans Sequencing Consortium, 1998), Anopheles gambiae (Holt et al., 2002), Fugu rubripes (Aparicio et al., 2002), Mus musculus (Waterston et al., 2002), and Homo sapiens (Lander et al., 2001) genomes and aligned using ClustalW analysis software. Our analysis suggests the synaptotagmin superfamily can be divided into eight subfamilies based on sequence relationships across species (Fig. 1 B). The Syt 1, Syt 4, Syt 7, Syt 12, and Syt 14 subfamilies contain at least one Drosophila member and one or more mammalian homologues. Isoforms of the Syt 1 and Syt 4 families were identified in all vertebrate and invertebrate genomes, suggesting that these two synaptotagmin families mediate an evolutionarily conserved function required in all animals. The Syt 7, Syt 12, and Syt 14 subfamilies contain Drosophila and vertebrate members, but lack homologues in other invertebrate genomes. Similar to the Drosophila homologues, the mammalian 12 and 14 isoforms lack the majority of consensus calcium binding aspartate residues, whereas Syt 7 contains highly conserved calcium binding sites. The three remaining synaptotagmin subfamilies are not highly conserved across evolution. The Syt 3 family consists of only vertebrate members, including the mammalian 3, 5, 6, and 10 isoforms. In contrast to the Syt 3 family, the Syt α and Syt β subfamiles do not contain any obvious vertebrate orthologues.

To characterize the expression profile of the Drosophila synaptotagmin family, we assayed their mRNA abundance and localization. The abundance of mRNA transcripts and their temporal expression in embryos was determined from developmental microarray expression experiments performed by the Berkeley Drosophila Genome Project. Embryonic mRNA was isolated at 1-h windows throughout the first 12 h of development and used to probe Affymetrix Drosophila genome arrays that include all seven Drosophila synaptotagmin isoforms (Fig. 2 A). The onset of expression of syt 1 coincides with formation of the nervous system. Similarly, none of the remaining synaptotagmins show a peak of expression before 11 h AEL, making it unlikely that they function at earlier stages of development. syt 4, syt 7, and syt β mRNA showed a similar developmental regulation, with increased expression from 10 to 12 h during nervous system development. The mRNAs for syt 12, syt 14, and syt α were expressed at very low levels throughout embryonic development.

To compare the expression levels between adults and embryos, we performed quantitative microarray analysis using Affymetrix genome arrays and mRNAs isolated from whole animals or from heads only (greatly enriching for neuronal transcripts). syt 1 was the most abundant synaptotagmin transcript, enriched in heads (Fig. 2 B). syt 4 and syt 7 mRNAs were also relatively abundant, whereas the remaining synaptotagmin mRNAs were expressed at low levels, suggesting low abundance or restricted patterns of expression. None of the synaptotagmins showed increased expression in whole animal versus head extracts, suggesting that preferential enrichment of the isoforms in nonneuronal tissue is unlikely.

To identify the expression patterns for the synaptotagmin family, we performed RNA in situ hybridization experiments on 0–22-h embryos using RNA probes specific to each isoform (Fig. 2, C–H). Similar to syt 1, syt 4 and syt 7 mRNAs were abundantly expressed throughout the central nervous system (CNS; Fig. 2, D and E). In addition to CNS staining, syt 7 mRNA was expressed in several tissues outside the nervous system, indicating a more ubiquitous expression pattern. syt 14 was expressed at low levels in the CNS (Fig. 2 F). syt α and syt β displayed a highly restricted expression pattern in subsets of CNS cells. syt β was expressed in a few bilaterally symmetrical large cells found in each segment of the ventral nerve cord (VNC; Fig. 2 G). Expression was also detected in peritracheal cells and within a small population of neurons in each brain lobe. The syt α isoform was found in a distinct population of smaller cells within each VNC segment, and in a subset of neurons within each brain lobe (Fig. 2 H). syt 12 mRNA was not detected by microarray or in situ analysis, suggesting it is expressed at levels below detection. Together with the microarray analysis, our data indicate that Syt 1 and Syt 4 are expressed in most, if not all, neurons. Syt 7 is also abundantly expressed, but in a ubiquitous pattern both within and outside of the nervous system. The remaining synaptotagmins display restricted expression in the nervous system, labeling only specific subpopulations of cells.

We had previously generated antisera to Drosophila Syt 4 and found the antisera recognized an antigen copurifying with synaptic vesicles and coIPing with Syt 1, leading us to conclude that Syt 4 was a synaptic vesicle protein (Littleton et al., 1999). The lack of a mutant in syt 4 prevented us from confirming the antibody was isoform specific. It is now clear that the original Syt 4 antisera is not isoform specific, as the antigen detected by the antisera is not removed in animals lacking the syt 4 locus, resulting in cross reactivity with the synaptic vesicle-localized Syt 1 protein. Therefore, to define the subcellular localization of the Drosophila synaptotagmins, we generated isoform-specific antisera to each synaptotagmin using multiple host animals and performed control experiments to confirm their specificity. The reactivity of the purified antisera to Drosophila synaptotagmins is shown in Fig. 3 A. The synaptotagmin antisera uniquely recognize their respective recombinant proteins and do not cross-react with other isoforms. In addition, preincubation of the antisera with excess recombinant antigen abolished the signal obtained on Westerns (Fig. 3 E) and immunostaining (Fig. S1, available). Further confirmation that the anti–Syt 7 antibody is specific was obtained by generating UAS-syt 7 transgenic animals and overexpressing the protein using elav-GAL4. Overexpression of Syt 7 resulted in up-regulation of the signal detected by anti–Syt 7 antisera. The most definitive confirmation of isoform specificity is to demonstrate that immunoreactivity is lost in mutant animals. This has been determined for our antisera to Syt 1 and Syt 4, proving that these antisera display isoform specificity (Fig. 3 D and Fig. 4 B). We have not yet generated mutations in the remaining synaptotagmins, so their specificity is still tentative using these rigorous requirements. However, as shown in Fig. 6 and Fig. 7, the localization patterns for each isoform are unique and correspond with in situ results.

To characterize the compartmental localization of synaptotagmins, we determined their subcellular distribution on gradients prepared from Drosophila brain homogenates. We performed velocity gradient subcellular fractionation experiments using 10–30% sucrose gradients to separate Canton-S head extracts. Fractions containing plasma membrane, synaptic vesicle, and cytosol compartments were identified using known markers (Syntaxin 1A [Schulze et al., 1995], n-Synaptobrevin [DiAntonio et al., 1993], and ROP ROP [Salzberg et al., 1993]). As shown in Fig. 3 B, Syt 1 comigrates with other synaptic vesicle proteins such as n-Synaptobrevin. In contrast, Syt 4 and Syt 7 did not comigrate with Syt 1 or plasma membrane fractions, suggesting that they reside in compartments that are biochemically separate from Syt 1–containing synaptic vesicles. Syt α was primarily enriched in fractions containing plasma membrane proteins, and was absent from the synaptic vesicle fraction. Antibodies against Syt 14, Syt 12, and Syt β did not give a signal from brain extracts, suggesting low expression. To confirm that Syt 1–containing synaptic vesicles can be separated from other synaptotagmin compartments, we performed equilibrium density gradient centrifugation experiments using a 26% self-forming Optiprep gradient. As with velocity gradients, the Syt 1 compartment was clearly separable from the remaining synaptotagmins (Fig. 3 C). We conclude from these experiments that the remaining synaptotagmins are not present on synaptic vesicles in vivo, indicating Syt 1 is the only synaptic vesicle isoform in Drosophila.

To characterize the subcellular distribution of the synaptotagmins, we examined their localization in Drosophila embryos and larvae using immunocytochemistry. Syt 1 has been previously localized to synaptic vesicles at presynaptic terminals (Littleton et al., 1993). Similar to Syt 1 and consistent with our in situ localization data, the Syt 4 protein was found concentrated in the neuropil of the larval CNS (Fig. 4 A), suggesting localization to mature synapses. This immunostaining is abolished in mutants that remove the syt 4 locus (Fig. 4 B). During embryonic development, the subcellular localization of Syt 4 is clearly distinct from Syt 1. As shown in Fig. 4 C, Syt 4 is abundant in neuronal cell bodies in the developing CNS at a time in which Syt 1 and other axonal markers such as Fas II have already trafficked to axons, indicating Syt 4 is differentially sorted during the establishment of neuronal polarity. The consequences of this differential sorting are apparent at mature third instar larval neuromuscular junctions (NMJs), where Syt 4, in contrast to Syt 1, is found postsynaptically (Fig. 4, D–I). Syt 4 antiserum labels the postsynaptic side of NMJs (Fig. 5, A and B), surrounding the outside of presynaptic terminals (labeled by anti-HRP) in a punctate pattern, suggesting Syt 4 resides in a postsynaptic vesicular compartment. Double labeling experiments performed in animals overexpressing a myc-tagged postsynaptic glutamate receptor subunit, mycGluRIIA, reveals that Syt 4 localizes to regions adjacent to postsynaptic receptor clusters (Fig. 5, C and D). Costaining with Syt 1 antisera demonstrates no overlap in the distribution of the two proteins, confirming Syt 4 is not a synaptic vesicle protein. In addition, overexpression of Syt 4 presynaptically in syt 4 mutant animals does not shift its localization to synaptic vesicles (Fig. 5, E and F). Immunostaining of Syt 4 and Syt 1 revealed a nonoverlapping pattern of expression, with Syt 4 excluded from Syt 1–positive synaptic vesicle microdomains (Fig. 5 F). Although there is no overlap between Syt 4 and Syt 1 staining (Fig. 5, E and F), and the majority of Syt 4 labeling is postsynaptic (Fig. 5, A and B), a smaller fraction of Syt 4 may also localize presynaptically. However, synaptic defects present in syt 4 mutants (unpublished data) are rescued by postsynaptic expression, suggesting that Syt 4 functions in postsynaptic trafficking steps required for synaptic growth and plasticity.

Similar to Syt 4, the Syt 7 isoform was not found in axonal tracts during embryonic neuronal development, but instead localized to neuronal cell bodies (Fig. 6 A). This segregation of the synaptotagmins was maintained in mature third instar larvae. Unlike Syt 1 or Syt 4, we could not detect Syt 7 at NMJ synapses, but rather in a distinct vesicular compartment that was present not only in muscles (Fig. 6, B and C) but in other tissues, including imaginal discs (Fig. 6 D). In muscle, anti–Syt 7 staining was observed in small clusters throughout the sarcoplasm. The remaining synaptotagmin isoforms were expressed at low abundance compared with Syt 1, Syt 4, and Syt 7. Consistent with our in situ and microarray analysis, antiserum to Syt 12 and Syt 14 revealed no staining for these two isoforms in either embryos or at mature synapses in third instar larva. Syt α and Syt β were detected in subsets of neurons in the CNS and periphery that corresponded with their in situ expression patterns. In mature embryos, the Syt α protein was found in a small population of bilaterally symmetrical VNC neurons (Fig. 7 B). In third instar larvae, Syt α immunoreactivity was observed in the mushroom body and in several large CNS cell bodies (Fig. 7 A). Subsets of synaptic tracts that innervate the ventral ganglion and several ventral ganglion cell bodies were also labeled. Syt α was not detected at any peripheral motor synapses, but rather localized to the neurosecretory lateral bipolar dendritic (LBD) neuron within each abdominal segment of the larva (Fig. 7 C). Specific localization in the LBD neuron and a CNS localization pattern similar to that observed for known neuropeptides suggest Syt α may function in trafficking of specific subclasses of neuropeptides and/or neuromodulators.

Similar to Syt α, Syt β was detected in a restricted population of cells. As observed with in situ hybridization experiments, Syt β was found in peritracheal cells that surround tracheal branchpoints in embryos (Fig. 7 F) and larvae. Syt β was also detected in several synaptic tracts and large cell bodies in the ventral ganglion. In the larval brain, Syt β immunoreactivity was absent from the mushroom bodies and instead concentrated in a pair of bilaterally symmetric cell bodies in the brain lobes that innervated the ventral ganglion. At peripheral NMJs, Syt β is present at synapses of a single motorneuron that innervates muscle fiber 8 and that release the neuropeptide leukokinin (Cantera and Nassel, 1992). In summary, our findings indicate that only the Syt 1 and Syt 4 isoforms are ubiquitously present at synapses, whereas the remaining isoforms were not detected at synapses (Syt 7), expressed at very low levels (Syt 12 and Syt 14), or in subsets of putative neurosecretory cells (Syt α and Syt β).

The localization of synaptotagmin isoforms to distinct subcellular compartments suggests that they function in unique trafficking pathways. To test this hypothesis, we examined if Syt 4 and Syt 7, which are coexpressed in neurons with Syt 1, could rescue the synaptic transmission and behavioral defects of syt 1 mutants when overexpressed in the nervous system. We overexpressed Syt 4 in the syt 1 null background (syt^AD4 in trans to Df(2L)N13) using C155^elav-GAL4 (Fig. 8, A and B). To obtain quantitative information on the behavioral rescue, we performed larval locomotion assays to examine the output of the central motor pattern generator. Representative traces of crawling patterns from the control line C155^elav-GAL4 and the syt 1 null mutant (syt^AD4/Df(2L)N13) are shown in Fig. 8 C. In contrast to the robust locomotion observed in control animals, the lack of Syt 1 dramatically slows larval locomotion. In addition to a decrease in distance traveled and locomotor cycle number (Fig. 8 D), syt 1 null mutants display an increase in the duration of a single locomotor cycle from 1 s to ∼6 s (Fig. 8 E). Transgenic expression of the syt 1 gene in the null background was able to partially restore all the behavioral defects observed (Fig. 8, C–E). In contrast to Syt 1, Syt 4 (Fig. 8, C–E) or Syt 7 (not depicted) overexpression did not rescue any aspect of the behavioral defects.

To directly examine synaptic transmission, we performed synaptic physiology at the third instar NMJ (Fig. 8, F–H). Consistent with previous observations, synaptic transmission was severely decreased in syt 1 null mutants, which showed the characteristic slow rise and decay reflecting asynchronous release and a loss of the synchronous component of fusion. Overexpression of Syt 1 was able to restore evoked excitatory junctional potential (EJP) amplitudes to near wild-type levels. Similar to the lack of behavioral rescue, overexpression of Syt 4 or Syt 7 in the syt 1 null mutant had no effect on the synaptic transmission defects. Only slow release reflecting the asynchronous component of fusion was observed in syt 1 null mutants overexpressing Syt 4 or Syt 7, and evoked EJP amplitude was unchanged from the null mutant alone (Fig. 8 F). Our results are in disagreement with a recent work indicating that overexpression of Syt 4 using our UAS-syt 4 construct can fully rescue synaptic transmission defects in syt 1 mutants in Drosophila (Robinson et al., 2002). Their results do not fit with our observation that Syt 4 is absent from synaptic vesicles and primarily localizes postsynaptically. In addition, their results seem contradictory to the finding that the syt 4 gene is coexpressed in all neurons with syt 1, yet there is a complete absence of the calcium-dependent synchronous component of release when Syt 1 is removed (Yoshihara and Littleton, 2002). To further examine these differences, we repeated our rescue experiments in saline containing 5.0 mM of extracellular calcium as previously reported (Robinson et al., 2002), but observed no rescue of the syt 1 null phenotype by UAS-syt 4 (Fig. 8 G). We also examined the possibility that the second chromosome elav-GAL4 driver used in their previous study was generating stronger expression of Syt 4 than the X chromosome C155^elav-GAL4 driver used in our rescue experiments. We generated animals containing both the X chromosome C155^elav-GAL4 and the previously used second chromosome elav-GAL4 to test if increased expression of Syt 4 would rescue syt 1 mutants. We found no rescue of the syt 1 null mutant by increased Syt 4 expression (Fig. 8 G). Instead, we observed that the increased expression of Syt 4 resulted in decreased viability. Compared with syt 1 null mutants alone that survive to the third instar stage, there is a 96% reduction in the expected Mendelian ratios when the UAS-syt 4 transgene is driven by both the X chromosome elav-GAL4 and second chromosome elav-GAL4 drivers. In summary, our behavioral and physiological data indicate that Syt 4 and Syt 7 cannot functionally substitute for Syt 1 when overexpressed, indicating synaptotagmins define unique membrane trafficking pathways within neurons.

Genetic analysis has demonstrated that Syt 1 is essential for calcium-dependent synchronous release, underlying the fourth order cooperativity of synaptic vesicle fusion, but does not abolish asynchronous calcium-dependent release (Geppert et al., 1994; Yoshihara and Littleton, 2002; Stevens and Sullivan, 2003). These observations are consistent with the current two calcium sensor model for synaptic transmission (Yamada and Zucker, 1992), with Syt 1 functioning as the calcium sensor regulating the fast synchronous component of release and an unidentified calcium sensor mediating the slow asynchronous component. Other synaptotagmin isoforms are obvious candidates for the asynchronous calcium sensor. In addition, synaptotagmins have unique calcium-binding properties (Sugita et al., 2002) and undergo heterooligomerization in vitro (Littleton et al., 1999; Desai et al., 2000). Several plasticity models have been proposed, suggesting differential expression of synaptotagmin isoforms on synaptic vesicles might regulate presynaptic release probability (Littleton et al., 1999; Wang et al., 2001) or transitions from full fusion to kiss-and-run (Wang et al., 2003). These hypotheses require that synaptotagmins have a similar expression pattern to Syt 1 and localize presynaptically at synaptic terminals. We have addressed these hypotheses in vivo by performing an extensive expression and localization study of the entire synaptotagmin family in D. melanogaster. Our localization data argue against the possibility that other synaptotagmin isoforms function with Syt 1 to regulate neurotransmitter release. Instead, the remaining synaptotagmin isoforms likely regulate distinct membrane trafficking steps in vivo.

Syt 4 was found in the postsynaptic compartment, suggesting it regulates a postsynaptic membrane trafficking pathway. We cannot rule out that a small fraction of Syt 4 may also be present in some presynaptic compartments, though it does not localize to Syt 1–positive synaptic vesicles. The detection of the Syt 4 protein by Western analysis and immunocytochemistry with our new antisera is abolished in syt 4 null mutants, confirming the antisera accurately reflects the subcellular localization of Syt 4. These results indicate that previous detection of Syt 4 on synaptic vesicles (Littleton et al., 1999) reflected cross-reactivity of the old antisera with Syt 1. Given that Syt 4 does not colocalize on Syt 1–positive synaptic vesicles, the reduction of neurotransmitter release by Syt 4 up-regulation observed in Drosophila (Littleton et al., 1999) is unlikely to be due to heteromultimerization of the two proteins on vesicles and may instead reflect competitive binding to Syt 1 effectors or altered presynaptic calcium buffering.

In terms of Syt 4's postsynaptic localization, there is evidence in several experimental systems for a regulated form of postsynaptic vesicular trafficking (Ludwig et al., 2002). Studies in hippocampal culture neurons indicate that long-term labeling with FM1–43 loads dendritic organelles that undergo rapid calcium-triggered exocytosis that is blocked by tetanus toxin (Maletic-Savatic and Malinow, 1998). In addition, pharmacological blockage of postsynaptic membrane fusion reduces LTP (Lledo et al., 1998), suggesting postsynaptic vesicle trafficking contributes to synaptic plasticity. Mammalian Syt 4 has been localized within dendrites and soma (Ibata et al., 2002), suggesting Syt 4 and the related homologue Syt 11 may also function postsynaptically. Although the exact role for regulated postsynaptic fusion remains unclear, possibilities include the release of retrograde signals, trafficking of postsynaptic receptors, and/or trafficking of synaptic cell adhesion proteins.

The remaining synaptotagmins were not ubiquitously localized to synapses. Unlike Syt 1 or Syt 4, we could not detect Syt 7 at synapses, but found it was expressed in both neuronal and nonneuronal tissues. Mammalian Syt 7 has been found in secretory lysosomes (Martinez et al., 2000) and in synaptic active zones where it has been postulated to function as a plasma membrane calcium sensor (Sugita et al., 2001). Genetic studies of Syt 7 will be required to determine if it also functions at Drosophila active zones. Peripheral Syt β staining was restricted to muscle fiber 8 synapses that are known to release the neuropeptide leukokinin (Cantera and Nassel, 1992). In the CNS, Syt β was observed in a pair of bilateral neurons that may be the DPM neurosecretory neurons known to secrete the amnesiac neuropeptide. The only staining outside the nervous system is detected at tracheal branch points, where a group of myomodulin-releasing neurosecretory cells are located (O'Brien and Taghert, 1998). These localization studies suggest Syt β is a candidate calcium sensor for mediating dense core vesicle fusion and release of neuropeptides. Similar to Syt β, Syt α showed specific expression in another set of putative CNS neuropeptide-releasing neurons, as well as within the mushroom bodies. In the periphery, staining was restricted to the LBD neurosecretory neuron, which is consistent with a role in neuropeptide release. In addition, the localization of Syt α in mushroom bodies and the possible localization of Syt β in DPM neurons makes these isoforms attractive candidates for potential roles in vesicular trafficking pathways contributing to neuronal plasticity. We were unable to localize the two remaining synaptotagmins, Syt 12 and Syt 14. It is likely that the proteins are below the detection level of our antisera, which is consistent with the microarray and in situ experiments, indicating that these isoforms are expressed at low levels in embryos and adults. Unlike the other synaptotagmins, these two isoforms lack most of the calcium coordination residues in C2A and C2B in both vertebrates and flies, indicating that they may function in trafficking pathways not regulated by calcium.

In summary, Drosophila synaptotagmin isoforms identify unique membrane-trafficking compartments. A summary of the expression of both the mRNA and protein for each synaptotagmin family member is shown in Fig. 9. Our data indicate that only the Syt 1 isoform is found on synaptic vesicles and so argue against heterooligomerization models. In addition, we find that Syt 4 and Syt 7 cannot rescue the behavioral or physiological defects in syt 1 mutants, suggesting that synaptotagmins define unique membrane trafficking pathways within neurons. It is possible synaptotagmins function in an analogous manner to control vesicle fusion, but do so in distinct compartments. Given that Syt 4 localizes to the postsynaptic compartment, our findings indicate that calcium-dependent membrane trafficking occurs on both sides of the synapse.

Drosophila were cultured on standard medium at 22°C. A null mutant in syt 4, syt 4^BA1, was generated by imprecise P-element excision of EY09259, an insertion located 100 bp 5′ of the syt 4 transcription start site. syt 4^BA1 is an intragenic deletion that removes Syt 4 immunoreactivity. UAS-syt 7 transgenic animals were obtained by subcloning a syt 7 cDNA into PUAST and generating transgenic animals via standard techniques.

Synaptotagmin protein sequences were collected from Drosophila, C. elegans, A. gambiae, F. rubripes, M. musculus, and H. sapiens genomes. Sequences were identified by BLAST analysis of Drosophila synaptotagmin protein sequences against the corresponding genomes deposited in GenBank. Collected sequences were then clustered based on homology using the ClustalW program (http://www.ch.embnet.org/software/ClustalW.html). Results were displayed as a tree diagram using the Phylodendron program (http://iubio.bio.indiana.edu/treeapp/treeprint-form.html).

Embryos aged 0–22 h were collected and processed according to standard procedures. Probes of 500–700 bp long were designed to the C2 domain region of each synaptotagmin gene.

Microarrays were performed with Affymetrix Drosophila Genechips using biotinylated cRNA using the laboratory methods described in the Affymetrix genechip expression manual (Affymetrix, Inc.). RNA was isolated from heads or heads and bodies of Canton-S males aged 3–4 d after eclosion at RT. All flies were killed between 12 and 2 p.m. to reduce any circadian-dependent transcriptional changes. Affymetrix high-density oligonucleotide arrays were probed, hybridized, stained, and washed in MIT's Biopolymers Facility according to the manufacturer's instructions. Microarray analysis was performed using Microarray Suite Vs.5 and Data Mining Tool Vs.3 statistics-based analysis software (Affymetrix, Inc.).

Western blots were done using standard laboratory procedures. All synaptotagmin antibodies were used at a 1:1,000 dilution and detected using a goat anti–rabbit antibody conjugated to HRP (Jackson ImmunoResearch Laboratories). Visualization of HRP was accomplished using a SuperSignal ECL kit (Pierce Chemical Co.).

Isolation of Canton-S head homogenates was performed as described previously (Littleton et al., 1999). For rate-zonal sedimentation experiments, a post-nuclear extract was layered onto a 10–30% sucrose gradient and centrifuged at 50,000 RPM for 1 h in a NVT65 rotor (Beckman Coulter). 1-ml fractions were collected beginning from the bottom of the gradient and proceeding to the top. After collection, fractions were mixed with an equal volume of 2× SDS-PAGE loading buffer and probed by Western analysis. For equilibrium sedimentation experiments, post-nuclear extract was combined with a 26% Optiprep (Axis Shield) solution. The mixed sample was centrifuged at 60,000 RPM for 3.5 h in a NVT65 rotor and fractions were collected as for velocity experiments.

Polyclonal antibodies were generated in rabbits (Invitrogen). For the Syt 4, Syt 7, Syt α, and Syt β isoforms, we generated antisera to recombinant proteins encompassing the C2 domains of each protein. For Syt 14, we prepared antisera to a recombinant protein that encompassed the linker between the TM domain and C2A. For Syt 12, we generated antisera against a peptide derived from the linker domain between C2A and C2B. Each respective sequence was cloned into pGEX vectors. Recombinant GST fusion proteins were expressed and processed in E. coli (BL21) according to standard laboratory protocols. The fusion proteins were purified in batch using glutathione-sepharose (Amersham Biosciences). To remove the GST affinity tag, protein samples were incubated with thrombin for 1 h at RT. Antisera was purified using affinity chromatography. The domain of each synaptotagmin was coupled to a 1-ml NHS-activated sepharose column (Amersham Biosciences). Antisera (2 ml) injection, subsequent washes, and elution from the columns were all performed on an AKTA FPLC (Amersham Biosciences). Columns were washed in 20 mM sodium phosphate and eluted with 0.1 M glycine, pH 2.7. To minimize denaturation of the antibody at low pH, the eluted fractions were immediately mixed with 1 M Tris, pH 9. Fractions containing the desired peak were concentrated using Amicon ultra centrifugal filter devices (Millipore), aliquoted, and stored at −80°C.

Embryos and larvae were immunostained as described previously (Yoshihara and Littleton, 2002; Rieckhof et al., 2003). The dilution of primary antibodies was as follows: Syt 1 (1:1,000) Syt 4 (1:500), Syt 7 (1:1,000), Syt α (1:2,000), and Syt β (1:500). To decrease background, antibodies were preabsorbed to 0–11-h embryos. Samples were washed and mounted in 70% glycerol. Cy2-conjugated goat α–rabbit secondary antibodies (Jackson ImmunoResearch Laboratories) were used at 1:200. Visualization was performed under light microscopy using a 40× oil-immersion lens. Images were taken using confocal microscopy on a microscope (model Axoplan 2; Carl Zeiss MicroImaging, Inc.) and processed with PASCAL software (Carl Zeiss MicroImaging, Inc.).

Electrophysiological analysis of wandering stage third instar larva was performed in Drosophila saline (70 mM NaCl, 5 mM KCl, 4 mM MgCl₂, 10 mM NaHCO₃, 5 mM Trehalose, 115 mM sucrose, and 5 mM Hepes-NaOH, pH 7.2) supplemented with either 1.5 mM or 5.0 mM CaCl₂ using an Axoclamp 2B amplifier (Axon Instruments, Inc.) at 22°C as described previously (Rieckhof et al., 2003).

To quantify larval locomotion, late third instar larvae grown at 25°C were collected and placed on a flat layer of 2.9% agar supplemented with grape juice. Quantification of larval locomotion parameters was performed as described previously (Saraswati et al., 2004). For quantification of cycle duration, video recording of locomotion was performed using a digital video camera (model XL1S; Canon) attached to a 16× zoom lens (field of 3 cm²). Cycle duration was reconstructed offline by digitizing frame-by-frame locomotor contractions.

The specificity of the Syt α and Syt β antibodies for immunocytochemistry is shown in Fig. S1.

We thank Mario Mikula for help with Syt 4 antisera production, Enrico Montana and Zhuo Guan for providing microarray results, and Avital Rodal and Enrico Montana for helpful discussions about the manuscript. The 8C3 antibody developed by Seymour Benzer was obtained from the Developmental Studies Hybridoma Bank under the auspices of the National Institute of Child Health and Human Development and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242.

This work was supported by grants from the National Institutes of Health, the Human Frontiers Science Program Organization, the Packard Foundation, and the Searle Scholars Program. J.T. Littleton is an Alfred P. Sloan Research Fellow.