by neuronal specific modifications to an evolutionarily conserved vesicle trafficking system. manner (8, 9). In addition, synaptotagmin has been

Transcription

1 Proc. Natd. Acad. Sci. USA Vol. 91, pp , November 1994 Neurobiology Calcium dependence of neurotransmitter release and rate of spontaneous vesicle fusions are altered in Drosophila synaptotagmin mutants (Ca2/synaptic ved/ne roa tter relse/exocytosis) J. TROY LITTLETON*t, MICHAEL STERNt, MARK PERIN*, A HuGo J. BELLEN*t *Division of Neuroscience, and thoward Hughes Medical Institute and Department of Molecular and Human Genetics, Baylor College of Medicine, Houston, TX 77030; and tdepartment of Biochemistry and Cell Biology, Rice University, Houston, TX Communicated by Charles F. Stevens, July 18, 1994 ABSTRACT Since the demonstration that Ca influx into the presynaptic terminal is esential for neurotransmitter release, there has been much speculation about the Ca2 receptor responsible for i,niting ex ts. Numerous experiments have shown that the protin, or protein complex, binds multiple Ca2 ions, resides nea the ite of Ca2f influx, and has a relatively low affinity for Ca. Synaptotagmin is an integral membrane protein of synaptic vesicles that contains two copies of a domain known to be involved in Ca2-dependent membrane interactions. Synaptotagmin has been shown to bind Ca in vitro with a relatively low affinity. In addition, synaptogmin has been shown to bind indirectly to Ca can nels, positioning the protein close to the site ofca2 influx. Recently, a negative regulatory role for synaptotagmin has been proposed, in which it unctions as a damp to prevent fusdon of synaptic vesiles with the presynaptic membrane. Release of the damp would allow exocytosis. Here we present genetic and electrophysological evidence that synaptotagmin forms a multimeric complex that can fnction as a clamp in vivo. However, upon nerve stimulation and Cae infu, all synaptoagmin mutations dramatically decrease the ability of Ca2 to promote release, suggesting that synaptotagmin probably plays a key role In activation of synaptic vesicle fusion. This activity cannot simply be attributed to the removal of a barrier to secretion, as we cap electrophyslologically separate the increase in rate of spontaneous vesicle ftuonfrom the decrease in evoked response. We also find that some syt mutations, including those that lack the second Ca2-binding domain, decrease the fourthorder dependence of release on Ca2 by approximately half, consistent with the hypothesis that a synaptotagmin complex functions as 'a Ca2 receptor for initiating exocytosis. The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C solely to indicate this fact In 1965 Katz and Miledi (1) demonstrated that Ca2W influx into the presynaptic terminal triggered neurotransmitter release. The molecular mechanisms underlying the docking and Ca2 -activated fusion of neurotransmitter-containing synaptic vesicles with the presynaptic membrane have been intensively studied in recent years. It has become apparent that a substantial fraction of proteins involved in this process are conserved from yeast through human (for review, see ref. 2), suggesting a conserved mechanism for membrane trafficking. In addition to these conserved membrane trafficking proteins, several synapse-specific proteins, including synaptotagmin (3), have been implicated in neurotransmitter release. Homologues for synaptotagmin have now been identified in several invertebrate species such as Caenorhabditis elegans (4) and Drosophila (5-7) but have not been seen in lower eukaryotes like yeast. It is therefore possible that the speed and accuracy of fast synaptic transmission are mediated by neuronal specific modifications to an evolutionarily conserved vesicle trafficking system. Synaptotagmin is a synaptic vesicle-specific protein that contains two repeats implicated in Ca2-dependent membrane interactions (3). These domains have recently been shown to bind Ca2 and phospholipids in vitro in a cooperative manner (8, 9). In addition, synaptotagmin has been reported to interact with several other synaptic proteins (10-12), suggesting a role in docking and/or fusion of synaptic vesicles with the presynaptic membrane. Drosophila syt is expressed in all neurons, and antibodies raised against the protein show that it is localized specifically to synapses (6). Mutations in Drosophila syt were isolated by using chemical and transposon mutagenesis, and a preliminary analysis of the mutant phenotype associated with a partial loss-offunction allele of syt (syttli) indicated a slight increase in the spontaneous release of synaptic vesicles and a dramatic decrease in evoked release (7). To further characterize the role of synaptotagmin in vivo we have generated eight additional syt mutations and show that many syt alleles exhibit intragenic complementation. Analysis of the electrophysiological properties of nine different heteroallelic mutations provides evidence that synaptotagmin acts in a multimeric complex and is a key component of the Ca2-sensing mechanism activating neurotransmitter release. MATERIAL A METHODS Isolation and Rescue of Synaptotagmin Mutants. Alleles D2, D3, D37, D39, and D45 were generated with ethyl methanesulfonate as described (7). Alleles N10, N13, and N28 were generated by imprecise excision of the 177 P-element insertion in syt. The isolation of these alleles is described elsewhere (13). Absence of synaptotagmin protein was determined by immunocytochemical staining with antibody DSYT2 as described (6). Lethal phase was assessed as described (7). A transgenic strain carrying a syt cdna under control of the elav promoter and regulatory sequences (14) was from A. DiAntonio and T. Schwarz (Stanford University Medical Center) (5) and used for rescue of mutants in trans to a deficiency (N6) that completely removes the syt locus (13). Electrphysology at the Third-Instar Neurmuscular Junction. Dissections, nerve stimulation, and muscle recordings were done as described (7). The Ca2 dependence of neurotransmitter release (n) was measured as described (15). Quantal values were corrected as described (16). cn bw sp is the parent chromosome for most mutant combinations examined and was used as control. Abbreviation: MEJP, miniature excitatory junctional potential. To whom reprint requests should be addressed.

2 Neurobiology: Littleton et al. RESULTS To test the role of synaptotagmin in neurotransmitter release, we have performed a genetic analysis of Drosophila syt. Table 1 shows that 19 different alleles of syt have now been isolated (refs. 5, 7, and 13 and this study). All but 6 of the 19 alleles can be partially rescued as homozygotes by using a transgene that expresses syt under the control of a neuralspecific promoter (5). In addition, five of the six remaining syt mutations can be partially rescued over a chromosome carrying a deficiency that removes the syt locus. In addition, all but the TI I allele are homozygous lethal, and all syt alleles fail to complement the lethality of null alleles like N13 (see Table 2). These data provide evidence that all phenotypes discussed below are likely due to mutations in the syt gene. A quantitative analysis of the hatching frequency of each syt mutation in trans to a deficiency that completely removes the syt gene is shown in Table 1. Many syt mutations cause homozygous embryonic or first-instar lethality, indicating that syt is required for hatching and locomotion of first-instar larvae. Table 2 shows that complementation tests between all syt alleles reveal that approximately a quarter of the heteroallelic genotypes can sometimes survive to adulthood. Some of these flies may survive because certain partial lack-of-function alleles, like Ti ), behave as dominant alleles in combination with other syt alleles. However, based on the data of Tables 1 and 2, we conclude that several severe lack-of-function syt alleles that are homozygous lethal (e.g., D3, D37, AD2, and others) complement each other partially. This form of intragenic complementation is typically observed in genes that encode proteins that are part of multimeric complexes, such that mutations in one domain of the protein may partially compensate for mutations in another domain, alleviating the mutant phenotype (17). These observations also suggest the presence of several independent domains within synaptotagmin that may play different roles in neurotransmitter release. Table 2. Table 1. Allele N6* T7* T11* T41* 66.4* D2t D3t D37t D39t D45t AD1t AD2* AD3t AD4O D27t T77* NlOt N13t N28t Rescue in trans to Summary of synaptotagmin mutations Rescue as Embryonic homo- Protein lethality, zygous deficiency Proc. Natl. Acad. Sci. USA 91 (1994) /- /- /-, hypomorph;, not determined. *Ref. 7. tthis study. tref Classification Null Weak Strong /null Strong Strong /null Strong /null Strong /null Null Strong /null One of the hallmarks of intragenic complementation is that the two defective proteins encoded by different alleles rarely function as wild-type proteins (17). The data summarized in Table 2 show that none of the syt mutants fully complement each other with respect to the expected Mendelian ratios of adult progeny recovered. In addition, all recovered adult flies exhibit behavioral defects. Fig. 1 shows the results of our attempt to qualitatively evaluate the synaptotagmin phenotype from most abnormal to most normal on the basis of Heteroallelic complementation of synaptotagmin mutations N6 77 Ti) T D2 D3 D37 D39 D45 AD) AD2 AD3 AD4 Deletion N6 - EMS-induced mutations TI) - - _* T * - - D D D * D D AD) - /- - / AD2 - -* -* /- /- -* AD3 - - /- 1- /- /- - /- - / AD * Inversion D * - -* -* Insertion * * - - Regulatory mutations NJO * /- /- - N N * -* - - EMS, ethyl methanesulfonate. -, No adults observed in transheterozygote crosses; -*, occasional adult survivor observed (<1%); /-, 10-50o expected progeny.

3 10890 Neurobiology: Littleton et al. Abnormal i T11 Afl3 D37 T11 T11 T41 T77 T D2 D37 D45 N10 TJJ/T77) that cause a reduced response at all external [Ca2] AD1 AD2 AD2 Normal AD2 tested and, in addition, dramatically alter the sigmoidal nature of the response curve. These shifts indicate that the fl t t s t fl t protein complex responsible for fusion has a significantly T4 1 <1% 25% 70% lower responsiveness for Ca2 in the presence of mutant T7 T11 D2 D37 synaptotagmin proteins. In addition, the data show that a T77 D3 functional synaptotagmin complex is required for the major- N28 D45 ity of evoked neurotransmitter release in wild-type larvae. D27 T7 Dodge and Rahaminoff (18) first measured the Ca2 de- Ni10 pendence of release and found a fourth-order relationship between extracellular [Ca2] and the amplitude of evoked FIG. 1. Behavioral severit) yof viable trans-heterozygote syt flies. end plate potentials. Subsequent work at the squid giant The phenotype of viable fliess shown in Table 2 was qualitatively synapse demonstrated a third-order relationship between evaluated on a subjective sceale and graphically represented from intracellular ICa2] and the release process (19). It was most severe (at left) to least severe (at right). The following behav- concluded from these studies that three to four Ca2 ions are ioral criteria were used: gailt, jump response, flight ability, and required in cooperative action to evoke release. We have fertility. measured the Ca2 dependence ofneurotransmitter release in criteria such as ability to walk, fly, and mate. Flies that carry wild-type larva and found a third- to fourth-order dependence alleles Ti), AD3, D37, A Dl, as well as syt alleles shown of neurotransmitter release upon external Ca2 (15) (Fig. 2C below the line in Fig. 1 di: splay very severe uncoordination and Table 3, genotype cn bw sp). One class of syt mutations and die hr after ecloision. Mutant combinations carry- (Tll/AD3, T41/AD3, D37/AD3) shifts this curve to the right, ing the AD2 allele fall into several distinct categories. Some indicating a decrease in the quantal content for all [Ca2] but AD2 transheterozygotes (tthose with alleles Ti), 777, N28, does not significantly affect the slope (average n = 3.6). and AD)) rarely survive tc) adulthood (<1%), but the survi- Hence, neurotransmitter release still shows a third- to fourth- are unable to mate. Other alleles order [Ca2] dependence in these mutants. However, a vors can walk and jump but (D2, D3, D45, 77, T41, anid N10) in combination with AD2 second class of mutations (77/AD2, T41/AD2, D37/AD2, produce 30% of the expectced progeny, whereas AD2 and D37 77/ADI, T41/ADJ, Tl I /777) not only shifts the curve to the almost fully complement e; ach other with respect to viability, right but also decreases the slope and lowers the Ca2 Flies of the last two genotypic groups can walk and jump and dependence of neurotransmitter release to a first- to second- carrying anad2 allele are smaller order relationship. This cooperative Ca2 binding must be are partially fertile. All fliess in body size and show uricoordination, although most are linked to the function of the synaptotagmin multimer, as the capable of limited flight. Hence, the defects caused by same mutant synaptotagmin protein (e.g., Tll/AD2 versus various mutant combinatioins of syt alleles that produce adult Tll/AD3, T41/AD3 versus T41/ADI, T41/AD2) can particvere uncoordination and death to ipate with different synaptotagmin mutant proteins in a progeny vary from very se relatively subtle behavior al defects affecting flight and re- complex that either lowers the Ca2 cooperativity (n) or productive abilities. alters the sensitivity to extracellular Ca2 without affecting n. The ability to generate vaiable syt mutant flies allowed us to Interestingly, the shifts in n are not randomly distributed but characterize the defects c; aused by these mutations in vivo. are clustered around a value of 1.8, approximately half of the As shown in Table 3 and ] Fig. 2, larvae of nine heteroallelic normal value of 3.6. combinations of syt alleles were examined electrophysiolog- In a second approach to monitor release properties of ically for synaptic defectss at the third-instar larval neuro- synapses in mutant larvae, we have measured the spontane- syt mutant combinations ous fusion of synaptic vesicles [miniature excitatory junc- muscular junction. All heteeroallelic show a severely decreased ability to promote evoked release tional potentials (MEJPs)] in the absence of nerve stimulation at lower extracellular [Ca24], although the [Ca2] required for and Ca2 influx. As shown in Fig. 2D and Table 3, all the onset of an evoked re-sponse varies from 0.3 to 0.6 mm combinations of syt mutants show an increase in the fre- [Ca2], compared with mm [Ca2] in wild type. The quency of MEJPs over wild type, ranging from 2- to 5-fold. mutations can be classifi Wd into two broad categories: (i) In addition, one often sees two- to three-vesicle fusions those that cause a reduced response at all external [Ca2] occurring simultaneously in the absence of any action potested (TJl/AD3, T41/ADU3, D37/AD3) but do not seem to tential, a pattern not seen in wild-type larva. However, the affect the sigmoidal curve for Ca2-evoked release, and (ii) amplitude resulting from a single vesicle fusion is the same in those (17/AD2, T41/AD2?, D37/AD2, 77/ADI, T41/ADI, mutant and wild type, indicating that the mutant muscle Table 3. Electrophysiological properties of heteroallelic synaptotagmin mutations Proc. Nad. Acad. Sci. USA 91 (1994) Allele Evoked responses to Ca2 Mini- Regression combination 0.4 mm 1.0 mm 1.8 mm 6.0 mm frequency n coefficient cn bw sp 17.9 ± 2 (6) 23.5 ± 1.9 (4).6 ± 2.6 (5) 43.3 ± 4.2 (4) 1.3 ± TJJ/AD3 0.1 ± 0.1 (4) 7.0 ± 1.2 (3) 16.1 ± 2.1 (3) 25.0 ± 1.7 (3) 4.0 ± T41/AD3 0.4 ± 0.1 (3) 7.3 ± 0.3 (3) 18.6 ± 2.6 (3) 38.5 ± 2.4 (4) 3.7 ± D37/AD3 0.2 ± 0.1 (3) 4.7 ± 0.9 (3) 14.9 ± 2.9 (3) 28.2 ± 7.0 (4) 2.7 ± /AD2 0.6 ± 0.2 (6) 2.1 ± 0.5 (4) 4.8 ± 1.4 (5) 10.4 ± 0.7 (4) 4.8 ± T41/AD (5) 4.0 ± 0.9 (5) 5.9 ± 1.9 (5) 11.4 ± 1.6 (5) 4.8 ± D37/AD (3) 4.9 ± 0.7 (3) 7.2 ± 0.7 (3) 15.1 ± 2.8 (3) 5.9 ± T7/ADI 2.2 ± 0.7 (3) 2.5 ± 0.8 (3) 9.7 ± 2.2 (3) 10.8 ± 1.5 (4) 5.5 ± T41/ADJ 0.8 ± 0.2 (3) 3.3 ± 1.2 (3) 6.6 ± 1.5 (3) 11.8 ± 4.6 (3) 5.1 ± T1J/T ± 0.4 (3) 1.5 ± 0.2 (4) 7.0 ± 1.0 (3) 3.1 ± Evoked responses are measured as peaks of depolarization (errors represent SEMs). The numbers of larvae tested for each data point are indicated in parentheses. n represents the slope derived from four to six data points on a double-logarithmic plot of mean quantal content versus [Ca2]. The regression coefficient resulting from analysis of n values is also shown.

4 Neurobiology: Littleton et al..a 100 a - a) 0 o 10 Cẹ a Ce A wr T1l/AD3 T41/AD1 T41/AD2 1 5 mvl ms Ca21concentration, mm co * ;> 40 '.., 30 co -4 2 o o C-O 0., a) 0~ 4p.4' C) I- a) E 6 D Proc. Natl. Acad. Sci. USA 91 (1994) Ca2 concentration, mm -I- WT Tll/AD3 T41/AD2H T41/AD1 Genotype FIG. 2. Synaptic defects in syt mutants. (A) Evoked responses in wild type and three syt mutant combinations. The open arrowhead indicates onset of stimulation. Representative traces from one larva of each genotype are superimposed. (B) Measurements represent the mean of responses from three to six larvae for each data point, as shown in Table 3. Error bars represent SEM and are shown where larger than symbol. (C) Ca2 dependence of neurotransmitter release in wild-type and mutant larva. Each point represents the mean quantal content of 50-0 responses measured in four to six larva (error bars are SEM). (D) Measurements are the mean frequency of MEJPs for 7-15 larvae in mm [Ca2]; MEJP frequency did not change significantly in this [Ca2] range (errors represent SEM). WT, wild type. responds normally to released neurotransmitter and that the defect we observe in evoked response is presynaptic. The observed increase in MEJPs, however, cannot fully account for the reduction in evoked response or the changes in the Ca2 dependence of release, as the MEJP increase in syt mutations can be electrophysiologically separated from the dramatic decreases in evoked responses. As shown in Fig. 2D, the TJJ/AD3 and T41/AD2 larvae have no significant differences in MEJP frequency but exhibit dramatic differences in the amplitude of evoked responses (Fig. 2A) and n (Fig. 2C). DISCUSSION Approximately mutations in Drosophila syt have now been isolated. Complementation tests between severe loss-offunction syt mutant alleles that are homozygous lethal show that many alleles can partially complement each other. This type of intragenic complementation is often seen with genes that encode proteins that are part of a multimeric complex. As synaptotagmin has been shown to form a homotetramer () and to interact with a number of presynaptic proteins such as syntaxins (10), neurexins (11), and Ca2 channels (12), the intragenic complementation data support the idea that these multimeric interactions play an important role in vivo and that multimerization of different syt alleles can partially restore critical aspects of syt function disrupted by individual mutations. These data also indicate that synaptotagmin has different and partially independent functional domains. Intragenic complementation has allowed us to recover viable third-instar larvae trans-heterozygous for many syt alleles. Our recordings from these larva reveal that transheterozygous syt mutants can be subdivided into two main categories: (i) those that exhibit a sigmoidal Ca2 response curve (Fig. 2B), reflecting an n of 3.6 (Fig. 2C), but require higher Ca2 concentrations to evoke release and (ii) those that show a more linear-response curve (Fig. 2B) with an n of , approximately half of wild-type value. In addition to the dramatic decrease in neurotransmitter release, any perturbation in synaptotagmin, regardless of changes in n, can partially uncouple the fusion process from Ca2 influx and increase spontaneous vesicle fusion, consistent with a role for synaptotagmin as a clamp to prevent vesicle fusion in the absence of Ca2. The changes in Ca2 dependence of neurotransmitter release in the second class of syt mutants suggest that synaptotagmin may function as a Ca2 receptor for initiating exocytosis. Because an individual C2 domain can bind Ca2 (9), because each synaptotagmin molecule contains two C2 domains, and because synaptotagmin is likely to form a tetramer (8, ), it is tempting to speculate that the reduction in Ca2 cooperativity of neurotransmitter release in many trans-heterozygous mutants reflects specific alterations in the multimeric structure of synaptotagmin or, alternatively, alterations in the interactions between synaptotagmin domains. This hypothesis is strengthened by the molecular defects found in several syt mutants used in this study (see Fig. 3). As shown in Table 2, AD4, N13, and D39 behave as null alleles, as no trans-heterozygotes can be recovered, and these alleles do not contain one complete C2 domain (refs. 13 and 21; Sandra Kooyer and H.J.B., unpublished data). The AD] allele lacks the second C2 domain (21), and relatively few mutant alleles are viable in combination with AD), but those that do survive have an n = 1.6 (Table 3). The AD3 mutation has an altered amino acid in the second C2 domain (21) and, hence, is likely to be a less severe loss-of-function allele than AD). Interestingly, all tested trans-heterozygotes that contain an AD3 allele exhibit a Ca2 cooperativity of n = 3.6, but the release mechanism requires higher levels of external Ca2. Comparison of the behavioral defects with the underlying electrophysiological properties in the mutants leads to several interesting observations. Mutants carrying the AD) and AD2 alleles decrease the third- to fourth-order Ca2 dependency but appear the most normal behaviorally, whereas mutants carrying the AD3 allele are the most be-

5 10892 Neurobiology: Littleton et al. \ usic le N toplasm Lumen / - NX1I) \l1ele N) 3 41)4 [IM/ REPEAT A REPEAT B I- Domain Structure I XJH * Stop D39st. 1_ *Stop.AV)D L - *Y -b N..-11 AD)3 FIG. 3. Molecular defects in syt mutations. The domain structure of synaptotagmin with respect to the synaptic vesicle membrane is shown. Highlighted are the transmembrane region (TM) and the two homologous C2 domains (Repeats A and B). Shown below are the domain structures of several syt mutations. Alleles N13, AD4, and D39 behave as null alleles, indicating that deletion of both C2 domains completely disrupts synaptotagmin function. Heteroallelic combinations with AD] change the Ca2 dependence of release (n) from 3.6 to -1.8, whereas AD3 heterozygotes do not change n but, instead, alter Ca2 sensitivity of the release process. haviorally defective. Hence, lowering the order of the Ca2 dependency of neurotransmitter release may partially compensate for the higher requirement for Ca2. Although the presented data provide in vivo evidence that synaptotagmin is a Ca2 sensor involved in activation of neurotransmitter release, we cannot exclude the possibility that synaptotagmin is involved in other processes, such as docking (21) and endocytosis. Mutations in one gene other than synaptotagmin, dunce, have also been reported to alter n (22). dunce mutations are known to cause elevated camp levels (23). In addition, one observes an increase in release probability and evoked responses in dunce mutations (22) instead of a decrease, as seen in syt mutants. The increase in camp concentration may alter the phosphorylation state of many proteins (24) involved in the release process, leading to many defects, including aberrant synaptic facilitation and potentiation (22). Concurrent work on docking and fusion of synaptic vesicles has led to a model of synaptic vesicle trafficking in which the synaptic vesicle protein VAMP/synaptobrevin forms a docking complex with the presynaptic membrane proteins SNAP-25 and syntaxin (25). Recruitment of the soluble factors N-ethylmaleimide-sensitive fusion protein (NSF) and a- and t-snap (for soluble NSF attachment protein) may then participate in the fusion of the synaptic vesicle with the presynaptic membrane. Our in vivo findings suggest that a multimeric synaptotagmin complex may act to prevent vesicle fusion in the absence of Ca2 influx. However, the conclusion that the sole function of synaptotagmin may be to release the clamp on vesicle fusion is insufficient to account for the dramatic decrease in evoked response upon reduced synaptotagmin function. Therefore, we propose that synaptotagmin must participate more directly in vesicle fusion, possibly via its Ca2-activated phospholipid-binding capabilities (8). This mechanism could serve to bring the two membrane compartments in closer contact upon Ca2 influx and facilitate vesicle fusion. The demonstration that syt mutations dramatically reduce evoked response at all [Ca2] values and also decrease the Ca2 cooperativity required for neurotransmitter release suggests that synaptotagmin is probably the main neuronal-specific Ca2-binding protein responsible for Ca2 activation of vesicle fusion. However, Proc. Natl. Acad. Sci. USA 91 (1994) the presence of dramatically reduced and very abnormal neurotransmission in syt null alleles (5, 7, ) suggests that there is an additional Ca2-dependent process to the aforementioned putative vesicle docking/fusion complex. It is possible that the constitutive fusion pathway of all cells shows a slight Ca2 sensitivity (27, 28), similar to that seen in syt null mutants (). It is likely that neuronal-specific proteins like synaptotagmin have evolved to allow the transformation of a conserved vesicle trafficking system used in all cells into the rapid and efficient Ca2-dependent exocytosis characteristic of neuronal communication. We thank Aaron DiAntonio and Tom Schwarz for promptly sending us their fly strains. We thank Dan Littleton for assisting in the early phases of this work. We also thank Karen Schulze, David Sweatt, Jim Patrick, Paul Pfaffinger, John Dani, and Dan Johnston for critical comments. This work was supported by a National Institute of Mental Health fellowship to J.T.L., a National Institutes of Health grant to M.P. and H.J.B., a National Institutes of Health grant to M.P., and a National Institutes of Health grant to M.S. H.J.B. is an Assistant Investigator of the Howard Hughes Medical Institute. 1. Katz, B. & Miledi, R. (1965) Proc. R. Soc. London B 161, Bennett, M. K. & Scheller, R. H. (1993) Proc. NatI. Acad. Sci. USA 90, Perin, M. S., Fried, V. A., Mignery, G. A., Jahn, R. & Sudhof, T. C. (1990) Nature (London) 345, Nonet, M. L., Grundahl, K., Meyer, B. J. & Rand, J. B. (1993) Cell 73, DiAntonio, A., Parfitt, K. D. & Schwarz, T. L. (1993) Cell 73, Littleton, J. T., Bellen, H. J. & Perin, M. S. (1993) Development (Cambridge, U.K.) 118, Littleton, J. T., Stem, M., Schulze, K., Perin, M. & Bellen, H. J. (1993) Cell 74, Brose, N., Petrenko, A. G., Sudhof, T. C. & Jahn, R. (1992) Science 256, Davletov, B. A. & Sudhof, T. C. (1993) J. Biol. Chem. 8, Bennett, M. K., Calakos, N. & Scheller, R. H. (1992) Science 257, Petrenko, A. G., Perin, M. S., Davletov, B. A., Ushkaryov, Y. A., Geppert, M. & Sudhof, T. C. (1991) Nature (London) 353, Leveque, C., Hoshino, T., David, P., Shoji-Kasai, Y., Leys, K., Omori, A., Lang, B., El Far, O., Sato, K., Martin-Moutot, K., Newsom-Davis, J., Takahashi, M. & Seagar, M. J. (1992) Proc. Natl. Acad. Sci. USA 89, Littleton, J. T. & Bellen, H. J. (1994) Genetics 138, Robinow, S. & White, K. (1988) Dev. Biol. 1, Jan, L. Y. & Jan, Y. N. (1976) J. Physiol. (London) 2, Martin, A. R. (1955) J. Physiol. (London) 130, Fincham, J. R. S. (1966) Genetic Complementation (Benjamin, Amsterdam). 18. Dodge, F. A. & Rahaminoff, R. (1967) J. Physiol. (London) 193, Augustine, G. J., Charlton, M. P. & Smith, S. J. (1985) J. Physiol. (London) 369, Perin, M. S., Brose, N., Jahn, R. & Sfidhof, T. C. (1991) J. Biol. Chem. 6, DiAntonio, A. & Schwarz, T. L. (1994) Neuron 12, Zhong, Y. & Wu, C.-F. (1991) Science 252, Davis, R. L. & Kiger, J. A. (1981) J. Cell Biol. 90, Devay, P. & Friederich, P. (1987) J. Neurogenet. 4, Sollner, T., Whiteheart, S. W., Brunner, M., Erdjument-Bromage, H., Geromanos, S., Tempst, P. & Rothman, J. E. (1993) Nature (London) 362, Broadie, K., Bellen, H. J., DiAntonnio, A., Littleton, J. T. & Schwarz, T. (1994) Proc. Natl. Acad. Sci. USA 91, Dan, Y. & Poo, M. M. (1992) Nature (London) 359, Popov, S. V. & Poo, M. M. (1993) Cell 73,