Two distinct effects on neurotransmission in a temperature-sensitive SNAP-25 mutant

SNAP-25^ts is a recessive mutation

To study the role of SNAP-25 in vivo, we generated a mutation in the Drosophila gene by employing a large scale F₁ temperature-sensitive (ts) paralytic screen as described (Watson et al., 2001). F₁ progeny were tested at 38°C and from this screen, one allele that was paralytic over a deficiency that removes SNAP-25, Df(3L)1-16, was selected for further study. We will refer to this mutant allele as SNAP-25^ts. At 22°C, SNAP-25^ts flies can crawl, fly and mate but at 38°C, they are paralyzed. To further define the paralytic behavior, SNAP-25^ts flies were examined over a range of temperatures. SNAP-25^ts heterozygotes (SNAP-25^ts/TM3, Sb), homozygotes and hemizygotes (SNAP-25^ts/Df(3L)1-16) were tested between 32 and 40°C (Figure 1). Both SNAP-25^ts homozygotes and hemizygotes are paralyzed within seconds at 35°C, while heterozygotes were indistinguishable from wild type and are not paralyzed at temperatures up to 40°C. These data indicate that SNAP-25^ts is recessive, and since the homozygote and hemizygote behaved similarly, this indicates that SNAP-25^ts behaves as an apparent loss-of-function mutation above 35°C. Moreover, the transgenic expression of SNAP-25 in the nervous system with the P-element vector P[nsyb–SNAP-25] in the mutant background rescued the paralytic phenotype as indicated by an upward shift of the paralytic temperature to 38°C. This is consistent with the paralysis being due to a mutation in SNAP-25.

Mutation in SNAP-25^ts affects amphipathic helix I

The Drosophila SNAP-25 gene is composed of eight exons (Risinger et al., 1997). DNA sequencing of the SNAP-25^ts allele revealed a single base pair change of a G to A in exon 4 of the gene that resulted in a Gly50 to Glu substitution in the hydrophobic face of amphipathic helix I (Figure 2A and B). This Gly residue is highly conserved in members of the SNAP-25 family and related proteins. A temperature-sensitive allele of the yeast homologue of SNAP-25, SEC9, changes the analogous Gly to Asp, resulting in a loss of protein secretory function at 37°C (Brennwald et al., 1994). In order to see whether the mutation affected SNAP-25 localization, we stained larvae with an antibody against SNAP-25. We observed no difference in distribution of SNAP-25 at the third instar larval NMJ of wild-type and mutant animals (Figure 2C).

Because exon 4 is also alternatively spliced in Drosophila (H.Bellen, personal communication), it was possible that the other splice form could substitute for the ts allele. However, when we examined the level of splice variant mRNA we found that it accounts for <15% of the total SNAP-25 transcript. Furthermore, staining with an antiserum specific for the splice variant revealed that, unlike the predominant form of SNAP-25 protein, the splice variant was not concentrated in synaptic regions and was not detectable at the larval NMJ (our unpublished results). Thus, this splice variant was unlikely to play a significant role in the synaptic release of neurotransmitter.

To understand the effect of this mutation on the structure of the SNARE complex, we used Swiss Protein Database Viewer (Guex and Peitsch, 1997) to simulate the Gly to Glu substitution in the mouse SNARE complex structure. We found that the Glu side chain was predicted to sterically hinder Met163 (mouse numbering) in SNAP-25’s second amphipathic helix (Figure 2D). We anticipated that this steric hindrance may reduce the stability of the SNARE complex and alter its conformation.

SNAP-25^ts does not affect steady state levels of SNARE complex in vivo

To test whether the SNAP-25^ts protein affects SNARE complex stability in vivo, we examined the steady state levels of SNARE complex from adult head extracts. We predicted that the Gly50 to Glu change would affect SNARE complex levels in the SNAP-25^ts flies. An SDS-resistant 73 kDa SNARE complex consisting of SNAP-25, syntaxin and n-syb had been characterized previously from unboiled adult fly head extracts with an anti-syntaxin antibody on western blots (Tolar and Pallanck, 1998). A 73 kDa band was detected in head extracts from control and SNAP-25^ts lines (Figure 3A) at both 22 and 37°C. The 73 kDa band disappeared if mutant or control (red e) extracts were boiled and only syntaxin monomer was detected. The amounts of detectable complex at 22°C from SNAP-25^ts and control extracts were not significantly different from each other (Figure 3B). At 37°C, both control and mutant lines show a temperature-dependent increase in SNARE complex levels. The temperature-dependent increase in SNARE complex levels in control flies is in agreement with a previous study (Tolar and Pallanck, 1998). Thus, the mutation in SNAP-25 had virtually no effect on SNARE complex levels at 22 or 37°C.

SNAP-25^ts affects the mobility and temperature stability of SNARE multimers

In order to further examine the nature of the mutant complex and its temperature-dependent phenotype, SNARE complex formation and temperature stability were examined in vitro using recombinant proteins. Recombinant syntaxin and neuronal synaptobrevin (minus their transmembrane domains) were mixed with either wild-type SNAP-25 or SNAP-25^ts and the amount of SNARE complex was assayed at different time intervals by SDS–PAGE and Coomassie Blue staining. We saw the expected SNARE complexes of ∼70 kDa and higher order SNARE complexes or SNARE multimers of ∼120 kDa in the unboiled assembly reactions. No SNARE complexes were detected if any of the three components were omitted from the reaction mixture or if the sample was boiled. Binary complexes between SNAP-25 and syntaxin were seen in unboiled reactions lacking n-syb, but these differed in size from SNARE complexes and SNARE multimers. Although SNARE complex formation rates were similar in both wild-type SNAP-25- and SNAP-25^ts-containing mixtures with maximal formation occurring by 20 h, the total amount of SNAP-25^ts complexes formed was reduced by 39% (p <0.001) as compared with wild type (Figure 3C). Mutant SNARE multimer formation at 20 h was reduced by 90% as compared with wild type (p <0.001, Figure 3C) but by only 28% at 96 h (p = 0.02).

We also examined the mutant complexes’ stability upon heating. Mutant and wild-type SNARE complexes were heated from temperatures of 22 to 80°C as described (Chen et al., 1999). Little difference was observed in the melting temperature curves of mutant and wild-type SNARE complexes (Figure 3D). However, mutant SNARE multimers were very unstable upon heating. After only a 5 min incubation at 37°C, mutant SNARE multimers were undetectable by Coomassie Blue staining or by western blotting whereas wild-type multimers were detectable up to 45°C (Figure 3E). This suggests that temperature-sensitive paralysis in the mutant may result from temperature instability of SNARE multimers. We also observed that mutant SNARE complexes had lower mobility than wild-type SNARE complexes on SDS–PAGE gels. These mobility changes were even more pronounced in SNARE multimers, with mutant SNARE multimers running significantly larger in size than the corresponding wild-type multimer (Figure 3E). The increased size of the mutant complex is consistent with the notion that its shape is altered. Mobility changes were also observed in 73 kDa SNARE complexes derived from some preparations of head extracts but these shifts were not found as consistently as those in vitro (data not shown). SNARE multimers were not routinely detected from head extracts, thus their mobility changes could not be assessed (data not shown).

In an effort to further define the structural change in the in vitro mutant complex, both mutant and wild-type SNARE complexes were treated with trypsin to look for any differences in digestion pattern. However, no difference was seen in the SNARE complex digestion pattern between the wild-type and mutant complexes (data not shown).

Evoked release is increased at 22°C and decreased at 37°C

To examine how the SNAP-25^ts mutation affects neurotransmitter release we recorded synaptic events from the NMJ of third instar larvae. At the permissive temperature, nerve stimulation surprisingly resulted in larger excitatory junctional currents (EJCs) in SNAP-25^ts homozygotes as compared with controls (Figure 4A and C). An ∼2-fold increase in EJC amplitude was observed at subsaturating concentrations of external calcium (e.g. p <0.01, ANOVA at 0.8 mM Ca²⁺).

To examine whether this increase resulted from a change in the sensitivity of evoked release to external Ca²⁺, recordings were performed on SNAP-25^ts and control larvae over a range of Ca²⁺ concentrations. A linear plot of EJC amplitude versus Ca²⁺ concentration revealed that the SNAP-25^ts mutant was indeed more sensitive to external Ca²⁺ than were controls (Figure 4B, left). Examination of log–log plots of EJC amplitude versus Ca²⁺ concentration showed that, though the curve is clearly shifted to the left, cooperativity of transmitter release was not different in the mutant as compared with the control, with both having cooperativity coefficients of ∼3 (Figure 4B, right). We further examined the synaptic phenotype of SNAP-25^ts/Df(3L)1-16 hemizygotes at 0.8 mM Ca²⁺ and found that it was identical to that of SNAP-25^ts homozygotes (Figure 4C), indicating that the SNAP-25^ts allele is recessive and acts as an apparent loss-of-function mutation at 22°C. Expression of a SNAP-25 transgene in the hemizygote background rescued the synaptic phenotype (Figure 4C).

Why is there enhanced evoked release in the SNAP-25^ts mutant? One possibility is that quantal size is larger in the mutant. However, recordings of miniature EJCs (mEJCs) from the mutant and control animals revealed that mEJC amplitude was unaltered in the mutant, indicating that quantal size was not changed (Figure 5A and B). Interestingly, we additionally found that the SNAP-25^ts mutant showed a 6-fold increase in mEJC frequency (p <0.01, ANOVA) as compared with controls (Figure 5C). Expression of a SNAP-25 transgene partially rescued the elevated mEJC frequency (Figure 5C). The measured mEJC frequency was not dependent on extracellular Ca²⁺ since we recorded similar frequencies for control and SNAP-25^ts in both 1 mM Ca²⁺ and 0 mM Ca²⁺ plus 2 mM EGTA [SNAP-25^ts, 0 mM Ca²⁺, 9.0 ± 0.9 Hz (n = 15); red e, 0 mM Ca²⁺, 1.4 ± 0.2 Hz (n = 6)].

Evoked release could also be elevated if the mutation caused an increase in the number of neuromuscular synapses. SNAP-25^ts and control larvae were stained with anti-synaptotagmin antiserum to determine whether the number of synaptic boutons was altered in the mutant. We counted the number of synaptic boutons per muscle in SNAP-25^ts (113.9 ± 7.3, mean ± SEM) and control (92.8 ± 5.5) larvae and found no significant difference.

From the behavioral data, we know that SNAP-25^ts flies are paralyzed at 35°C. Is this paralysis due to synaptic failure? To address this question, single electrode intracellular recordings were performed on SNAP-25^ts larvae at 37°C, since two electrode voltage clamp recordings proved deleterious to the preparation. In addition, we found that the high temperature caused a high rate of synaptic failures in HL3 saline with the standard 0.8 mM Ca²⁺, even in control animals. To counter the failure rate, we chose to carry out the elevated temperature recordings at 2 mM Ca²⁺, a concentration at which transmitter release is saturated for both genotypes at 22°C (Figure 6A, left, also see Figure 4B, left). These recording conditions would have the tendency to minimize the difference between control and SNAP-25^ts since this Ca²⁺ level yields the same amplitude of the excitatory junctional potentials (EJPs) from both genotypes at 22°C. None the less, we found that at 37°C, evoked release in the SNAP-25^ts mutant was significantly reduced, by 60% (p <0.001, ANOVA) as compared with SNAP-25^ts at 22°C, whereas controls showed no significant temperature-dependent decrease (Figure 6A and B). As we found at the permissive temperature, the phenotype at the elevated temperature of SNAP-25^ts/Df(3L)1-16 hemizygotes was identical to that of SNAP-25^ts homozygotes and the phenotype of the hemizygote was rescued by the SNAP-25 transgene (Figure 6B). We next analyzed spontaneous transmitter release at 37°C (Figure 6C) and found that the frequency of release in controls was increased to a level comparable to that of SNAP-25^ts at 22°C. Interestingly, despite the reduction in evoked release, there was no detectable difference in the frequency of spontaneous release at 37°C in SNAP-25^ts mutants as compared with controls. The spontaneous frequency in SNAP-25^ts/Df(3L)1-16 hemizygotes was also similar to that of SNAP-25^ts homozygotes at 37°C.

SNAP-25^ts alters the structure of the active zone in larvae

To identify morphological effects of the mutation, we examined the synaptic ultrastructure of the larval NMJ. The synapses of SNAP-25^ts larvae appeared to be grossly normal at 22 and 37°C. The nerve terminals were filled with synaptic vesicles (30 nm in diameter), with many vesicles clustered near the active zone, known as the T-bar; vesicles were also found in morphologically docked positions near the T-bar (Figure 7A).

To determine whether the enhanced evoked release at 22°C and the reduced release at 37°C is due to changes in the distribution or number of synaptic vesicles, the number of vesicles clustered within 200 nm of the T-bar was counted in SNAP-25^ts animals and in controls at both temperatures. No difference in vesicle density was detected in SNAP-25^ts as compared with controls at 22°C (188 ± 10 versus 185 ± 11 vesicles/µm²). While a slight decrease in vesicle density was observed in SNAP-25^ts as compared with controls at 37°C (205 ± 19 versus 229 ± 8 vesicles/µm², respectively), this decrease was not statistically significant (Mann–Whitney test, p >0.05). The vesicle densities reported here are consistent with previous reports where vesicle density in wild-type animals ranged from 155 to 200 vesicles/µm² (Jia et al., 1993; Renger et al., 2000). We also determined whether the number of morphologically docked vesicles was different in the mutant as compared with control at both 22 and 37°C. As described in Reist et al. (1998), morphologically docked vesicles are defined as those vesicles that are within 200 nm of the T-bar and whose centers are within 12–18 nm (one vesicle radius ± 3 nm) of the presynaptic membrane. No difference in vesicle docking was observed at either temperature with SNAP-25^ts having 0.79 ± 0.15 and 0.67 ± 0.21 docked vesicles at 22 and 37°C, respectively, and control having 1.09 ± 0.21 and 0.93 ± 0.18 docked vesicles at 22 and 37°C, respectively (Mann–Whitney test; p >>0.05).

Although the vesicle distribution near the T-bar was unchanged at the SNAP-25^ts larval NMJ, some qualitative changes in the structure of the synapse were evident. In particular, the T-bar in the SNAP-25^ts larvae at 22°C appeared to be more dense and filamentous as compared with controls (Figure 7B and C). Such T-bars were rarely seen in control animals at both 22 and 37°C and were less frequently observed in SNAP-25^ts at 37°C. Because docked vesicles very near the central beam of the T-bar were difficult to discern due to the filamentous nature of the T-bar, this may have resulted in an underestimate of vesicle docking in SNAP-25^ts at 22°C.

Role of SNARE complex in neurotransmitter release

Neurotransmitter release is affected in two opposing ways in SNAP-25^ts mutants depending upon temperature. How could these divergent phenotypes arise from a single amino acid substitution in SNAP-25? The region around Gly50, which is mutated in SNAP-25^ts, is one of the most mobile in the SNARE complex and conformational changes in this complex are likely to be centered within this region (Poirier et al., 1998). CD spectroscopy and EPR studies have indicated that SNARE proteins undergo dramatic conformational changes during complex assembly in vitro (Fasshauer et al., 1997; Margittai et al., 2001). After the complex assembles, different conformations of the trans-SNARE complex (loose and tight) have been proposed to explain changes in the kinetics of release and suggest that relaxed conformations may be more fusion competent (Xu et al., 1999). Our in vitro electrophoretic mobility data suggest that the mutant SNARE multimer exists in an altered, more fusion-ready conformation, since the evoked release of neurotransmitter was increased 2-fold at 22°C. The in vitro data also suggest an explanation for the decrease in neurotransmitter release at 37°C. Mutant SNARE multimers are very unstable at 37°C. Two recent reports indicate that SNARE multimers, specifically trimers, may be the form responsible for vesicle fusion (Stewart et al., 2000; Hua and Scheller, 2001). Thus, decreased levels of mutant SNARE multimers at 37°C could lead to the blockade in evoked release of neurotransmitter observed in the SNAP-25^ts mutant.

We also observed that the mutation in SNAP-25 reduced its ability to form SNARE complexes in vitro but not in vivo. How might this discrepancy be explained? The in vitro SNARE assembly reactions used syntaxin and synaptobrevin without their transmembrane domains. Transmembrane domains have been shown to stabilize in vitro-generated SNARE complexes (Poirier et al., 1998). It is presumed that other cellular factors aid in the very rapid process of SNARE assembly in vivo but these, too, are lacking in in vitro systems. The increased level of SNAREs in heated flies also differs from in vitro results. This effect appears to be due to an accumulation of misfolded, SDS-resistant SNARE complexes that are not effectively resolved by N-ethylmaleimide-sensitive factor (NSF) (Tolar and Pollanck, 1998; Mohtashami et al., 2001; this study) and this form of the SNARE complex is unlikely to be involved in any fusion events.

What is the role of Ca²⁺ in the phenotype of the mutant? The SNAP-25^ts mutant shows increased evoked release at a given external Ca²⁺ concentration. One possibility is that the SNARE complex may act in Ca²⁺ sensing and this function may be enhanced by the mutation. Classical studies have indicated that the blockage of neurotransmitter release by botulinum toxin A treatment and SNAP-25 cleavage could be partially reversed by increasing Ca²⁺ levels (Simpson, 1978). Treatment with toxins that cleave syntaxin or synaptobrevin cannot be overcome with Ca²⁺, suggesting that SNAP-25 in particular is important in Ca²⁺ sensing. The crystal structure of the SNARE complex revealed Sr²⁺ binding sites, which may also bind Ca²⁺ (Fasshauer et al., 1998; Sutton et al., 1998). In addition, recent evidence indicates that hypomorphic mutations of n-syb and syntaxin in Drosophila reduce the Ca²⁺ cooperativity of release (Stewart et al., 2000). These data collectively suggest a Ca²⁺ sensing role for SNARE proteins.

It is possible that the enhanced Ca²⁺ sensitivity in the mutant could also be mediated by other proteins. It has recently been shown that there is Ca²⁺-dependent interaction between the C-terminus of SNAP-25 and synaptotagmin (Gerona et al., 2000). However, the site of the mutation in SNAP-25^ts is near the N-terminus not the C-terminus, arguing against this mechanism of action. Recent work also suggests that synaptotagmin may mediate SNARE complex formation or oligomerization (Littleton et al., 2001). It is possible that synaptotagmin may more favorably assemble the mutant SNARE complex. Another hypothesis is that enhanced Ca²⁺ sensitivity could also result from alteration of resting levels of Ca²⁺ in the nerve terminal. Elevated levels of resting Ca²⁺ could explain both the elevated evoked and spontaneous release in the mutant. Our data argue against this hypothesis as removal of extracellular Ca²⁺ has no effect on the frequency of spontaneous release in the mutant. How ever, the enhanced evoked release of neurotransmitter could result from increased Ca²⁺ entry through Ca²⁺ channels during stimulation. Although SNAREs interact with Ca²⁺ channels through the conserved synprint peptide in mammals (Sheng et al., 1996), this sequence does not appear to be present in the major Drosophila Ca²⁺ channel (Kawasaki et al., 2000). Another Ca²⁺ channel modulatory site has been found to interact with syntaxin (Bezprozvanny et al., 2000) so it is possible that this site may cause greater Ca²⁺ entry during stimulation.

SNAP-25^ts may affect the tethering of vesicles to the active zone material

The SNAP-25^ts mutation caused the active zone material to have a thickened and filamentous appearance. Three-dimensional reconstructions using EM tomography of frog NMJs have discerned ‘ribs’ that attach vesicles to the active zone material (Harlow et al., 2001). The filamentous projections that we observed at T-bars in SNAP-25^ts may correspond to an extended version of these ribs and they may be involved in either the elevated mini-frequency or elevated evoked release.

In summary, our characterization of SNAP-25^ts indicates that the conformation of SNAP-25 plays a key role in regulating the fusion of synaptic vesicles. Our in vitro data suggest that a less compact or ‘relaxed’ SNARE multimer may support vesicle fusion. Further studies may define how SNAP-25 more precisely carries out this function.

Plasmid construction and P-element transformation

P[nsyb–SNAP-25] is a Casper2 derivative containing a pan-neural promoter driving the expression of a SNAP-25 cDNA. It was constructed as follows. A 2 kb EcoRI genomic fragment of the n-syb promoter (L.Vosshall, personal communication) was subcloned into Casper2. The resulting plasmid was digested with HpaI and a blunted HindIII fragment containing a chimeric intron from pCAT3-Basic (Promega, Madison, WI) was ligated into the plasmid. Then, a blunted BamHI–XbaI fragment of the SV40 late poly(A) region from pCAT3-Basic was subcloned into the StuI cloning site of the above construct. An SspI fragment of the SNAP-25 cDNA containing the entire ORF was then subcloned into the blunted BglII site of the above construct to yield the P-element vector, P[nsyb–SNAP-25]. P[nsyb–SNAP-25] was used to transform Drosophila.

Fly stocks and genetic crosses

Deficiency line Df(3L)1-16/TM3, Sb (Marchant and Holm, 1988), which removes SNAP-25 (our unpublished results) was obtained from the Bloomington Stock Center. Isozygotic red e males were fed a sucrose solution containing 25 mM ethylmethane sulfonate and mated en masse with Df(3L)1-16/TM3, Sb virgin females; 450 000 F₁ progeny were collected and temperature tested in a heated box. A single line was identified that, upon repeated testing, was paralyzed in combination with Df(3L)1-16. A stock was established from this line. Genetic rescue experiments were performed by crossing SNAP-25^ts with P[nsyb– SNAP-25]/P[nsyb–SNAP-25]; Df(3L)1-16/TM6B, Hu Tb. Larvae that were SNAP-25^ts/Df(3L)1-16 were selected based on non-Tb appearance and adult flies were selected based upon those lacking Hu.

RT–PCR and DNA sequencing

RNA was prepared from heads of red e and SNAP-25^ts flies using Trizol (Life Technologies, Gaithersburg, MD). The resulting total RNA was reverse transcribed using Superscript II (Life Technologies) and the entire ORF was amplified by PCR with primers 5′-GCGCGCGCACGTGTTTAAG-3′ and 5′-GGGCTAGGTTGCACTCGTATGTATC-3′, using Tfl DNA polymerase (Epicentre Technologies, Madison, WI). The PCR product was purified using Qiagen spin columns (Qiagen, Valencia, CA) and sequenced by the Cornell University Sequencing Facility.

Antibodies

The peptide MPADPSEEVAPQVPKT(norL)C containing N-terminal sequence from Drosophila SNAP-25 was synthesized (Cornell University BioResource Center), coupled to keyhole limpet hemocyanin (KLH) and used to raise antisera in rabbits by Cornell Veterinary College Animal Services. The alternative exon peptide AEVGMRSIVMLDC was synthesized by Bethyl Laboratories (Montgomery, TX), coupled to KLH, and used to raise antisera as described above. Antisera were affinity purified using the Sulfolink Kit (Pierce Chemical, Rockford, IL). The anti-synaptotagmin antiserum was a kind gift from Dr Hugo Bellen and the 8C3 anti-syntaxin monoclonal was obtained from the Developmental Studies Hybridoma Bank.

Western blotting and Coomassie staining

SNARE complexes were isolated by a modification of the method of Tolar and Pallanck (1998). Briefly, two freshly dissociated heads from SNAP-25^ts or red e adults were placed in 5 µl of HL3 saline containing 1 mM Ca²⁺ (Stewart et al., 1994) and incubated at 22 or 37°C for 5 min. Heads were crushed on ice in 25 µl of 10% glycerol, 1% SDS, 62.5 mM Tris–HCl pH 6.8, 1.2% β-mercaptoethanol and 0.01% bromophenol blue; 20 µl was loaded on a 8% gel and blotted onto Immobilon membranes (Millipore, Bedford, MA). The blot was probed with the 8C3 anti-syntaxin monoclonal antibody at 1:1000 dilution, processed and visualized with enhanced chemiluminescence (Amersham Pharmacia Biotech, Piscataway, NJ). Blots were exposed to Kodak BioMax ML film (Kodak, Rochester, NY) within the linear range of the film and bands were scanned and quantitated using Quantity One software (Bio-Rad, Richmond, CA). The average SNARE complex levels were calculated from nine independent samples. The in vitro SNARE complex formation assay was a modification of the one used in Pevsner et al. (1994). Drosophila n-syb (minus the transmembrane domain), syntaxin (minus the transmembrane domain), SNAP-25 and SNAP-25^ts were cloned into a pGEX-KG vector, expressed in XL-1 Blue, purified with glutathione–Sepharose-4B beads (Amersham Pharmacia Biotech) and cleaved by thrombin. N-syb, syntaxin and SNAP-25 or SNAP-25^ts were mixed (10 µM each) in 50 mM Tris–HCl pH 8.0, 150 mM NaCl, 0.5% Tween-20, 5 mM dithiothreitol and 2 mM EDTA and incubated for the times indicated at 4°C. The amount of complex formation was calculated from four independent experiments and one-way ANOVA was used for statistical comparisons. After incubation in binding buffer, samples were mixed with an equal volume of sample loading buffer (10% glycerol, 1.34% SDS, 62 mM Tris–HCl pH 7.5, 0.17 M β-mercaptoethanol) and separated by SDS–PAGE at 4°C. Temperature stability of the complexes was tested by incubating 20 h assembly reactions for an additional 5 min at the temperatures indicated, in sample loading buffer before SDS–PAGE. Gels were stained with Coomassie Blue, scanned, and bands were quantitated by ImageQuant 5.0 (Molecular Dynamics, Sunnyvale, CA). The regression melting curve was calculated using StatView. Wild-type and mutant SNARE melting temperature regression curves had R² values of 0.996 and 0.998, respectively. Western transfer of SNARE complexes were probed with antibodies against syntaxin (1:1000), synaptobrevin (1:750) and SNAP-25 (1:1000) to confirm the presence of each component in the SNARE complex. Controls lacking individual proteins failed to produce SNARE complexes and higher order complexes.

Electrophysiology

Electrophysiological procedures and physiological solutions are described in Stewart et al. (1994). In brief, wandering third instar larvae were dissected at room temperature in HL3 saline. For two-electrode voltage clamp recordings a holding potential of –80 mV was used. Each electrode was ∼15–20 MΩ when filled with 3 M KCl. Data were digitized at 10 kHz and low-pass filtered at 2 kHz for EJCs, or 600–800 Hz for mEJCs, with pClamp8 software (Axon Instruments, Foster City, CA). For mEJC analysis 1 min epochs of data were analyzed for amplitude and frequency using Clampfit (Axon Instruments, Foster City, CA). For recordings at restrictive temperature (37°C), larvae were dissected at room temperature, then immersed directly into the 37°C recording chamber and allowed to incubate for at least 5 min before recording. Comparisons of EJP amplitude and mEJP frequency were thus made between populations of responses recorded from independent larvae at the permissive and restrictive temperatures. The bath temperature was controlled by a Peltier device (Model PSMI) and temperature controller (Model TC-202A) from Medical Systems Research Products.

Immunocytochemistry

Third instar wandering larvae were dissected as described in Stewart et al. (1994), and fixed in non-alcoholic Bouin’s fluid. Preparations were stained with either 1:1000 rabbit anti-synaptotagmin or 1:100 rabbit anti-SNAP-25 diluted in PBST. For SNAP-25 staining, preparations were incubated with 1:500 dilution of a goat anti-rabbit Alexa Fluor 488 (Molecular Probes, Eugene, OR), rinsed and mounted with Vectashield (Vector Laboratories, Inc., Burlingame, CA), and viewed with a Nikon Eclipse E600-FN microscope (Nikon, Melville, NY) equipped with 40×, 0.8 NA water-immersion lens, 100 W mercury lamp and an FITC/GFP filter (Chroma Technology, Brattleboro, VT). Images from mutant and control animals were acquired with a cooled CCD SPOT2 camera using the SPOT software v2.2 (Diagnostic Instruments, Sterling Heights, MI) with identical camera settings. For synaptotagmin staining, preparations were incubated with biotinylated goat anti-rabbit IgG and processed using a Vectastain Elite ABC kit (Vector Laboratories, Inc.). We counted the number of anti-synaptotagmin stained boutons on muscle fibers 6 and 7 of the third abdominal segment (A3) in SNAP-25^ts (n = 10) and red e controls (n = 8). For statistical comparison of bouton counts, 95% confidence intervals (CI) on each mean were calculated and the counts were considered significantly different when their CI did not overlap.

Electron microscopy

Third instar wandering larvae were dissected in HL3 saline as described above and incubated in HL3 saline at 22 or 37°C for at least 5 min before fixation. Dissected larvae were processed for electron microscopy as described (Jia et al., 1993) with the addition of 4–9% sucrose to the fixative. Tissue was embedded in EMbed (Electron Microscopy Sciences, Inc), and several series of transverse sections (50–150 sections/series, 75–100 nm thick) were collected.

Only type Ib boutons (Atwood et al., 1993) primarily from muscle fibers 6 and 7, and, occassionally, fibers 12 and 13, of A3 were used. NIH image software was used to measure the distance from the center of a vesicle to the presynaptic membrane. The number of docked vesicles observed in a single section through the T-bar is reported in this study. We counted the number of vesicles within a 200 nm radius of the center of the T-bar (defined as the point where the top of the bar intersects its base). The area of the sampled region was measured and the density was calculated as number of vesicles/µm². Statistical significance was determined using a Mann–Whitney test where p <0.05 was considered significant.