At the synapse, SNAP-25, along with syntaxin/HPC-1 and synaptobrevin/VAMP, forms SNARE N-ethylmaleimide-sensitive factor [soluble (NSF) attachment protein receptor] complexes that are thought to catalyze membrane fusion. Results from neuronal cultures of synaptobrevin-2 knockout (KO) mice showed that loss of synaptobrevin has a more severe effect on calcium-evoked release than on spontaneous release or on release evoked by hypertonicity. In this study, we recorded neurotransmitter release from neuronal cultures of SNAP-25 KO mice to determine whether they share this property. In neurons lacking SNAP-25, as those deficient in synaptobrevin-2, we found that ∼10–12% of calcium-independent excitatory and inhibitory neurotransmitter release persisted. However, in contrast to synaptobrevin-2 knockouts, this remaining readily releasable pool in SNAP-25-deficient synapses was virtually insensitive to calcium-dependent–evoked stimulation. Although field stimulation reliably evoked neurotransmitter release in synaptobrevin-2 KO neurons, responses were rare in neurons lacking SNAP-25, and unlike synaptobrevin-2–deficient synapses, SNAP-25–deficient synapses did not exhibit facilitation of release during high-frequency stimulation. This severe loss of evoked exocytosis was matched by a reduction, but not a complete loss, of endocytosis during evoked stimulation. Moreover, synaptic vesicle turnover probed by FM-dye uptake and release during hypertonic stimulation was relatively unaffected by the absence of SNAP-25. This last difference indicates that in contrast to synaptobrevin, SNAP-25 does not directly function in endocytosis. Together, these results suggest that SNAP-25 has a more significant role in calcium-secretion coupling than synaptobrevin-2.
Synaptic vesicle fusion is mediated by the formation of SNARE [for “soluble N-ethylmaleimide-sensitive factor (NSF)-attachment protein receptor proteins”] complexes (Sollner et al. 1993) from the SNARE proteins synaptobrevin/VAMP, syntaxin-1, and SNAP-25 (Jahn and Scheller 2006; Rizo and Sudhof 2002). Syntaxin and synaptobrevin are anchored on the plasma and the synaptic vesicle membrane, respectively, by a transmembrane region, whereas SNAP-25 is attached to the plasma membrane by palmitoylated cysteines (Hess et al. 1992). All SNARE proteins contain a sequence called the SNARE motif that associates into parallel four-helical bundles to form SNARE complexes, with SNAP-25 contributing two SNARE motifs, and syntaxin and synaptobrevin each contributing one SNARE motif to the synaptic SNARE complex (Poirier et al. 1998; Sutton et al. 1998). SNARE complexes are thought to assemble by “zippering” in an N- to C-terminal direction, thereby forcing their resident membranes closely together (Sorensen et al. 2006). SNAREs alone seem sufficient to fuse lipid bilayer vesicles (Weber et al. 1998) and fibroblast plasma membranes (Hu et al. 2003), and so may represent the minimal fusion machinery. It is less clear whether SNAREs alone are sufficient to execute fusion physiologically, but likely cooperate with SM proteins (Sec1/Munc18-like proteins) in a poorly understood reaction (Rizo and Sudhof 2002).
Initial analyses of genetic deletions of individual SNARE proteins present a complicated view of SNARE function in neurotransmitter release. In Drosophila, loss of either syntaxin or synaptobrevin results in a complete loss of calcium-evoked release, but some spontaneous release persists (Deitcher et al. 1998; Schulze et al. 1995), as well as a residual hyperosmotic saline response (Broadie et al. 1995). In Caenorhabditis elegans, syntaxin null mutants were almost completely paralyzed (Saifee et al. 1998), whereas synaptobrevin nulls exhibit reduced but not absent movements, such as pharyngeal pumping (Nonet et al. 1998). In mice, loss of syntaxin 1A results in normal basic neurotransmitter release, but there is a phenotype in hippocampal long-term potentiation and conditioned fear memory (Fujiwara et al. 2006). This relatively weak phenotype might be due to compensation by syntaxin 1B. Deletion of mouse synaptobrevin-2 also causes only a partial impairment of neurotransmitter release; here, spontaneous and hypertonic sucrose-induced neurotransmitter release is relatively less affected than evoked release (Schoch et al. 2001). Moreover, both in flies and mice, synaptobrevin null mutants exhibit a facilitation of release during 10-Hz stimulation (Deak et al. 2004; Yoshihara et al. 1999). In contrast to synaptobrevin, loss of SNAP-25 in flies did not diminish neurotransmission substantially, partly because of potential compensation from SNAP-24, a protein closely related to SNAP-25 (Niemeyer and Schwarz 2000; Vilinsky et al. 2002). In mice, SNAP-25 deletion leads to lethality at birth, and secretion, in particular stimulus evoked secretion, is severely impaired (Sorensen et al. 2003; Tafoya et al. 2006; Washbourne et al. 2002). However, a detailed analysis of neurotransmission and vesicle trafficking after loss of SNAP-25 has not been performed. Here, we analyzed the remaining neurotransmission in mature hippocampal cultures from SNAP-25–deficient mice and tested whether loss of SNAP-25 causes a differential impairment of Ca2+-dependent and -independent synaptic vesicle trafficking. In SNAP-25 knockout (KO) neuronal cultures, in agreement with earlier findings (Washbourne et al. 2002), we detected almost no calcium-evoked release. Even strong stimulation with elevated potassium could barely elicit responses. In contrast, spontaneous neurotransmission occurs reliably in SNAP-25 KO neurons, albeit at a lower frequency than controls. In addition, SNAP-25 mutants always respond to hypertonic sucrose application, a calcium-independent form of stimulation. Furthermore, SNAP-25–deficient synapses are capable of synaptic vesicle recycling monitored by uptake and release of FM dyes in response to hypertonic sucrose stimulation.
KO mice were bred by crossing heterozygous SNAP-25 KO mice (Washbourne et al. 2002). All mouse handling and embryo harvesting techniques were approved by the U.T. Southwestern Medical Center Institutional Animal Care and Use Committee. Hippocampal neurons from embryonic day 17.5–18.5 littermate mice were dissociated by trypsin (5 mg/ml for 10 min at 37°C), triturated with a siliconized pipette, and plated (100 μl) onto each 12-mm coverslip coated with Matrigel (BD Biosciences; ∼1 whole hippocampus for 3–4 coverslips). After 10–15 min, the wells were flooded with plating medium [1 ml/well; minimal essential media (Gibco) containing 5 g/l glucose, 0.2 g/l NaHCO3 (Sigma), 0.1 g/l transferrin (Calbiochem), 0.25 g/l insulin (Sigma), 0.3 g/l l-glutamine (Gibco), and 10% heat inactivated fetal bovine serum]. This method of plating usually results in a dense area of neurons in the center of the coverslip, and SNAP-25 KO neurons survived best if the cells were separated by 1 cell diameter or less. Neurons were cultured at 37°C in a humidified incubator with 95% air-5% CO2 in minimal essential media (Gibco) containing 5 g/l glucose, 0.2 g/l NaHCO3 (Sigma), 0.1 g/l transferrin (Calbiochem), 0.3 g/l l-glutamine (Gibco), 5% heat inactivated fetal bovine serum, 2% B-27 supplement (Gibco), and 0.5–1 μM cytosine arabinoside (Sigma) (added 24 h after plating), and used for experiments after 13–17 days in vitro (DIV).
Fourteen or 21 DIV hippocampal cultures were washed 3 times with room temperature PBS and incubated in cold 100% methanol (−20°C) in the freezer for 10 min. The neurons were washed 3 times with PBS and incubated for 15 min in blocking solution (PBSS: 3% milk and 0.1% saponin; Sigma). Primary incubation with synapsin (E028, polyclonal) and Map2 (Sigma, monoclonal) antibodies at 1:1,000 dilution in PBSS was done at room temperature for 1 h followed by four to five washes 5–10 min each in PBSS. Secondary antibodies (Alexa 488 and 546) at 1:500 dilution were incubated for 1 h at room temperature in PBSS and washed as before in PBSS. The coverslips were rinsed with water, mounted on slides, and viewed on a Leica confocal microscope.
Fourteen to 16 DIV cultures were treated for 90 s with horseradish peroxidase (HRP; 2 mg/well) in modified Tyrode solution [in mM: 150 NaCl, 4 KCl, 2 MgCl2, 10 glucose, 10 HEPES-NaOH (pH 7.4), 0.01 CNQX, 0.05 AP-5, and 2 mM CaCl2 (∼310 mOsm)]. Stimulated cultures were treated with bath solution containing 90 mM KCl substituted for NaCl. The cells were quickly rinsed 2 times in nominally calcium-free Tyrode solution containing 6 mM MgCl2 and fixed in 4% paraformaldehyde (PFA) +0.1% glutaraldehyde (GA) in PBS for 20 min at room temperature. After fixing, the cells were rinsed 2 times with 100 mM Tris followed by incubation in a 100 mM Tris solution containing 0.1% diaminobenzidine (DAB; Sigma) and 0.02% H2O2 for 10 min. To stop the DAB reaction, cells were rinsed 3 times with distilled water. The neurons were fixed for 1 h with 2% glutaraldehyde in 0.1 M sodium phosphate buffer (pH 7.4) at 37°C. Cells were rinsed 2 times in buffer and incubated in 0.5% OsO4 for 30 min at room temperature. Specimens were dehydrated in ethanol and embedded in Poly/bed 812 for 24 h. Sections (60 nm) were poststained with uranyl acetate and viewed with a JOEL 1200 EX transmission microscope.
FM dye imaging
Modified Tyrode bath solution was used for all experiments. Synaptic terminals were loaded with either N-(3-triethylammoniumpropyl)-4-[4-(diethylamino)styryl]pyridinium dibromide (FM2-10; 400 μM; Invitrogen) or N-(3-triethylammoniumpropyl)-4-[4-(dibutylamino)styryl]pyridinium dibromide (FM1-43; 8 μM; Invitrogen) by a 90-s exposure to bath solution containing 47 mM KCl substituted for NaCl or a 30-s exposure to nominally Ca2+-free bath solution containing 500 mOsm sucrose. Both stimulation solutions were added to the bath (200 μl) as a 2 times, 200 μl dye containing solution and mixed by pipetting up and down four times. Before destaining, neurons were washed for 10 min in dye-free, nominally Ca2+-free solution to minimize spontaneous release. In all experiments, isolated boutons (∼1 μm2) were selected for analysis. Destaining was triggered with a hyperkalemic bath solution containing 90 mM KCl substituted for NaCl or nominally Ca2+-free bath solution containing 500 mOsm sucrose applied to the bath by gravity perfusion onto the field of interest (2 ml/min) for the given time of stimulation. Images were acquired with a cooled CCD camera (CoolSnapHQ, Roper Scientific) during illumination (1 Hz and 100 ms) at 480 ± 20 nm (505 dichroic long pass and 534 ± 25 band-pass) through an optical switch (Sutter Instruments) and analyzed using Metafluor Software (Universal Imaging).
Postsynaptic currents were recorded from hippocampal pyramidal cells (13–17 DIV) in modified Tyrode bath solution in whole cell patch-clamp configuration. Data were acquired with a Multiclamp 700B amplifier interfaced through a Digidata 1322A A/D converter to Clampex 8.2 software (Molecular Devices) filtered at 2 kHz and sampled at 100 μs. For excitatory responses, the internal solution contained (in mM) 115 Cs-MeSO3, 10 CsCl, 5 NaCl, 10 HEPES, 20 tetraethylammonium-Cl, 4 Mg-ATP, 0.3 Na2-GTP, 0.6 EGTA, and 10 lidocaine N-ethyl-bromide, pH 7.35 (∼310 mOsm). For inhibitory responses, the internal solution contained (in mM) 135 CsCl, 10 HEPES, 0.6 EGTA, 4 Mg-ATP, 0.3 Na2-GTP, and 10 lidocaine N-ethyl-bromide, pH 7.35 (∼310 mOsm). For sucrose stimulation, a hypertonic solution prepared by adding 500 mM sucrose to the nominally Ca2+-free Tyrode solution was perfused directly into the region containing the patched cell. Field stimulations (30 mA pulses of 0.1 ms) were applied with parallel platinum electrodes immersed in the bath solution. Ionomycin (Sigma) was applied directly to the bath by pipette and mixed briefly while patch clamping the neurons after a control recording from the same cell without ionomycin.
All error bars indicate SE. Statistical comparisons were made using Student's t-test (2-tailed, heteroscedastic, Microsoft Excel Software) unless otherwise stated, and P < 0.05 was considered significant.
Development of mature SNAP-25–deficient synapses in cultured hippocampal neurons
Previous studies of low-density neuronal cultures from SNAP-25–deficient mice showed that apparently normal, immature synapses were present until ∼7 DIV (Washbourne et al. 2002). Beginning at 8–10 DIV, however, a loss of neuronal processes leading to neuronal degeneration was observed, although some synapses may have persisted in these cultures (Tafoya et al. 2006). To test whether loss of SNAP-25 in itself generally leads to a loss of neuronal viability, we used high-density mixed (neuron and glia) cultures of embryonic hippocampal neurons (Kavalali et al. 1999). In these cultures, SNAP-25–deficient neurons survived for more than three weeks, whereas SNAP-25 KO neurons plated at lower density degenerated at ∼8 DIV (data not shown). In mature hippocampal cultures (Mozhayeva et al. 2002), SNAP-25 KO neurons maintained healthy processes with synapses and synaptic densities that were not significantly different from those observed in control neurons cultured from a mix of heterozygous (majority) and wild-type littermates to avoid artifacts caused by genetic background or culture conditions (Fig. 1A). However, at 21 DIV, we observed a small decrease in synapse density of ∼20% (P = 0.012) in SNAP-25–deficient neurons compared with control neurons (Fig. 1B). Electron micrographs of cultured neurons revealed no significant difference in vesicle distribution between SNAP-25 KO and littermate control neurons, and the distribution of vesicle number per terminal per section as well as vesicle size was the same for both genotypes (Fig. 1C). Overall, SNAP-25 KO cultures, in parallel with our earlier observations in synaptobrevin-2–deficient cultures, lacked signs of neurodegeneration. Typically, neurons in both types of cultures maintained membrane integrity as judged by their high membrane resistances at rest (>200 MΩ). Therefore we surmise that the synapse loss we see in SNAP-25 KOs may be caused by relative lack of synaptic activity.
Analysis of excitatory neurotransmission in SNAP-25–deficient synapses
In SNAP-25–deficient neurons, spontaneous events were significantly reduced in frequency (P < 0.0001) but were not abolished (Fig. 2, A and B). The reduction in spontaneous miniature excitatory postsynaptic current (mEPSC) frequency did not seem to be caused by reduced synaptic contacts (see Fig. 1A) as suggested previously (Washbourne et al. 2002), but is consistent with more recent results on SNAP-25 KO cultures (Tafoya et al. 2006). mEPSCs were reduced to a similar degree as hippocampal cultures of synaptobrevin-2 KO mice (Deak et al. 2004; Schoch et al. 2001), indicating that fusion can still occur in the absence of either synaptobrevin-2 or SNAP-25. The amplitude distribution of mEPSCs is slightly, but significantly (Kolmogorov-Smirnov test), smaller in SNAP-25–deficient neurons than in control neurons (Fig. 2C). Because there was no significant change in the size of synaptic vesicles in SNAP-25–deficient synapses (Fig. 1C) and we did not observe the same decrease in mIPSCs, the decreased mEPSC amplitudes may reflect a decrease in the number of postsynaptic glutamate receptors. This premise is consistent with earlier work suggesting a role for SNAP-25 in N-methyl-d-aspartate (NMDA) receptor trafficking (Lan et al. 2001). Although we were not recording NMDA currents, it is possible that SNAP-25 also has a role in trafficking AMPA receptors. This potential role of SNAP-25 in postsynaptic receptor trafficking seems to be mostly compensated because application of kainate or NMDA to SNAP-25–deficient neurons did not elicit a significantly reduced response (Washbourne et al. 2002). We also observed no significant reduction in the response to glutamate puffed onto SNAP-25 KO neurons (1,056.6 ± 162.5 pA, n = 7 cells compared with controls: 1,169.2 ± 115.0 pA, n = 5 cells).
We next examined evoked neurotransmitter release in mature SNAP-25–deficient neurons, using extracellular field stimulation (30 mA, 0.1 ms). Two thirds of SNAP-25–deficient neurons (21/31) exhibited no synaptic responses (Fig. 3A), whereas one third (10/31) initially displayed EPSCs (Fig. 3A, bottom trace), but the synaptic currents quickly disappeared (Fig. 3B). The lack of evoked release in the mutant neurons was not caused by a failure of action potentials because normal action potentials (Fig. 3A, right) could be triggered with current injection. The threshold potential in control neurons, −43.7 ± 1.2 mV, was similar to SNAP-25 KO neurons, −45.0 ± 1 mV. Figure 3B shows diary plots of two sample cells from control, SNAP-25 KO, and synaptobrevin-2 KO cultures. Averages for the first EPSC of each of the responding SNAP-25 KO neurons are plotted to the right of the diary plots. We could only average the first responses because of the rapid depression, even at low-frequency (0.2 Hz) stimulation. In this set of experiments, we also tested synaptobrevin-2 KO neurons from a similar culture obtained at the same time. This direct comparison of SNAP-25–and synaptobrevin-2–deficient cultures revealed that calcium-triggered release is more defective in SNAP-25 KO neurons than in synaptobrevin-2 KOs.
Despite the low probability of stimulating neurotransmitter release with field stimulation, hypertonic sucrose (500 mOsm) always triggered neurotransmitter release in SNAP-25–deficient neurons. Neurons that did not respond during field stimulation reliably responded to hypertonic sucrose application. Hypertonic sucrose-elicited responses, however, were dramatically reduced as charge transfer during the peak response was 6.9% of the peak charge transfer in controls (47.02 ± 14.9 vs. 676.66 ± 167.3 pC; Fig. 3C). The total charge over the interval graphed in Fig. 3C for mutant neurons was 1.6 nC, 12% of the control value of 13.4 nC. The SNAP-25 KO responses represented actual neurotransmitter release because they could be blocked by the addition of CNQX and AP-5 to the bath solution already containing picrotoxin (Fig. 3C, bottom inset). Earlier experiments in hippocampal cultures obtained from SNAP-25 KOs did not reveal significant vesicle mobilization in response to hypertonic sucrose stimulation (Tafoya et al. 2006). This discrepancy may arise from the relative maturity of cultures used here compared with earlier work. We did record sucrose responses from SNAP-25–deficient neurons at 8 DIV and did not observe a significant response beyond a small number of mEPSC events. In the absence of SNAP-25, synapse maturation may be delayed leading to a late onset of responsiveness to hypertonic sucrose (Mozhayeva et al. 2002).
Analysis of inhibitory neurotransmission in SNAP-25–deficient synapses
Because SNAP-25 is part of the basic core fusion machinery, one would expect that it should also participate in neurotransmitter release from inhibitory synapses. Past studies, however, have raised conflicting results about this hypothesis. It has been suggested that SNAP-25 is absent from hippocampal GABAergic synapses and is substituted with SNAP-23 (Frassoni et al. 2005; Verderio et al. 2004). However, recent analysis showed co-localization of SNAP-25 and inhibitory neuron specific markers in several mouse brain regions (Tafoya et al. 2006). There was also no evoked release from GABAergic neurons in cortical slices from SNAP-25 KO embryos. Furthermore, GABAergic terminals in SNAP-25 KO hippocampal cultures could not be loaded with FM1-43, indicating a defect in release from GABAergic and glutamatergic synapses lacking SNAP-25.
We compared SNAP-25–deficient excitatory and inhibitory neurotransmitter release directly in hippocampal cultures by recording inhibitory responses in the same type of hippocampal cultures we used for analyzing excitatory responses. To maximize the amplitude of inhibitory currents, we used a cesium chloride internal solution in the whole cell patch pipette to shift the chloride reversal potential to 0 mV and held the cells at −70 mV.
When we recorded spontaneous inhibitory events from SNAP-25 KO hippocampal neurons, the reduction in spontaneous event frequency was similar to that seen in excitatory synapses (Fig. 4, A–C). As in excitatory synapses, evoked release from inhibitory synapses was severely impaired, with only 3/15 cells responding (Fig. 5, A and B). Hypertonic sucrose was always able to induce synaptic release, but similar to excitatory sucrose responses, inhibitory responses were reduced in SNAP-25 KO neurons (Fig. 5C). Therefore our results add support to an equally critical role for SNAP-25 in excitatory and inhibitory neurotransmission, at least throughout the extensive synaptic maturation process enabled in these mixed hippocampal neuronal/glial cultures.
Effect of increased calcium on neurotransmission in the absence of SNAP-25
Because calcium-triggered neurotransmitter release in SNAP-25 KOs was more affected than that evoked by a hyperosmotic stimulus, we tested the possibility that the low probability of release could be overcome with increased calcium either from increased extracellular concentrations or accumulation of residual intracellular calcium during higher frequency stimulation (Capogna et al. 1997). Consistent with this idea, synaptobrevin-2 KO neurons, which also are severely affected in calcium-evoked release, showed a facilitation of release during 5-, 10-, or 30-Hz stimulation (Deak et al. 2004). As mentioned above, in the SNAP-25 KO cultures, evoked neurotransmission was detectable only in a small number of cases, and a 10-Hz train of 200 stimuli did not increase the probability of neurotransmitter release in cells that did not show evoked synaptic responses (Fig. 6A). Trains of stimuli at 30 Hz gave the same results (data not shown). Ten-hertz stimulation of the few responding cells in SNAP-25 KO cultures (Fig. 6, A, bottom traces, and B, bottom) failed to induce facilitation. We always saw a depression of release even at 1 Hz (Fig. 6B, top). For cells that had no evoked response, raising extracellular calcium concentrations to 5 or 10 mM did not increase the probability of evoking neurotransmitter release (data not shown).
Given the rare occurrence of evoked neurotransmitter release in SNAP-25 KOs, we analyzed the more reliable spontaneous release to gain further information about the remaining calcium sensitivity. When we compared spontaneous mEPSCs at 10 mM extracellular calcium to 2 mM calcium, we found a slight increase in frequency of SNAP-25 KO spontaneous events (Fig. 7A), which did not reach significance (P = 0.2). There was, however, a clear increase in mEPSC frequency in littermate control neurons (P = 0.016). After the weak effect of residual calcium and raising extracellular calcium, we reasoned that we might have to bypass the calcium channels to evaluate any residual effect of calcium on release in SNAP-25 KO neurons.
To bypass calcium channels, we applied the calcium ionophore ionomycin (2 μM) onto the SNAP-25 KO cultures. We first recorded ionomycin responses in the presence of 0.2 mM extracellular calcium to determine the effect of moderate calcium concentrations. Similar to controls, SNAP-25 KO neurons showed a significant increase in mEPSC frequency after ionomycin application (Fig. 7B). Increasing the extracellular calcium to 2 mM, however, caused a greater relative increase in both genotypes, but the SNAP-25 KO neurons still showed a reduced response (Fig. 7C). This result indicates that the defect in calcium sensitivity of release is attributable to the calcium sensing or calcium transduction mechanism of the release machinery because bypassing calcium channels does not overcome the deficit in calcium-dependent neurotransmitter release. Alternatively, the ∼7-fold increase in event frequency seen in the SNAP-25 KOs during ionomycin application in 2 mM calcium represents the limit of fusable vesicles. This possibility seems unlikely because hypertonic sucrose-induced release in mutants (total charge = 1,609.7 pC) exceeds the level of ionomycin-induced release for the ionomycin concentration we used (total charge = 5.1 pC). Hypertonic sucrose application is typically more effective in mobilizing a large fraction of readily releasable vesicles compared with Ca2+-dependent mechanisms (Moulder and Mennerick 2005).
Analysis of endocytosis in the absence of SNAP-25
Exo- and endocytosis are tightly coupled to maintain reliable neurotransmitter release and the integrity of the synaptic plasma membrane. In view of this coupling, one would expect that a severe deficit in exocytosis would lead to very little endocytosis, if any. However, significant endocytosis has been observed after a block of exocytosis with botulinum toxin A, which cleaves SNAP-25 at the C terminus (Neale et al. 1999). The particular cleavage by this toxin uniquely uncoupled exo- and endocytosis.
To determine whether the complete loss of SNAP-25 uncouples exo- and endocytosis, or merely reduces endocytosis to match the severely reduced exocytosis, we loaded the synaptic vesicles of mutant synapses with HRP during application of 90 mM K+ and visualized the HRP containing vesicles with an electron microscope (Heuser and Reese 1973). Although sucrose application would have induced more reliable endocytosis, the hypertonicity would have altered the morphology of the synapses. Sample electron micrographs show synaptic terminals containing HRP particles (arrows) without stimulation and after stimulation with 90 mM K+ (Fig. 8A). Figure 8B shows that, even with no stimulation, SNAP-25 KO terminals contain an average number of HRP-labeled vesicles similar to littermate controls. Upon stimulation with 90 mM K+, however, there was no significant increase in the average number of HRP-labeled vesicles per mutant terminal in contrast to control terminals (Fig. 8B), showing that the degree of endocytosis is still coupled to the very low rate of exocytosis.
Analysis of FM dye uptake and release in SNAP-25 KOs
Even though elevated extracellular potassium was not able to increase the amount of HRP uptake, we surmised that the ability of FM dyes to partition into membranes might increase the efficiency of FM dye uptake relative to HRP. We stimulated the cultures for 90 s with 47 mM K+ in the presence of FM2-10 or FM1-43. After this dye labeling protocol, we observed clear FM stained puncta in control cultures (arrows), but only a few dim puncta in SNAP-25 KO cultures (Fig. 9A). In response to 90 mM K+ stimulation, neither FM2-10 nor FM1-43 (data not shown) could be detectably released from the FM-positive puncta in SNAP-25 KO synapses. Furthermore, these few FM dye-labeled spots were unresponsive to hypertonic sucrose stimulation, which causes relatively robust exocytosis in SNAP-25–deficient cultures (Fig. 9A). FM dye-labeled puncta in SNAP-25 KO cultures had relative ΔF values similar to values resulting from the application of a wash solution (4 mM K+, nominally Ca2+-free modified Tyrode's solution). The relative ΔF was calculated by subtracting an average of three consecutive fluorescence values 30 s after the point of a sharp drop in fluorescence from a baseline fluorescence average. Therefore similar to HRP uptake, these FM dye positive puncta may either correspond to sites of high spontaneous vesicle turnover (Sara et al. 2005) or they may originate from non–synapse-specific staining that would not be expected to respond to stimulation. The absence of stimulation-evoked endocytosis prompted us to use electrophysiology and test neurotransmitter release during elevated K+ stimulation, which can trigger release from synaptobrevin-2–deficient synapses (Deak et al. 2004; Schoch et al. 2001). When we recorded EPSCs during application of 47 mM K+ (Fig. 9B), wild-type synapses showed a significant amount of release (black trace) superimposed over a depolarizing current (gray trace) detected on the cell held in whole cell voltage clamp. This depolarizing current is independent of neurotransmission as it is insensitive to blockade of all neurotransmitter receptors. Similar to field stimulation, SNAP25 KO synapses showed nominal release during stimulation with 47 mM K+ (Fig. 9B, bottom traces) because there was little difference between the response with receptor blockers and the response with only picrotoxin.
Because absence of unequivocal dye uptake with high K+ stimulation did not allow analysis of dye release, we took advantage of hypertonic sucrose stimulation, which results in significant neurotransmitter release from SNAP-25–deficient synapses. We needed to load and destain with hypertonic (+500 mOsm sucrose) stimulation to observe a convincing destaining profile in the SNAP-25 KO synapses (Fig. 10). Figure 10A shows FM1-43–positive SNAP-25 KO puncta (arrows) that surprisingly have an average initial absolute fluorescence value that is similar to controls. These SNAP-25–deficient puncta showed a drop in fluorescence (ΔF) in the first 30 s of stimulation with hypertonic sucrose that is significantly different from the ΔF during application of the wash solution (Fig. 10A, right). It should be noted that the SNAP-25 FM1-43–positive puncta loaded on sucrose stimulation could not be destained with 90 mM K+ (data not shown). Measuring the actual size of the synaptic vesicle pool labeled by application of hypertonic sucrose requires repeated destaining to mobilize all labeled vesicles and completely release all the trapped dye. The inefficacy of high K+ stimulation on SNAP-25 KO synapses and impracticality of repeated destaining with hypertonic sucrose application hampered our ability to reliably measure synaptic vesicle pool sizes. Therefore we used relative fluorescence difference values (ΔF) to compare hypertonic sucrose-induced dye release after hypertonic sucrose-induced dye uptake. The relative ΔF was calculated as in Fig. 9 (Fig. 10A, inset). It is surprising that the absolute initial fluorescence and ΔF for SNAP-25 KO synapses is similar to control synapses given that the EPSCs during hypertonic sucrose reached a peak average current that was ∼10% of control currents. This is presumably because of a selection bias toward brightly stained fluorescent puncta in the SNAP-25 KO cultures for FM dye analysis.
Earlier work has shown that wild-type synapses release endocytosed FM1-43 dye slowly during hypertonic sucrose application, presumably because the rapid retrieval of exocytosed vesicles in the readily releasable pool gives the dye little time to dissociate from the membrane (Pyle et al. 2000). FM2-10, which dissociates more rapidly (Klingauf et al. 1998; Ryan et al. 1996; Schote and Seelig 1998), is released more swiftly from vesicles during hypertonic sucrose application. In these experiments, we compared the rate of dye release using F/F0 plots from KO and wild-type synapses because it is not possible to completely destain KO synapses to normalize our measurements to the total pool size. This analysis showed that SNAP-25 KO synapses loaded with FM1-43 dye by hypertonic sucrose, release FM1-43 during hypertonic sucrose application at a slightly faster rate (τfast = 9.8 ± 2.9 s) than controls (23.7 ± 3.3 s), which did not reach significance (P > 0.1; Fig. 10, B and C). The more rapidly dissociating FM2-10 is released with the same τfast in SNAP-25 KO synapses (2.5 ± 0.3 s) as controls (2.7 ± 0.5 s). This result is different from reported in the synaptobrevin-2 KOs in which the rate of FM2-10 dye release was not different from FM1-43 (Deak et al. 2004). Apparently, the loss of SNAP-25 does not modify fusion pore openings or the rate of endocytosis to a degree that would alter FM dye release kinetics significantly.
Analysis of the loss of the SNARE protein synaptobrevin-2 in mice indicated that its function encompasses more than membrane fusion (Deak et al. 2004, 2006a; Schoch et al. 2001). In this study, we analyzed hippocampal synapses lacking the SNARE protein SNAP-25 with similar techniques to determine functions unique to SNAP-25. Previous analysis of immature neurons lacking SNAP-25 reported a loss of evoked neurotransmitter release, whereas some low level of spontaneous release remained (Tafoya et al. 2006; Washbourne et al. 2002). We found that 2-wk-old SNAP-25 KO neurons had a reduced frequency of spontaneous miniature events, almost no evoked release, and nominal calcium sensitivity. The phenotypes of the SNAP-25 KO neurons were similar to the synaptobrevin-2 KOs in superficial aspects of neurotransmitter release, but detailed analysis revealed three main differences between the two mutants (Table 1). First, although synaptobrevin-2 KO neurons have significantly reduced evoked currents, all neurons exhibited some responses, whereas most of the SNAP-25 KO neurons did not respond at all when stimulated with field stimulation or high K+ stimulation. This lack of responsiveness to calcium-evoked neurotransmitter release cannot be caused by a postsynaptic defect because we can reliably record spontaneous and hypertonic sucrose-induced postsynaptic currents. Second, the facilitation of neurotransmitter release seen in synaptobrevin-2 KOs during stimulation frequencies of 5 Hz or above is absent in the SNAP-25 KOs. Third, SNAP-25 KO synapses release FM1-43 and FM2-10 at different rates matching controls; in contrast, synaptobrevin-2 KO synapses release each dye at the same rate. Such differences hint at potential functions of SNAP-25 not shared by synaptobrevin-2 as well as functions unique (among SNAREs) to synaptobrevin-2. Because SNAREs are thought to drive fusion as a tightly bound complex, unique functions suggest SNAREs have additional roles before and after fusion.
Critical role of SNAP-25 in calcium-evoked release
We found that the average amplitude of evoked EPSCs and IPSCs (in the few cells that respond) is ∼2–3% of control amplitudes (Figs. 3 and 5; including the “0” responses this would amount to 0.5–1%), whereas the frequency of spontaneous events is ∼14% of controls (Figs. 2 and 4). Hypertonic sucrose reliably induced neurotransmitter release that was ∼10% of controls (Figs. 3 and 5). Therefore, although the rate of synaptic vesicle fusion is by no means normal, it is still easily detectable. SNAP-25 seems to have a role in calcium sensitivity that can only be moderately compensated by flooding the terminals with calcium (Fig. 7).
The deficit in calcium sensitivity is consistent with the strictly calcium-dependent interaction of synaptotagmin-1 with SNARE complexes. This interaction is disrupted by complexins competing with synaptotagmin-1/SNARE complex binding during fast synchronous neurotransmitter release (Tang et al. 2006). Perhaps the relatively reduced effect of deleting SNAP-25 or synaptobrevin-2 on spontaneous release is because spontaneous release does not depend on the interaction of synaptotagmin with SNAREs. An insertion of 12 amino acids between the SNARE motif and the transmembrane domain of synaptobrevin-2 fully rescues spontaneous miniature frequency in synaptobrevin-2 KO cultures, but only marginally improves calcium-evoked release suggesting that spontaneous release has a less stringent requirement for SNAREs (Deak et al. 2006a).
Cleavage near the end of the C-terminus of SNAP-25 by botulinum toxins A or C reduces calcium sensitivity, but increasing calcium partially overcomes this deficit (Capogna et al. 1997; Sakaba et al. 2005; Trudeau et al. 1998). Mutant constructs of SNAP-25 with a truncated C-terminus were only able to partially restore evoked neurotransmitter release in botulinum toxin E treated terminals under conditions of elevated extracellular calcium or paired pulse stimulation (Finley et al. 2002). In the complete absence of SNAP-25, we found that elevating calcium did not improve the probability of evoked neurotransmitter release. All we observed was a moderate increase in spontaneous release with ionomycin (Fig. 7). Our results suggest that other regions of SNAP-25 may interact with the calcium-sensing machinery, because removing the whole protein reduces calcium sensitivity more severely than just removing the end of the C-terminus.
SNAP-25 has also been shown to interact with the rab3 effector protein, rabphillin, through its C2B domain (Deak et al. 2006b). Neurons from the double KO of synaptobrevin-2 and rabphillin show depression of neurotransmitter release during 10-Hz stimulation rather than facilitation. Our results in this study show that removing SNAP-25 yields a similar depression of release presumably because rabphillin cannot interact with SNAP-25 to limit release and allow facilitation.
Synaptobrevin-2 KO neurons also show a change in calcium sensitivity, but their evoked release is more robust than in SNAP-25 KOs, and the facilitation during high-frequency stimulation indicates that elevated residual calcium can moderately overcome this deficit. These distinctions between SNAP-25 and synaptobrevin-2 are consistent with experiments in the calyx of Held showing that tetanus toxin cleaved synaptobrevin did not affect the peak release rate per vesicle with caged calcium stimulation (Sakaba et al. 2005). Cleavage of SNAP-25 with botulinum toxin A under the same conditions, on the other hand, dramatically reduced the peak release rate (Sakaba et al. 2005).
Exocytosis/endocytosis coupling in SNAP-25 KO mice
In SNAP-25 KOs, we found no severe deficit in endocytosis beyond what would be expected from very limited exocytosis (Fig. 8). Because elevated potassium triggers minimal release (Figs. 8B and 9B), the lack of increased endocytosis of HRP during stimulation is likely caused by insufficient exocytosis. Endocytosis does increase with more robust exocytosis in SNAP-25 KOs, as we see in Fig. 10, using a stimulus of hypertonic sucrose to load and destain FM dyes. Therefore the coupling of exo- and endocytosis seems to be intact in the absence of SNAP-25.
Under conditions where we see a convincing release of FM dyes above background, the relative rate of FM dye release in SNAP-25 KO synapses is comparable with controls (Fig. 10). This result suggests that, at the resolution of FM dye kinetics, in the SNAP-25 KO there is no significant difference in the time the luminal membrane of synaptic vesicles is exposed to the external medium compared with control vesicles. However, we cannot exclude a subtle role of SNAP-25 in the regulation of fusion pore opening kinetics. Amperometry data recording release from SNAP-25 KO chromaffin cells showed a shorter foot preceding the opening of the fusion pore (Sorensen et al. 2003), which suggests the fusion pore might be open longer in the absence of SNAP-25. The longer open time may not be sufficient to increase the rate of FM1-43 release from SNAP-25 KO synapses. Recently, it has been suggested that serotonin-mediated presynaptic inhibition of the lamprey glutamate-releasing reticulospinal/motorneuron giant synapse occurs through the action of Gβγ on the C-terminus of SNAP-25, limiting fusion pore openings as measured by FM1-43 release and neurotransmitter release (Gerachshenko et al. 2005; Photowala et al. 2006). This earlier result implies that loss of SNAP-25 might diminish FM dye release. In our system under control conditions, FM1-43 release seems to be impaired substantially, which makes detection of a further decrease in dye release difficult. In contrast to SNAP-25–deficient synapses, synaptobrevin-2 KO synapses have a defect in fast endocytosis and release FM1-43 and FM2-10 at the same rate (Deak et al. 2004).
A major issue in analyzing KO neuronal cultures is the possibility of compensation by another protein in the same family as the missing protein, or another isoform of the same protein. In the case of SNAP-25, both isoforms are knocked out in the mice analyzed (Washbourne et al. 2002), but SNAP-23, SNAP-29, and SNAP-47 are also expressed in mice. Could the residual neurotransmitter release be caused by these other members? In the Drosophila neuromuscular junction, null mutants of SNAP-25 do not exhibit a severe phenotype because SNAP-24 is able to compensate almost completely (Vilinsky et al. 2002). In contrast, the fast release component of chromaffin cells from SNAP-25 KO mice could not be rescued by exogenous expression of SNAP-23 (Sorensen et al. 2003). In this same study, overexpression of SNAP-23 in wild-type chromaffin cells inhibits fast release. Furthermore, studies of SNAP-29 overexpression in hippocampal neurons shows an inhibition of neurotransmission (Pan et al. 2005). SNAP-47 forms ternary complexes with syntaxin-1a and synaptobrevin-2 and facilitates fusion of two populations of liposomes containing either syntaxin-1a or synaptobrevin-2; however, the fusion reaction is slower compared with SNAP-25, and SNAP-47 is not primarily expressed at the plasma membrane of neurons (Holt et al. 2006). Synaptobrevin, on the other hand, seems to be more interchangeable with the other members of its protein family. In Drosophila, the non-neuronal synaptobrevin is able to rescue the phenotype in synaptic transmission caused by a deletion of the neuronal synaptobrevin (Bhattacharya et al. 2002). This result suggests that even a synaptobrevin that normally participates in slow constitutive membrane trafficking pathways is able to participate in fast fusion events. In mice, the synaptobrevin-related protein, cellubrevin, can rescue evoked and spontaneous release in synaptobrevin-2 KO synapses (Deak et al. 2006a). A similar degree of rescue with cellubrevin and synaptobrevin-2 has also been shown in chromaffin cells of cellubrevin/synaptobrevin-2 double KO mice (Borisovska et al. 2005). Although the phenotypic differences between SNAP-25 and synaptobrevin-2 KOs might possibly be caused by relatively better compensation in synaptobrevin-2 KOs, we propose that, if this were the case, the phenotypes would merely be less severe in synaptobrevin-2 KOs and not selectively different. We cannot, however, rule out the possibility that the different phenotypes are caused by the relative differences between the protein(s) that might possibly be mediating the remaining exocytosis.
The ubiquitous involvement of SNAREs and related proteins in membrane trafficking throughout evolution suggests they play an integral role in the merging of membranes. By necessity, these fusion events must be regulated, and neurotransmitter release may be one of the most highly regulated events. At this point, it is difficult to determine whether different apparent functions of individual SNAREs are just a result of different regulation. It will be necessary, therefore to specifically disrupt particular regulatory interactions and perhaps identify additional regulatory components to address this question.
This work was supported by the National Institute of Mental Health Grants MH-066198 to E. T. Kavalali, MH-070207 to P. Bronk, and MH-048989 to M. C. Wilson. T. C. Südhof is a Howard Hughes Medical Institute investigator. E. T. Kavalali is an Established Investigator of American Heart Association.
We thank A. Roth, I. Kornblum, and E. Borowicz for excellent technical assistance and J. Mitchell for help with mouse husbandry.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 by the American Physiological Society