Report

Reconstitution of Ca2+-Regulated Membrane Fusion by Synaptotagmin and SNAREs

See allHide authors and affiliations

Science  16 Apr 2004:
Vol. 304, Issue 5669, pp. 435-438
DOI: 10.1126/science.1097196

Abstract

We investigated the effect of synaptotagmin I on membrane fusion mediated by neuronal SNARE proteins, SNAP-25, syntaxin, and synaptobrevin, which were reconstituted into vesicles. In the presence of Ca2+, the cytoplasmic domain of synaptotagmin I (syt) strongly stimulated membrane fusion when synaptobrevin densities were similar to those found in native synaptic vesicles. The Ca2+ dependence of syt-stimulated fusion was modulated by changes in lipid composition of the vesicles and by a truncation that mimics cleavage of SNAP-25 by botulinum neurotoxin A. Stimulation of fusion was abolished by disrupting the Ca2+-binding activity, or by severing the tandem C2 domains, of syt. Thus, syt and SNAREs are likely to represent the minimal protein complement for Ca2+-triggered exocytosis.

Soluble N-ethylmaleimide–sensitive factor attachment protein receptors (SNAREs) are thought to form the minimal machinery needed to mediate intracellular membrane fusion (1). In neurons, the SNARE complex is composed of the target membrane SNAREs (t-SNAREs), syntaxin and SNAP-25, and the vesicle membrane SNARE (v-SNARE), synaptobrevin (syb) (2). Reconstituted v-SNARE vesicles fuse with t-SNARE vesicles during the assembly of cognate trans-SNARE complexes (1, 3). Fusion occurs with a half-time on the order of minutes (4) and is not regulated by Ca2+.

At synapses, exocytosis of neurotransmitters is strictly controlled by Ca2+ (5). The integral membrane protein, synaptotagmin I, binds Ca2+ and has been proposed to function as the Ca2+ sensor that triggers rapid release (6, 7). Ca2+ promoted interactions of synaptotagmin with anionic phospholipids and t-SNAREs have emerged as putative coupling steps in evoked secretion (68). Here, we directly addressed this issue by examining the ability of synaptotagmin to regulate SNARE-catalyzed membrane fusion, using a defined reconstituted model system (1). In this assay, lipidic fluorescence resonance energy transfer (FRET) donor and acceptor pairs are incorporated into v-SNARE vesicles. Fusion of the labeled v-SNARE vesicles with unlabeled t-SNARE vesicles results in dilution of the donor and acceptor. Thus, fusion can be monitored by following the increase in donor fluorescence (1, 9, 10).

The cytoplasmic domain of synaptotagmin I (syt) was mixed with t-SNARE and v-SNARE vesicles at 37°C. In the presence of Ca2+, syt markedly enhanced both the rate and extent of fusion (Fig. 1A). In the absence of Ca2+, syt suppressed fusion to a small but reproducible extent. Thus, Ca2+ triggered a transition in the activity of syt. In the absence of syt, fusion was largely unaffected by Ca2+.

Fig. 1.

Reconstitution of Ca2+-dependent membrane fusion. (A) Syt (10 μM) stimulates SNARE-mediated fusion between v-SNARE vesicles (∼90 syb/vesicle) and t-SNARE vesicles (∼80 syntaxin/SNAP-25 complexes per vesicle) in the presence of 1mM Ca2+ but not in the presence of 0.2 mM EGTA. Raw donor fluorescence (left) was normalized by using the maximum fluorescence obtained after addition of the detergent n-dodecylmaltoside (0.5%; open arrowhead) and converted to rounds of fusion (right), as described (4, 10). (B) v-SNARE vesicles were reconstituted with varying amounts of syb. Purified vesicles (7.5 μl) were subjected to SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and visualized by staining with Coomassie blue. The asterisk denotes proteolyzed syb. (C) t-SNARE vesicles (∼80 syntaxin/SNAP-25 complexes per vesicle) were incubated with v-SNARE vesicles containing the indicated amounts of syb in the presence of increasing concentrations of syt in either 1 mM Ca2+ (open circles) or 0.2 mM EGTA (filled circles). Plots depict the total amount of fusion obtained after 2 hours at 37°C as a function of (syt). (D) t-SNARE vesicles reconstituted with varying amounts of syntaxin and SNAP-25. Purified vesicles (5 μl) were subjected to SDS-PAGE and visualized by staining with Coomassie blue. (E) v-SNARE vesicles (∼90 syb/vesicle) were incubated with t-SNARE vesicles containing 0, ∼30, ∼80, or ∼250 copies of syntaxin/SNAP-25 complexes per vesicle in the presence of increasing [syt] plus 1 mM Ca2+. The amount of fusion (t = 120 min) was plotted as a function of [syt] (left) or normalized to the maximum amount of fusion obtained for each t-SNARE vesicle condition (right).

The ability of syt to stimulate membrane fusion was dependent on the density of syb on the v-SNARE vesicles (Fig. 1, B and C). Ca2+·syt stimulation of membrane fusion was most evident at or below ∼150 syb/vesicle, similar to densities reported for native synaptic vesicles (11, 12). At much higher densities of syb, syt inhibited fusion. However, at ∼760 syb/vesicle, the mass ratio of syb to lipids would be ∼1: 2; thus, the surface of the liposome would be strongly influenced by molecular crowding, and the observed inhibition might be secondary to complex effects of nonphysiological levels of syb.

Ca2+·syt stimulated SNARE-mediated fusion at all t-SNARE densities tested (Fig. 1, D and E). Increasing the number of t-SNARE complexes per vesicle did, however, require higher concentrations of syt to saturate the reaction (Fig. 1E), which suggested that syt acted, at least in part, via direct interactions with the t-SNAREs, syntaxin and SNAP-25.

Substitution of either v-SNARE or t-SNARE vesicles with protein-free vesicles resulted in complete loss of fusion; under these conditions, syt and Ca2+ were without effect (Fig. 1, C and E). Furthermore, fusion in the presence and absence of Ca2+·syt was efficiently blocked by the cytoplasmic domain of syb (syb1-94) (Fig. 2A), presumably because it prevents trans SNARE pairing (1). The syb1-94 dose responses in the presence and absence of syt were identical. Thus Ca2+·syt acts by facilitating the SNARE-mediated fusion pathway, rather than by causing lipid mixing via an alternative mechanism. Ca2+·syt stimulation of membrane fusion was also inhibited by soluble t-SNARE complexes and by cleavage of reconstituted syb by botulinum neurotoxin B (BoNT/B) (Fig. 2B) (13).

Fig. 2.

Ca2+·syt stimulation of membrane fusion proceeds through the formation of fusion-competent SNARE complexes. (A) v-SNARE and t-SNARE vesicles were incubated with increasing amounts of the cytoplasmic domain of syb (syb1-94) in the absence (filled circles) or presence (open circles) of syt (10 μM); 1mMCa2+ was present throughout all reactions. The total amount of fusion (t = 120 min) was plotted as a function of [syb1-94]; data were also normalized to the maximum amount of fusion obtained in the absence of syb1-94 (inset). (B) BoNT/B treatment of v-SNARE vesicles, or incubation with soluble t-SNAREs (syx1-265/SNAP-25; 5 μM), leads to loss of membrane fusion both in the absence (black bars) and presence of syt (white bars); 1 mM Ca2+ was present in all the samples. (Inset) Untreated or BoNT/B-treated v-SNARE vesicles (7.5 μl) were subjected to SDS-PAGE, and protein was visualized by staining with Coomassie blue; syb was efficiently cleaved by the toxin (13). Some syb was protected from digestion because of the lumenal orientation of its cytoplasmic domain (1). Error bars indicate standard deviations for three independent determinations.

We incubated a mutant version of syt, in which the Ca2+-binding sites in both C2 domains had been disrupted (1416), with v- and t-SNARE vesicles. The Ca2+ ligand mutant (sytCLM) was completely unable to stimulate fusion in the presence of Ca2+ (Fig. 3A). Thus, Ca2+ binding to syt is critical for regulation of SNARE-mediated fusion.

Fig. 3.

Stimulation of fusion involves the Ca2+- and t-SNARE–binding activities of syt. (A) v-SNARE and t-SNARE vesicles were incubated with increasing amounts of wild-type syt (circles), syt that harbors Ca2+-ligand mutations (sytCLM) in both C2 domains (C2A: D232, 230N; C2B: D363, 365N) (squares, left), and a mixture of isolated C2A and C2B domains (diamonds, right; [syt] refers to the concentration of each C2 domain). The amount of stimulation obtained after 2 hours as compared with control (–syt, +Ca2+) was plotted as a function of the [syt]. (B) Syt was incubated with vesicles before mixing with Accudenz density medium. The mixture was overlaid with decreasing concentrations of Accudenz. After centrifugation, the vesicles floated to the 0/30% Accudenz interface along with any bound syt. Samples collected from the 0/30% Accudenz interface were analyzed by SDS-PAGE and stained with Coomassie blue. Unbound syt remained in the bottom portion of the tube. (C) Syt (10 μM) cofloated with vesicles composed of 15% PS, 85% PC (PS/PC) in the presence of Ca2+ (1 mM) but did not cofloat with vesicles composed 100% PC (PC). (D) The indicated syt constructs were incubated with t-SNARE vesicles (100% PC) or protein-free (pf) vesicles (100% PC) in the presence or absence of 1 mM Ca2+; samples were applied to density gradients as described in (B). (E) The t-SNARE–binding activity shown in (D) was quantified by densitometry.

The tandem C2 domains of syt must be tethered together in order for syt to bind t-SNAREs—severing the linker that connects C2A and C2B disrupts t-SNARE–binding activity (17, 18). Isolated C2A and C2B failed to stimulate fusion, even when added simultaneously (Fig. 3A). Because isolated C2A retains robust lipid-binding activity (18), stimulation of fusion by Ca2+·syt is not due to perturbation of the vesicle through, for example, coating of membranes with C2 domains. These data are consistent with the idea that Ca2+·syt stimulates fusion, at least in part, via interactions with t-SNAREs.

The interaction of syt with membrane-embedded t-SNAREs has not been measured directly. To assay for these interactions we used a coflotation assay (Fig. 3B). When the vesicles were composed of 15% phosphatidylserine (PS) and 85% phosphatidylcholine (PC), syt efficiently cofloated with vesicles in a density gradient, which reflected its ability to bind anionic phospholipids in the presence of Ca2+ (Fig. 3C) (7). Syt did not bind nor cofloat with vesicles composed of 100% PC. Next, t-SNARE vesicles that had been reconstituted in 100% PC were incubated with syt. Syt efficiently bound the reconstituted t-SNARE vesicles in the presence of Ca2+; only weak binding was observed in the absence of Ca2+ (Fig. 3D). Disruption of the Ca2+-binding sites of syt abolished Ca2+-stimulated coflotation with t-SNARE vesicles. C2A exhibited no detectable binding, whereas C2B exhibited a faint amount of Ca2+-dependent binding, similar to levels previously shown in pull-down experiments (17, 18). Thus, syt efficiently and stoichiometrically binds to t-SNAREs that are embedded in lipid bilayers, and loss of Ca2+-dependent t-SNARE–binding activity is correlated with a loss of Ca2+-dependent stimulation of membrane fusion (Fig. 3E).

We explored the Ca2+ requirements for fusion in the minimal regulated fusion assay. When we used vesicles that contained 15% PS, Ca2+·syt stimulated fusion with a [Ca2+]1/2 of 129 μM (Fig. 4A); increasing the PS to 25% shifted the [Ca2+]1/2 to 82 μM. These findings agree with previous studies showing that the apparent affinity of syt for Ca2+ is strongly dependent on the mole fraction of PS (7). Thus, the interaction of syt with membranes, and the ability of syt to simulate membrane fusion in vitro, occurs at physiologically relevant [Ca2+] (6).

Fig. 4.

The Ca2+-dependence of syt-mediated stimulation of fusion can be modulated by changes in membrane composition or by C-terminal truncation of SNAP-25. (A) v-SNARE and t-SNARE vesicles, reconstituted with either 15% (circles) or 25% (squares) PS, were incubated in the presence (open symbols) or absence (filled symbols) of 10 μM syt at the indicated [Ca2+]. The total amount of fusion (t = 120 min) was plotted as a function of [Ca2+]. At 15% PS, the [Ca2+]1/2 (as indicated by the dashed line) was 129 μM, and the Hill slope was 1.9; at 25% PS the [Ca2+]1/2 was 82 μM, and the Hill slope was 1.0. (Inset) v-SNARE (7.5 μl) and t-SNARE (5 μl) vesicles were subjected to SDS-PAGE, and protein was visualized by staining with Coomassie blue. (B) We generated t-SNARE complexes using either wild-type SNAP-25 or truncated SNAP-25 constructs that correspond to the BoNT/A and BoNT/E cleavage products (residues 1 to 197 and 1 to 180, respectively) (13). The t-SNARE complexes were reconstituted into vesicles, and fusion assays were carried out in the presence (left) or absence (right) of 10 μM syt at the indicated [Ca2+]. (Inset) t-SNARE vesicles (5 μl) were subjected to SDS-PAGE and stained with Coomassie blue. Error bars indicate standard deviations from three independent determinations.

Finally, we asked whether alterations in t-SNAREs could impact the Ca2+ sensitivity of the fusion reaction. Cleavage of SNAP-25 by BoNT/A causes a reduction in secretion that can be overcome, at least in part, by elevating [Ca2+] (13, 19, 20). t-SNARE complexes containing truncated versions of SNAP-25 that mimic cleavage by BoNT/A (residues 1 to 197) were reconstituted into vesicles (with 15% PS). As a control, a truncated version of SNAP-25 that mimics cleavage by BoNT/E (corresponding to residues 1 to 180), was tested in parallel [this cleavage event results in a more profound block of exocytosis (13)]. Fusion was abolished by the “BoNT/E” truncation (Fig. 4B). In contrast, the “BoNT/A” truncation supported a low level of fusion that could be enhanced by increasing [Ca2+]; the Ca2+ response was too impaired to determine the precise [Ca2+]1/2, but this value is >360 μM (Fig. 4B). Thus, the reconstituted system recapitulates the functional effect of BoNT/A and E treatment on neurons (13, 19, 20).

Membrane-embedded syt has been reported to stimulate membrane fusion in a Ca2+-independent manner (21). We have repeated these experiments and observed the same phenomena. The lack of an effect of Ca2+ is surprising, because the ability of syt to interact with its targets in the reduced fusion assay is promoted by Ca2+. The easiest explanation, however, is that the bacterially expressed full-length syt is not fully functional. Variants of a number of isoforms of syt, including syt I, that lack a transmembrane domain are expressed in cells where they may also regulate membrane traffic in vivo, supporting the idea that studies with the cytoplasmic domain of syt are physiologically relevant (2224).

The data reported here indicate that a complex of syt, membranes, and SNARE proteins forms the core of the Ca2+-triggered fusion apparatus. With this reconstitution approach, it should be possible to test additional factors to construct a Ca2+-triggered membrane fusion complex that operates on the rapid (millisecond) time scale observed during synaptic transmission.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1097196/DC1

Materials and Methods

References and Notes

References and Notes

View Abstract

Navigate This Article