Synaptotagmin Isoforms Couple Distinct Ranges of Ca²⁺, Ba²⁺, and Sr²⁺ Concentration to SNARE-mediated Membrane Fusion^D⃞

Plasmids and Protein Purification

cDNA encoding rat syt I (Perin et al., 1990) was provided by T. C. Südhof (University of Texas Southwestern Medical Institute, Dallas, TX). cDNA encoding mouse syts VII and IX (Fukuda et al., 1999) was provided by M. Fukuda (Institute of Physical and Chemical Research, Saitama, Japan). Syt I C2AB (amino acids 96–421), syt VII C2AB (amino acids 134–403), and syt IX C2AB (amino acids 104–386) were subcloned into pTrc-His vector (Invitrogen, Carlsbad, CA), expressed in Escherichia coli, and purified as described previously (Chapman et al., 1996) with modifications. Briefly, bacterial extracts were mixed with Ni-NTA agarose (0.8 ml of a 50% slurry; QIAGEN, Valencia, CA) for 30 min at 4°C. Beads were washed two times in wash buffer (25 mM HEPES-KOH, 400 mM KCl, 10% glycerol, 20 mM imidazole, and 5 mM β-mercaptoethanol) plus 10 μg/ml DNase and RNase (Roche Diagnostics, Indianapolis, IN). Two more washes were carried out in the same buffer lacking nucleases. Proteins were eluted from the beads in wash buffer with 500 mM imidazole and dialyzed overnight against buffer A (25 mM HEPES-KOH, pH 7.4, 100 mM KCl, 10% glycerol, 1 mM dithiothreitol).

Syt linker mutants (Bai et al., 2004b), which have either two (2X-linker) or three (3X-linker) copies of the linker sequence between C2A and C2B (residues 264–272), and Ca²⁺ ligand mutants, C2A_MB I and C2AB_M I, in which the subscript M corresponds to mutations that disrupt Ca²⁺-binding to either C2A (D230,232N) or C2B (D363,365N) (Earles et al., 2001; Bai et al., 2002), were expressed and purified as glutathione S-transferase (GST) fusion proteins and cleaved from the GST tag using thrombin, as described previously (Tucker et al., 2003).

Plasmids to generate recombinant synaptobrevin 2 (also called vesicle-associated membrane protein [VAMP] II) (pTW2) and the t-SNARE heterodimer (syntaxin 1A and SNAP-25) (pTW34) were kindly provided by J. E. Rothman (Columbia University, New York, NY), and proteins were expressed and purified as described previously (Weber et al., 1998; Tucker et al., 2004). The cytoplasmic domain of synaptobrevin 2 (cd VAMP, amino acids 1–94) was purified as described previously (Tucker et al., 2004). Full-length GST-SNAP-25B and GST-syntaxin 1A were purified as described previously (Earles et al., 2001). The syt IX-c-myc construct used for human embryonic kidney (HEK)-293 cell transfection was generated by cloning full-length syt IX into the pcDNA 3.1/c-myc-His vector (Invitrogen), resulting in a tag at the C terminus.

Preparation of v-SNARE- and t-SNARE-reconstituted Vesicles

All lipids were obtained from Avanti Polar Lipids (Alabaster, AL). Reconstitution of v-SNARE and t-SNARE vesicles was carried out as described previously (Tucker et al., 2004). v-SNARE vesicles were reconstituted using a lipid mix composed of 82% 1-palmitoyl, 2-oleoyl phosphatidylcholine (PC), 15% 1,2-dioleoyl phosphatidylserine, 1.5% N-(7-nitro-2–1,3-benzoxadiazol-4-yl)-1,2-dipalmitoyl phosphatidylethanolamine (NBD-PE, donor), and 1.5% N-(lissamine rhodamine B sulfonyl)-1,2-dipalmitoyl phosphatidylethanolamine (Rhodamine-PE, acceptor). t-SNARE vesicles were reconstituted in 85% PC and 15% PS (mole/mole). When PS was omitted from the vesicles, PC was increased to 97 and 100% for v-SNARE and t-SNARE vesicles, respectively. v-SNARE and t-SNARE vesicles were reconstituted to give ∼100–150 copies or ∼80 copies per vesicle, respectively, as described previously (Tucker et al., 2004). At ∼80 copies/vesicle, the t-SNARE concentration in the fusion assay is ∼2.7–3 μM.

Fusion Assays

Fusion assays were conducted as described previously (Tucker et al., 2004). Briefly, each reaction consisted of 45 μl of purified t-SNARE vesicles and 5 μl of purified, NBD/Rhodamine-PE labeled v-SNARE vesicles along with 0.2 mM EGTA, the indicated concentrations of divalent cation and syt, and buffer A in a total volume of 75 μl. The v- and t-SNARE vesicles stocks had equal concentrations of phospholipids; hence, t-SNARE vesicles were added in excess. After 2 h, 0.5% n-dodecylmaltoside (Roche Diagnostics) was added to dequench the NBD fluorescence. The raw fluorescence was converted to rounds of fusion as described previously (Parlati et al., 1999; Tucker et al., 2004). [Divalent cation]_free was calculated using WebMaxC version 2.22 software (C. Patton, Stanford University, Stanford, CA). Curve fitting (for Figure 1C) was carried out using Prism 4.02 software (GraphPad Software, San Diego, CA) with the variable slope sigmoidal dose-response function. For the syt VII/Ca²⁺ curve, data up to the plateau (at 100 μM Ca²⁺) was used for fitting to calculate EC₅₀ values.

Flotation Assays

Flotation assays were carried out as described previously (Tucker et al., 2004). For binding of syt cytoplasmic domains to protein-free vesicles shown in Figure 2C, 100-nm vesicles were prepared using an Avanti Polar lipids extruder according to the manufacturers instructions (Davis et al., 1999).

Immunoprecipitation Assays

Rat brain detergent extracts were prepared as described, and all incubations and binding reactions were carried out at 4°C (Chapman et al., 1995). One milligram of total protein extract was incubated with the indicated [divalent cation] for 1 h before the addition of 1.5 μl of the α-syntaxin 1A monoclonal antibody (mAb) HPC-1 (Barnstable et al., 1985). After incubating for an additional 2 h, 30 μl of protein-G Fast-Flow beads (of a 50% slurry; GE Healthcare, Piscataway, NJ) was added and incubated for another hour. Samples were then washed three times in buffer B (50 mM HEPES-NaOH, pH 7.4, 100 mM NaCl, 0.2 mM EGTA, and 1 mM [divalent cation]_free). Bound protein was eluted from the beads by boiling in SDS sample buffer; 25% of the bound material was subjected to SDS-PAGE and immunoblot analysis. Native syt I was probed using mAb 41.1 (Chapman and Jahn, 1994) and syntaxin was probed using HPC-1. For all blots in this study, immunoreactive bands were detected using enhanced chemiluminescence; signals were quantified by densitometry and bar graphs depict averages from at least three independent experiments. Error bars represent the SD. Significance was determined using an unpaired student t test with a 95% confidence interval. HPC-1 and 41.1 were provided by R. Jahn (Max Planck Institute for Biophysical Chemistry, Gottingen, Germany).

Transfection of HEK-293 Cells and Coimmunoprecipitation

Twelve micrograms of syt IX-c-myc (p-CDNA3.1) and syntaxin 1A (p-IRES-GFP) were cotransfected into HEK-293 cells using calcium phosphate (Jordan and Wurm, 2004). Four plates (100 mm) of confluent cells were harvested and lysed 48–72 h posttransfection using 0.5% cholate and 1% Triton X-100. The protein concentration was determined using the BCA assay and bovine serum albumin as a standard (Pierce Chemical, Rockford, IL). Coimmunoprecipitation (CoIP) assays were carried out as described above with 75 μg of protein from the detergent extract and 1.5 μl of α-c-myc antibody (9E-10; obtained from Harlan Bioproducts for Science, Indianapolis, IN; Evan et al., 1985). Fifty percent of the bound material was subjected to SDS-PAGE and immunoblot analysis. Syntaxin was probed using HPC-1, and syt IX was detected using an α-c-myc antibody.

GST-Pull-Down Assays

GST-binding assays were carried out in a total volume of 150 μl using 15 μg of GST-syntaxin or GST-SNAP-25 immobilized on 25 μl of glutathione-Sepharose beads (GE Healthcare) and 1.5 μM cytoplasmic domain of syt (Earles et al., 2001). Binding reactions were carried out for 1 h at 4°C in buffer B and beads were washed three times. Five percent of the bound material was subjected to SDS-PAGE and immunoblot analysis. Syt I was detected using mAb 41.1. For experiments with the cytoplasmic domain of syt IX, 35% of the bound material was subjected to SDS-PAGE and syt IX was detected using a polyclonal antibody directed against the cytoplasmic domain of the protein (Tucker et al., 2003).

Divalent Cation Dependency of syt I-, VII-, and IX-regulated Membrane Fusion

In a recent study, we demonstrated that the cytoplasmic domain of syt I stimulated SNARE-mediated membrane fusion in the presence of Ca²⁺ (Tucker et al., 2004). Here, we asked whether cytoplasmic domains derived from other members of the syt family can also stimulate SNARE-mediated membrane fusion in response to Ca²⁺ (Figure 1A). For these studies, we compared syt I, VII, and IX, because these are the major isoforms expressed in PC12 cells, which are a common model system to study secretion (Kishimoto et al., 2001; Zhang et al., 2002; Tucker et al., 2003). In addition, we selected syt VII because it has been suggested to be a high-affinity Ca²⁺ sensor in comparison with syt I (Sugita et al., 2002, but see Li et al., 1995a,b). We note that all experiments reported in this study use the cytoplasmic domain of either syt I, VII, or IX, because the full-length proteins have a tendency to aggregate.

We found that the cytoplasmic domains of all three syt isoforms tested stimulated SNARE-mediated membrane fusion in response to Ca²⁺ (Figure 1B). The [syt] dependence for stimulation was nearly identical for each of the three isoforms tested, with maximal stimulation occurring at ∼7 μM protein (Supplemental Figure 1). Although different syt isoforms stimulated membrane fusion to different extents, it is clear that stimulation of SNARE-mediated fusion is not limited to syt I but rather extends to other members of the syt family.

In these experiments, syt I-stimulated membrane fusion with a [Ca²⁺]_1/2 of 116 μM, in agreement with previous work (Tucker et al., 2004). In contrast, syt VII stimulated fusion with a [Ca²⁺]_1/2 of 0.30 μM, which is roughly 400-fold lower than for syt I. The Ca²⁺ dose response for syt IX-stimulated membrane fusion was more complex, yielding two components with [Ca²⁺]_1/2 values of 0.40 and 32 μM (Figure 1C and Table 1). The major conclusion drawn for these experiments is that the syt family is capable of sensing a wide range of Ca²⁺ concentrations. It should be noted that these [Ca²⁺]_1/2 values were obtained using liposomes containing 15% PS. For syt I, increasing the PS to 25% significantly reduced the [Ca²⁺]_1/2 for fusion by enhancing the ability of syt to bind Ca²⁺ (Brose et al., 1992; Tucker et al., 2004). Thus, all of the values reported here are likely to be modulated by changes in lipid composition.

The cytoplasmic domain of synaptobrevin 2 (cd VAMP) effectively blocked Ca²⁺·syt-stimulated membrane fusion for all isoforms tested, as shown previously for syt I (Tucker et al., 2004; Figure 1D). This suggests that trans pairing of SNAREs is required for syt-stimulated membrane fusion. Furthermore, the cytoplasmic domain of each syt isoform had no effect on NBD dequenching when either v- or t-SNAREs were omitted from the vesicles (Tucker et al., 2004; our unpublished observations), suggesting that both SNARE populations are required for syt action during fusion.

As outlined in Introduction, other alkaline metals such as Ba²⁺ and Sr²⁺ can substitute for Ca²⁺ in triggering release from numerous cell types, including PC12 cells (Kishimoto et al., 2001). One question that has arisen is whether syt I can couple these alternative metals to membrane fusion or whether other syt isoforms selectively serve as Sr²⁺ and Ba²⁺ sensors. We therefore assayed the ability of Sr²⁺ and Ba²⁺ to regulate the activity of syt I, VII, and IX in the reconstituted fusion assay (Figure 1, B and C, and Table 1). We found that Sr²⁺ activated only syt VII and IX ([Sr²⁺]_1/2 values of 260 and 983 μM, respectively) and Ba²⁺ activated only syt VII in the fusion assay ([Ba²⁺]_1/2 value of 563 μM). Thus, although all syt isoforms tested responded to Ca²⁺, only a subset of syts are capable of coupling Sr²⁺ and Ba²⁺ to membrane fusion under the conditions of our assay system. These data suggest that in PC12 cells, syt VII and IX might function as Sr²⁺ sensors and that syt VII might also function as a Ba²⁺ sensor.

In all cases, syt stimulation of SNARE-mediated membrane fusion was strictly regulated by Ca²⁺, Ba²⁺, and/or Sr²⁺; in the absence of these divalent cations, none of the syt isoforms stimulated fusion (Tucker et al., 2004). The slight increase in fusion seen in the presence of Mg²⁺ (Figure 1B) is within error of the assay, and raising the [Mg²⁺] from 10 nM to 3 mM (Figure 1C) did not result in significant increases in fusion. Furthermore, none of the divalent cations tested affected fusion in the absence of syt (our unpublished observations).

Divalent Cations Trigger Binding of syt to t-SNAREs and PS

Syt I binds anionic lipids and the t-SNAREs, syntaxin and SNAP-25, in a Ca²⁺-promoted manner (Zhang et al., 2002; Bai et al., 2004b). These interactions have been suggested to couple Ca²⁺ influx to regulated secretion, although the role of SNARE binding activity in Ca²⁺-regulated secretion has been questioned (Shin et al., 2003). We therefore carried out experiments to determine whether there is a correlation between stimulation of fusion and divalent cation-induced binding of syt to t-SNAREs and/or to anionic lipids.

A coflotation assay was used to monitor the binding of syt to t-SNARE liposomes (Figure 2A). This assay is unique because it makes it possible to measure binding of syt to t-SNAREs that are embedded into membranes rather than to t-SNAREs in detergent micelles or soluble t-SNARE fragments. In the first series of experiments, we assayed for binding of syt to t-SNARE liposomes composed of 100% PC; under these conditions (i.e., in the absence of PS) all of the proteoliposome binding activity is mediated by interactions with t-SNAREs and not membranes, because syt does not bind to PC (Tucker et al., 2004). We observed that all three divalent cations were able to trigger binding of syt I to t-SNAREs (Figure 2B), which is consistent with previous results (Chapman et al., 1995; Tucker et al., 2004, but see also Shin et al., 2003). For syt VII, Ca²⁺, Sr²⁺, and Ba²⁺ promoted binding to t-SNAREs and also stimulated membrane fusion. Syt IX exhibited strong t-SNARE binding activity in the presence of Ca²⁺; weaker, but significant levels of binding in response to Sr²⁺; and little binding in Ba²⁺, similar to the pattern of stimulation seen in the reduced fusion assay. Thus, in all cases where divalent cations stimulated fusion, they also triggered syt·t-SNARE interactions. However, in some cases, binding was not sufficient to regulate fusion (e.g., Ba²⁺/Sr²⁺, and syt I). We propose that syt operates by first binding to effectors and then driving subsequent conformational changes/rearrangements that result in the regulation of fusion. Apparently, some metals can drive binding but fail to drive the subsequent steps that regulate fusion.

Next, we examined the interaction of the cytoplasmic domains of syt I, VII, and IX with protein free PS-harboring liposomes in response to different divalent cations. Under these conditions, syt associates with the liposomes by virtue of its high-affinity interaction with PS (Tucker et al., 2004). Syt I bound protein-free vesicles containing 15% PS in the presence of Ca²⁺ but weakly in the presence of Ba²⁺ and Sr²⁺ (Figure 2C); these results correlate well with syt I stimulation of membrane fusion (Figure 1, B and C). Likewise, there was a tight correlation between stimulation of membrane fusion and the PS binding activity of syt VII and IX in response to each divalent cation. Specifically, syt VII stimulated membrane fusion in response to Ca²⁺, Ba²⁺, and Sr²⁺ and also bound to PS in the presence of all three divalent cations. Syt IX stimulated membrane fusion in the presence of Ca²⁺, weakly in response to Sr²⁺, and not at all in Ba²⁺—the same pattern of behavior as seen for lipid binding.

Binding of syt I to vesicles in response to Ba²⁺ and Sr²⁺ could be induced by increasing the mole fraction of PS from 15 to 25% (Figure 2C). However, reconstitution of v- and t-SNAREs into vesicles that harbored 25% PS did not result in syt I stimulation of membrane fusion in response to Ba²⁺ and Sr²⁺ (Figure 2, D and E). Thus, in at least some cases, Sr²⁺- and Ba²⁺-triggered binding of syt to PS, and to t-SNAREs (discussed above), can be uncoupled from membrane fusion. These data indicate that syt regulation of membrane fusion may involve additional divalent cation-induced activation steps that occur after binding of syt to these effector molecules.

Recent studies suggested that syt I·t-SNARE interactions are not essential for the fast component of divalent cation regulated secretion. This was largely based on the observation that syt I did not bind t-SNAREs in response to Sr²⁺ (Shin et al., 2003), and native syt IX was reported not to coimmunoprecipitate (IP) with t-SNAREs in a Ca²⁺-dependent manner (Shin et al., 2004). However, we found that recombinant syt I and IX bind reconstituted t-SNAREs in the presence of Ca²⁺ and Sr²⁺, although syt IX bound t-SNAREs somewhat weakly in the presence of Sr²⁺ (Figure 2B). To clarify this issue, we performed coIP experiments from rat brain detergent extracts. Native syt I coIPed with the t-SNARE syntaxin to a small extent in EGTA; binding was significantly increased in response to Ca²⁺ as well as Ba²⁺ and Sr²⁺ (Figure 3, A and B). As a control, we probed for SNAP-25 in the immunoprecipitates and found the levels of SNAP-25 were unaffected by divalent cations (our unpublished observations). Because syt IX is expressed at >10-fold lower levels in rat brain than syt I (Chicka, Edwardson, Bhalla, and Chapman, unpublished observations), we cannot reliably detect syt IX in t-SNARE IPs. This observation might explain why Shin et al. (2004) failed to detect syt IX·t-SNARE interactions from rat brain. To further explore this issue, full-length syt IX was tagged with a c-myc epitope and coexpressed with full-length syntaxin in HEK-293 cells. CoIP experiments were then performed using an α-c-myc antibody. Low levels of syntaxin bound to syt IX in EGTA with a significant increase in binding in the presence of both Ca²⁺ (p < 0.05) and Sr²⁺ (p = 0.05; Figure 3, C and D). These data are in agreement with data obtained using recombinant proteins (Figure 2B). As a final confirmation, pull-down experiments were carried out using the cytoplasmic domain of syt and immobilized GST-tagged SNAP-25 and GST-tagged syntaxin. The pull-down data confirmed that syt I and IX bind t-SNAREs in a Ca²⁺, Ba²⁺, and Sr²⁺ promoted manner (Supplemental Figure 2) and are also in agreement with previous studies showing that in PC12 cell membranes, native syts I and IX interact with the other t-SNARE, SNAP-25, in a Ca²⁺-promoted manner (Zhang et al., 2002). Although there are some differences in the extent of binding, the interaction of syt I and IX with t-SNAREs was always stimulated by divalent cations that promote fusion.

Syt Stimulates Membrane Fusion by Engaging t-SNAREs

The data mentioned above suggest that t-SNARE binding is not sufficient for syt stimulation of membrane fusion. To address more directly whether syt·t-SNARE interactions do in fact regulate fusion, we determined whether mutations in syt I, that selectively disrupt t-SNARE binding activity, also disrupt stimulation in the reduced fusion assay. In a previous study, we showed that doubling or tripling the length of the linker that connects the C2A and C2B domains of syt I results in a graded reduction in Ca²⁺-dependent t-SNARE binding activity (Bai et al., 2004b). This is a selective loss-of-function mutation; increasing the linker length does not affect the interaction of syt I with anionic lipids (Bai et al., 2004b). Coflotation of the linker mutants with reconstituted t-SNAREs confirmed the loss of t-SNARE binding activity with increased linker length, although some degree of Ca²⁺-dependent binding persisted even with the 3X-linker construct (Figure 4, A and B) (Bai et al., 2004b).

Increasing the length of the linker that connects the C2A and C2B domains of syt I resulted in a graded loss of Ca²⁺·syt stimulated SNARE-catalyzed membrane fusion (Figure 4C). For example, the extent of stimulation was reduced 32% for the 3X-linker mutant compared with wild-type (wt) syt I. These data indicate that reduction of t-SNARE binding activity correlates with a loss of Ca²⁺-dependent stimulation of fusion. Thus, syt regulates membrane fusion, at least in part, via interactions with t-SNAREs. In the absence of t-SNAREs or v-SNAREs, Ca²⁺·syt is without effect in the fusion assay (Tucker et al., 2004; our unpublished observations).

Disruption of the Ca²⁺-binding sites in both C2 domains of syt I abolished Ca²⁺-stimulated fusion of v- and t-SNARE vesicles (Tucker et al., 2004). However, it is not known whether Ca²⁺ binding to the C2A domain, C2B domain, or both domains is necessary for stimulation. We therefore analyzed two mutant versions of syt I; one in which the Ca²⁺-binding activity of C2A was disrupted (C2A_MB), and one in which the Ca²⁺-binding activity of C2B was disrupted (C2AB_M). These mutations reduce t-SNARE and PS/PC liposome binding activity (Figure 4, A and B; Earles et al., 2001). Disruption of Ca²⁺-binding to either C2A or C2B reduced the ability of syt to stimulate fusion (Figure 4D). Similar reductions in activity for the C2A and C2B mutants were observed when liposomes contained 25% PS and 15 μM syt was added to the reaction (Figure 4E). These data indicate that both C2-domains of syt I play roles in Ca²⁺ triggered, SNARE-mediated membrane fusion. However, at lower levels of PS (e.g., 15%; Figure 4D) and soluble syt I, the mutations in C2A had a greater effect than mutations in C2B (Figure 4, D compared with E).

PS Is Required for Syt Stimulation of Fusion

Finally, we addressed the role of syt·PS interactions during membrane fusion. We had previously shown that changes in the PS content of reconstituted v- and t-SNARE vesicles results in a shift in the Ca²⁺ dependence of syt-stimulated membrane fusion, indicating that syt executes its function via interactions with anionic lipids in addition to interactions with t-SNAREs (Tucker et al., 2004). Here, we asked whether PS is essential for Ca²⁺·syt-stimulated fusion by selectively omitting this lipid from the v- and/or t-SNARE vesicles. Removal of PS from both vesicle populations did not affect SNARE catalyzed fusion but abolished Ca²⁺·syt stimulation of fusion (Figure 5, A and B). Thus, PS is an essential effector for the action of Ca²⁺·syt.

Removal of PS from the v-SNARE membranes also resulted in a loss of syt-stimulated membrane fusion. However, we observed that syt VII and IX, but not syt I, could tolerate removal of PS from the t-SNARE vesicles (Figure 5, A and B; our unpublished observations). Thus, PS is essential for the ability of all isoforms of syt studied here to stimulate membrane fusion. However, the precise localization of PS needed for syt activity differs among syt isoforms. Although the reason for these latter differences is unclear, these experiments nevertheless demonstrate that PS is an essential effector for Ca²⁺·syt action.