Ca 2 and N -Ethylmaleimide-sensitive Factor Differentially Regulate Disassembly of SNARE Complexes on Early Endosomes*

The endosome-associated protein Hrs inhibits the homotypic fusion of early endosomes. A helical region of Hrs containing a Q-SNARE motif mediates this effect as well as its endosomal membrane association via SNAP-25, an endosomal receptor for Hrs. Hrs inhibits formation of an early endosomal SNARE complex by displac-ing VAMP-2 from the complex, suggesting a mechanism by which Hrs inhibits early endosome fusion. We examined the regulation of endosomal SNARE complexes to probe how Hrs may function as a negative regulator. We show that although NSF dissociates the VAMP-2 (cid:1) SNAP-25 (cid:1) syntaxin 13 complex, it has no effect on the Hrs-con-taining complex. Whereas Ca 2 (cid:1) dissociates the Hrs-con-taining complex but not the VAMP-2-containing SNARE complex. This is the first demonstration of differential regulation of R/Q-SNARE and all Q-SNARE-containing SNARE complexes. Ca 2 (cid:1) also reverses the Hrs-induced inhibition of early endosome fusion in a tetanus toxin-sensitive manner and removes Hrs from early endosomal membranes. Moreover, Hrs inhibition of endosome fusion and its endosomal localization are sensitive to bafilomycin, Early Endosome Fusion— The homotypic fusion of early endosomes was measured as previously described (17). Effects of various molecules were assessed after incubation on ice for 15 min with the donor and acceptor membranes, cytosol, and ATP. The reactions were then trans- ferred to 37 °C for 60 min. Free Ca 2 (cid:1) concentrations (0, 0.003, 0.03, 0.1, 0.3, 1, 10, and 30 m M ) and 1 m M of Cu 2 (cid:1) , Ba 2 (cid:1) , and Mn 2 (cid:1) were calculated in the presence of 2 m M EGTA using the WebMaxCalc program (www.stanford.edu/ (cid:3) cpatton/webmaxcS.htm) (18). TeTx treatment was performed by isolating the early endosomes as described above and resuspending the vesicles with varying concentrations of TeTx (5–6400 n M in a final volume of 30 (cid:3) l) (9, 24) followed by incubation at 37 °C for 30 or 60 min. The membranes were collected by centrifugation (100,000 (cid:4) g for 10 min) and resuspended in 100 (cid:3) l of homogenization buffer for use in the fusion reactions. For bafilomycin treatment, the cells were incubated with Dulbecco’s minimal essential medium containing 1% bovine

Endocytosis is a fundamental process essential for all eukaryotic cells. It functions in nutrient uptake, regulation of the protein and lipid composition in the plasma membrane, and modulation of cellular responses by affecting exocytosis and receptor signaling. Molecules transit through the endocytic pathway by passing from one compartment to another through a series of membrane fission and fusion reactions. The ultimate role of this pathway is to allow the sorting of molecules to be recycled from those to be degraded. To function correctly, this system must regulate both the sorting events and the fusion events that promote proper targeting of internalized small molecules, lipids, and proteins.
Protein machinery is required for fusion of biological membranes. Interactions among SNARE 1 proteins (1-4) associated with donor membranes (e.g. VAMP/synaptobrevin) and acceptor membranes (e.g. syntaxin and SNAP-25) are thought to be essential for fusion (1)(2)(3)(4)(5). SNAREs are sufficient for membrane fusion in artificial membranes, suggesting that they form the core membrane fusion machinery (6). SNAREs form cytoplasmic helical bundles that bridge two membranes (trans-SNARE complex) to enable membrane fusion (1,4,7). SNAREs are characterized by a helical "SNARE" motif that contains a glutamine or arginine at its center (4,8). Botulinum and tetanus toxins are zinc endoproteases that cleave the SNAREs, inhibit the formation of SNARE complexes, and block fusion (1,2,9). Once membrane fusion has occurred, the cytoplasmic adapter protein ␣SNAP binds to the SNARE complex and recruits Nethylmaleimide-sensitive factor (NSF) from the cytoplasm (1,2,7). The ATPase activity of NSF dissociates the cis-SNARE complex, allowing the proteins to be available for trans-pairing and subsequent fusion events (1,2,7).
The trans-SNARE complex may be the catalyst for membrane fusion, although regulatory events or molecules may influence this process. For example, intraorganellar Ca 2ϩ release from the yeast vacuole, mammalian endosome, or nuclear vesicles is required for fusion events involving those compartments (10 -13). A model that has emerged from these studies is that an unknown event triggers Ca 2ϩ release from the organelle. This local pool of Ca 2ϩ is required for the fusion event, although the nature of its effector is unclear. Evidence from studies of homotypic vacuole fusion suggest that calmodulin may be a Ca 2ϩ target because it binds to vacuoles upon Ca 2ϩ release and appears to promote bilayer mixing of vacuoles and endosomes (11)(12)(13), although the effect of calmodulin may be via regulation of SNARE complexes (14,15).
The endosome-associated protein Hrs (hepatocyte-growth factor-regulated tyrosine kinase substrate) has been shown to bind the Q-SNARE SNAP-25 (16) and inhibits the homotypic fusion of early endosomes (17). A heptad repeat region of Hrs containing a Q-SNARE motif mediates this effect as well as its endosomal membrane association (17). SNAP-25 is an endosomal receptor for Hrs, and Hrs inhibits the formation of an early endosomal SNARE complex by disallowing VAMP incorporation into the complex (17). Hrs is also likely involved in cargo sorting at the level of the early/late endosome by recruiting sorting or signaling components to the endosomal membrane. Therefore, Hrs may bind to SNAP-25 using its Q-SNARE domain and inhibit endosomal fusion while it is involved in cargo sorting or endosome motility using NH 2 -terminal VHS (Vps27, Hrs, STAM, a domain found in a number of proteins involved in trafficking that, in some cases, binds to GGA proteins); FYVE (Fab 1, YotB, Vac1, EEA1, a dual zinc finger domain found in a number of proteins involved in trafficking some of which bind to phosphatidylinositol 3-phosphate, phosphatidylinositol 3-phosphate); or ubiquitin-interacting motif, a domain found in a number of proteins that binds ubiquitin with low affinity) domains (Ref. 17 and references therein) or via other protein interactions. Interestingly, the binding of Hrs to SNAP-25 is negatively regulated by Ca 2ϩ such that Ca 2ϩ inhibits the binding of Hrs and SNAP-25 (18,19). In the present study, we have examined the regulation of Hrs-and VAMP-2-containing endosomal SNARE complexes and find that although NSF regulates the dissociation of the VAMP-2⅐SNAP-25⅐syntaxin 13 complex, it has no effect on the Hrs-containing complex. Conversely, Ca 2ϩ dissociates the Hrs-containing complex but not the VAMP-2-containing SNARE complex. Ca 2ϩ removes Hrs from early endosomal membranes at concentrations similar to those that dissociate the Hrs-containing SNARE complex. Similar Ca 2ϩ concentrations reverse the Hrs-induced inhibition of early endosome fusion in a tetanus toxin-sensitive manner. Moreover, the sensitivity of Hrs inhibition of endosome fusion to bafilomycin implies a role for luminal Ca 2ϩ release in the dissociation of Hrs from SNAP-25/early endosomal membranes. These data suggest that Hrs provides a Ca 2ϩ -dependent negative influence on early endosome fusion and that a VAMP-2-containing SNARE complex can form to enable fusion after the dissociation of Hrs.
Early Endosome Fusion-The homotypic fusion of early endosomes was measured as previously described (17). Effects of various molecules were assessed after incubation on ice for 15 min with the donor and acceptor membranes, cytosol, and ATP. The reactions were then transferred to 37°C for 60 min. Free Ca 2ϩ concentrations (0, 0.003, 0.03, 0.1, 0.3, 1, 10, and 30 mM) and 1 mM of Cu 2ϩ , Ba 2ϩ , and Mn 2ϩ were calculated in the presence of 2 mM EGTA using the WebMaxCalc program (www.stanford.edu/ϳcpatton/webmaxcS.htm) (18). TeTx treatment was performed by isolating the early endosomes as described above and resuspending the vesicles with varying concentrations of TeTx (5-6400 nM in a final volume of 30 l) (9, 24) followed by incubation at 37°C for 30 or 60 min. The membranes were collected by centrifugation (100,000 ϫ g for 10 min) and resuspended in 100 l of homogenization buffer for use in the fusion reactions. For bafilomycin treatment, the cells were incubated with Dulbecco's minimal essential medium containing 1% bovine serum albumin and 10 mM Ca 2ϩ for 60 min. Subsequently, the cells were labeled with fluorescent epidermal growth factor (15 min at 37°C) in the presence of bafilomycin (100 nM). After labeling, the cells were washed, homogenized, and the endosomes were isolated in the presence of bafilomycin. The endosomes were incubated with Hrs (180 nM) in the presence of bafilomycin (15 min on ice). Bafilomycin was removed from some membranes by dilution with 10 volumes of reaction buffer, and all of the membranes were subjected to centrifugation followed by resuspension of endosomes in reaction buffer. The reactions were incubated (37°C for 60 min) in the presence or absence of bafilomycin.
SNARE Complex Disassembly-A constant amount of GST-syntaxin 13 (2 g/reaction) bound to glutathione-agarose was incubated with constant amounts of SNAP-25 (1 g), VAMP-2 (1 g), or Hrs (2 g) in the presence or absence of NSF/␣SNAP (0.25 mg/ml), MgATP, or ATP␥S (0.5 mM) or EDTA in phosphate-buffered saline binding buffer to a final reaction volume of 50 l. In addition, the Hrs-containing complex was incubated in 50 M free Ca 2ϩ . After incubation (4°C for 60 min), the samples were washed three times with binding buffer, boiled in SDS sample buffer, and separated by SDS-PAGE. Immunoblot analysis was performed using anti-Hrs and anti-VAMP-2 antibodies followed by appropriate 125 I secondary antibodies. Parallel experiments in which Hrsor VAMP-2-containing SNARE complexes were preformed prior to incubation with NSF/␣SNAP or Ca 2ϩ yielded similar results (see Fig. 1b). Immunoreactive bands were visualized and quantitated using phosphorimaging. The experiment shown is representative of six experiments.
Endosomal Membrane Binding-The endosomal membranes were purified from HeLa cells by centrifugation on a discontinuous sucrose gradient (17). One 10-cm plate (approximately 80% confluent) was scraped in homogenization buffer (20 mM HEPES, pH 7.4, 0.25 M sucrose, 2 mM EGTA, 2 mM EDTA, 0.1 mM dithiothreitol, 0.4 ml) and passed through a 30-gauge needle 30 times. The resulting lysate was subjected to centrifugation (100,000 ϫ g 10 min), and the pellet was resuspended in homogenization buffer (0.17 ml) and then mixed with 61% sucrose to a final concentration of 46% sucrose (0.5 ml total). The 46% sucrose cushion was overlaid with two additional layers of sucrose 35% (0.65 ml) and 30% (0.45 ml) and then with homogenization buffer (0.4 ml). The gradients were subjected to centrifugation in a Beckman TLS55 rotor (124,000 ϫ g for 60 min), and the early endosomes (interface between 30 and 35% sucrose) were collected. A constant amount of His-tagged Hrs (180 nM) was added to reactions containing purified endosomal membranes (as in a fusion reaction), along with various concentration of free Ca 2ϩ (0, 0.003, 0.03, 0.1, 0.3, 1, 3, 10, and 30 mM) and 1 mM of Cu 2ϩ , Ba 2ϩ , and Mn 2ϩ . The reactions were incubated at 37°C for 60 min and stopped by subjecting them to centrifugation (100,000 ϫ g for 10 min). The amount of His-Hrs was determined in pellet and supernatant fractions by quantitative Western blotting using anti-His antibodies and 125 I-conjugated secondary antibodies. The experiment shown is representative of 11 experiments.

Ca 2ϩ but Not NSF Dissociates an Hrs-containing SNARE
Complex-Because Ca 2ϩ inhibits Hrs from binding to SNAP-25 (18,19), we tested the possibility that Ca 2ϩ might prevent the formation of the recently identified SNARE complex containing Hrs, syntaxin 13, and SNAP-25 (17). In the presence of free Ca 2ϩ (50 M), Ͻ10% of Hrs bound to a complex of SNAP-25 and syntaxin 13 (Fig. 1a, lane 5) compared with the binding in the absence of Ca 2ϩ (Fig. 1a, lanes 1-4). Even in the presence of ␣SNAP and NSF, two proteins known to disassemble SNARE complexes, Hrs was incorporated into a complex with SNAP-25 and syntaxin 13 (Fig. 1a, lane 2). In contrast, NSF and ␣SNAP dissociated the VAMP-2-containing SNARE complex (Fig. 1a, lane 7). The effect of NSF and ␣SNAP required MgATP because the VAMP-2-containing complex did not dissociate when EDTA was present (Fig. 1a, lane 8). Moreover, the ATPase activity of NSF was required for SNARE complex dissociation because the complex did not dissociate in the presence of the nonhydrolyzable ATP analog ATP␥S (Fig. 1a, lane 9). Parallel experiments in which Hrs or VAMP-2 SNARE complexes were preformed prior to incubation with Ca 2ϩ or NSF/␣SNAP yielded similar results ( Fig. 1b and data not shown). The half-maximal concentration of Ca 2ϩ required to inhibit Hrs binding to the SNAP-25⅐syntaxin 13 complex was ϳ20 M (Fig. 1b). These results suggested that Ca 2ϩ could dissociate a SNARE complex containing Hrs, SNAP-25, and syntaxin, whereas NSF and ␣SNAP have no effect on the formation or dissociation of this complex.
Ca 2ϩ Dissociates an Hrs-containing SNARE Complex Allowing VAMP-2 Binding-The effect of Ca 2ϩ dissociation of Hrs on the ability of VAMP-2 to enter a SNARE complex was tested. When VAMP-2 and Hrs alone were added to binding reactions with SNAP-25 and syntaxin 13, they were both able to form SNARE complexes (Fig. 1c, lanes 1 and 3, respectively). However, when VAMP-2 and Hrs were added together, Ͼ80% of VAMP-2 was not bound (Fig. 1c, lane 2), consistent with our previous studies (23). Free calcium inclusion resulted in a loss of Hrs binding (Fig. 1c, lane 4). However, when VAMP-2 was included with Hrs and Ca 2ϩ , VAMP-2 binding was comparable with control (Fig. 1c, lane 5 compared with lane 1). This result demonstrates that Ca 2ϩ has no effect on formation/dissociation of a VAMP-2⅐SNAP-25⅐syntaxin 13 SNARE complex and that in the presence of Ca 2ϩ VAMP-2 is able to displace Hrs for SNAP-25-syntaxin 13 binding.
Ca 2ϩ Relieves the Hrs-dependent Inhibition of Early Endosome Fusion-The ability of Ca 2ϩ to displace Hrs from the SNARE complex and allow complex assembly with VAMP-2 suggests that calcium might reverse the inhibition of early endosome fusion produced by Hrs. We previously developed an assay for endosome fusion in which different populations of HeLa cells engage in receptor-mediated endocytosis of epidermal growth factor linked to either Alexa 488 or tetramethylrhodamine, allowing isolation of donor and acceptor pools of endosomes. These compartments are used in fusion reactions that are analyzed by examining resonance energy transfer between the fluorophores to detect content mixing. This assay is dependent on temperature, time, energy, and cytosol (17). Fusion reactions containing 180 nM Hrs were inhibited 70.2 Ϯ 1.4% compared with control reactions (Fig. 2). However, when Ca 2ϩ was added to these reactions in the range of 3 nM to 30 mM, fusion became more efficient in a concentration-dependent fashion reaching a maximum of 84.0 Ϯ 5.4% of control reactions (Fig. 2). This effect had a half-maximal value of ϳ90 M Ca 2ϩ and was specific to calcium because other divalent cations such as Cu 2ϩ , Ba 2ϩ , and Mn 2ϩ (all at 1 mM) were unable to relieve the inhibition (Fig. 2). Although not shown, Cu 2ϩ , Ba 2ϩ , and Mn 2ϩ did not prevent Hrs from binding in SNARE complexes formed in vitro as performed earlier (Figs. 1 and 2). The concentration of calcium required and the cation specificity necessary for relieving the Hrs-dependent inhibition of early endosome fusion correlated well with the block in formation of the Hrs⅐SNAP-25⅐syntaxin 13 SNARE complex. This provided evidence to suggest that calcium might function to dissociate a "nonfusogenic" SNARE complex (containing Hrs) and allow a "fusogenic" SNARE complex (containing VAMP-2) to assemble on early endosomes.
Calcium Blocks Hrs from Binding to Early Endosomes-To further examine the notion that calcium might block binding of Hrs to SNARE complexes on membranes, we examined the effect of divalent cations on the binding of Hrs to purified early endosomes. When Hrs was added (180 nM) to early endosomes (purified as described under "Experimental Procedures") in the presence of increasing Ca 2ϩ , a concentration-dependent decrease of Hrs binding was observed (Fig. 3a, lanes 1-7). The half-maximal value for calcium in this inhibition was ϳ30 M. As seen previously (Fig. 2), Cu 2ϩ , Ba 2ϩ , and Mn 2ϩ had no effect (Fig. 3a, lanes 8 -10). Importantly, Ca 2ϩ did not affect the binding of an early endosomal protein, EEA1, suggesting specificity in its effect on Hrs (Fig. 3a, lanes 1-7). The inhibition of Hrs binding to early endosomes and the SNAP-25⅐syntaxin 13 SNARE complex suggested that Ca 2ϩ might regulate how Hrs interacts with a receptor on membranes such as SNAP-25 (17). SNAP-25 is considered to be a strictly neuroendocrine or neuronal protein (although see Ref. 25). Our use of HeLa cells for not only purified early endosomes but also homotypic endosome fusion assays and the dependence of these assays on SNAP-25 as suggested by their sensitivity to botulinum toxin E suggests that SNAP-25 resides in this non-neuronal cell. To demonstrate that this is indeed the case, we attempted to detect SNAP-25 in HeLa cell lysates and purified early endosomes. As expected, markers for the ER (calnexin), plasma membrane (Na/K ATPase), lysosomes (LAMP1), and early endosomes (EEA1) were detectable in crude lysates from HeLa cells (Fig.  4, lane 1). The HeLa cell lysate also contained SNAP-25 (Fig. 4,  lane 1). Most importantly, purified membranes from HeLa cells containing the early endosomal marker, EEA1, also contained SNAP-25 (Fig. 4, lane 2). These membranes did not contain detectable amounts of the Na/K ATPase, calnexin, or LAMP1, suggesting the lack of plasma membrane, ER, and lysosomes, respectively (Fig. 4, lane 2). This demonstrated that the membranes were not contaminated with other organelles and allowed us to conclude that SNAP-25 was indeed located on early endosomes from HeLa cells.
VAMP-2 Is Required for Calcium to Efficiently Relieve Hrs Inhibition of Early Endosome Fusion-Certain bacterial toxins are valuable tools for dissecting the molecular details of SNARE-dependent membrane fusion. These toxins are proteases with highly specific cleavage recognition sites (9, 24). For example, botulinum toxin E (BoNT/E) is a zinc endopro- tease that cleaves the COOH-terminal 26 amino acids of SNAP-25 and thereby blocks membrane fusion that requires a four helical SNARE complex containing SNAP-25 (9). Likewise, TeTx cleaves the COOH-terminal 41 amino acids of VAMP-2 (24) and blocks membrane fusion that requires a four helical SNARE complex containing VAMP-2 (1, 7, 9).
Pretreatment of early endosomes with TeTx inhibited mem-brane fusion in a dose-dependent manner (Fig. 5a). Treatment of early endosomal membranes resulted in 56.2 Ϯ 9.7% maximal fusion efficiency (Fig. 5, a and b). The half-maximal value

FIG. 2. Calcium relieves the Hrs-dependent inhibition of early endosome fusion.
Separate cultures of HeLa cells were treated with epidermal growth factor linked to either Alexa 488 or tetramethylrhodamine, and early endosomes were prepared for use as donor and acceptor membranes (17). Fusion assays using standard conditions were performed in the presence of various concentrations of free Ca 2ϩ , Cu 2ϩ , Ba 2ϩ , or Mn 2ϩ as indicated. The fluorescent signal at 580 nm (acceptor emission) was measured as an indicator of content mixing (17). The results are representative of three such experiments.
FIG. 3. Calcium inhibits Hrs binding to early endosomes. a, early endosomes were purified from HeLa cells (described under "Experimental Procedures"). The binding reactions contained purified early endosomes, recombinant His 6 -Hrs (180 nM), and various concentrations of divalent cations as indicated (lanes 2-10). After incubations (60 min at 37°C), the membranes were collected by centrifugation at 100,000 ϫ g, and the proteins were separated by SDS-PAGE. His 6 -Hrs was detected with a monoclonal antibody against the His tag and quantified using an 125 I secondary antibody and phosphorimaging. The early endosomal marker, EEA1, was detected with chemiluminescence. b, plot of the Hrs amount bound to early endosomes from data in a. The results shown are representative of 11 experiments.

FIG. 4. Early endosomes from HeLa cells contain SNAP-25.
HeLa cell lysate (lane 1) and purified early endosomes (lane 2) were subjected to SDS-PAGE, and proteins were detected on Western blots with specific antibodies using chemiluminescence. In addition to SNAP-25, marker proteins for the ER (calnexin), plasma membrane (Na/K ATPase), lysosomes (LAMP1), and early endosomes (EEA1) were detected. The bottom blot shows Ponceau S staining before antibody detection.
of TeTx for this inhibition was ϳ80 nM, comparable with the concentrations required to inhibit neurotransmitter secretion. This result suggested that a TeTx substrate, presumably-VAMP-2, played a role in early endosome fusion. The combination of TeTx and Hrs was no more efficacious than Hrs alone in inhibiting early endosome fusion (Figs. 5b or 3). When Ca 2ϩ was added along with Hrs, the fusion efficiency increased to 85.6 Ϯ 5.6% (Figs. 5b and 2). However, if membranes were first treated with TeTx and then incubated in the presence of Hrs and free Ca 2ϩ , the fusion efficiency was just 57.4 Ϯ 4.8% (Fig.  5b). This suggested that a TeTx substrate, presumably VAMP-2, was in part required for calcium to reverse the Hrsdependent inhibition of early endosome fusion, accounting for a loss of 28.3% (85-57%; Fig. 5b).

Release of Calcium from Luminal Stores Affects Hrs Function and Localization-An experiment was performed to test whether release of Ca 2ϩ from luminal stores could reverse
Hrs-dependent inhibition of early endosome fusion. The steadystate level of Ca 2ϩ in HeLa cells is estimated to be about 100 nM (26), which is below the concentration that affects Hrs-mediated block in early endosome fusion. During isolation, endosomes lose their Ca 2ϩ in 5-10 min through a vacuolar-type ATPase (27). To overcome this, we loaded HeLa cells with extracellular calcium in the presence of bafilomycin, an inhibitor of vacuolar-type ATPases (28,29). Bafilomycin was maintained at a constant level during the lysis and isolation of endosomal membranes used in the fusion assay, which would block release of luminal Ca 2ϩ . In control cells, Hrs inhibited early endosome fusion by 74.5 Ϯ 3.7% when it was present throughout the fusion reaction (Fig. 6a, left panel). In the continued presence of bafilomycin, fusion was inhibited by 79.3 Ϯ 5.6% when Hrs was incubated with the endosomes and washed out prior to the beginning of fusion reactions (Fig. 6a,  rightpanel). However, in the absence of bafilomycin, the effect of Hrs addition and removal prior to the fusion reactions was significantly decreased (74.5 Ϯ 3.7% versus 20.7 Ϯ 5.6% of control; Fig. 6a, right panel). This nearly 4-fold increase sug-gested that luminal Ca 2ϩ loss though a bafilomycin-sensitive V-ATPase can decrease the Hrs-dependent inhibition of early endosome fusion. To examine whether Ca 2ϩ release from luminal stores could dissociate Hrs from endosomal membranes, we performed a similar experiment and quantitatively examined the amount of Hrs present on the resulting endosomes. Similar amounts of Hrs were found on control endosomes and those in which bafilomycin was present throughout the fusion reaction (Fig. 6b, lanes 1 and 3). However, in the absence of bafilomycin, the amount of Hrs on the endosomal membranes decreased to 15.4 Ϯ 5.3% of control values (Fig. 6b, lane 2). These data suggest that luminal Ca 2ϩ can both remove Hrs from endosomal membranes and reverse the inhibition of endosomal fusion produced by Hrs. DISCUSSION The SNARE hypothesis offers a molecular explanation for how membranes can overcome the energy barrier for fusion. Pairing of SNARE proteins on opposing membranes is required for, and may underlie the specificity of, the fusion reaction (1, 7). Regulation of SNARE complex formation/dissociation is crit- ical for the control of whether and when membrane fusion occurs. However, the control of SNARE complex assembly/ disassembly is poorly understood. The endosomal SNAREs, SNAP-25 and syntaxin 13, exist in at least two distinct heterotrimeric complexes; one is fusogenic and includes VAMP-2, and the other is nonfusogenic and includes Hrs (17,20). The Hrscontaining complex appears to predominate in vitro and prevents formation of the VAMP-2-containing complex (17). This suggests a mechanism by which Hrs can inhibit endosome fusion. We have examined the regulation of the Hrs-and VAMP-2-containing complexes and find that although the VAMP-2-containing complex is dissociated by the action of the NSF ATPase, NSF does not dissociate the Hrs-containing complex. In contrast, Ca 2ϩ dissociates the Hrs-containing complex without affecting the VAMP-2-containing complex. The micromolar concentrations of Ca 2ϩ required for the dissociation of the Hrs-containing complex are similar to those that remove Hrs from endosomal membranes and that reverse the Hrsinduced inhibition of VAMP-2-dependent homotypic endosome fusion. This is the first demonstration of differential regulation of R/Q-SNARE and all Q-SNARE containing SNARE complexes. Furthermore, these data suggest a mechanism by which the regulation of Hrs-and VAMP-2-containing complexes allows for the regulation of homotypic endosome fusion. Thus, Hrs binding to SNAP-25 on early endosomal membranes negatively regulates trans-SNARE pairing, and once Ca 2ϩ dissociates Hrs, a VAMP-2-containing complex forms, allowing fusion to occur.
The ATPase activity of NSF has been shown to dissociate SNARE complex protein interactions (1, 3, 6, 7) and recently to dissociate other protein complexes (30 -33), suggesting a con-served function as a dissociating factor. NSF binds indirectly to syntaxin family members via interactions with ␣SNAP and in the presence of Mg/ATP disrupts syntaxin⅐SNAP-25⅐VAMP complexes. The simplest explanation for the NSF resistance of the Hrs-containing complex is steric hindrance. Perhaps Hrs, being relatively larger than the SNAREs, blocks syntaxin from binding to ␣SNAP and NSF. Steric hindrance would likely not block calcium from dissociating an Hrs⅐SNAP-25⅐syntaxin 13 complex. The effect of Ca 2ϩ is specific because other divalent cations such as barium, copper, and manganese have no effect on the Hrs-containing complex. The role of Ca 2ϩ in fusion of organelle membranes has been recognized for years, although there are many possible effector proteins to which Ca 2ϩ binds (1,7,(11)(12)(13)34), making understanding of the mechanism of Ca 2ϩ action unclear. Ca 2ϩ has been suggested to play a role in conformational/structural changes in proteins, actions that could lead to protein complex assembly or dissociation.
SNAP-25 has been recently shown to have a role in early endosome fusion (17) in accord with its presence on endosome membranes (37). However, SNAP-25 has been almost exclusively associated with a function in exocytosis. Moreover, SNAP-25 has been thought to be present solely in neuronal and neuroendocrine tissues and cell lines (although see Ref. 25). We have presented evidence that SNAP-25 is present in HeLa cell lysate and on purified early endosomes isolated from HeLa cells, a cervical tumor-derived cell line. The presence of SNAP-25 on endosomes in these cells is a further suggestion that SNAP-25 may be present in various cells from different lineages where it may be involved in membrane fusion events (25).
The mechanism of the membrane association of Hrs has been FIG. 7. A model of how calcium may regulate the formation of fusogenic and nonfusogenic SNARE complexes on early endosomes. a, the concentration of calcium near two early endosomes is below that amount required to release Hrs from a Q-SNARE complex with SNAP-25 and syntaxin 13. Fusion of the endosomes is blocked. b, an unknown trigger activates a calcium channel in an early endosome (one endosome shown for simplicity). The concentration of calcium rises locally to the point that causes Hrs to dissociate from the Q-SNARE complex, most likely because of a putative conformational change. c, once calcium removes Hrs, VAMP-2 (or other R-SNARES) can then form a fusogenic SNARE complex with SNAP-25 and syntaxin 13, facilitating fusion of early endosomes. This model is based in part on a model describing how calcium might regulate vacuole fusion in yeast (11). the subject of some debate (Ref . 17 and references therein). Although the FYVE domain may provide a link to the membrane through an interaction with phosphatidylinositol 3-phosphate (38), the Q-SNARE domain of Hrs interacts with SNAP-25 and SNAP-25 can act as a saturable binding site for Hrs on endosomal membranes (17). Because the FYVE domain may partially penetrate membranes and regulate residence time (38), it is possible that SNAP-25 is the protein receptor and that the FYVE-lipid interaction may be a regulatory influence for Hrs binding to endosomes. Ca 2ϩ dissociates Hrs from early endosomal membranes at concentrations similar to that required for the dissociation of the Hrs⅐SNAP-25 proteins complex. How relevant are these Ca 2ϩ concentrations for membrane fusion in the endocytic pathway? The luminal concentration of Ca 2ϩ in endosomes has been estimated at 1 mM (35), 360 M in the Golgi, 350 M in the ER (26), and 2 mM in yeast vacuole (11). Resting Ca 2ϩ in HeLa cells is ϳ100 nM creating a large driving force for Ca 2ϩ , and the concentration of Ca 2ϩ at the endosome surface directly outside of the pore could likely reach 100 M or greater (36). Ca 2ϩ in the micromolar range can reverse the effect of chelators and support endosome fusion (12,13). For homotypic vacuolar fusion, 100 M Ca 2ϩ in the medium is required to establish the normal luminal Ca 2ϩ concentration and support fusion (11). These data are consistent with a model in which luminal Ca 2ϩ release, triggered by an unknown mechanism, results in Ն100 M Ca 2ϩ in discrete domains on the endosome membrane, resulting in dissociation of Hrs from SNAP-25 and allowing the VAMP-2⅐SNAP-25⅐syntaxin 13 complex to form that enables fusion (Fig. 7).
Although tetanus toxin light chain inhibits the fusion of early endosomes in a saturable and concentration-dependent manner, the maximal inhibition obtained is ϳ50%. In this regard, Hrs can completely inhibit the formation of the syntaxin 13⅐SNAP-25⅐VAMP-2 complex, and it does not completely inhibit early endosome fusion. Moreover, BoNT/E inhibits early endosome fusion to the same extent as Hrs. The incomplete inhibition of endosome fusion may be related to the inability of tetanus toxin to affect previously formed trans-SNARE complexes. Alternatively, these data may suggest the involvement of multiple R-SNAREs (39,40) in early endosomal fusion.
Hrs inhibits early endosome fusion in a Ca 2ϩ -reversable manner. The concentrations of Ca 2ϩ required for the reversal of fusion, the dissociation of Hrs from the SNAP-25⅐syntaxin 13 complex, and the removal of Hrs from endosomal membranes are consistent. If VAMP-2 is present when Ca 2ϩ induces the removal of Hrs from the SNAP-25⅐syntaxin 13 complex, then VAMP-2 will enter the SNARE complex, even in the presence of Ca 2ϩ , and can support fusion. These data suggest differential regulation of Hrs-and VAMP-2-containing early endosomal SNARE complexes and suggest a biochemical pathway by which luminal Ca 2ϩ may regulate endosome fusion.