Ca2+-dependent formation of a dynamin-synaptophysin complex: potential role in synaptic vesicle endocytosis.

Synaptophysin is a synaptic vesicle (SV) protein of unknown function. Here we show that a repeated sequence in the cytoplasmic tail of synaptophysin mediates the formation of a protein complex containing the GTPase dynamin. The formation of this complex requires a high Ca(2+) concentration, suggesting that it occurs preferentially at the sites of SV exocytosis. Coimmunoprecipitation of a dynamin-synaptophysin complex from brain extracts is promoted by dissociation of vesicle-associated membrane protein 2 from synaptophysin. This finding suggests that dynamin only associates with synaptophysin in vivo after vesicle-associated membrane protein 2 (VAMP2) enters the SNARE complex. GTP binding releases dynamin from synaptophysin, possibly serving to regulate dynamin selfassembly during endocytosis. Our results suggest that synaptophysin plays a role in SV recycling by recruiting dynamin to the vesicle membrane.

Synaptophysin is a synaptic vesicle (SV) protein of unknown function. Here we show that a repeated sequence in the cytoplasmic tail of synaptophysin mediates the formation of a protein complex containing the GTPase dynamin. The formation of this complex requires a high Ca 2؉ concentration, suggesting that it occurs preferentially at the sites of SV exocytosis. Coimmunoprecipitation of a dynamin-synaptophysin complex from brain extracts is promoted by dissociation of vesicle-associated membrane protein 2 from synaptophysin. This finding suggests that dynamin only associates with synaptophysin in vivo after vesicleassociated membrane protein 2 (VAMP2) enters the SNARE complex. GTP binding releases dynamin from synaptophysin, possibly serving to regulate dynamin selfassembly during endocytosis. Our results suggest that synaptophysin plays a role in SV recycling by recruiting dynamin to the vesicle membrane.
Synaptophysin is one member of a family of SV 1 proteins with four transmembrane domains (1,2). In addition to synaptophysin, SVs contain the closely related protein synaptoporin (3,4) and the distantly related protein synaptogyrin (5). The function of these proteins is currently unknown. Synaptic transmission, synaptic plasticity, and SV number are unaffected in synaptophysin knockout mice (6,7), although synaptophysin-synaptogyrin double knockout mice exhibit defects in synaptic plasticity (8). The molecular basis for these defects is unknown.
The only protein known to interact directly with synaptophysin is vesicle-associated membrane protein 2 (VAMP 2) (9, 10), which mediates vesicle fusion via SNARE complex formation with syntaxin and SNAP 25. Synaptophysin-VAMP 2 complexes purified from synaptosomal extracts do not contain syntaxin 1 or SNAP 25, suggesting that VAMP 2 dissociates from synaptophysin prior to entering the SNARE complex (10). These data raise the possibility that the association of VAMP 2 with synaptophysin serves to regulate the availability of VAMP 2 for SNARE complex formation. This hypothesis is supported by a study demonstrating an inverse correlation between the basal rate of SV exocytosis and the extent of synaptophysin-VAMP 2 association (11).
Synaptophysin has also been shown to bind cholesterol, which appears to be required for SV formation (12). This observation suggests a function for synaptophysin in the biogenesis of SVs, a role that had been proposed previously based on the ability of synaptophysin to induce the formation of small clear vesicles in non-neuronal cells (13).
To help understand synaptophysin function, we sought to identify synaptophysin binding partners. In this report, we describe a Ca 2ϩ -dependent interaction between synaptophysin and dynamin I, a GTPase required for SV endocytosis (14,15). Our results suggest that synaptophysin may function in vesicle recycling. Indeed, this hypothesis is supported by recent in vivo experiments performed in the squid giant synapse (16).

EXPERIMENTAL PROCEDURES
Plasmid Construction-Fragments of the synaptophysin cDNA (17) (a gift of T. Sudhof) encoding various regions of the synaptophysin C terminus were amplified by PCR and cloned into the BamHI-SalI sites of the pGEX-4T-1 vector (Amersham Biosciences, Inc.), creating plasmids that encode GST-SYP fusion proteins. Fragments of the synaptophysin II (3) and synaptogyrin (5) cDNAs encoding their C-terminal cytoplasmic domains (amino acids 199 -265 of synaptophysin II and amino acids 170 -234 of synaptogyrin) and a fragment of the amphiphysin II cDNA encoding amino acids 491-588 of the SH3 domain were amplified from a rat brain cDNA library by PCR and cloned into pGEX-4T-1.
Binding Assays Using Immobilized GST Fusion Proteins-The various GST fusion proteins were expressed in Escherichia coli and purified on glutathione-agarose beads. A washed synaptosome fraction was prepared from the cerebral cortex of 6 -8-week-old Sprague-Dawley rats essentially as described previously (18). Synaptosomes were extracted in 50 mM NaCl, 20 mM HEPES, pH 7.4, 1% Triton X-100, and a protease inhibitor mixture (Complete TM protease inhibitor without EDTA, Roche Molecular Biochemicals) for 1 h at 4°C and then clarified by centrifugation at 15,000 ϫ g for 10 min. Synaptosomal extracts at a protein concentration of ϳ7.5 mg/ml (unless stated otherwise) were incubated in a volume of 1 ml with ϳ50 g of the indicated GST fusion proteins immobilized on 20 l of glutathione-agarose beads (Santa Cruz Biotechnology, Inc.) for 2-3 h at 4°C. Binding reactions generally contained either 2 mM EGTA or CaCl 2 unless noted otherwise. After binding, the beads were washed three times with 1 ml of extraction buffer containing EGTA or Ca 2ϩ . Bound proteins were eluted by boiling in SDS gel sample buffer and resolved on an SDS gel.
Effect of GTP on Dynamin Association with Synaptophysin-To test the effect of nucleotides on the ability of dynamin to bind to synaptophysin (Fig. 5A), synaptosomal extracts were incubated with GST-SYP in the presence of Ca 2ϩ and either no nucleotide or 10 M GDP or 100 M GTP␥S. All samples contained 1 mM Mg 2ϩ . To test whether GTP could release dynamin from synaptophysin (Fig. 5B), synaptosomal extracts were incubated with GST-SYP in the presence of Ca 2ϩ . The beads containing the dynamin-synaptophysin complexes were then washed three times with buffer containing no nucleotide or 10 M GDP, 100 M GTP, or 100 M GTP␥S. All wash buffers contained 1 mM Mg 2ϩ . After washing, the amount of dynamin remaining bound to synaptophysin was assessed by immunoblot.
Immunoprecipitations-Synaptosomes were extracted in 0.15 M NaCl, 20 mM HEPES, pH 7.4, 1% Triton X-100, and a protease inhibitor mixture without EDTA for 1 h at 4°C and then clarified at 15,000 ϫ g for 10 min. Synaptosomal extracts at a protein concentration of ϳ2.5 mg/ml in a volume of 1 ml were precleared with 15 l of protein A or protein G-agarose beads (Santa Cruz Biotechnology, Inc.) for 1 h. Extracts were then incubated at 4 or 30°C for 30 min in the presence of 2 mM EGTA or 2 mM CaCl 2 , 2 mM MgCl 2 as indicated in the legend to Fig.  4. The Ca 2ϩ /Mg 2ϩ treatment served to dissociate synaptophysin from VAMP, the dissociation does not require the addition of ATP (19)). Extracts were then placed on ice and immunoprecipitated with 3 g of a monoclonal antibody against dynamin I (D25520, Transduction Laboratories), 10 l of a rabbit antiserum against VAMP 2 (SV006, Stressgen Biotechnology), or 1-2 l (ascites fluid) of a monoclonal antibody against syntaxin I (HPC-1, Sigma). Immunoglobulins were collected with 15 l of either protein A-or protein G-agarose beads for 1 h at 4°C. Immunoprecipitates were washed three times with 1 ml of extraction buffer containing EGTA or Ca 2ϩ . Bound proteins were eluted and resolved on an SDS gel.
Immunoblots-The following antibodies were used in immunoblots at the indicated concentrations: monoclonal antibody against dynamin I (D25520, Transduction Laboratories) at 0.5 g/ml; monoclonal antibody against synaptophysin (SY38, Roche Molecular Biochemicals) at 0.5 g/ml; monoclonal antibody against syntaxin 1 (HPC-1, Sigma) at 1:1000 dilution; polyclonal antibody against VAMP 2 (SV006, Stressgen Biotech) at 1:1000 dilution (  (21). Tail-To investigate whether synaptophysin associates with synaptic proteins other than VAMP 2, a fusion protein of GST and the 89 amino acid C terminus of synaptophysin (GST-SYP) was immobilized on beads and incubated with a synaptosomal extract. Because Ca 2ϩ is known to modulate many protein-protein interactions in synapses, the binding was done in the presence of EGTA or Ca 2ϩ . Synaptosomal proteins that bound to GST or GST-SYP were eluted, resolved on an SDS gel, and visualized by Coomassie Blue staining. In the presence of Ca 2ϩ but not EGTA, GST-SYP bound to a protein of approximately 100 kDa (Fig. 1B). This protein, which was not bound by GST in the presence of either EGTA or Ca 2ϩ (Fig. 1B), was subsequently identified by immunoblot as dynamin I, a neuron-specific GTPase required for synaptic vesicle endocytosis (14,15) (Fig. 1C). GST-SYP did not bind to the synaptic proteins syntaxin 1A and synaptojanin, confirming the specificity of the interaction (Fig. 1C).

Ca 2ϩ -dependent Association of Dynamin with the Synaptophysin-cytoplasmic
To determine the concentration of Ca 2ϩ required for the association of dynamin with synaptophysin, synaptosomal extracts were incubated with GST-SYP in the presence of EGTA or Ca 2ϩ at concentrations of 50, 100, 200, 300, 400, or 500 M. Bound proteins were eluted and subjected to immunoblot with anti-dynamin antibody ( Fig. 2A). The binding of dynamin was very low at 50 M Ca 2ϩ , increased a little at 100 M, and then increased dramatically at 200 M after which binding leveled off ( Fig. 2A). Thus, the association of dynamin and synaptophysin requires a high Ca 2ϩ concentration with half-maximal binding occurring at approximately 150 M (Fig. 2B). This level of Ca 2ϩ closely matches that which is present in microdomains at sites of SV exocytosis (22) as a result of the physical association of voltage-gated calcium channels with the SV fusion machinery (23). Thus, our results suggest that the association of dynamin with synaptophysin is restricted to sites of SV exocytosis, consistent with the possibility that this complex has a role in vesicle recycling. Because dynamin has been shown to bind Ca 2ϩ with low affinity (24), the effect of Ca 2ϩ on the formation of the dynamin-synaptophysin complex probably occurs through an effect of Ca 2ϩ on dynamin conformation, al-though other possibilities can be envisioned.
Synaptic vesicles contain two proteins that are structurally related to synaptophysin, synaptoporin-synaptophysin II (3,4) and synaptogyrin (5). Synaptophysin has a 58% amino acid identity with synaptophysin II (3) but has little sequence homology to synaptogyrin (5). The C termini of synaptophysin and synaptophysin II are divergent with the exception of the five juxtamembrane amino acids, KETGW, and the last eight amino acids, PTSF(S/N)NQ(M/I) (Fig. 3A, underlined). The synaptophysin C terminus is characterized by nine repeats of a degenerate pentapeptide with the consensus YG(P/Q)QG (2) (Fig. 3A, boxed). GST fusion proteins containing the C termini of synaptophysin II and synaptogyrin were tested for their ability to pull down dynamin from brain extract in the presence of EGTA or Ca 2ϩ . The C termini of both synaptophysin II and synaptogyrin associated with dynamin in a Ca 2ϩ -dependent fashion but much less efficiently than synaptophysin (Fig. 3B). Thus, these three SV proteins are not entirely redundant with respect to this function, as dynamin associates preferentially with the cytoplasmic tail of synaptophysin.
To map the synaptophysin sequence required for association with dynamin, we employed a series of truncated GST-SYP fusion proteins. Each was tested for its ability to bind dynamin from brain extract in the presence of Ca 2ϩ . A progressive removal of the YG(P/Q)QG repeats resulted in a progressive loss of dynamin binding (Fig. 3C). Thus, the ability of the synaptophysin C terminus to associate with dynamin is dependent on FIG. 1. Ca 2؉ -dependent binding of dynamin to the C-terminal cytoplasmic tail of synaptophysin. A, topology of synaptophysin in the synaptic vesicle membrane. The arrow indicates the region of the protein (amino acids 219 -307), which was fused to GST and used as bait in the binding studies. B, the GST-SYP fusion protein and GST alone were immobilized on glutathione-agarose beads and incubated with a synaptosomal extract (protein concentration ϳ10.5 mg/ml) in the presence of EGTA (E) or Ca 2ϩ (Ca). Bound proteins were eluted by boiling in SDS gel sample buffer, resolved on an SDS gel, and visualized by staining with Coomassie Blue. The arrow indicates a protein of ϳ100 kDa that was bound specifically by GST-SYP in the presence of Ca 2ϩ . C, a synaptosomal extract was incubated with GST or GST-SYP in the presence of EGTA or Ca 2ϩ . Bound proteins were eluted, resolved on an SDS gel along with an aliquot of the extract, and subjected to immunoblot with an anti-dynamin antibody (top), an anti-syntaxin 1 antibody (middle), or an anti-synaptojanin antibody (bottom). a region of approximately 60 amino acids that is composed almost entirely of the nine YG(P/Q)QG repeats. The precise requirements for the formation of the dynamin-synaptophysin complex will require more extensive mutagenesis of the synaptophysin C terminus. However, this result may explain the ability of synaptophysin II and synaptogyrin to associate with dynamin albeit with much lower affinity (Fig. 3B). Both synaptophysin II and synaptogyrin have sequences that are related to the synaptophysin repeats, for example, YGSSG and YSQQA in synaptophysin II and YQSQG in synaptogyrin (Fig. 3A).
Association of Dynamin with Synaptophysin Is Promoted by Dissociation of Synaptophysin from VAMP-It has been shown recently that the synaptophysin-VAMP 2 complex can be dissociated in synaptosomal extracts by incubation at 30°C in the presence of Ca 2ϩ and Mg 2ϩ (19). Although the mechanism underlying this effect is undefined, it seems likely that the addition of Ca 2ϩ to the extract activates a process that is normally triggered in vivo by depolarization-induced Ca 2ϩ influx (19). A period of strong synaptic activity, by promoting synaptophysin-VAMP 2 dissociation, would thereby replenish the pool of SVs in which VAMP 2 is free to enter the SNARE complex. Because dynamin presumably binds synaptophysin after SNARE complex formation, it seemed possible that the disruption of synaptophysin-VAMP 2 complexes in synaptosomal extracts might release synaptophysin to bind dynamin. The incubation of synaptosomal extracts at 30°C for 30 min in the presence of Ca 2ϩ but not in the presence of EGTA almost completely dissociated synaptophysin from VAMP 2 as assessed by coimmunoprecipitation of synaptophysin with anti-VAMP 2 antibody (Fig. 4A). As shown in Fig. 4B, following incubation of extracts at 30°C in the presence of Ca 2ϩ but not in the presence of EGTA, synaptophysin coimmunoprecipitated with dynamin. Thus, under conditions in which synaptophysin dissociates from VAMP 2, a dynamin-synaptophysin complex forms. Anti-dynamin antibody did not immunoprecipitate syntaxin 1 in the presence of either EGTA or Ca 2ϩ , and antisyntaxin 1 antibody did not immunoprecipitate either synap-

FIG. 2. Binding of dynamin to synaptophysin requires a Ca 2؉ concentration found only at sites of synaptic vesicle exocytosis.
A, a synaptosomal extract was incubated with GST-SYP in the presence of EGTA (E) or Ca 2ϩ at concentrations of 50, 100, 200, 300, 400, or 500 M. Bound proteins were eluted, resolved on an SDS gel, and subjected to immunoblot with an anti-dynamin antibody. B, the graph illustrates relative dynamin binding to GST-SYP as a function of Ca 2ϩ concentration. Data points (% maximal binding Ϯ S.E.) were generated by quantitating signals from immunoblots using NIH Image software.

FIG. 3. Dynamin binds to a repeated sequence in the synaptophysin C terminus.
A, sequences of the cytoplasmic C-terminal domains of synaptophysin (syp), synaptophysin II (syp II), and synaptogyrin (sgyrin). The conserved juxtamembrane and extreme C-terminal sequences of synaptophysins are underlined. The nine pentapeptide repeats in the synaptophysin C terminus with consensus sequence YG(P/Q)QG are boxed. B, GST fusion proteins containing the C-terminal cytoplasmic domains of synaptophysin (SYP), synaptophysin II (SYP II), or synaptogyrin (SGYRIN) were immobilized on glutathione-agarose beads and incubated with a synaptosomal extract in the presence of EGTA (E) or Ca 2ϩ (Ca). Bound proteins were eluted by boiling in SDS gel sample buffer, resolved on an SDS gel, and subjected to immunoblot with an anti-dynamin antibody. After immunoblotting (IB), the filter was stained with Ponceau S to reveal the GST fusion proteins. C, GST fusion proteins containing the indicated regions of the synaptophysin C terminus were immobilized on glutathione-agarose beads and incubated with a synaptosomal extract in the presence of Ca 2ϩ . Bound proteins were eluted by boiling in SDS gel sample buffer, resolved on an SDS gel, and subjected to immunoblot with an anti-dynamin antibody. After immunoblotting, the filter was stained with Ponceau S to reveal the GST fusion proteins. The sites of the truncations are indicated in A, ⌬257-274 is a deletion of the indicated amino acids. tophysin or dynamin, demonstrating the specificity of the immunoprecipitations (Fig. 4B). As shown in Fig. 4C, after incubation of extracts at 4°C in the presence of Ca 2ϩ , the synaptophysin-VAMP 2 complex remained intact, and antidynamin antibody coimmunoprecipitated very little synaptophysin. In contrast, after incubation at 30°C in the presence of Ca 2ϩ , the synaptophysin-VAMP 2 interaction was disrupted, and at the same time the amount of synaptophysin coimmunoprecipitated by anti-dynamin antibody increased significantly (Fig. 4C). This finding suggests that in addition to requiring Ca 2ϩ , the interaction of dynamin with synaptophysin is promoted by dissociation of synaptophysin from VAMP 2. This result indicates that synaptophysin forms two distinct complexes in nerve terminals, one with VAMP 2 and one with dynamin. The fact that the association of synaptophysin and dynamin is dependent on dissociation of synaptophysin and VAMP 2 is consistent with a role for the dynamin-synaptophysin complex in a step of the SV cycle after SNARE complex formation.
GTP Binding Releases Dynamin from Synaptophysin-Dynamin function in endocytosis requires that it dissociate from the proteins responsible for recruiting it to the membrane and then self-assemble into a ring around the vesicle neck (14,15). The binding of GTP by dynamin has been proposed to play a role in this redistribution presumably by inducing a change in dynamin conformation (25,26). Thus, we examined the effect of nucleotides on the dynamin-synaptophysin complex. Because of the high intrinsic rate of GTP hydrolysis exhibited by dynamin (14,15), GTP␥S was used to test the properties of GTP-bound dynamin. As shown in Fig. 5A, the binding of dynamin to GST-SYP was dramatically inhibited by GTP␥S but not by GDP, indicating that dynamin can associate with synaptophysin in its unoccupied or GDP-bound state but not in its GTP-bound state. In contrast, the association of dynamin with amphiphysin was unaffected by GTP␥S (Fig. 5A, right). To test whether GTP binding releases dynamin from synaptophysin after complex formation has occurred, dynamin was bound to GST-SYP in the absence of nucleotide after which the complex was washed with buffer containing no nucleotide or either GDP or GTP. Washing in the presence of GTP specifically released prebound dynamin from the complex (Fig. 5B, left). This finding suggests that the change in dynamin conformation induced by GTP binding promotes its dissociation from synaptophysin. To test whether the release of dynamin from synaptophysin requires GTP hydrolysis, the wash was performed with GTP␥S instead of GTP (Fig. 5B, middle). GTP␥S also released dynamin from synaptophysin, indicating that GTP binding but not GTP hydrolysis is required for release. In contrast, the amphiphysin-dynamin complex was unaffected by the GTP wash (Fig. 5B, right). The sensitivity of the synaptophysin-dynamin complex to the GTP-induced change in dynamin conformation suggests a possible mechanism for regulated release of dynamin from synaptophysin after recruitment to SVs. This outcome might serve to free dynamin for selfassembly and subsequent vesicle fission. DISCUSSION The results presented in this report indicate that dynamin interacts with synaptophysin in a Ca 2ϩ -dependent fashion. In addition, we show that the interaction is modulated by dissociation of synaptophysin from VAMP and by GTP binding to dynamin. Given the well established function of dynamin in endocytosis (14,15), it seems most probable that the association of dynamin with synaptophysin plays a role in vesicle recycling by targeting dynamin to the SV membrane. Indeed, we have recently shown that injection of the synaptophysin C terminus into the squid giant synapse results in a block of vesicle recycling (16). The results presented here suggest that the recycling block results from a disruption of the dynaminsynaptophysin complex.
Several important questions regarding the role of a dynamin-synaptophysin complex in SV recycling remain to be addressed. First, do dynamin and synaptophysin interact directly, or does the association require additional synaptic proteins? Attempts to reconstitute a Ca 2ϩ -dependent interaction between GST-SYP and dynamin purified from rat brain have yielded inconclusive results (data not shown). One possible explanation for this is that a required protein is missing from the binding reaction. Coomassie Blue staining of the synaptic proteins which bind to the synaptophysin C terminus reveals that dynamin is the major protein (Fig. 1B). However, another protein(s) essential for the interaction might be present in very FIG. 4. Ca 2؉ -dependent coimmunoprecipitation of dynamin and synaptophysin from synaptosomal extracts is promoted by dissociation of synaptophysin from VAMP 2. A, synaptosomal extracts (ϳ2.5 mg/ml protein) in a volume of 1 ml were precleared with 15 l of protein A-agarose for 1 h. Extracts were then incubated at 30°C in the presence of EGTA (E) or Ca 2ϩ /Mg 2ϩ (Ca) for 30 min, placed on ice, and subjected to immunoprecipitation with anti-VAMP 2 antibody. Bound proteins were eluted, resolved on an SDS gel, and subjected to immunoblot with anti-synaptophysin antibody (top) or anti-VAMP 2 antibody (bottom). B, synaptosomal extracts were precleared with 15 l of protein G-agarose for 1 h. Extracts were then incubated at 30°C in the presence of EGTA or Ca 2ϩ /Mg 2ϩ for 30 min, placed on ice, and subjected to immunoprecipitation with anti-dynamin antibody (␣ dyn) or anti-syntaxin 1 antibody (␣ syn 1). Bound proteins were eluted, resolved on an SDS gel, and subjected to immunoblot with anti-synaptophysin antibody (top), anti-dynamin antibody (middle), or anti-syntaxin 1 antibody (bottom). C, synaptosomal extracts were precleared with 15 l of protein A-agarose for ␣-VAMP 2 immunoprecipitation (IP) or protein G-agarose for ␣-dynamin IP for 1 h. Extracts were then incubated at 4°C or at 30°C for 30 min in the presence of Ca 2ϩ /Mg 2ϩ . Extracts were placed on ice and subjected to immunoprecipitation with anti-dynamin antibody (␣ dyn) or anti-VAMP 2 antibody (␣ V2). Bound proteins were eluted, resolved on an SDS gel, and subjected to immunoblot with anti-synaptophysin antibody (top), anti-dynamin antibody (middle), or anti-VAMP 2 antibody (bottom). low amounts or obscured by the GST-SYP fusion protein. Alternatively, the interaction may be dependent on a particular conformation or post-translational modification of dynamin, which is lost in the purification process.
A second question concerns the relatively low amount of dynamin that associates with the GST-SYP fusion protein (Fig.  1B). Several factors may contribute to this result. Given the wide array of SH3 domain proteins with which dynamin can interact (27), it is possible that only a small pool of dynamin in the synapse is available to bind synaptophysin. Alternatively, as mentioned above, post-translational modifications might limit the pool of dynamin in a brain extract which is competent to associate with synaptophysin. For example, the dynamin phosphorylation state is known to be dramatically regulated by synaptic activity (28,29). Clearly, further investigation is required to define the parameters that control the formation of the dynamin-synaptophysin complex. However, despite these uncertainties, we have demonstrated that dynamin and synaptophysin can be coimmunoprecipitated from brain extracts, strongly suggesting that these proteins can form a complex in vivo.
One of the most intriguing aspects of the dynamin-synaptophysin complex is its dependence on a high Ca 2ϩ concentration (Fig. 2). This finding suggests that the formation of the complex is restricted to sites of SV exocytosis, at which high Ca 2ϩ concentrations are found (22). Although dynamin does not possess a conventional Ca 2ϩ binding site, it has been shown to directly bind Ca 2ϩ (24). In addition, Ca 2ϩ has been shown to modulate the activity of dynamin in liposome vesiculation assays (30), indicating that Ca 2ϩ can regulate dynamin function. Furthermore, these experiments suggested that dynamin binds Ca 2ϩ with low affinity, implying that Ca 2ϩ modulation of dynamin function occurs preferentially around vesicle release sites. Taken together, these results suggest the possibility that the Ca 2ϩ dependence of the dynamin-synaptophysin association reflects a Ca 2ϩ -induced change in dynamin conformation.
We have shown that the dynamin-synaptophysin complex is dissociated by GTP (Fig. 5), suggesting that it is sensitive to dynamin conformation. It is unclear what fraction of dynamin in vivo is normally GTP bound. Although cytosolic GTP concentrations are generally high, dynamin has a very low affinity for GTP and has a high intrinsic GTPase activity (14,15). Thus, it is not unreasonable to propose that there is a pool of dynamin at the synapse which is bound to GDP or not bound by nucleotide, and is therefore competent to associate with synaptophysin. Presently, it is unclear as to what is the functional significance of the GTP regulation of the dynamin-synaptophysin complex. One possibility is that it promotes dynamin release from synaptophysin after recruitment to the vesicle, allowing dynamin self-assembly and subsequent endocytosis. This model assumes that after recruitment to the vesicle, dynamin would rapidly associate with GTP. It is possible that the association of dynamin with synaptophysin or with lipids in the vesicle membrane promotes GTP binding, although at present there is no evidence in support of this hypothesis.
The most critical question concerning the dynamin-synaptophysin complex is what is its importance to SV recycling? The evidence supporting a functional role for the complex is that injection of the synaptophysin C terminus into the squid giant synapse preterminal disrupts vesicle recycling (16). It is presently unknown to what extent a synaptophysin-dependent recycling pathway plays a significant role in other synapses. If there is more than one mechanism to recruit dynamin to vesicles (i.e. besides the well characterized amphiphysin (14)), different synapses might rely to varying degrees on a mechanism that requires synaptophysin. In addition, a single synapse under different stimulation conditions might rely preferentially on different endocytic mechanisms.
An analysis of synaptophysin knockout and synaptophysinsynaptogyrin double knockout mice has not thus far revealed a role for synaptophysin in vesicle recycling (6 -8). For instance, hippocampal synapses in synaptophysin knockout mice do not exhibit any decrease in the number of SVs (7) as might be expected if synaptophysin is essential for vesicle recycling. However, the clathrin-dependent recycling pathway in which dynamin is recruited to vesicles via amphiphysin (14) is presumably functional at these synapses and serves to maintain the vesicle pool. In addition, as mentioned above, it is possible that the loss of synaptophysin affects vesicle recycling at a particular subset of synapses. Further work will be required to evaluate the importance of the dynamin-synaptophysin complex in SV recycling at various synapses.
Acknowledgments-We thank P. Robinson and J. Hinshaw for purified dynamin, R. Jahn and P. McPherson for antibodies, T. Sudhof for FIG. 5. GTP binding releases dynamin from synaptophysin. A, GST-SYP or GST-Amph SH3 were immobilized on glutathione-agarose beads and incubated with a synaptosomal extract. The binding reactions were supplemented with Mg 2ϩ (1 mM) and GDP (10 M) or GTP␥S (100 M), as indicated. Bound proteins were eluted, resolved on an SDS gel, and subjected to immunoblot with antidynamin antibody. B, dynamin-synaptophysin or dynamin-amphiphysin complexes were formed by incubating synaptosomal extracts with GST-SYP in the presence of Ca 2ϩ (left) or with GST-Amph SH3 (right). Half of each sample was then processed for immunoblotting as usual to confirm that approximately equal levels of dynamin binding had occurred (top). The rest of each sample was washed with buffer containing Mg 2ϩ (1 mM) and GDP (10 M), GTP (100 M), or GTP␥S (100 M), as indicated. The dynamin, which remained bound after washing, was eluted, resolved on an SDS gel, and subjected to immunoblot with anti-dynamin antibody (bottom).