Identification of the major synaptojanin-binding proteins in brain.

Synaptojanin is a nerve-terminal enriched inositol 5-phosphatase thought to function in synaptic vesicle endocytosis, in part through interactions with the Src homology 3 domain of amphiphysin. We have used synaptojanin purified from Sf9 cells after baculovirus mediated expression in overlay assays to identify two major synaptojanin-binding proteins in rat brain. The first, at 125 kDa, is amphiphysin. The second, at 40 kDa, is the major synaptojanin-binding protein detected, is highly enriched in brain, is concentrated in a soluble synaptic fraction, and co-immunoprecipitates with synaptojanin. The 40-kDa protein does not bind to a synaptojanin construct lacking the proline-rich C terminus, suggesting that its interaction with synaptojanin is mediated through an Src homology 3 domain. The 40-kDa synaptojanin-binding protein was partially purified from rat brain cytosol through a three-step procedure involving ammonium sulfate precipitation, sucrose density gradient centrifugation, and DEAE ion-exchange chromatography. Peptide sequence analysis identified the 40-kDa protein as SH3P4, a member of a novel family of Src homology 3 domain-containing proteins. These data suggest an important role for SH3P4 in synaptic vesicle endocytosis.

Synaptic vesicles are specialized organelles that neurons use to secrete nonpeptide neurotransmitters. Following neurotransmitter release, synaptic vesicle membranes are retrieved by a process thought to involve clathrin-coated pits and vesicles (1,2), and recent data suggest that this endocytic mechanism is active at both high and low rates of exocytosis (3). We have identified a 145-kDa protein, referred to as synaptojanin, which is enriched in nerve terminals and appears to function in synaptic vesicle endocytosis (4 -6). Synaptojanin is an inositol 5-phosphatase that dephosphorylates inositol 1,4,5-trisphosphate, inositol 1,3,4,5-tetrakisphosphate, and phosphatidylinositol 4,5-bisphosphate at the 5Ј position of the inositol ring (6). Inositol phosphate metabolism has been implicated in a variety of membrane trafficking events including endocytosis (7). In addition, synaptojanin has an N-terminal domain that is homologous to the cytosolic domain of the yeast SacI protein.
SacI mutants show genetic interactions with actin as well as with the yeast secretory mutants sec6, sec9, and sec14 (8,9), and more recently SacIp has been demonstrated to mediate ATP transport into the yeast endoplasmic reticulum (10).
Synaptojanin was initially identified based on its ability to bind to the Src homology 3 (SH3) 1 domains of Grb2 (4). Cloning of synaptojanin revealed a 250-amino acid proline-rich domain at its C terminus (6) that contains at least five sequences forming potential SH3 domain-binding sites (11). A second, 170-kDa isoform of synaptojanin is present in a wide variety of tissues including neonate brain but is not detected in adult brain (6,12). The 170-kDa synaptojanin isoform is generated by alternative splicing of the synaptojanin gene leading to the presence of an additional 266-amino acid proline-rich domain with at least three additional SH3 domain-binding consensus sequences as compared with the 145-kDa isoform (12).
Synaptojanin also binds to the SH3 domain of amphiphysin (6). Amphiphysin was first identified in chicken synaptic fractions (13) and mammalian amphiphysin, which is concentrated in presynaptic nerve terminals, has been implicated in synaptic vesicle endocytosis (14). A role for amphiphysin in endocytosis is supported by studies on its yeast homologues, RVS 161 and RVS 167 (15)(16)(17). Mutations in these genes cause an endocytosis defect characterized in part by an impairment in ␣-factor receptor internalization (18). Further, amphiphysin is known to interact with AP2 (14,19), a component of the plasma membrane clathrin coat (20). Evidence of a role for the SH3 domain of amphiphysin in synaptic vesicle endocytosis is provided by its interaction with dynamin. A role for dynamin in endocytosis was first determined based on its identity with the gene product of the Drosphila shibire mutant (21,22). Mutations in Drosophila dynamin leads to a block in synaptic vesicle endocytosis (23), and recent data suggest that dynamin functions in the nerve terminal by mediating the fission of endocytic vesicles (24,25). Thus, it appears likely that SH3 domain-mediated interactions of amphiphysin with synaptojanin are important to the endocytic function of synaptojanin in vivo. SH3 domain interactions involving Grb2 have also been recently demonstrated to be important for clathrin-mediated endocytosis of the epidermal growth factor receptor in non-neuronal cells (26). Specifically, disruption of Grb2 interactions with the epidermal growth factor receptor blocks receptor endocytosis, and epidermal growth factor can stimulate a transient association of Grb2 with dynamin (26). Thus, SH3 domain-mediated interactions appear to function widely in clathrin-mediated endocytosis.
Here, we have used purified synaptojanin in overlay assays to identify its preferred binding targets in brain. In addition to amphiphysin, we identified a 40-kDa synaptojanin-binding protein that is highly enriched in brain, is concentrated in soluble synaptic fractions, and co-immunoprecipitates with synaptojanin. Purification and peptide sequence analysis revealed the 40-kDa protein as SH3P4, a novel SH3 domaincontaining protein that was identified from a mouse library screened with a Src SH3 ligand peptide (27). SH3P4, along with SH3P8 and SH3P13, define a family of similar proteins of unknown function (27). Our data strongly implicate SH3P4, and perhaps other family members, in synaptic vesicle endocytosis.
Expression and Purification of Synaptojanin Baculovirus Constructs-Spodoptera frugiperda (Sf9; Invitrogen) cells were grown at 27°C in suspension cultures in Sf-900 II SFM optimized serum-free medium (Life Technologies, Inc.) supplemented with gentamycin. The baculovirus transfer vectors were co-transfected with linear baculovirus into Sf9 cells, and recombinant baculovirus was selected by plaque assay as described (28). Positive colonies were confirmed by protein purification and Western blot, and high titer stocks (10 8 -10 9 plaque forming units/ml) were generated as described (28). For purification of synaptojanin constructs, 200-ml cultures of Sf9 cells (1.5 ϫ 10 6 cells/ml) were infected with ϳ1 ϫ 10 9 plaque forming units of baculovirus. After 72 h of growth, cells were washed with 4°C phosphate-buffered saline (20 mM NaPO 4 monobasic, 0.9% NaCl, pH 7.4) and lysed in 30 ml of buffer A (300 mM NaCl, 0.83 mM benzamidine, 0.23 mM phenylmethylsulfonyl fluoride, 0.5 g/ml aprotinin, 0.5 g/ml leupeptin, 50 mM HEPES-OH, pH 8.0) by homogenization in a glass Teflon homogenizer and two passes through a 255 ⁄8 gauge needle. Homogenates were spun at 12,000 ϫ g for 10 min, and Triton X-100 (0.1% final) and imidazole (20 mM final) were added to the supernatant before the addition of 0.5 ml of Ni-NTA-agarose (Qiagen Corp.). The samples were incubated overnight at 4°C and washed three times in 20 ml of ice-cold buffer A with 0.1% Triton X-100 and 20 mM imidazole, and bound proteins were eluted with 4 ϫ 1-ml incubations in buffer A with 0.1% Triton X-100 and 200 mM imidazole. Dynamin was purified from Sf9 cells after baculovirus-induced expression as described (29) and was a generous gift of Dr. Sandra Schmid (Scripps Research Institute).
Overlay Assays-Overlay assays using a glutathione S-transferase/ amphiphysin SH3 domain fusion protein were performed as described (4). For synaptojanin overlay assays, protein fractions on nitrocellulose membranes were blocked for 1 h in blotto (phosphate-buffered saline with 5% (w/v) nonfat dry milk), rinsed in water, and incubated overnight at 4°C in buffer B (150 mM NaCl, 3% bovine serum albumin, 0.1% Tween 20, 1 mM dithiothreitol, 0.83 mM benzamidine, 0.23 mM phenylmethylsulfonyl fluoride, 20 mM Tris-Cl, pH 7.4) containing approximately 10 g/ml of purified synaptojanin or the synaptojanin R-2 deletion construct. Transfers were then washed and incubated with affinity purified anti-synaptojanin antibody (Milo) (5) or 1852 (described below) in buffer B without dithiothreitol for 1 h at room temperature. After washing, transfers were incubated in goat anti-rabbit secondary antibody conjugated to horseradish peroxidase for 1 h in blotto and devel-oped using the ECL kit (Amersham Corp.).
Preparation of Membrane Fractions-Various tissues were dissected from adult male rats and were homogenized at 1:10 (w/v) in buffer C (0.83 mM benzamidine, 0.23 mM phenylmethylsulfonyl fluoride, 0.5 g/ml aprotinin, 0.5 g/ml leupeptin, 20 mM HEPES-OH, pH 7.4) with a polytron or glass Teflon homogenizer, followed by centrifugation for 5 min at 800 ϫ g max . The supernatant fractions were separated on SDS-PAGE on 5-16% or 3-12% gradient gels. Subcellular fractionation of brain homogenates to generate synaptic fractions was performed as described (5).
Immunoprecipitation Analysis-Amphiphysin immunoprecipitations were performed as described (14). For synaptojanin immunoprecipitations, a rat brain was homogenized at 1:10 (w/v) in buffer D (0.3 M sucrose, 0.83 mM benzamidine, 0.23 mM phenylmethylsulfonyl fluoride, 0.5 g/ml aprotinin, 0.5 g/ml leupeptin, 10 mM HEPES-OH, pH 7.4) with a polytron. The homogenate was centrifuged at 800 ϫ g max for 5 min, and the supernatant was then centrifuged at 180,000 ϫ g max for 2 h. The soluble supernatant was precleared with protein G-agarose for 4 h at 4°C and the precleared extracts were incubated overnight at 4°C with a monoclonal antibody against synaptojanin or a control monoclonal antibody raised against p75 LNGFR precoupled to protein G-agarose. Samples were washed five times in 1 ml of buffer D and were eluted with SDS sample buffer.
Purification and Identification of the 40-kDa Synaptojanin-binding Protein-Adult rat brains (typically four for each preparation) were homogenized at 1:5 (w/v) in buffer C with a polytron, and the extracts were centrifuged at 180,000 ϫ g max for 1 h. Ammonium sulfate powder was added slowly to the soluble supernatant with stirring until 20% saturation. After 45 min on ice, the sample was centrifuged at 2700 ϫ g max for 30 min, and the supernatant was removed and precipitated with ammonium sulfate to 40% saturation. The 20 -40% ammonium sulfate precipitate was resuspended in 16 ml of buffer C and was loaded on four 40-ml 2.5-15% linear sucrose gradients prepared in buffer C. The gradients were centrifuged in a Beckman VTi 50 rotor for 6 h at 45,000 rpm with slow acceleration and no brake. Gradient fractions (20 ϫ 2 ml) were analyzed by synaptojanin overlay assay, and peak 40-kDa synaptojanin-binding protein fractions were pooled and passed over a 5-ml column of DEAE-Sephacel (Pharmacia Biotech Inc.) equilibrated in buffer C. Samples were recirculated over the column at a flow rate of 0.2 ml/min for 16 h, and the column was then eluted into 20 4-ml fractions at 2 ml/min with an 80-ml linear gradient of 0 -0.5 M NaCl prepared in buffer C. Eluted fractions (80 l/fraction) were analyzed for the 40-kDa synaptojanin-binding protein by overlay assay. Alternatively, proteins from eluted fractions (1 ml/fraction) were precipitated with 50% ice-cold trichloroacetic acid with 0.03% sodium deoxycholate as a carrier and analyzed by Coomassie Blue staining of protein gels. The peak 40-kDa synaptojanin-binding protein fraction was concentrated, run on SDS-PAGE, and transferred to PVDF membranes. The 40-kDa protein was excised and subjected to Edman degradation but was found to have a blocked N terminus. Therefore, the sample was treated with cyanogen bromide (CNBr) to affect peptide bond cleavage at the C-terminal side of methionyl residues (30). The PVDF membrane was immersed in 100 l of freshly prepared CNBr cleavage solution (70 mg CNBr/ml in 70% formic acid), flushed with argon, sealed, and left 18 h at room temperature. The cleavage mixture was then dried in a speed vacuum centrifuge, and the PVDF pieces were subjected to sequence analysis in an Applied Biosystems model 470A protein sequencer equipped with an on-line Applied Biosystems model 120A phenylthiohydantoin analyzer (31) according to procedures as recommended by the manufacturer. Sequence analysis revealed multiple sequencing signals which were manually analyzed by overlaying sucessive high pressure liquid chromatography traces. The strengths of the multiple signals were ranked, and probable sequences were searched against protein data bases employing the Blastp algorithm.
Antibodies-A polyclonal anti-synaptojanin antibody (1852) was prepared by injection of a rabbit with 200 g of synaptojanin R-2 deletion construct in Titer-Max adjuvant (CytRx Corporation) using standard protocols. Serum was tested for immunoreactivity by Western blot against brain extracts and purified synaptojanin. Antibodies were affinity purified from serum against purified synaptojanin on PVDF membranes as described (5). A polyclonal antibody against synaptojanin purified from rat brain (Milo) was described previously (5). Polyclonal antibodies against amphiphysin were prepared as described (15) and were a generous gift of Drs. Carol David and Pietro De Camilli (Yale University). The monoclonal antibody against synaptojanin was raised against a glutathione S-transferase fusion protein encoding amino acids 1156 -1286 of synaptojanin in the laboratory of Dr. Pietro De Camilli and was a generous gift of Drs. Amy Hudson and Pietro De Camilli (Yale University). The monoclonal antibody against p75 LNGFR was prepared as described (32) and was a generous gift of Dr. Phil Barker (Montreal Neurological Institute) and Dr. Eric Shooter (Stanford University).

Expression and Purification of Synaptojanin Constructs in
Sf9 Cells-To study the SH3 domain-binding properties of synaptojanin, we generated full-length synaptojanin and a synaptojanin deletion construct (synaptojanin R-2) lacking the proline-rich C terminus for expression in Sf9 cells using baculovirus. The constructs had six histidine residues introduced at the N terminus to allow for their purification with nickel-agarose. Purification of nickel-binding proteins from Sf9 cell cultures infected with the full-length synaptojanin construct leads to the isolation of a 145-kDa protein (Coomassie, Fig. 1) that is strongly reactive with a polyclonal antibody against synaptojanin (Milo Western, Fig. 1). Infection of cultures with the synaptojanin R-2 construct leads to the production of a 120-kDa protein (Coomassie, Fig. 1) that does not react with the polyclonal antiserum raised against full-length synaptojanin (5) (Milo Western, Fig. 1), indicating that the antibodies are directed entirely against the last 231 amino acids of the proline-rich C terminus of synaptojanin. However, a rabbit antiserum raised against synaptojanin R-2 reacted strongly with both synaptojanin constructs (1852 Western, Fig. 1).
To further characterize the baculovirus expressed synaptojanin, various dilutions of dynamin and synaptojanin, both purified from baculovirus infected Sf9 cells, were run on SDS-PAGE and overlaid (4) with a glutathione S-transferase fusion protein encoding the SH3 domain of amphiphysin. The synaptojanin construct strongly binds the SH3 domain of amphiphysin in this assay. Interestingly, when equal amounts of the two proteins are compared directly, amphiphysin demonstrates a greater relative affinity for synaptojanin than dynamin (Fig. 2).
Overlay of Brain Extracts with Synaptojanin-To identify the major binding partners for synaptojanin in brain, we used purified synaptojanin in overlay assays. Nitrocellulose transfers containing proteins from brain extracts were incubated with purified synaptojanin, and bound synaptojanin was detected with the antibody raised against full-length synaptojanin (Milo). Control overlay assays were performed with purified proteins from control infected Sf9 cells. Synaptojanin is seen to bind to two major (stars, Fig. 3) and two minor (arrowheads, Fig. 3) proteins in crude rat brain extracts. In addition, synaptojanin itself is detected (diamond, Fig. 3) and is the only protein seen in control overlays (Fig. 3).
Identification of a Major Synaptojanin-binding Protein as Amphiphysin-Based on previous results (6,14), we predicted that amphiphysin would be detected in the synaptojanin overlay assay. One of the major synaptojanin-binding proteins migrates at approximately 125 kDa, consistent with the molecular mass of amphiphysin (13,14). In fact, amphiphysin and the 125-kDa synaptojanin-binding protein have an identical mobility on SDS-PAGE (Fig. 4A). To confirm the identity of this protein, we performed an immunoprecipitation assay using two different amphiphysin antibodies (CD5 and CD6). As seen in Fig. 4B, and in agreement with previous data (14), both amphiphysin antibodies immunoprecipitate amphiphysin from a rat brain extract, although CD5 is much more effective than CD6. A synaptojanin overlay assay of the amphiphysin immunoprecipitates demonstrates that the 125-kDa synaptojanin-

FIG. 2. Amphiphysin overlay of synaptojanin and dynamin.
Purified synaptojanin and purified dynamin (500 -20 ng as indicated) were separated on SDS-PAGE, transferred to nitrocellulose, and overlaid with glutathione S-transferase/amphiphysin SH3 domain (amphiphysin overlay) as described (4). Transfers were stained with ponceau S to ensure even electrophoretic transfer. The arrows on the right indicate the migratory position of the proteins detected on the blots .   FIG. 3. Synaptojanin overlay of a rat brain extract. Rat brain post-nuclear supernatant fractions were separated on SDS-PAGE, transferred to nitrocellulose, and overlaid with synaptojanin (synaptojanin overlay) or with protein purified from mock infected Sf9 cells (control overlay). The symbols on the right denote the migratory positions of the two major (stars) and two minor (arrowheads) synaptojaninbinding proteins detected on the blot. The diamond denotes the migratory position of synaptojanin, which is also detected in this assay.

FIG. 1. Purification of baculovirus expressed synaptojanin constructs.
Sf9 cells were infected with baculovirus encoding fulllength synaptojanin (synaptojanin) or a synaptojanin deletion construct lacking the proline-rich C terminus (synaptojanin R-2), and the synaptojanin proteins were purified with nickel-agarose. Approximately 4 g of protein from each sample were separated on SDS-PAGE and stained with Coomassie Blue (Coomassie) or were transferred to nitrocellulose and blotted with a polyclonal antibody against synaptojanin purified from rat brain (5) (Milo Western) or with a polyclonal antibody raised against synaptojanin R-2 (1852 Western).
binding protein is amphiphysin.
Characterization of the 40-kDa Synaptojanin-binding Protein-A protein at approximately 40 kDa is the strongest synaptojanin-binding protein in brain (Figs. 3 and 4A). As determined by overlay, the 40-kDa protein is enriched in brain, although it is also detected in extracts from rat testis (Fig. 5A). Synaptojanin is also enriched in adult brain (Fig. 5A) although lower levels are seen in a wide variety of tissues (12). Synaptojanin, which is concentrated in presynaptic nerve terminals (5), is enriched in synaptic membrane fractions (Fig. 5B, LP 2 ). As determined by overlay, the 40-kDa synaptojanin-binding protein is enriched in soluble fractions and is concentrated in the LS 2 fraction that corresponds to cytosol isolated from lysed synaptosomes (Fig. 5B).
Co-immunoprecipitation of Synaptojanin and the 40-kDa Synaptojanin-binding Protein-We used a monoclonal antibody against synaptojanin to immunoprecipitate the protein from soluble fractions of rat brain. Synaptojanin is enriched in the precipitated material (Fig. 6). The 125-kDa synaptojaninbinding protein, which we identified as amphiphysin, does not co-immunoprecipitate with synaptojanin ( Fig. 6; see "Discussion"). However, the 40-kDa synaptojanin-binding protein does co-immunoprecipitate with synaptojanin and is enriched in the synaptojanin immunoprecipitate as compared with the starting material (Fig. 6). These data confirm the interaction between synaptojanin and the 40-kDa synaptojanin-binding protein in the brain.
Identification of the 40-kDa Synaptojanin-binding Protein-Rat brain cytosol was fractionated using various concentrations of ammonium sulfate, and the 40-kDa synaptojanin-binding protein was found exclusively in the 20 -40% ammonium sulfate precipitate (data not shown). This fraction was then subjected to size fractionation on 2.5-15% linear sucrose density gradients, and the 40-kDa protein was found in a narrow peak near the top of the gradient (data not shown). Peak A soluble fraction from rat brain was subjected to immunoprecipitation with a monoclonal antibody against p75 LNGFR (anti-p75) or a monoclonal antibody against synaptojanin (anti-synaptojanin). Precipitated proteins were separated on SDS-PAGE along with an aliquot of the soluble extract (starting material, SM), transferred to nitrocellulose and subjected to a synaptojanin Western blot (top panel) or a synaptojanin overlay (middle and bottom panels). The arrows on the right indicate the molecular masses of the proteins detected on the blots. sucrose density gradient fractions were pooled and subjected to anion exchange chromatography on DEAE-Sephacel. The column was eluted with a linear gradient of NaCl from 0 to 0.5 M. A Coomassie Blue-stained gel of the proteins eluted from the DEAE column is shown in Fig. 7A. A band at 40-kDa was apparent that was strongly reactive in the synaptojanin overlay assay (Fig. 7A, synaptojanin overlay).
To further characterize the 40-kDa protein, partially purified samples were overlaid with synaptojanin or synaptojanin R-2 deletion mutant (Fig. 7B) using the antiserum that recognizes the N-terminal domain of synaptojanin (1852 Western, Fig. 1). Synaptojanin, but not synaptojanin R-2, binds to the 40-kDa synaptojanin-binding protein. This demonstrates that the interaction of the 40-kDa protein with synaptojanin is mediated through synaptojanin's proline-rich C terminus and suggests that the 40-kDa protein contains an SH3 domain.
To identify the 40-kDa protein, fraction 14 from the DEAE column elution (Fig. 7A) was concentrated and transferred to PVDF membranes, and the 40-kDa protein band was subjected to peptide sequence analysis. The sample was refractive to automated Edman degradation, suggesting a blocked N terminus. Therefore, the sample was cleaved at methionyl residues with CNBr and resubjected to sequence analysis. The mixture resequencing revealed 2-3 major sets of sequencing signals. The strengths of the multiple signals were ranked, and the best guess sequence (Met 0 -Glu 1 -Val 2 -Phe 3 -Gln 4 -Asn 5 -Phe 6 -Ile 7 -Asp 8 -Pro 9 -Asp 10 -Gln 11 -Asn 12 -Gln 13 -His 14 -His 15 -Ala 16 -Asp 17 -Leu 18 -Arg 19 ) was searched against the protein data base and was found to align to internal sequences of mouse SH3P4, SH3P8, and SH3P13, three members of a novel family of SH3 domain-containing proteins (27) (Fig. 7C). SH3P4 is the likely homologue of the 40-kDa synaptojanin-binding protein because its predicted protein sequence contains the required methionyl residue as well as 15 of 20 identities found (Fig. 7C). Further, a second major peptide that was sequenced from the 40-kDa protein aligns with a peptide from SH3P4 that contains three of the five mismatches from the best guess sequence in the proper position in relationship to the methionyl residue (Fig. 7C). In contrast, SH3P8 and SH3P13 lack the methionyl residue at the start of the sequence, and only 12 and 9, respectively, of the above 20 residues are identical (Fig. 7C). Furthermore, the major signals in the mixture resequencing data are accounted for by CNBr cleavage at Met 96 , Met 121 , Met 133 , Met 201 , and Met 207 of mouse SH3P4 (data not shown) whereas the predicted sequences of the two SH3P4 homologues SH3P8 and SH3P13 (27) bottom panel). B, the partially purified 40-kDa synaptojanin-binding protein was overlaid with synaptojanin or with the synaptojanin construct lacking the proline-rich C terminus (synaptojanin R-2 overlay). C, the best guess sequence predicted from the major mixture sequencing data is shown in bold. The aligned sequences from SH3P4, SH3P8, and SH3P13 (27) are indicated and matches to the best guess sequence are in bold. The sequence of a second region of SH3P4, which was also identified in the sequencing mixture, is indicated, and amino acids that align with three of the five mismatches from the best guess sequence are in bold.
(Met 96 and Met 201 ) of the required methionyl residues, respectively. On this basis, the 40-kDa protein is the rat homologue of mouse SH3P4. DISCUSSION Synaptojanin is an inositol 5-phosphatase implicated in synaptic vesicle endocytosis (4 -6, 14). Synaptojanin was initially isolated based on its ability to bind to the SH3 domains of Grb2 (4), and a role for Grb2 in endocytosis was recently demonstrated (26). Synaptojanin also binds to the SH3 domain of amphiphysin (6,14), and several pieces of evidence implicate amphiphysin in endocytosis, including its SH3 domain-dependent interaction with dynamin (14). Thus, it appears that SH3 domain-mediated interactions play a general role in endocytosis. In an effort to better characterize the nature of SH3 domain-mediated protein-protein interactions with synaptojanin, we generated synaptojanin and a synaptojanin deletion construct in Sf9 cells using the baculovirus system. The proteins were then purified on nickel-agarose using a His 6 tag engineered into the N terminus of the constructs. To characterize the baculovirus expressed synaptojanin, we compared the affinity of amphiphysin binding to synaptojanin versus dynamin. When amphiphysin or Grb2 are used as substrates for the purification of SH3 domain-binding proteins from brain extracts, greater amounts of dynamin than synaptojanin are isolated (5,14), likely owing to higher levels of dynamin expression in brain. However, as shown here, when equal amounts of purified dynamin and synaptojanin are analyzed, amphiphysin shows stronger binding to synaptojanin than dynamin. It has been proposed (14) that amphiphysin may serve to target dynamin to sites of synaptic vesicle endocytosis via its dual interactions with AP2 (14,19) and dynamin. Amphiphysin may also play a role in targeting synaptojanin to endocytic sites. The higher affinity of synaptojanin than dynamin for amphiphysin binding may be important to allow for synaptojanin targeting in the presence of high dynamin concentrations in the nerve terminal.
We used synaptojanin purified from Sf9 cells in a gel overlay assay to identify two major synaptojanin-binding proteins with molecular masses of approximately 125 and 40 kDa. The 125-kDa synaptojanin-binding protein was identified as amphiphysin based on its co-migration with amphiphysin on SDS-PAGE and its precipitation with amphiphysin antibodies. The identification of amphiphysin as a major synaptojanin-binding protein strongly suggests that the assay is effective in identifying relevant synaptojanin-binding partners in vitro and further suggests that amphiphysin and the 40-kDa protein are the major synaptojanin-binding proteins in vivo.
Further characterization of the 40-kDa synaptojanin-binding protein demonstrates that it is highly concentrated in brain and is predominantly a soluble protein that is enriched in cytosol isolated from lysed synaptosomes. Proteins that function in clathrin-mediated endocytosis are often expressed at levels 10 -50-fold higher in neuronal versus non-neuronal cells (33). For example, both dynamin and synaptojanin are highly expressed in neurons, whereas these proteins or related isoforms are expressed at lower levels in non-neuronal cells (12, 34 -36). The 40-kDa synaptojanin-binding protein is concentrated in brain but is also detected in testis, a tissue with little or no expression of the 145-kDa isoform of synaptojanin (12). However, the testis does express the 170-kDa synaptojanin isoform (12), and this protein also binds strongly to the 40-kDa synaptojanin-binding protein (data not shown). An important role for the 40-kDa synaptojanin-binding protein is also supported by the observation that it co-immunoprecipitates with synaptojanin from rat brain cytosol. This is in contrast to amphiphysin, which does not co-immunoprecipitate with syn-aptojanin (Fig. 6). The reason for the lack of amphiphysin/ synaptojanin co-immunoprecipitation is unclear, but it may be due to a technical reason such as steric interference of the synaptojanin antibody with the site of amphiphysin binding. A more interesting explanation may be that the binding of synaptojanin to the 40-kDa synaptojanin-binding protein excludes amphiphysin binding. Thus, it is possible that the 40-kDa synaptojanin-binding protein could regulate the ability of synaptojanin to bind to amphiphysin, and this could play a key role in regulating the targeting of synaptojanin to sites of endocytosis.
To identify the 40-kDa protein, we purified it from rat brain cytosol and subjected it to peptide sequence analysis. The sequence analysis identifies the 40-kDa synaptojanin-binding protein as SH3P4, a novel SH3 domain-containing protein with a predicted molecular mass of 39,880 Da (27). The identification of the 40-kDa synaptojanin-binding protein as an SH3 domain-containing protein is consistent with our observation that the 40-kDa protein does not bind to a synaptojanin deletion construct lacking the proline-rich C terminus. Further, the predicted isoelectric point of 5.3 for SH3P4 (27) is consistent with its elution from the DEAE ion exchange column in high salt. SH3P4, which was identified from a mouse library screened with a Src SH3 ligand peptide, is 75 and 63% identical to SH3P8 and SH3P13, respectively, two other proteins identified in the same screen (27). These three proteins define a novel protein family of unknown function. Our data strongly implicate SH3P4, and perhaps other family members, in synaptic vesicle endocytosis. It will be of interest to determine if the interaction of SH3P4 can regulate the ability of synaptojanin to bind to amphiphysin and thus regulate the targeting of synaptojanin to sites of endocytosis.