Phosphorylation of a Synaptic Vesicle-associated Protein by an Inositol Hexakisphosphate-regulated Protein Kinase*

Despite the fact that inositol hexakisphosphate (InsP 6 ) is the most abundant inositol metabolite in cells, its cellular function has remained an enigma. In the present study, we present the first evidence of a protein kinase identified in rat cerebral cortex/hippocampus that is activated by InsP 6 . The substrate for the InsP 6 - regulated protein kinase was found to be the synaptic vesicle-associated protein, pacsin/syndapin I. This brain-specific protein, which is highly enriched at nerve terminals, is proposed to act as a molecular link coupling components of the synaptic vesicle endocytic machinery to the cytoskeleton. We show here that the association between pacsin/syndapin I and dynamin I can be increased by InsP 6 -dependent phosphorylation of pacsin/syndapin I. These data provide a model by which InsP 6 -dependent phosphorylation regulates synaptic vesicle recycling by increasing the interaction between endocytic proteins at desorption and Purification of the GST-Pacsin/Syndapin I Bacterial Fusion Protein— Production of murine GST-pacsin/syndapin has been previously described (14). To the GST-pacsin/syndapin I SH3 domain deletion mutant, site-directed mutagenesis (Transformer Site-directed Mutagenesis Kit, CLONTECH) was used to introduce two Bam HI sites, one upstream of the startcodon and the other at position 1153 in the murine pacsin/syndapin I coding sequence. The mutation primers used were 5 9 -CGCTACAGGATCCCCATGTCTGGC-3

A large number of G-protein-coupled and growth factor receptors activate phospholipase C to stimulate the hydrolysis of the phospholipid, phosphatidylinositol 4,5-bisphosphate, producing inositol 1,4,5-trisphosphate (InsP 3 ) 1 and diacylglycerol. InsP 3 has an established second messenger role in mobilizing intracellular calcium stores (1) and can be phosphorylated to inositol 1,3,4,5-tetrakisphosphate (InsP 4 ), which is proposed to play a role in capacitative calcium entry (2) and as an activator of ras-GAP activity (3). The sequential dephosphorylation of InsP 3 and InsP 4 gives rise to a host of inositol phosphate isomers in a cyclic metabolic pathway that results in inositol returning to the phosphoinositide pool (4).
In contrast to this well understood signaling pathway the role of inositol hexakisphosphate (InsP 6 ), the most abundant cellular inositol metabolite, is very poorly understood. Present at concentrations 10 -100 times greater than InsP 3 (15-60 M) (5,6), InsP 6 has long been thought of as metabolically rather inert with levels that do not change in response to cell surface receptor activation. However, the discovery of higher phosphorylated forms of InsP 6 has led to the realization that the InsP 6 pool is rapidly turning over in a continual cycle of phosphorylation and dephosphorylation (7,8). Furthermore, a role for InsP 6 in synaptic vesicle trafficking has emerged from studies showing that InsP 6 binds with high affinity to a number of proteins that are involved in exo/endocytosis including the clathrin assembly proteins AP2 and AP3 and the synaptic vesicle calcium-sensing protein synaptotagmin (9 -12).
In the present study, we have investigated the possibility of the existence of a protein kinase that is regulated by InsP 6 . We found that the pacsin/syndapin I, a synaptic vesicle-associated protein that acts as a molecular link coupling the endocytic machinery to the cytoskeleton, is phosphorylated by a protein kinase that is regulated by InsP 6 . In addition, InsP 6 -regulated phosphorylation of pacsin/syndapin I increases the interaction between pacsin/syndapin I and dynamin I. These data provide a novel model by which InsP 6 can regulate synaptic vesicle endocytosis.

EXPERIMENTAL PROCEDURES
Anion Exchange Fractionation of Soluble Rat Brain Cortex/Hippocampus Extract-Rat brain cortex/hippocampus was homogenized in 15 ml of TE buffer (2.5 mM Tris-HCl, 2.5 mM EDTA, pH 7.4) containing 1 mM phenylmethylsulfonyl fluoride. The homogenate was centrifuged at 50,000 ϫ g for 10 min, and the pellet was discarded. The supernatant was further centrifuged at 50,000 ϫ g for 20 min. The supernatant (13 ml at ϳ2.5 mg of protein/ml) was applied to a 1-ml Resource Q (Amersham Pharmacia Biotech) anion-exchange column (6.4 ϫ 30 mm). The column was washed with 5 ml of TE buffer and eluted with a linear gradient of 0 -1 M NaCl in TE buffer over 20 ml (flow rate ϭ 1 ml/min). One ml fractions were collected and assayed for kinase activity. The entire procedure was conducted at 4°C.
Assay for Inositol Polyphosphate Kinase Activity in Brain Fractions-Aliquots of fractions (5 l, ϳ5 g of protein) obtained from the anion exchange brain fractionation were added to kinase buffer (20 mM Tris-HCl, 10 mM MgCl 2 , 1 mM EGTA, 0.5 mM dithiothreitol, 0.2 mM orthovanadate, 25 mM glycerol phosphate, pH 7.5) containing 10 M [ 32 P]ATP (1-4 cpm/fmol ATP), Ϯ InsP 6 (50 M), in a final volume of 25 l. The reactions were continued for 10 min at 37°C and were terminated by the addition of 10 l (2ϫ) SDS-PAGE sample buffer, and the proteins were resolved by 10% SDS-PAGE.
Large scale reactions that were to be fractionated on a hydroxyapatite column were conducted using 100 l of anion-exchange fraction in kinase buffer containing 10 M [ 32 P]ATP (1-3.0 cpm/fmol ATP), Ϯ inositol polyphosphates (50 M) in a final volume of 500 l. The reactions were continued for 10 min at 37°C and terminated by the addition of 500 l of buffer A (5 mM Tris-HCl, 10 mM KHPO 4 , pH 7.4).
Fractionation of Kinase Reactions on Hydroxyapatite-One ml of the above large scale kinase reaction was applied to a hydroxyapatite column with a bed volume of 1 ml (Bio-Rad). The column was washed with 5 ml of buffer A and eluted with a linear gradient of 0 -100% buffer * This work was supported by Grant 059333 from the Wellcome Trust, Grant GM31278 from the National Institutes of Health, and the Köln Fortune Program. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Gels were routinely silver stained (Bio-Rad) before autoradiography to ensure that the chromatography of each reaction was the same and that equal amounts of protein had been loaded.
Immunoprecipitation of Pacsin/Syndapin I-Immunoprecipitations were carried out essentially as reported previously (13). Large scale kinase reactions using fraction 9 (100 l) from the anion-exchange fractionation of rat brain cytosolic extract were carried out as described above. The reactions were stopped by the addition of ice-cold TE buffer (500 l). Pacsin/syndapin I specific antiserum (14) was then added (ϳ5 g) and incubated for 60 min. The immune complexes were isolated on protein A-Sepharose beads. The beads were washed 3ϫ in TE buffer and resuspended in 20 l of SDS-PAGE sample buffer. Proteins were then resolved by 10% SDS-PAGE, and an autoradiograph was obtained.
Construction and Purification of the GST-Pacsin/Syndapin I Bacterial Fusion Protein-Production of murine GST-pacsin/syndapin I has been previously described (14). To produce the GST-pacsin/syndapin I SH3 domain deletion mutant, site-directed mutagenesis (Transformer Site-directed Mutagenesis Kit, CLONTECH) was used to introduce two BamHI sites, one upstream of the startcodon and the other at position 1153 in the murine pacsin/syndapin I coding sequence. The mutation primers used were 5Ј-CGCTACAGGATCCCCATGTCTGGC-3Ј and 5Ј-GATGCCAAGGGGATCCGTGTACGGGC-3Ј, respectively. The resulting pacsin/syndapin I was subcloned into the BamHI site of the pGEX-3X vector (Amersham Pharmacia Biotech) resulting in the production of GST-pacsin/syndapin I lacking the majority of the SH3 domain (Arg 386 -Ile 441 was deleted).
Phosphorylation of the GST-Pacsin/Syndapin I Fusion Protein-Aliquots of each fraction from the anion exchange fractionation (5 l containing ϳ5 g of protein) was added to kinase buffer containing ϳ5 g of GST-pacsin/syndapin I, 10 M [ 32 P]ATP (1-4 cpm/fmol ATP), Ϯ inositol polyphosphates in a final volume of 100 l. The reactions were continued for 10 min at 37°C and terminated by the addition of ice-cold TE buffer (100 l). GSH-Sepharose slurry (giving 20 l of packed beads) was added. Sepharose beads were pelleted in a microcentrifuge and washed in TE-buffer three times. SDS-PAGE sample buffer (20 l) was added to the beads, and the proteins were resolved by 10% SDS-PAGE.
Pacsin/Syndapin I Association with Dynamin I-Kinase reactions were performed as described above except with 2 mM ATP. The GSTpacsin/syndapin (ϳ5 g) bound to glutathione-Sepharose beads was then washed in ice-cold TE buffer and incubated with rat cytosol (1 mg protein) in a final volume of 1 ml for 60 min at 4°C. The Sepharose beads were then washed three times in ice-cold TE buffer, and the associated proteins were resolved by 10% SDS-PAGE. The gels were transferred to nitrocellulose, and the membranes were probed with anti-dynamin I polyclonal antibody (Santa Cruz Biotechnology). To determine whether there was equal transfer and loading of proteins, membranes were stained with Ponceau before being probed with dynamin I antibodies. Quantification of protein bands in the immunoblots was determined using NIH Image.
Mass Spectrometry-Samples were separated on 10 -20% tricine/ SDS-polyacrylamide gel prior to staining with GELCODE colloidal Coomassie stain (Pierce) or a mass spectrometry-compatible silver stain. Excised bands were destained and carbamidomethylated and digested with trypsin prior to analysis of a portion of the digested supernatant by MALDI-TOF mass spectrometry on a TOFSPEC-S.E. instrument (Micromass). Nanoelectrospray mass spectrometry was performed on a API-III triple quadruple instrument (PE-Sciex).

Identification of the Synaptic Vesicle-associated Protein, Pacsin/Syndapin I, as a Substrate for an InsP 6 -regulated Protein
Kinase-The strategy we adopted to identify the presence of an InsP 6 -regulated protein kinase was based on the fact that many signaling molecules, including protein kinases and their substrates, form complexes with each other that are robust enough to be purified using techniques such as co-immunoprecipitation (15,16). It was, therefore, considered possible that the protein substrates for a putative inositol polyphosphate- The reaction was stopped and fractionated on a hydroxyapatite column. Each fraction from the hydroxyapatite column was concentrated and resolved by 10% SDS-PAGE. The gel was first silverstained to establish equal protein loading and to identify the proteins present. An autoradiograph of the gel was then obtained to identify the position of the phosphoproteins. The position of a 52-kDa protein that increases in its phosphorylation state in response to InsP 6 is marked with an arrow. Shown are fractions obtained from the elution gradient of the column (fraction 14 ϭ 90% buffer B). The results shown are of an autoradiograph and silver stain of the same gel and are typical of at least three experiments. The positions of molecular mass markers are shown in kDa. regulated protein kinase may form a complex with the kinase that will allow them to co-purify over at least one chromatographic step.
A low speed supernatant extract from rat brain cortex/hippocampus was fractionated over an anion-exchange column. The presence of an InsP 6 -regulated protein kinase was assessed by incubating an aliquot of each fraction with [P 32 ]ATP, in the absence or presence of InsP 6 (50 M) and resolving the phosphoproteins by SDS-PAGE. The level of phosphorylation of the majority of phosphoproteins contained in these kinase reactions were not affected by the presence of InsP 6 (Fig. 1). However, in fractions 9 and 10, the phosphorylation state of three proteins (Fig. 1, labeled 1-3) appeared to be altered by InsP 6 . Interestingly, band 3 appears to have decreased in its level of phosphorylation, whereas the other two protein bands have increased levels of phosphorylation in the presence of InsP 6 . We decided to center our attention on the identity of the ϳ52-kDa protein band labeled 1 in Fig. 1.
To determine the identity of the 52-kDa phosphoprotein, large scale kinase reactions containing vehicle, InsP 3 , or InsP 6 , were performed. The kinase reactions were then fractionated on a hydroxyapatite column, and the phosphoproteins were resolved by 10% SDS-PAGE. Before the gels were exposed to film, they were silver stained to establish the fidelity of the chromatography. The silver stain demonstrates that the chromatography of each of the three kinase reactions on the hydroxyapatite matrix and the amount of protein loaded on each lane were identical (Fig. 2). The autoradiograph of the silverstained gels revealed that the phosphorylation state of a ϳ52-kDa protein band eluting in fractions 14 and 15 was increased in the presence of InsP 6 but not InsP 3 (Fig. 2). By aligning the silver-stained gel with the autoradiograph, it was possible to determine that the 52-kDa phosphoprotein in fraction 14 lined up exactly with a 52-kDa protein on the silver-stained gel (Fig. 2).
MALDI-TOF analysis of tryptic peptide fragments of the 52-kDa protein band excised from the SDS-PAGE gel showed 28 peptides covering 63% of the amino acid sequence that matched exactly with the synaptic vesicle-associated protein, pacsin/syndapin I (Fig. 3). The amino acid sequence of five selected peptides were determined using MS/MS and shown to match exactly with the sequence of rat pacsin/syndapin I (Fig. 3).
The identity of pacsin/syndapin I as the substrate for the putative InsP 6 -regulated protein kinase was confirmed in immunoprecipation studies using a pacsin/syndapin I-specific antiserum (14). In these experiments, the level of phosphorylation of pacsin/syndapin I present in kinase reactions was shown to be increased in the presence of InsP 6 (Fig. 4). We also tested the ability of a bacterial GST fusion protein containing the full FIG. 3. Amino acid sequence analysis of the 52-kDa protein identified to be phosphorylated in an IP 6 -dependent manner. A, mass spectra analysis (MALDI-TOF) of tryptic peptide fragments of the 52-kDa protein shown to be phosphorylated in an InsP 6 -dependent manner. 28 peptides (marked with a diamond) matched exactly with that of pacsin/syndapin I, covering 63% of the amino acid sequence. B, rat pacsin/syndapin I showing the sequence of five tryptic peptides (in bold) as determined by MS/MS. pacsin/syndapin I sequence to act as a substrate for the InsP 6regulated protein kinase. GST-pacsin/syndapin I was phosphorylated in an InsP 6 -sensitive manner by fractions 8 to 11 from the anion exchange chromatography of rat brain cytosolic extract (Fig. 5A). There was no phosphorylation of the control GST protein (Fig. 5B).
Phosphoamino acid analysis revealed that the GST-pacsin/ syndapin I fusion protein was phosphorylated in an InsP 6sensitive manner on serine (data not shown). Furthermore, a mutant of the GST-pacsin/syndapin I, where the C-terminal SH3 domain was deleted, also appeared to be a substrate for the InsP 6 -regulated protein kinase (Fig. 6), indicating that the site of phosphorylation on pacsin/syndapin I was not within the SH3 domain, which has previously been shown to be involved with coupling pacsin/syndapin I to the synaptic vesicle endocytic machinery (17).

Specificity of Inositol Phosphate Regulation of Pacsin/Syndapin I Phosphorylation-
The phosphorylation of GST-pacsin/ syndapin I was not stimulated by any of the lower inositol polyphosphates indicating that a phosphate group on each of the 6 positions on the inositol ring was essential for stimulating protein kinase activity (Fig. 7A). Importantly, inositol hexasulfate was also unable to activate protein kinase activity, eliminating the possibility that InsP 6 may be exerting its activity via nonspecific effects such as low affinity charge interactions or chelation of divalent or trivalent cations. The polyphosphoinositides, phosphatidylinositol (4,5)-bisphosphate and phosphatidylinositol (3,4,5)-trisphosphate, were found to be unable to stimulate the phosphorylation of GST-pacsin/syndapin I (Fig. 7B). Concentration-response analysis demonstrated that InsP 6 was able to stimulate protein kinase activity in a dosedependent manner with maximal effects at 25-50 M (Fig. 7C). Cellular InsP 6 can be phosphorylated to produce the inositol pyrophosphates termed diphosphoinositol pentakisphosphate (PP-InsP 5 , also known as InsP 7 ) and bis(diphospho)inositol tetrakisphosphate (bis-PP-InsP 4 or InsP 8 ) (7,8). The predominant PP-InsP 5 isomer found in mammalian cells is 5-PP-InsP 5 , where the pyrophosphate is in the 5-position on the inositol ring (18). We used this isomer and an isomer that does not naturally occur (2-PP-InsP 5 ) to determine the ability of inositol pyrophosphates to stimulate phosphorylation of pacsin/syndapin I. We found that both isomers of PP-InsP 5 stimulated phosphorylation of GST-pacsin/syndapin I (Fig. 8A). Analysis of the concentration-response curve demonstrated that the potency of 5-PP-InsP 5 was similar to that of InsP 6 (Fig. 8B).
Phosphorylation of Pacsin/Syndapin I Regulates Its Interaction with Dynamin I-Pacsin/syndapin I is proposed to regulate synaptic vesicle recycling by acting as a molecular link between the cytoskeleton and the endocytosis machinery (14,17). In vitro binding studies have demonstrated that one of the binding partners for pacsin/syndapin I is dynamin I (17,19). Consistent with these previous studies, we show here that GST-pacsin/syndapin I interacts with dynamin I present in a cytosolic rat brain preparation, whereas GST alone was unable to interact (Fig. 9A). Also consistent with earlier studies we found that this interaction appears to be via the SH3 domain because the SH3 domain deletion mutant of pacsin/syndapin I was unable to associate with dynamin I (data not shown).
The possibility that phosphorylation of pacsin/syndapin I contributes to its ability to interact with proteins involved in synaptic vesicle endocytosis was tested by analysis of the association of the phosphorylated form of pacsin/syndapin I and dynamin I. InsP 6 -dependent phosphorylation of GST-pacsin/ syndapin I increased the association between pacsin/syndapin I and dynamin I by 3.28 Ϯ 1.09-fold (n ϭ 3 Ϯ S.E.; Fig. 9B).

DISCUSSION
In the present study, we have shown that the synaptic vesicle-associated protein, pacsin/syndapin I, is phosphorylated by a novel protein kinase that is regulated by InsP 6 . Furthermore, InsP 6 -dependent phosphorylation of pacsin/syndapin I increases the ability of pacsin/syndapin I to interact with dynamin I.
The strategy we adopted for the discovery of the InsP 6regulated protein kinase and its protein substrate was based on the hypothesis that many signaling proteins form complexes with each other that are robust enough to allow them to be co-purified. This phenomenon has been described for the platelet-derived growth factor receptor, which forms a so-called "signal transduction particle" with PLC␥, phosphatidylinositol 3-kinase, GAP, and raf (15). Other signal tranduction protein kinases, such as those of the MAP kinase pathway, form signaling complexes that require the presence of scaffolding proteins (16). In the present study we analyzed chromatographic fractions from rat brain for the presence of co-purifying InsP 6regulated protein kinase and its substrate proteins. We found that a 52-kDa protein eluting from an anion exchange column was phosphorylated in response to InsP 6 . This protein was determined to be the synaptic vesicle protein, pacsin/syndapin I, on the basis of MALDI-TOF and amino acid sequence analysis and immunoprecipitation experiments. The identification of the phosphorylation of pacsin/syndapin I is the first demonstration of a protein phosphorylation that is regulated by InsP 6 . We are currently using the InsP 6 -dependent phosphorylation of the GST-pacsin/syndapin I fusion protein in an assay to purify the InsP 6 -regulated protein kinase.
Pacsin/syndapin I is a 52-kDa protein that is highly enriched in brain and associated with presynaptic terminals where it co-localizes with dynamin I (14,17). In vitro binding studies have demonstrated that the SH3 domain of pacsin/syndapin I is able to bind to the proline-rich C terminus of dynamin I, as well as a number of other brain-specific proteins including synaptojanin, synapsin Ia/Ib and the neuronal Wiskott-Aldrich syndrome protein (17,19). These proteins are known to form part of the multicomponent protein complex involved in clathrin-mediated synaptic vesicle recycling at nerve terminals (20).
The assembly of this multiprotein complex has been demonstrated to be regulated by a phosphorylation/dephosphorylation cycle that is controlled by calcium entry at the nerve terminal (21,22). Interestingly, these previous studies have demonstrated that phosphorylation of synaptic vesicle proteins such as dynamin I, synaptojanin and amphiphysin results in a decreased ability of these proteins to associate with other synaptic vesicle endocytic proteins. Assembly of the endocytic machinery occurs on calcium entry via dephosphorylation mediated by the calcium-sensitive phosphatase calcineurin (21,22). Our data indicate the presence of an additional novel mechanism in the regulation of synaptic vesicle endocytsis where InsP 6 -mediated phosphorylation of pacsin/syndapin I is able to increase the ability of pacsin/syndapin I to interact with dynamin I. This process may either contribute to the assembly of the endocytic machinery or tether the endocytic multiprotein complex to the cytoskeleton.
Our study, therefore, adds to a growing body of evidence to suggest that InsP 6 is involved in synaptic vesicle trafficking and membrane trafficking. The clathrin assembly proteins AP2 and AP3, together with the visual and non-visual arrestins and synaptotagmin, have been shown to bind InsP 6 with high affinity (9 -12, 23). The binding of InsP 6 to these proteins regulates their interaction with the endo/exocytic pathways. The present study extends a role for InsP 6 in endocytosis by demonstrating that InsP 6 -dependent phosphorylation is a mechanism that can increase the interaction between specific endocytic proteins.
The discovery of an InsP 6 -regulated protein kinase supports a signaling role for this inositol polyphosphate. Importantly, none of the lower inositol phosphates stimulated kinase activity. Furthermore, kinase activity was not stimulated by the polyphosphoinositides. This is of significance because a number of high affinity InsP 6 -binding proteins have also been shown to bind the polyphosphoinositides, and it is in fact the binding of polyphosphoinositides that is thought in some instances to be the physiological regulators of these proteins (11,24). We also tested the ability of the inositol pyrophosphates to stimulate kinase activity. Both isomers of PP-InsP 5 stimulated kinase activity with a concentration-response curve that was very similar to InsP 6 . Thus, the InsP 6 -regulated protein kinase is activated by inositol that contains a phosphate group on all 6 positions of the inositol ring and that addition of a further phosphate to make a pyrophosphate at positions 5 or 2 does not change the activation profile. It is, therefore, possible that both InsP 6 and/or 5-PP-InsP 5 could be the physiological activators of the protein kinase identified here. However, it is interesting to note that the concentration-response curve for InsP 6 is consistent with a protein kinase that would be expected to respond to cellular concentrations of InsP 6 , which are thought to be in the M range (5,6).
Early work on InsP 6 demonstrated that it is ubiquitously present at high cellular levels, and it was thought to be involved in metal ion chelation or act as a phosphate store (25,26). However, studies demonstrating that InsP 6 levels can change in response to receptor stimulation suggested that InsP 6 may have a signaling role (27,28). Recently, InsP 6 has been shown to be involved in the regulation of mRNA export from the nucleus (29), the control of phosphatase activity in pancreatic ␤ cells (28), and to act as a key factor in DNA repair (30). Our studies demonstrating the presence of a protein kinase that is specifically activated by InsP 6 adds a novel signaling pathway for this inositol polyphosphate.
FIG. 9. InsP 6 -dependent phosphorylation of pacsin/syndapin I regulates its interaction with dynamin I. A, GST or GST-pacsin/ syndapin I associated with glutathione-Sepharose beads was incubated with rat brain cytosol. The beads were then washed and associated proteins resolved by 10% SDS-PAGE, and the gels were transferred to nitrocellulose and immunoblotted with anti-dynamin I antibodies. B, GST-pacsin/syndapin I was first incubated in the presence or absence of InsP 6 (50 M) in kinase reactions using fraction 9 from the anion exchange fractionation of rat brain cytosol. The GST-pacsin/syndapin I bound to glutathione beads was then washed and incubated with rat brain cytosol (pacsin/syndapin ϩ brain) or with vehicle (pacsin/syndapin Ϫ brain) for 60 min (4°C). The presence of associated dynamin I was determined by immunoblotting with anti-dynamin I antibodies. Standard, 20 g of rat brain cytosol. The results shown are typical of at least four experiments. In the example shown the -fold increase in dynamin to pacsin/syndapin I interaction following InsP 6 -dependent phosphorylation of pacsin/syndapin I was determined to be 2.43-fold. The positions of molecular mass markers are shown in kDa.