Synapsin III, a Novel Synapsin with an Unusual Regulation by Ca2+ *

Synapsins I and II are synaptic vesicle proteins essential for normal Ca2+ regulation of neurotransmitter release. Synapsins are composed of combinations of common and variable sequences, with the central C-domain as the largest conserved domain. The C-domain is structurally homologous to ATPases, suggesting that synapsins function as ATP-dependent phosphotransfer enzymes. We have now identified an unanticipated third synapsin gene that is also expressed in brain. The product of this gene, synapsin IIIa, shares with synapsins Ia and IIa three conserved domains that are connected by variable sequences: the phosphorylated A-domain at the amino terminus, the large ATP-binding C-domain in the center, and the E-domain at the carboxyl terminus. Like other synapsins, synapsin IIIa binds ATP with high affinity and ADP with a lower affinity, consistent with a cycle of ATP binding and hydrolysis. ATP binding to the different synapsins is directly regulated by Ca2+ in a dramatically different fashion: Ca2+activates ATP binding to synapsin I, has no effect on synapsin II, and inhibits synapsin III. Thus vertebrates express three distinct synapsins that utilize ATP but are specialized for different modes of direct Ca2+ regulation in synaptic function.

Synapsins are abundant phosphoproteins of synaptic vesicles (1). Two genes for synapsins were described, the synapsin I gene, which produces alternatively spliced transcripts encoding synapsins Ia and Ib, and the synapsin II gene encoding synapsins IIa and IIb (2). Sequence comparisons revealed that synapsins are composed of a mosaic of domains. All synapsins share a short amino-terminal domain (A-domain) that is phosphorylated by CaM kinase I and protein kinase A, a linker sequence (B-domain) that is rich in short-chain amino acids (proline/alanine/glycine/serine), and a large central domain (Cdomain) that comprises approximately one-half of the total synapsin sequences. After the C-domain, different combinations of domains are observed (the D-, E-, F-, and G-domains). Interestingly, at the very carboxyl terminus synapsins Ia and IIa contain an additional short common domain of 50 residues (the E-domain) despite distinct sequences between the C-and E-domains (2). The E-domain is absent from synapsins Ib and IIb. A synapsin homologue was described in Drosophila in which the C-and E-domains are the only conserved domains, suggesting that these domains are responsible for the function of synapsins (3,4).
Synapsins have been studied extensively both biochemically and genetically (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15). Although a large number of synapsin functions have been proposed, their precise biological activity has proved elusive. Based on the analysis of mouse knockouts, currently the best defined role of synapsins exists in the regulation of synaptic vesicle function in mature nerve terminals (16 -18). Synapsins I and II are phosphorylated at the amino terminus during stimulation, but the physiological consequences of this phosphorylation are unknown. Synapsins interact with a number of proteins with high affinity and stoichiometry (e.g. actin filaments, neurofilaments, microtubules, calmodulin, spectrin, annexin VI; Refs. 5-15); however, it seems unlikely that synapsins will bind to all of these proteins in vivo, and the biological significance of these interactions is unclear.
Recently the crystal structure of the C-domain of synapsin I was solved, giving unexpected insights into its function (19). The structure showed that the C-domain is an independently folding domain that forms a stable dimer. Data bank searches revealed that the C-domain is structurally closely related to five ATP-utilizing enzymes: glutathione synthetase, D-alanine: D-alanine ligase, biotin carboxylase ␣-chain, succinyl-CoA synthetase ␤-chain, and pyruvate,orthophosphate dikinase. More than 80% of the C␣ carbon atoms of the synapsin I C-domain can be superimposed on those of glutathione synthetase or D-alanine:D-alanine ligase with a root mean square deviation of 0.32 nm, suggesting a close structural and evolutionary similarity between these enzymes and synapsins (19). The enzymes to which synapsin I is structurally related bind ATP and transfer phosphate from bound ATP to a substrate (20,21). Synapsins also bind ATP with high affinity and ADP with a lower affinity, suggesting that synapsins also represent phosphotransfer enzymes (19,22). Surprisingly, ATP binding to the C-domains of synapsins I and II was found to be differentially regulated: ATP binding to synapsin I required Ca 2ϩ which was not necessary for ATP binding to synapsin II.
We now report that a third unanticipated synapsin gene is expressed in vertebrates. Synapsin IIIa is closely homologous to synapsins Ia and IIa, with the highest similarity observed in the A-, C-, and E-domains. As in the other synapsins, the C-domain of synapsin IIIa binds ATP with high affinity. Different from synapsins I and II, however, Ca 2ϩ inhibits ATP binding to synapsin IIIa at micromolar concentrations. As a consequence, Ca 2ϩ has distinct regulatory effects on the three synapsins: it activates ATP binding in synapsin I, the most abundant synapsin; it is without effect with synapsin II; and it inhibits ATP binding to synapsin IIIa, the least abundant synapsin. Our data suggest an unexpectedly direct and diverse regulatory role of Ca 2ϩ in a family of synaptic vesicle phosphoproteins.

EXPERIMENTAL PROCEDURES
cDNA Cloning of Rat Synapsin III-Data bank searches with the rat synapsin I and II sequences (2) revealed that sequences from cosmids * This study was supported by a postdoctoral fellowship grant from the Human Frontiers Science Program (to M. H.). 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF056704.
‡ To whom correspondence should be addressed. Tel.: 214-648-5022; Fax: 214-648-6426. N104C7, N28H9, E86D10, and N80H12 from human chromosome 22q11.2-22q12-qter (accession numbers Z71183, Z80902, Z82181, and Z82246) contained multiple exons of a gene highly homologous to synapsins I and II. A 0.24-kilobase fragment of the corresponding cDNA was cloned by PCR 1 from human first-strand brain cDNA (obtained from CLONTECH) and used as a probe for screening rat brain cDNA libraries (2,23). Multiple overlapping clones were isolated and sequenced. Data bank analyses were performed using DNA-Star and BLAST softwares. The synapsin III sequence was submitted to Gen-Bank TM (accession number AF056704).
Northern Blotting-RNA blotting experiments were performed using multiple rat tissue blots obtained from CLONTECH and a rat synapsin III probe corresponding to amino acid residues 440 -526. (24) with C-domain sequences were obtained by PCR with oligonucleotide primers containing flanking restriction sites essentially as described (22). The following pGEX constructs were used in the current study: pGEXrSynI-C (residues 110 -420 of rat synapsin I), pGEXrSynII-C (residues 113-421 of rat synapsin II), and pGEXrSynIII-C (residues 89 -399 of rat synapsin III), together with control plasmids described previously (22). All recombinant proteins were purified using standard techniques, analyzed by SDS-polyacrylamide gel electrophoresis and Coomassie staining, and quantified using known amounts of bovine serum albumin examined on the same gel.

Construction of Expression Vectors and Expression of Recombinant Proteins-Synapsin expression vectors in pGEX-KG
ATP␥ 35 S Binding Measurement-Purified GST-fusion proteins immobilized on glutathione-agarose beads were washed 3ϫ with buffer A (50 mM Hepes-NaOH, pH 7.4, 25 mM NaCl) containing either 2 mM EGTA Ϯ 2.1 mM calcium or 2 mM EDTA. Aliquots of the beads (5 pmol of recombinant protein) were used in 0.1-ml binding assays containing buffer A with 10 nM ATP␥ 35 S and the indicated addition of nucleotides, Ca 2ϩ , Mg 2ϩ , and/or chelator. After a 1-h incubation at room temperature, beads were washed 3ϫ in the incubation buffer without ATP␥ 35 S, and the radioactivity bound to the beads was determined by scintillation counting. For determination of the Ca 2ϩ concentration dependence of ATP binding, Ca 2ϩ /EGTA buffers were used. Binding assays were performed essentially as described above except that all steps were performed in 75 mM Hepes-NaOH, pH 7.0, 25 mM NaCl, and 4 mM EGTA. The concentration of free Ca 2ϩ was calculated using the Chelator program.

RESULTS
Identification of Synapsin IIIa-Data bank searches with synapsin sequences uncovered exonic sequences from a novel, unidentified synapsin gene on chromosome 22. These sequences were highly homologous to those of synapsins I and II and exhibited the same exon-intron structure as the synapsin I gene (23). Nevertheless, two reasons led us to conclude that the chromosome 22 gene encodes a novel gene, named here synapsin III. Firstly, its predicted sequences differ at several positions from the human synapsin I and II sequences, and secondly the human genes for synapsins I and II map to the X-chromosome and to chromosome 3, respectively, whereas the new gene was localized on chromosome 22.
The synapsin III exons were distributed over four distinct cosmids (N104C7, N80H12, E86D10, and N28H9), suggesting that the synapsin III gene is very large. We were unable to assemble a complete sequence for synapsin III from the cosmids and found no ESTs corresponding to synapsin III in the EST data banks. Therefore we were uncertain whether the synapsin III gene corresponds to an expressed gene or a pseudogene. To address this, we searched for a processed synapsin III transcript by PCR using human brain cDNA. We obtained a single major PCR product with a sequence predicted from the synapsin III gene, demonstrating that synapsin III is indeed an expressed gene (data not shown). We then used the PCR product to isolate rat cDNA clones encoding synapsin III from a brain cDNA library and isolated multiple overlapping cDNA clones that allowed us to assemble a complete synapsin III sequence that is similar to the "a" variants of synapsins I and II (Fig. 1).
Domain Structure of Synapsins IIIa-Alignment of the synapsin Ia, IIa, and IIIa sequences reveals a similar domain organization in all three synapsins. The sequence of the aminoterminal A-domain is nearly identical between synapsins and includes a phosphorylation site for CaM kinase I/cAMP-dependent protein kinase in all synapsins (arrow in Fig. 1). The B-domain that follows the A-domain is a less well conserved domain rich in prolines, glutamines, serines, and alanines. The central C-domain is the largest domain in all synapsins (308 residues) and highly conserved (62% invariant residues between all three synapsins). Following the C-domain, all synapsins again have a poorly conserved domain that is rich in prolines, glutamines, and alanines. This domain contains phosphorylation sites for multiple protein kinases in synapsin I. The phosphorylation sites are not conserved in the other synapsins, although it is possible that synapsin IIIa will also be a substrate for CaM kinase II (see arrows in Fig. 1). At the carboxyl terminus, synapsins Ia, IIa, and IIIa again have a highly homologous domain, the E-domain, that is absent from synapsins Ib and IIb. The E-domain consists of two blocks of conserved sequences separated by a short proline-rich loop. Overall, the sequence alignment reveals a common architecture for synapsins Ia, IIa, and IIIa. A large central conserved domain (the C-domain) is flanked by small amino-and carboxyl-terminal domains that are also very homologous. These conserved domains are connected to each other by linker sequences rich in prolines, glutamines, alanines, serines, and glycines, short side-chain amino acids characteristic of non-structured protein regions. Together with the crystallographic data characterizing the C-domain as an independently folding dimeric module, these data suggest that the central C-domain of synapsins is linked to the two other domains by flexible loops.
Tissue Distribution of Synapsin IIIa Expression-To test which tissues express synapsin IIIa, an RNA blot was probed with a specific probe. A single band corresponding to the right size was detected in brain, indicating that synapsin IIIa is expressed primarily in brain similar to other synapsins (Fig. 2).
ATP Binding by Synapsin III-The crystal structure of the C-domain of synapsin I and biochemical studies with the Cdomains of synapsins I and II revealed that the C-domain constitutes an ATP-binding module related to several ATPases with functions as phosphotransfer enzymes (19,22). The Cdomain of synapsin III is highly homologous to that of the other synapsins, and most of the residues involved in ATP binding are conserved. This suggests that synapsin III may also be an ATP-utilizing protein. To test this, we measured ATP␥ 35 S binding to the recombinant C-domain of synapsin III. ATP␥ 35 S bound specifically and was displaced by ATP and, to a lesser degree, GTP (data not shown). To test the relative affinity of ATP and ADP, we performed displacement titrations of bound ATP␥ 35 S with cold ATP or ADP. The half-maximal concentrations needed for competition were approximately 0.5 M ATP and 12 M ADP (Fig. 3), suggesting that as in other synapsins, ATP is bound much more tightly than ADP. The half-maximal displacement concentrations indicate that synapsin III is a high-affinity ATP-binding protein, although its ATP affinity is slightly lower than that of the other synapsins. The difference in affinity for ATP and ADP are consistent with a catalytic role of the C-domain in which ADP would be dissociated by cellular ATP after bound ATP was hydrolyzed. A catalytic role as an enzyme of unknown specificity for synapsin III is also supported by the fact that the sequence in synapsins that corresponds to the catalytic loop in the structurally related enzymes is highly conserved, including the position of a catalytically active lysine residue (underlined in Fig. 1).
Regulation of ATP Binding by Ca 2ϩ -Previous studies showed that ATP␥S binding to the C-domain of synapsin I or to full-length synapsin I required Ca 2ϩ whereas binding was Ca 2ϩ -independent for synapsin II (22). Surprisingly, Ca 2ϩ inhibited ATP␥S binding to the C-domain of synapsin III despite its high degree of sequence homology to synapsins I and II and despite their similar ATP affinity (data not shown). The inhibition of ATP␥S binding to synapsin III was not an artifact caused by Ca 2ϩ -triggered proteolysis because analysis of pro- teins after the ATP incubations showed that Ca 2ϩ did not change the amount of protein left (data not shown). These data suggest that despite their close similarity, synapsins I, II, and III are specialized for different types of synaptic regulation. Synapsin I is activated by Ca 2ϩ , synapsin II is not affected by Ca 2ϩ , and synapsin III is inhibited by Ca 2ϩ .
To test whether the regulation of synapsins by Ca 2ϩ occurred at physiologically meaningful concentrations, we measured the Ca 2ϩ concentration dependence of ATP␥S binding (Fig. 4). All three synapsin C-domains were compared in the same experiment. Ca 2ϩ had no effect on ATP binding to synapsin II but regulated ATP binding to synapsins I and III with Ca 2ϩ concentration dependences that were mirror image patterns. Whereas Ca 2ϩ -activated ATP binding to synapsin I with a half-maximal activation concentration (EC 50 ) of Ϸ2 M free Ca 2ϩ , it inhibited ATP binding to synapsin III with an almost identical EC 50 of Ϸ3 M free Ca 2ϩ . These data demonstrate that synapsins are similar to each other in structure and ATP binding but diverge dramatically in regulation.

DISCUSSION
Synapsins I and II are abundant peripheral membrane proteins of synaptic vesicles that are phosphorylated in nerve terminals (reviewed in Refs. 1 and 25). Although synapsins bind to a number of proteins, the physiological functions of synapsins are unclear. Knockouts revealed that synapsins are essential for synaptic regulation but are not required for the establishment of synapses, clustering of vesicles, or long term synaptic plasticity (16,17). The crystal structure of the Cdomain of synapsin I uncovered an unexpected structural homology to a special class of ATPases (19). These ATPases transfer active phosphates to substrates and mediate ligations or syntheses of small molecules. In agreement with the possibility that synapsins belong to these ATPases, the C-domains of synapsins I and II bind ATP and, with much lower affinity, ADP. In addition, the flexible loop in the ATPases that is catalytically active is conserved in synapsins.
The current study describes data suggesting that the function of synapsins in nerve terminal regulation may be more complex and interesting than previously envisioned. With the discovery of synapsin III, we now show that the synapsin family is larger than anticipated. The sequence of synapsin III allowed a further definition of the domain structure of synapsins, reinforcing and refining the previous model (2). The sequence alignment confirms the role of the central C-domain of synapsins as their largest, conserved domain that presumably carries the activity of the protein. All C-domains of synapsins constitute high affinity ATP-binding modules that may function as phosphotransfer enzymes. In addition, the alignment demonstrates that the amino-terminal phosphorylation domain (the A-domain) is a fixture in synapsins. The fact that this phosphorylation domain is so highly conserved makes the necessity to elucidate the physiological importance of the amino-terminal phosphorylation site even more important. Finally, the structure also confirms the role of the carboxylterminal domain (the E-domain) as a constant part of synapsins, again of unknown function. The sequence alignment also highlights the high content of proline, glutamine, serine, and alanine residues in the two variable regions linking the three conserved domains. These variable regions are not conserved between synapsins, and the phosphorylation sites in synapsin I in these sequences are also not conserved. Nevertheless, these variable sequences are characterized by similar amino acid compositions in all synapsins.
Our most surprising result, however, was the divergence in Ca 2ϩ regulation of synapsins. Analysis of the synapsin I and II knockouts suggested a physiological role for synapsins in short term regulation of neurotransmitter release which is Ca 2ϩ -dependent (16,17). Therefore a direct regulation of synapsins by Ca 2ϩ would agree very well with their essential functions. However, the fact that the three synapsins, despite their high homology and similar ATP affinity, are differentially regulated was unexpected. Synapsin I is Ca 2ϩ -activated, synapsin II Ca 2ϩ independent, and synapsin III Ca 2ϩ -inhibited. For synapsins I and III, Ca 2ϩ acts at concentrations that are similar and physiologically relevant. These data suggest that whatever the enzymatic function may be that is being performed by synapsins in nerve terminals, this function responds differently to increases in nerve terminal Ca 2ϩ during an action potential. Such a difference in Ca 2ϩ regulation for members of a closely related family of proteins is unprecedented and suggests that the three synapsin isoforms specifically evolved to allow different types of synaptic Ca 2ϩ regulation.