Inositol 1,3,4,5-Tetrakisphosphate Binding Activities of Neuronal and Non-neuronal Synaptotagmins

Synaptotagmins I and II are essential for Ca2+-regulated exocytosis of synaptic vesicles from neurons, probably serving as Ca2+ sensors. This Ca2+-sensing function is thought to be disrupted by binding of an inositol 1,3,4,5-tetrakisphosphate (IP4) to the C2B domain of synaptotagmin I or II (Fukuda, M., Moreira, J. E., Lewis, F. M. T., Sugimori, M., Niinobe, M., Mikoshiba, K., and Llinás, R. (1995) Proc. Natl. Acad. Sci. U. S. A.92, 10708–10712). Recently, several synaptotagmin isoforms, expressed outside the nervous system, have been identified in rats and proposed to be involved in constitutive vesicle traffic. To test whether the inositol high polyphosphates also regulate constitutive vesicle traffic by binding to the non-neuronal synaptotagmins, we examined the IP4 binding properties of the recombinant C2 domains of both neuronal (III, V, X, and XI) and non-neuronal (VI–VIII and IX) synaptotagmins. The C2B domains of synaptotagmins VII–IX and XI had strong IP4 binding activity, but the C2B domain of synaptotagmin VI showed very weak IP4 binding activity. In contrast, there was no significant IP4 binding activity of the C2B domains of synaptotagmins III, V, and X or any of the C2A domains. A phylogenetic tree of the C2 domains of 11 isoforms revealed that synaptotagmins III, V, VI, and X (IP4-insensitive or very weak IP4-binding isoforms) belong to the same branch. Based on the sequence comparison between the IP4-sensitive and -insensitive isoforms, we performed site-directed mutagenesis of synaptotagmin III and identified several amino acid substitutions that abolish IP4 binding activity. Our data suggest that the inositol high polyphosphates might also regulate constitutive vesicle traffic via binding to the IP4-sensitive non-neuronal synaptotagmins.

Synaptotagmins are a family of vesicle membrane proteins characterized by a short intravesicular amino terminus, a single transmembrane region, and two copies of highly conserved repeats homologous to the C2 regulatory region of protein kinase C (named the C2A and C2B domains) in the cytoplasmic domain (reviewed in Ref. 1). To date, at least 11 isoforms (synaptotagmins I-XI) have been described in rats or mice (2)(3)(4)(5)(6)(7)(8)(9)(10). Based on their expression patterns in tissues, synaptotagmins I-V, X, and XI are classified as neuronal types (expressed abundantly in neurons), and others (synaptotagmins VI-IX) are expressed in a wide variety of tissues other than brain (so-called ubiquitous types). Most of the proteins involved in Ca 2ϩ -regulated exocytosis in neurons (e.g. synaptobrevin or syntaxin) have been reported to have homologues involved in constitutive membrane trafficking; and therefore, it has been suggested that the same protein family governs both constitutive and regulated vesicle traffic (11)(12)(13). Based on this idea, the ubiquitous isoforms of synaptotagmin are also thought to be involved in constitutive vesicle trafficking because synaptotagmin I (the best characterized neuronal type) is essential for Ca 2ϩ -regulated exocytosis in neurons and some endocrine cells (probably functioning as a Ca 2ϩ sensor) (reviewed in Ref. 1). However, the exact localization and functions of the ubiquitous synaptotagmins remain unknown.
Recently, we demonstrated the distinct roles of two C2 domains of synaptotagmin I (or II) in Ca 2ϩ -regulated exocytosis in the squid giant presynapse (14,15), superior cervical ganglion cells (16), chromaffin cells (17), and insulin-secreting cells (18) by using specific antibodies against each C2 domain. The C2A domain is crucial for Ca 2ϩ -regulated exocytosis and is directly involved in the fusion of synaptic vesicles with the presynaptic plasma membrane. This fusion step was strongly inhibited by binding of an inositol high polyphosphate (inositol 1,3,4,5-tetrakisphosphate (IP 4 ), 1 inositol 1,3,4,5,6-pentakisphosphate (IP 5 ), and inositol hexakisphosphate (IP 6 )) to the C2B domain of synaptotagmin I (or II) (15-17, 19 -23). In chromaffin cells particularly, IP 5 is suggested to function as a fusion clamp for exocytosis because IP 5 is rapidly accumulated after depolarizing stimulation (17). These observations raised the possibility that inositol high polyphosphates may also regulate other types of vesicle traffic (e.g. constitutive) via binding to ubiquitous members of the synaptotagmin family (24).
To address this question, we examined the inositol high polyphosphate binding properties of the C2 domains of all * This work was supported in part by grants from the Japanese Ministry of Education, Science, Sports, and Culture (to K. M. and M. F.) and from the Science and Technology Agency of Japan (to K. M.). 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  synaptotagmin isoforms identified to date (synaptotagmins I-XI) as an indicator of IP 4 binding activity. In this study, we show that the C2B domains, but not the C2A domains, of non-neuronal synaptotagmin isoforms (VII-IX) also have strong IP 4 binding activities. In addition, we newly identified a subclass of synaptotagmins deficient in IP 4 binding activity, despite having a putative IP 4 -binding sequence as determined previously (20,22). Interestingly, this class of synaptotagmins (III, V, VI, and X) is structurally related and distinguished from other isoforms by phylogenetic trees of the C2 domains. We further determined the conserved amino acid substitutions that abolish IP 4 binding activity by site-directed mutagenesis. On the basis of these results, we discuss the functional difference between IP 4 -sensitive and -insensitive synaptotagmins in vesicular trafficking.
Molecular Cloning of Mouse Synaptotagmin B/K and SrgI (Synaptotagmin-related Gene I)-cDNAs encoding two C2 domains of SrgI and SytB/K were also amplified by reverse transcriptase-PCR from mouse cerebellum cDNAs using the following primers designed on the basis of rat sequences (26,27) with the addition of appropriate restriction endonuclease sites: SrgI, 5Ј-GAAGATCTATGGCCGTGGACGTGACAGA (sense) and 3Ј-CCAATTGTTAGTTTCGCCGGACTGGAT-3Ј (antisense); and SytB/K, 5Ј-CGGGATCCATGGCGTACATCCAGTTGGA-3Ј (sense) and 5Ј-GGAATTCTCAGGTCACCTCCAGCGAGG-3Ј (antisense). Reaction conditions were the same as described above. After digestion with BamHI or BglII and EcoRI or MunI, the PCR products were purified on an agarose gel and extracted with a Geneclean II kit and then inserted into the BamHI-EcoRI site of the pGEX-2T vector. Only the nucleotide sequences coding for two C2 domains were sequenced in both directions using a BcaBEST dideoxy sequencing kit. As compared with rat sequences, three amino acid substitutions were found in mouse SrgI (D315E, T317S, and A393V). These changes are probably not due to PCR-induced errors because they were also found in two independent PCR products. Using primers based on the mouse nucleotide sequences obtained above, fragments encoding the C2A or C2B domains of SrgI and SytB/K were amplified by PCR. After digestion with BamHI or BglII and EcoRI, the PCR products were inserted into the BamHI-EcoRI site of the pGEX-2T vector and verified by DNA sequencing. GST-Srg-C2A, -C2B, and -C2B⌬C contained amino acids 149 -278, 278 -404, and 278 -356, respectively, of mouse SrgI; and GST-STB/K-C2A, -C2B, and -C2B⌬C contained amino acids 180 -315, 315-443, and 315-394, respectively, of mouse SytB/K.
Site-directed Mutagenesis of the Synaptotagmin III C2B Domain-Site-directed mutagenesis of GST-STIII-C2B␣-(P505F,E509Q,N510K) and GST-STIII-C2B␤7-(H525K,V531K,C532I,R533F) was carried out by means of two-step PCR as follows (22). In GST-STIII-C2B␣-(P505F,E509Q,N510K), for example, the right and left halves of the C2B domain were separately amplified with two pairs of oligonucleotides (primer A (5Ј-CGGGATCCGAAAAGGCAGATCTTGGGGA-3Ј) and mutagenic primer B (5Ј-GCGACGTCGAACACCAGGGCCT-3Ј) (left half); mutagenic primer C (5Ј-TCGACGTCGCTTTCGAGAGCGTGCA-GAAAGTGGGTCTCAG-3Ј) and primer D (5Ј-GGAATTCATCTCTGC-CCAGTGTTCTC-3Ј) (right half)). The two resulting PCR fragments were digested with AatII (underlined), ligated to each other, and reamplified with primers A and D. The obtained PCR fragment encoding the mutant C2B domain of synaptotagmin III (P505F,E509Q,N510K) was subcloned into the BamHI-EcoRI site of pGEX-2T and verified by DNA sequencing. Site-directed mutagenesis of GST-STIII-C2Bloop␤7-8-(E537N,A539T,D540G,G543L) was achieved by PCR using primer A and a mutagenic primer (primer E, 5Ј-GGAATTCATCTCTGCCCAG-TGTTCTCTGAGGTGTGGGCCGGTAGCGTTTGGGCCCACG-3Ј). The obtained PCR fragment encoding the mutant C2B domain of synaptotagmin III (E537N,A539T,D540G,G543L) was subcloned into the BamHI-EcoRI site of the pGEX-2T vector and verified by DNA sequencing. Other plasmids encoding the mutant C2B domain of synaptotagmin were similarly constructed by means of PCR using mutagenic primers. 4 Binding to GST Fusion Proteins-Measurement of IP 4 binding was performed as described previously (20) with slight modifications. Briefly, the buffer system was changed from 20 mM Tris-HCl (pH 8.0) to 50 mM HEPES-KOH (pH 7.2) because GST-STII-C2B showed stronger IP 4 binding activity in the latter buffer. GST fusion proteins (1-2.5 g) were incubated with 9.6 nM [ 3 H]IP 4 (specific radioactivity of 777 GBq/mmol; NEN Life Science Products) in 49 l of 50 mM HEPES-KOH (pH 7.2) for 10 min at 4°C. The sample was then mixed with 1 l of 50 mg/ml ␥-globulins and 50 l of a solution con-taining 30% (w/v) polyethylene glycol 6000 and 50 mM HEPES-KOH (pH 7.2) and placed on ice for 5 min. The precipitate obtained by centrifugation at 10,000 ϫ g for 5 min was solubilized in 500 l of Solvable (Packard Instrument Co.), and radioactivity was measured in Aquasol 2 (Packard Instrument Co.) with a liquid scintillation counter. Nonspecific binding was determined in the presence of 10 M nonradioactive IP 4 Sequence Analyses-Multiple sequence alignment of the C2 domains of synaptotagmin isoforms was performed using the PILEUP program of the GCG program (Version 8.1). Calculation of genetic distance and suitable depiction of the phylogenetic tree using the neighbor joining method were performed with the SINCA program (Fujitsu). Putative IP 4 -binding sites were aligned referring to the multiple alignment results.

Measurements of [ 3 H]IP
Data Processing-Statistical analysis and curve fitting were done using the GraphPad PRISM computer program (Version 2.0).

IP 4 Binding Activity of Synaptotagmin Isoforms (V-XI)-The
C2 domain, originally identified as a sequence motif of protein kinase C, is a conserved protein module of ϳ120 amino acids found in many proteins (28). Among them, synaptotagmins are apparently distinguished from other proteins in that they have a single transmembrane region and tandem C2 domains (C2A and C2B domains) with a short spacer. In our previous studies, we showed that neuronal synaptotagmins I, II, and IV, but not synaptotagmin III, are IP 4 -or inositol high polyphosphatebinding proteins (20,22,23). To further examine whether other neuronal and non-neuronal isoforms of synaptotagmins are also regulated by inositol high polyphosphates like synaptotagmin I, we prepared GST fusion proteins of C2 domains of synaptotagmins V-XI and tested for their IP 4 binding activity ( Fig. 1 and Table I). GST-STVII, -STVIII, -STIX, and -STXI-C2B had strong IP 4 binding activity like synaptotagmin II, but GST-STVI-C2B showed weak IP 4 binding activity (Ͻ20% of that of GST-STII-C2B). In contrast, GST-STV and -STX and all the GST-ST-C2A fusion proteins showed no significant IP 4 binding activity under our experimental conditions.
Analysis of the Phylogenetic Tree of Synaptotagmin C2 Domains-To understand the relationship between the molecular evolution of the C2 domains of synaptotagmins and IP 4 binding capacity, a phylogenetic tree of the C2 domains of synaptotagmin isoforms was constructed using the neighbor joining method (Fig. 2). In this phylogenetic tree, the C2B domain of synaptotagmin I is expressed as the most primitive or original form of the C2B domain because this domain of invertebrate synaptotagmins has been identified in Drosophila (31), Caenorhabditis elegans (32), Aplysia (33), and squid (Loligo pealei) (14) as having a less distant genetic relationship to mouse synaptotagmin I than the other isoforms (data not shown). According to this phylogenetic tree, the C2 domains of synaptotagmin isoforms are classified into two distinct groups, C2A and C2B, and the C2 domain from the same isoform is located at very similar positions in the two groups, suggesting that mammalian synaptotagmin isoforms were separated after the tandem C2 domains had been produced. When synaptotagmins that bind IP 4 strongly or weakly are solid-boxed or brokenboxed, respectively, it is apparent that the C2B domains of synaptotagmins III, V, VI, and X (IP 4 -insensitive or weak binding isoforms) form a small but distinct branch (Fig. 2).
Sequence Comparison of Putative IP 4 -binding Domains-To further examine whether these synaptotagmins (III, V, VI, and X) have a common sequence responsible for the lack of IP 4 binding at the amino acid level, we compared the putative IP 4 -binding sites of all synaptotagmin isoforms as determined previously (20,22) (Fig. 3). However, in this region, no apparent differences were observed between IP 4 -sensitive and -insensitive synaptotagmin isoforms. Within the putative IP 4binding domain, three positively charged amino acids responsible for high affinity IP 4 binding activity (Lys at positions 327, 328, and 332 of synaptotagmin II (22); asterisks in Fig. 3) were highly conserved among isoforms, whereas the corresponding positions in the C2A domains of virtually none of these molecules are occupied by positively charged amino acids (data not shown) (24). Since SytVIII-C2B lacks one of the important Lys residues (Ser at position 252, Ser-252), its IP 4 binding activity was weaker than that of synaptotagmin II (Fig. 3), which is consistent with our previous mutational analysis (22).
Although no substitutions corresponding to the three important positively charged amino acids were found in the C2B domains of synaptotagmins III, V, and X, they showed no IP 4 binding activity whatsoever. The results indicate that these three residues are important for IP 4 binding activity, but are not solely responsible for this binding. Furthermore, chimeric and deletion analyses of synaptotagmin III revealed the loss of IP 4 binding activity to be due mainly to one-third of the C terminus of the C2B domain of synaptotagmin III (22). Since  Ϫ Ϫ 30 a Ϫ, no significant IP 4 binding activity; ϩϩϩ, 75-100% of the IP 4 binding activity of GST-STII-C2B (see Fig. 1); ϩϩ, 50 -75% of the IP 4 binding activity of GST-STII-C2B; ϩ, 5-25% of the IP 4 binding activity of GST-STII-C2B. synaptotagmins III, V, VI, and X belong to the same small branch of the phylogenetic tree (Fig. 2), it seems likely that loss of the IP 4 binding activity of synaptotagmins V and X or the weak IP 4 binding ability of synaptotagmin VI also results from alterations in the sequence of the C termini of their C2B domains. To address this question, C-terminal deletion mutants of synaptotagmins V and X (GST-STV-C2B⌬C and -STX-C2B⌬C) were produced (Fig. 4B). As expected, both GST-STV-C2B⌬C and -STX-C2B⌬C showed IP 4 binding activity (70% of that of GST-STII-C2B⌬C) (Fig. 4C), similar to the observations in our previous study using GST-STIII-C2B⌬C. Such recovery of IP 4 binding by deletion of one-third of the C terminus seems to be a unique event occurring only in synaptotagmins III, V, and X because other tandem C2 proteins closely related to synaptotagmin, SrgI (26) and SytB/K (27), did not bind IP 4 even when their C termini were deleted (Fig. 4C).

DISCUSSION
In this study, we have demonstrated that IP 4 binding to the C2 domains is unique to the C2B domains of certain synaptotagmin isoforms, but not to other tandem C2 domains of SrgI, SytB/K, rabphilin 3A (20) and Doc2␣, and Doc2␤ (30) (summarized in Table I). Based on a sequence comparison between IP 4 -sensitive and -insensitive isoforms (Figs. 3 and 4A) and deletion or mutational analysis of synaptotagmin III (Figs. 4 and 5), we propose that the primitive form of synaptotagmin originally bound IP 4 and that, during the diversification of synaptotagmin in vertebrates, some isoforms (III, V, and X) lost their IP 4 binding activity as a consequence of several amino acid substitutions mainly in the C-terminal region of the C2B domain, but not in the putative IP 4 -binding site itself. This hypothesis was supported by the following evidence. (i) In invertebrates (Drosophila, C. elegans, Aplysia, and squid), only one synaptotagmin, closely related to mouse synaptotagmin I, has been reported (14,(31)(32)(33). Since the putative IP 4 -binding site of synaptotagmin I is highly conserved between vertebrates and invertebrates and squid synaptotagmin also bound IP 4 (15), the IP 4 binding properties are likely to have been retained during evolution. (ii) IP 4 -insensitive or weak IP 4 -binding synaptotagmins (III, V, VI, and X) also have a putative IP 4 -binding site containing a KK(K/R)TXXK(K/R) basic sequence (Fig. 3) (22). (iii) The C-terminal truncated form of synaptotagmins III, V, and X showed IP 4 binding activities comparable to that of synaptotagmin II (Fig. 4C). (iv) Substi-  (29). Among these subdomains, the ␣-helix between the ␤5 and ␤6 strands, the ␤7 strand, and the loop between the ␤7 and ␤8 strands contain conserved amino acids only among IP 4 -insensitive isoforms (see below the sequence and Fig. 5A). The amino acid (Asn-471) probably involved in the weak IP 4 binding activity of synaptotagmin VI (see "Results") is indicated (#). B, schematic representation of synaptotagmin III and its C-terminal truncated mutant and the wild-type GST fusion protein. Other C-terminal truncated mutants (GST-STII-C2B⌬C, -STV-C2B⌬C, -STX-C2B⌬C, -Srg-C2B⌬C, and -STB/K-C2B⌬C) lack a corresponding C-terminal region like GST-STIII-C2B⌬C. The transmembrane region (TM), the two C2 domains, and GST are designated by open, hatched, and cross-hatched boxes, respectively. The putative IP 4 -binding sites (amino acid residues 470 -501; see Fig. 3) are indicated below. C, effect of C-terminal deletion of the C2B domain of SytII, SytIII, SytV, SytX, SrgI, and SytB/K on IP 4 binding activity. GST fusion proteins (1-2.5 g) were analyzed by [ 3 H]IP 4 binding assay as described under "Materials and Methods." Black, open, and hatched bars indicate the IP 4 binding activities of GST-C2A, -C2B, and -C2B⌬C fusion proteins, respectively. Note that GST-STIII-C2B⌬C, -STV-C2B⌬C, and -STX-C2B⌬C showed IP 4 binding activity similar to that of GST-STII-C2B⌬C, but GST-Srg-C2A, -C2B, or -C2B⌬C, and -STB/K-C2A, -C2B, or C2B⌬C did not show any significant IP 4 binding activity. The data are means Ϯ S.D. of three or four measurements, normalized to 100% for binding to GST-STII-C2B (100% specific binding ϭ 6.93 Ϯ 0.25 pmol/nmol of protein).
Based on the three-dimensional structure of the C2A domain of synaptotagmin I (29), Lys-rich basic residues in the C2B domain are thought to be located in the fourth strand of the eight-stranded ␤ sandwich, named C2 key. As shown in Fig.  4A, one-third of the C terminus of the C2B domain corresponds to the ␣-helix and the ␤6, ␤7, and ␤8 strands; and conserved substitutions among synaptotagmins III, V, VI, and X are found in the ␣-helix, the ␤7 strand, and the loop between the ␤7 and ␤8 strands. Among these, the ␣-helix between the ␤5 and ␤6 strands is predicted to be structurally closer to the IP 4binding site than the others (29) and may shield the basic region of synaptotagmin III from IP 4 because of the presence of negatively charged rather than positively charged amino acids (e.g. Glu-509 of synaptotagmin III) or the presence of an ␣-helix breaker, i.e. Pro-505. Consistent with this finding, the ␣-helix mutation (P505F,E509Q,N510K) more effectively restored IP 4 binding than the mutation in the ␤7 strand (H525K, V531K,C532I,R533F) or in the loop between the ␤7 and ␤8 strands (E537N,A539T,D540G,G543L). Although, at this stage, we do not know how ␤7-8 loop substitutions affect the IP 4 -binding site, the combination of ␣-helix and loop ␤7-8 mutations of synaptotagmin III greatly enhances IP 4 binding activity (Fig. 5). Taken together, these results indicate that abolishing the IP 4 binding capacity of the C2B domain of synaptotagmin requires a combination of amino acid substitutions. In contrast, the phospholipid binding properties of the C2A domain of synaptotagmins IV and XI changed dramatically with only one mutation (e.g. D244S of synaptotagmin IV) (10,25).
In neurons, synaptotagmins I and II are thought to function as low affinity Ca 2ϩ sensors for neurotransmitter release (1) and to be directly involved in the vesicular fusion step (14). In endoplasmic recticulum-Golgi transport, Ca 2ϩ is now known to be required at a stage between vesicle docking and the actual membrane fusion event because EGTA inhibits this stage (34 -36). Thus, it is possible that non-neuronal synaptotagmins VII-IX participate in this Ca 2ϩ -requiring event and that their function may be regulated by binding of IP 4 to the C2B domain.
What is the functional difference between IP 4 -sensitive and -insensitive (SytIII, SytV, or SytX) isoforms of synaptotagmins? If synaptotagmin III and X function as Ca 2ϩ sensors for vesicular exocytosis, it is most likely that their functions are unaffected by inositol high polyphosphates. The effects of inositol high polyphosphates on the in vitro biochemical nature of synaptotagmin III (e.g. Ca 2ϩ -dependent phospholipid and syntaxin binding) were not examined in this study. However, very recently, we demonstrated binding of the clathrin assembly protein AP-2 to the C2B domain of synaptotagmin II to be inhibited by IP 6 , whereas the SytIII-C2B/AP-2 interaction was not affected by IP 6 (37). Thus, if IP 6 functions as an inhibitor of endocytosis as proposed previously (24,38), synaptotagmin III is a target of AP-2 even in the presence of IP 6 .
In summary, we investigated the IP 4 binding properties of the C2 domains of neuronal and non-neuronal isoforms of synaptotagmin and showed that most of the C2B domains can bind IP 4 , suggesting that the inositol high polyphosphates may function as negative modulators of both regulated and constitutive vesicle trafficking, although the exact functions of isoforms other than synaptotagmins I and II remain to be clarified.