Synaptotagmins: why so many?

Synaptotagmins constitute a family of membrane-trafficking proteins that are characterized by an N-terminal TMR, a variable linker, and two C-terminal C2-domains (1). Synaptotagmin 1 (Syt 1) was identified as p65 in a monoclonal antibody screen for synaptic proteins (2) and proposed as a potential Ca sensor for regulated exocytosis when its cloning revealed the presence of two C2-domains (3). Twelve additional synaptotagmins were subsequently discovered (Fig. 1 and Table I; Refs. 4–14). Extensive work showed that Syts 1 and 2 likely function as Ca sensors in synaptic vesicle exocytosis (15–19). Much less is known about the other synaptotagmins, although many of them are abundantly co-expressed with Syts 1 and 2 in brain and are evolutionarily conserved. The most abundant of these “other” synaptotagmins, Syts 3 and 7, are localized on the plasma membrane opposite to synaptic vesicles and exhibit distinct Ca affinities, suggesting that plasma membrane and vesicular synaptotagmins may function as complementary Ca sensors in exocytosis with a hierarchy of Ca affinities (20–22). In the present short review, I shall discuss the synaptotagmin family as a whole and evaluate possible biological reasons for the co-expression of multiple isoforms.

Synaptotagmins constitute a family of membrane-trafficking proteins that are characterized by an N-terminal TMR, 1 a variable linker, and two C-terminal C 2 -domains (1). Synaptotagmin 1 (Syt 1) was identified as p65 in a monoclonal antibody screen for synaptic proteins (2) and proposed as a potential Ca 2ϩ sensor for regulated exocytosis when its cloning revealed the presence of two C 2 -domains (3). Twelve additional synaptotagmins were subsequently discovered ( Fig. 1 and Table I; Refs. 4 -14). Extensive work showed that Syts 1 and 2 likely function as Ca 2ϩ sensors in synaptic vesicle exocytosis (15)(16)(17)(18)(19). Much less is known about the other synaptotagmins, although many of them are abundantly co-expressed with Syts 1 and 2 in brain and are evolutionarily conserved. The most abundant of these "other" synaptotagmins, Syts 3 and 7, are localized on the plasma membrane opposite to synaptic vesicles and exhibit distinct Ca 2ϩ affinities, suggesting that plasma membrane and vesicular synaptotagmins may function as complementary Ca 2ϩ sensors in exocytosis with a hierarchy of Ca 2ϩ affinities (20 -22). In the present short review, I shall discuss the synaptotagmin family as a whole and evaluate possible biological reasons for the co-expression of multiple isoforms.

Structures of Synaptotagmins
We classify as a synaptotagmin any protein that has an Nterminal TMR, a central linker, and two C-terminal C 2 -domains ( Fig. 1). Based on their sequences and properties, the 13 vertebrate synaptotagmins can be grouped into 6 classes (Table I). Fig. 1 displays the domain structures of synaptotagmins and the positions of the introns in the corresponding human genes. 2 The N-terminal sequences preceding the TMRs are usually short (4 -62 residues) and exhibit significant sequence similarity only within a given class. Syts 1 and 2 are the only synaptotagmins with an N-glycosylated N terminus (4,23), and Syts 3, 5, 6, and 10 are the only synaptotagmins with disulfide-bonded cysteines at the N terminus (24). The TMRs of synaptotagmins also lack significant sequence similarity. In all synapotagmins except for Syt 13, an intron is placed just N-terminal to the TMR, 2 and all synaptotagmins except for Syt 12 contain cysteine residues at the cytoplasmic boundary of the TMR. At least in Syts 1 and 7, some of these cysteines are palmitoylated (25)(26)(27). Synaptotagmins form constitutive dimers via their TMRs in a reaction that may require palmitoylation (23,27). One possible reason for the constitutive dimerization of synaptotagmins is to protect them from cleavage by ␥-secretase since ␥-secretase promiscuously cleaves monomeric but not dimeric membrane proteins containing a short extracellular stub (28). The size of the linker between the TMR and the C 2domains varies from 43 residues for Syt 8 to 417 residues for the longest splice variant of Syt 7 ( Fig. 1). At least in Syt 1, the linker is phosphorylated by multiple protein kinases (29,30).
The C-terminal C 2 -domains of synaptotagmins, referred to as the C 2 A-and C 2 B-domains, account for the majority of their sequences and represent their only homologous sequences (Fig. 1). C 2 -domains are widespread modules of 130 -140 residues that were initially defined as the second constant sequence (hence "C 2 ") in protein kinase C isoforms (31) and were first shown in Syt 1 to constitute Ca 2ϩ -binding modules (32)(33)(34). Atomic structures revealed that the synaptotagmin C 2 A-and C 2 B-domains are similarly composed of a ␤-sandwich containing eight ␤-strands, with flexible loops emerging from the top and the bottom (35,36). C 2 Adomains generally bind three Ca 2ϩ ions, whereas C 2 B-domains bind only two Ca 2ϩ ions ( Fig. 3) (Refs. 36 -38; but see below for exceptions). All C 2 B-domains contain a bottom ␣-helix between the 7th and 8th ␤-strands that is absent from C 2 A-domains (35)(36)(37)(38). The fact that the differences between the C 2 A-and C 2 B-domains are conserved indicates a common ancestry and suggests that the C 2 -domains are functionally specialized in all synaptotagmins (1).
In addition to synaptotagmins, proteins such as rabphilin and Doc2s contain similar C 2 A-and C 2 B-domains but are not classified as synaptotagmins because they lack TMRs (reviewed in Ref. 1). Furthermore, two other protein families (called Vp115 and ferlins) contain multiple C 2 -domains and a TMR similar to synaptotagmins but differ from synaptotagmins in that they have more than two C 2 -domains (39 -44).

Genes and Alternative Splicing of Synaptotagmins
Analysis of the human genome revealed that synaptotagmin genes vary from Ͻ10 kilobases (Syts 8 and 9) to Ͼ60 kilobases (Syt 1), and are not physically linked. 2 The exon/intron structures of synaptotagmin genes are nearly identical for synaptotagmins within a class but differ dramatically between classes. Some synaptotagmin genes are relatively simple (e.g. the Syt 4 and 11 genes contain only four exons), whereas others are complex (the Syt 7 gene with 14 exons; Fig. 1). Only a single exon (at the beginning of the C 2 B-domain) is shared among all synaptotagmins, making this exon a signature of synaptotagmin genes.
Alternative splicing of the linker sequence between the TMR and the C 2 -domains of synaptotagmins and of the TMR itself have been reported. Alternative splicing of a short sequence in the linker of Syt 1 (3) probably represents a sliding exon/intron junction. Strikingly, the Syt 7 linker region is subject to extensive alternative splicing with dramatic changes in protein structure (21). Five exons are variably included or excluded in the mRNA ( Fig. 1), resulting in a 10-fold expansion or contraction of the linker. One of the five alternatively spliced exons features an evolutionarily conserved stop codon; insertion of this exon creates a truncated Syt 7 protein (21). The alternative splicing of Syt 7 is developmentally and regionally regulated, suggesting that it regulates the coupling of the C 2 -domains to the TMR (21).
PCR experiments suggested that for many synaptotagmins, variants lacking TMRs are produced by "exon skipping" (45,46). However, the gene structures of these synaptotagmins show that exon skipping would create out-of-frame junctions between the N-and C-terminal exons (Fig. 1). This result together with the fact that no soluble synaptotagmins were observed make the alternative splicing of synaptotagmin TMRs doubtful.

Synaptotagmin C 2 -Domains
The synaptotagmin C 2 -domains are functionally polarized in that Ca 2ϩ exclusively binds to the top loops but not the bottom loops (36 -38, 47, 48). Diverse activities have been reported for synaptotagmin C 2 -domains. In assessing potential interactions of C 2 -domains, structural implications are unfortunately often ignored in favor of "functional" interpretations. For example, experiments with peptide injections into nerve terminals (49) led to the conclusion that the characteristic C-terminal WHXL sequence of the Syt 1 C 2 B-domain functions in synaptic vesicle docking. In the folded C 2 B-domain, however, the WHXL sequence forms a stably bonded ␤-strand (36) that is unlikely to unfold under physiological conditions. Another example is the AD3 mutant in Drosophila (18,19). The AD3 mutation changes a conserved ␤-strand of the C 2 Bdomain and likely interferes with its folding. Claims about testing specific binding activities of the C 2 B-domain with the AD3 mutant domain (50,51) are thus difficult to interpret. Similar reservations apply to other studies of synaptotagmins that involve sequences with conserved secondary structures.
Ca 2ϩ and Phospholipid Binding-C 2 -domains bind Ca 2ϩ exclusively via their top loops (33,47). Fig. 2 shows a model of the Ca 2ϩbinding sites of the Syt 1 C 2 -domains (36,38). The Ca 2ϩ -binding sites of C 2 -domains are unusual because Ca 2ϩ ions are coordinated by residues that are widely separated in the primary sequence and because the coordination spheres for the Ca 2ϩ ions are incomplete (16,47). The intrinsic Ca 2ϩ affinities of the C 2 -domains are low (K d Ͼ1 mM and Ͼ0.3 mM for the Syt 1 C 2 A-and C 2 B-domains, respectively (16,36)). Ca 2ϩ binding to C 2 -domains (except for the unusual piccolo/aczonin C 2 -domain (52)) does not cause significant conformational changes (36 -38), suggesting that C 2 -domains function as electrostatic switches whereby Ca 2ϩ binding triggers interactions by altering their surface charge (48).
The C 2 -domains of Syt 1 bind phospholipids as a function of Ca 2ϩ with apparent Ca 2ϩ affinities (1-20 M Ca 2ϩ ) that are 100 -10,000fold the intrinsic Ca 2ϩ affinities (16,36). This dramatic increase probably occurs because the docked phospholipids on top of the C 2 -domains provide additional coordination sites for the incomplete Ca 2ϩ -binding spheres (16,53). It is likely that phospholipid binding by most synaptotagmins is functionally important but not all synaptotagmins bind Ca 2ϩ and phospholipids. For example, Syts 4 and 11 have an evolutionarily conserved substitution in the Ca 2ϩ -binding site of the C 2 A-domain ( Fig. 2; Ref. 11) that prevents binding of a full complement of Ca 2ϩ ions. In Syts 8, 12, and 13, the top loops of the C 2 A-and the C 2 B-domains lack almost all of the aspartate and glutamate residues involved in Ca 2ϩ binding.
Ca 2ϩ -dependent Self-association-The recombinant C 2 B-domains, but not C 2 A-domains, of most synaptotagmins pull down native Syt 1 and Syt 2 from brain in a Ca 2ϩ -dependent manner (54 -57). This activity suggests a potential mechanism by which C 2 B-domains could mediate Ca 2ϩ -dependent polymerization of synaptotagmins during exocytosis. Some studies detected Ca 2ϩ -dependent multimerization of recombinant C 2 -domains (56 -58), whereas others did not (54,55). Structural characterization of the Syt 1 C 2 B-domain revealed that the electrophoretically pure C 2 Bdomain contains stoichiometric amounts of bacterial contaminants (probably polynucleotides and lipids) that are tightly attached to a polybasic sequence (LKKKKT) on the C 2 B-domain, cause domain aggregation, and are only removed by harsh washes (59). C 2 Bdomain that was cleared of bacterial contaminants exhibited competent Ca 2ϩ binding but no self-association even at millimolar Ca 2ϩ (59), but native Syt 1 formed Ca 2ϩ -dependent high molecular weight complexes, indicating that Ca 2ϩ -dependent self-association (54, 55) may prove relevant but utilize a more complex mechanism.
Interactions with SNAREs-Most synaptotagmins bind to syntaxin 1 in a Ca 2ϩ -dependent manner (7), suggesting an intersection of the mechanism of fusion (via the SNARE protein syntaxin) with the Ca 2ϩ -dependent regulation of fusion (via Syt 1) (reviewed in Ref. 60). However, the precise nature of this interaction is unclear. Binding of Syt 1 to syntaxin, SNAP-25 (another SNARE protein), and assembled SNARE complexes has been reported (7,51,(61)(62)(63). Synaptotagminbinding sites on syntaxin have been localized on the N-terminal Habc domain (63) and the C-terminal SNARE motif (61).
Binding to Endocytosis Proteins-All synaptotagmins tested (Syts 1-7) interact with AP-2 with high affinity (7,64), and binding is enhanced by hydrophobic peptides (65). Transfection of truncated synaptotagmins inhibits clathrin coat assembly and endocytosis in fibroblasts, suggesting that synaptotagmins have a physiological role in endocytosis (26). In Drosophila, mutations in the stoned gene cause a defect in synaptic vesicle endocytosis and also a mislocalization and destabilization of Syt 1 (66,67). Mammalian stoned proteins may also interact with Syt 1, suggesting that this could be a general mechanism in synaptic vesicle endocytosis (68,69).
Binding of Inositol Phosphates-The C 2 B-domains of most synaptotagmins bind to inositol polyphosphates via the short polybasic sequence that contains bacterial contaminants in recombinant C 2 B- in the corresponding human genes as identified in the human genome sequence. 2 The numbers next to the arrows describe the position in the codon at which the coding sequence is interrupted by the intron (0, at the codon junction; 1 and 2, after the first and second codon position, respectively). The N-terminal TMR is marked with a T, and the C 2 Aand C 2 B-domains are labeled.  (70). Binding does not require the folded C 2 B-domain but is obtained with the isolated polybasic sequence. Inhibition of neurotransmitter release by inositol polyphosphates suggests that inositol phosphate binding to synaptotagmin physiologically regulates release (71). However, hundreds of synaptic proteins may bind to inositol phosphates, and the specific role of synaptotagmin is unknown.
Functional Properties of Individual Synaptotagmins Class 1: Syts 1 and 2 as Vesicular Ca 2ϩ Sensors-Syts 1 and 2 are localized to synaptic vesicles and secretory granules (2)(3)(4)22), are closely related, and are differentially expressed in a largely non-overlapping pattern (4,72,73). Syts 1 and 2 probably function as an essential Ca 2ϩ sensor in fast synaptic vesicle exocytosis (15) as shown by a point mutation that was introduced into the endogenous mouse genome (16). This mutation caused an approximately 2-fold decrease in the apparent Ca 2ϩ affinity of Syt 1 and a similar decrease in the Ca 2ϩ sensitivity of synaptic vesicle exocytosis, indicating that Ca 2ϩ binding to Syt 1 is an essential step in triggering exocytosis. However, Syts 1 and 2 are not the only Ca 2ϩ sensor because some Ca 2ϩ -dependent exocytosis remains in their absence (15,17,18). In addition to exocytosis, Syt 1 may also function in endocytosis as discussed above.
Class 2: Syt 7 as an Active Zone Ca 2ϩ Sensor-Syt 7, together with Syt 3, is the most abundant synaptotagmin after Syts 1 and 2. Syt 7 is expressed predominantly in brain (7,21,72,73) where it is primarily present on plasma membranes (20,21). Confusingly, in cultured cells Syt 7 has been reported in lysosomes (76). However, the antibody used in that study did not detect Syt 7 in brain, making the localization of Syt 7 in non-neuronal cells uncertain. The C 2 -domains of Syt 7 have a 10 -20-fold higher Ca 2ϩ affinity than those of Syts 1 and 2 (22) and self-associate as a function of Ca 2ϩ (77). In permeabilized PC12 cells, recombinant C 2 A-and C 2 B-domains of Syt 7 severely inhibit exocytosis (21). C 2 -domains of Syt 1, by contrast, had no effect consistent with the fact that Syt 1 is not essential for Ca 2ϩ -triggered exocytosis in PC12 cells as opposed to synapses (78). Inhibition by recombinant Syt 7 C 2 -domains required intact Ca 2ϩ -binding sites, suggesting that plasma membrane Syt 7 in PC12 cells functions as a Ca 2ϩ sensor for exocytosis.
Class 4: Syts 4 and 11-Syts 4 and 11 are encoded by the simplest genes among synaptotagmins (Fig. 1). Their characteristic feature is a conserved substitution of an aspartate for a serine residue in the C 2 A-domain (Fig. 2) that changes the Ca 2ϩ -binding properties of the C 2 A-domain (11). Syt 4 is expressed at highest levels early during postnatal development (79) but is strongly induced by seizures in adult brain (80) and by increases in cAMP or Ca 2ϩ in PC12 cells (81). Immunocytochemistry of brain sections, cultured neurons, and transfected cells localized Syt 4 in the Golgi apparatus (79,82). In contrast, the Drosophila Syt 4 homolog was observed in synaptic vesicles (83).
The function of Syts 4 and 11 is also uncertain. Microinjection of Syt 4 fragments into PC12 cells (84) and overexpression of Drosophila Syt 4 inhibited exocytosis (83). A knockout of Syt 4 in mice uncovered a subtle but interesting phenotype that is consistent with a modulatory role at the synapse (85). Nothing is known about the expression and localization of Syt 11, but the similarity between Syts 4 and 11 suggests that they are redundant.
Class 6: Atypical Syts 8, 12, and 13 That Lack Consensus Ca 2ϩbinding Sites-Very little is also known about these synaptotagmins. Syt 12 was initially discovered as a thyroid hormone-regulated transcript (12), indicating a specialized function in brain that could be related to thyroid-dependent developmental changes.
A Working Model for Synaptotagmin Function Vertebrates express at least 13 synaptotagmins that are synthesized exclusively or predominantly in neurons and neuroendocrine cells. By virtue of their C 2 -domains, most synaptotagmins are membrane-bound Ca 2ϩ -signaling machines. The presence of distinct synaptotagmins on the membranes of synaptic vesicles and active zones, the membranes that have to fuse during neurotransmitter release, suggests a potential explanation for the existence of multiple synaptotagmins. The model in Fig. 3 proposes that at least Syts 1, 2, 3, 6, and 7 perform complementary functions in Ca 2ϩtriggered exocytosis whereby Ca 2ϩ binding to each class of synaptotagmins contributes differently to triggering exocytosis. As Ca 2ϩ sensors, all synaptotagmins share the same Ca 2ϩ cooperativity and Ca 2ϩ -dependent phospholipid binding but have distinct Ca 2ϩ binding properties. Vesicular synaptotagmins have a lower Ca 2ϩ affin- FIG. 3. Geometry of synaptotagmins during exocytosis. One class of synaptotagmins (Syts 1 and 2 and possibly Syts 4, 9, and 11) are located on synaptic vesicles or secretory granules, whereas a second set (Syts 3, 6, and 7 and possibly others) are targeted to the plasma membrane, creating a symmetrical arrangement of opposing Ca 2ϩsensors. All synaptotagmins are probably dimers via interactions in the TMR and/or disulfide bonds between the extracytoplasmic N-terminal regions (for Syts 3, 5, 6, and 10). In addition to Ca 2ϩ -dependent interactions of the C 2 -domains of synaptotagmins with phospholipids, at least some synaptotagmins also bind to themselves and each other via Ca 2ϩ -dependent interactions that may be mediated by the C 2 B-domains (modified from Ref. 22).
FIG. 2. Structure of the Ca 2؉ -binding sites of the C 2 A-and C 2 B-domains of synaptotagmins. The diagram shows a generic model for synaptotagmin Ca 2ϩ -binding sites that is based on the structure of the C 2 A-and C 2 B-domains of Syt 1 (36,38). The Ca 2ϩ -binding sites at the top of the synaptotagmin C 2 -domains are formed by loops 1 (right blue line) and 3 (left blue line). The C 2 A-domain ligates three Ca 2ϩ ions via five aspartate and one serine residue, whereas the C 2 B-domain lacks the binding site for Ca3 and ligates only two Ca 2ϩ ions. The model is proposed for Syts 1-7 and 9 -11, whereas Syts 8, 12, and 13 lack almost all of the Ca 2ϩ -binding residues. In Syts 4 and 11, one of five Ca 2ϩ -binding aspartates in the C 2 A-domain is substituted for a serine residue, suggesting that this C 2 A-domain may only bind one or two Ca 2ϩ ions (11).
Minireview: Synaptotagmins: Why So Many? 7631 ity and are more important in fast synaptic exocytosis than in endocrine exocytosis, the bulk of which is much slower (15,16,75). Plasma membrane synaptotagmins, in contrast, have a higher Ca 2ϩ affinity and may be more important for endocrine exocytosis (21,22). With these distinct properties, the combination and relative abundance of various synaptotagmins could contribute to shaping the characteristic Ca 2ϩ responses of synapses. Although the model proposed for synaptotagmin function (Fig. 3) is attractive because it provides an explanation for the unprecedented Ca 2ϩ control of neurotransmitter release, it is as yet only a working hypothesis. Questions abound. For example, is Ca 2ϩ binding to a single type of synaptotagmin sufficient to trigger exocytosis, what is the functional relation between different C 2 -domains of synaptotagmins, and how do synaptotagmins precisely work in exocytosis? If synaptotagmins are Ca 2ϩ -signaling machines, why do some synaptotagmins lack Ca 2ϩ -binding sites? Another set of questions regards the potential role of synaptotagmins in endocytosis and the possibility that at least some of them may be expressed outside of neurons and endocrine cells. Addressing these questions will be essential for further progress.