The Evolutionary Pressure to Inactivate

Synaptotagmin I is a Ca2+-binding protein of synaptic vesicles that serves as a Ca2+ sensor for neurotransmitter release and was the first member found of a large family of trafficking proteins. We have now identified a novel synaptotagmin, synaptotagmin XI, that is highly expressed in brain and at lower levels in other tissues. Like other synaptotagmins, synaptotagmin XI has a single transmembrane region and two cytoplasmic C2-domains but is most closely related to synaptotagmin IV with which it forms a new subclass of synaptotagmins. The first C2-domain of synaptotagmin I (the C2A-domain) binds phospholipids as a function of Ca2+ and contains a Ca2+-binding site, the C2-motif, that binds at least two Ca2+ ions via five aspartate residues and is conserved in most C2-domains (Shao, X., Davletov, B., Sutton, B., Südhof, T. C., Rizo, J. R. (1996) Science 273, 248–253). In the C2A-domains of synaptotagmins IV and XI, however, one of the five Ca2+-binding aspartates in the C2-motif is substituted for a serine, suggesting that these C2-domains do not bind Ca2+. To test this, we produced recombinant C2A-domains from synaptotagmins IV and XI with either wild type serine or mutant aspartate in the C2-motif. Circular dichroism showed that Ca2+stabilizes both mutant but not wild type C2-domains against temperature-induced denaturation, indicating that the mutations restore Ca2+-binding to the wild type C2-domains. Furthermore, wild type C2A-domains of synaptotagmins IV and XI exhibited no Ca2+-dependent phospholipid binding, whereas mutant C2A-domains bound phospholipids as a function of Ca2+ similarly to wild type synaptotagmin I. These experiments suggest that a class of synaptotagmins was selected during evolution in which the Ca2+-binding site of the C2A-domain was inactivated by a single point mutation. Thus, synaptotagmins must have Ca2+-independent functions as well as Ca2+-dependent functions that are selectively maintained in distinct members of this gene family.

Synaptotagmins represent a large protein family with at least ten genes that probably function in membrane traffic (1)(2)(3)(4)(5)(6)(7)(8)(9). All synaptotagmins are characterized by a single Nterminal transmembrane region and two cytoplasmic C 2 -do-mains followed by a short conserved C terminus. Synaptotagmins I and II probably serve as Ca 2ϩ sensors in fast Ca 2ϩdependent exocytosis (10,11). Synaptotagmin III may have a distinct function at the synapse, possibly in mediating the slow component of Ca 2ϩ -dependent release (12), while the other synaptotagmins are thought to function in related neuronal and nonneuronal membrane trafficking reactions (6). However, the total number of synaptotagmins and their localizations and functions are unknown. The current study was initiated to determine if there are additional synaptotagmins in mammals that could be grouped into subclasses with distinct properties.
Most synaptotagmins are Ca 2ϩ -binding proteins (1). In synaptotagmin I, the first C 2 -domain (C 2 A-domain) binds to phospholipids and syntaxin as a function of Ca 2ϩ (6,13). The second C 2 -domain mediates the Ca 2ϩ -dependent binding of synaptotagmin to itself, leading to Ca 2ϩ -dependent homomultimers (14,15). A crystal structure of the C 2 A-domain of synaptotagmin I revealed that it represents a compact domain composed of eight ␤-strands forming two ␤-sheets (16). Three sequence loops emerge from the top, and four from the bottom of the domain. Ca 2ϩ binding to the C 2 A-domain stabilizes it (17) but does not induce a major conformational change (18,19). Detailed studies of Ca 2ϩ binding to the C 2 A-domain by NMR spectroscopy demonstrated that it binds at least two Ca 2ϩ ions at a site formed by two of the top three loops (19). Ca 2ϩ is coordinated by five aspartate residues, three of which coordinate both Ca 2ϩ ions. Thus the C 2 -domain contains an unusual Ca 2ϩ -binding site, designated the C 2 -motif, that is formed by aspartate residues on discontinuous sequence loops.
More than 50 C 2 -domain sequences are present in the data banks, suggesting that it is a widespread domain (20). The C 2 -motif is conserved in many of these C 2 -domains that are thus likely to bind Ca 2ϩ similarly to the C 2 A-domain of synaptotagmin I. In agreement with the binding of at least two Ca 2ϩ ions to the C 2 A-domain of synaptotagmin I, Ca 2ϩ cooperatively activates phospholipid binding to native synaptotagmin I (12,21) and to recombinant C 2 A-domain (13). Phospholipid binding is promiscuous and only requires negatively charged phospholipids (12). In addition to phospholipids, C 2 A-domains bind syntaxin as a function of Ca 2ϩ but with distinct Ca 2ϩ affinities that suggest different functions in membrane traffic (6).
Although most C 2 A-domains from synaptotagmins contain the C 2 -motif and bind phospholipids as a function of Ca 2ϩ , those of synaptotagmins IV and VIII do not (1,6,11). Synaptotagmin VIII contains many changes in the sequences of the Ca 2ϩ -binding loops, indicating that the C 2 -motif is not formed in this synaptotagmin. By contrast, the C 2 A-domain of synaptotagmin IV is very similar to that of the other synaptotagmins and contains all residues of the C 2 -motif except that one of the five Ca 2ϩ -coordinating aspartates is substituted for a serine. In this study, we now describe the identification of a novel synaptotagmin, synaptotagmin XI, that is highly homologous to synaptotagmin IV and also contains the aspartate to serine substitution in the C 2 -motif. Since the changed aspartate coordinates both Ca 2ϩ ions in the C 2 -motif, the substitution is predicted to abolish Ca 2ϩ binding, and indeed no Ca 2ϩ -dependent phospholipid binding was observed with standard liposomes for the C 2 A-domain of synaptotagmin IV (11).
In addition to participating in Ca 2ϩ -dependent interactions, C 2 -domains from synaptotagmins also exhibit Ca 2ϩ -independent activities. For example, the second C 2 -domain of all synaptotagmins tested binds AP2, a clathrin adaptor protein complex involved in endocytosis (6,22). Since this and other Ca 2ϩindependent interactions of C 2 -domains are not regulated, it is difficult to determine if they are physiologically relevant. Data showing that some C 2 -domains do not exhibit Ca 2ϩ -dependent interactions do not actually prove that these C 2 -domains are indeed Ca 2ϩ -independent (6,11). Such a demonstration could be obtained, however, if reversal of an inactivating amino acid substitution restored Ca 2ϩ -dependent properties. Indeed, the notion that the C 2 A-domain of synaptotagmin IV is a Ca 2ϩindependent domain was challenged by a report of Ca 2ϩ -dependent binding of the C 2 A-domain of synaptotagmin IV to phospholipids consisting of 100% PS 1 (23). It is thus important to establish if synaptotagmins IV and XI are Ca 2ϩ -independent because this would support a general function for C 2 -domains in Ca 2ϩ -independent reactions. Therefore, a further goal of the current study was to determine if the C 2 A-domains of synaptotagmins IV and XI are selectively inactivated in evolution as Ca 2ϩ -binding modules by a single substitution, or if these C 2domains contain additional changes that differentiate them from other synaptotagmins. Our data reveal that synaptotagmins IV and XI contain a single point mutation that selectively abolishes Ca 2ϩ binding but leaves other properties of the C 2domain intact.

EXPERIMENTAL PROCEDURES
Cloning of synaptotagmin XI and sequence analysis-Searches of GenBank with a consensus sequence from the C-terminal domain of synaptotagmins and double C 2 -domain proteins (1) uncovered a human EST sequence (accession number D38522) that represents a novel member of this family. PCR primers were designed based on the human sequence (sequences: GCGGAATTCCAGGTAATCCTTATGTCAAGGT-GAA[C, T]GT and GGCCGTCGACTAGTA CTCGCTCAGACTGTGCC-A[C, T]TT[C, T,A, G]GC; letters in brackets indicate redundant posi-tions) and used for PCRs with total rat brain cDNA to isolate the corresponding rat sequence (24). A single product of the correct size (0.325 kb) was obtained, verified by sequencing, and used to screen a rat brain cDNA library by standard techniques (24). Positive clones were sequenced, with two clones containing the entire coding region as judged by comparison with the sequence of synaptotagmin IV and the presence of in-frame stop codons in the longest clone. The cDNA sequence has been deposited in GenBank™ (accession number AF000423).

Construction of Expression Vectors and Purification of Recombinant
Proteins-The C 2 A-domains of synaptotagmins IV and XI were subcloned into pGEX-KG (25) with wild type or mutant sequence by PCR as described (6,14), resulting in the following expression plasmids. pGEX-T1082/1083 and pGEX-T1082/1083 S244D encode residues 151-282 of wild type and mutant synaptotagmin IV, respectively, with the mutant containing a substitution of serine 244 to aspartate; pGEX-T1454/1455 and pGEX-T1454/1455 S247D encode residues 153-287 of wild type and mutant synaptotagmin XI, respectively, with the mutant containing a substitution of serine 247 to aspartate; both synaptotagmin XI constructs contain an additional C-terminal serine residue before the termination codon; and pGEX65-4 encodes residues 140 -267 of synaptotagmin I. Recombinant GST-fusion proteins were purified on glutathione-agarose. Proteins were used for phospholipid binding measurements immobilized on glutathione-agarose without elution. For the CD studies, the C 2 A-domains were cleaved from the GST-fusion proteins on the column by thrombin (0.25 mg/ml of resin at room temperature for 3 h) and further purified by gel filtration on a Superdex-75 FPLC column.
Phospholipid Binding Measurements-Phospholipids (3.5 mg total, obtained from Avanti Poalr Lipids) were dissolved in chloroform, mixed in the indicated weight ratios with a trace amount of 3 H-labeled PC (Ͻ0.01% total; Amersham), and dried under a stream of nitrogen. Dried lipids were resuspended in 20 ml of 50 mM HEPES-NaOH, pH 7.4, 0.1 M NaCl (buffer A) by vigorous shaking for 1 min. Suspensions were sonicated for 20 s in a Branson probe sonicator (model 450) at an intensity setting of 5 and centrifuged for 20 min at approximately 5,000 ϫ g to remove aggregates. The standard binding assay contained 25 g of recombinant protein with 1 g of protein/l wet glutathione beads. Beads were prewashed and resuspended in the respective incubation buffers (0.1 ml buffer A containing 1 mM EGTA, Ϸ9 g of phospholipids with Ϸ0.025 Ci 3 H-labeled PC, and either no additions or 1.12 or 2 mM either Mg 2ϩ or Ca 2ϩ as stated in the legend to Figs. 5 and 6). The mixture was incubated for 10 min at room temperature with vigorous shaking, briefly centrifuged, and washed eight times with 0.8 ml of the respective incubation buffer. Phospholipid binding was quantified by scintillation counting.
Ca 2ϩ -dependent Thermal Denaturation Monitored by Circular Dichroism-Thermal denaturation data were obtained on an Aviv model 62DS spectropolarimeter using a 1-mm path-length cell. Approximately 8 M C 2 A-domain in 40 mM Tris-HCl, pH 8.0, 0.1 M NaCl and 0.5 mM EGTA with or without 5.5 mM CaCl 2 were used. Thermal denaturation was monitored by changes in CD absorption at 217 nm from 25 to 95°C in 1°C steps. The fraction of unfolded protein at each temperature was calculated as (I obs -I f )/(I u -I f ), where I obs is the observed signal intensity and I u and I f are the signal intensities of the unfolded and folded states, respectively. I u and I f were obtained by extrapolation of the linear regions of the unfolding curves.
Miscellaneous Procedures-Amounts of proteins used were standardized based on Coomassie Blue stained SDS-gels and UV absorption at 280 nm. RNA blotting experiments were performed at high stringency with blots loaded with poly(A) ϩ -enriched RNA purchased from CLON-TECH as described previously (24).

RESULTS
Cloning of Synaptotagmin XI-At the C terminus, synaptotagmins and other double C 2 -domain proteins contain a unique conserved domain after the second C 2 -domain (1). Searches of GenBank TM with a consensus sequence derived from the Cterminal domain identified a human EST sequence (accession number D38522) from a myeloblast cell line that was related to, but distinct from, all currently known synaptotagmins and other double C 2 -domain proteins. To determine if a homologue of this EST was present in rat brain, we synthesized PCR primers based on the human sequence and amplified a corresponding fragment from rat brain total cDNA. A single band of the correct size was obtained that was confirmed by sequencing to correspond to the rat homologue of the human EST sequence (data not shown). Using the PCR product as a probe, we isolated multiple independent cDNA clones from a rat brain cDNA library. Two of the cDNA clones were found to contain the complete coding region, and the sequence of the entire protein was assembled from the sequences of the cDNA clones (Fig. 1).
The amino acid sequence translated from the cDNA sequence identifies the new double C 2 -domain protein as a synaptotagmin based on the criteria established for this protein family (1). It has no signal sequence, contains a single transmembrane region with multiple neighboring cysteines, has two C 2 -domains, and has a short C-terminal sequence homologous to the C-terminal domains of other double C 2 -domain proteins (Fig.  1). In continuation of the naming of other members of this protein family, we named this protein synaptotagmin XI.
Expression of Synaptotagmin XI-To learn which rat tissues express synaptotagmin XI, we performed an RNA blot analysis. High levels of synaptotagmin XI mRNA were observed in brain (Fig. 2). In addition, low levels of the mRNA could be detected in almost all tissues tested, suggesting that synaptotagmin XI, similar to synaptotagmins IV, VI, and VII (6), is not brainspecific but ubiquitously expressed in low abundance.
Definition of a Subclass of Synaptotagmins-With the new synaptotagmin described here, synaptotagmins now constitute a family of 11 proteins (2)(3)(4)(5)(6)(7)(8)(9). Although all synaptotagmins are composed of similar domains, their N-terminal regions (intravesicular sequence, transmembrane region, and linker between the transmembrane region and the C 2 A-domain) exhibit little similarity to each other except for pairwise similarities between synaptotagmins I and II and between III and VI (1, 6). By contrast, the C-terminal domains (the two C 2 -domains and the short C-terminal domain) are highly homologous between all synaptotagmins. Sequence analysis of synaptotagmin XI demonstrates that it is most closely related to synaptotagmin IV (53% overall identity; Fig. 1). Synaptotagmins IV and XI are homologous to each other in their N-terminal 73 amino acids which exhibit no sequence similarity to other synaptotagmins, defining them as a separate subgroup. In addition, the C 2domains of synaptotagmins IV and XI are more closely related to each other than to those of other synaptotagmins. The sequence differences between different synaptotagmins are not random variations since they are evolutionarily conserved (98% identity between rat and mouse synaptotagmin IV) (11). Similarly, the C-terminal 104 amino acids of human synaptotagmin XI encoded by EST D38522 exhibits only a single amino acid change compared with the rat sequence.
The C 2 A-domain of synaptotagmin I binds two Ca 2ϩ ions through five aspartate residues located in two loops (19). Three of the aspartates are bifunctional and ligate both Ca 2ϩ ions. Alignment of different C 2 A-domains shows that most contain either aspartates or glutamates at the five positions that are involved in Ca 2ϩ binding (Fig. 3). The C 2 A-domains of synaptotagmins IV and XI, however, are different. The general spacing is the same, but, in one position (corresponding to aspartate 230 in synaptotagmin I), a serine is substituted for the aspartate (Fig. 3). The aspartate to serine substitution appears not to be a sequencing or cloning artifact since it is present in both synaptotagmin IV and XI and since the serine is evolutionarily conserved in synaptotagmin IV (11).
Aspartate 230 is a critical residue in synaptotagmin I since it coordinates both Ca 2ϩ ions and is at the center of the C 2 -motif (19). Modeling of the analogous Ca 2ϩ -binding site in synaptotagmins IV and XI suggests that in the presence of serine instead of aspartate, Ca 2ϩ binding is unlikely (Fig. 4A). This model also suggests that Ca 2ϩ binding could potentially be restored if the serine is mutated to aspartate (Fig. 4B). To test these hypotheses, we constructed expression vectors that direct the synthesis of GST-fusion proteins with the C 2 A-domains from synaptotagmins IV and XI either in the wild-type forms containing serine or in mutant forms containing aspartate instead of serine at the appropriate position.

Substitution of Serine to Aspartate in the C 2 -motif of the C 2 A-domain of synaptotagmins IV and XI Restores Ca 2ϩ
Binding-The GST-fusion proteins encoding the C 2 A-domains of synaptotagmins IV and XI either as wild type proteins or with serine to aspartate substitutions were isolated, the linkers between GST and the fused C 2 A-domain were cut with thrombin, and the C 2 A-domains were purified to homogeneity (data not shown). As a test of Ca 2ϩ binding, we used temperature denaturation experiments since binding of a ligand to a protein usually leads to stabilization of the protein structure. Temperature denaturation can be monitored by CD measurements at a conformation-dependent wavelength. This provides a sensitive method to analyze Ca 2ϩ binding to C 2 -domains as shown previously for the wild type and mutant C 2 A-domains from synaptotagmin I (18,19).
Wild type C 2 A-domains from synaptotagmins IV and XI exhibited no change in denaturation temperature as a function of Ca 2ϩ (Fig. 5, A and C), confirming that these C 2 -domains do not bind Ca 2ϩ at concentrations of up to 5 mM although the data do not exclude the possibility of a binding site with even lower affinity. By contrast, C 2 A-domains carrying a single amino acid substitution that exchanged serine for aspartate at the position corresponding to aspartate 230 in synaptotagmin I experience Ca 2ϩ -dependent shifts in denaturation temperature (approximately 15 and 9°C for synaptotagmins IV and XI, respectively; Fig. 5, B and D). Thus, the single amino acid substitution restores the ability to bind Ca 2ϩ to these C 2 -domains.
Ca 2ϩ -dependent Phospholipid Binding Properties of Wild Type and Mutant C 2 A-domains from Synaptotagmins IV and XI-The C 2 A-domains of most synaptotagmins but not of synaptotagmin IV bind phospholipids as a function of Ca 2ϩ , whereas Mg 2ϩ is ineffective (6,11). This agrees well with the aspartate to serine substitution in the C 2 -motif of the C 2 Adomain of synaptotagmin IV (Fig. 4). The questions now arise if synaptotagmin XI, with a similar substitution, is also unable to bind phospholipids as a function of Ca 2ϩ and if mutating the wild type serine back to aspartate would confer Ca 2ϩ -dependent phospholipid binding to these two C 2 A-domains.
Fusion proteins containing the wild type and mutant C 2 Adomains were immobilized on glutathione beads and used for phospholipid binding measurements. As shown in Fig. 6, A-C,   FIG. 4. Model of the Ca 2ϩ -binding site in the C 2 -motif of the C 2 A-domain from synaptotagmin XI as wild type sequence containing serine (A) and as mutant with the serine to aspartate substitution (B). The diagram is based on the bipartite Ca 2ϩ -binding motif modeled for the C 2 A-domain of synaptotagmin I (19); residues are identified by numbers (Fig. 1). In the wild type form (A), serine 247 is predicted to be unable to act as a bidentate ligand for the two Ca 2ϩ ions. In the serine to aspartate substitution mutant (B), aspartate 247 can contribute to bind the two Ca 2ϩ ions as in the C 2 A-domain of synaptotagmin I. An analogous model is proposed for synaptotagmin IV, with the corresponding changes in residue numbers. Dashed lines illustrate the coordination of the two Ca 2ϩ ions. The protein backbone linking aspartate residues in the same loop is represented by solid curves.
FIG. 5. Effect of Ca 2ϩ on the temperature-dependent denaturation of the wild type and mutant C 2 A-domains from synaptotagmins IV and XI. Purified recombinant wild type C 2 A-domains from synaptotagmins IV and XI (A and C, respectively) and mutant C 2 A-domains from the same synaptotagmins (B and D, respectively) were incubated with 0.5 mM EGTA in the absence of Ca 2ϩ (open circles) or in the presence of 5.5 mM Ca 2ϩ (closed circles). Unfolding of the domains as a function of temperature was monitored by CD absorbance at conformationdependent wavelength (217 nm). The mutants contain a single amino acid substitution exchanging the serine in the C 2 -motif for aspartate (see Figs. 3 and 4).
no Ca 2ϩ -dependent binding of phospholipids consisting of 29% PS and 71% PC was observed to the wild type C 2 A-domains from synaptotagmins IV and XI. In contrast, the C 2 A-domain of synaptotagmin I exhibited robust Ca 2ϩ -specific phospholipid binding. No Ca 2ϩ -independent phospholipid binding was observed as the background level of phospholipids bound corresponds to that obtained with GST alone. The mutant C 2 Adomains, however, exhibited marked Ca 2ϩ -dependent phospholipid binding that was not observed with Mg 2ϩ (Fig. 6). These data suggest that the C 2 A-domains of the synaptotagmin IV/XI subclass contain a selective, evolutionarily conserved single amino acid substitution that inactivates Ca 2ϩ binding and Ca 2ϩ -dependent phospholipid binding.
Dependence of Ca 2ϩ -dependent Phospholipid Binding on Liposome Composition-Recently, it was reported that the C 2 Adomain of synaptotagmin IV can bind 100% PS in a Ca 2ϩ -dependent manner although it does not bind to liposomes composed of mixtures of PS and PC (23). Given the structure of the Ca 2ϩ -binding site in C 2 -domains (Fig. 4), this is a surprising finding since the C 2 A-domain is not expected to bind Ca 2ϩ with the aspartate to serine substitution. However, PS itself binds Ca 2ϩ at the concentrations used, and it is possible that clusters of negative charges by PS may also allow Ca 2ϩ binding to the protein with unanticipated properties. Furthermore, stable bilayers are difficult to obtain with 100% PS, which tends to aggregate. Since the assay used by Fukuda et al. (23) utilized soluble GST-fusion proteins that were co-precipitated with lipids, the possibility arises that Ca 2ϩ -dependent PS aggregation could artifactually trap the C 2 A-domain.
To test if 100% PS induces a novel Ca 2ϩ -binding site in wild type synaptotagmin IV, we investigated the Ca 2ϩ -dependent interactions of the wild type and mutant C 2 A-domains from synaptotagmin IV with 100% PS in an assay that does not depend on lipid co-precipitation to avoid artifacts. No binding of 100% PS to the wild type C 2 A-domain was observed (Fig. 6, D and E). Even with the mutant C 2 A-domain, which exhibits robust Ca 2ϩ -dependent binding to liposomes composed of 29% PS, 71% PC, Ca 2ϩ only had an insignificant effect on binding of 100% PS. Since, in the C 2 A-domain of synaptotagmin I, only negatively charged phospholipids bind as a function of Ca 2ϩ , the absence of Ca 2ϩ -dependent binding of 100% PS to the mutant C 2 A-domain of synaptotagmin IV is paradoxical. This result can best be explained by the assumption that a stable lipid bilayer may be required. To test this hypothesis, we made liposomes from 80% PS, 20% PC, which are more likely to contain stable bilayers. Now, Ca 2ϩ -dependent phospholipid binding to the mutant C 2 A-domain was observed although it was not as strong as that observed with our standard liposomes. The wild type C 2 A-domain, however, was still unable to bind. Together, these data provide further evidence for the conclusion that the wild type C 2 A-domain of synaptotagmin IV is not a Ca 2ϩ -binding domain and suggest that, even under conditions of PS enrichment, no Ca 2ϩ -dependent properties can be demonstrated. DISCUSSION Synaptotagmins form a large family of genes with putative functions in membrane traffic (1). We now report the structure and properties of the 11th member of this family. Since it seems likely that additional synaptotagmins remain to be discovered, synaptotagmins now form one of the largest protein families in membrane traffic second only to rab proteins.
The structure of synaptotagmin XI reveals that it defines a new subclass of synaptotagmins together with synaptotagmin IV. Synaptotagmins IV and XI are highly expressed in brain, and at lower levels in other tissues ( Fig. 2; see Ref. 6). This indicates a function that is concentrated in brain but also FIG. 6. Ca 2ϩ -dependent phospholipid binding to wild type and mutant C 2 A-domains from synaptotagmins I, IV, and XI. GST alone and the C 2 A-domain of synaptotagmin I (GST-Syt I) were used as negative and positive controls, respectively (A). C 2 A-domain fusion proteins for wild type and mutant synaptotagmins IV and XI (B-E; GST-Syt IV and GST-Syt XI; superscript identifies mutation) were studied in standard liposome binding assays (13). Experiments in A-C were performed with liposomes composed of 29% PS, 71% PC; in D with 80% PS, 20% PC liposomes; and in E with 100% PS sonicated lipid suspensions. GST-fusion proteins were immobilized on glutathione beads and incubated with 3 H-labeled liposomes in the presence of 1 mM EGTA with or without additions of divalent cations (2 mM Mg 2ϩ or Ca 2ϩ in A-C and 1.12 mM in D and E to reproduce the concentrations in Fukuda, et al. (23). Open bars, EGTA; cross-hatched bars, Mg 2ϩ ; solid bars, Ca 2ϩ . operative in nonneural tissues. Synaptotagmins I and II probably serve as Ca 2ϩ sensors in neurotransmitter release (10,11), suggesting that the other synaptotagmins also represent membrane trafficking proteins at the plasma membrane, possibly in regulating different forms of exocytosis. In addition, synaptotagmins may function in endocytosis since all synaptotagmins tested are high affinity binding proteins for the clathrin adaptor protein complex, AP2 (6,11,22).
The subgroup of synaptotagmins composed of synaptotagmins IV and XI is characterized by a homologous N-terminal region and by identical deviations from the C 2 -domain consensus sequence shared by most synaptotagmins (Figs. 1 and 3). The most remarkable change in the C 2 -domains of synaptotagmins IV and XI is the substitution of one of the aspartates of the Ca 2ϩ -binding site for a serine. This raises the question if a C 2 A-domain with this substitution is capable of Ca 2ϩ binding. We show by two assays, the Ca 2ϩ -dependent shift in the heat denaturation curve and Ca 2ϩ -dependent phospholipid binding, that Ca 2ϩ has no effect on the C 2 A-domains from both synaptotagmins at concentrations up to 5 mM. A point mutation that reverses the evolutionary serine to aspartate substitution, however, restores both Ca 2ϩ -dependent properties to both synaptotagmins. This result indicates that this amino acid change is the only change in the C 2 A-domain that made the C 2 Adomains from synaptotagmins IV and XI Ca 2ϩ -independent. Thus, this subgroup of synaptotagmins is characterized by an evolutionarily conserved substitution in the C 2 A-domain that abolishes Ca 2ϩ binding.
The most amazing aspect of our findings is that evolution seems to have selected for the inactivation of Ca 2ϩ -binding in two different synaptotagmins by a single amino acid substitution while leaving the remaining structure intact. Based on their overall sequence similarity, it seems likely that synaptotagmins IV and XI share a common ancestor derived from the original "ur-synaptotagmin". Further evolution then created the diversification into synaptotagmins IV and XI with many amino acid changes, including several in their C 2 A-domains. Nevertheless, these changes did not abolish the ability of these synaptotagmins to bind Ca 2ϩ and to exhibit Ca 2ϩ -dependent phospholipid binding if only the serine in the Ca 2ϩ -binding site was rechanged into aspartate.
The structure of the C 2 -domain consists of a stable core composed of two ␤-sheets with three loops emerging on top and four on the bottom (16). Ca 2ϩ binding occurs only to the top and induces no electrostatic or conformational change in the bottom (18,19). It is possible that C 2 -domains serve as janus-faced interaction domains, a Ca 2ϩ -dependent top and a Ca 2ϩ -independent bottom (1). This would imply that in the synaptotagmin IV/XI subgroup, evolutionary pressure led to a selective retention of only the Ca 2ϩ -independent properties. Unfortunately no Ca 2ϩ -independent activities of a C 2 A-domain have been identified yet. Once these have been discovered, however, it will be important to test them in different synaptotagmins to determine if they are selectively retained in the Ca 2ϩ -independent forms.
In a very interesting study, synaptotagmin IV was identified as an immediate early gene (26). This suggests that a switch from Ca 2ϩ -dependent to Ca 2ϩ -independent synaptotagmins may occur during strong stimulation of neurons. Such a switch could be particularly useful during pathological hyperexcitation that is accompanied by unimpeded Ca 2ϩ influx. A switch to a Ca 2ϩ -unresponsive synaptotagmin under those conditions would eliminate a Ca 2ϩ target and maybe inhibit excessive neurotransmitter release. Future studies will have to test this hypothesis and also investigate if synaptotagmin XI is also an immediate early gene.