Structure and Ca²⁺-binding Properties of the Tandem C₂ domains of E-Syt2

SUMMARY

Contacts between the endoplasmic reticulum and the plasma membrane involve extended synaptotagmins (E-Syts) in mammals or tricalbins in yeast, proteins with multiple C₂ domains. One of the tandem C₂ domains of E-Syt2 is predicted to bind Ca²⁺, but no Ca²⁺-dependent function has been attributed to this protein. We have determined the crystal structures of the tandem C₂ domains of E-Syt2 in the absence and presence of Ca²⁺, and analyzed their Ca²⁺-binding properties by NMR spectroscopy. Our data reveal an unexpected V-shaped structure with a rigid orientation between the two C₂ domains that is not substantially altered by Ca²⁺. The E-Syt2 C₂A domain binds up to four Ca²⁺ ions, whereas the C₂B domain does not bind Ca²⁺. These results suggest that E-Syt2 performs an as yet unidentified Ca²⁺-dependent function through its C²A domain, and uncover fundamental differences between the properties of the tandem C₂ domains of E-Syts and synaptotagmins.

INTRODUCTION

Membrane contact sites (MCSs) are areas where the membranes from two distinct organelles come into close apposition, facilitating the transfer of signals and molecules between them. Increasing evidence suggests that these sites play important roles in cell physiology (Levine and Loewen, 2006; Toulmay and Prinz, 2011; Elbaz and Schuldiner, 2011). Particularly prominent are contacts between the endoplasmic reticulum (ER) and the plasma membrane, which appear to be a general feature of eukaryotic cells and have been implicated in regulation of cell signaling, metabolism, ER architecture and plasma membrane domain organization (Carrasco and Meyer, 2011; Stefan et al., 2013). However, the molecular machineries that form MCSs and the principles underlying their functions are just beginning to be uncovered.

In yeast, three ER proteins known as tricalbins (Tcb1p, Tcb2p and Tcb3p) were recently shown to play key functions in ER-plasma membrane tethering (Manford et al., 2012). In mammalian cells, such tethering was found to also depend on three proteins called E-Syts [for extended synaptotagmins; (Min et al., 2007)] that are related to tricalbins (Giordano et al., 2013). The term E-Syts reflects their similarity to synaptotagmins (Min et al., 2007), a family of proteins that are involved in Ca²⁺-regulated secretion and are characterized by an N-terminal transmembrane (TM) sequence and two C-terminal C₂ domains (C₂A and C₂B) (Rizo and Rosenmund, 2008). Tricalbins and E-Syts share similar domain structures comprising multiple C₂ domains (5 for E-Syt1, and 3 for E-Syt2 and E-Syt3) and a synaptotagmin-like-mitochondrial-lipid binding protein (SMP) domain that has been implicated in targeting to MCSs (Creutz et al., 2004; Min et al., 2007; Toulmay and Prinz, 2012) (Figure 1A for E-Syt2).

Figure 1.

Crystal structure of the Ca²⁺-free E-Syt2 C₂AB fragment. (A) Domain structure of E-Syt2. The black box represents a hydrophobic region involved in membrane anchoring (Giordano et al., 2013). SMP = synaptotagmin-like-mitochondrial-lipid binding protein domain. (B) Ribbon diagram of the Ca²⁺-free E-Syt2 C₂AB fragment. The C₂A domain is red, the C₂B domain cyan and the linker between them yellow. The loops at the top, which mediate Ca²⁺ binding in some C₂ domains, are labeled loop 1–4. N and C indicate the N- and C-termini, respectively. (C) Superposition of ribbon diagrams of the E-Syt2 C₂A and C₂B domains. (D) Superpositions of ribbon diagrams of the C₂A (red) and C₂B domains (cyan) of E-Syt2 with the C₂ domain of intersectin 2 (wheat) and the Munc13–1 C₂B domain (pale yellow) (PDB accession codes 3JZY and 3KWT, respectively). See also Figure S1.

C₂ domains are widespread modules that commonly bind Ca²⁺ and phospholipids (Rizo and Sudhof, 1998) but can also be Ca²⁺-independent and function in protein-protein interactions [e.g. (Lu et al., 2006)]. C₂ domains occur in tandem in proteins involved in membrane traffic such as synaptotagmins and also in E-Syts (C₂A-C₂B in E-Syt2 and E-Syt3; C₂A-C₂B and C₂C-C₂D in E-Syt1). The synaptotagmin-1 C₂ domains were shown to form β-sandwich structures that bind two or three Ca²⁺ ions via five aspartate side chains from loops located at the top (Sutton et al., 1995; Shao et al., 1996; Ubach et al., 1998; Fernandez et al., 2001). These loops also mediate Ca²⁺-dependent phospholipid binding (Zhang et al., 1998; Chapman and Davis, 1998). Among E-Syt C₂ domains, the entire set of five aspartate residues is conserved only in the C₂A domains of E-Syt2 and E-Syt3, and in the C₂A and C₂C domains of E-Syt1 (both homologous to the E-Syt2 and E-Syt3 C₂A domains), suggesting that these C₂ domains but not others bind Ca²⁺ (Min et al., 2007). However, localization of E-Syt2 to the plasma membrane required its C₂C domain (Min et al., 2007), and studies of the function of E-Syts in ER-plasma membrane tethering assigned a Ca²⁺-dependent role only for E-Syt1 (Giordano et al., 2013). In yeast, no Ca²⁺-dependent function in ER-plasma membrane tethering was attributed to tricalbins (Manford et al., 2012), although the C₂D domains of Tcb1p and Tcb3p were reported to bind Ca²⁺ while other tricalbin C₂ domains contain substitutions in some of the canonical aspartate residues (Schulz and Creutz, 2004). However, predictions of C₂ domains properties from sequence analyses need to be interpreted with caution, as distinctive sequences inserted between β-strands often yield unexpected structural features (Ubach et al., 1999; Garcia et al., 2004; Shin et al., 2010), and the presence of a complete or incomplete set of five canonical aspartate residues does not guarantee Ca²⁺ binding or lack of this activity, respectively (Dai et al., 2004).

While important advances have recently been made in establishing the importance of E-Syts and tricalbins for ER-plasma membrane tethering, much remains to be learned about how the structural and biochemical properties of the C₂ domains from these proteins underlie their functions. Particularly important is to unambiguously establish whether the tandem C₂ domain-regions of E-Syts bind Ca²⁺ and contain unique structural features that might be important for their functions. It is also of interest to investigate the similarity between the tandem C₂ domains of E-Syts and those of synaptotagmins, which have been extensively studied [e.g. (Sutton et al., 1999; Fuson et al., 2007; Arac et al., 2006; Herrick et al., 2009; Vrljic et al., 2010; Shin et al., 2009)].

To address these questions, we have analyzed the Ca²⁺-binding properties of a fragment spanning the C₂A and C₂B domains of E-Syt2 using NMR spectroscopy, and we have determined its crystal structure in the absence and presence of Ca²⁺. Our data reveal a V-shaped structure with a rigid orientation of the two C₂ domains, in contrast to the flexible arrangement of the synaptotagmin C₂ domains. The E-Syt2 C₂B domain does not bind Ca²⁺ but the C₂A domain binds up to four Ca²⁺ ions with a wide range of affinities, including a novel Ca²⁺-binding site that arises from the highly acidic nature of one of the Ca²⁺-binding loops. These results show that the C₂A domain of E-Syt2 shares common features with the synaptotagmin C₂A domains but suggest that, overall, the tandem C₂ domains of E-Syt2 function by markedly different mechanisms from those characteristic of synaptotagmins. Moreover, our data indicate that the C₂A domains of E-Syts perform a general Ca²⁺-dependent role at ER-plasma membrane contact sites, suggesting that their general function may be dynamically regulated by cell-signaling.

RESULTS

Structure of the Ca²⁺-free tandem C₂ domains of E-Syt2

C₂ domains can contain extensions in their termini that are critical for proper folding or optimal solubility [e.g. (Guan et al., 2007; Shin et al., 2010)]. To determine the structure of the E-Syt2 tandem C₂ domains, we first tried using a fragment containing the C₂A-C₂B domains of mouse E-Syt2 with additional N-terminal sequences (residues 283–584), but could not get suitable crystals. We then turned to a shorter fragment spanning the C₂A-C₂B domains of human E-Syt2 (residues 363–660; below referred to as E-Syt2 C₂AB). Crystallization screens and optimizations performed with Ca²⁺-free protein yielded crystals of E-Syt2 C₂AB that had the symmetry of space group P4₃ and diffracted to 2.1 Å. We were unable to solve the structure by molecular replacement, but we obtained selenomethionine-derivatized crystals and solved the structure of the Ca²⁺-free E-Syt2 C₂AB fragment using phases from single-wavelength anomalous dispersion data. Structural statistics are described in Table 1.

Table 1.

Data collection, phasing and refinement statistics for E-Syt2 C₂AB fragment structures.

Data collection
Crystal	ESyt	ESyt-Ca2+	ESyt Seapeak
Space group	P43	C2	P43
Energy (eV)	8961.6	12,684.9	12,684.9
Resolution range (Å)	41.9 – 2.10 (2.14 – 2.10)	30.0 – 2.55 (2.59 – 2.55)	43.3 – 2.33 (2.37 – 2.33)
Unique reflections	48,539 (1,830)	12,318 (263)	36,177 (1,608)
Multiplicity	5.5 (3.2)	3.6 (2.4)	4.4 (2.9)
Data completeness (%)	97.0 (73.8)	88.7 (38.0)	98.9 (87.2)
Rmerge (%)b	6.7 (50.1)	5.8 (27.6)	8.0 (44.6)
I/σ(I)	23.4 (1.46)	33.4 (2.45)	23.5 (1.38)
Wilson B-value (Å2)	40.3	66.6	52.4
Phase determination
Anomalous scatterers	selenium, 10 out of 10 possible sites
Figure of merit (43.3 – 2.33 Å)	0.16
Refinement statistics
Crystal	Native	ESyt-Ca2+
Resolution range (Å)	41.9 – 2.10 (2.14 – 2.10)	29.0 – 2.55 (2.81 – 2.55)
No. of reflections RWork/Rfree	48,422/2.468 (1,979/108)	12,224/604 (1,782/93)
Data completeness (%)	96.9 (73.8)	87.4 (54.0)
Atoms (non-H protein/solvent/metals)	4,650/146/0	2,178/4/3
Rwork (%)	20.7 (29.5)	22.2 (31.5)
Rfree (%)	24.7 (34.1)	28.7 (41.4)
R.m.s.d. bond length (Å)	0.012	0.012
R.m.s.d. bond angle (°)	1.32	1.26
Mean B-value (Å2) (protein/solvent/metals)	58.3/46.1/NA	121.7/105.8/126.38
Ramachandran plot (%) (favored/additional/disallowed)c	95.7/4.1/0.2	94.7/3.8/1.5
Maximum likelihood coordinate error	0.31	0.45
Missing residues	A:552–556, 660 B: 363, 403–409, 550–556, 660	363–379, 526–531, 551–555, 660

Data for the outermost shell are given in parentheses.

^aBijvoet-pairs were kept separate for data processing.

^bR_merge =100 Σ_hΣ_i|I_{h, i} — 〈I_h〉|/Σ_hΣ_iI_{h, i}, where the outer sum (h) is over the unique reflections and the inner sum (i) is over the set of independent observations of each unique reflection.

^cAs defined by the validation suite MolProbity (Chen et al, 2010).

The asymmetric unit contains two molecules of E-Syt2 C₂AB that have very similar structures [0.4 Å root mean square (rms) deviation between 3,362 equivalent atoms] (Figure S1A available online). The N-terminal sequence spanning residues 363–383 forms an extended structure that is not part of the C₂A domain (Figures S1A,B) and is involved in extensive crystal contacts. As expected, the E-Syt2 C₂A and C₂B domains form β-sandwich structures with two four-stranded β-sheets (Figure 1B). The C₂A domain spans residues 384–509 while residues 535–656 form the C₂B domain (red and cyan, respectively, in Figure 1B). Both C₂ domains have a type II topology where the N- and C-termini are located at the bottom of the β-sandwich, opposite to the top loops that mediate Ca²⁺ binding in C₂ domains (Figure 1C) [see (Rizo and Sudhof, 1998)]. This topology is analogous to that found originally in the PLC-δ1 C₂ domain (Essen et al., 1996) and differs from the type I topology of the synaptotagmin C₂ domains, where the N- and C-termini are located at the top of the β-sandwich (Sutton et al., 1995). As a consequence, the top of the E-Syt2 C₂A and C₂B domains contain four loops (labeled loop 1–4 in Figures 1B,C), while the bottom side has three loops, one of which forms a short α-helix in both domains. The linker sequence joining the C₂A and C₂B domains (residues 510–534; yellow in Figures 1B and S1C) forms an α-helix and two loops that pack against both domains, conferring a V-shape to the overall structure.

The structures of the E-Syt2 C₂A and C₂B domains are very similar, with the main differences observed in the loops joining the β-strands (Figure 1C). Structural superposition using DaliLite (Holm and Park, 2000) yielded an rms deviation of 1.5 Å for 114 equivalent Cα carbons. Searches of the Protein Data Bank (PDB) using DALI (Holm and Sander, 1993) revealed that the known C₂ domain structures that are most similar to both the C₂A and the C₂B domain of E-Syt2 correspond to the C₂ domain of intersectin 2 and the C₂B domain of Munc13-1, both of which also have topology II (Figure 1D). Pairwise comparisons of the intersectin 2 C₂ domain and the Munc13-1 C₂B domain with the E-Syt2 C₂A and C₂B domains yielded Z scores ranging from 16.6 to 19.1 and 1.3–1.5 rms deviations for 116–124 equivalent Cα carbons. A noteworthy feature of the E-Syt2 C₂A and C₂B domains is that the sequence corresponding to loop 1 is considerably longer than usual. This is more apparent in the structure of the E-Syt2 C₂A domain, where loop 1 protrudes at the top of the domain more markedly than in the intersectin 2 C₂ domain and the Munc13-1 C₂B domain (Figure 1D). The sequence of loop 1 is highly conserved evolutionarily, suggesting that it is functionally significant. No electron density is observed for part of the loop 1 sequence of the C₂A domain for the other molecule in the asymmetric unit and for part of the loop 1 sequence of the E-Syt2 C₂B domain in both molecules of the asymmetric unit (Figures 1B and S1A), suggesting that loop 1 is flexible in both domains. It is also worth noting that the β-sandwiches of the E-Syt2 C₂ domains are also highly similar to many C₂ domains with topology I such as those of synaptotagmin-1, but the two α-helices characteristic of the synaptotagmin-1 C₂B domain are not observed in the E-Syt2 C₂B domain (Figure S1D,E).

Ca²⁺-binding to the E-Syt2 C₂A domain

Because of difficulties in expressing the isolated E-Syt2 C₂A and C₂B domains in soluble form, we used the E-Syt2 C₂AB fragment to study its Ca²⁺-binding properties. We first used temperature denaturation monitored by circular dichroism (CD), which provided a sensitive tool to test for Ca²⁺ binding to the synaptotagmin-1 C₂A domain (Shao et al., 1997). Ca²⁺ caused a dramatic decrease in the melting temperature of the wild type (WT) E-Syt2 C₂AB fragment, which is unusual in C₂ domains (see discussion), but shows that this fragment indeed binds Ca²⁺ (Figure 2A). Practically no shift in melting temperature was observed for a mutant E-Syt2 C₂AB fragment where two of the canonical aspartate residues predicted to form the Ca²⁺-binding sites of the C₂A domain (D401 and D413) were replaced by alanines (Figure 2B), showing that, as expected, Ca²⁺ binding to the WT E-Syt2 C₂AB fragment is mediated by the C₂A domain.

Figure 2.

The E-Syt2 C₂AB fragment binds Ca²⁺. (A,B) Thermal denaturation curves monitored by the CD absorption at 215 nM of the WT (A) and D401A,D413A mutant (B) E-Syt2 C₂AB fragment in the presence of 0.1 mM EGTA (closed circles) or 1 mM Ca²⁺ (open circles).

To analyze the Ca²⁺-binding properties in detail, we used transverse relaxation optimized spectroscopy (TROSY)-enhanced ¹H-¹⁵N heteronuclear single quantum coherence (HSQC) spectra. In these NMR spectra, Ca²⁺ binding with slow exchange rates characteristic of high affinity sites leads to gradual disappearance of the NH cross-peaks corresponding to the Ca²⁺ free state and concomitant appearance of cross-peaks from the Ca²⁺ bound state, whereas Ca²⁺ binding with fast exchange rates characteristic of low affinity sites leads to gradual movement of cross-peaks from their positions in the Ca²⁺-free state to those of the Ca²⁺-bound state (Rizo et al., 2012). ¹H-¹⁵N TROSY-HSQC spectra of 0.16 mM ¹⁵N-labeled E-Syt2 C₂AB fragment exhibited multiple changes upon addition of 1 mM Ca²⁺ (Figure 3A) and additional changes for a few cross-peaks at 30 mM Ca²⁺ (Figure 3B), suggesting the presence of at least two Ca²⁺-binding sites with different affinities. Indeed, Ca²⁺ titrations showed clearly distinct behaviors in different cross-peaks that define up to four Ca²⁺-binding sites.

Figure 3.

The E-Syt2 C₂AB fragments contains up to four Ca²⁺-binding sites. (A,B) ¹H-¹⁵N TROSY-HSQC spectra of 0.15 M E-Syt2 C₂AB fragment acquired in the absence of Ca²⁺ (black contours) or in the presence of 1 mM Ca²⁺ (red contours) or 30 mM Ca²⁺ (blue contours). (C–G) Expansions of different regions of ¹H-¹⁵N TROSY-HSQC spectra of the E-Syt2 C₂AB fragment (blue squares in A, B) showing distinct Ca²⁺-dependent behaviors. The numbers indicate the Ca²⁺ concentrations used (in mM units) and are color-coded as the cross-peaks from the corresponding spectrum. In panels C and D, the 1D traces shown above the contour plots were taken along the ¹⁵N chemical shifts indicated by the arrows and have the same color codes.

Titration from 0 to 0.15 mM Ca²⁺ led to complete disappearance of a few cross-peaks (Figure 3C) and full appearance of a few new cross-peaks (Figure 3D). This behavior corresponds to a stoichiometric saturation-binding curve characteristic of a Ca²⁺-binding site with relatively high affinity (K_D < 10 μM; we refer to this site as Ca1). Several other cross-peaks moved gradually from 0 to ca. 0.8 mM Ca²⁺ and exhibited strong broadening at intermediate Ca²⁺ concentrations (Figure 3E). This behavior must arise from the presence of a second Ca²⁺ binding site (Ca2) with a lower affinity than Ca1, but the strong broadening suggests that these cross-peaks are also affected by binding to Ca1. This is not unexpected, since the two most common Ca²⁺-binding sites in C₂ domains are very close to each other (Shao et al., 1996). This feature hinders accurate measurement of the affinity of site Ca2 but we estimate the K_D to be in the 100–200 μM range based on the mid points observed for cross-peaks with this behavior.

We also observed the appearance of a few new cross-peaks at 2 mM Ca²⁺, and higher Ca²⁺ concentrations induced movements of these cross-peaks (Figures 3F,G). For one of the cross-peaks (Figure 3F), the Ca²⁺-dependent movement was clearly curved and therefore included at least two components arising from two distinct Ca²⁺-binding sites (Ca3 and Ca4). The two components and the fact that the positions of the cross-peaks below 2 mM Ca²⁺ were unknown hindered accurate measurement of the corresponding affinities, but we estimate that the K_Ds of these sites are 2–3 mM and 15–20 mM, respectively. For comparison, the K_Ds of the three Ca²⁺-binding sites of the synaptotagmin-1 C₂A domain are 60 μM, 0.5 μM and > 2 mM (Ubach et al., 1998; Fernandez-Chacon et al., 2001). Note that phospholipids dramatically enhance these affinities because their headgroups fill empty coordination sites of the Ca²⁺ ions bound to the synaptotagmin-1 C₂A domain (Zhang et al., 1998; Fernandez-Chacon et al., 2001). Correspondingly, Ca²⁺-dependent phospholipid binding to a fragment spanning the SMP, C₂A and C₂B domains of E-Syt2 (SMP-C₂AB fragment) occurs with an apparent Ca²⁺ affinity in the low micromolar range, although it is unclear to what extent the SMP domain might contribute to this apparent affinity (Min et al., 2007).

Structure of the Ca²⁺-bound E-Syt2 C₂AB fragment

To define the Ca²⁺-binding mode of the E-Syt2 C₂AB fragment, we performed crystallization screens at different Ca²⁺ concentrations. We obtained crystals in 2 mM Ca²⁺ that diffracted to 2.55 Å and had the symmetry of space group C2, with one molecule per asymmetric unit. The structure was solved by molecular replacement using the Ca²⁺-free structure as search model (Table 1).

The structure of the Ca²⁺-bound E-Syt2 C₂AB fragment also exhibits a V-shape formed by two β-sandwiches and reveals three Ca²⁺ ions bound to the C₂A domain (Figure 4A). Superposition of the Ca²⁺-bound and Ca²⁺-free structures showed that the structures are very similar (Figure 4B). The rms deviations between the Ca²⁺-bound structure and the two structures in the Ca²⁺-free asymmetric unit are 1.0 Å and 1.1 Å for 3,656 and 3,525 equivalent atoms, respectively. Because the crystals of the Ca²⁺-free and Ca²⁺-bound forms were obtained under very different conditions and exhibited symmetries from different space groups, the similarity in the relative orientations of the C₂A and C₂B domain observed in the Ca²⁺-free and Ca²⁺-bound structures strongly suggests that this orientation is rigid and is not affected substantially by Ca²⁺ binding. The most overt differences between the Ca²⁺-free and Ca²⁺-bound structures of the E-Syt2 C₂AB fragment are observed in two regions. One region corresponds to the N-terminal sequence, which is disordered in the Ca²⁺-bound structure but is ordered in the Ca²⁺-free structures. The second region is in the long loop 1 of the C₂A domain, which forms a β-hairpin upon Ca²⁺ binding. It is unclear to what extent the structural differences observed in this loop occur in solution, since the loop makes different types of crystal contacts in the Ca²⁺-bound structure and in one of the Ca²⁺-free structures, whereas this loop is partially disordered in the other Ca²⁺-free structure (Figure S1A). However, formation of the β-hairpin appears to be necessary to place some of the Ca²⁺ ligands in the proper orientation for Ca²⁺ binding (Figure S2). Hence, it seems likely that, in solution, loop 1 of the C₂A domain is flexible in the absence of Ca²⁺ and forms the β-hairpin upon Ca²⁺ binding.

Figure 4.

Structure of the Ca²⁺-bound E-Syt2 C₂AB fragment. (A) Ribbon diagram of Ca²⁺-bound E-Syt2 C₂AB. The three Ca²⁺ ions are shown as yellow spheres. The loops at the top region of the β-sandwiches are labeled loop 1–4. N and C indicate the N- and C-termini, respectively. (B) Superposition of the Ca²⁺-free (orange ribbon) and Ca²⁺-bound (blue ribbon with yellow spheres) E-Syt2 C₂AB fragment. The position of loop 1 of the C₂A domain is indicated. (C) Surface representations illustrating the electrostatic potentials of the Ca²⁺-free and Ca²⁺-bound E-Syt2 C₂AB fragment. The electrostatic potentials were calculated with GRASP (Nicholls et al., 1991). The gradients of electrostatic potential shown ranged from ≥ −7 k_BT/e (red) to ≤ 7 k_BT/e (blue). See also Figure S2.

Since a Ca²⁺-induced switch in the electrostatic potential of the synaptotagmin-1 C₂ domains is believed to be important for synaptotagmin-1 function (Shao et al., 1997; Arac et al., 2006; Dai et al., 2007), we analyzed the changes caused by Ca²⁺ binding on the electrostatic properties of the E-Syt2 C₂AB fragment. The electrostatic potential of the Ca²⁺-free C₂A domain is highly acidic and becomes somewhat positive around the Ca²⁺-binding region upon Ca²⁺ binding (Figure 4C), but to a lesser extent than that observed for the synaptotagmin-1 C₂ domains (Fernandez et al., 1998; Fernandez et al., 2001). This finding arises because of the large excess of acidic residues (26) over basic residues (16) in the E-Syt2 C₂A domain, and likely underlies the fact that the E-Syt2 SMP-C₂AB fragment binds similarly to neutral or acidic phospholipids (Min et al., 2007) whereas the synaptotagmin-1 C₂ domains bind preferentially to negatively charged phospholipids (Davletov and Sudhof, 1993; Fernandez et al., 2001). An even more marked distinction is found for the E-Syt2 and synaptotagmin-1 C₂B domains, as the E-Syt2 C₂B domain is zwitterionic and does not bind Ca²⁺ (Figure 4A,C) while the synaptotagmin-1 C₂B domain is highly basic and becomes even more positively charged upon Ca²⁺ binding (Fernandez et al., 2001).

The Ca²⁺-binding mode of the E-Syt2 C₂ A domain

Two of the three Ca²⁺ ions bound to the C₂A domain in the crystal structure of the E-Syt2 C₂AB fragment are located in a cup-shaped cavity formed by loops 1, 2 and 3 at the top of the β-sandwich, which is common in C₂ domains (Rizo and Sudhof, 1998), whereas the other Ca²⁺ ion binds at a site that is formed by loop 3 outside the cavity (Figure 5A). The two sites within the cup-shaped cavity are the most common in C₂ domains (Rizo and Sudhof, 1998) and are found in both synaptotagmin-1 C₂ domains. We refer to these two sites as Ca1 and Ca2. Site Ca1 normally has the highest affinity and is critical to position several of the Ca²⁺-binding ligands in the correct orientations for binding to site Ca2, which in turn has the second highest affinity (Ubach et al., 1998; Fernandez et al., 2001). Hence, it is natural to assume that sites Ca1 and Ca2 correspond to the two sites with the highest affinity observed by NMR spectroscopy (Figure 3). The five canonical aspartate residues are D401, D413, D460, D462 and D467. Site Ca1 is coordinated by seven ligands, including the backbone carbonyl of E461, two oxygens from the D401 side chain, one oxygen each from the D413, D460 and D462 side chains, and on ordered water molecule (Figure 5B). Seven ligands also coordinate site Ca2: the backbone carbonyl of K400, two oxygens from the D462 side chain, one oxygen each from the D401, D460 and D467 side chains, and an ordered water molecule.

Figure 5.

Ca²⁺-binding mode of the C₂A domain in the crystal structure of E-Syt2 C₂AB fragment. (A) Ribbon diagram of the C₂A domain showing the three bound Ca²⁺ ions. The top loops are also labeled. N and C indicate the N- and C-termini, respectively. (B) The Ca²⁺-binding sites of the E-Syt2 C₂A domain. The backbone is represented by blue a ribbon diagram, the Ca²⁺ ligands are labeled and are shown as stick models with carbon atoms colored in cyan, oxygen atoms in red and nitrogen atoms in blue. The bound Ca²⁺ ions are shown as yellow spheres. Sites Ca1, Ca2 and Ca3 are labeled 1, 2 and 3, respectively. Water molecules involved in Ca²⁺ coordination are shown as small red spheres and labeled w1–w4. Black lines indicate the coordination of Ca²⁺ ions by their ligands except when the ligands appear to be in contact with the ions because of the two-dimensional nature of the drawing. See also Figures S2 and S3.

The third Ca²⁺-binding site in the E-Syt2 C₂A domain is formed by only five ligands: the backbone carbonyls of D462 and D464, one oxygen from the D466 side chain, and two ordered water molecules (Figure 5B; see electron density map in Figure S3A). Hence, it is not surprising that this Ca²⁺-binding site has lower affinity than sites Ca1 and Ca2, and can be attributed to one of the two low-affinity Ca²⁺-binding sites detected in our NMR data (Figure 3). We thus tentatively assign this site to that with an estimated K_D of 2–3 mM (site Ca3). It is noteworthy that, while were able to grow crystals of the E-Syt2 C₂AB fragment in 2 mM Ca²⁺, we barely obtained crystals under similar conditions in the presence of 5 mM Ca²⁺, perhaps because binding to the fourth site detected by NMR spectroscopy (Ca4) perturbs the electrostatic balance of the crystal lattice. Since the estimated K_D of this site is 15–20 mM, it is not surprising that binding to this site is not observed in our crystal structure. However, we note that the backbone carbonyl of K465, two oxygens from the D467 side chain and one oxygen from the D462 side chain appear to be poised to form another Ca²⁺-binding site that would be analogous to the third site found in the synaptotagmin-1 C₂A domain (Ubach et al., 1998) and the synaptotagmin-7 C₂B domain (Xue et al., 2010), and might also be coordinated by D464 (Figure S3). Hence, we speculate that this putative site corresponds to Ca4. Overall, the abundance of Ca²⁺-binding sites in the E-Syt2 C₂A domain can be attributed to the highly acidic nature of the sequence that forms loop 3 (DEDPDKDD).

To test the Ca²⁺-binding site assignments, we performed Ca²⁺ titrations monitored by ¹H-¹⁵N TROSY-HSQC spectra with mutant E-Syt2 C₂AB fragments containing either a D413N or a D466A substitution (Figures 6,7). In the absence of Ca²⁺, both mutations caused shifts in only a few cross-peaks compared to WT E-Syt2 C₂AB (Figures 6A,7A), showing that the mutations did not disrupt the proper folding of the protein. The D413N mutation in E-Syt2 is analogous to the D178N and D309N mutations that disrupt Ca²⁺ binding to site Ca1 of the synaptotagmin-1 C₂A and C₂B domain, respectively, and also impair Ca²⁺ binding to other sites because of the importance of filling site Ca1 to organize the overall Ca²⁺-binding mode (Ubach et al., 1998; Fernandez et al., 2001). Correspondingly, 1 mM Ca²⁺ caused very little perturbations in the ¹H-¹⁵N TROSY-HSQC spectrum of the E-Syt2 C₂AB D413N mutant (Figure 6B), showing that the mutation strongly disrupts overall Ca²⁺ binding. A few cross-peak changes were observed at Ca²⁺ concentrations above 1 mM (Figures 6C,D), which can be attributed to residual binding to sites Ca1 or Ca2, and/or to lack of perturbation of sites Ca3 and Ca4. These results strongly support the proposal that sites Ca1 and Ca2 in the crystal structure constitute the two Ca²⁺-binding sites of the E-Syt2 C₂A domain with highest affinity observed in our NMR data.

Figure 6.

The D413N mutation strongly disrupts Ca²⁺ binding to the E-Syt2 C₂A domain. (A)¹H-¹⁵N TROSY-HSQC spectra of WT (black contours) and D413N mutant (cyan contours) E-Syt2 C₂AB in the absence of Ca²⁺. (B)¹H-¹⁵N TROSY-HSQC spectra of D413N mutant E-Syt2 C₂AB in the absence of Ca²⁺ (cyan contours) and in the presence of 1 mM Ca²⁺ (red contours). (C,D) Expansions of different regions of ¹H-¹⁵N TROSY-HSQC spectra of the D413N mutant E-Syt2 C₂AB fragment (indicated by blue squares in panel B) showing distinct Ca²⁺-dependent behaviors. The numbers indicate the Ca²⁺ concentrations used (in mM units) and are color-coded as the cross-peaks from the corresponding spectrum.

Figure 7.

The D466A mutation disrupts one of the low-affinity Ca²⁺ binding sites of the E-Syt2 C A domain. (A) ¹H-¹⁵N TROSY-HSQC spectra of WT (black contours) and D466A mutant (green contours) E-Syt2 C₂AB in the absence of Ca²⁺. (B)¹H-¹⁵N TROSY-HSQC spectra of D466A mutant E-Syt2 C₂AB in the absence of Ca²⁺ (green contours) and in the presence of 1 mM Ca²⁺ (red contours). (C–E) Expansions of different regions of ¹H-¹⁵N TROSY-HSQC spectra of the D466A mutant E-Syt2 C₂AB fragment (indicated by blue squares in panel B) showing distinct Ca²⁺-dependent behaviors. The numbers indicate the Ca²⁺ concentrations used (in mM units) and are color-coded as the cross-peaks from the corresponding spectrum. In panel C, the 1D traces shown above the contour plots were taken along the ¹⁵N chemical shifts indicated by the arrow and have the same color codes.

The D466A mutation was designed to disrupt binding to site Ca3. As expected, 1 mM Ca²⁺ caused similar changes in the ¹H-¹⁵N TROSY-HSQC spectra of this mutant (Figures 7B) and WT E-Syt2 C₂AB (Figures 3A). A few cross-peaks of the D466A mutant disappeared almost completely at 0.15 mM Ca²⁺ and completely at 0.3 mM Ca²⁺ (Figure 7C), resembling the behavior of cross-peaks that manifest binding to site Ca1 of WT E-Syt2C₂AB (Figure 3C). Other cross-peaks of the D466A mutant exhibited gradual movements from 0 to 1 mM Ca²⁺, and strong broadening around 0.4 mM Ca²⁺ concentrations (Figure 7D), similar to cross-peaks that reflect binding to sites Ca1 and Ca2 of WT E-Syt2 C₂AB (Figure 3E). Thus, the D466A mutation causes little perturbation of sites Ca1 and Ca2, although their affinities appear to be somewhat lower in the mutant, perhaps because the mutation decreases the overall negative charge in the region. Interestingly, the cross-peak that appeared at the top of the spectra of WT E-Syt2 C₂AB at 2 mM Ca²⁺ (Figure 3G) appeared only at 5 mM Ca²⁺ for the D466A mutant and exhibited much smaller changes than the WT protein at higher Ca²⁺ concentrations (Figure 7E). These results strongly support the notion that site Ca3 identified by NMR spectroscopy in the WT E-Syt2 C₂AB corresponds to the third Ca²⁺-binding observed in the crystal structure.

DISCUSSION

Investigation of MCSs in general and of ER-plasma membrane contacts in particular has become a subject of intense biological interest because of the identification of diverse functions that occur at these sites and the realization that they may underlie a wide variety of cellular processes. Research in yeast (Manford et al., 2012) and mammalian cells (Giordano et al., 2013) has demonstrated the importance of E-Syts and their yeast homologues, the tricalbins, for formation of ER-plasma membrane contacts. However, the functions of these proteins are poorly understood, in part because no high-resolution structural information on their domains is available. To help bridge this gap in our knowledge, we have elucidated the structure of the E-Syt2 C₂AB fragment and analyzed its Ca²⁺-binding properties. Our data reveal that the E-Syt2 tandem C₂AB fragment exhibits an unexpected V-shaped structure with a well-defined orientation of the two C₂ domains and show that the E-Syt2 C₂A domain binds up to four Ca²⁺ ions, thus suggesting that E-Syt2 and E-Syt3 physiologically perform a Ca²⁺-dependent function that remains to be discovered. Overall, our results provide a critical framework to rationalize this and other functions performed by E-Syt tandem C₂ domains, and are also important for understanding how C₂ domains function in general.

In agreement with the previous observation that the E-Syt2 and E-Syt3 C₂C domain mediates plasma membrane localization (Min et al., 2007), recent studies showed that formation of ER-plasma membrane contacts requires the C₂C domain of E-Syt2 and E-Syt3, and provided evidence that this function depends on Ca²⁺-independent interactions with PI(4,5)P₂ (Giordano et al., 2013). These studies also indicated that E-Syt1 is recruited to the plasma membrane in a Ca²⁺-dependent manner, but potential roles of Ca²⁺ binding to the C₂A domains of E-Syt2 and E-Syt3 were not studied. Our results now demonstrate that the E-Syt2 C₂A domain binds multiple Ca²⁺ ions. This observation suggests that the C₂A domains from E-Syt2 and the highly similar E-Syt3 must have a Ca²⁺-dependent function. Moreover, our data indicate that E-Syt1 may have two independent tandem C₂ domain modules (C₂AB and C₂CD) that are also Ca²⁺-regulated.

Ca²⁺-dependent phospholipid binding to the C₂A domain could enhance the Ca²⁺-independent ER-plasma membrane tethering activity of the C₂C domain (Figure 8). However, since the C₂C domain is sufficient for plasma membrane localization, we favor the possibility that Ca²⁺ binding to the C₂A domain plays a role in physiological processes that are Ca²⁺-regulated, including some form of Ca²⁺ signaling. For instance, the SMP domain has been implicated in lipid metabolism and lipid transfer between membranes (Kopec et al., 2010; Toulmay and Prinz, 2012), and appears to cooperate with the C₂A domain in lipid binding (Min et al., 2007). Thus, Ca²⁺-dependent phospholipid binding to the C₂A domain could favor localization of the SMP domain between the two membranes (Figure 8), facilitating the putative lipid-transfer activity (or other activities) in response to a Ca²⁺ signal.

Figure 8.

Model of E-Syt function incorporating a Ca²⁺-dependent function for the C₂A domain. The model postulates that, in addition to promoting ER-plasma membrane contact through Ca²⁺-independent binding of their C₂C domain to the plasma membrane, E-Syt2 and E-Syt3 have a Ca²⁺-dependent function mediated by the C₂A domain. This function likely involves Ca²⁺-induced binding of the C₂A domain to the plasma membrane, which could help bring the SMP domain close to the plasma membrane. The C₂B domain is not expected to contribute to membrane binding, but could bind to a protein target on the plasma membrane (orange circle labeled X), thus cooperating with the C₂A domain.

Our studies reveal similarities but also fundamental differences between the tandem C₂ domains of E-Syt2 and synaptotagmin-1. The Ca²⁺-binding mode of the E-Syt2 C₂A domain (Figure 5B) is clearly similar to that of the synaptotagmin-1 C₂A domain (Ubach et al., 1998), sharing at least two Ca²⁺-binding sites with practically identical ligands (sites Ca1 and Ca2) and probably a third site (site Ca4 site of the E-Syt2 C₂A domain, called Ca3 for the synaptotagmin-1 C₂A domain). The E-Syt2 C₂A domain contains an additional site (Ca3) that has not been observed previously. The presence of four Ca²⁺ binding sites in the E-Syt2 C₂A domain is likely facilitated by the highly acidic nature of loop 3 and of the overall domain. This feature probably underlies also the lack of a preference for negatively charged phospholipids (Min et al., 2007) exhibited by the synaptotagmin-1 C₂A domain (Davletov and Sudhof, 1993).

More drastic differences are observed between the C₂B domains of E-Syt2 and synaptotagmin-1. Thus, the synaptotagmin-1 C₂B domain binds Ca²⁺, which is critical for its function in triggering neurotransmitter release (Mackler et al., 2002), and is characterized by a highly basic nature that is likely also crucial for synaptotagmin-1 function, enabling simultaneous binding to the synaptic vesicle and plasma membranes as well as to the SNARE complex to cooperate in membrane fusion (Arac et al., 2006; Dai et al., 2007; Seven et al., 2013). Clearly, the E-Syt2 C₂B domain must perform a different function that is currently unknown. The findings that the E-Syt2 C₂B domain does not bind Ca² and is zwitterionic suggest that it might act as a protein-protein interaction domain. We speculate that binding of the E-Syt2 C₂B domain to a receptor on the plasma membrane (labeled 'X' in Figure 8) might be coupled to the activities of the other domains of E-Syt2. Such coupling could be facilitated by the V-shape of the E-Syt2 C₂AB fragment resulting from the rigid orientation of the C₂A and C₂B domains, which constitutes a unique feature of the E-Syt2 C₂AB fragment and contrasts with the flexibility in the relative orientation of the synaptotagmin-1 C₂ domains (Arac et al., 2006; Herrick et al., 2009; Choi et al., 2010). However, the sequence of the linker between the E-Syt2 C₂A and C₂B domains is markedly less conserved through evolution than the sequence of the C₂ domains (Min et al., 2007) and hence it is unclear whether the V-shape has a critical functional significance.

The decrease in thermal stability of the tandem C₂AB fragment (Figure 2) constitutes another unexpected property of E-Syt2 that contrasts with the increased stability normally induced by Ca²⁺ binding to C₂ domains [e.g. (Shao et al., 1997)]. This feature is unlikely to have biological significance but is interesting from the point of view of protein biophysics because ligand binding normally stabilizes proteins. Potential explanations for such a decrease include the possibility that the thermal denaturations of the E-Syt2 C₂AB fragment do not reflect protein stability and are primarily governed by kinetic effects, or that Ca²⁺ binds more tightly to the unfolded protein than the folded protein (Cimmperman et al., 2008).

All these observations suggest that the tandem C₂ domains of E-Syt2 and synaptotagmin-1 function by considerably different mechanisms. This is not surprising given the fact that the ESyt2 and synaptotagmin-1 C₂ domains have different topologies and have highest structural similarity to C₂ domains with topology II such as the intersectin 2 C₂ domain and the Munc13-1 C₂B domain. Nevertheless, it is also unclear to what extent the E-Syt2 C₂ domains share functional properties with these C₂ domains, which do not occur in tandem. For instance, the Munc13-1 C₂B domain has an unusually long loop 3 that confers in part its phosphoinositide binding properties (Shin et al., 2010) and is not observed in the E-Syt2 C₂ domains; the intersectin 2 C₂ domain exhibits normal lengths in its top loops; and the E-Syt2 C₂A and C₂B domains have an unusually long loop 1 that is not shared by the intersectin 2 C₂ domain and the Munc13-1 C₂B domain. In the E-Syt2 C₂A domain, loop 1 appears to undergo a considerable conformational change upon Ca²⁺ binding (Figures 4B and S2). While the physiological significance of this observation is unknown, it is interesting that part of the sequence of this loop is alternatively spliced (Min et al., 2007) and that splicing of short sequences can dramatically change the Ca²⁺-depedendent properties of C₂ domains (Garcia et al., 2004). Clearly, much research will be required to learn more about E-Syts, about how they function and about their regulation, but the structures of the E-Syt2 C₂AB fragment and its Ca²⁺-binding properties described here provide a foundation to rationalize future studies.

EXPERIMENTAL PROCEDURES

Protein expression and purification

The cDNA sequence of the human E-Syt2 C2AB fragment (residues 363–660) from a previously described construct (Min et al., 2007) was subcloned into a pGEX-KG vector (Guan and Dixon, 1991), to express the protein fused at the N-terminus with glutathione S-transferase (GST) and with a linker containing a TEV cleavage site. All mutants were obtained by site-directed mutagenesis using PCR and custom-designed primers. Proteins were expressed in E. coli BL21 (DE3) cells, inducing with 0.1 mM IPTG at 16 °C for 18 hours. Cells were resuspended in buffer containing 0.15 M NaCl, 0.01 M Tris pH 8.0, 0.1 mM EDTA, 2 mM DTT and 0.5 mM AEBSF. The fusion protein was isolated by affinity chromatography on glutathione-agarose and cleaved with TEV, and the cleaved protein was purified by gel filtration on a Hiload^TM 16/60 Superdex^TM 75 column in 0.125 M NaCl, 0.025 M Tris pH 7.4, 0.5 mM TCEP. Uniform ¹⁵N labeling was achieved by growing the cells in M9 medium containing 15NH₄Cl as nitrogen source. Selenomethionine(SeMet)-derivatized protein was overexpressed in BL21 (DE3) cells grown in M9 medium, adding SeMet and amino acids (Thr, Lys, Phe, Leu, Ile and Val; for inhibition of Met synthesis) 15 minutes before IPTG induction (Van Duyne et al., 1993). Yields were similar for all the proteins (ca. 6 mg of purified protein from one liter of culture). Proteins prepared for crystallization were kept at 4°C or −80°C in gel filtration buffer including 2.5% glycerol, until use.

Circular dichroism

CD spectra were acquired on an Aviv 62DS spectropolarimeter. Thermal denaturations were performed with 15 mM samples of E-Syt2 C₂AB fragments in 0.04 M Tris pH7.4, 0.1 M NaCl, 0.1 mM EGTA with or without 1.1 mM CaCl₂, monitoring the CD absorption at 215 nM. The fraction of unfolded protein f was calculated as f = (I_obs − I_f)/(I_u − I_f), where I_obs is the observed signal intensity, and I_f and I_u are the intensities corresponding to the folded and unfolded states, respectively. I_f and I_u were calculated by extrapolation of the linear regions of the denaturation curves.

NMR spectroscopy

All spectra were acquired at 25°C on a Varian INOVA 800 MHz spectrometer equipped with a cold probe. Samples of 0.16 mM E-Syt2 C₂AB fragments were dissolved in 25 mM Tris pH 7.4, 125 mM NaCl, 0.5 mM TCEP and 5% D₂O. One-dimensional (1D) ¹H-NMR spectra of the initial samples were acquired before and after addition of 100 μM EDTA to ensure that there were negligible amounts of Ca²⁺ at the start of the titrations (Pan et al., 2011). ¹H-¹⁵N TROSY-HSQC spectra were acquired after successive additions of different Ca²⁺ concentrations. The data were processed with NMRPipe (Delaglio et al., 1995) and analyzed with NMRView (Johnson and Blevins, 1994).

Crystallization and X-ray diffraction data collection

Crystals of the Ca²⁺-free E-Syt2 C₂AB fragment were grown using the hanging-drop vapor-diffusion method from drops containing 1 μL protein (12.9 mg/ml) and 1 μL of reservoir solution (1.5 M LiSO₄, 0.1 M NaAc pH 4.9), and equilibrated over reservoir solution at 20°C. Cryoprotection was performed by transferring the crystals to a final solution of 2.25 M LiCl, 0.125 M NaCl, 0.1 M sodium acetate pH 4.9, 2.5% glycerol and 25% ethylene glycol, and were flash-cooled in liquid nitrogen. E-Syt2 C₂AB fragment crystals exhibited the symmetry of space group P4₃ with cell dimensions of a = 62.30 Å, c = 226.40 Å, contained two molecules per asymmetric unit, and diffracted to a minimum Bragg spacing (d_min) of 2.10 Å when exposed to synchrotron radiation. The SeMet-derivatized E-Syt2 C₂AB fragment was crystallized in a similar manner to the native protein, with reservoir solution containing 1.6 M Li₂SO₄, 0.1 M NaAc pH 5.2, and 0.5 mM TCEP. SeMet E-Syt2 C₂AB fragment crystals were cryo-protected with reservoir solution supplemented with 15% ethylene glycol, and were isomorphous to the native crystals but diffracted to a d_min of 2.33 Å.

The Ca²⁺-bound E-Syt2 C₂AB fragment (dissolved in 12.4 mg/ml in 0.125 M NaCl, 0.025 M Tris pH 7.4, 0.5 mM TCEP and 2 mM CaCl₂) was crystallized at 20°C using the hanging-drop vapor-diffusion method from drops set up in a 1:1.5 ratio with a reservoir solution containing 4% PEG 8,000, 0.1 M Hepes pH 8.0. The crystals were cryo-protected with solution of 0.1 M Hepes pH8.0, 0.125 M NaCl, 2 mM CaCl₂, 0.5 mM TCEP, 6% PEG 8,000 and 40% ethylene glycol. Ca²⁺-bound E-Syt2 C₂AB fragment crystals exhibited the symmetry of space group C2 with cell dimensions of a = 95.84 Å, b= 53.19 Å, c = 85.28 Å, β = 99.958° and contained one molecule per asymmetric unit. The crystals diffracted anisotropically to a minimum Bragg spacing (d_min) of 2.55 Å when exposed to synchrotron radiation. Data were indexed, integrated and scaled using the HKL-3000 program package (Minor et al., 2006). Data collection statistics are provided in Table 1.

Phase determination and structure refinement

Phases for the Ca²⁺-free E-Syt2 C₂AB fragment were obtained from a single-wavelength anomalous dispersion experiment using SeMet-labeled protein with data to a d_min of 2.33 Å. Ten selenium sites were located using the program SHELXD (Schneider and Sheldrick, 2002) and phases were refined with the program Mlphare (Otwinowsky, 1991), resulting in an overall figure-of-merit of 0.16 for data between 43.3 and 2.33 Å. Phases were further improved by density modification with two-fold non-crystallographic symmetry averaging with the program Parrot(4) (Cowtan, 2010) resulting in a figure-of-merit of 0.57. An initial model containing 97% of all the E-Syt2 C₂AB fragment residues was automatically generated in the program Buccaneer (Cowtan, 2006).

Additional residues of the E-Syt2 C₂AB fragment were manually modeled in the programs O (Jones et al., 1991) and Coot (Emsley et al., 2010). Refinement was performed to a resolution of 2.10 Å using the program Phenix (Afonine et al., 2010) with a random 5% of all data set aside for an R_free calculation. The current model contains two E-Syt2 C₂AB fragment monomers; included are residues 363–551 and 557–659 of monomer A, residues 364–402, 410–549 and 557–659 of monomer B and 146 solvent atoms. The R_work is 0.207, and the R_free is 0.247. A Ramachandran plot generated with Molprobity (Chen et al., 2010) indicated that 95.7% of all protein residues are in the most favored regions, and 0.2% (one residue) in the disallowed regions. This residue is located in a surface loop with weak electron density.

Phases for the Ca²⁺-bound E-Syt2 C₂AB fragment were obtained via molecular replacement in the program Phaser (McCoy et al., 2007) using a search model constructed from the coordinates of monomer A of the Ca²⁺-free E-Syt2 C₂AB fragment with the first 20 N-terminal amino acids removed. Model building and refinement was performed as described above, with the following modification: due to the high anisotropy and lower resolution of the data, monomer A of the Ca²⁺-free E-Syt2 C₂AB fragment was used as a reference model to provide an additional source of geometry restraints. The current model contains one E-Syt2 C₂AB fragment monomer, three Ca²⁺ ions and four waters complexed to the calciums. Included are E-Syt2 C₂AB fragment residues 380–525, 532–550, 556–659, the R_work is 0.222, and the R_free is 0.287. A Ramachandran plot generated with Molprobity indicated that 94.7% of all protein residues are in the most favored regions, and 1.5% (three residues) in the disallowed regions. These residues are located in regions with poorly defined electron density. Phasing and model refinement statistics are provided in Table 1. The structures of the Ca²⁺-free and Ca²⁺-bound E-Syt2 C₂AB fragment have been deposited in the PDB with accession codes 4NPJ and 4NPK, respectively.

Highlights

• The crystal structure of the tandem C₂AB domains of E-Syt2 has been determined.

• The structure has a V-shape with a rigid orientation of the two C₂ domains.

• The E-Syt2 C₂A domain binds four Ca²⁺ ions while the C₂B domain does not bind Ca²⁺.

• E-Syts most likely have an as yet unidentified Ca²⁺-dependent function.