Exceptionally tight membrane-binding may explain the key role of the synaptotagmin-7 C₂A domain in asynchronous neurotransmitter release

Significance

Synaptotagmins-1 and -7, act as Ca²⁺ sensors for the fast and slow components of neurotransmitter release, respectively, through the two C₂ domains that form their cytoplasmic region. Surprisingly, Ca²⁺-binding to the synaptotagmin-7 C₂A domain is more critical for slow release than Ca²⁺-binding to the C₂B domain, whereas the opposite was found for the synaptotagmin-1 C₂ domains and fast release. This paper suggests an explanation for this apparent contradiction, showing that the C₂A domain dominates binding of synaptotagmin-7 to membranes, whereas such binding is dominated by the C₂B domain in the case of synaptotagmin-1. Thus, membrane affinity may be a key determinant of the relative functional importance of synaptotagmin C₂ domains.

Abstract

Synaptotagmins (Syts) act as Ca²⁺ sensors in neurotransmitter release by virtue of Ca²⁺-binding to their two C₂ domains, but their mechanisms of action remain unclear. Puzzlingly, Ca²⁺-binding to the C₂B domain appears to dominate Syt1 function in synchronous release, whereas Ca²⁺-binding to the C₂A domain mediates Syt7 function in asynchronous release. Here we show that crystal structures of the Syt7 C₂A domain and C₂AB region, and analyses of intrinsic Ca²⁺-binding to the Syt7 C2 domains using isothermal titration calorimetry, did not reveal major differences that could explain functional differentiation between Syt7 and Syt1. However, using liposome titrations under Ca²⁺ saturating conditions, we show that the Syt7 C₂A domain has a very high membrane affinity and dominates phospholipid binding to Syt7 in the presence or absence of l-α-phosphatidylinositol 4,5-diphosphate (PIP₂). For Syt1, the two Ca²⁺-saturated C₂ domains have similar affinities for membranes lacking PIP₂, but the C₂B domain dominates binding to PIP₂-containing membranes. Mutagenesis revealed that the dramatic differences in membrane affinity between the Syt1 and Syt7 C₂A domains arise in part from apparently conservative residue substitutions, showing how striking biochemical and functional differences can result from the cumulative effects of subtle residue substitutions. Viewed together, our results suggest that membrane affinity may be a key determinant of the functions of Syt C₂ domains in neurotransmitter release.

Neurons communicate with each other via the release of neurotransmitters by synaptic vesicle exocytosis. Neurotransmitter release is triggered by entry of Ca²⁺ into a presynaptic terminal, exhibiting a fast synchronous component that occurs within less than 1 ms after Ca²⁺ influx and a slower, asynchronous component (1). The sophisticated machinery that controls release includes core proteins, such as Munc18-1, Munc13s, and the SNAP receptors (SNAREs) syntaxin-1, SNAP-25, and synaptobrevin, which form a tight SNARE complex that brings the vesicle and plasma membranes together and is key for membrane fusion (2–4). In addition, the exquisite regulation of synaptic exocytosis depends on specialized factors, including the Ca²⁺ sensors from the synaptotagmin (Syt) family (5). Syt1 and the closely related Syt2 and Syt9 act as the Ca²⁺ sensors for the major, synchronous components of neurotransmitter release (6–8), whereas Syt7 mediates at least in part the asynchronous component (9). While major advances have been made in characterizing how all these proteins function, and reconstitution experiments have suggested that basic steps of synaptic vesicle fusion can be recapitulated with eight central proteins (10), fundamental questions remain about the overall mechanism of Ca²⁺-dependent membrane fusion and in particular about the specific actions of Syts.

Syts contain tandem C₂ domains (referred to as C₂A and C₂B domains) that span most of their cytoplasmic region and adopt characteristic β-sandwich structures (11–13). The Syt1 C₂A and C₂B domains bind three and two Ca²⁺ ions, respectively, through loops at the top of the these β-sandwiches (13–15), which does not cause substantial conformational changes (12) but drastically changes the electrostatic potential of the Ca²⁺-binding region (13, 15, 16). These top loops also mediate Ca²⁺-dependent binding to negatively charged phospholipid membranes (17, 18). The key functional importance of this activity was demonstrated by the finding that mutations that increase or decrease the apparent Ca²⁺ affinity of Syt1 in phospholipid binding lead to parallel changes in the Ca²⁺ sensitivity of release (6, 19). The Syt1 C₂B domain can bind simultaneously to two membranes in a Ca²⁺-dependent manner via the Ca²⁺-binding loops at the top and two arginines (R398 and R399) at the bottom of the β-sandwich, which may cooperate with SNARE complex formation in bringing the membranes together and inducing membrane fusion upon Ca²⁺ influx (20–22). In support of this model, NMR studies of Syt1–SNARE complex interactions in solution revealed a dynamic binding mode that allows simultaneous interactions with the SNAREs and two closely apposed membranes (23). However, SNARE binding in this structure is mediated by a polybasic region of the C₂B domain that also binds to l-α-phosphatidylinositol 4,5-diphosphate (PIP₂) (24, 25), and multiple Syt1–SNARE complex binding modes likely exist (23). Indeed, a crystal structure revealed three different Syt1–SNARE complex binding modes, all distinct from that observed in solution (26). Hence, it is still unclear how the functions of Syt1 and the SNARE complex are coupled.

The structural and biochemical properties uncovered by the extensive studies of the Syt1 C₂ domains are paradigmatic in general terms for C₂ domains of other Ca²⁺-dependent Syts, but differences in these properties can range from subtle to dramatic (5). Such differences can be difficult to predict from sequence alignments alone (e.g., ref. 27) and can result in fine or drastic functional differences that may also be difficult to rationalize from sequence analyses. Particularly intriguing in this respect are the differences between Syt1 and Syt7, as both isoforms function as Ca²⁺ sensors for exocytosis in chromaffin cells (28) and presynaptic terminals (6, 9), but Syt7 cannot functionally replace Syt1 to trigger fast synchronous neurotransmitter release (8). In fact, replacing the C₂B domain in Syt1 with the Syt7 C₂B domain prevents rescue of synchronous release in Syt1 KO neurons even after extensive mutagenesis to replace residues of the Syt7 C₂B domain back with those present in Syt1 (29). The crystal structure of the Syt7 C₂B domain did not reveal substantial differences from that of the Syt1 C₂B domain that could clarify the basis for these results (29). Moreover, mutating the Ca²⁺-binding sites of the Syt1 C₂B domain impairs synchronous release much more drastically than analogous mutations in the Ca²⁺-binding sites of the Syt1 C₂A domain (30–33), whereas asynchronous release is more strongly impaired by mutations in the Ca²⁺-binding sites of the Syt7 C₂A domain than mutations in the Syt7 C₂B domain Ca²⁺-binding sites (9). In principle, these results suggest that the relative functional importance of the two C₂ domains is switched in Syt1 and Syt7, but this conclusion represents a simplification because substitutions in the C₂B domain Ca²⁺-binding sites render Syt1 dominant-negative (30, 34). Strikingly, expressing such mutant Syt1 in WT cells blocks both synchronous and asynchronous Ca²⁺-triggered release, suggesting that C2B-domain mutant Syt1 locks synaptic vesicles into a Ca²⁺-unresponsive state. Overall, these findings show that understanding the key determinants for functional differentiation between Syt1 and Syt7 remains as a major challenge in this field.

In the study presented herein, we have investigated the structural and biochemical basis for the differences in the relative functional importance of Ca²⁺-binding to the Syt7 C₂ domains. We have determined crystal structures of the Syt7 C₂A domain and of a fragment spanning the two Syt7 C₂ domains (C₂AB), and found no structural features that could readily explain their distinct functional properties. Isothermal titration calorimetry (ITC) data reveal that the intrinsic Ca²⁺ affinities of the Syt7 C₂ domains are similar to those of the Syt1 C₂ domains. However, in contrast to Syt1, we find that under Ca²⁺ saturating conditions the Syt7 C₂A domain binds to phospholipids with much higher affinity than the Syt7 C₂B domain, and that disruption of the C₂A domain Ca²⁺-binding sites impairs binding of the Syt7 C₂AB fragment to phospholipids more strongly than disruption of the C₂B domain Ca²⁺-binding sites. Moreover, the Ca²⁺-saturated Syt7 C₂A domain clusters liposomes and hence has the ability to bring two membranes together, whereas the Syt7 C₂B domain does not, again in contrast to results obtained with Syt1. These results suggest that the dominant function of Ca²⁺ binding to the Syt7 C₂A domain versus the C₂B domain in asynchronous release arises because of its preponderant role in membrane binding and perhaps in bringing the vesicle and plasma membranes together. Furthermore, mutagenesis shows that cooperative effects of apparently conservative amino acid substitutions underlie at least in part the much higher phospholipid binding affinity of the Syt7 C₂A domain compared with the Syt1 C₂A domain, illustrating how the cumulative effects of subtle mutations can have profound effects in protein biochemistry and function.

Results

Crystal Structure of the Ca²⁺-Bound Syt7 C2A Domain.

The crystal structure of the Ca²⁺-bound Syt7 C₂B domain has been determined and found to be very similar to that of the Syt1 C₂B domain, although three Ca²⁺-binding sites were observed in the Syt7 C₂B domain instead of two (29). A solution structure of the Ca²⁺-free Syt7 C₂A domain determined by NMR spectroscopy was deposited in the Protein Data Bank as part of a structural genomics initiative (PDB ID code 2D8K). To investigate whether Ca²⁺ induces conformational changes in the Syt7 C₂A domain, we performed crystallization screens in the presence of Ca²⁺ and were able to obtain crystals that diffracted to 1.7 Å. The structure of the Ca²⁺-bound Syt7 C₂A domain was determined by molecular replacement using the Ca²⁺-free solution structure as the search model. Data collection and refinement statistics are described in Table S1.

Table S1.

Data collection and refinement statistics

Data collection
Crystal	Syt7 C2A	Syt7 C2AB
Space group	P6(5)	P2(1)
Cell constants, Å, °	55.30, 89.81	60.64, 75.06, 70.98, 90.00, 114.51, 90.00
Wavelength, Å	0.97915	0.97918
Resolution range, Å	27.65–1.70 (1.73–1.70)	44.45–2.25 (2.33–2.25)
Unique reflections	17,195 (866)	25,080 (774)
Multiplicity	6.6 (4.1)	3.7 (2.4)
Data completeness, %	99.7 (99.5)	91.6 (56.3)
Rmerge,* %	5.4 (98.5)	7.9 (76.6)
Rpim,† %	2.2 (52.8)	4.6 (54.1)
CC1/2 (last resolution shell)	0.542	0.698
I/σ(I)	34.6 (1.42)	15.4 (1.28)
Wilson B-value, Å2		44.6
Wilson B-value, Å2, sharpened‡	11.4	33.3
Mean |L|§	0.483 (0.500/0.375)	0.537 (0.500/0.375)
Mean L2	0.314 (0.333/0.200)	0.375 (0.333/0.200)
Refinement statistics
Resolution range, Å	27.65–1.69 (1.75–1.69)	44.45–2.25 (2.33–2.25)
No. of reflections Rwork/Rfree	14,830/1,638 (897/96)	20,026/2,011 (737/67)
Data completeness, %	95.7 (51.0)	80.2 (26.0)
Atoms (non-H Protein/Ca2+/solvent)	1,169/3/188	4,341/8/103
Rwork, %	14.4 (22.9)	21.4 (30.2)
Rfree, %	18.1 (34.3)	25.6 (31.3)
Rmsd bond length, Å	0.017	0.014
Rmsd bond angle, °	0.675	0.489
Mean B-value, Å2 (protein/Ca2+/solvent)	16.2/18.9/29.9	Chain A, 50.1/chain B, 51.8/50.5/52.2
Ramachandran plot, % (favored/additional/disallowed)¶	97.8/2.2/0	94.0/6.0/0
Clashscore/overall score¶	0.00/0.53	0.68/1.12
Maximum-likelihood coordinate error	0.16	0.29
Missing residues	−19 to −9	−19 to −1, 134, 260–266

Data for the outermost shell are given in parentheses.

^*R_merge = 100 Σ_hΣ_i|I_h,i −〈I_h〉|/Σ_hΣ_i〈I_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.

^†R_pim = 100 Σ_hΣ_i [1/(n_h − 1)]^1/2|I_h,i −〈I_h〉|/Σ_hΣ_i〈I_h,i〉, where n_h is the number of observations of reflections h.

^‡B-factor sharpening was performed in the autocorrection mode of HKL-3000 (57).

^§L values are calculated using PHENIX (60) for acentric reflections in the resolution range 10–3.5 Å. Values in parentheses are the expected values for untwinned and perfectly twinned data, respectively.

^¶As defined by the validation suite MolProbity (64).

Fig. 1A shows the ribbon diagram of the Ca²⁺-bound Syt7 C2A domain, which has a typical β-sandwich architecture composed of two four-stranded β-sheets characteristic of C₂ domains (35). The structure is very similar to the Ca²⁺-free solution structure (Fig. 1B), with a 1.13 Å rmsd for 127 equivalent Cα carbons. Differences observed in some of the loops can readily arise from the known flexibility of these loops in C₂ domains (12, 35) and potential adaptations induced by crystal contacts. Hence, Ca²⁺ binding does not appear to induce substantial structural changes in the Syt7 C₂A domain, as observed for the Syt1 C₂A domain (12). Indeed, the structures of the Ca²⁺-bound C₂A domains of Syt1 and Syt7 are also very similar (Fig. 1C), with a 0.92 Å rmsd for 113 equivalent Cα carbons. Note that the perturbations induced by Ca²⁺ in the ¹H-¹⁵N heteronuclear single quantum coherence (HSQC) spectra of the Syt7 C₂A domain (36) are also comparable to those observed for the Syt1 C₂A domain (15). Correspondingly, the Syt7 C₂A domain binds three Ca²⁺ ions via the same ligands (Fig. 1 D and E) observed in the three Ca²⁺-binding sites of the Syt1 C₂A domain (12, 15). Overall, these results show that there are no overt differences in the structure and Ca²⁺-binding mode of the C₂A domains from Syt1 and Syt7 that might explain the predominant role of Ca²⁺ binding to the Syt7 C₂A domain in asynchronous neurotransmitter release.

Fig. 1.

Crystal structure of the Ca²⁺-bound Syt7 C₂A domain. (A) Ribbon diagram of the crystal structure of the Ca²⁺-bound Syt7 C₂A domain. The three bound Ca²⁺ ions are shown as yellow spheres. The two loops involved in Ca²⁺ binding are indicated (loop 1 and loop 3), and the N and C termini are labeled N and C, respectively. (B) Superposition of the crystal structure of the Ca²⁺-bound Syt7 C₂A domain (cyan) with the solution structure of Ca²⁺-free Syt7 C₂A domain (red) (PDB ID code 2D8K). (C) Superposition of the crystal structure of the Ca²⁺-bound Syt7 C₂A domain (cyan) with the solution structure of Ca²⁺-bound Syt1 C₂A domain (orange) (PDB ID code 1BYN). (D) Diagram illustrating the Ca²⁺-binding sites of the Syt7 C₂A domain. The three bound Ca²⁺ ions are labeled Ca1, Ca2, and Ca3. The Ca²⁺ ligands are shown as stick models and labeled. (E) Diagram of the Ca²⁺-binding sites of Syt7 C₂A domain summarizing all of the residues in loops 1 and 3 (black letters). The residues that are different in the Syt1 C₂A domain are indicated in parenthesis in red letters.

Crystal Structure of the Syt7 C2AB Fragment.

A crystal structure of a Syt1 C₂AB fragment in the absence of Ca²⁺ revealed extensive contacts between the two C₂ domains that distort the Ca²⁺-binding region of the C₂A domain and hence would be expected to be disrupted upon Ca²⁺ binding to this domain, which could provide a mechanism for Ca²⁺-regulation of neurotransmitter release (37). Although such contacts did not appear to occur in solution (20), we performed extensive crystallization screens with the Syt7 C₂AB fragment to examine the possibility of intramolecular interactions between the two Syt7 C₂ domains. Unfortunately, we were not able to obtain crystals for Ca²⁺-free Syt7 C₂AB, but we did obtain crystals in the presence of Ca²⁺ and we determined its structure by molecular replacement using the crystal structures of the Syt7 C₂A and C₂B domains (Fig. 1A and ref. 29) as search models (Fig. 2A). Data collection and refinement statistics are summarized in Table S1.

Fig. 2.

Crystal structure of the Ca²⁺-bound Syt7 C₂AB fragment. (A) Ribbon diagram of the crystal structure of the Ca²⁺-bound Syt7 C₂AB fragment, with the C₂A domain in cyan and the C₂B domain in blue. The bound Ca²⁺ ions are shown as yellow spheres. The loops involved in Ca²⁺ binding are indicated (loop 1 and loop 3), and the N and C termini of both domains are labeled N and C, respectively. The sequence linking the two C₂ domains, for which there is insufficient electronic density to build the structure, is illustrated by a dotted line. (B) Superposition of the crystal structure of the Ca²⁺-bound Syt7 C₂AB (cyan and blue) with the crystal structures of Ca²⁺-free Syt1 C₂AB (wheat; PDB ID code 2R83), Syt1 C₂AB bound to the SNARE complex (red; PDB ID code 5CCG) and Ca²⁺-free Syt3 C₂AB (green; PDB ID code 1DQV). Only the C₂A domain of each structure was used for the superposition to illustrate the diversity of relative orientations between the two C₂ domains in these structures. The position of all of the C₂A domains is indicated by C₂A and those of the different C₂B domains are also labeled; Syt1 (SC) C₂B domain refers to Syt1 C₂AB bound to the SNARE complex.

The structures of the two C₂ domains in the Syt7 C₂AB structure are very similar to those observed in the crystal structures of the isolated C₂A (described above) and C₂B (29) domains. Comparison of the isolated C₂A domain with this domain in the two structures of C₂AB in the asymmetric unit of its crystals revealed rmsds of 0.49 and 0.58 Å for 125 equivalent Cα carbons, whereas for the C₂B domain the rmsds were 0.45 and 0.41 Å for 138 and 137 equivalent Cα carbons, respectively. No density could be modeled for the linker sequence between the two C₂ domains in the structure of the Syt7 C₂AB fragment, and the observed contacts between the domains were typical of crystal packing interfaces, suggesting that the relative orientation between the two C₂ domains is a result of crystal packing. Comparison of the crystal structure of Syt7 C₂AB with those of isolated Syt1 C₂AB (37), one of the C₂AB molecules in the crystal structure of Syt1 C₂AB bound to the SNARE complex (26) and the crystal structure of Syt3 C₂AB (38), shows that the relative orientation of the two C₂ domains is very different in each structure (Fig. 2B), supporting the conclusion drawn from NMR studies of Syt1 C₂AB that this relative orientation is flexible, and hence adaptable (20). We also analyzed the Syt7 C₂AB fragment by NMR spectroscopy and found that its ¹H-¹⁵N HSQC spectra in the presence and absence of Ca²⁺ are very similar to the superposition of ¹H-¹⁵N HSQC acquired separately for the isolated C₂A and C₂B domains (Fig. S1). There are a few cross-peaks that do not overlap, but they can be attributed to residues from the linker or adjacent sequences, and the overall coincidence of cross-peaks is inconsistent with extensive contacts between the two C₂ domains, thus supporting the conclusion that their relative orientation is flexible.

Fig. S1.

There are no stable interactions between the C₂A and C₂B domains of Syt7 in solution. Superpositions of ¹H-¹⁵N HSQC spectra of the Syt7 C₂AB fragment (black contours), the Syt7 C₂A domain (green contours) or C₂B domain (red contours) in the presence of 1 mM EDTA (A) or 2 mM Ca²⁺ (B).

Intrinsic Calcium Binding to the Syt7 C₂A and C₂B Domains.

In the crystal structure of the Syt7 C₂AB fragment we observed only two Ca²⁺ ions bound to the C₂A domain and two Ca²⁺ ions bound to the C₂B domain (Fig. 2A), corresponding in both cases to sites Ca1 and Ca2 as defined for the C₂A domain in Fig. 1E. These findings contrast with the observation of three bound Ca²⁺ ions in the structure of the isolated Syt7 C₂A domain (Fig. 1D) and four bound Ca²⁺ ions in the crystal structure of the isolated C₂B domain (29). These differences are not surprising, as the third Ca²⁺-binding site observed in the isolated Syt7 C₂A domain (Ca3) is likely of lower affinity because it is formed by only four protein ligands (Fig. 1D) and, in the final refined structure, Ca3 displayed a higher B-factor (11.8 Ų) than Ca1 (7.8 Ų) or Ca2 (7.2 Ų). The Ca3 site may not be occupied in the crystals of the Syt7 C₂AB fragment because of the high sulfate concentration present in the crystallization conditions. For the isolated Syt7 C₂B domain, one of the two additional Ca²⁺ ions was bound to site Ca3, which is also formed by four protein ligands, and the other was bound to an additional site formed by only three protein ligands on the outside of the cup shape formed by the Ca²⁺-binding loops, near a lattice contact (29). In the final refined Syt7 C₂B structure, Ca3 also displayed a higher B-factor (16.2 Ų) than Ca1 (8.8 Ų) or Ca2 (10.1 Ų). Although the crystals of the isolated Syt7 C₂B domain also contained high sulfate concentration, the lower pH used to crystalize the Syt7 C₂AB fragment (6.0 instead of 6.5) may have decreased the already weak Ca²⁺ affinity of these two sites and, hence, no binding is observed.

To analyze the intrinsic affinities of the Ca²⁺-binding sites of the Syt7 C₂ domains in solution, we used ITC (Fig. 3). For this purpose, we used the isolated Syt7 C₂A and C₂B domains because the expected existence of multiple Ca²⁺ binding sites would hinder fitting the data obtained with the C₂AB fragment, and because the two C₂ domains are not expected to influence each other in Ca²⁺ binding given their flexible relative orientation. We found that fitting the ITC data obtained for the Syt7 C₂A domain (Fig. 3A) required a three-sequential-binding-site model, consistent with the crystal structure of the isolated domain (Fig. 1A). The K_Ds measured for the three Ca²⁺ binding sites were 90 ± 15 μM, 180 ± 42 μM, and 11 ± 5 mM. These K_Ds should be examined with caution due to the natural difficulty in obtaining accurate K_Ds from such multiple binding-site data, but it is reassuring that the values are comparable to those measured previously by NMR spectroscopy for the Syt7 C₂A domain (∼150 μM, 200–300 μM and >2 mM) (36). They are also similar to K_Ds obtained for the Syt1 C₂A domain by NMR spectroscopy (54 μM, 530 μM and >2 mM) (6) and ITC (119 μM, 465 μM and 1.7 mM) (39).

Fig. 3.

ITC analysis of intrinsic Ca²⁺ binding to the Syt7 C₂ domains. Illustrative ITC data obtained by titration of Ca²⁺ versus the isolated Syt7 C₂A (A) or C₂B (B) domain. The curves represent the fits of the data to a three- (A) or four- (B) sequential-binding-site model.

Fitting the ITC data obtained for the Syt7 C₂B domain required a four-sequential-binding-site model, as attempts to fit the data assuming only three binding sites yielded a systematic error at the end of the titration, which reflects binding to a very low-affinity Ca²⁺-binding site. Hence, these data are consistent with the crystal structure of the isolated Syt7 C₂B domain (29). Unfortunately, it was difficult to obtain reliable K_Ds from these data because of the multiplicity of binding sites and because the values obtained depended strongly on the position chosen for the zero baseline (note the drift observed even at molar ratio = 40) (Fig. 3B). Nevertheless, the ranges of K_D values for the four sites (75–900 μM, 0.5–1.2 mM, 3–5 mM, and 5–27 mM) obtained in different fits give an idea of the intrinsic affinities of the four sites. The K_Ds of the two higher affinity sites are comparable to those of the two Ca²⁺-binding sites of the Syt1 C₂B domain, for which NMR data yielded estimated K_Ds of 300–400 and 500–600 μM (13) and ITC experiments yielded an estimated overall K_D of 200 μM for the two sites (39). It is currently unclear whether all four Ca²⁺-binding sites of the Syt7 C₂B domain, or perhaps only three or two, are physiologically relevant but, regardless of these possibilities, these results indicate that the intrinsic Ca²⁺-binding properties of the C₂ domains from Syt7 are similar to those of the Syt1 C₂ domains and do not provide an explanation for the differences in the relative functional importance of the Syt7 C₂ domains.

The Ca²⁺-Bound Syt7 C₂A Domain Binds to Membranes with Higher Affinity than the Ca²⁺-Bound C₂B Domain.

The apparent Ca²⁺ affinities of Syt C₂ domains in Ca²⁺-dependent phospholipid binding assays are much stronger (apparent K_Ds < 10 μM) than their intrinsic Ca²⁺ affinities because the phospholipids contribute to coordinate the bound Ca²⁺ ions (6, 17, 40). Notably, the apparent Ca²⁺-affinity of the Syt7 C₂A domain in Ca²⁺-dependent phospholipid binding is considerably higher than that of the Syt1 C₂A domain (40). This higher apparent Ca²⁺ affinity can be associated to slower dissociation rates from the membranes (41), which has been also observed in experiments performed with Syt1 and Syt7 C₂AB fragments (42). Since the intrinsic Ca²⁺ affinities of the Syt1 and Syt7 C₂ domains are similar, their differences in Ca²⁺-dependent phospholipid binding can be ascribed to stronger interactions with the membranes in the Ca²⁺-bound state. Thus, we decided to investigate whether differences in membrane affinity might underlie the relative contributions of the C₂A and C₂B domains to Syt7 function. For this purpose, we used a similar approach to that described previously for the Syt1 C₂ domains, which used FRET to detect phospholipid binding and titrations with liposomes under Ca²⁺ saturating conditions to compare relative membrane affinities (39).

In our FRET assays, we used liposomes containing Rhodamine-PE as a FRET acceptor and a BODIPY-FL donor probe was attached to the native cysteines of the Syt7 C₂A and C₂B domain (C260 and C275, respectively), which are located in regions that are not expected to participate in phospholipid binding based on extensive analyses of Syt1 (e.g., refs. 17, 18, 43, and 44) (Fig. S2A). For experiments with the Syt7 C₂AB fragment, the BODIPY-FL probe was attached to either C260 or C275 (indicated by an asterisk; i.e., C₂A*B or C₂AB*) and the other cysteine was mutated to alanine. The position of the probe did not substantially alter the phospholipid binding curves (Fig. S2B). Most of the data shown were obtained with the probe on C260 of the C₂A domain, except for experiments where the Ca²⁺-binding sites of the C₂A domain were mutated, in which case the probe was attached to C275 of the C₂B domain. In initial experiments and those described below for Syt1 fragments, we performed titrations with increasing amounts of liposomes onto 100 nM protein under Ca²⁺ saturating conditions. However, because of the very high liposome affinities observed for fragments including the Syt7 C₂A domain, we lowered the protein concentration to 20 nM to allow better discrimination of relative liposome affinities. Even at these low protein concentrations, excellent consistency was observed in separate experiments performed under the same conditions with different liposome preparations (e.g., Fig. S2 D and E). All of the data could be fit well to a Hill function (Figs. 4–7), with Hill coefficients ranging from 1 to 3 (Table 1) that in principle could indicate some cooperativity. However, it is important to note that it is difficult to interpret Hill coefficients in these experiments because at low concentrations of liposomes (high protein-to-lipid ratios) their surface may be covered by bound proteins, which is expected to occlude some of the available binding sites because of molecular crowding and at the same time can favor liposome clustering. On the opposite end, at high liposome concentrations (low protein-to-lipid ratios) molecular crowding effects are not expected and liposome clustering is less favorable (20). Hence, we did not attempt to rationalize the Hill coefficients and only used the K_Ds derived from fitting the data (Table 1) as apparent values that report on the relative affinities of the Ca²⁺-saturated proteins for the liposomes.

Fig. S2.

Placing the fluorescent probe on the C₂A or C₂B domain does not alter phospholipid binding to the Syt1 or Syt7 C₂AB fragment. (A) Ribbon diagrams of the structures of the Syt7 and Syt1 C₂A or C₂B domains, as indicated. The structure of the Syt7 C₂A domain is described here. Those of the Syt7 C₂B domain, the Syt1 C₂A domain, and the Syt1 C₂B domain correspond to PDB ID codes 3N5A, 1BYN, and 1K5W, respectively. The bound Ca²⁺ ions are shown as yellow spheres. The two loops involved in Ca²⁺ binding are indicated (loop 1 and loop 3), and the N and C termini are labeled N and C, respectively. The native cysteine side chains of the Syt7 C₂A and C₂B domains, as well as the side chains that were mutated to cysteine in the Syt1 C₂A and C₂B domains to attach a fluorescent probe, are shown as stick models. (B) Titrations of 20 nM Syt7 C₂A*B and C₂AB* fragments (where the * indicates the domain where the fluorescent probe was attached) with liposomes in the presence of 1 mM Ca²⁺. Binding was monitored from the FRET developed between a donor BODIPY-FL probe attached to the proteins and a rhodamine acceptor present in the liposomes. Each point represents the average of at least three measurements performed with different liposome preparations. (C) Analogous liposome titrations of 100 nM Syt1 C₂A*B and C₂AB* fragments. All of the data were fit to a Hill function. (D and E) Superpositions of the data obtained in analogous titrations of 20 nM Syt7 C₂A domain (D) or Syt7 C₂B domain (E), illustrating the reproducibility of the results. Each symbol represents one dataset obtained with a different liposome preparation. The apparent K_Ds derived for the six datasets shown in D are 1.49, 2.35, 2.62, 2.42, 2.37, and 3.17 μM, and those derived for the four datasets shown in E are 34.1, 31.4, 38.1, and 36.4 μM. The Syt7 C₂B domain data provide an example with small variability among different titrations while the Syt7 C₂A domain data illustrate an example with the largest variability, which is reflected in a larger SD in relative terms (Table 1). Even in this case, the data exhibit excellent consistency that allows accurate comparison with the binding curves obtained for other proteins.

Fig. 4.

The Syt7 C₂A domain dominates binding of Syt7 to membranes and clusters liposomes. (A) Titrations of 20 nM Syt7 C₂A domain, C₂B domain and C₂A*B fragment (the * in C₂A*B denotes that the fluorescent probe is attached to C260 of the C₂A domain) with liposomes lacking PIP₂ in the presence of 1 mM Ca²⁺. Binding was monitored from the FRET developed between a donor BODIPY-FL probe attached to the proteins and a rhodamine acceptor present in the liposomes. Each point represents the average of at least three measurements performed with different liposome preparations. (B) The same titrations shown in A but changing the x axis to better show the points of the titrations obtained at low liposome concentrations. (C) Liposome titrations of 20 nM WT Syt7 C₂AB fragment and mutant fragments where the Ca²⁺-binding sites of the C₂A domain (C₂aB*) or C₂B domain (C₂A*b) were disrupted with D225A,D227A,D233A or D357A,D359A mutations. All of the data in A–C were fit to a Hill function. (D–F) Distribution of particle size measured by DLS on samples containing liposomes alone (D) or liposomes in the presence of the Syt7 C₂A domain (E) or Syt7 C₂B domain (F). In E and F, the diagrams show superpositions of data obtained in the absence (green bars) or presence (red bars) of Ca²⁺.

Fig. 7.

Subtle residue substitutions underlie in part the stronger membrane binding affinity of the Syt7 C₂A domain compared with the Syt1 C₂A domain. (A) Ribbon diagrams of the Syt7 C₂A domain (cyan) and Syt1 C₂A domain (orange) showing the Cα and side chain atoms of residues that are different in their Ca²⁺-binding regions and were mutated to analyze the basis for their different membrane affinities (oxygens are in red, nitrogens in blue, sulfur atoms in yellow, and carbon atoms in cyan for Syt7 and in orange for Syt1). (B and C) Titrations of 20 nM WT and mutant Syt7 C₂A domains with liposomes in the presence of 1 mM Ca²⁺. Binding was monitored from the FRET developed between a donor BODIPY-FL probe attached to the proteins and a rhodamine acceptor present in the liposomes. Each point represents the average of at least three measurements performed with different liposome preparations. (D) Analogous liposome titrations of 100 nM WT and mutant Syt1 C₂A domains. (E) Analogous liposome titrations of 100 nM WT and mutant Syt1 and Syt7 C₂A domains. Note that the data obtained for WT and M173F,K236R Syt1 C₂A domains (orange and gray circles, respectively) are the same as those shown in D, but the scale of the x axis is different to allow comparison with the data obtained for the WT and mutant Syt7 C₂A domains. All of the data were fit to a Hill function.

Table 1.

Apparent liposome affinities of Ca²⁺-saturated Syt7 and Syt1 C₂ domain fragments

Protein fragment	KD (µM)	SD	h	SD
Syt7 C2A	2.40	0.54	2.18	0.41
Syt7 C2B	34.97	2.92	1.02	0.09
Syt7 C2A*B	4.12	0.80	3.1	0.52
Syt7 C2A*b	4.29	0.18	1.74	0.26
Syt7 C2aB*	8.06	0.39	2.03	0.26
Syt7 C2A F167M,R231K	8.68	0.44	1.61	0.17
Syt1 C2A	551.1	134.26	1.34	0.159
Syt1 C2B	637.37	146.89	1.78	0.38
Syt1 C2A*B	24.05	1.77	2.15	0.46
Syt1 C2A*b	245.5	4.49	1.95	0.24
Syt1 C2aB*	267.23	38.96	2.09	0.25
Syt1 C2A M173F,K236R	95.48	25.9	1.51	0.3

The listed apparent K_Ds and Hill coefficients (h), as well as their SDs, were obtained from fitting the FRET data obtained in liposome titrations of the indicated Ca²⁺-saturated Syt7 and Syt1 C₂ domain fragments. At least three titrations were performed for each fragment. K_Ds and h values were derived for each titration, and the averages as well as SDs were calculated. The protein concentrations were 20 nM for all Syt7 fragments and 100 nM for all Syt1 fragments.

We first used liposomes containing a lipid composition that resembles that of physiological membranes (33) but did not include PIP₂ (Fig. 4). Importantly, liposome titrations revealed that the Syt7 C₂A domain binds with much higher affinity than the Syt7 C₂B domain and a similar affinity to that observed for Syt7 C₂AB (Fig. 4 A and B and Table 1). In fact, the curves observed for Syt7 C₂AB were consistently shifted to the right (i.e., larger K_D) compared with those obtained for the Syt7 C₂A domain (Fig. 4B), but this observation does not necessarily imply a higher affinity of the C₂A domain versus C₂AB. Considering that a C₂ domain can bind multiple phospholipids and has a much larger area than that of a phospholipid headgroup, that the liposomes contained 17% of negatively charged phospholipids, that only half of the lipids are in the outer leaflets, and that molecular crowding limits the availability of more than 70% of the liposome surface (20), the amounts of phospholipids required to observed maximum FRET for the Syt7 C₂A domain and C₂AB fragment suggest that we are observing saturation binding curves. Hence, the right shift of the curves observed for the Syt7 C₂AB fragment compared with the C₂A domain most likely reflects the fact that C₂AB is larger than C₂A, thus requiring more lipids for quantitative binding. We also performed titrations with Syt7 C₂AB fragments containing mutations that disrupt the Ca²⁺-binding sites of the C₂A domain (denoted C₂aB* because C₂A cannot bind Ca²⁺ and the fluorescent probe is on C₂B) or the C₂B domain (C₂A*b). Interestingly, impairing Ca²⁺-binding to the C₂A domain markedly decreased the membrane affinity of Syt7 C₂AB, whereas disrupting Ca²⁺ binding to the C₂B domain had little effect (Fig. 4C and Table 1). These results show that the C₂A domain dominates binding of Syt7 to membranes and reveal a clear correlation with the finding that disrupting Ca²⁺ binding to the C₂A domain has a much stronger effect on the function of Syt7 in asynchronous release than abrogating Ca²⁺ binding to the C₂B domain (9).

The Ca²⁺-Bound Syt7 C₂A Domain Has Higher Membrane-Bridging Activity than the Ca²⁺-Bound C₂B Domain.

The finding that the Syt1 C₂B domain can bridge two membranes, bringing them into close proximity, suggested that this property could be crucial for Syt1 function and could explain the critical functional importance of Ca²⁺-binding to the Syt1 C₂B domain (20, 22). Thus, we also tested the membrane bridging activity of the Syt7 C₂ domains. For this purpose, we analyzed the formation of liposome clusters due to membrane bridging using dynamic light scattering (DLS). Isolated liposomes had radii below 100 nm (Fig. 4D) that did not change substantially upon addition of the Syt7 C₂A domain or C₂B domain in the absence of Ca²⁺ (Fig. 4 E and F). However, in the presence of Ca²⁺, the Syt7 C₂A domain caused a dramatic increase in particle size reflecting robust vesicle clustering, whereas the C₂B domain had practically no activity (Fig. 4 E and F). These results again correlate with the electrophysiological data and suggest that the preponderant role of Ca²⁺ binding to the Syt7 C₂A domain in triggering asynchronous release may arise not only from its the higher membrane affinity, compared with the C₂B domain, but also from its stronger ability to bridge two membranes.

The Ca²⁺-Bound Syt1 C₂B Domain Binds to PIP₂-Containing Membranes with Higher Affinity than the Ca²⁺-Bound Syt1 C₂A Domain.

Phospholipid binding to the Syt1 C₂ domains has been extensively investigated (1, 45, 46), but most studies analyzed the Ca²⁺ dependence of lipid binding, and analysis of phospholipid binding under Ca²⁺ saturating conditions was mostly focused on Syt1 C₂AB fragments (39). To compare the properties of the Syt1 C₂ domains using the same methodology used for our experiments with the Syt7 C₂ domains and examine whether the asymmetry in the functional importance of Ca²⁺-binding to the Syt1 C₂ domains may also be reflected in their membrane affinities, we performed liposome titrations of the Ca²⁺-saturated Syt1 C₂A domain, C₂B domain, and C₂AB fragment. Because the Syt1 C₂ domains only contain one cysteine side chain (C277) and this side chain is buried, we mutated this cysteine to alanine and introduced single cysteine mutations in regions of the C₂A and C₂B domain that have not been implicated in lipid binding (Fig. S2A). As in the case of Syt7 C₂AB, placing the fluorescent probe on the C₂A or C₂B domain did not substantially affect the phospholipid curves observed for the Syt1 C₂AB fragment (Fig. S2C).

Our data showed that the Ca²⁺-saturated Syt1 C₂A and C₂B domains bind to the liposomes with similar affinities, which were much weaker than that observed for the Syt7 C₂A domain and were far from leading to saturation binding curves (Fig. 5A). Syt1 C₂AB bound much more avidly than the individual Syt1 C₂ domains (Fig. 5 A and B and Table 1), showing that they cooperate strongly in membrane binding. Mutation of the Ca²⁺-binding sites of the C₂A or C₂B domain considerably impaired phospholipid binding to the Syt1 C₂AB fragment, but to similar extents (Fig. 5C) that do not correlate with the stronger disruption of Syt1 function caused by mutation of the C₂B domain Ca²⁺-binding sites.

Fig. 5.

The Syt1 C₂ domains have similar membrane affinities and cooperate in membrane binding. (A) Titrations of 100 nM Syt1 C₂A domain, C₂B domain and C₂AB* fragment (the * in C₂AB* denotes that the fluorescent probe is attached to the C₂B domain) with liposomes lacking PIP₂ in the presence of 1 mM Ca²⁺. Binding was monitored from the FRET developed between a donor BODIPY-FL probe attached to the proteins and a rhodamine acceptor present in the liposomes. Each point represents the average of at least three measurements performed with different liposome preparations. (B) The same titrations shown in A but changing the x axis to better show the points of the titrations obtained at low liposome concentrations. (C) Liposome titrations of 100 nM WT Syt1 C₂AB* fragment and mutant fragments where the Ca²⁺-binding sites of the C₂A domain (C₂aB*) or C₂B domain (C₂A*b) were disrupted with D178A,D230A,D232A or D309A,D363A,D365A mutations. All of the data were fit to a Hill function.

Because PIP₂ is known to increase the membrane affinity of the C₂B domain but not the C₂A domain (47), we performed additional experiments with liposomes that contained the same lipid composition except for the inclusion of 1% PIP₂. Our data confirmed that, indeed, the presence of PIP₂ in the liposomes did not alter the affinity for the Syt1 C₂A domain but dramatically increased the affinity for the Syt1 C₂B domain (Fig. 6 A and B). Hence, in the presence of PIP₂, the stronger membrane affinity of the Ca²⁺-saturated Syt1 C₂B domain compared with the C₂A domain does correlate with the functional effects of disrupting Ca²⁺ binding. These results led us to also test the effects of including PIP₂ on the membrane affinity of the Syt7 C₂ domains. PIP₂ clearly increased the membrane affinity of the Syt7 C₂B domain but did not appear to alter liposome binding to the Syt7 C₂A domain (Fig. 6 C and D). However, the latter result needs to be interpreted with caution because the titrations with Syt7 C₂A domain most likely reflect saturation binding (see above), which is expected to prevent observation of any increase in membrane affinity due to inclusion of PIP₂. Note also that, even in the presence of PIP₂, the Ca²⁺-saturated Syt7 C₂A domain still binds membranes with higher affinity than the Syt7 C₂B domain (Fig. S3).

Fig. 6.

PIP₂ enhances the membrane affinity of the Syt1 and Syt7 C₂B domains. Titrations of 100 nM Syt1 C₂A domain (A) or C₂B domain (B), or 20 nM Syt7 C₂A domain (C) and C₂B domain (D) with liposomes lacking (black circles) or containing (red circles) PIP₂ in the presence of 1 mM Ca²⁺. Binding was monitored from the FRET developed between a donor BODIPY-FL probe attached to the proteins and a rhodamine acceptor present in the liposomes. Each point represents the average of at least three measurements performed with different liposome preparations. All of the data were fit to a Hill function.

Fig. S3.

The Syt7 C₂A domain binds to PIP₂-containing liposomes with higher affinity than the Syt7 C₂B domain. Titrations of 20 nM Syt7 C₂A domain (red circles) or C₂B domain (blue circles) with liposomes containing PIP₂ in the presence of 1 mM Ca²⁺. Binding was monitored from the FRET developed between a donor BODIPY-FL probe attached to the proteins and a rhodamine acceptor present in the liposomes. Each point represents the average of at least three measurements. All of the data were fit to a Hill function. The data are the same shown in Fig. 6 C and D with liposomes containing PIP₂.

Subtle Residue Substitutions Underlie in Part the Differential Membrane Affinities of the Syt1 and Syt7 C₂A Domains.

Overall, the data presented above and other previous results suggest that phospholipid binding affinity is a key factor that determines the functional importance of Syt C₂ domains. When comparing the properties of Syt1 and Syt7, it is particularly striking that the Syt7 C₂A domain has such a higher membrane-binding affinity than the Syt1 C₂A domain (Figs. 4 and 5), with an apparent K_D at least 200 times smaller (Table 1). This finding is not surprising given previous results showing a higher apparent Ca²⁺ affinity (40) and a slower membrane dissociation rate (41) for the Syt7 C₂A domain compared with the Syt1 C₂A domain, but it seems difficult to rationalize considering the similarities of the 3D structures, the intrinsic Ca²⁺ affinities, and the sequences of these domains. Given the key functional importance of membrane binding, elucidating the sequence determinants of membrane affinity is critical to understand the functional differences between Syts, and the C₂A domains of Syt1 and Syt7 provide an ideal benchmark to address this question.

We hypothesized that differences in the membrane affinities of the two domains arise because of cumulative small or moderate effects caused by residues that are distinct in the Ca²⁺-binding regions of the Syt1 and Syt7 C₂A domains. To test this hypothesis, we prepared six single-site mutants of the Syt7 C₂A domain where single residues were replaced by the (distinct) residue present in the homologous position of the Syt1 C₂A domain (Fig. 7A; see also Fig. 1E). Titrations with liposomes under Ca²⁺ saturating conditions did not reveal substantial differences between the membrane affinities of any of the single mutants and the WT Syt7 C₂A domain (Fig. 7 B and C). We reasoned that perhaps some of these single site mutations decreased the membrane affinity of the Syt7 C₂A domain but, since the WT affinity is so high, a modest decrease in affinity may not be observable because the experiments still yield saturation binding curves. To try to cause a stronger decrease in membrane affinity that may be observable, we combined two of the single mutations (F167M and R231K) into a double mutant. Note that these mutations would normally be considered conservative, but they might decrease membrane affinity to some extent because arginines can form hydrogen bonds more efficiently with phosphate groups than lysines, and phenylalanines have a larger hydrophobic surface than methionines. Indeed, the double F167M,R231K mutation markedly impaired liposome binding to the Syt7 C₂A domain (Fig. 7C and Table 1).

If the differences in membrane affinities between the Syt1 and Syt7 C₂A domains arise in part because of the nature of these two side chains, rather than differences in 3D structure or intrinsic Ca²⁺ affinities, we could expect that making the reverse mutations in the Syt1 C₂A domain would increase its membrane affinity. Indeed, single M173F and K236R mutations both increased the membrane affinity of the Syt1 C₂A domain, and a double M173F,K236R mutation revealed an additional increase in membrane affinity that reflects the cumulative effects of the two mutations (Fig. 7D). Titrations of Syt1 and Syt7 fragments were generally performed at 100 nM and 20 nM protein concentrations, respectively, to better dissect the differences in membrane affinities between the fragments of each isoform. To be able to directly compare the membrane affinities of the Syt1 and Syt7 C₂A domains in this set of experiments, we repeated the titrations of the WT and F167M,R231K mutant Syt7 C₂A domain at 100 nM protein concentration. Comparison of the results with those obtained for the WT and M173F,K236R Syt1 C₂A domains (Fig. 7E) showed that the Syt7 double mutant still has a higher affinity than the Syt1 double mutant, but it is clear that the double mutations helped to close the large gap existing between the membrane affinities of the WT Syt1 and Syt7 C₂A domains. Thus, the ratio between the apparent K_Ds was reduced from more than 200 to 11 (Table 1). These results show that apparently conservative residue substitutions in the membrane-binding regions of C₂ domains can have significant effects on membrane affinity that can accumulate to yield dramatic biochemical and functional differences.

Discussion

Much has been learned about the neurotransmitter release machinery and, among its components, Syts have been studied extensively because of their roles as Ca²⁺ sensors. Although Syt function in release is broadly understood and plausible models for the mechanism of action of Syts in fusion have emerged, the structural determinants that shape the functional differences between fast and slow Syts remain unclear. Particularly intriguing was the observation that disrupting Ca²⁺ binding to the Syt7 C₂A domain impaired its function in asynchronous release much more strongly than disruption of the C₂B domain Ca²⁺ binding sites, in contrast to the opposite effects caused by mutations in the Ca²⁺-binding sites of the Syt1 C₂ domains on synchronous release (9). The results presented here strongly suggest that the predominant role of the Syt7 C₂A domain in asynchronous release, compared with the C₂B domain, arises because of its stronger contribution to membrane binding, and perhaps to bridging the vesicle and plasma membranes. More generally, our data suggest that the contribution to membrane affinity is a major determinant of the relative functional importance of the C₂ domains of Syts. Furthermore, our results illustrate how small or moderate effects from what appear to be subtle residue substitutions can accumulate to yield dramatic functional differences in Syt C₂ domains.

Ca²⁺-dependent phospholipid binding was clearly shown to be critical for Syt1 function based on the correlations that were established between the effects of mutations that increase or decrease the apparent Ca²⁺ affinity of Syt1 in the presence of phospholipids and their functional effects on neurotransmitter release (6, 19). This activity is also believed to underlie at least in part the functions of other Ca²⁺-dependent Syts (5). Hence, our finding that Ca²⁺-dependent phospholipid binding to Syt7 is dominated by the C₂A domain (Fig. 4) provides a natural explanation for the stronger disruption of asynchronous release caused by mutations in the Ca²⁺-binding sites of the Syt7 C₂A domain compared with those induced by analogous mutations in the Syt7 C₂B domain (9). Moreover, although the Syt1 C₂A and C₂B domains bind with similar affinities to membranes lacking PIP₂ (Fig. 5), the Syt1 C₂B domain clearly dominates membrane binding in the presence of PIP₂ (Fig. 6), and mutations in the Ca²⁺-binding sites of the C₂B domain impair membrane binding to Syt1 C₂AB more strongly than mutations in the C₂A domain Ca²⁺ binding sites, in correlation with the functional effects of these mutations (33, 39, 47). A remaining unresolved key issue, however, is the precise role of Ca²⁺-binding to the Syt1 C₂B-domain. Since the C₂B-domain mutant Syt1 abolishes even asynchronous release in a dominant-negative fashion, although it does not normally mediate such release, such mutant Syt1 likely locks the prefusion complex at the active zone in a Ca²⁺-unresponsive state, possibly by preventing the Ca²⁺-dependent unlocking of a complexin-induced prefusion intermediate (48). Although the lack of rescue of Ca²⁺-dependent release by C₂B-domain mutant Syt1 can thus not be used to argue that Ca²⁺-dependent phospholipid binding by the Syt1 C₂B-domain is of central importance in release, it seems likely that the “unlocking” of the prefusion complex is actually mediated by Ca²⁺-dependent phospholipid binding by the Syt1 C₂B-domain. In this sense, Ca²⁺-dependent phospholipid binding would constitute an intrinsic aspect of Syt1 function even if it did not underlie the differential effects of mutations in the Ca²⁺-binding sites of the Syt1 C₂ domains, which are opposite to those observed for Syt7.

The intrinsic Ca²⁺ affinities of the C₂A and C₂B domains are similar for both Syt1 and Syt7. What appears to be critical for their relative functional importance is the membrane affinity of the Ca²⁺-saturated C₂ domain, which is indirectly reflected in the apparent Ca²⁺ affinities observed in Ca²⁺-dependent phospholipid binding experiments. Thus, the very high affinity of the Ca²⁺-saturated Syt7 C₂A domain for membranes, dramatically stronger than that of the Syt1 C₂A domain (Figs. 4, 5, and 7), was already reflected in its higher apparent Ca²⁺ affinity (40). Our data show that this dramatic difference in membrane affinity arises in part from cooperative effects resulting from apparently subtle substitutions, including a methionine-to-phenylalanine substitution that is expected to increase hydrophobic interactions with the acyl chains of the membranes and a lysine-to-arginine substitution that is expected to enhance hydrogen bonding to phospholipid headgroups (Fig. 7). It is likely that the overall basic nature of the Syt7 C₂A domain, compared with the acidic nature of the Syt1 C₂A domain (see below), also contributes to the stronger affinity for negatively charged membranes. Regardless of this possibility, the effects of the F167M,R231K mutation in the Syt7 C₂A domain and the reverse M173F,K236R mutation in the Syt1 C₂A domain on membrane binding provide an emphatic illustration of how such subtle mutations can have profound consequences for the biochemical properties of C₂ domains and proteins in general.

Another biochemical property that is intrinsically related to phospholipid binding and could underlie a dominant role for the Syt1 C₂B domain is its ability to bring two membranes into close proximity in a Ca²⁺-dependent manner (20, 22), which arises because of binding of its Ca²⁺-binding loops at the top of the β-sandwich to one membrane and interactions of R398 and R399 at the bottom of the sandwich with another membrane, and may also be aided by its basic nature (#basic − #acidic residues = 5). However, there was a priori no clear reason why this property might have switched to the C₂A domain for Syt7. Thus, although the Syt7 C₂A domain (#basic − #acidic residues = 4) is more basic than the Syt1 C₂A domain (#basic − #acidic residues = −2), the Syt7 C₂B domain is even more basic (#basic − #acidic residues = 10) and also contains arginines at the bottom of the β-sandwich. Nevertheless, our DLS measurements (Fig. 4 D–F) show that the Syt7 C₂A domain does have a much stronger activity in bridging two membranes than the Syt7 C₂B domain. The sequence determinants for this unexpected activity of the Syt7 C₂A domain are currently unclear, but the switch in this property, which is characteristic of the C₂B domain in Syt1 and of the C₂A domain in Syt7, provides a correlation with the critical importance of Ca²⁺ binding to the Syt1 C₂B domain for synchronous release and of Ca²⁺ binding to the Syt7 C₂A domain for asynchronous release. Because Ca²⁺-dependent bridging of the two membranes could be crucial to trigger Ca²⁺-dependent fusion, in cooperation with the SNAREs, it is tempting to speculate that this activity is key to determine the relative functional importance of Ca²⁺ binding to the Syt1 and Syt7 C₂ domains. However, it is worth noting that there are other mechanisms by which Ca²⁺-dependent membrane binding could help to cause membrane fusion, for example by creating tension in the membranes (1), perturbing the packing of the bilayers, or inducing membrane curvature (20, 49).

The notion that membrane affinity or the ability to bridge two membranes might determine the relative contributions of the C₂ domains to Syt function does not imply that other interactions of Syts, such as those with PIP₂ or the SNARE complex, are not important. However, it is worth noting that the primary regions that have been implicated in these interactions [the polybasic region and one face of the β-sandwich of the Syt1 C₂B domain (23–26)] do not include the Ca²⁺-binding regions, even though Ca²⁺ enhances these interactions. At the high local concentrations of these reagents existing in the primed state of a synaptic vesicle, PIP₂ or SNARE complex binding to the C₂B domain (of either Syt1 or Syt7) could occur in the absence of Ca²⁺ and could help setting up Ca²⁺-dependent membrane fusion, but the key action of Ca²⁺ binding to Syts that triggers neurotransmitter release is most likely the stimulation of interactions of the Syt Ca²⁺-binding loops with one or both membranes. Binding of Syt1 to the SNARE complex could facilitate this action, as SNARE complex and membrane binding to Syt1 cooperate with each other (50), but the nature of such cooperation remains unclear given the diversity of Syt1–SNARE complex binding modes observed in recently reported structures (23, 26). Moreover, the interactions observed in these structures involve primarily the Syt1 C₂B domain and the key residues involved in binding are not conserved in the Syt7 C₂A domain. Hence, the mechanisms of coupling of Syt1 and Syt7 with the SNAREs may be more complex than currently envisioned.

Materials and Methods

Recombinant Proteins.

N-terminal GST fusion proteins of rat Syt7 fragments 134–262 (C₂A domain), 263–403 (C₂B domain), and 134–403 (C₂AB fragment) were expressed in Escherichia coli BL21 (DE3) cells. The bacteria were grown in Lennox L Broth media at 25 °C for 16 h and induced with 0.5 mM isopropyl β-d-1-thiogalactopyranoside. Cells were resuspended in a buffer containing 40 mM Tris⋅HCl (pH 8.0), 1 M NaCl, 2 mM DTT, 1% Triton and Sigma protease inhibitor mixture (P2714-1BTL), and lysed using an Avestin EmulsiFlex-C5 homogenizer. The soluble fraction of the cell lysate was collected after centrifugation at 48,000 × g for 30 min and incubated with 100 mg protamine sulfate (Sigma-Aldrich) per 35 mL supernatant for 1 h at room temperature. The mixture was spun again at 48,000 × g for 30 min and the soluble fraction was incubated with Glutathione Sepharose 4B (GE Healthcare) at 4 °C for 16 h. The resin was washed with buffer containing 40 mM Tris⋅HCl (pH 8.0) and 200 mM NaCl (buffer A), buffer A + 50 mM CaCl₂, and buffer A + 50 mM CaCl₂ + 1 M NaCl. Remaining nucleic acid contaminants bound to synaptotagmin fragments were then cleared with benzonase treatment (40 units per milliliter of solution, corresponding to ∼1,000 units per liter of cell culture) in 50 mM Tris (pH 8.0), 2 mM MgCl₂ for 2 h at room temperature with gentle rotation of the beads. Thrombin cleavage of the GST tag was carried out at room temperature for 3 h in 10 mL thrombin cleavage buffer (50 mM Tris⋅HCl pH 8.0, 150 mM NaCl, 2.5 mM CaCl₂) and 0.08 mg/mL thrombin (Sigma-Aldrich). Eighteen residues from the GST tag (SPGISGGGGGILDSMGRL) remained at the N terminus of Syt7 fragments after thrombin cleavage. Syt7 fragments were further purified with an ion exchange Source S column [buffers: 50 mM NaAc (pH 6.2), 5 mM CaCl₂; and 50 mM NaAc (pH 6.2), 5 mM CaCl₂, 1 M NaCl]. All proteins were further purified using size-exclusion chromatography on a Superdex 75 16/60 column using buffer containing 20 mM Hepes (pH 7.4), 125 mM KCl, and 1 mM TCEP for Syt7 C₂A and C₂B domains, and buffer containing 20 mM Hepes (pH 7.4), 500 mM NaCl, 1 mM TCEP for the Syt7 C₂AB fragment. Following chromatography, the Syt7 C₂AB fragment was buffer exchanged into 20 mM Hepes (pH 7.4), 125 mM KCl, and 1 mM TCEP by serial concentration and dilution. Syt1 C₂A domain, C₂B domain and C₂AB fragment were purified as described previously (15, 20, 51, 52). All mutations were performed using the QuikChange site-directed mutagenesis kit (Stratagene). These included the cysteine mutations indicated in the text and the following mutations to disrupt the Ca²⁺-binding sites: for Syt7 C₂A, D225A,D227A,D233A; for Syt7 C₂B, D357A,D359A; for Syt1 C₂A, D178A,D230A,D232A; for Syt1 C₂B, D309A,D363A,D365A.

Crystallization and X-Ray Data Collection.

Purified Syt7 C₂A domain and C₂AB fragment were concentrated to 20 mg/mL in a 20 mM Hepes (pH 7.4), 125 mM KCl, 1 mM CaCl₂ buffer, and used for crystal screening. Crystals of Syt7 C₂A domain were grown in 21% t-butanol, 0.1 M Tris (pH 8.5), 0.1 M CaCl₂ using the hanging-drop method at 20 °C, and the crystals were cryoprotected by transferring to a final solution of 15% (vol/vol) glycerol, 21% t-butanol, 0.1 M Tris (pH 8.5), 0.1 M CaCl₂, then flash-cooled in liquid nitrogen. Crystals of Syt7 C₂AB fragment were grown in 20% PEG 3350, 0.2 M Li₂SO₄, 0.1 M Bis⋅Tris (pH 6.0), 0.1 M CaCl₂ using the hanging-drop method at 20 °C, and the crystals were cryoprotected by transferring to a final solution of 20% (vol/vol) ethylene glycol, 22% PEG 3350, 0.2 M Li₂SO₄, 0.1 M Bis⋅Tris (pH 6.0), 0.1 M CaCl₂.

Data were collected at the APS beamline 19-ID (SBC-CAT) at the Advanced Photon Source (Argonne National Laboratory, Argonne, IL) at 100 K, and were indexed, integrated and scaled using HKL-3000 (53), with applied corrections for effects resulting from absorption in a crystal and for radiation damage (54, 55), the calculation of an optimal error model, and corrections to compensate the phasing signal for a radiation-induced increase of nonisomorphism within the crystal (56, 57). Syt7 C₂A crystals exhibited the symmetry of space group P6₅ with unit cell parameters of a = 55.30 Å, c = 89.81 Å, and contained one molecule of Syt7 C₂A per asymmetric unit, with a solvent content of 50%. Syt7 C₂AB fragment crystals belonged to space group P2₁ with unit cell parameters of a = 60.64 Å, b = 75.06 Å, c = 70.98 Å, β = 114.5°, and contained two molecules each of Syt7 C₂AB fragment per asymmetric unit, with a solvent content of 45%. Syt7 C₂A crystals diffracted isotropically to a d_min of 1.70 Å when exposed to synchrotron radiation. Syt7 C₂AB fragment crystals diffracted anisotropically to a d_min of 2.25 Å when exposed to synchrotron radiation, and displayed strong translational noncrystallographic symmetry characteristic of a pseudo B-centered lattice, as evidenced by a peak in the Patterson function at approximately (1/2, 0, 1/2) with a height ≈49.9% of the origin peak. This resulted in numerous systematically weak reflections in the diffraction pattern, namely for the following reflection conditions: hkl, h + l = 2n + 1; hk0, h = 2n + 1; 0kl, l = 2n + 1; h0l, h + l = 2n + 1; 00l, l = 2n + 1; and h00, h = 2n + 1. Data collection statistics are provided in Table S1.

Phase Determination and Structure Refinement.

Phases for the Syt7 C₂A domain were obtained via molecular replacement in the program Phaser (58) using a search model derived from the NMR structure of the Ca²⁺-free Syt7 C₂A domain (PDB ID code 2D8K). Phases for the Syt7 C₂AB fragment were obtained via molecular replacement in the program Phaser using search models derived from the structure of the Syt7 C₂A domain described herein and the crystal structure of the Syt7 C₂B domain (PDB ID code 3N5A) (29). Additional protein residues and calcium ions were manually modeled into the electron density maps via the program COOT (59). Refinement was performed using the program Phenix (60) with a random 5% of all data set aside for an R_free calculation. The final model for Syt7 C₂A (R_work = 14.4%, R_free = 18.1%) contained 138 residues and three Ca²⁺ ions. The final model for Syt7 C₂AB fragment (R_work = 21.4%, R_free = 25.6%) contained 262 residues in chain A, 263 residues in chain B, and 8 Ca²⁺ ions, or 2 Ca²⁺ ions per C₂ domain. The presence of the pseudocentering operator in the Syt7 C₂AB fragment lattice translates into a systematic modulation of the observed intensities (61, 62) and thus resulted in lower than 100% completeness in the highest resolution shells between 2.70 and 2.25 Å. The L-value twinning test (63), which is not influenced by the presence of pseudocentering, shows no significant deviation from the expected values for untwinned data. Ramachandran plots generated with MolProbity (64) indicated no outliers in the Ramachandran plot for Syt7 fragments. Data collection and structure refinement statistics are summarized in Table S1. Coordinates and structure factors have been deposited in the Protein Data Bank with ID codes 6ANJ (Syt7 C₂A domain) and 6ANK (Syt7 C₂AB fragment).

Isothermal Titration Calorimetry.

The protein solutions were dialyzed twice against the ITC buffer (20 mM Hepes, pH 7.4, 125 mM KCl, 1 mM TCEP). The buffer was pretreated with Chelex-100 (Bio-Rad; calcium binding constant ≈4.6 × 10³ M⁻¹) to remove residual calcium ions bound with moderate affinity. The Chelex-100 beads were first washed with water and then added directly to the dialysis buffer for 2 h. The Chelex-100 beads were removed by filtration. The buffer pH was adjusted to 7.4 after filtration. ITC experiments were performed using a VP-ITC system (MicroCal) at 37 °C in ITC buffer. The protein solution (200 µM) was loaded into the sample cell, and CaCl₂ solution (20 mM for the Syt7 C₂A domain and 40 mM for the Syt7 C₂B domain) was loaded in the syringe. The CaCl₂ solution was prepared by diluting a 1 M stock solution with ITC buffer to the appropriate concentration. CaCl₂ was injected at 3-μL volumes 100 times, and the heat evolved per injection was measured. All ITC data were analyzed using the Microcal Origin ITC software packet.

Labeling Proteins with BODIPY-FL Maleimide.

Purified proteins at a concentration of 100 µM were incubated with 1 mM BODIPY FL N-(2-aminoethyl) maleimide (Molecular Probes) for 2 h at room temperature or for 16 h at 4 °C with rotation. The reaction was quenched by adding 10 mM DTT to the mixture. The unlabeled dye was separated from the labeled protein by ion exchange chromatography on a Source S column (GE Healthcare) [buffers: 50 mM NaAc (pH 6.2), 5 mM CaCl₂; and 50 mM NaAc (pH 6.2), 5 mM CaCl₂, 1 M NaCl].

Preparation of Phospholipid Vesicles.

For preparation of phospholipid vesicles, 1,2-dioleoyl-sn-glycero-3-[phospho-L-serine] (DOPS), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), Cholesterol, PIP₂, l-α-phosphatidylinositol (PI), and 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N- (lissamine rhodamine B sulfonyl) (ammonium salt; Liss Rho PE; Avanti Polar Lipids) in chloroform were mixed in a glass test tube in a desired ratio, and chloroform was evaporated using a dry nitrogen stream. The lipids were placed in a vacuum chamber overnight for complete removal of organic solvent. Lipid films were hydrated with 20 mM Hepes (pH 7.4) 150 mM KCl buffer in an appropriate volume yielding 10 mM lipids. Lipids were vortexed for >5 min, then frozen and thawed five times. Large unilamellar vesicles were prepared by extruding the hydrated lipid solution through 80 nm polycarbonate membranes 23 times using an Avanti Mini-Extruder. For most FRET assays, we used liposomes containing 41% POPC, 31% POPE, 1% Liss Rho PE, 1% cholesterol, 12% DOPS, and 5% PI. PIP₂-containing liposomes were made by mixing 40% POPC, 31% POPE, 1% Liss Rho PE, 1% cholesterol, 12% DOPS, 1% PIP₂, and 5% PI.

Phospholipid Binding Assays.

The FRET experiments were performed at 22 °C on a PTI spectrofluorimeter. All of the experiments were carried out in a buffer containing 20 mM Hepes (pH 7.4) and 150 mM KCl. The labeled protein (20 nM in case of Syt7 fragments and 100 nM in the case of Syt1 fragments, unless otherwise indicated) was mixed with variable amounts of liposomes and 1 mM CaCl₂. Note that, although the intrinsic Ca²⁺ affinities of the Syt1 and Syt7 C₂ domains are weak, the apparent Ca²⁺ affinities in the presence of phospholipids are much higher (6, 40), and hence 1 mM Ca²⁺ is sufficient for Ca²⁺ saturation. These samples were excited at 485 nm and the emission spectra from 500 to 600 nm were acquired. A fluorescence spectrum of the same sample after the addition of 2 mM EDTA was also collected. To correct for the bleed-through from Rhodamine at the emission maxima of BODIPY-FL, a spectrum of liposomes alone was collected and subtracted from the corresponding spectrum acquired in the presence of the protein. The FRET efficiency (E) was calculated with the formula: E = (I_EDTA − I_Ca)/I_EDTA, where I_Ca is the fluorescence intensity of the protein-liposome sample at the emission maxima of BODIPY-FL (512 nm) in the presence of 1 mM Ca²⁺ and I_EDTA is the fluorescence intensity of the protein-liposome sample at 512 nm after the addition of 2 mM EDTA. The binding data were fitted to a Hill function using GraphPad PRISM7.

Dynamic Light Scattering.

The clustering ability of Syt7 C₂A and C₂B domains was measured by DLS using a DynaPro (Wyatt Technology) instrument equipped with a temperature controlled Microsampler. Liposomes containing 70% POPC and 30% DOPS were used for the DLS experiments. Samples were prepared by mixing 100 μM liposomes and 2 μM protein in 20 mM Hepes, pH 7.4, 125 mM KCl buffer. DLS measurements were made at 22 °C.

NMR Spectroscopy.

¹H-¹⁵N HSQC spectra were acquired at 25 °C on Agilent DD2 spectrometers operating at 800 or 600 MHz. Samples contained 30–50 μM uniformly ¹⁵N-labeled proteins dissolved in buffer with 20 mM Hepes (pH 7.4), 125 mM KCl containing 5% D₂O. All data were processed with NMRpipe (65) and analyzed with NMRView (66).