The tendency of the Syt1 C₂ domains to precipitate with the SNARE complex formed by the SNARE motifs of synaptobrevin, syntaxin-1 and SNAP-25, particularly in the presence of Ca²⁺, hindered analysis of their interactions in solution by NMR spectroscopy (Dai et al., 2007; Zhou et al., 2013). Inclusion of 125 mM KSCN allowed NMR analyses in the presence of Ca²⁺ (Brewer et al., 2015), but KSCN might have favored binding of the SNARE complex to the C₂B domain polybasic region over binding to the primary interface. In attempts to solve the solubility problem without addition of KSCN, we found that including the complexin-1 fragment that we used for crystallization with the SNARE complex [residues 26–83; Cpx1(26-83)] (Chen et al., 2002) dramatically improved the solubility of mixtures containing the Ca²⁺-bound C₂B domain and the SNARE complex (Figure 1—figure supplement 1). Taking advantage of this observation, we analyzed the binding sites of the SNARE complex on the Syt1 C₂B domain in the presence of Cpx1(26-83) using NMR spectroscopy. Note that the Syt1 C₂ domains do not bind to complexin-1 under the conditions that we used for the NMR experiments and that Cpx1(26-83) binds to the SNARE complex with high affinity (K_D ca. 25 nM) (Xu et al., 2013). Hence, the assembly formed by Cpx1(26-83) and the SNARE motifs of synaptobrevin, syntaxin-1 and SNAP-25 can be considered as a single complex that we refer to as CpxSC. For improved sensitivity, we prepared samples of the C₂B domain that were ²H,¹⁵N-labeled and specifically ¹³CH₃-labeled at the Ile δ1 and Met methyl groups (²H,¹⁵N-IM-¹³CH₃-C₂B). First we analyzed the perturbations caused by addition of increasing amounts of CpxSC on transverse relaxation optimized spectroscopy (TROSY)-enhanced ¹H-¹⁵N heteronuclear single quantum coherence (HSQC) spectra of ²H,¹⁵N-IM-¹³CH₃-C₂B. Titrations of ²H,¹⁵N-IM-¹³CH₃-C₂B with CpxSC were performed in the presence of Ca²⁺ at physiological ionic strength and in the absence of Ca²⁺ at lower salt concentration (100 mM KCl) to enhance binding, as Syt1-SNARE complex binding is weaker in the absence than in the presence of Ca²⁺ (Zhou et al., 2013). Similar cross-peak shifts were observed in both sets of experiments (Figure 1H, Figure 1—figure supplement 2), indicating that Ca²⁺ does not affect the binding mode(s).

To analyze the data, we obtained assignments of the ¹H-¹⁵N TROSY-HSQC spectrum of Ca²⁺-free ²H,¹⁵N-IM-¹³CH₃-C₂B (Figure 1—figure supplement 3) based on the assignments obtained previously in a different buffer (Fernandez et al., 2001) and a titration where the buffer composition was gradually changed from one buffer to the other. Mapping of the residues corresponding to the cross-peaks that exhibited the largest shifts induced by CpxSC (Figure 1I) onto the structure of the C₂B domain clearly showed that the residues are clustered on two sides of the β-sandwich, one containing the polybasic region and another corresponding to the primary interface (Figure 1J). There were no significant shifts in cross-peaks from residues in the tripartite interface, which is largely formed by the α-helix spanning residues 384–395, except for residues at the very C-terminus of this helix that are near the primary interface. We could not obtain accurate K_Ds for the interactions of CpxSC with the polybasic region and the primary interface from these data because we did not achieved saturation at the highest concentration of CpxSC that we reached (85 μM) and, at these concentrations, some cross-peaks broadened beyond detection or exhibited odd behavior that did not correlate with the previous titration points (illustrated in Figure 1K by the cross-peaks from two residues corresponding to the polybasic region, K325 and K369, and two residues from the primary interface, V283 and Y339). However, it was clear that the affinities of both binding sites are comparable, and the K_Ds are larger than 20 μM.

We next examined the effects of mutations in the two binding sites of the C₂B domain for the SNARE complex, using ²H,¹⁵N-IM-¹³CH₃-labeled samples of C₂B domain mutants. Since the C₂B domain Ca²⁺-binding region is not involved in binding and Ca²⁺ did not affect the binding modes, these experiments were performed in the absence of Ca²⁺. We first analyzed the effects of the R322E/K325E mutation that we designed previously to disrupt SNARE complex binding to the polybasic region and strongly impaired neurotransmitter release (Brewer et al., 2015). A titration with CpxSC revealed substantial shifts for selected cross-peaks of the C₂B mutant HSQC spectrum, but the patterns were somewhat distinct from those observed for the WT C₂B domain (compare Figure 2A with 1H). The cross-peaks from the β-strand containing the mutated residues shifted and their positions could not be ascertained. However, we could identify the diagnostic cross-peak from K369, which is in the same face of the C₂B domain as the polybasic region. The position of this cross-peak was unaffected even at 85 μM CpxSC (Figure 2C), confirming that binding to the polybasic region was abolished by the R322E/K325E mutation. We also observed that the diagnostic cross-peaks from the primary interface corresponding to V283 and Y339 shifted faster in the titration of the R322E/K325E mutant than observed for the WT C₂B domain (Figure 2C), which is a natural consequence of the lack of competition with the polybasic region for binding to CpxSC. Interestingly, multiple cross-peaks from the primary interface exhibited much more substantial CpxSC-induced shifts for the mutant than for the WT C₂B domain, including cross-peaks from R398 and I401 (Figure 2A,D), residues that are near V283 (Figure 1E). These results suggest that binding of CpxSC to the primary interface is more extensive for the R322E/K325E mutant C₂B and that the polybasic region not only competes with the primary interface for binding to CpxSC but also hinders full C₂B-SNARE engagement at the primary interface.

R398 and R399 at the bottom of the C₂B domain are in the primary interface (Figure 1E), but the R398Q/R399Q mutation that almost abolishes neurotransmitter release (Xue et al., 2008) did not significantly impair Syt1-SNARE co-IP (Zhou et al., 2015). To investigate the contribution of these tandem arginines to Syt1-SNARE complex binding at the primary interface, we analyzed binding of CpxSC to the R398Q/R399Q mutant C₂B domain. We observed that the CpxSC-induced shifts of the cross-peaks from the primary interface were almost abolished (Figure 2B and expansions for the V283 and Y339 cross-peaks in Figure 2C), showing that the mutation strongly impairs binding of CpxSC to this interface. As expected, CpxSC still bound to the polybasic region of R398Q/R399Q C₂B (e.g. K369 cross-peak, Figure 2C). Since individual R398Q and R399Q mutations also impair release considerably (Xue et al., 2008), we also tested binding of CpxSC to R398Q C₂B and R399Q C₂B mutants. Both single mutations impaired CpxSC binding substantially (Figure 2—figure supplement 1), although to a smaller extent than the double R398Q/R399Q mutation. These results show that both R398 and R399 play important roles in binding of Syt1 to CpxSC through the primary interface.

We also prepared a mutant C₂B domain bearing both the R322E/K325E mutation in the polybasic region and the R398Q/R399Q mutation in the primary interface. Titration with CpxSC did not induce significant shifts in the ¹H-¹⁵N TROSY-HSQC spectrum of the R322E/K325E/R398Q/R399Q C₂B mutant even when CpxSC was added at 85 μM concentration (Figure 2E). Binding of an unlabeled protein or complex to a ¹⁵N-labeled protein is expected to cause not only cross-peak shifts but also decreased cross-peak intensities due to the larger size of the resulting complex compared to the isolated ¹⁵N-labeled protein (Rizo et al., 2012), as illustrated in Figure 2—figure supplement 2A–C by the gradual decreases in intensities of selected cross-peaks of WT C₂B caused by increasing CpxSC concentrations. In contrast, the cross-peak intensities of R322E/K325E/R398Q/R399Q C₂B did not decrease appreciably as increasing concentrations of CpxSC were added. Thus, the ratios of the intensities in the presence of 85 μM CpxSC versus the intensities in the absence of CpxSC for the same selected cross-peaks of R322E/K325E/R398Q/R399Q C₂B were close to 1 (Figure 2—figure supplement 2A–C). Analysis of these ratios for all cross-peaks showed a relatively homogeneous distribution, with an average ratio of 0.947 and some natural variability due to the noise in the data, particularly for the weakest cross-peaks (Figure 2—figure supplement 2D). The average ratio for five cross-peaks that are in well-resolved regions of the spectrum and correspond to residues in the α-helix involved in the tripartite interface (T383, G384, L387, R388 and S391) is 0.945. These data show that the quadruple R322E/K325E/R398Q/R399Q mutation abolishes binding of the Syt1 C₂B domain to CpxSC and the mutant does not bind through the tripartite interface under these conditions.

To corroborate this conclusion, we also analyzed samples where Cpx1(26-83) was ²H,¹⁵N-labeled and formed ²H,¹⁵N-CpxSC to analyze perturbations on Cpx1(26-83) upon C₂B domain binding. Addition of WT Syt1 C₂B domain caused substantial broadening in the ¹H-¹⁵N TROSY-HSQC spectrum of CpxSC, as manifested by decreased intensities in the cross-peaks corresponding to structured parts of Cpx1(26-83) in the complex upon binding to C₂B (Figure 3A). Importantly, C₂B did not induce significant shifts in any of the cross-peaks from residues at or near the tripartite interface (residues 66–75; see Figure 3A). We note that K73, K74 and K75 are flexible in the complex but their cross-peaks are in unique positions that are very sensitive to changes in their environment (Chen et al., 2002; Pabst et al., 2000; Trimbuch et al., 2014). Hence, the absence of shifts in these cross-peaks and those of residues 66–75 in general show that the C₂B domain does not bind to CpxSC through the tripartite interface under these conditions. Moreover, R322E/K325E/R398Q/R399Q C₂B did not cause shifts or intensity decreases in the ¹H-¹⁵N TROSY-HSQC spectrum of ²H,¹⁵N-CpxSC (Figure 3B), showing again that the quadruple mutation abrogates binding of the Syt1 C₂B domain to CpxSC. Hence, we do not find any evidence for binding of the C₂B domain to CpxSC through the tripartite interface even after abolishing the interactions involving the polybasic region and the primary interface. These results contrast with ITC data suggesting the existence of a Syt1-complexin-1-SNARE complex interaction involving the tripartite interface in experiments performed with Syt1 C₂B bearing seven mutations designed to disrupt binding via the polybasic region and the primary interface (KA-Q mutant), and a complexin-1-SNARE complex bearing five mutations to further disrupt such binding (Zhou et al., 2017). We are collaborating with the laboratory of Axel Brunger to determine the reasons underlying these conflicting results, and the results of these efforts will be published elsewhere.

Since the L387Q/L394Q mutation in the α-helix of the C₂B domain that forms the tripartite interface was reported to disrupt binding through this interfaces based on ITC data (Zhou et al., 2017), we tried to analyze whether this mutation perturbs binding of C₂B to CpxSC. The ¹H-¹⁵N TROSY-HSQC spectrum of ²H,¹⁵N-IM-¹³CH₃-labeled L387Q/L394Q C₂B exhibited well-dispersed cross-peaks characteristic of a folded protein domain, but also contained broad cross-peaks in the center of the spectrum that are commonly observed in unstable proteins that unfold and/or aggregate (Rizo et al., 2012; Figure 3—figure supplement 1). Addition of 20 μM CpxSC led to strong decreases in cross-peak intensities that arose at least in part from sample precipitation. Hence it was not possible to continue the titration. These results show that the L387Q/L394Q causes a considerable destabilization of the Syt1 C₂B domain, which is consistent with the finding that this mutation decreases the thermal denaturation temperature of the C₂B domain by about 10°C (Zhou et al., 2017). Such destabilization may arise because this mutation involves the replacement of a hydrophobic side chain that packs inside the C₂B domain and barely contacts the SNAREs (L394; Figure 1G) with a polar residue. Hence, it is plausible that the disruption of neurotransmitter release caused by the L387Q/L394Q mutation (Zhou et al., 2017) arose from a general loss of function caused by protein instability.

In summary, our data show that binding of the Syt1 C₂B domain to the Cpx1(26-83)-SNARE complex in solution involves the polybasic region and the primary interface, while binding via the tripartite interface is undetectable under the conditions of our NMR experiments even when CpxSC is added at 85 μM concentration and binding through the polybasic region and the tripartite interface is abolished. These results indicate that the tripartite interface observed in the complexin-1-Syt1-SNARE complex crystals might have arisen from crystal packing, but further research will be required to clarify this issue.

The primary interface involves two regions of the C₂B domain, one containing R398 and R399 (region II), and the other E295 and Y338 (region I). The functional importance of region I was demonstrated by the strong disruption of neurotransmitter release caused by the E295A/Y338W mutation in this region (Zhou et al., 2015). To examine the effects of this mutation on Syt1-SNARE binding, we performed titrations of ²H,¹⁵N-IM-¹³CH₃-E295A/Y338W C₂B with CpxSC. Interestingly, we found that CpxSC caused more extensive changes in the ¹H-¹⁵N TROSY-HSQC spectrum of this mutant (Figure 4A) than those observed for WT C₂B domain (Figure 1H). Analysis of these changes showed that cross-peaks from residues in or near the polybasic region shifted less extensively and more slowly than observed for WT C₂B, particularly in the initial points of the titration (Figure 4B,C). Conversely, cross-peaks from the primary interface moved faster and more extensively for E295A/Y338W C₂B than for WT C₂B; indeed, some cross-peaks that did not shift or barely shifted in WT C₂B exhibited considerable shifts for E295A/Y338W C₂B domain, including the R398 cross-peak (Figure 4B,D). These effects in cross-peaks from the primary interface are reminiscent of those observed for the R322E/K325E mutant C₂B domain (Figure 2A,C,D), but the shifts observed for the primary interface of the E295A/Y338W C₂B mutant are even larger. These results show that the E295A/Y338W mutation actually increases the affinity of the Syt1 C₂B domain for CpxSC rather than impairs binding, and also appears to make the interaction at the primary interface more extensive, as observed for the R322E/K325E mutation. The basis for this behavior is unclear, but the increased affinity caused by the E295A/Y338W mutation may arise from replacing a tyrosine with a tryptophan, which increases the hydrophobic surface area of this residue. This change may be readily accommodated because the packing in this region of the interface with the SNARE complex is not optimal. We note that the movement of some cross-peaks with increasing CpxSC concentration was curved in some cases (e.g. those of K325 and R398, Figure 4C,D). This finding suggests that there is an interplay between binding of CpxSC to the primary interface and to the polybasic region. CpxSC appears to bind exclusively to one or the other site at low concentrations, likely because of steric hindrance disfavors simultaneous binding of two CpxSC complexes to one C₂B domain. However, simultaneous binding to both sites might be allowed at higher CpxSC concentrations by slight alterations in both binding modes, leading to the curved cross-peak movement.

Because the WT and mutant C₂B domains used for all the titrations with CpxSC were specifically ¹³CH₃-labeled at the Met and Ile δ1 methyl groups, we also acquired ¹H-¹³C heteronuclear multiple quantum coherence (HMQC) spectra of each sample, as these spectra offers very high sensitivity (Ruschak and Kay, 2010). The HMQC spectra contained only a small number of probes (Figure 4—figure supplement 1A) and hence provided more limited information than the ¹H-¹⁵N TROSY-HSQC spectra, but corroborated the conclusions obtained from the latter regarding binding to the primary interface. Thus, CpxSC caused shifts in the cross-peak from the I401 δ1 methyl group at the primary interface that were more marked for the R322E/K325E mutant, were even larger for the E295A/Y338W mutant, and were abolished for the R398Q/R399Q and R322E/K325E/R398Q/R399Q mutants (Figure 4—figure supplement 1B). The cross-peak from the δ1 methyl of I293, another residue at the primary interface, was shifted by CpxSC only for the E295A/Y338W mutant.

Our NMR data show that there are two major binding modes between the Syt1 C₂B domain and the SNARE complex in solution, one involving the polybasic region of C₂B and the other involving the primary interface, which includes the tandem arginines R398,R399. Since the polybasic region and the tandem arginines have been implicated also in membrane binding [e.g. (Araç et al., 2006; Bai et al., 2004; Li et al., 2006; Xue et al., 2008)], it is critical to analyze interactions between Syt1 and membrane-anchored SNARE complexes to assess whether either of these two binding modes still remain in the presence of membranes, or a different type of interaction might occur. For this purpose, we designed a strategy based on anchoring SNARE complexes on nanodiscs. For these studies we used the Syt1 C₂AB fragment that spans both C₂ domains because the C₂A domain contributes to binding of Syt1 to membranes and could contribute to binding to the SNARE complex directly or indirectly, for example through cooperativity between SNARE and lipid interactions of Syt1. As a scaffold for the nanodiscs, we chose MSP1E3D1 because it yields stable nanodiscs with a diameter of ca. 13 nm (Nath et al., 2007) that can accommodate the SNARE complex and allow potential simultaneous interactions of Syt1 C₂AB with the SNARE complex and the lipids. We prepared SNARE complexes with full-length syntaxin-1 and the SNARE motifs of SNAP-25 and synaptobrevin, anchoring the complexes on the nanodiscs through the syntaxin-1 transmembrane (TM) region to mimic the configuration expected to occur on the plasma membrane. The stoichiometry of MSP1D3 to syntaxin-1 was adjusted to form nanodiscs that on average contained one SNARE complex. We refer to these macromolecular assemblies as cisSC-NDs, while control nanodiscs prepared without the SNAREs are abbreviated as NDs (Figure 5A). Compared to experiments using liposomes, this modular design facilitates analysis of individual SNARE complex-Syt1-membrane assemblies without complications that might arise from liposome clustering induced by C₂AB (Araç et al., 2006).

To monitor binding of C₂AB to NDs or cisSC-NDs by FRET, we generated a single-cysteine C₂AB mutant with the cysteine replacing E346, and labeled it with an Alex488 donor fluorescent probe (C₂AB*). This position was chosen because residue 346 is not located on any of the interfaces that have been implicated in SNARE complex binding (Figure 5—figure supplement 1A) and placing the fluorescent probe on this residue is not expected to disrupt these interfaces. Unless otherwise indicated, NDs contained 5% rhodamine-labeled phosphatidylethanolamine (Rho-PE), which constitutes a suitable acceptor probe for highly efficient FRET with Alexa488. Indeed, titration of C₂AB* with cisSC-NDs formed with a 80:15:5 mixture of phosphatildylcholine (PC), phosphatidylserine (PS) and Rho-PE in the presence of 125 mM KCl and 1 mM Ca²⁺ led to progressively more efficient FRET that maximized at a FRET efficiency of ca. 0.88 (Figure 5B and C, blue curve). Control experiments with analogous NDs lacking SNARE complex showed that saturation required higher ND concentrations (Figure 5C, red curve), yielding a substantially higher apparent K_D (129.5 nM compared to 26.6 nM for the cisSC-NDs). A summary of the K_Ds obtained under these and other conditions described below is presented in Supplementary file 1. Supplementary file 2 lists cooperativity factors calculated from K_D NDs/K_D cisSC-NDs, which yield an idea of the synergy between interactions of C₂AB* with the SNAREs and the lipids, for each condition. Supplementary file 1 also lists repeat experiments performed under selected conditions with different nanodisc preparations that show the reproducibility of the data (see statistics in Materials and methods). We observed a natural variability in the apparent K_Ds that may arise in part from different incorporation of SNARE complexes into the nanodiscs. Note also that some of the K_Ds are in the low nM range and may not be accurate because the concentration of C₂AB* used for all experiments was 50 nM and the corresponding titration curves thus approach saturating binding conditions. Hence, these K_Ds and corresponding cooperativity factors must be interpreted with caution. For all these reasons, the conclusions described below were obtained by comparing experiments performed on the same day or over period of two days with the same preparations, and were confirmed by additional comparisons made with other preparations on different days. Importantly, the key conclusions are supported by the overall consistency of the data obtained under different conditions.

The higher affinity of C₂AB* for cisSC-NDs than for NDs (Figure 5C, blue and red curves, respectively) shows that C₂AB* binds simultaneously to the SNARE complex and the lipids under these conditions. Since Syt1 C₂AB is able to bind simultaneously to two membranes in the presence of Ca²⁺ (Araç et al., 2006), which could cooperate with binding to the SNARE complex (Brewer et al., 2015), we also investigated binding of C₂AB to trans-SNARE complexes formed between two nanodiscs (transSC-NDs). Because we wanted to compare binding to cis and trans SNARE complexes, in these experiments we placed a tetramethylrhodamine (TMR) acceptor fluorescent probe on the SNARE complex rather than on the nanodiscs to allow direct quantification of the SNARE complex concentrations from the UV-vis absorption of the probe. We chose residue 34 of SNAP-25 to place the acceptor probe because it is predicted to be sufficiently close to the donor probe at residue 346 of C₂AB in any of the three structures of Syt1-SNARE complexes that have been elucidated (Figure 5—figure supplement 1B), such that binding would be detected regardless of which binding mode occurs. Titrations of C₂AB* with transSC-NDs and cisSC-NDs yielded similar results (Figure 5D) and comparable apparent K_Ds (118 nM and 96 nM, respectively). The lower K_D obtained with cisSC-NDs labeled with 5% Rho-PE can be attributed to the negative charge added by the labeled lipids, which should increase their membrane affinity for C₂AB (Zhang et al., 1998). These results suggest that the affinity of C₂AB* for cisSC-NDs and transSC-NDs is similar and ensuing experiments were performed with cisSC-NDs for simplicity.

We also analyzed binding of C₂AB* to cisSC-NDs that were labeled at the N- or C-terminus of the SNARE complex (residue 16 or 76 of SNAP-25; Figure 5—figure supplement 1B). We obtained similar binding curves and comparable K_Ds to those obtained with the label at residue 34 (Figure 5E; Supplementary file 1). The FRET efficiencies observed at saturating concentrations were similar for the labels at residues 34 and 76 of SNAP-25, and somewhat lower for the label at residue 16. The FRET efficiencies are consistent with the binding modes involving the polybasic region and the primary interface, which predict that the probe on C₂AB* is located at comparable, short distances from residues 34 and 76, and farther from residues 16, but are not consistent with the tripartite complex, where residue 346 of C₂AB* is expected to be much closer to residue 16 than to residue 76 (Figure 5—figure supplement 1B). However, these results are not conclusive, as other binding modes could also be consistent with the observed FRET efficiencies. We also analyzed the effect of including complexin-1 on binding of C₂AB* to cisSC-NDs labeled at residue 34 of SNAP-25 and observed similar binding curves and apparent K_Ds (Figure 5F; Supplementary file 1), suggesting that complexin-1 does not substantially alter the interaction of C₂AB* with the nanodisc-anchored SNARE complex. This observation is also consistent with binding of C₂AB to the SNARE complex through the primary and polybasic interfaces, which is not expected to be affected by complexin-1.

In subsequent experiments we focused on comparing binding of C₂AB* to NDs and cisSC-NDs labeled with Rho-PE to analyze the increases in affinity caused by the presence of the SNARE complex, and analyzed how the conditions of the experiments affect the underlying affinities. In experiments performed in 125 mM KCl and 1 mM EGTA to analyze Ca²⁺-independent interactions, binding to NDs was very weak. We lowered the KCl concentration to 50 mM to facilitate binding and were able to observe efficient binding at relatively high ND concentrations (Figure 5C, black curve), with an apparent K_D of 8.7 μM. Importantly, we observed a much higher affinity for cisSC-NDs (Figure 5C, green curve), with an apparent K_D of 141 nM. These results suggest that there is a strong synergy between binding of Ca²⁺-free C₂AB* to the SNARE complex and the nanodisc phospholipids. Since PIP₂ enhances binding of the C₂B domain to membranes due to interactions with the polybasic region (Bai et al., 2004; Li et al., 2006), we performed titrations with NDs and cisSC NDs containing 1% PIP₂ and indeed observed much higher affinities at 50 mM KCl (Figure 5G, black and red curves; apparent K_Ds 286 nM and 43 nM, respectively). Binding was weaker in 125 mM KCl (Figure 5G, blue and green curves) but there was still a large difference in the K_Ds observed for NDs and cisSC-NDs (apparent K_Ds 4.9 μM and 366 nM, respectively). We also analyzed binding of C₂AB* to NDs and cisSC-NDs containing 1% PIP₂ in the presence of 1 mM Ca²⁺ and again observed increased affinity for the latter (Figure 5H, black and red curves; apparent K_Ds 22 nM and 12 nM, respectively). Since the high affinity for cisSC-NDs containing PIP₂ implied that we were close to saturation binding conditions, which hindered analysis of the SNARE-induced enhancement on binding, we also performed experiments with NDs and cisSC-NDs containing 10% instead of 15% PS (i.e. composed of PC:PS:Rho-PE:PIP₂ 84:10:5:1) and indeed observed somewhat weaker binding (Figure 5H, green and orange curves; apparent K_Ds 29 nM and 20 nM, respectively). Overall, the increases in affinity caused by simultaneous binding of C₂AB* to the nanodiscs and the SNARE complex were considerably lower in the presence than in the absence of Ca²⁺, particularly when the nanodiscs contained PIP₂ (Supplementary file 1, 2).

To dissect the contributions of the polybasic region and R398,R399 at the primary interface to binding of C₂AB* to NDs and cisSC-NDs, we performed titrations under various conditions using WT and mutant versions of C₂AB* that were labeled with a donor probe at residue 346 and bore the R322E/K325E, R398Q/R399Q or R322E/K325E/R398Q/R399Q mutations (Figure 6; Figure 6—figure supplements 1–6). Since the much stronger effects of the R322E/K325E mutation than the K324E/K326E mutation on SNARE complex binding and on neurotransmitter release supported the physiological relevance of Syt1-SNARE complex binding through the polybasic region (Brewer et al., 2015), we also included C₂AB* with the K324E/K326E mutation in these analyses. All NDs and cisSC-NDs included 5% Rho-PE. Ca²⁺-independent binding of C₂AB* to NDs was markedly impaired by all mutations, but mutations in the polybasic region impaired binding to NDs containing PIP₂ much more strongly than the R398Q/R399Q mutation (Figure 6A,C). Hence, binding to these nanodiscs is mediated largely by the polybasic region, as expected, but the tandem arginines also participate in binding to some extent. Binding to cisSC-NDs in EGTA was also impaired strongly by mutations in the polybasic region, but in this case the impairment caused by the R398Q/R399Q mutation was almost as strong (Figure 6B,D). All mutations strongly decreased the cooperativity factors calculated from K_D NDs/K_D cisSC-NDs, which provide an idea of the contribution of C₂AB*-SNARE complex interactions to cisSC-ND binding and ranged from 6.64 to 62.1 for WT C₂AB* in the three conditions including EGTA (Supplementary file 2). Thus, both the tandem arginines (R398,R399) and the polybasic region are important for Ca²⁺-independent, simultaneous binding of C₂AB* to the SNARE complex and the lipids. These findings suggest that the underlying interactions are dynamic and involve at least two binding modes whereby either the primary interface containing R398,R399 or the polybasic region interacts with the SNAREs and the other basic sequence (the polybasic region or R398,R399) binds to the lipids (Figure 5—figure supplement 2A,B).

An interestingly different picture emerged in experiments performed in the presence of Ca²⁺. Importantly, the R322E/K325E mutation impaired Ca²⁺-dependent binding to PIP₂-containing NDs much more dramatically than the K324E/K326E mutation (Figure 6G, Figure 6—figure supplement 6A; apparent K_Ds 49–62 nM for the K324E/K326E mutant and 1.5–2.9 μM for R322E/K325E). This striking difference correlates with the effects of these mutations on neurotransmitter release (Brewer et al., 2015) and was not observed for Ca²⁺-dependent binding of C₂AB* to NDs lacking PIP₂ (Figure 6E), or Ca²⁺-independent binding to PIP₂-containing NDs (Figure 6A,C) or liposomes (Brewer et al., 2015), all of which were similarly disrupted by the R322E/K325E and K324E/K326E mutations. These findings show that there is a specific Ca²⁺-dependent binding mode of Syt1 to PIP₂-containing membranes that involves R322,K325 but not K324,K326, and suggest that the specific, strong disruption of neurotransmitter release induced by the R322E/K325E mutation but not the K324E/K326E mutation arises from impairment of Ca²⁺-dependent binding of Syt1 to the (PIP₂-containing) plasma membrane rather than to the SNARE complex.

The cooperativity factors calculated from K_D NDs/K_D cisSC-NDs in the presence of Ca²⁺ under various conditions (Figure 6E–H, Figure 6—figure supplements 4–6) ranged from 1.39 to 3.11 and hence were considerably smaller than those observed in EGTA (Supplementary file 2), indicating that interactions of WT C₂AB* with the SNARE complex contributed much less to binding to cisSC-NDs in the presence of Ca²⁺ than in its absence. Interestingly, the cooperativity factors observed for WT C₂AB* in the presence of PIP₂ were particularly small, suggesting that Ca²⁺-dependent interactions of C2AB* with PIP₂ on the nanodiscs preclude interactions with the SNARE complex. The diminished cooperativity factors in the presence of PIP₂ were decreased further by the R398Q/R399Q, K324E/K326E and R322E/K325E/R398Q/R399Q mutations, but where substantially increased by the R322E/K325E mutation, presumably because the specific interactions of C₂AB* with PIP₂ mediated by R322E,K325E were disrupted, allowing binding to the SNARE complex. These findings can be readily rationalized by modeling how the C₂B domain binds to a PIP₂-containing membrane in a Ca²⁺-dependent manner. Because Ca²⁺ induces binding of the C₂B domain in an approximately perpendicular orientation to the membrane that favors insertion of both Ca²⁺-binding loops into the bilayer (Araç et al., 2006; Bai et al., 2004; Rufener et al., 2005), this orientation allows binding of R322 and K325 to PIP₂, but K324 and K326 point away from the membrane because they are on the opposite side of the same β-strand (Figure 5—figure supplement 2C; see also Figure 10B). In this configuration, interaction of the SNARE complex with the primary interface of the C₂B domain is impossible because it would place the syntaxin-1 C-terminus far from the membrane where it is anchored (Figure 5—figure supplement 2C). Simultaneous Ca²⁺-dependent binding of the C₂B domain to the membrane through its Ca²⁺-binding loops and to the SNARE complex through its polybasic region is in principle compatible with anchoring of syntaxin-1 to the membrane (Figure 5—figure supplement 2D), but this binding mode is expected to be prevented by PIP₂ because PIP₂ binds to K322,R325, the same residues of the polybasic region that are key for SNARE complex binding (Brewer et al., 2015).

In summary, Ca²⁺-dependent binding of C₂AB* to PIP₂-containing membranes is incompatible with the two major Syt1-SNARE complex binding modes, which involve the primary and polybasic interfaces. However, upon Ca²⁺- and PIP₂-dependent membrane binding, R398 and R399 can still be involved in non-specific interactions with negative residues of the SNARE complex, and such interactions are also possible if the C₂B domain binds in a more slanted orientation to membranes lacking PIP₂ (Figure 5—figure supplement 2E,F). Such non-specific interactions might underlie the modest decreases in Ca²⁺-dependent binding of C₂AB* to cisSC-NDs caused by the R398Q/R399Q mutation (Figure 6F,H) and, similarly, the impairments in such binding caused by the K324E/K326E mutation might arise from other non-specific binding modes.

We also analyzed the effects of the E295A/Y338W mutation in region I of the primary interface on binding to NDs and cisSC-NDs. Interestingly, this mutation did not alter binding to NDs under various conditions (Figure 7) but caused a considerable increase in affinity for cisSC-NDs containing 15% PS and 1% PIP₂ in the absence of Ca²⁺, compared to WT C₂AB* (Figure 7A; apparent K_Ds 366 nM for WT C₂AB* and 80 nM for the E295A/Y338W mutant). In contrast, we observed only very slight increases in Ca²⁺-dependent binding of E295A/Y338W C₂AB* to cisSC-NDs containing 15% PS with or without 1% PIP₂, or 10% PS and 1% PIP₂, compared to WT C₂AB* (Figure 7B–D). The results obtained in the absence of Ca²⁺ correlate with the increased affinity of the Syt1 C₂B domain for CpxSC observed in our NMR experiments (Figure 4) and show that this mutations enhances Ca²⁺-independent binding of C₂AB* to the SNARE complex within cisSC-NDs via the primary interface. However, the lack of an overt effect of the E295A/Y338W mutation on binding to cisSC-NDs in the presence of Ca²⁺ strongly supports the notion that the primary interface is not involved in Ca²⁺-dependent binding.

To gain insights in how the phospholipids in the nanodiscs influence Syt1-SNARE complex interactions, we performed experiments where we monitored direct binding of C₂AB* to the SNARE complex in cisSC-NDs by placing the FRET acceptor on residue 76 of SNAP-25 (Figure 8; Supplementary file 1). In these experiments we focused on WT C₂AB* and C₂AB* bearing the R398Q/R399Q, R322E/K325E or E295A/Y338W mutations that strongly impair neurotransmitter release. Comparisons of the results obtained with cisSC-NDS and those of parallel experiments performed with soluble SNARE complexes formed with the cytoplasmic region of syntaxin-1 (residues 2–253) (referred to as solubleSC) provided information on how interactions of C₂AB* with the lipids enhance SNARE complex binding. To examine how negatively charged phospholipid head groups contribute to such enhancement, we performed experiments with cisSC-NDs that contained only PC (cisSC-PC-NDs) or 15% PS,1% PIP₂ and 84% PC (cisSC-PC/PS/PIP₂-NDs). We also performed experiments with a soluble SNARE complex formed with the syntaxin-1 SNARE motif (residues 191–253) instead of its cytoplasmic region (Figure 8—figure supplement 1) to mimic the SNARE complex used in the NMR experiments (referred to as mcc). The results were similar to those obtained with solubleSC but revealed somewhat weaker affinities for C₂AB* that are consistent with the NMR data obtained with the C₂B domain and CpxSC (Figure 1) and suggest that the N-terminal region of syntaxin-1 contributes to Syt1-SNARE interactions, as proposed previously (Fernandez et al., 1998).

In the absence of Ca²⁺, C₂AB* bound more strongly to cisSC-PC-NDs than to solubleSC (K_D 0.93 and 5.5 μM, respectively), and binding to cisSC-PC/PS/PIP₂-NDs was even tighter (K_D0.49 μM) (Figure 8A). Binding of C₂AB* to these complexes was strongly impaired by the R322E/K325E mutation and to a lesser extent by the R398Q/R399Q mutation, and was strengthened by the E295A/Y338W mutation (Figure 8C,E,G). The cooperativity factors calculated from K_D solubleSC/K_D cisSC-PC-NDs and K_D solubleSC/K_D cisSC-PC/PS/PIP₂-NDs were 5.91 and 11.09 for WT C₂AB*, were dramatically decreased by the R322E/K325E and R398Q/R399Q mutations, and slightly increased by the E295A/Y338W mutation (Supplementary file 2). These findings are fully consistent with the conclusion that interactions of both the polybasic region and R398,R399 with the lipids contribute to Ca²⁺-independent binding of C₂AB* to cisSC-NDs, and hence that such binding involves at least two types of interactions where either the primary interface binds to the SNAREs and the polybasic region to the lipids, or the polybasic region binds to the SNAREs and R398,R339 to the lipids (Figure 5—figure supplement 2A,B). As observed in the previous experiments, the E295A/Y338W mutation increased the former of these two binding modes. As expected, the PC/PS/PIP₂-NDs enhanced binding to the SNARE complex more than the PC-NDs, but it is noteworthy that even interactions of C₂AB* basic residues with neutral phospholipids such as PC can enhance binding to the SNARE complex.

As expected, Ca²⁺ increased the affinity of C₂AB* for solubleSC, cisSC-PC-NDs and cisSC-PC/PS/PIP₂-NDs, and binding was tightest for the latter (Figure 8A,B; Supplementary file 1). The effects of the R398Q/R399Q, R322E/K325E and E295A/Y338W mutations on Ca²⁺-dependent binding of C₂AB* to solubleSC and cis-SC-PC-NDs (Figure 8D,F) were similar to those observed in the absence of Ca²⁺, and highly efficient FRET was also observed at saturating concentrations, as expected for binding of C₂AB* to the SNARE complex through the primary or polybasic interfaces with the acceptor probe on residue 76 of SNAP-25 (Figure 5—figure supplement 1B). However, the highest FRET efficiency observed upon binding of WT C₂AB* to cisSC-PC/PS/PIP₂-NDs was 0.4 (Figure 5B). The highest FRET efficiency was increased to some extent by the E295A/Y338W mutation and even more by the R322E/K325E mutation, but was decreased by the R398Q/R399Q mutation, which consistently led to a bimodal binding curve with a maximum at about 20 nM cisSC-PC/PS/PIP₂-NDs in four independent experiments (Figure 8D,F,H). These results are consistent with the notion that Ca²⁺-dependent binding of C₂AB* to PIP₂-containing membranes is incompatible with SNARE complex binding through the primary and polybasic interfaces. In the absence of interactions with the SNARE complex, C₂AB* may bind with similar probabilities to both sides of the nanodiscs, perhaps with some preference for the side lacking SNAREs where there are no steric clashes with the SNARE complex. Non-specific interactions of R398,R399 of WT C₂AB* with acidic residues from the SNAREs may facilitate binding to the SNARE-containing side (Figure 5—figure supplement 2E), leading to some FRET, but mutation of these arginines may lead to preferential binding of the R398Q/R399Q mutant to the other side, particularly at high nanodisc concentrations when membrane availability is not limiting. The highest FRET efficiency was likely increased by the E295A/Y338W mutation because it enhances SNARE binding. High FRET efficiencies were reached for binding of the R322E/K325E mutant to cisSC-PC/PS/PIP₂-NDs or for binding of WT C₂AB* and the three mutants to solubleSC or cisSC-PC-NDs (Figure 8D,F,H) because the specific Ca²⁺-dependent interaction of C₂AB* with PIP₂ that competes with SNARE binding is precluded.

Physiological conditions including ATP and PIP₂-containing membranes were reported to disrupt Syt1-SNARE complex binding, leading to the conclusion that such binding is not biologically relevant (Park et al., 2015). However, a fluorescent probe attached to residue 342 of the C₂B domain to monitor binding to the SNARE complex in this study may have disrupted binding through the primary interface because residue 342 is very close to this interface (Figure 5—figure supplement 1A), and another study concluded that Syt1-SNARE complex binding persists in the presence of ATP (Wang et al., 2016). To clarify this controversy and examine how ATP affects Syt1-SNARE complex interactions, we first analyzed the effects of adding 2.5 mM Mg²⁺ with or without 2 mM ATP on the FRET observed between C₂AB* and NDs or cisSC-NDs containing 5% Rho-PE, 15% PS and 1% PIP₂. Mg²⁺ and Mg-ATP decreased the FRET with both NDs and cisSC-NDs to some extent in the absence of Ca²⁺, but the FRET with cisSC-NDs was still much stronger that the FRET with NDs (Figure 9—figure supplement 1). However, the difference in FRET with cisSC-NDs and NDs was much smaller in the presence of Ca²⁺ when Mg²⁺ and ATP were added (Figure 9—figure supplement 2). These results suggested that Ca²⁺-independent binding of C₂AB* to the SNARE complex persists in the presence of Mg-ATP, but Ca²⁺-dependent binding is strongly impaired. These conclusions were supported by titrations of C₂AB* with NDs and cisSC-NDs, which revealed a considerable difference in affinity in the absence of Ca²⁺ but a much smaller difference in Ca²⁺-dependent binding (Figure 9A,B).

To investigate the nature of the Ca²⁺-independent interaction of C₂AB* with cisSC-NDs in the presence of Mg-ATP, we analyzed the effects of mutations on Ca²⁺-independent binding to NDs and cisSC-NDs. Binding to cisSC-NDs was enhanced by the E295A/Y338W mutation and markedly impaired by the R398Q/R399Q, R322E/K325E, K324E/K326E and R322E/K325E, R398Q/R399Q mutations, whereas all the mutations had small effects on ND binding (Figure 9C–D). The E295A/Y338W mutant bound much tighter to cisSC-NDs than to NDs but all other mutants exhibited similar affinity for NDs and cisSC-NDs (Figure 9—figure supplement 3). These results suggest that, in the presence of Mg-ATP, Ca²⁺-independent binding of WT C₂AB* to the SNARE complex on nanodiscs also involves at least two binding modes mediated by either the primary or the polybasic interface, and the former is stabilized by the E295A/Y338W mutation. We also analyzed whether complexin-1 altered the affinity of WT C₂AB* for cisSC-NDs, but did not observe any significant effects (Figure 9E). We also verified that the R322E/K325E mutation disrupts Ca²⁺-dependent binding to PIP₂-containing NDs much more strongly than the K324E/K326E mutation in the presence of Mg-ATP (Figure 9F), confirming the specificity of such impairment under these conditions.

Constructs to express the following proteins or protein fragments were described previously: rat synaptobrevin-2 SNARE motif (residues 29–93), rat synaptobrevin-2 (residues 49–93), human SNAP-25A fragments encoding its SNARE motifs (residues 11–82 and 141–203), full-length rat synaptobrevin, full-length rat syntaxin-1A, rat syntaxin-1A (residues 191–253), rat syntaxin-1A (residues 2–253), human SNAP-25A (residues 11–82 and 141–203), rat synaptotagmin-1 C₂B domain with a C277A mutation (residues 271–421; referred to as WT C₂B domain), the same rat synaptotagmin-1 fragment with a R398Q/R399Q mutation, rat synaptotagmin-1 C2AB fragment with a C277A mutation (residues 140–421), full-length rat complexin-1, rat complexin-1 (residues 26–83) and MSP1E3D1 (pMSP1E3D1 was a kind gift from Stephen Sligar; Addgene plasmid # 20066; http://n2t.net/addgene:20066; RRID:Addgene_20066) (Araç et al., 2006; Brewer et al., 2015; Chen et al., 2006; Chen et al., 2008; Chen et al., 2002; Denisov et al., 2007; Ma et al., 2013; Xu et al., 2013; Zhou et al., 2013). All these proteins were expressed in E. coli BL21 (DE3) cells and purified as previously described in these references, with the following exceptions.

The synaptobrevin-2(29–93), syntaxin-1A (191–253), SNAP-25A(11–82), SNAP-25A(141–203) and complexin-1(26–83) fragments were expressed E. coli BL21 (DE3) cells with the pET-duet vector. Both SNAP-25 fragments included tryptophan at the N-terminus to facilitate detection by UV spectroscopy. Cells were harvested and re-suspended in PBS pH 7.4 with 10 mM imidazole and supplemented with Sigma protease inhibitors (P2714-1BTL). Cleared lysates were applied to Ni-NTA resin (Thermo Fisher), washed with PBS pH 7.4 with 10 mM imidazole, PBS pH 7.4 with 10 mM imidazole and 10% Triton, PBS pH 7.4 with 10 mM imidazole and 1 M NaCl, and eluted in PBS pH 7.4, 500 mM imidazole. The affinity tag was cleaved with TEV protease overnight at 4°C in PBS pH 7.4. After affinity tag cleavage, all proteins were further purified using size exclusion chromatography on a Superdex 75 column (GE 16/60) equilibrated with 20 mM Tris pH 7.4 125 mM NaCl. MSP1E3D1 was expressed in E. coli BL21 (DE3) cells grown in Terrific broth media to OD600 = 2.0, then induced with 1 mM Isopropylβ-D-1-thiogalactopyranoside (IPTG) for 4 hr at 37°C. Cells were re-suspended in 40 mM Tris pH 8.0 300 mM NaCl 1% TritonX-100 5 mM Imidazole containing Sigma protease inhibitors (P2714-1BTL) and lysed using an Avestin EmulsiFlex-C5 homogenizer. The soluble fraction of the cell lysate was collected after centrifugation at 48,000 x g for 45 min and incubated with Ni-NTA resin (Thermo Fisher) for 2 hr at 4°C. The resin was washed with the following buffers sequentially: 40 mM Tris pH 8.0 300 mM NaCl 1% Triton, 40 mM Tris pH 8.0 300 mM NaCl 50 mM Sodium Cholate 20 mM Imidazole and 40 mM Tris pH 8.0 300 mM NaCl 50 mM Imidazole followed by elution using 40 mM Tris pH 8.0 300 mM NaCl 500 mM imidazole. This was followed by size exclusion chromatography on a Superdex 200 column (GE 16/60) in 20 mM Tris pH 7.4 100 mM NaCl 0.5 mM EDTA.

Expression vectors for mutant proteins were generated using a combination of the QuickChange site-directed mutagenesis kit (Strategene) and standard PCR-based techniques with custom designed primers. These mutants included: Syt1 C₂B (residues 271–421) E295A/Y338W, L387Q/L394Q, R398Q, R399Q, R322E/K325E, and R322E/K325E/R399Q/R398Q; Syt1 C₂AB (residues 140–421) C277A/E346C (referred to as WT C₂AB*), C277A/E346C/R398Q/R399Q, C277A/E346C/R322E/K325E, C277A,E346C,R322E/K325E/R398Q/R399Q, C277A/E346C/E295A/Y338W and C277A/E346C/K324E/K326E; and SNAP-25 (residues 11–82) R16C, Q34C and K76C. All mutant proteins were purified as the WT proteins, including 0.5 mM TCEP in the final purification step for cysteine containing proteins. Isotopically labeled and perdeuterated Syt1 C₂B domain and Cpx-1 (26–83) fragments were expressed using M9 expression media in 99.9% D₂O with D-glucose (1,2,3,4,5,6,6-D₇, 97–98%) as the sole carbon source (3 g/L) and ¹⁵NH₄Cl as the sole nitrogen source (1 g/L). Specific ¹³CH₃-labeling at the Met and Ile δ1 methyl groups of Syt1 C₂B domain fragments was achieved by adding [3,3–²H] ¹³C-methyl α-ketobutyric acid (80 mg/L) and 13C-methyl methionine (250 mg/L) (Cambridge Isotope Laboratories) to the cell cultures 30 min prior to IPTG induction.

The SNARE motifs were mixed in the equimolar ratio in following order: synaptobrevin-2 (29–93), SNAP-25A(141–203), SNAP-25A(11–82) and syntaxin-1A (191–253), in the presence of 1 M NaCl. The mixture contained the following protease inhibitors (protease inhibitor cocktail A): Antipain Dihydrochloride 0.016 mg/ml (Thermo Fischer Scientific: 50488492); Leupeptin 0.33 mg/ml (Gold Bio: L01025); Aprotinin 0.08 mg/ml (Gold Bio: A655100). The assembly reaction was incubated in room temperature overnight while rotating. The SNARE motifs that did not incorporate into complex were removed by concentration-dilution at room temperature using 30 kDa molecular weight cutoff (MWCO) Amicon centrifugation filters.

Samples containing 80 µM WT Syt1 C₂B domain were mixed with 30 µM SNARE complex +/- Complexin-1 (26–83) in 20 mM HEPES, 125 mM KCl, 1 mM CaCl₂, pH 7.4. The total reaction volume was 30 µl. After mixing, the UV spectra of both samples were acquired using Nanodrop (Thermofisher). Next, the samples were centrifuged at 13,000 rpm for 5 min spin in a benchtop centrifuge (Eppendorf). The UV spectra of both samples were acquired again after centrifugation.

All NMR spectra were acquired at 25°C on Agilent DD2 spectrometers operating at 600 or 800 MHz and equipped with cold probes. ¹H-¹⁵N TROSY HSQC and ¹H-¹³ HMQC spectra were acquired on samples with 10% D₂O as the solvent. For titrations of WT and mutant Syt1 C₂B domains with SNARE-Complexin-1 (26–83) complex, assembled soluble SNARE complex was concentrated at room temperature to a concentration above 250 μM using 30 kDa (Amicon) centrifugation filters and buffer exchanged into 20 mM HEPES containing 1 mM TCEP and 100 or 125 mM KCl 10% D2O buffer using Zeba Spin Desalting Columns, 7K MWCO, 10 mL (Thermofisher). Complexin-1 (26–83) was also concentrated above 250 μM using 3 kDa (Amicon) centrifugation filters and buffer exchanged to matching buffer. SNARE-Complexin-1 (26–83) complex was preassembled before mixing with C₂B Syt-1 fragment with 20% excess Complexin-1 (26–83), and 1 mM EDTA or 1 mM CaCl₂ were included in the mix depending on the final sample buffer. After cation exchange, the corresponding ²H,¹⁵N-IM-¹³CH₃-C₂B domain fragment was buffer exchanged using a PD-10 Desalting Column (GE Healthcare) to matching buffer. Final samples containing 32 μM C₂B domain were prepared by adding 1 mM EDTA or 1 mM CaCl2 and protease inhibitor cocktail A. ¹H-¹⁵N TROSY-HSQC and ¹H-¹³C HMQC spectra were acquired first for the isolated C₂B domain and then with the same sample after adding increasing concentrations of complexin-1 (26–83)/SNARE complex. The concentrations for each titration step for each mutant are indicated in the figure legends. For titrations of ²H,¹⁵N-CpxSC with WT and R322E/K325E/R398Q/R399Q mutant Syt1 C₂B domains, the proteins were prepared by the same procedures described above. ²H,¹⁵N-CpxSC was formed with ²H,¹⁵N-Cpx1(26-83) and 12.5% excess assembled soluble SNARE complex. ¹H,¹⁵N TROSY-HSQC spectra were acquired for ²H,¹⁵N-CpxSC alone and for 40 μM ²H,¹⁵N-CpxSC in the presence of 40 μM WT or R322E/K325E/R398Q/R399Q mutant Syt1 C₂B domain in 20 mM HEPES 100 mM KCl 1 mM EDTA 1 mM TCEP. All NMR samples included protease inhibitor cocktail A. Total acquisition times ranged from 3.5 to 87.5 hr, depending on the sensitivity of the spectra. NMR data were processed with NMRPipe (Delaglio et al., 1995) and analyzed with NMRView (Johnson and Blevins, 1994).

Single cysteine mutants of Synaptotagmin-1 C₂AB (140-421) were labeled with Alexa488 using maleimide reactions with Alexa Fluor488 C5 Maleimide reagent from Thermo Fisher Scientific (A10254). Synaptotagmin-1 C₂AB single cysteine mutants were first buffer exchanged into 20 mM HEPES pH 7.4 125 mM KCl 0.5 mM TCEP using concentration and dilution. Buffer exchanged proteins at a concentration of 100–150 μM were incubated with 10-fold excess dye for 4 hr at room temperature or overnight at 4°C. The reaction was quenched by addition of 10 mM DTT. Unreacted dye was separated from labeled protein through cation exchange chromatography on a HiTrap SP column (GE) in 50 mM Sodium Acetate pH 6.2 (Buffer A) using a linear gradient from 0 to 1000 mM NaCl. Prior to elution, the column with the bound protein was washed with 100 mL of buffer A. Single cysteine mutants of SNAP-25A(11-82) were labeled with TMR using maleimide reactions with TMR-5-Maleimide reagent from Thermo Fisher Scientific (T6027). Proteins were first buffer exchanged into 20 mM HEPES pH 7.4 125 mM KCl 0.5 mM TCEP using concentration dilution. Buffer exchanged proteins at a concentration of 100–150 μM were incubated with 10-fold excess dye for 4 hr at room temperature or overnight at 4°C. The reaction was quenched by addition of 10 mM DTT. Unreacted dye was separated from the labeled protein by using two runs on PD Miditrap G25 column followed by size exclusion chromatography on Superdex 75 column (GE 10/300 or 16/60). The concentration and the labeling efficiency of the tagged proteins were determined by using UV-vis absorbance and a Bradford assay.

Appropriate lipid mixtures (specific to each experiment as indicated in the figures and Supplementary file 1) were prepared by mixing chloroform stocks in glass test tubes. These mixtures were dried under a stream of nitrogen and stored overnight in a vacuum desiccator. The lipids were solubilized in a 20 mM HEPES pH 7.4 125 mM KCl 1% β-OG (octyl-beta-glucoside) buffer by vortexing for 5 min. To form nanodiscs, MSP1E3D1 was incubated with solubilized lipids at a ratio of 1:110 in the presence of 1% β-OG (final concentration) at 4°C for 30 min. The mixture was passed over a 4 cm-high Thermo Scientific Pierce Detergent Removal Resin (87780) column (approximately 3 mL of the slurry; the final volume of the mixture was always between 3–4 mL). The nanodiscs were further purified by size exclusion chromatography using a Superdex 200 column (GE 16/60). Appropriate fractions were pooled and concentrated to a desired concentration using a 30 kDa MWCO Amicon centrigufation filter.

To prepare cisSC-NDs, detergent solubilized cisSNARE-complex was formed by incubating 10 μM full-length rat syntaxin-1A with 15 μM of synaptobrevin-2 (29–93), SNAP-25A(11-82) and SNAP-25A(141-203) in the presence of 1% β−OG, 0.5 M NaCl and protease inhibitor cocktail A overnight at 4°C. To obtain transSC-NDs, we first prepared detergent solubilized activated t-SNARE complex [see Prinslow et al., 2019 by incubating 10 μM full-length syntaxin-1A with 15 μM synaptobrevin-2 (49–93), SNAP-25A(11-82) and SNAP-25A(141-203) in the presence of 1% β-OG, 0.5 M NaCl and protease inhibitor cocktail A overnight at 4°C. For incorporation into nanodiscs, cisSNARE-complex, activated t-SNARE-complex or full-length synaptobrevin-2 were mixed with MSP1E3D1 and solubilized lipids at a ratio of 1:3:300 and incubated at 4°C for 30 min. The detergent was removed using Thermo Scientific Pierce Detergent Removal Resin, as described for the isolated nanodiscs, and size exclusion chromatography using a Superdex 200 column (GE 16/60). Fractions from size exclusion chromatography were assessed by SDS-PAGE and fractions that contained approximately one cisSNARE-complex, activated t-SNARE-complex or full-length synaptobrevin-2 per 2 MSP1E3D1 molecules were pooled together, mixed with protease inhibitor cocktail A and concentrated. TransSNARE-complex was formed by incubating nanodiscs containing activated t-SNARE complex with 2-fold excess of synaptobrevin nanodiscs overnight at 4°C. This method leads to quantitative formation of trans-SNARE complexes between nanodiscs based on the complete release of the synaptobrevin-2 (49–93) peptide, which was assessed by labeling the peptide with ¹⁵N and acquiring ¹H-¹⁵N HSQC spectra, or labeling the peptide with Cy5 at residue 79 and monitoring the fluorescence anisotropy of the dye. For experiments comparing binding of C₂AB* to cisSC-NDs and transSC-NDs, the cisSC-NDs were formed by incubating nanodiscs containing activated t-SNARE complex with 2-fold excess of synaptobrevin-2 (29–93) overnight at 4°C.

To estimate the concentration of the resulting ND, cisSC-ND or transSC-ND samples, a stock solution of solubilized lipids containing 5% RhoPE was prepared as previously described in the Materials and methods section. Dilutions of this stock solution were made and the absorbance at 570 nm was measured to make a standard curve. Samples were solubilized in 1% b-OG and their absorbance at 570 nm was measured. This measurement and the standard curve were used to estimate the total lipids in nanodisc samples. The total lipid concentration was divided by 200 to get the approximate concentration of the nanodiscs.

To estimate the concentration of SNARE complex samples formed using SNAP-25(11–82) R16C, Q34C or K76C labeled with tetramethylrhodamine, the dye absorbance at 555 nm was used. For this purpose, the extinction coefficient of tetramethylrhodamine at 555 nm in 20 mM HEPES pH 7.4 125 mM KCl buffer was first determined from a standard curve obtained from the absorbance at 555 nm of several samples made after serial dilution of a stock solution of Tetramethylrhodamine-5-Maleimide.

The SNARE motifs were mixed in an equimolar ratio in the following order: synaptobrevin-2 (29–93), SNAP-25A(141–203), SNAP-25A(11–82) K76C labeled with tetramethylrhodamine and syntaxin-1A (191–253) (for mcc) or syntaxin-1A (2–253) (for solubleSC), in the presence of 1 M NaCl. The assembly reaction was incubated in the presence of protease inhibitor cocktail A at room temperature overnight while rotating. The SNARE motifs that did not incorporate into the complex were removed by concentration-dilution at room temperature using 30 kDa Amicon centrifugation filters (Chen et al., 2002). The concentration of the SNARE complex was estimated by using UV-vis absorbance.

A stock solution of 0.5 µM Alexa 488-tagged C₂AB (C₂AB*), containing 5 µM BSA (Sigma Aldrich: A3059) to prevent binding of C₂AB* to the cuvette, was prepared in the appropriate buffer (20 mM HEPES pH 7.4 and varying amounts of KCl and other components) on the day of the experiment. The ND, cisSC-ND, transSC-ND, solubleSC or mcc stock solutions were buffer exchanged into the proper buffer by using concentration-dilution or during the size exclusion chromatography step. Other components such as 1 mM CaCl₂, 1 mM EGTA, 2.5 mM MgCl₂, 2 mM ATP or a combination of these reagents were added after the concentration step. All the stock solutions were kept on ice and the buffers were kept at room temperature. The ND samples can be kept on ice and used for experiments during a period of three days in the presence of protease inhibitor cocktail A.

C₂AB* fragments at 50 nM concentration were mixed with varying amount of NDs, cisSC-NDs or transSC-NDs in buffer containing 20 mM HEPES pH 7.4 and the concentrations of EGTA, Ca²⁺, Mg²⁺, ATP and KCl indicated in the figure legends. Each sample of a titration was prepared freshly by adding 20 µL of the stock C₂AB* solution to an appropriate amount of ND, cisSC-ND or transSC-ND in a 1.5 mL Eppendorf tube. Dilutions of the ND stock solutions were used if less than 5 µL of the original stock had to be pipetted to minimize errors caused by pipetting small volumes. The volume was made up to 200 µL by adding buffer. The tube was vortexed for 2-3 s and 180 µL of the solution was pipetted into an SOG cuvette (Fischer Scientific: NC9040289). Emission scan of the sample was immediately collected on a PTI Quantamaster 400 spectrofluorometer (T-format) at room temperature. All slits were set to 1.2 mm. The samples were excited at 480 nm and emission spectra from 500 to 620 nm were acquired. After the scan was collected, the sample was discarded, and the cuvette was washed with water followed by ethanol and dried using a stream of air. Multiple cuvettes were used during the experiment. This did not change the emission scans. Collecting the spectra after 5 min incubation at room temperature also did not change the emission scan. FRET efficiency was calculated using the formula: FRET = (I₀ –I)/ I₀, where I₀ is the fluorescence intensity of 50 nM Syt1 C₂AB* at 518 nm and I is the fluorescence intensity of the sample at 518 nm. The binding data were fitted to a Hill function (FRET=Bmax∗(C∧h)/((Kd∧h)+(C∧h)), where C is the concentration of NDs, cisSC-NDs, transSC-NDs or solubleSCs, using GraphPad PRISM7.

The titrations of WT and mutant C₂AB* with NDs and cisSC-NDs under a variety of conditions and with different lipid compositions were performed over a long period of time as the project developed and the results suggested new experiments to be performed. The data shown in Figures 5–9 and associated figure supplements, and summarized in Supplementary file 1, correspond to the last sets of experiments performed for presentation in this paper, all of which were consistent with the previous data. Each titration involving a given set of reagents and conditions was repeated at least twice and in most cases three or more times over the course of this work, but a comprehensive statistical analysis for groups of data corresponding to each figure panel is hindered by the fact that the groups of experiments performed previously as the project developed varied (e.g. with regard to which mutants were included). The reproducibility of the data that we obtained in this type of experiments is illustrated in Figure 9—figure supplement 4 by the statistical analysis of the results of independent repeat experiments performed under the conditions of Figure 6G,H with the same set of mutants and three different preparations of NDs and cisSC-NDs. This analysis and, more importantly, the overall consistency of the data obtained in the large number of experiments that we performed strongly support the key conclusions drawn from the titrations.