Synaptic proteins promote calcium-triggered fast transition from point contact to full fusion

Abstract

The molecular underpinnings of synaptic vesicle fusion for fast neurotransmitter release are still unclear. Here, we used a single vesicle-vesicle system with reconstituted SNARE and synaptotagmin-1 proteoliposomes to decipher the temporal sequence of membrane states upon Ca(2+)-injection at 250-500 μM on a 100-ms timescale. Furthermore, detailed membrane morphologies were imaged with cryo-electron microscopy before and after Ca(2+)-injection. We discovered a heterogeneous network of immediate and delayed fusion pathways. Remarkably, all instances of Ca(2+)-triggered immediate fusion started from a membrane-membrane point-contact and proceeded to complete fusion without discernible hemifusion intermediates. In contrast, pathways that involved a stable hemifusion diaphragm only resulted in fusion after many seconds, if at all. When complexin was included, the Ca(2+)-triggered fusion network shifted towards the immediate pathway, effectively synchronizing fusion, especially at lower Ca(2+)-concentration. Synaptic proteins may have evolved to select this immediate pathway out of a heterogeneous network of possible membrane fusion pathways.DOI:http://dx.doi.org/10.7554/eLife.00109.001.

Keywords: Other; SNARE; complexin; neurotransmitter release; synaptic vesicle fusion; synaptotagmin.

Figure 1.

Single vesicle–vesicle microscopy for monitoring changes in the membrane and contents upon Ca²⁺-injection. (A) Labeling and reconstitution scheme of our single-vesicle lipid/content mixing system. Other factors and proteins can be added to the system. The detailed experimental protocol is described in (Kyoung et al., 2012) with modifications as described in ‘Materials and methods’. (B) Representative real-time fluorescence intensity traces with donor (synaptobrevin and synaptotagmin 1) and acceptor (syntaxin and SNAP-25) vesicles (red/upper traces: lipid dye fluorescence intensity, green/lower traces: content dye fluorescence intensity). Time point 0 indicates the instance of Ca²⁺ injection at 500 μM (blue vertical lines). Shown are instances of ‘immediate’ fusion, as defined by a content dye fluorescence intensity jump during the first 600 ms time bin upon Ca²⁺-injection along with a simultaneous jump in lipid dye fluorescence intensity, ‘delayed’ fusion, as defined by a content dye fluorescence intensity jump during a subsequent time bin, and ‘hemifusion-only’, as defined by a lipid dye fluorescence intensity jump without a content dye fluorescence intensity jump during the observation period of 50 s. For illustration purposes, only 30 s of the observation period are shown since there was no change in the subsequent 20 s. (C) Bar graph of the percentage of immediate, delayed fusion, and hemifusion-only events involving SNAREs and synaptotagmin 1. The bar graph was normalized with respect to the number of fluorescent spots that exhibited at least one lipid-mixing event during the observation period. Note, that there are fewer immediate fusion events compared to our previous experiments that were carried out at higher Ca²⁺ concentration (Kyoung et al., 2011). DOI: http://dx.doi.org/10.7554/eLife.00109.003

Figure 2.

Imaging of donor/acceptor interface morphologies by cryo-EM before and after Ca²⁺ addition. (A–F) Cryo-EM images of mixtures of donor (synaptobrevin and synaptotagmin 1) and acceptor (syntaxin and SNAP-25) vesicles before (A–C) and approximately 35 s after (D–F) 500 μM Ca²⁺ addition (panels B, C, E, and F are close-up views). Vesicles were imaged in the holes of the substrate carbon film, visible as the darker areas in the image, in conditions that clearly show the lipid bilayers (‘Materials and methods’). The particular images shown in this figure were selected out of total of 16 (before Ca²⁺ addition) and 21 (after Ca²⁺ addition) EM micrographs, respectively, with emphasis on showing point contacts before Ca²⁺ addition and extended interfaces after Ca²⁺ addition in order to illustrate the variety of these particular states. Arrows indicate interfaces between vesicles that are approximately perpendicular to the direction of the projection (see ‘Materials and methods’ for definitions of all contact and interface types). Large black arrow: point contact (representative close-up in B), small black/white arrow: hemifusion diaphragm (representative close-ups in C and E), and large white arrow: extended close contact (representative close-up in F). Scale bars in A and D are 100 nm, and 20 nm in B, C, E, and F. (G) Distribution of vesicle sizes before and after Ca²⁺ addition. Vesicle diameters were calculated from all cryo-EM images in both conditions (‘Materials and methods’). (H) Bar graph of the percentage of various vesicle interfaces, that is, point contacts, hemifusion diaphragms, and extended close contacts (including a few instances of mixed, i.e., extended/hemifused, interfaces), normalized with respect to the total number of interfaces observed before and after addition of 500 μM Ca²⁺, respectively (‘Materials and methods’). In addition to the changes in the distribution of vesicle interfaces upon Ca²⁺ addition, some amount of complete fusion between vesicles occurred as indicated by the shift of the diameter distribution in panel G towards larger values. Source files of all cryo-EM micrographs used for the quantitative analysis are available in the Figure 2—source data 1. DOI: http://dx.doi.org/10.7554/eLife.00109.004

Figure 3.

Image density profile analysis of selected vesicle–vesicle interfaces. (A) Image density profile analysis of two representative point contacts between vesicles at zero Ca²⁺. Dotted lines indicate 2 nm thick sections selected from the cryo-EM images. These sections were used to generate the profiles shown to the left and the right of the close-up views in the center (for details, see ‘Materials and methods’). (B) Image density profile analysis of a transition from extended close contact to hemifusion diaphragm after addition of 500 μM Ca²⁺. As in panel A, selected sections are indicated by dotted lines, and the corresponding profiles are shown the left and right of the close-up view in the center. DOI: http://dx.doi.org/10.7554/eLife.00109.006

Figure 4.

Upon Ca²⁺-injection, immediate fusion events start form hemifusion-free point contacts whereas initially hemifused states are slow to fuse. (A) Representative real-time fluorescence intensity traces of instances of immediate fusion with imaging of initial florescence intensity levels right after addition of donor (synaptobrevin and synaptotagmin 1) to acceptor vesicles (syntaxin and SNAP-25) (red/upper traces: lipid dye fluorescence intensity, green/lower traces: content dye fluorescence intensity). Time point 0 indicates the instance of Ca²⁺ injection at 500 μM (blue vertical lines). As in Figure 1, immediate fusion is defined as a content dye fluorescence intensity jump that occurs during the first 600 ms time bin upon Ca²⁺ injection. In all cases, no significant (i.e., above noise level) lipid fluorescence intensity change was observed during the incubation period, which excludes the possibility of a hemifused initial state before Ca²⁺ triggering. Moreover, ‘all’ (seven) observed traces with immediate fusion upon Ca²⁺ injection exhibited this behavior. (B) Representative real-time fluorescence intensity traces of cases where the membranes are hemifused prior to Ca²⁺-injection. Initial hemifusion was characterized by a significant increase (i.e., above noise level) in only lipid dye fluorescence intensity between the initial recording and at the end of incubation period at 0 Ca²⁺. For approximately 400 fluorescent spots co-localized in both the initial recording and at the end of the incubation period, we observed 51 traces showing (1) an increase in lipid dye fluorescence and (2) no change in content dye fluorescence intensity during the incubation period but before the Ca²⁺ injection. For most of the 51 cases, no content mixing was observed upon Ca²⁺-injection during the observation period of 50 s except for two cases (see one example in the bottom panel) where delayed content mixing occurred at a later time. For illustration purposes only 20 s of the 50 s observation period are shown all traces but one since there was no change in the subsequent 30 s. DOI: http://dx.doi.org/10.7554/eLife.00109.007

Figure 5.

Modeling of putative protein density in a pre-fusion point contact state. (A) Close up view of Figure 3A, top middle panel, around a significant density feature (greater than a factor of two above noise level) that is not part of the membranes (red arrow). Only the density features on the left-hand side of the panel were considered since the carbon grid (dark feature in the lower right corner in Figure 3A, top middle panel) may have affected protein interactions. (B) Overlay of the model of the complex of the C2A–C2B fragment of synaptotagmin 1 (cyan) and the neuronal SNARE complex (synaptobrevin—blue, syntaxin—red, SNAP-25—green) based on 34 single-molecule FRET experiments (Choi et al., 2010). The last four C-terminal helical turns of synaptobrevin were modeled as a random coil. (C) Overlay of a model of the complex of the C2B domain of synaptotagmin 1 and the neuronal SNARE complex based on proximity between the C2B domain polybasic region and residues D186, D193 of SNAP-25, both of which show sizeable chemical shift perturbations in NMR experiments upon complex formation, among other perturbations (Dai et al., 2007). Six C-terminal α-helical turns of synaptobrevin and one α-helical turn of syntaxin were modeled as a random coil. The transmembrane domains were modeled as α-helices, and the linkers as random coils. See ‘Materials and methods’ for details. DOI: http://dx.doi.org/10.7554/eLife.00109.008

Figure 6.

Effect of complexin on Ca²⁺-triggered fusion probability. Probability of fusion vs time upon 500 μM (A) or 250 μM (B) Ca²⁺-injection (i.e., occurrence of Ca²⁺-triggered content-mixing events vs time) with and without 5 μM complexin. The same donor (synaptobrevin and synaptotagmin 1) and acceptor (syntaxin and SNAP-25) vesicles were used as in Figure 1. Histograms were normalized to the number of respective fluorescence intensity time traces showing at least one lipid-mixing event during the observation period of 50 s (see ‘Materials and methods’ for details). For illustration purposes only 20 s of the 50 s observation period are shown since there were few events in the subsequent 30 s; a time binning of 600 ms was used. Black lines are fits to exponential decay functions over the entire 50 secs. For 500 μM Ca²⁺, in the absence of complexin the fitted function is f(t) = 0.0015 + 0.096e^−t/1.93 and it is 10 times more likely compared to a two-exponential fit, and in the presence of 5 μM complexin the fitted function is f(t) = 0.0021 + 0.084e^−t/2.38 + 0.18e^−t/0.36 and it is 10⁷ times more likely compared to a single exponential fit. For 250 μM Ca²⁺, in the absence of complexin the fitted function is f(t) = 0.0026 + 0.027e^−t/2.98 and it is three times more likely compared to a two-exponential fit, and in the presence of 5 μM complexin the fitted function is f(t) = 0.0005 + 0.10e^−t/0.89 + 0.011e^−t/14.5 and it is 10¹⁴ times more likely compared to a single exponential fit. DOI: http://dx.doi.org/10.7554/eLife.00109.009

Figure 7.

Model of multiple fusion pathways upon Ca²⁺ injection starting from a point contact. Complexin synchronizes Ca²⁺-triggered fusion by increasing the number of immediate fusion processes from point contact to fusion pore opening, relative to delayed fusion pathways involving stable hemifusion diaphragms and other long-lived intermediates. DOI: http://dx.doi.org/10.7554/eLife.00109.010

Figure 8.

Histograms of the occurrence of Ca²⁺-triggered lipid-mixing events with and without 5 μM complexin, corresponding to the experiments shown in Figure 6. (A) Ca²⁺ injection at 500 μM. (B) Ca²⁺ injection at 250 μM. Black lines are fits to exponential decay functions over the entire 50 s. For 500 μM Ca²⁺, in the absence of complexin the fitted function is f(t) = 0.0014 + 2.66e^−t/0.11 + 0.22e^−t/1.94 and it is 10¹⁸ times more likely compared to a single exponential fit, and in the presence of 5 μM complexin the fitted function is f(t) = 0.0006 + 0.42e^−t/0.81 + 0.05e^−t/4.76 and it is 10¹⁰ times more likely compared to a single exponential fit. For 250 μM Ca²⁺, in the absence of complexin the fitted function is f(t) = 0.0023 + 0.13e^−t/0.61 + 0.05e^−t/9.05 and it is 10⁶ times more likely compared to a single exponential fit, and in the presence of 5 μM complexin the fitted function is f(t) = 0.003 + 0.33e^−t/0.51 + 0.05e^−t/6.34 and it is 10¹⁴ times more likely compared to a single exponential fit. DOI: http://dx.doi.org/10.7554/eLife.00109.011