Complexin-1 regulated assembly of single neuronal SNARE complex revealed by single-molecule optical tweezers

The dynamic assembly of the Synaptic-soluble N-ethylmaleimide-sensitive factor Attachment REceptor (SNARE) complex is crucial to understand membrane fusion. Traditional ensemble study meets the challenge to dissect the dynamic assembly of the protein complex. Here, we apply minute force on a tethered protein complex through dual-trap optical tweezers and study the folding dynamics of SNARE complex under mechanical force regulated by complexin-1 (CpxI). We reconstruct the clamp and facilitate functions of CpxI in vitro and identify different interplay mechanism of CpxI fragment binding on the SNARE complex. Specially, while the N-terminal domain (NTD) plays a dominant role of the facilitate function, CTD is mainly related to clamping. And the mixture of 1-83aa and CTD of CpxI can efficiently reconstitute the inhibitory signal identical to that the full-length CpxI functions. Our observation identifies the important chaperone role of the CpxI molecule in the dynamic assembly of SNARE complex under mechanical tension, and elucidates the specific function of each fragment of CpxI molecules in the chaperone process.

where ℎ is assumed as the spatial length of the folded portion (such as coiled coil) projected along pulling direction, which also contributes to the final extension. The !"# is stable under the same average force % (fixed trap), so we focus on the protein extension, i.e., $ + ℎ, which we named % = $ + ℎ, = 1, … ,5 . In the fully folded state (native state), ℎ = 2 nm is determined from the x-ray structure of the protein 1 , whereas for the fully unfolded state, ℎ = 0.
Then % , = 1, … , 5, is the contour length of the unfolded polypeptide in different SNARE assembly states, = 0.15/0.365 = 0.41 is the ratio of the contour length of an amino acid (aa) in the helical conformation to that in the coil conformation. The extension of the unfolded polypeptide is related to the contour length and force through the Marko-Siggia formula. The contour length difference at this force point is evaluated based on the model shown in Fig. S1 as where ∆ % is the extension difference between the two states now under the same tension % . Finally, the average of these contour length differences is calculated. In the case of the α-helix, every 3.6 amino acid residues, the helix rises by a circle, the pitch is 0.54nm, and the span of each residue is 0.15nm. When the polypeptide chain is fully extended, the length of each amino acid residue is 0.365 nm.

Supplementary
Protein extension for state 1: Protein extension for state 2: Protein extension for state 5: Protein extension for state 6: We could get the number of unfolded amino acids ( % , = 1,6) is 34 aa and 147 aa and ℎ ' = Specifically, the Syntaxin unfolded 47 aa from state 5 to state 6. As Syntaxin contributed 49 aa in the state 2 (A254 to L205C), 2 aa unfolded between state 2 and state 3.
Since the crystal structure analysis only tackles the ensemble averaged conformations in the absence of mechanical force, the protein extension changes in two intermediate States 3 and 4 can only be inferred through the following, Protein extension for state 3: Protein extension for state 4: to the unzippering of +2 layer to -1 layer in VAMP (-2 layer remain folded, Fig. 1d, state 4).

Interaction test between SNAREs and CpxI
The Binding assay between SNAREs and complexin is completed (FigS2a: SNARE monomers, FigS2b: SNARE complex). Among them, the mole ratio of CpxI and SNAREs is 5:1 (due to the larger time cost of SNARE complex sample acquisition, CpxI is selected as excess protein), and they were incubated for 1hr at room temperature. The loading quantity of the same protein in all swimming lanes is strictly quantitative control, and the same stock buffer containing TCEP (strong reducing agent) is added in advance for processing. In Figure S2a, the first lane is CpxI, the second/forth/sixth lane is Syntaxin/VAMP/SNAP25, the third/fifth/seventh lane is SNAREs and CpxI after incubation. For we did not see significant difference on the gel, complexin does not binding to any SNARE monomers. In Figure 2Sb

Primers for the preparation of DNA handles
Primer sequences for the biotin-and digoxigenin-modified 2260-bp handles were:

Middle cross-linked constructs of SNARE complex
The half-zipped state of the N-terminal cross-linked SNARE complex is too short, so we needed to build a new SNARE pulling system with sable and long half-zipper state. To got four The sequences with underline were the native sequences, red marked amino acids were mutated amino acids, the FLAG tag (DYKDDDDK) at the C-terminal was used for protein purification, but was not used in this assay. SAGG at N-terminal and GGSGNGSGG at C-terminal were protein linkers, which were designed to increase the flexibility of protein.

Single molecule experiment on -2 layer and -6 layer cross-linked SNARE complex
We have tried all the four crosslinking sites (-1, -2, -3, -6), and found that the crosslinking at -1 or -3 layer was unstable, as a result, the efficiency to form a tether is very low, and the tether would easily break under a force of 16 pN. We have successfully formed tether with high efficiency for the crosslinking at -3 and -6 layers. Although the success rate is low for the crosslinked SNARE at -3 layer, we have successfully collected the dynamic transition signal at ~ 16 pN. In figure S5a, the de-assembly signals obtained by the experiments of the SNARE complex of -2 layer was hopping at 10 nm, corresponding to the de-assembly process of Cterminal and partial N-terminal of 0 to -2 layer (about 28+8 amino acids in total). In figure S5b, at the same force, the de-assembly signal hops at 20nm, which correspond to the de-assembly process of the C-terminal of the SNARE complex and partial N-terminal of the 0-6 layer (about 28+22 amino acids in total). Because the -2 layer is too close to the 0 layer, and the de-assembly protein fragment covering the N-terminal is too short, insufficient for us to study the interplay of the Cplx with the N-segment of SNARE complexes. Therefore, the subsequent experiments are mainly based on the structure of the -6 layer with longer hopping. We also recorded the dynamic transition of the -2 layer crosslinked SNARE complex. The -2 layer SNARE complex is stretched to a hopping signal in presence of a fixed trap separation.

Supplementary
We observed the transition among three state of SNARE complex in equilibrium (Fig. S5c).
Then CpxI was added through protein channels, and significantly long stay pauses were observed during the rapid transition from SNARE complexes (in green, Fig. S5c), where the SNARE complexes were stabilized in a quad helical bundle state, revealing a new CpxI dependent state. This is consistent with the signal that the -6 layer SNARE complex could be stabilized in the four-helical bundle state (i.e., the C-terminal stabilized state in Fig. 2).
In summary, we have tried to crosslink the SNARE complex at various layers and verified that the -6 layer allows stable performance and maximum interaction region on the SNARE complex with the Cplx molecule. Although different crosslinking site shows slightly different force-extension graph (mainly the open distance is different), the interaction of the Cplx and SNARE complex in the same region suggests consistent mechanism. On the data interpretation, the major principle is the same, the only difference is that when we fit the force-extension curve with the worm-like chain (WLC) model, a different contour length for each of the four-helix bundle, polypeptide was used according to the crosslinking site. In order to maximize the interaction region on the SNARE, most of our experiment was conducted with -6 layer crosslinking on SNARE complex.

Chamber fabrication
Standard cover-slides were cleaned with dishwashing liquid, then washed with clean water, drained, and ultrasonic cleaned in deionized water for 5 minutes. After rinsing in deionized water, the slides were ultrasonic cleaned in anhydrous ethanol for 5 minutes, and then dried.
The glass slide was cut with 6 inlets/outlets using the Laser engraving machine. Meanwhile, the parafilm was cut into three channels as described in Supplementary figure S2.
Supplementary Figure 6. Sketch map of chamber cell. It is constructed by sandwiching one p arafilm between two slides, with parafilm thickness as the sample storage space.
The custom chamber was made using the following steps： (1) Place one parafilm on the slide without laser engraving, then mount the glass pipettes to form the connection channels between the top/bottom channels with the main channel. Put another pipette to serve as the protein channel if necessary. Then put on top of them another parafilm and coverslide, and gently press on the assembly. Such that the parafilms and the capillary are sandwiched in between.
(2) The sample pool is turned over and placed on the metal hot plate to melt the parafilm, and the appropriate pressure is applied to make the slide and the melted parafilm tightly bonded for about 20 minutes, and then it can be cooled after taking out.
(3) Assemble the chamber cell according to the diagram and install it on the optical tweezers instrument. Fill the chamber cell with 0.2% NaN3 to prevent bacterial growth when finish the experiment.
When assembling the chamber, be careful to test the leakage using clean water, and make sure there will be no air bubbles inside the channels. The tiny air bubbles can be removed by gently tapping on the plastic tubing to generate gentle vibration.

Configurations of the SNARE with and without Cpx
In In summary, based on the explanation, the so-called "C-terminal stabilized", "Middle-clamped", and "C-terminal blocked" states demonstrate the function of Cpx molecule on the SNARE complex, while State 1 through 5 denote four different configurations of SNARE complex.
They are associated but not exactly equal.

Definition of signal rate
The fragments of or complete CpxI interacts with the SNARE complex, and the Extension trace For we injected only buffer (first column in fig.6a) though protein channel as control, and the majority (81% of 31 molecules under test) showed no change after the injection. And once the fragment 83-134 aa (mostly part of CTD, without CH) was injected, and 18 of 27 SNARE complexes showed no change (second column of Fig. 6a), which was a negative control of experiments in Fig. 5. In Fig 6a, we define the signal rate as,

HMM analysis of C-terminal stabilized state and middle clamp state
In the presence of 8 μM full-length CpxI, 49 SNARE complexes changed their state after the addition of CpxI. We observed that SNARE complex could be clamped by CpxI into the exactly half-zippered state (middle-clamped state, 7 molecules, 14% signal rate, Fig. 2a, 2b, 2d).
Accordingly, the dynamic folding of SNARE complex only took place between 3~4 states at this situation. Surprisingly, the SNARE complex can also maintain in C-terminal stabilized state after the addition of CpxI (C-terminal stabilized state, 20 molecules, 41% signal rate, Fig.   2a, 2b, 2d). Accordingly, SNARE complexes have changed from hopping among 2-5 states to being maintained only in the second state with the addition of CpxI.

Unfolding force statistics of SNARE complex with full-length and 1-83aa Cpx
We also counted the unfolding force of the SNARE complex in the absence and presence of either full length CpxI or the fragment 1-83 aa. The hopping takes place at ~21 pN in the absence of CpxI, while hopping happens at a higher force level in presence of full or fragmental CpxI. The maximum likely force increased from ~21 pN to ~24 pN and some events happen at even higher force of ~43 pN in the presence of full length CpxI (top panel in Fig. S7). Although similar phenomenon has been observed with the fragment 1-83 aa of CpxI (bottom panel in Fig. S7), it is worth to note that there are more high-force signals in the full-length Cpx, which may imply more stable binding of full-length Cpx to SNARE complexes.

The stabilization function of CpxI NTD
To pinpoint the possible role of the NTD in the CpxI -dependent SNARE disassembly, we removed the NTD in the CpxI construct. In the presence of CpxI 26-83 aa, 38 dynamic molecules changed their state after the addition of CpxI (Fig. 4a, 4b). Interestingly, the extension-time traces of 25 molecules (66%) were still stabilized at C-terminal stabilized state (Fig. S8). The rate of C-terminal stabilized state decreased dramatically, showing that the removal of NTD caused significant changes in the CpxI-dependent SNARE disassembly (Fig.   4d), which indicates that the CpxI stabilizes the SNARE complex critically depending on its N-

No interaction between 1-83 aa and 83-134 aa
In order to test whether mixed CpxI 1-83 aa and 83-134 aa would interact in advance, we  Figure S10 show the FECs of C-terminal stabilized state (Fig.S10a), C-terminal blocked state ( Fig.S10b), middle clamped state (Fig.S10c) after the addition of CpxI.