Lipid Regulated Intramolecular Conformational Dynamics of SNARE-Protein Ykt6

Cellular informational and metabolic processes are propagated with specific membrane fusions governed by soluble N-ethylmaleimide sensitive factor attachment protein receptors (SNARE). SNARE protein Ykt6 is highly expressed in brain neurons and plays a critical role in the membrane-trafficking process. Studies suggested that Ykt6 undergoes a conformational change at the interface between its longin domain and the SNARE core. In this work, we study the conformational state distributions and dynamics of rat Ykt6 by means of single-molecule Förster Resonance Energy Transfer (smFRET) and Fluorescence Cross-Correlation Spectroscopy (FCCS). We observed that intramolecular conformational dynamics between longin domain and SNARE core occurred at the timescale ~200 μs. Furthermore, this dynamics can be regulated and even eliminated by the presence of lipid dodecylphoshpocholine (DPC). Our molecular dynamic (MD) simulations have shown that, the SNARE core exhibits a flexible structure while the longin domain retains relatively stable in apo state. Combining single molecule experiments and theoretical MD simulations, we are the first to provide a quantitative dynamics of Ykt6 and explain the functional conformational change from a qualitative point of view.


Note 1: Design of rYkt6ΔC FRET construct
Single molecule FRET (smFRET) is a direct and powerful technique for demonstrating the conformational changes between the longin domain and the SNARE core of rYkt6∆C. Nevertheless, for smFRET measurements to perform well, the distances to be determined have to fall around the Förster radii R0 of a chosen FRET pair since smFRET measurements are most sensitive there. The two labeling sites on the molecule, rYkt6ΔC, has to be chosen such that the distances between them will optimally report any conformational changes between the longin domain and the SNARE core.
There is a native Cysteine residue, C66, in the sequence of Ykt6 from Rattus norvegicus besides the CCAIM terminal. To minimize the perturbation of the original construct, we reserved this C66 in the longin domain as one of the FRET labeling sites.
We then narrowed down our labeling site selection to the central part of the core domain.
To choose the proper site, we fetched the longin domain sequences from Pfam and UniProt, ran multiple sequence alignment with ClustalWS, and calculated the conservation scores for each residue on rYkt6. Residues with low conservation scores was chosen as potential labeling sites. We have chosen E175 as our labeling residue on rYkt6ΔC. Since the only crystal structure of rYkt6 available (PDB-ID: 3KYQ) is in the closed form bound with DPC, we used MD simulation result to predict an open structure. respectively. The Förster radii R0 for these two pairs are 51 Å and 56 Å, which is suitable for reporting conformational changes between 20 Å and 62 Å. As a remark, previous work Hasegawa et al. 2004 found that targeting function of Ykt6 is regulated by its longin domain (sequence: 1-137) and normal Ykt6 requires a soluble intermediate. In our earlier attempts, we found that the native cysteine C66 is a possible site related to the solubility of Ykt6. The final construct Ykt6/E175C keeps the soluble features as the wildtype one. Although the native C66 labeling site is not fully exposed to solution in the structure with PDB id 3KYQ, the construct rYkt6 E175C can be readily doubly labeled which is observed in single-molecule experiment. (see Figure S2).

Note 2: Structure diversity analyzed by MD simulations at a longer distances range
As the Cys66-Glu175 C distance falls within longer distance range, i.e. 3.2-3.9 nm, more prominent structural variability of the SNARE core was observed while the longin domain retains its structure. The conformations were divided into 49 clusters with a cutoff of 1.1 nm ( Figure S7). The SNARE core becomes more stretched and less compact. The positions of the Glu175 Cα are obviously distributed in a larger space than that of the 'closed' state, ranging around the αA helix, the βC-βD turn, the αD helix and the βD-βE turn ( Figure S8(a), (b)). In spite of the more extended conformations in the 'open' state, two clusters were still found with the longin-SNARE BSA of more than 12 nm2. The 1st cluster adopts a similar αG orientation as the 3rd and 5-8th clusters of the 'closed' state with the helix lying parallel with αA ( Figure S8(c)). The 2nd cluster shows an antiparallel orientation between the αG and αA helices, reminiscent of the 9th cluster of the 'closed' state ( Figure S8(d)). The N-terminus of αG helix is uncoiled in these two clusters, resulting in the longer distance between Glu175 and Cys66. We also inspect the centre structures with the smallest BSA ( Figure S8(e-j)). In these structures, the αG helix is more than 0.9 nm away from the αA helix and has little contact with the longin domain.
Nevertheless, the αG helix tends to pack with SNARE core itself in the last six clusters,     Fig. S3(b) shows that the normalized correlations owning the same correlation decay. This confirms that there is no time dependent difference between the four correlations. Since Poly-Proline15 can be seen as a rigid molecule without intramolecular conformational dynamics, these measurements are consistent with theoretical expectations. Regarding the yielded Poly-Proline fitting results shown in tab. S1 , one can use the p-value marking the ratio of lateral to axial dimensions of the focal region for further measurements. The threshold of this parameter were set to 6.85 as the lower and 6.9 as the upper boundary conditions.