Electron Microscopy

To ensure putting good quality sample for cryoEM imaging, in the negative stain samples we look for better filament separation, twisting of filaments and overall quantity of filaments. The dilution factor that meets all those criteria will be used to image in Titan Krios microscope for cryoEM imaging. Briefly, amplified aSyn filaments will be centrifuged for 20k g for 1 hour and the palette will be resuspended in 10 mM Tris 50 mM NaCl buffer of pH 7.5. The concentrated filaments will be diluted in the above-mentioned buffer for several dilutions and briefly sonicated before placing them in the formvar carbon supported copper grids for negative stain electron microscopy imaging where uranyl acetate will be used for staining. Next samples will be imaged in JEOL 1400 microscope where we will monitor the overall quality of the amplified aSyn filaments.

CryoEM imaging

Samples selected in the negative stain EM will be used for further imaging on a Thermo Fischer Titan Krios microscope operating at 300 kV equipped with post-column BioQuantum Energy Filter with K2 Summit direct detector. The process of sample preparation for cryoEM imaging is as follows.

Sample will be applied to the glow-discharged 2/2 holey carbon coated copper grids (Quantifoil R2/2, 200 mesh) for 60s. After blotting for 3.5 s sample will be plunge-frozen in liquid ethane using an FEI Vitrobot Mark IV. Next, samples will be imaged on Titan Krios microscope. Initially samples will be screened for ice quality. Grids that show more amorphous and moderately thin ice will be used for collecting data. Data will be collected in either counting or super resolution mode depending on the need of overall resolution.

Image processing

Filaments will be reconstructed in RELION-3.1 using helical reconstruction. Movie frames will be corrected for beam-induced motions and dose-weighted using MotionCor2. Non-dose-weighted micrographs will be used for CTF estimation with CTFFIND-4.1. We will manually pick around 11000-15000 filaments. Particle segments will be extracted using a box size of 900 pixels and an interbox distance of 14 Å and downscaled to 300 pixels for 2D classification. Crossover-distances will be obtained by manual measurements in the 2D classes and will be used to calculate initial estimates for the helical twist of the different filament types. Relion_helix_inimodel2d program will then be used to reconstruct the de novo 3D initial models from the 2D classes. Later, segments will be re-extracted without down-sampling in boxes of 320*320 pixels for use in 3D classifications and auto refinements. Final reconstructions of all the maps will be done after CTF refinement followed by 3D auto-refinement and standard RELION post-processing with a soft solvent mask that extended to 20 % of the box height.

Model building

Initial segmentation of individual subunits for each filament will be done using Segger in UCSF Chimera.  Known αSyn structures will then fit to the individual subunits. In filaments where the known structures will appear to agree with the overall fold observed in the density the known structures will be trimmed to the core residues that fit into the density. Where the folds of the known structure and density appeared to be different, Pathwalker will run to generate an initial backbone trace; the density will be seeded with 100 pseudoatoms and a threshold at which point the overall fold can be seen will be provided for each run of Pathwalker. As with the filaments that fit the known structures, models based on Pathwalker will be trimmed to contain just the core portions of the fold. These initial models, both from known structures and Pathwalker, will be then completed in Coot using the density map and the Baton Build option. The initial models will then converted into full atom models in Coot and refined with Phenix real space refinement. Refined models will then imported back into Coot and adjusted manually to improve fit to the map. For each filament, 10 copies of the corresponding individual subunit will then propagated into the density map at adjacent positions. Phenix real space refinement and Coot manual optimization will be iterated, maximizing the model quality, model clash and fit to density. Final models will then be analyzed using the comprehensive validation tools in Phenix. Model visualization and interpretation will be carried out in Coot and UCSF Chimera.