Optimizing the Protein Stability in Thick Filament Cryo-EM Sample Preparation Using a PEGylation Technique

Vertebrate myosin filaments are especially important for two reasons: first, they have not been resolved to high-resolution, and second, they have more similarities with human muscle which makes them a suitable model to study muscle diseases. Even though cryo-EM has made numerous advancements in this field, the structural analysis of these filaments in vertebrates is limited to low-resolution negatively stained specimens [1-4]. Preserving the vertebrate thick filaments in the isolation and freezing step seems to be the main challenge in the way of cryoEM studies, which is a much smaller problem when working on invertebrate myosin filaments. We believe that these fragile particles tend to aggregate and/or become denaturated, due to their adhesion to the air-water interface during vitrification. In addition to this apparent fragility, the structural arrangement of myosin heads has shown to be sensitive to temperature in the relaxed state [5] which makes the cryo-EM sample preparation process even more challenging. As a solution to this problem, we have found that using a PEGylation technique [6] we can protect the structure of both backbone and myosin heads from these issues during vitrification. This will enable an easier cryo-EM structural determination. The preliminary results showed a significant improvement in 2D class averages in samples treated with Polyethylene Glycol (PEG). Myosin filaments were isolated from rabbit psoas muscle under relaxing conditions [7]. Using negative staining electron microscopy, the concentration and ordered arrangement of the heads were assessed. Myosin filaments suspended in a relaxing solution containing 80 mM KCl, 20 mM MOPS, 5 mM EGTA, 2 mM dithiothreitol, 5 mM MgCl2, and 2.5 mM ATP (pH 7) were incubated with 4% PEG3350 for 30 min at 27°C. No attempt to stop the reaction was performed and 5ul of the suspension was applied to the grid bar side of a glow-discharged R1.2/R1.3 Quantifoil grid, incubated for 60 s, rinsed with 4% PEG 4000, and backblotted for 1 second in 100% humidity at 27C. Another grid was prepared an in the same condition except filaments were not incubated in PEG. Grids were plunged into liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific) and examined on the Titan Krios operated at 300keV for 30 min at 27°C. Movies were recorded with a total dose of ∼ 60 e−/Å2 and defocus range of -0.5 to -2. Data pre-processing was performed using MOTIONCORR2 [8] and GCTF [9] through RELION software. Keeping the polarity of the filaments in mind we manually picked filaments where the first crown of myosin heads (crown zero) can be clearly clearly distinguished from the filament backbone. We only extract particles from the myosin binding protein C (MyBP-C) zone. Selected filaments were then helically extracted into 1024 x 1024 pixels boxes, with an inter-box distance of 429Å. 2D classification was done using cisTEM on a stack of 29,576 extracted segments of the PEG treated samples and a stack of 27,477 extracted segments of the non-treated native sample. The result of the classification shown in fig. 2. It has been shown that PEGylated proteins will not aggregate or become denaturated during vitrification and Cryo-EM reconstructions of these proteins are similar to those of native proteins [6]. In our experiments, cryo-EM of native filaments without pegylation showed disordered heads and disassembled backbone (Fig 1b). We realized that even increasing the temperature of the Vitrobot chamber to 27°C during the plunge freezing will not prevent the nature of the disordered state populated by lowering temperature at rest. However, filaments treated with PEG looked homogenous and heads appear to lie on the surface of the filament (Fig 1c). As opposed to invertebrates, vertebrate thick filaments lack a perfect helical symmetry. In addition, the 4- or 7-fold rotational symmetry of invertebrates is replaced by 3-fold symmetry in vertebrates. While the helical symmetry is not present, a pseudo helical symmetry can be assumed as every three crowns along the axis of the filament are related by a 120∘ turn and a 429Å rise. MyBP-C is located between the 3rd and 10th 429A axial repeats from the bare zone [10]. Considering these features, we tried to align thick filament segments. 2D class averages of the PEG treated filaments showed a significant difference from the native ones (Fig. 2). We visualized some details both in the backbone and on the surface of the PEG treated filaments which are not seen on the 2D class averages of the native filaments. We used Relion3.1 helical initial model, initially developed for structure determination in amyloids, with a cross-over distance of 860A ̊ and C3 in-plane symmetry, to produce a featureless helix. Then we selected filaments from classes with similar class averages from the PEG treated group (Fig. 2b). These segments were then 3D refined in cisTEM, using 3D Autorefine, while imposing the existing C3 symmetry. The projection of the reconstructed map at ∼3 nm resolution, resembles the features seen in 2D reference-free class averages (Fig. 2c). In this preliminary reconstruction, our aim is to develop a method capable of reconstructing filaments with pseudo or limited helical symmetry. Further efforts are in progress to collect a new larger dataset, development of methods to relax the helical symmetry, and 3D reconstruction of various parts of the filaments. Supported by NIH/NIAMSD. Electron micrographs of isolated vertebrate thick filaments. (a) A negatively stained image of a rabbit psoas thick filament. Heads look ordered on the backbone. (b): Filament suspension vitrified. Heads look disordered and filament backbones look disarrayed. (c) Frozen hydrated human Cardiac thick filament treated with 4% PEG 3350. Heads look ordered on the backbone. Class averages from electron micrographs of frozen-hydrated filaments. (a) Various class averages from 2D classification of PEG treated filaments show mainly similar features, while including ~3 repeating units. Some variabilities can be seen, at what seems to be the surface of the filaments. (b) 2D class averages of native filaments show less features compared to 2D class averages of PEG treated filaments in fig 2a. (c) The result of a cross-section of the 3D map is similar to the reference-free 2D classification in Relion3.1 shown in (a).