Cryo-S(T)EM instrumentation
Our developed cryo-S(T)EM instrument including of a newly developed cryo-transfer holder and an anti-contamination trap was fully functional as a standard cryo-electron microscope (Fig. S1). Both the cryo-transfer holder and the anti-contamination trap have a double pipe structure containing a pipe for venting gas, so each can be easily filled to the tip with liquid nitrogen. The specimen stage of the cryo-transfer holder could be cooled to -190°C by a liquid nitrogen slush obtained by evaporation with a vacuum pump (Fig. S2). Similarly, the anti-contamination trap could be cooled to -210°C by using a liquid nitrogen slush (Fig. S3). However, the temperature of the specimen stage used for observing samples was substantially dependent on the temperature of the anti-contamination trap. If the temperature of the anti-contamination trap was the same as or higher than the temperature of the specimen stage, the water molecules sublimated from the ice around the sample were not efficiently adsorbed onto the anti-contamination trap and were refrozen on the sample surface as frost, causing contamination. To prevent this issue, the temperature of the specimen stage had to be set to approximately 20 ℃ above the temperature of the anti-contamination trap. Thus, the sample had to be observed above − 180°C. The water adsorption capacity of the anti-contamination trap was very important, especially for observing cells embedded in large amounts of ice, which differs from the requirements for observing purified protein molecules. Therefore, the tip shape of the anti-contamination trap was designed to enclose the specimen stage of the cryo-transfer holder, as shown in Fig. 1. On the other hand, the specimen stage of the cryo-transfer holder slides into the holder to prevent frost formation after setting samples in a cryo-workstation (Fig. S4) and is pulled out during observation (Fig. 2). The cryo-S(T)EM instrument, which was developed based on a SEM instrument (Hitachi SU9000) with detectors for secondary electrons, transmission electrons and backscattered electrons, should, in principle, be able to capture three images simultaneously. However, since the irradiation dose was limited to ~ 30 electrons/Å2 to protect the frozen cells from irradiation damage, backscattered electrons from the ice surface were too weak to generate an image. On the other hand, a secondary electron image (SEM image) was observed with little charge-up due to the electron conductivity of ice. The transmission images obtained by cryo-S(T)EM were formed by scanning an extremely thin electron beam emitted from a cold FEG (diameter of beam: 0.4 nm, spherical aberration coefficient: 2 mm, optimum opening angle: 11 mrad) 24. The quality of the scanning transmission image (STEM image) was evaluated by image processing of the micrographs of negatively stained actin filaments. The raw STEM image (Fig. 3A) clearly showed fine details with high contrast. The molecular model reconstructed by image processing also showed higher resolution than the model reconstructed from images obtained by conventional TEM (compare B with C in Fig. 3). These results showed that the scanning transmission electron microscopy (STEM) mechanism within the above-described cryo-S(T)EM method (or Hitachi SU9000 which is the basic machine for development) had the ability to display STEM images of biological samples equal to or better than conventional TEM images. Unfortunately, however, at -180°C, the resolution obtained at room temperature was not achieved due to drifting of the specimen stage. Therefore, high-resolution analysis will require the fabrication of a new drift-free cryo-transfer holder.
Application of cryo-S(T)EM to the analysis of cell structures
Since quick freezing can be used to preserve not only purified proteins but also the fine structure of cells in a nearly native state, cells also need to be observed in the frozen state to understand their real fine structure. Of course, the slight drift of the specimen stage mentioned above makes high-resolution cell observation difficult, but cryo-S(T)EM combined with the unroofing method can yield valuable results for thick cell structures even at low resolution. Cultured cells are extremely thick, reaching a few microns, and are frozen together with a large amount of water compared to purified proteins. Therefore, it is necessary to reduce the thickness of frozen samples enough to allow electrons to pass through. In previous studies, frozen samples were thinned by using a cryo-ultramicrotome (CEMOVIS) 7 or focused ion beam (FIB) 19–22. In this study, unroofing 23, 25, 26 was used instead. Initially, unroofing is a unique method for exposing the cytoplasmic surface of the ventral cell membrane to observe clathrin coats, caveolae, etc., in freeze-etching replica EM 27,28. In practice, however, all cells are not unroofed in the same way. Unroofed cells vary in thickness and cytoplasmic contents, ranging from fully unroofed cells to non-unroofed cells. The observations of various unroofed cells by cryo-S(T)EM are described below, and the cells are classified into fully unroofed cells, partially unroofed cells and small (micro-) unroofed cells (see Fig. S5 for better understanding). In many cases, frozen cells were embedded in enough ice to prevent the transmission of electrons. To reduce the thickness of the ice layer, the temperature of the specimen stage was first increased to -100°C to accelerate ice sublimation, and thereafter, the temperature was decreased to -180°C again for observation. This ability to easily adjust the ice thickness by raising or lowering the temperature of the specimen stage is another advantage of this cryo-S(T)EM technique that is very useful for observing various unroofed cells. In fully unroofed cells (cells with little cytoplasm), the cytoplasmic surface (inner surface) of the ventral cell membrane was exposed thoroughly, and clathrin coats, cortical actin filaments, and microtubules were found in close contact with the surface (Fig. 4). Actin filaments and microtubules grew without fragmentation or branching while bending gently. Partially unroofed cells reached a few microns in thickness due to residual organelles and cytoplasm, although most of the cell membrane was removed. In general, it is difficult for electrons to pass through such a thick sample to generate an image. However, surprisingly, sufficient (deep) sublimation of ice exposed intracellular fine structures, regardless of the degree of unroofing. The reason the structures became visible after ice sublimation even though the cell thickness did not change will be explained below (DISCUSSION section). Regardless, the intracellular structures became clearer with increasing duration of ice sublimation. Figure 5 shows partially unroofed cells whose cytoplasm and water were frozen to an appropriate extent. Actin filaments, ribosomes, endoplasmic reticulum (ER), etc., appeared after slight ice sublimation. These cells are completely embedded in ice, and thus, the SEM image shows a flat surface. Cells unroofed to this extent provide the most information about structures. Figure 5B shows a high-magnification micrograph of another part of the same cell shown in Fig. 5A. The ER, ribosomal chain, microtubules, and actin filaments clearly overlap with each other. On the other hand, in the case of another partially unroofed cell with considerable thickness (Fig. 6), ice sublimation for 20–30 minutes at -100°C was required for TEM images to be obtained. Figure 6 shows the nucleus, mitochondria, ER and filaments. Sufficient (deep) sublimation also revealed organelles embedded in ice in an SEM image (Fig. 6 right image). Thus, simultaneous measurement of SEM images was useful for investigating the freeze-drying state. Notably, in this study, the intracellular fine structure of micro-unroofed cells, which is similar to the whole cell, appeared after deep sublimation of ice even during 30 kV cryo-S(T)EM. In fact, as shown in Fig. 7, since the SEM image was flat even after sufficient sublimation, it was considered that most parts of these cells are covered with a cell membrane. Nevertheless, the mitochondria, ER, and cytoskeleton became visible upon deep sublimation. Many flat ERs were located beneath the apical cell membrane (Fig. 7) and seemed to be interconnected with each other to form a network. These ER networks were also well preserved in mitotic telophase cells, as shown in Fig. 8. These structures had never before been observed in thin sections of cells. The ability to observe thick cells in combination with deep sublimation of ice is one of the advantages of this developed cryo-S(T)EM technique.