Although visible light is replaced by shorter-wavelength, high-energy electron beams to achieve nanometer-level resolution, electron microscope was initially considered applicable only to inorganic materials instead of bio-tissues. The primary reason was that the sample would be destroyed by vacuum, dehydration, and bombardment of high-energy electrons; it is difficult to leave shadows on film because biological tissues are nearly transparent to electrons. Besides heavy metal negative staining, another solution is the cryo-electron microscope or Cryo-EM. The rapidly frozen proteins or tissues are imaged in 2D projections by a transmission electron microscope, then their 3D structures are reconstructed by algorithm and computer from these 2D projections.
History of Cryo-EM: high-pressure vitrified specimen
As early as 1934 Marton proposed that samples should be frozen to overcome various drawbacks of TEM. However, scientists found that this path seemed infeasible: once water freezes, it arranges into regular crystals to cut and compress bio samples. What you see in TEM is a severely damaged sample. Freezing also brought more severe issues: electron beams passing through ice crystals were strongly diffracted, leaving unintelligible diffraction patterns on the screen. The advantage of direct imaging was entirely disappeared. Later, scientists found that if liquid water was placed at extremely low temperatures, it would directly transform into vitreous ice whose molecules arranged in disordered patterns. Vitreous ice is more like an extremely viscous liquid without crystal structure. Therefore, damaged samples and diffraction are avoided.
The first practical cryo-electron microscope was made by Professor Dubochet and his team at the European Molecular Biology Laboratory. Initially, liquid nitrogen was expected to quickly freeze biological samples, but it was so cold that the Leidenfrost effect emerged. The sample was enveloped by air layer to block heat dissipation, and ice crystals still existed. Later, Scientists chose liquid ethane with a higher boiling point, which would contact with samples directly, and freeze them instantly. High-pressure freezing was also invented by Dubochet's team to increase vitrification depth. Vitrified biological samples were made into ultra-thin slices where living-state ultrastructures were preserved
It was surprising that the vitreous state almost solved all the drawbacks of traditional TEM. Cryo-electron microscope reduces radiation damage by 4 to 5 times, and enhances resolution and contrast. At the same time, cryo-electron microscope has an unparalleled advantage: it does not require crystal preparation. Not all substances can be crystallized, especially proteins embedded in biological membranes, but almost all substances can be frozen, whether proteins, viruses, or organelles.
3D Reconstruction and DDD in Cryo-EM
A 2D projection contains many randomly arranged proteins and a little of impurities, so Frank and Marin van Heel developed an algorithm to extract each molecule’s features and orientations hidden in projection. Based on this information, impurities were removed, and proteins were classified. Protein projections in the same cluster were similar enough to be superimposed and averaged to improve the signal-to-noise ratio. Frank also invented Random Conical Tilt to convert 2D projections into 3D images. Although scientists have developed other algorithms to identify proteins and enhance the signal-to-noise ratio, Cryo-EM resolution was only about 10 Å around 2010.
The electron direct detection device (DDD) sparked a resolution revolution in electron microscopes around 2010. DDD directly captures electrons to convert them into digital images (electrons → current → image). Compared to the traditional charge-coupled device, converting electrons into light signals is discarded by this new camera, so the noise introduced during signal conversion is minimized. Just like recording a movie, DDD can rapidly capture signals to reveal the dynamic structure of proteins during biochemical reactions, and correct the drift caused by electron irradiation and thermal motion. Current Cryo-EM resolution has reached nearly 1 Å, enough to rival X-ray diffraction.