ET is a variant of EM in which samples are rotated and imaged from various angles, yielding multiple projection images that can be reconstructed into three-dimensional pictures of cells, organelles and other structures. Instead of removing any liquid water, samples are rapidly frozen such that they become embedded in ‘vitreous’ ice. In this medium, as with glass, water molecules are arranged randomly rather than in the crystalline form that can damage biological samples.
The resulting technique of cryo-ET thus avoids the artefacts associated with conventional EM, and delivers highly detailed and spatially accurate reconstructions of the intact cellular environment at nanometer resolution. However, the absence of staining agents means that scientists must contend with reduced contrast, making it more difficult to distinguish structures of interest against the background of the cellular environment. This is exacerbated by the fact that ice-embedded samples are vulnerable to damage from prolonged exposure to the electron beam, meaning that researchers must minimize exposure, which results in a low signal-to-noise ratio.
In order to maximize the quality of the information that can be obtained from these tomograms without compromising sample integrity, scientists have paired clever strategies for specimen preparation with sophisticated computer algorithms for data processing and analysis. This has allowed scientists to sucessfully employ cryo-ET to visualize a number of complicated cellular systems including the following: the nuclear-pore complex, which is the ‘gatekeeper’ to the nucleus; the presynaptic terminals, from which neurons issue stimulatory and inhibitory signals to their neighbours; and strings of polyribosomes, which are the assembly lines that manage the production of protein from RNA (Fig. 1). For smaller cells, such as bacteria, it even becomes possible to characterize the distribution of populations of large multi-protein complexes within the full cellular volume, and early efforts at performing organism-wide visual proteomics have been described for the human pathogens Leptospira interrogans and Mycobacterium pneumoniae9.10 .