These electron micrographs can be treated as two-dimensional projections (Fourier transforms) of the particles. Three-dimensional structures can be reconstructed from such two-dimensional projections by mathematically combining the information included in different views of the particles. For the purpose of reconstruction, the images of different particles are treated as different views of the same structure.
For reconstruction, micrographs are digitized for computer processing. Each particle to be analyzed is then centered inside a box, and its orientation is determined by application of programs that orient the particle on the basis of its icosahedral symmetry. In cryo-electron tomography, images are collected with the sample at different angles to the electron beam and combined computationally to reconstruct a three-dimensional structure. The advantage of this approach is that no assumptions about the symmetry of the structure are required. The parameters that define the orientation of the particle must be determined with a high degree of accuracy, for example, to within 1° for even a low-resolution reconstruction (~40 Å).
Once the orientations of a number of particles sufficient to represent all parts of the asymmetric unit have been determined, a low-resolution three-dimensional reconstruction is calculated from the initial set of two-dimensional projections by using computational methods.
This reconstruction is refined by including data from additional views (particles). The number of views required depends on the size of the particle and the resolution sought. The reconstruction is initially interpreted in terms of the external features of the virus particle. Various computational and computer graphics procedures have been developed to facilitate interpretation of internal features. Courtesy of B.V.V. Prasad, Baylor College of Medicine.
And is it not true that even the small step of a glimpse through the microscope reveals to us images that we should deem fantastic and over-imaginative if we were to see them somewhere accidentally, and lacked the sense to understand them.
Paul Klee, On Modern Art, translated by Paul Findlay (London, United Kingdom, 1948)
Figure 4.4 Determination of virus structure by X-ray diffraction. This method requires crystallization of the sample of interest, such as a virus particle. The conditions under which a supersaturated solution of a protein or virus particles will form crystals suitable for X-ray diffraction, while retaining its native structure, cannot be predicted. Consequently, highly concentrated and purified virus particles are incubated under many different conditions, varying such parameters as concentrations of salt or metal ions, the presence of other polymers, and temperature. This empirical approach has been facilitated by the development of crystallization screening kits and robotic devices to set up crystallization trials. Nevertheless, it can be time-consuming, and success is not guaranteed. A virus crystal is composed of virus particles arranged in a well-ordered three-dimensional lattice. When the crystal is bombarded with a monochromatic X-ray beam traveling through the pinhole, each atom within the virus particle scatters the radiation. Interactions of the scattered rays with one another form a diffraction pattern that is recorded. Each spot contains information about the position and the identity of the atoms in the crystal. The locations and intensities of the spots are stored electronically. Determination of the three-dimensional structure of the virus from the diffraction pattern requires information that is lost in the X-ray diffraction experiment needed for calculating the positions of the atoms. The diffraction pattern collected from the crystal is now most usually interpreted by using the phases from the structure of a related molecule as a starting point, and subsequently applying computer algorithms to calculate the actual values of the phases. This method is known as molecular replacement. Once the phases are known, the intensities and spot positions from the diffraction pattern are used to calculate the locations of the atoms within the crystal.
Not all viruses can be examined directly by X-ray crystallography: some do not form suitable crystals, and the larger viruses lie beyond the power of the current procedures by which X-ray diffraction spots are converted into a structural model. However, their architectures can be determined by using a combination of structural methods. Individual viral proteins can be examined by X-ray crystallography and by multidimensional nuclear magnetic resonance techniques. The latter methods, which allow structural models to be constructed from knowledge of the distances between specific atoms in a polypeptide chain, can be applied to proteins in solution, a significant advantage.
High-resolution structures of individual proteins have been particularly important in deciphering mechanisms of attachment and entry of enveloped viruses. Even more valuable are methods in which high-resolution structures of individual viral proteins are combined with cryo-EM reconstructions of intact virus particles. For example, in difference imaging, the structures of individual proteins are in essence subtracted from the reconstruction of the particle to yield new structural insights (Fig. 4.5). This powerful approach has provided fascinating views of interactions of viral envelope proteins embedded in lipid bilayers and of internal surfaces and components of virus particles.
Atomic-resolution structures of individual proteins or domains can also be modeled into lower-resolution views (currently ~15 Å) obtained by small-angle X-ray scattering. This technique, which is applied to proteins in solution, provides information about the overall size and shape of flexible, asymmetric proteins, and has provided valuable information about viral proteins with multiple functional domains (see Chapter 10). It can also reveal dynamic properties, such as conformational change, a property shared with serial femto-second X-ray crystallography, in which as many as hundreds of thousands of images of small crystals are recorded in a very short time.
Building a Protective Coat
Regardless of their size and architectural sophistication, all virions contain at least one protein coat, the capsid or nucleocapsid, which encases and protects the nucleic acid genome (Table 4.2). As first pointed out by Francis Crick and James Watson in 1956, most virus particles appear to be rod shaped or spherical under the electron microscope. Because the coding capacities of viral genomes known at that time were very limited, these authors proposed that construction of capsids from a small number of subunits would minimize the genetic cost of encoding structural proteins. Such genetic economy dictates that capsids be built from identical copies of a small number of viral proteins with properties that permit regular and repetitive interactions among them. These protein molecules are arranged to provide maximal contact and noncovalent bonding among subunits and structural units. We now know that the capsids of even the largest viruses, with genomes of >1 Mbp, are also built from a small number of proteins. This property indicates that optimization of regular protein-protein interactions is the primary determinant of virus architecture. The repetition of such interactions among a limited number of proteins results in a regular structure, with symmetry that is determined by the spatial patterns of the interactions. The helical or icosahedral symmetry common to many viruses not only satisfies such protein limitations but also has considerable practical value (Box 4.2).
Figure 4.5 Difference mapping illustrated by a 6-Å-resolution reconstruction of adenovirus. (A) Comparison of α-helices of the penton base in the cryo-electron microscopic (cryo-EM) density (gray mesh) and the crystal structure of this protein bound to a fiber peptide (ribbon). The excellent agreement established that α-helices could be reliably discerned in the 6-Å cryo-EM reconstruction.