key: cord-0927949-qnh6tlnm
authors: Rossmann, Michael G.; Morais, Marc C.; Leiman, Petr G.; Zhang, Wei
title: Combining X-Ray Crystallography and Electron Microscopy
date: 2005-03-31
journal: Structure
DOI: 10.1016/j.str.2005.01.005
sha: 3c68d0840199c90aaa522360400dd18262badbf8
doc_id: 927949
cord_uid: qnh6tlnm

The combination of cryo-electron microscopy to study large biological assemblies at low resolution with crystallography to determine near atomic structures of assembly fragments is quickly expanding the horizon of structural biology. This technique can be used to advantage in the study of large structures that cannot be crystallized, to follow dynamic processes, and to “purify” samples by visual selection of particles. Factors affecting the quality of cryo-electron microscopy maps and limits of accuracy in fitting known structural fragments are discussed.

Introduction or molecular assembly, the less likely it is that the mole-Three-dimensional structure determinations based on cule can be packed into a crystal, such that the surface cryo-electron microscopy (cryo-EM) have become a area of contact between molecules is large compared standard tool of structural biology in recent years. Just with the total area of the molecular assembly itself. An as in the practice of crystallography (Rodgers, 2001 which about 40 code for structural proteins, many of images with minimal distortion or artifacts. Provided which occur in multiple copies in the virion. Crystallizathere is a plentiful supply of near identical frozen partion can also be difficult or impossible when the structicles in random orientations, these projections can be ture of interest has components that are flexible, implycombined to form three-dimensional images. The resoing that each structure has specific regions on its lution of these images has been improving rapidly, surface that vary from molecule to molecule. This might largely because of improvements in reconstruction be caused by a hinge between domains (e.g., the fibers techniques (Tao and Zhang, 2000; van Heel et al., 2000) . of T4; Figure 2 ) or a surface carbohydrate moiety (Fig-In particular, efforts have centered on the accurate deure 3). termination of the contrast function that corrects the Another form of flexibility that can inhibit crystallizatwo-dimensional images for the experimental out-oftion is the presence of lipid membranes, as occurs in focus distance, the accurate determination of the relamany viruses. However, cryo-EM has made it possible tive orientation of the projected images, and the use of to visualize not only the virus structure as a whole, but a far greater number of particles. As a result, it is now also the membrane proteins in situ (Figure 4) , as opquite usual for cryo-EM image reconstructions to have posed to the usually artificial hydrophobic environan estimated resolution of 10 Å, and sometimes even ments used for crystallizing membrane proteins. Amino as good as 7 Å, with expectations of reconstructions acid sequence information and model building have going to 4 Å (Henderson, 2004; van Heel et al., 2000) produced a reasonable structure showing how the soon. These improvements have made it possible to transmembrane helices interact with their surrounding accurately fit atomic resolution crystal structures of lipid (Zhang et al., 2002) . molecular fragments into the lower resolution EM den-Crystallization also fails in the study of labile comsity to produce "pseudo" atomic structures of the complexes that would degrade or come apart in long crysplex (Figure 1 ). Although the fitting procedure is fretallization processes. An example is the complex of a quently done "by hand" using visual graphics programs, virus with its receptor and accessory receptors. . The packaging process was stopped by freezing about 2 minutes after dation could be slowed by cooling to 4°C, but certainly not for long enough to allow crystallization.

initiation. The original micrographs showed roughly two types of particles, those that appeared to be empty and Crystallization requires significant amount of sample to search for conditions that produce well-diffracting those that appeared to be partially filled. Separate image reconstructions showed not only the partial pres-crystals. For cryo-EM, it is only necessary to have enough sample to collect sufficient data to produce a ence of DNA in the fuller particles, but also a significantly different and larger structure around the unique reconstruction which might amount to 10 5 particles or so to attain resolutions of better than 10 Å, perhaps. In vertex containing the portal for DNA entry. It was concluded that the additional density is due to the ATPase comparison, a minimally sized crystal with dimensions of about 200 m in each direction would contain about (gene product 16) known to be essential for DNA packaging. The portal vertex density could be fitted with the 10 10 particles of 1,000 Å diameter. A further advantage of cryo-EM is that sample purity is not as critical as is crystal structure of the dodecameric "connector" (Simpson et al., 2000) , the central component of the required for crystallization, as images of the molecular assemblies being studied can be selected from the DNA-packaging machine, and the difference density of the structural prohead RNA (pRNA; Figure 7) .

Crystallization requires the presence of a large number of essentially identical particles. Although this is virus, are often mixtures of good and broken particles that would be impossible to crystallize. Or, in studying also required for single-particle reconstructions, the tomographic technique does permit the reconstruction of virus-receptor or virus-antibody complexes, it is often necessary to have excess ligand present to assure sat-three-dimensional images to low resolution (Baumeister and Steven, 2000; Grunewald et al., 2003). In this tech-uration of all sites on the virus. Another frequent occurrence is that there are two or more different modifica-nique, the EM grid is exposed to various tilt angles, allowing for the collection of a series of images pro-tions of the sample under study that are difficult to separate. jected in different directions for the particles on the grid. The limitations are, however, that the exposures A further example of the power of being able to select specific images on a micrograph is in the study of dy-have to be few and low dose to avoid excessive radiation damage, resulting in low-resolution reconstruc-namic processes such as stages in virus assembly including DNA packaging of proheads. This process was tions. Nevertheless, there is promise of three-dimensional analyses of whole cells and pleomorphic, membrane-used in the analysis of dsDNA packaging into the proheads of the small tailed f29 phage (Figure 6 ; Morais enveloped viruses, such as influenza or coronaviruses. 

The limit of resolution for which actual data are available on a particular micrograph or for a specific particle can be assessed by looking at the averaged Fourier transformed distribution ( larger the particle (corresponding to a smaller reciprocal cell), the greater will be the need for more particles . Another factor that impacts the transfer function has amplitude close to zero. Thus, if all images were taken at the same out-of-focus dis-quality of the reconstructed image is the out-of-focus distance used in recording the micrograph. This dis-tance, there would be shells of resolution where there would be few effective data. Hence, it is necessary to tance determines the resolution at which the contrast Given a good sample and the most perfect instrumentation conditions, such as lack of astigmatism, mechanical or magnetic vibrations, thermal motions of the specimen, and more, there are also other factors that determine the quality of the reconstruction. These include the accuracy with which the contrast function is determined, the accuracy with which the relative orientation and position of the particles are determined, the evaluation of the background that underlies every par- ature factor" correction. Not surprisingly, as resolution

When relatively few particle images are included in a reconstruction, only a low-resolution region of reciprocal space has moderately complete sampling (inside solid circle). Many more particle orientations are required for a fuller sampling at higher resolution (inside dashed circle).

Various types of models can be used for establishing the structure associated with a cryo-EM reconstruction. At lower resolution (worse than about 12 Å), it is necessary to interpret the density in terms of the structures of whole proteins or fairly large components of the mo- The known crystal structure of E1 was fitted into the cryo-EM density, assuming 3-fold symmetry, aided by the glycosylation sites at residues Asn139 and Asn245. The carbohydrate positions (crosses) had been determined from difference maps between wildtype and deglycosylated virus. After the E1 molecules had been fitted, the density at all grid points in the map that were within 4 Å of each atom in E1 was set to zero, leaving the density of E2. Hence, the three E2 molecules (cyan, brown, and blue) were shown to be long and thin ( A variety of criteria can be used when fitting rigid symmetry-related molecules or between different molecules, the number of atoms that are outside the bound-molecular structures into cryo-EM density as implemented in the EMfit program . ary of the available density, and the chemical sense of the interaction between fitted fragments. Other types These need to be suitably weighted to produce a combined overall criterion of best fit. In order to place the of information can also be considered, such as the dis- tance between known positions in the density map tion cryo-EM results will yield better information at pseudoatomic resolution, while lower resolution cryo-(e.g., carbohydrate sites found in difference maps; Zhang et al., 2002) and the corresponding residue posi-EM results of larger complexes have become an essential tool of cell biology. tion in the fitted structure, or the distance between the carboxy-terminal C α atom of one domain and the amino-terminal C α atom of the following, independently 

Three-dimensional structure of bacteriophage T4 baseplate. cryo-electron microscopy of dengue virus

Conformational Lenches

Structure of the dengue virus: implications for flavivirus organization, maturation, and fusion

Automatic particle selection: results of a comparative study

Structure and morphogenesis of bacteriophage T4

Three-dimensional rearrangement of proteins in the tail of bacteriophage T4 on infection of its host

A ligand-binding pocket in the dengue virus envelope glycoprotein

Cryoelectronmicroscopy image reconstruction of symmetry mismatches in bacteriophage f29

Review: automatic particle detection in electron microscopy

The envelope glycoprotein from tick-borne encephalitis virus at 2 Å resolution

Cryocrystallography techniques and devices

Docking structures of domains into maps from cryo-electron microscopy using local correlation

Combining electron microscopic with X-ray crystallographic structures

Structure of the bacteriophage f29 DNA packaging motor

Exploring global distortions of biological macromolecules and assemblies from lowresolution structural information and elastic network theory

Recent developments in cryoelectron microscopy reconstruction of single particles

Assembly of a tailed bacterial virus and its genome release studied in three dimensions

Singleparticle electron cryo-microscopy: towards atomic resolution

Docking of atomic models into reconstructions from electron microscopy

Modeling tricks and fitting techniques for multiresolution structures

Situs: a package for docking crystal structures into low-resolution maps from electron microscopy

Placement of the structural proteins in Sindbis virus