Overview. Three-dimensional cryo-electron microscopy holds great potential for the investigation of macromolecular complexes, including viruses, the cytoskeleton, and the nanomachines that carry out many vital cellular functions. Such complexes often contain protein subunits of different kinds, with cumulative molecular weights running well into the megadalton range. To rationalize their functional assignments in structural terms, the first order of business is to map the locations of the component subunits of a given complex. A more detailed objective then becomes to understand the organization of their domains and the overall layout of the polypeptide chains. With current methodology, resolutions below 20Å are often achievable in cryo-EM, extending below 10 Å in favorable cases. Although resolutions of this order yield valuable morphological information, they do not usually suffice to identify individual subunits within a complex. Generalized difference imaging offers a solution to this problem of molecular mapping.

Coliphage T4 is a large, double-stranded DNA virus (family Myoviridae) which infects Escherichia coli. It is the classic example of a contractile-tailed phage and a wealth of information on viral morphogenesis processes have been obtained on it. Extensive studies of T4 by negative-stain and metal-shadow electron microscopy have shown that mature virions, which contain over 40 different polypeptides, have a prolate capsid that encapsidates the genome (∼175 × 106Da), and several other structural components, including the collar with whiskers, neck, a contractile sheath, and a baseplate with fibers. A complete understanding of processes such as assembly of the viral capsids will require atomic-scale knowledge of the various protein interactions. Details of the process can be extrapolated from knowledge of the initial, transitional, and final states of a structure.

We have initiated an examination of T4 head morphogenesis by studying mutant, isometric capsids ('isometrics’) using cryo-electron microscopy and three-dimensional (3D) image reconstruction methods as applied to particles with icosahedral symmetry.

Several structures of single particles have been resolved to 7-9 Å using electron cryomicroscopy. This field is now poised to make the next step, to near atomic (3-4 Å) resolution. Signal to noise ratio is the critical factor determining the amount of data required for a structural determination at higher resolution. When properly weighted, signal to noise ratio is additive as image data is averaged coherently, however, this relies on perfect alignment of the noisy images. Typical refinement algorithms require as high a signal to noise ratio as possible for accurate alignment. We report a method, which allows accurate, routine determination of signal to noise ratio and other image parameters, both for data quality assessment and microscope quality tracking.

Figure la shows a typical 400 kV electron image of ice-embedded herpesvirus capsids. The incoherent sum of the Fourier transforms of particle images (Fig. lb) shows that the data contains information out to a resolution of 7 Å. When rotationally averaged, this power spectrum can be expressed as:

The GroEL/GroES chaperonin system of E. coli facilitates nucleotide dependent folding of select proteins. The structure of GroEL has been described by three-dimensional electron microscopy and at higher resolution by X-ray crystallography. The GroEL oligomer is a cylindrical tetradecamer composed of two heptameric rings of 57 kDa protein subunits, stacked back to back. Each subunit is comprised of two large domains, equatorial and apical, connected by a smaller intermediate domain. The intermediate “hinge” domain links the apical and equatorial domain and provides flexibility for en bloc rearrangement associated with nucleotide and GroES binding. Equatorial domains are responsible for interactions between the two heptamers and contain the ATPase activity of GroEL. Each ring of GroEL has a central cavity that is the binding site for denatured protein substrate. GroES and denatured substrate binding sites are located on the apical domains facing the central cavity.

ClpAP, an ATP-dependent protease of E. coli, recognizes and unfolds protein substrates via ClpA, its chaperonelike ATPase component, and digests them in ClpP, its protease component . ClpA forms hexameric rings with a two-layered structure, and stacks axially on either face of the double heptameric rings of ClpP. Protein substrates can bind to ClpAP in the presence of ATPγS, which is not hydrolyzed by ClpA, but are not degraded unless ATP is added. This property makes it possible to synchronize degradation in vitro by forming enzymesubstrate complexes in the presence of ATPγS and then adding ATP to trigger subsequent steps. We have used image averaging of electron micrographs of frozen hydrated and negatively stained specimens to characterize interactions of ClpA and ClpAP complexes with the model substrate, bacteriophage P1 protein, RepA.

Three-dimensional cryo-electron microscopy (EM) of macromolecular assemblies with low or nonexistent symmetry in single-particle form is now recognized as a powerful, legitimate method of structure research. The progress in the resolution achieved has always been related to the introduction of new image processing methods and to improvements in understanding of the overall strategy of single particle analysis. The new techniques either were adopted from methods developed in other fields, or were specifically designed to deal with extremely low Signal-to-Noise Ratio (SNR) data in EM. In the former category we can place multivariate statistical analysis, clustering techniques, correlation techniques, and tomographic methods, in the latter alignment methods, three-dimensional (3D) projection alignment, contrast transfer function (CTF) correction methods, and common-lines based orientational search. In addition, there have been steady improvements of the quality of the data by seeking better electron microscopy conditions (usage of higher microscope voltage and of microscopes equipped with a field emission gun).

Co-translational translocation at the endoplasmic reticulum (ER) plays a critical role in the targeting of both soluble and membrane proteins to their correct intra- and intercellular compartments. We are studying the 3D architecture of the ribosome-Sec61p complex (translocon), with the aim of understanding the physical mechanisms of gating and transport. To this end, we are using single particle electron cryo-microscopy and 3D reconstruction of frozen hydrated channel complexes, to obtain interpretable and biologically relevant maps.

Previously, we have shown that both co- and post-translational translocation utilize a common central channel comprised of a ring-like Sec61p oligomer. Moreover, this channel morphology is shared with the related Sec YE complex from B. subtilus. Mass analysis, volume calculations and ribosome binding experiments suggest a stoichiometry of 3-4 Sec61p heterotrimers per ring. We currently favor 4 copies of the Sec61p complex per channel, as projection maps demonstrate 4 nearly equi-spaced peaks around the central pore.

The mechanism of encapsidation of DNA into a bacteriophage head is a most intriguing problem. The portal protein is essential for the assembly of tailed double-stranded DNA (dsDNA) bacteriophages. These turbine-like homo-oligomers consist of 12 or 13 subunits surrounding a central channel. Portal oligomers are located at the vertex of the icosahedral head that binds to the phage tail. It was shown that gp6 portal protein from bacteriophage SPP1 has 13 subunits, prior to their incorporation into the viral procapsid structure. After packaging of the viral DNA inside the phage capsid, additional proteins have been found attached to the portal oligomer: gpl5 and gp 16. A complex of gp6/gpl5/gpl7 forms the connector structure that provides the interface for attachment of the phage tail. We here present the three-dimensional (3D) reconstructions of the gp6 wild-type protein alone and of the gp6/gpl5/gpl6 complex - both at 10Å resolution - based on electron cryo-microscopy using the angular reconstitution single particle methodology.

The vault complex is a ubiquitous and predominantly cytoplasmic 13 MDa ribonucleoprotein assembly, composed of three proteins (TEP1, 240 kDa; VPARP, 193 kDa; and MVP, 100 kDa) and an untranslated RNA (vRNA) Although the basic cellular function of the vault is still unclear, recently it has been shown that induction of the major vault protein (MVP) has a direct negative influence on the nuclear uptake of the anti-cancer drug, doxorubicin. We have been applying cryo-EM single particle reconstruction methods to study the structure and molecular architecture of this cellular component. A published reconstruction of the intact rat vault at 31 Å resolution revealed that the complex is hollow and is structurally well suited to serve in macromolecular transport or sequestration. Higher resolution, 23 Å, was achieved for a reconstruction of the RNase-treated rat vault and difference imaging with the intact rat vault localized the vRNA to the ends of the vault caps.

Fatty acid synthase (FAS) is the enzyme responsible for de novo synthesis of fatty acids from acetyl-CoA, malonyl-CoA and NADPH. FAS (550 kDa) is a homodimer of two multifunctional polypeptides, each with seven distinct catalytic activities and a site for the prosthetic group, 4’- phosphopantetheine, acyl carrier protein (ACP). These domains are organized from the N- to the C-terminus as follows: keto acylsynthase, acetyl/malonyl transacetylase, dehydratase, the interdomain, enoyl reductase, ketoreductase, ACP and thioesterase. The two polypeptides are held together through the interdomain and oriented in an anti-parallel manner, each contributing complementary half sites and giving rise to two independently active centers for palmitate synthesis. Interest in FAS arises from its involvement in human disorders, such as obesity, hyperlipidemia and carcinogenesis.

Human FAS purified from a breast cancer cell line, ZR75-1, was vitrified at 50-70 μg/ml on holey grids in the presence of NADPH and acetyl-CoA and kept at -166°C in a Gatan 626 cryo-holder.

Strand separation in double stranded DNA is achieved in vivo by a class of enzymes called helicases in a process fuelled by hydrolysis of nucleoside triphosphates. DnaB is the major replicative helicase in E.coli. For chromosomal replication to initiate, DnaB needs to interact with a partner protein, namely DnaC, which after properly loading DnaB onto the DNA template at the origin of replication is subsequently released from the complex. DnaB turns to be functionally active as a helicase only after DnaC has been released from the complex. The native DnaB is a homohexamer of molecular weight 318 kD. In the presence of ATP and Mg2+, the hexameric DnaB has been shown to form a complex with six molecules of DnaC (total molecular weight of the complex: 480 kD).

The reconstructed 3D volume of the DnaB hexamer obtained from frozen-hydrated specimens showed the DnaB oligomer as a particle possessing three-fold rather than six-fold symmetry, despite DnaB being made up by six identical subunits.

For over 25 years a major effort in electron microscopy of macromolecules has been the determination of the three dimensional structure from the two-dimensional electron micrographs of such specimens. Great success has been realized when the macromolecule or complex takes the form of an array such as a 2D crystal (1,2), a helical structure (3,4) or one with icosahedral symmetry (5,6). However, for molecules which do not form such arrays, and in the limit only exist as single particles, a number of challenges have had to be addressed. No easy averaging of noisy low dose images is possible due to the lack of lateral and rotational symmetry. Random unknown orientations of the particles have to be determined, a process exacerbated by noise if low dose images are used as input. Alternatively, higher dose images result in radiation-induced structural alterations of the macromolecule.

We have reconstructed the three-dimensional quaternary structure of the complete 480 kDa insulin receptor (IR), complexed with Nanogold - labeled insulin, and the 290 kDa Xanthine Dehydrogenase (XDH). Both molecules were imaged by low-dose, low-temperature dark field scanning transmission electron microscopy (STEM). XDH and IR were both freeze-plunged in liquid ethane, and transferred to the STEM (VG HB 601) where they were freeze dried at -130°C. Reconstruction was carried out using the method of quaternion-assisted angular reconstruction (IQAD) as described previously. XDH was further refined by an iterative process in which the IQAD produced volume was used as a reference for further refinements. Separate reconstructions of two sets of half the images indicated an inter-reconstruction resolution of 20 Å and 9 Å by phase residual criteria for the IR and XDH reconstructions, respectively.

Training of students, postdocs and scientists in the techniques and instrumentation of Cryo-electron microscopy has many challenges. This is especially the case at the Wadsworth center where there are a variety of techniques as well as instruments to choose from, each of them having certain nuances that need to be recognized before they can be fully taken advantage of. There are also users with many levels of experience that need to be trained to use the equipment and techniques effectively. For this reason we have chosen to incorporate many of the protocols which are used into an internal web page. The web browser is a very effective tool for conveying detailed information on a variety of levels.

A novice user is asked to go through the EM training web page (Figure 1) and then select the Philips EM420 use link to get a feel for the steps that will be involved to set up and operate the microscope.

Macromolecular microscopy is becoming an increasingly important tool for structural biology. The development of improved capabilities for three-dimensional electron microscopy is critical for optimal progress in emerging integrative research in molecular cell biology. These techniques currently suffer from several severe disadvantages related to the tremendous time and effort required to acquire and analyze the data.

For several years we have been developing software for automated and intelligent acquisition of transmission electron micrographs [1,2,3]. Our overall goal is to develop a system for rapid and routine structure determination of macromolecular assemblies from specimens preserved in vitreous ice. Ultimately we plan to develop an integrated system that can produce an electron density map of a structure within a few hours of inserting a specimen into the electron microscope. With this goal in mind it is essential that the images be acquired using a digital camera rather than film.

Sophisticated tools are needed to examine the results of cyro-microscopy. As the size and resolution of three dimensional macromolecular structures steadily improve, and the speed at with which they can be generated increases, researchers are finding they are inundated with larger datasets and at the same time are compelled to expediently evaluate these structures in unforeseen ways. Integration of EM data with other types of information is becoming necessary and routine; for example X-ray data, 3D EM reconstructions, and theoretical models, must be evaluated in concert to discount or propose hypothesis. To create such tools, the developer must take into account not only the empirical and theoretical possibilities, but also they must master the human factors and computational limits. During the last five years, the National Center for Macromolecular Imaging (NCMI) has progressed from a remedial 3D visualization capability to a collection of visualization tools allowing researchers to focus on the discovery phase of biological research.

PRD1 is a ds-DNA bacteriophage from the Tectiviridae family with an unusual structural feature: the viral genome is enclosed by a protein-rich membrane, which is in turn enclosed by an external icosahedral protein shell (capsid). Three-dimensional reconstructions from cryo-electron microscopy (cryo-EM) images have revealed the structure of the PRD1 capsid at moderate resolution (28 Å), while X-ray crystallographic studies have recently provided a high resolution (1.85 Å) picture of the major coat protein, P3. We have now combined these results from different imaging methods to obtain a more detailed understanding of the virion organization. The combination has been made in a cyclic process: a preliminary fitting of the atomic structure of P3 to each one of its independent positions in the cryo-EM maps of the capsids provided initial models that could be used to improve the reconstructions; the refined maps then served as a base frame for an optimized fit. This process allows us to study the viral particle structure at “quasi-atomic” resolution.

Human adenovirus (Ad) provides unique model system for studying the multistage events in viral entry. We are applying cryo-electron microscopy and single particle reconstruction methods to study the cell entry pathway of Ad as well as to examine the structure of reengineered Ad vectors. Ad cell entry is mediated by interactions with two different cellular receptors: the fiber receptor, CAR for most Ad serotypes or a 50 kDa receptor for a subset of serotypes including Ad37, for attachment; and the αvβ3 and αvβ5 integrins for internalization. We have previously reconstructed both Ad2 and Adl2 virus particles complexed with a soluble form of αvβ5 integrin. We learned during our comparative cryo-EM study of the Ad2/αvβ5 and Adl2/αvβ5 complexes that the length of the variable RGD loop in the penton base affects the relative height of the integrin density over the penton base.

For several years we have been developing a software application, called Leginon, for the automated control and acquisition of images from a transmission electron microscope. The system has been developed around a Philips CM200 and a Gatan MSC CCD camera. One of the primary motivations in developing this software is to provide for a system that can acquire many hundreds of images under low dose conditions from a specimen embedded in vitreous ice. In order to set up and manage this system we have developed a number of simple interactive graphical tools that enable the user to design, oversee and manage protocols for controlling the microscope and collecting the images. Many of these simple tools have also proved generally useful as stand alone applications.

For several years we have been developing a system for the automated control and acquisition of images from a transmission electron microscope [1,2,3]. The system has been developed around a Philips CM200 equipped with a Gatan cryostage and a Gatan MSC CCD camera. One of the primary motivations in developing this software is to provide for a system that can acquire many hundreds of images over several days of completely unattended operation.

The current implementation of our automated system allows data to be automatically collected from the TEM for more than 36 hours. There is no intervention required from the operator except for one critical exception, refilling the Gatan cryostage. The cryostage, which maintains the temperature of the vitreous ice specimens, must be refilled with liquid nitrogen (LN2) approximately every two hours.