Dihydrohpoamide acetyl transferase (E2), a catalytic and structural component of a multienzyme complex that catalyzes the oxidative decarboxylation of pyruvate, forms the central core to which the other components are bound. We have utilized protein engineering and 3-D electron microscopy to study the structural organization of the largest multienzyme complex known (Mr ∼ 107). The structures of the truncated 60-mer core (tE2) and complexes of the tE2 associated with a binding protein (BP), and the BP associated with its dihydrohpoamide dehydrogenase (BP'E3) and the intact E2 associated with BP and the pyruvate dehydrogenase (E1) were determined (Figs. 1 and 2). The tE2 core is a pentagonal dodecahedron consisting of 20 cone-shaped trimers interconnected by 30 bridges.

Previous studies have given rise to the generally accepted belief that BP and BP'E3 components are bound on the outside of the E2 scaffold and that E1 is similarly bound to the core in variable positions by flexible tethers.

The nuclear envelope (NE) physically separates the genetic machinery residing in the nucleus from protein synthesis taking place in the cytoplasm. Bidirectional transport of proteins, RNAs and RNP particles between these two compartments is mediated by the nuclear pore complexes (NPCs), large (-120 MDa) supramolecular assemblies embedded in the NE and being built of -100 different polypeptides, called nucleoporins. Extensive structural studies have revealed the 3D architecture of NPCs, and epitopes of several nucleoporins have been localized within their 3D structure.

In an attempt to go beyond the current consensus model of the NPC (Fig. lc, inset), we have embarked on a more systematic structural analysis of the molecular architecture of native NPCs. This is achieved by field-emission (FETEM; Fig. la) or energy-filtering (EFTEM; Fig. lb) TEM of completely unfixed and unstained Xenopus oocyte NEs embedded in thick (i.e. -200 nm) amorphous ice so that the 3D organization of the cytoplasmic and nuclear periphery of the NPCs (i.e. the cytoplasmic fibrils and nuclear baskets) is fully preserved (Fig. lb).

Recent initiatives have resulted in a considerable increase in our understanding of the structure of the nuclear pore complex (NPC). The biochemical factors involved in both import and export have been rapidly characterised, with steady progress in the molecular dissection of the structural elements of the NPC, which is a unit of considerable molecular architecture (MW 125 kD), comprising an estimated 50- 100 different proteins. Despite this progress, the crucial molecular interactions involved in the mechanics of transport through the central transporter of the NPC remain unclear. NPC structure in Diptera, fish, (Fig 1) amphibians, birds and mammals shows a high degree of evolutionary conservation. 3D reconstructions of isolated yeast NPCs, show that the core structure is very similar to ‘higher’ organisms, but peripheral structures may be considerably reduced in structural complexity (1).

Individual NPC components have been accessed in FEISEM by a variety of methods, including proteolysis,

Protein translocation plays a critical role in the targeting of both soluble and membrane proteins to their correct intra- and inter-cellular compartments. We are studying the 3D architecture of two rather different translocation machines, the Nuclear Pore Complex (NPC) and the ribosome-Sec61p complex (translocon), with the aim of understanding their physical mechanisms of gating and transport. Towards this end, we are using single particle electron cryomicroscopy and 3D reconstruction of frozen hydrated specimens to obtain interpretable maps that are biologically relevant.

Previous work suggested that a central channel complex (termed the transporter) is present in vertebrate NPCs. Based on classification studies of the transporter and STEM images of Chironomus NPCs caught translocating large mRNPs, we have hypothesized that the transport mechanism utilizes a double iris-like gating mechanism, in which oppositely facing gates located at either end of the transporter open asynchronously. Recently, we have extended our studies to yeast NPCs and shown that this organelle is markedly smaller than its vertebrate cousin.

To date, cryo-electron microscopy has become the most successful technique for exploring the structure of the ribosome and for studying binding positions of its various ligands, with the resolution slowly extending toward 10 Å. Obstacles in the attempts to improve resolution are the limited stability and coherence of the electron microscope, the statistics of data collection, and the conformational heterogeneity of the specimen. The last factor in this list proved to be the reason why it has been difficult to go past 18-20 Å with many specimens despite the use of state-of-the-art electron microscopes and inclusion of tens of thousand of projections.

A breakthrough has been achieved with a protein synthesis initiation-like complex in which mRNA and fMet-tRNA is bound to the E. coli ribosome. The high occupancy and extraordinary conformational homogeneity of this specimen has enabled us to reach a resolution of 15 Å.

Vaults are ribonucleoprotein complexes highly conserved between species as diverse as Dictyostelium, rat, and human. Thousands of vault particles exist per cell in various tissues including liver, lung, kidney and spleen. These complexes have been found in the cytoplasm as well as associated with the nuclear membrane. Vaults have a large barrel-like middle section with a protruding cap on each end. These structures have an average mass of 12.9 ± 1 MDa, as measured by STEM, and are composed of 3 proteins, p210, pi92 and pi04, as well as an RNA strand of approximately 90 bases. The major vault protein, pl04, constitutes the majority of the particle mass, existing in 96 copies per structure. The major vault protein was recently recognized as the lung-resistance related protein (LRP), a known multidrug resistance gene that is over-expressed in certain cancer cells, including ovarian and myeloid leukemia cells.

A three-dimensional structure of the vault complex was generated using the QVIEW and IMAGIC-5 software packages (Fig. 1).

The capsid structure of the Hepatitis B virus (HBV) has been studied to resolutions below 10Å by cryo-electron microscopy, revealing much of its a-helical substructure and an apparently novel fold for a capsid protein. Although this represents a significant improvement in resolution for such studies, it is nonetheless still too low for complete tracing of the polypeptide chain. With the aim of establishing fiducial markers to aid in the process of chain-tracing, we have used cryo-microscopy to definitively localize specific peptides on the surface of the capsid. In one such study a gold cluster label was attached to a single cysteine residue engineered on to the C-terminus of the HBcAg assembly domain. The reconstructed density reveals a single gold cluster under each of the icosahedral 5-fold and 2-fold axes and connected to sites at either ends of the undersides of the dimers, thus pin-pointing the location of the C-terminus.

Ryanodine receptors (RyRs) function as the major intracellular calcium release channels in striated muscle, where they also play a central role in excitation-contraction (e-c) coupling, the signal transduction process by which neuron-induced depolarization of the muscle plasma membrane leads to release of Ca from the sarcoplasmic reticulum. Structurally, RyRs are the largest ion channels known, being composed of 4 identical large subunits (565 kDa). In situ, RyRs interact with numerous proteins that are essential for e-c coupling or regulation thereof. Some of these ligands include calmodulin, a 12-kDa FK506-binding protein (FKBP, an immunophi1 in), calsequestrin, triadin, and the dihydropyridine receptor (DHPR).

Detergent-solubilized, purified RyRs appear to retain their native structure as assessed by electron cryo-microscopy, and are amenable to three-dimensional reconstruction by single-particle image processing techniques. In Fig. 1, a solid-body representation of the reconstructed skeletal muscle RyR shows the structural complexity that is revealed at moderate resolutions (3-4 nm).

3D electron microscopy of macromolecules is now moving beyond the very successful tomographic approach of using images of a tilt-series of 2D crystals at known angles to single randomly oriented particles with less and less internal symmetry. The success of 3D reconstruction of symmetrical objects was predicated primarily on the redundancy of information due to lateral symmetry in 2D crystals, or to internal rotational symmetry in the icosahedral, helical or cylindrically symmetric particles. Such a spatially predictable redundancy permitted both a reduction in irradiation dose, to reduce radiationinduced structural alterations of the molecules, and an averaging of image noise due to the low number of electrons used and from the specimen support.

For 3D reconstruction from single particles the noise level in the images is crucial for the accurate determination of the relative orientation of the particles. In principle the central line theorem was recognized to be as applicable to asymmetric particles as it was to symmetrical viruses.

Nucleic acid polymerase structure has been studied by both X-ray and electron crystallography. To date, only the smaller, single subunit polymerases have been subjected to X-ray analysis, including the bacteriophage T7 RNA polymerase, which is the only RNA polymerase whose structure is known to atomic resolution. Lower resolution structures of several multisubunit polymerases have been determined by electron crystallography, including a mutant form of yeast RNA polymerase II which lacks subunits Rpb4 and Rpb7 (denoted A4/7 polymerase). All polymerase structures obtained by both X-ray and electron crystallography show a large cleft appropriate in size for binding duplex DNA, and further appear to contain a mobile arm allowing open and closed conformations of the cleft, presumably permitting entry and retention of DNA. Subunits Rpb4 and Rpb7 of RNA polymerase II form a dissociable subcomplex that has been implicated in the stress response and in the initiation of transcription. Human homologs of Rpb4 and Rpb7 have been identified.

Herpesviruses form an extensive family of DNA viruses, in which three subfamilies called alpha-, beta- and gammaherpesviruses, respectively, are distinguished on the basis of biological properties. In terms of overall structure, all herpesviruses conform to a common plan: the virion consists of a thick-walled capsid, ∼ 1250 Å in diameter, containing the genome, surrounded by a complex layer of proteins called the tegument (a feature unique to herpesviruses), and a lipoprotein envelope. At a more detailed level, it has not been clear whether the three subfamilies are also differentiated structurally. The molecular architecture of alphaherpesvirus capsids has been studied by cryoelectron microscopy of herpes simplex virus 1, revealing its icosahedral (T=16) surface lattice and the locations of the four abundant shell proteins. To ascertain how closely these features are emulated in a betaherpesvirus, we have studied capsids of simian cytomegalovirus (SCMV) by similar methods. By comparing the structures of nuclear capsids and cytoplasmic capsids (which have bound some tegument proteins),

Vaults are highly conserved ribonucleoprotein complexes found in thousands of copies per cell in various tissues. The major vault protein, pi 04, accounts for over 70% of the particle mass and has been recently found to be involved in cancer cell multidrug resistance. We have utilized cryoelectron microscopy (cryo-EM), and three-dimensional image reconstruction to generate a 3-D structure of the vault particle. Individual particle images were isolated using the QVIEW software package, which simultaneously performs density exclusion, planar background subtraction, and application of an elliptical mask. The IMAGIC-5 software package was used for the subsequent particle orientation and reconstruction steps (Fig. I).

A symmetry self-search test was used to measure the signal strength for a variety of point groups and the highest correlation was found for the cyclic 8-fold point group. We have utilized the sinogram correlation algorithms in IMAGIC-5 to find the Euler angles of both 1 μm and 2 μm defocus particle images.

ATP-dependent proteases play a major role in regulatory protein degradation in both prokaryotic and eukaryotic cells. ATP-dependent proteases in E. coli fall into two classes. The first class requires the interaction of structurally separate proteases with an ATPase, whereas in the second class both the protease and ATPase are formed from regions of the same polypeptide chain. We have studied the structure of several of these protein degrading complexes in E. coli and have found a remarkable similarity in the architecture of these macromolecular assemblies.

The prototypical protease of the first class has as its proteolytic component ClpP, a 14 subunit (MW 21,500) complex arranged as two lOnm-diameter stacked rings of seven subunits each. ClpP can interact with either one of two ATPases, ClpA or ClpX, each with unique substrate specificity. ClpA has two ATP-binding sites per subunit (MW 84,000), and its subunits are arranged as a 13nm (diameter) hexameric ring (MW -500,000).

Since cellular structures are rebuilt continually, protein degradation is essential for the maintenance of homeostasis. Misfolded proteins ensuing from genetic defects or environmental stress, are prone to aggregation; chaperones and proteases cooperate in minimizing such unproductive interactions. Last, but not least, protein degradation serves to terminate the lifespan of many regulatory proteins at distinct times and thus features as a key regulatory element itself. Proteins destined for degradation must be recognized and selected within the crowded environment of the cell. The stratagem of self-compartmentalization is key to controlling cellular proteolysis (1).

In recent years, a number of multisubunit proteolytic complexes have been described which possess large internal cavities or nano-compartments. This allows them to confine the proteolytic action to their interior; access to these inner compartments is usually restricted to the unfolded proteins. This, in turn, makes it necessary for these proteases to interact - either in a transient or in a permanent manner

The eukaryotic 20S proteasome (700 kDa) serves as the core complex for the breakdown of most cytosolic proteins by the ubiquitin-dependent pathway. This cylinder-shaped proteolytic assembly (17 nm long, 11 nm diameter) consists of four stacked heptameric rings that help to form a central cavity where degradation occurs. The catalytic sites are located on the cavity surface of the two middle (3 rings while the outer a rings serve as docking sites for various regulatory complexes.

PA700 (700 kDa, 20 protein subunits) is the major regulatory complex of the proteasome, forming a “cap” that attaches to either or both proteasome ends. The resulting superassembly structurally resembles the larger, more active 26S proteasome and exhibits both a higher intrinsic peptidase activity as well as a wider range of substrate specificity, including ubiquitin-tagged proteins. A second regulatory complex, the modulator (300 kDa), further stimulates the effects of PA700 at subsaturating concentrations of the activator with no detectable effect on the proteasome in the absence of PA700.

HK97 is a tailed temperate bacteriophage of E. coli that builds an icosahedral capsid using steps that include regulated assembly, proteolysis, radical conformational changes and the formation of novel covalent bonds (Fig. 1). This pathway is being exploited as a model system to explore how the formation of multiprotein complexes can be regulated by each of these mechanisms. We have identified and purified at least four intermediates (Prohead I, Prohead II, Head I and Head II) and examined them by cryo-electron microscopy and three dimensional reconstruction procedures (Fig. 2). Comparison of particle reconstructions at resolution of about 25 - 30 A have lead to major insights into the causes and purposes of the regulated changes that we have also characterized biochemically and genetically.

Prohead I consists of 420 copies of the 42 kDa gp5 capsid protein arranged as 72 blister-shaped morphological capsomers in a thick walled hollow T=7 icosahedral particle with a diameter of -470 Å.

Human α2 macroglobulin (α2M) has structural and functional properties that contribute to its uniqueness as proteinase inhibitor. It is the largest known (Mr=720,000) and the only natural proteinase inhibitor which has a broad range of reactivity and for which the reaction is irreversible. It has a vital role in the clearance of proteinases from the circulation and in regulating their activity in fibrinolysis, coagulation and complement activation.

An α2M molecule can entrap two proteinase molecules such as chymotrypsin and trypsin and can therefore be considered to contain two functional domains. Each subunit in the homotetramer has a bait region with cleavage sites for nearly all known endoproteinases and an internal thiol ester bond. A proteinase cleaves the two bait regions within both functional units leading to an activation and cleavage of the thiol ester bonds. Consequently, α2M undergoes a major structural change resulting in the irreversible entrapment of the proteinase.

Gold particles as markers in immunocytochemical procedures have undergone tremendous changes since their introduction by Faulk and Taylor (1). With the development of both small colloidal gold (<3 nm) and 1.4 nm gold clusters (Nanogold™) for immunocytochemical labeling, steric hindrance has been significantly reduced. In a comparative study using colloidal gold-anti CD3 complexes and particle sizes of 5 nm, 3 nm and less than 3 nm on surface labeling of lymphocytes there was a 1.5 fold increase in labeling density when 3 nm particles were used. There was no increase when smaller particles were used (unpublished data). In another unpublished experiment, labeling densities on an astrocytoma cell line were compared between 3nm colloidal gold-CD44 Fab’ fragments and 1.4 nm gold-CD44 Fab’ fragments. Again there was no detectable differences between particle sizes. The limiting factors remain the size of the IgG or Fab’ fragment, the ability to bind it whether by absorption or by covalent bonding without rendering it inactive and also the accessibility of the target molecule in the tissue.

The field of immunocytochemistry comprises a large body of methodologies whose purpose is to obtain spatial and temporal information about biological samples with a high degree of chemical specificity. The major differences in immunocytochemical methods are: (1) type of sample preparation (i.e., pre-embedding or post-embedding) and (2) type of detection system (i.e., chromogenic, fluorochromes, or particulate). The history of advances in immunocytochemistry can be understood, to a large extent, through the development of the various detection systems. All of these labeling systems have their appropriate applications and are in use today.

The introduction of colloidal gold as a particulate immunoprobe revolutionized immunocytochemistry, particularly at the ultrastructural level. The fact that these probes are discrete particles that are electron dense contributes to their wide acceptance. Due to the particulate nature of these probes, labeling density over a given structure or region of a cell can be determined by simple counting procedures.

Fluorescein and the 1.4 nm Nanogold cluster have been separately attached to polyclonal antibody Fab’ fragments to generate combined fluorescent and gold secondary antibody probes (1, 2), and in comparisons with Nanogold, colloidal gold and fluorescein-labeled secondary antibodies, neither label was found to be significantly compromised by the presence of the other. The proximity to a gold label with which a fluorophore retains sufficient fluorescence emission intensity to be useful is limited by non-radiative fluorescence resonance energy transfer (FRET) (3). This is a function of the degree of overlap of the fluorophore emission spectrum with the metal particle absorbtion spectrum, and of the separation of the metal particle and the fluorophore (4); for an effective dual-labeled probe, separation should be greater than the Forster distance (3), at which 50 % of excited-state decay occurs by FRET.