In science’s approach towards a comprehensive understanding of cell biology, sequencing programs are delivering complete lists of genes; proteomic efforts are cataloguing the functions, relative expression levels, and actual amounts of each protein present in cells at various times; comprehensive tagging studies are fluorescently labeling every protein in several cell types and roughly locating them at different stages of the cell cycle; and structural genomics initiatives are automating the tools required to solve the structure of every protein domain found in nature. Yet if we were to combine all this information in an attempted cell simulation, we would find one critical piece missing: the ultrastructure of the cell at the molecular level. Stated simply, cells are not merely “bags” of enzymes. Rather, “the entire cell can be viewed as a factory that contains an elaborate network of interlocking assembly lines, each of which is composed of a set of large protein machines” (B. Alberts, Cell 92:291-294). Unfortunately, the structural details of those networks of machines are yet to be discovered for lack of an effective tool. I believe cryoelectron microscopy (cryoEM) will become that tool, and my present interests are in both advancing the technique and applying it to important biological problems where it is most needed.

More specifically, my lab is developing and applying a technique called single particle analysis to solve the structures of large protein machines to near-atomic resolution, and a technique called electron tomography to image how those protein machines work together in the context of a living cell. Briefly, single particle analysis entails freezing a solution of a purified protein complex into thin films across EM grids, recording tens of thousands of images of individual, frozen complexes to high resolution in the electron microscope, and calculating a three-dimensional reconstruction of the complex based on those images. In electron tomography, a single unique object (like a cell) is incrementally rotated and imaged and a three-dimensional reconstruction is calculated.

Technology development. We are attempting to solve, for the first time, the structure of a large protein complex to atomic resolution without any type of crystallization, through cryoEM single particle analysis. Our model target is the large, rigid, symmetric, and readily available 20S proteasome, whose structure is already known for comparisons. This project involves automating our new, state-of-the-art microscope to record ~10 million molecular images and implementing effective image processing algorithms on the supercomputer to merge the data, as well as solving a variety of instrumental and technical challenges along the way.

Another development effort is planned to program the microscope to automatically record ~50 tomograms covering an entire eukaryotic cell (using yeast as an initial target), and merge them computationally to produce a three-dimensional model at “protein-machine” resolution.

Initial biological applications. One of the best examples of an assembly line of protein machines is seen in gene expression, where chromatin remodeling complexes, mediator complexes, basal transcription factors, RNA polymerase, RNA processing enzymes, and nuclear export machinery all appear to be physically linked. Understanding the mechanics of gene expression is a formidable but tremendously important problem in structural biology. Building on my own graduate and post-doctoral work on the structure of RNA polymerase, we are purifying various complexes involved in gene expression and solving their structures by single particle analysis. As we proceed to larger and larger complexes, which will soon include chromatin templates, the structures will become inhomogeneous and tomographic imaging will be required. Ultimately I envision fitting atomic models from X-ray crystallography into single particle reconstructions of complexes, which are in turn fit into tomograms of frozen "snapshots" of transcription in process on chromatin templates.

Another major project is comprehensive structural analysis of a "minimal" cell, namely Mycoplasma genitalium (MG). There is good reason to believe that state-of-the-art microscopes will deliver "molecular resolution" tomograms of these small bacteria, such that large protein complexes will be visible within their native cellular context. I expect that within the next decade, all the information needed to simulate this most simple of all free-living organisms will be obtained. My lab will contribute the essential medium resolution (~20Å) structural information: the spatial organization of the cell, the identification and location of large complexes, the nature of the cytoskeleton, the characterization of diffusion constraints, etc., or in other words all the ways in which the cell is not just a "bag" of freely diffusing small enzymes and substrates. This may include purifying protein complexes from MG and determining their structures by single particle analysis, as well as recording tomograms of whole cells. Ultimately we will fit together the entire cellular puzzle using atomic models of individual domains, near-atomic resolution single particle reconstructions of complexes, and medium resolution tomograms of entire cells. MG is ideal because it is simultaneously (1) the smallest free-living organism known, which will improve our imaging; (2) the simplest free-living organism, having only ~500 genes; and (3) the organism which will soon have the most comprehensively characterized proteome, since it is the target of the Berkeley high-throughput structural genomics consortium dedicated to solve the structure of each of its proteins.

In a similar study targeting a more complex but more intensely studied bacteria that localizes many of its proteins and nucleic acids internally, we are recording tomograms of Caulobacter crescentus.

We are also using tomography to study the molecular structure of HIV. A non-infectious form of HIV-1 is imaged while arrested at various stages from budding to maturation. The virions are composed of matrix, capsid, and nucleocapsid protein shells, which are composed of pseudo-crystalline arrangements of protein subunits. The image analysis includes novel combinations of two-dimensional crystallographic, single particle analysis, and atomic model "docking" methods. The ultimate goal is to visualize the various conformations and arrangements of the subunits throughout the virus life cycle, as well as resolve several outstanding questions about HIV structure.