Complex cellular processes such as signal transduction, gene expression, motility and energy metabolism are often implemented using multi-component molecular assemblies. Understanding how these molecular machines function is an emerging frontier in cell biology, which will begin to bridge the gap that exists between our knowledge of the structures of individual proteins and those of cellular organelles. As more networks of interacting proteins emerge from genomics and proteomics, the need for innovative methods to illuminate these potentially disordered complexes will amplify. A major focus of my laboratory is the structure determination of large multiprotein complexes by analysis of high resolution images of single molecules. In single particle electron microscopy, images containing large numbers of well-separated protein molecules are recorded using low-dose electron microscopy of frozen-hydrated samples. Individual molecules are computationally selected, sorted into distinct classes, and averaged together to obtain distinct views of the molecule that have a high signal-to-noise ratio. The averaged views are then oriented with respect to each other, and used to reconstruct a model of the three-dimensional structure, which is subsequently improved using specific refinement algorithms. Pyruvate dehydrogenase multienzyme complexes are among the largest protein assemblies within cells, and catalyze the synthesis of acetyl CoA from pyruvate, a key metabolic reaction. While NMR and X-ray crystallographic studies have provided atomic models for the individual protein components that constitute this complex, determination of the structure of the entire complex, as well as an understanding of the mechanism of active site coupling have remained elusive goals. Using high resolution electron microscopy, we have arrived at an atomic interpretation of an 11,000 kDa icosahedral complex containing 60 copies of 2-oxo-acid decarboxylase (E1 enzyme) and 60 copies of dihydrolipoyl acetyltransferase (E2 enzyme). The molecular model provides a unique glimpse of the architecture and inner workings of a fascinating cellular machine, showing how the complex uses a swinging arm mechanism to couple the active sites of the E1 and E2 enzymes which are located at a distance of approximately 100 ? from each other. We have also analyzed the 1,800 kDa icosahedron formed by the E2 core enzyme to obtain a 14 ? model that is in excellent agreement with the 4.5 ? structure determined by X-ray crystallography. We are using this model system to optimize methods required to accurately orient the molecules, to correct distortions introduced during image collection on the electron microscope, and to enhance the speed of data processing so that it will be possible to analyze the hundreds of thousands of molecular images that will be required to attain near-atomic resolution three-dimensional models of non-symmetrical molecules. Continued refinement of single particle methods to facilitate the analysis of large dynamic complexes may provide a powerful tool to investigate other important macromolecular complexes present in normal and malignant cells.