Tools for Generating and Assessing Pseudo-atomic Models from 3D Electron Microscopy Maps of Macromolecular Assemblies.

Cellular processes are governed by the complex coordination and dynamics of biological macromolecules called proteins and nucleic acids. These macromolecules do not act in isolation but form assemblies. Understanding the structure of these macromolecular assemblies can often teach us how they function, which is important for the basic understanding of the cell as well as for developing cures for disease.

Structural biology aids this project by providing "pictures" of macromolecular assemblies. This is done through the use of experimental techniques such as cryogenic electron microscopy, a method whereby cell components are frozen and then bombarded with electrons, yielding an image of the sample. These images can be combined to give us a low-resolution 3-dimensional picture of the assembly structure. Though this technique has led to great discoveries, there are limitations to what it can accomplish due to the low-resolution of the picture, leaving us with significant gaps in our knowledge.

Our research aims to overcome this gap by developing computational methods that pull together information from a variety of experimental sources to construct clearer and more complete pictures of the assembly structures imaged by electron microscopy. Like a jigsaw puzzle, we will fit atomic structures of proteins (known from other experiments) together, within the low-resolution picture, creating a refined "pseudo-atomic" picture of the entire assembly.

Many cellular processes are governed by the complex coordination and dynamics of macromolecular assemblies. A detailed description of the structures of these assemblies can be extremely useful in understanding these processes. In the last decade, 3D electron microscopy (EM) techniques have become essential in achieving this goal, with an exponentially increasing number of structures solved. Although 3D EM allows us to determine the structural organisation of assemblies not amenable to other methods, the resolution range of the resulting density maps usually reveals the overall molecular shape but does not allow for an atomic description. For this reason, 3D EM maps are almost always further interpreted by fitting in them atomic structures of assembly components determined using X-ray crystallography, NMR, and atomic models from protein structure prediction methods.

We recently introduced a testing procedure for determining the quality of component fits and a number of scores that could be used for that purpose. We propose to build upon this work and create a computational tool for scoring and assessment of fits into 3D EM density maps. We also aim to develop a tool for simultaneous fitting of multiple components into low-resolution 3D EM maps of large assemblies. Our tools will be combined into a fully automated open-source software package, which will be made publicly available, especially to serve the EM community and related structural and computational biology fields. Ultimately, the software will be applied to provide pseudo-atomic models of many macromolecular assemblies, involved in a variety of cellular functions.

The computational tools for model generation and assessment from 3D electron microscopy (EM) data proposed in this project will primarily be of benefit to the increasingly growing EM community in the UK and internationally due to recent development in the field and a dramatic increase in data generation on structures of macromolecular assemblies. These methodologies and tools will be of general interest to bioscience researchers via systems level analysis of biological processes, as highlighted as one of BBSRC principal strategic aims ("Developing and embedding a 'systems' approach to biosciences in order to advance fundamental understanding of complex biological processes"). Additionally, since the assemblies expected to be analysed by the proposed tools are involved in many cellular functions (including those related to infection and disease), researchers in other fields of life science and medical research are also likely to benefit.

More specifically, the project could also have a great impact on human health since the resulting methods will be applied to 3D EM data on Type 4 secretion systems, which have been shown to be involved in antibiotic resistance and infection. A detailed description of the interactions within the entire system will be of high interest to the pharmaceutical industry, potentially allowing development of inhibitory drugs that target protein-protein interactions. Additionally, the software will contribute to the understanding of the mechanism of membrane pore forming proteins that are important virulence factors in diseases such as pneumonia and meningitis. Thus, results may have impact on the economic competitiveness of the United Kingdom in pharmaceutical research.