EM-Fold: De Novo Folding of α-Helical Proteins Guided by Intermediate-Resolution Electron Microscopy Density Maps

In medium-resolution (7-10 angstrom) cryo-electron microscopy (cryo-EM) density maps, alpha helices can be identified as density rods whereas beta-strand or loop regions are not as easily discerned. We are proposing a computational protein structure prediction algorithm "EM-Fold" that resolves the density rod connectivity ambiguity by placing predicted alpha helices into the density rods and adding missing backbone coordinates in loop regions. In a benchmark of 11 mainly alpha-helical proteins of known structure a native-like model is identified in eight cases (rmsd 3.9-7.9 angstrom). The three failures can be attributed to inaccuracies in the secondary structure prediction step that precedes EM-Fold. EM-Fold has been applied to the similar to 6 angstrom resolution cryo-EM density map of protein Illa from human adenovirus. We report the first topological model for the alpha-helical 400 residue N-terminal region of protein Illa. EM-Fold also has the potential to interpret medium-resolution density maps in X-ray crystallography.