Computational modeling of fluorescent dyes for accurate FRET-assisted in vivo protein structural modeling

Proteins naturally occur in crowded cellular environments and interact with other proteins, nucleic acids, and organelles. Since most protein structures have been studied in non-physiological conditions, the effects of the cellular environment on protein structure at the atomic scale are largely unexplored. Recently, Forster resonance energy transfer (FRET) has been shown to be an effective experimental method for investigating protein structure in vivo. In FRET experiments, the distance between fluorescent dyes attached to a given pair of amino acids can be determined with angstrom-level resolution, but this inter-dye distance is different from the inter-residue C_α-C_α distance. To model protein structure with atomistic detail in vivo, it is important to determine the mapping between C_α-C_α distances and inter-dye distances. How can we predict inter-dye distances from a known protein structure? Further, how can we infer unknown C_α-C_α distances from a set of FRET measurements of inter-dye distances? Here, we determine the minimum number of dye conformations that accurately recapitulate the 40 inter-dye distances measured by FRET experiments on two distinct conformations of T4 Lysozyme. These results establish the feasibility of accurate FRET-assisted protein structural modeling in vivo at atomic scale.