An experimental and computational framework to build a dynamic protein atlas of human cell division

Essential biological functions of human cells, such as division, require the tight coordination of the activity of hundreds of proteins in space and time. While live cell imaging is a powerful tool to study the distribution and dynamics of individual proteins after fluorescence tagging, it has not yet been used to map protein networks due to the lack of systematic and quantitative experimental and computational approaches. Using the cell and nuclear boundaries as landmarks, we generated a 4D image data-driven, canonical, computational model for the morphological changes during mitotic progression of human cells. We show that this model can be used to integrate the dynamic distribution of 3D concentration data for many mitotic proteins imaged by absolutely calibrated fluorescence microscopy in a large number of dividing cells. Analysis of a pilot data set, containing 28 proteins of interest imaged in dividing cells, allowed us to automatically identify sub-cellular structures and quantify the timing and magnitude of protein fluxes between them, as well as predicting dynamic multi-molecular biological processes such as organelle dis/assembly. Our integrated experimental and computational method enables building a 4D protein atlas of the dividing human cell. As part of the MitoSys and Systems Microscopy consortia, we provide here an approach that is generic and allows the systematic mapping and mining of dynamic protein localization networks that drive cellular functions.