Looking at Protein Interactions from the "Top"

Modeling with physics' most powerful tools coupled with those of bioinformatics, biochemical/molecular biology, and genetics/genomics is long overdue and has the potential to create a significant advance in our understanding of the complex orchestration of the components that underly cell function and all the modules responsible. Herein we propose an approach that uses all the best available empirical data to capture the essential structural features of dynamic protein interactions network in mammalian whole cell lysate. We amassed multiple metrics of protein abundances, chemical cross-linking, high salt AP-MS assays and genomic data to capture different yet complementary aspects of protein associations. We then implemented a statistical topological score (TopS) and used it in conjunction with non-linear dimensionality reduction algorithm and molecular modeling, pinpointing the structural map of protein assemblies. Using HP1 proteins as our case study, we show that HP1 proteins which are the key players in epigenetic repression, heterochromatin formation and maintenance were organized hierarchically and subsequently we classified the links of the network in different classes of different importance. HP1- proteins, associate with multiple repressor complexes. We also show that although HP1 copurifies with Zinc finger proteins which are amongst the most stable associations of HP1-alpha (CBX5), the association is mediated by TRIM28/KAP1. Finally, we show that within cells, the chromodomain folds in close proximity to the RBCC domain of TRIM28. The strategies developed here may generalized to other protein assemblies that are organized into hierarchical structures.