SUMMARY
The complexities of the proteome are both immense and dynamic, reflecting diverse, context-specific
expression of genes in individual cells and distinct isoforms for many proteins. In addition, individual proteins
may participate in several distinct protein assemblies during their lifetimes and undergo dynamic signal-
dependent re-organization in order to impart unique functions and cellular attributes. Moreover, numerous
protein assemblies self-combine and compartmentalize to generate organelles and signaling modules within
the cell, which are inherently dynamic. During the past 9 years, we have designed, validated, and applied a
platform for the large-scale analysis of protein interaction partners using affinity purification-mass spectrometry
(AP-MS) termed BioPlex, which has allowed us to profile interaction partners for over 10,000 nonredundant
human bait proteins in two contrasting human cell lines: HEK 293T and HCT116. In addition, we repeated
2000 of these IP’s in three additional cell lines – U2OS, RPE1, and HeLa – to broaden our exploration of cell-
specific interactomes. In aggregate, these 25,000 AP-MS experiments in five distinct cell lines have enabled us
to define an atlas of nearly 280,000 protein-protein interactions among 15,500 human proteins. The majority
have not been reported through independent efforts. The robustness of BioPlex, when benchmarked against
other studies and when compared across cell lines, parallels or exceeds available resources, allowing us to
broadly define human protein communities, predict functions and localizations of unstudied proteins based on
interaction partners, and define a large number of domain-domain enrichments that begin to impart structural
architecture upon the network. Finally, by repeating AP-MS experiments for thousands of baits in multiple cell
lines, we have created the first and largest proteome-scale, context-specific models of the human interactome,
revealing extensive remodeling that reflects the unique biology of each cell line. Yet, as valuable as these
BioPlex networks have been for biological discovery, they remain incomplete models of the human
interactome. Nearly 1 in 4 proteins do not appear in the BioPlex network at all, and a similar number of proteins
appear in these networks as preys only, providing just a partial view of their interactions. Finally, we have only
just begun to explore BioPlex in the context of other complementary genomics resources. In this renewal, we
seek to greatly enhance and extend our efforts in three major ways: First, we will prosecute an additional
7,500 genes in HEK293T cells. In this way, the BioPlex network will transition from utilizing 50% of all human
genes as baits to approximately 90% used as baits, revealing many novel interactions not possible previously.
Second, the same 7,500 clones will be examined in HCT116 cells to provide validation and explore cell line
specificity. Third, we will integrate BioPlex with complementary genome-scale resources including AlphaFold
and the Human Protein Atlas to model protein interactions at scales ranging from molecular interactions to
cellular structures.