Mapping protein-protein interactions is critical for deciphering complex cellular functions. For this reason, high performance mass spectrometers have been developed, allowing investigators to analyze hundreds of thousands of peptides from thousands of proteins. There is, however, a stumbling block: The ability to generate data in proteomic experiments far outstrips the ability to analyze it. As a result, enormous amounts of data have been submitted to public databases, while only a small portion of these have been studied beyond confirming the presence or absence of a group of proteins.

UBC investigators at the Biomedical Research Centre and the Department of Chemistry, led by CBR scientist Juergen Kast, have been tackling this problem head-on. They surmised that by extracting the data sets that are the most likely to contain information on protein-protein interactions for a protein of interest, they should be able to identify the proteins that are frequently observed, which are either known or unknown interaction partners or nonspecifically binding proteins. Using this approach, they could then generate new hypotheses for novel protein-protein interactions by performing an in silico protein interaction analysis.

Using the Global Proteome Machine database (GPMDB), which is the largest curated and publicly available proteomics data repository, they developed an in silico protein interaction analysis tool to identify candidate protein-protein interactions and proteins with shared functions in protein networks. They then applied this method to study human integrin αIIbβ3, a major receptor involved in the activation of platelets. They identified 28 proteins, including talin1, kindling-3 and Rap1b, along with integrin αIIbβ3, which constituted a protein network for integrin and platelet activation. They also identified other proteins that are involved in platelet activation and aggregation, including fibrinogen, coagulation factor XIII and vinculin. They went on to validate the method by performing co-immunoprecipitation experiments, and by confirming the findings in protein-protein interaction databases and in the literature.

This general method for in silico protein interaction analysis using the publicly available data in the GPMDB http://gpmdb.thegpm.org/thegpm-cgi/pvip.pl may be widely applied as a hypothesis generator for identifying and studying known/candidate protein interactions and mapping of protein networks. The potential value of such an approach is enormous, as it is through an understanding of the intricacies of protein-protein interactions that scientists will be able to develop novel therapeutics and diagnostics for a range of diseases.

Get the whole story in Journal of Proteomics Research (2011) 10, 656-668