NEW YORK (GenomeWeb News) – Protein-protein interaction networks are a powerful tool for gaining insights into pathogen-human relationships and, potentially, for assisting in therapeutic development, according to new research.
In an effort to understand how pathogens gain access to and exploit humans cells, Virginia Tech researchers drew from seven publicly available protein interaction databases, using computational analyses to create networks between human proteins and those of nearly 200 pathogens. The findings, published online in PLoS Pathogens today, suggest pathogen proteins tend to interact with central or widely interacting human proteins.
“[O]nly recently have large data sets for protein interactions become publicly available,” lead author Matt Dyer, a bioinformatician at Virginia Tech’s Bioinformatics Institute, said in a statement. “We have leveraged this opportunity to compare protein interactions between human and pathogen proteins … to provide important insights into the strategies used by pathogens to infect human cells.”
Dyer and his colleagues created their human-pathogen protein interaction networks by mapping 10,477 previously documented interactions between human proteins and proteins from 190 pathogens from 54 taxonomically related groups — 35 viral, 17 bacterial, and two protozoan.
They found that pathogen proteins interacted with 1,233 different human proteins — both inside and on the surface of cells. Based on the authors’ findings using gene set enrichment analysis, pathogen proteins tend to interact with human proteins that are either at protein communication hubs, interacting with many other proteins, or at bottlenecks that are integral to many different pathways.
However, because of the data available, nearly 80 percent of these protein-protein interactions were in the human-HIV system. So to avoid bias, the researchers performed several iterations of their analysis, creating networks that contained protein-protein interaction data from only high-throughput studies or from only manually curated interactions and looking at interactions in the presence or absence of human-HIV data. Again, in general the so-called hub or bottleneck human proteins interacted most often.
They also narrowed their focus, honing in on human proteins that interacted with proteins from at least two groups of viruses or bacteria. By determining which Gene Ontology functions were enriched in the various networks, the researchers pinpointed some of the most commonly affected pathways. Viruses, for example, often interacted with proteins mediating cell cycle regulation and transport across the nuclear membrane.
“Our results provide a striking example of how pathogens may have evolved the ability to exploit the structure of interactions between human proteins in order to promote infection,” Virginia Tech computer scientist T.M. Murali, an author on the paper, said in a statement. “This global study also suggests that many viruses share similar strategies to control the human cell cycle, regulate programmed cell death, and transport viral genetic material across the nuclear membrane in the human cell.”
Interestingly, the team also reported that pathogen proteins often target proteins in established cancer networks. This is particularly intriguing given the fact that some pathogens, for instance papillomavirus, can cause cancer in some cases.
Although they caution against over-generalizing the results from this study, the authors argue that the network approach may be a valuable resource for those developing therapeutics against pathogenic viruses or bacteria.