期刊名称：Proceedings of the National Academy of Sciences
印刷版ISSN：0027-8424
电子版ISSN：1091-6490
出版年度：2006
卷号：103
期号：31
页码：11527-11532
DOI：10.1073/pnas.0604316103
语种：English
出版社：The National Academy of Sciences of the United States of America
摘要：Major advances in large-scale yeast two-hybrid screening have provided a global view of binary protein-protein interactions across species as dissimilar as human, yeast, and bacteria. Remarkably, these analyses have revealed that all species studied have a degree distribution of protein-protein binding that is approximately scale-free (varies as a power law) even though their evolutionary divergence times differ by billions of years. The universal power law shows only the surface of the rich information harbored by these high-throughput data. We develop a detailed mathematical model of the protein-protein interaction network based on association free energy, the biochemical quantity that determines protein-protein interaction strength. This model reproduces the degree distribution of all of the large-scale yeast two-hybrid data sets available and allows us to extract the distribution of free energy, the likelihood that a pair of proteins of a given species will bind. We find that across-species interactomes have significant differences that reflect the strengths of the protein-protein interaction. Our results identify a global evolutionary shift: more evolved organisms have weaker binary protein-protein binding. This result is consistent with the evolution of increased protein unfoldedness and challenges the dogma that only specific protein-protein interactions can be biologically functional.
