Evolutionary Analyses of Protein Interaction Networks

Introduction

Protein-protein interactions (PPIs) are one of the most important components of biological networks. An understanding of the evolution of PPIs is crucial to elucidating how the evolution of biological networks has contributed to diversification of extant organisms. The amount of information about PPIs has grown rapidly due to the development of a high-throughput two-hybrid system, mass spectrometry of co-immunoprecipitated protein complexes, and bioinformatics approaches such as text mining from the many individual studies reported in the literature. The extensive data allow us to analyze the protein interaction networks from an evolutionary aspect.

Evolutionary studies of protein interaction networks can be classified into at least five topics based on their focus (Fig. 1). The most studied topic is the relationship between the number of PPIs of a protein (connectivity) and protein evolution (Fig. 1A). It has been shown that protein connectivity correlates with evolutionary rates (Fraser et al., 2002; Fraser et al., 2003). Protein connectivity can be directly calculated from the protein interaction network. To estimate evolutionary rates of proteins, we must first make multiple sequence alignments by using an application such as CLUSTAL W (Thompson et al., 1994). The number of amino acid substitutions is estimated from the number of differences between the aligned sequences with a biologically realistic statistical model, e.g. Kimura’s method. Many of the most widely used methods are implemented in the PHYLIP software package (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi) is the most commonly used sequence similarity search software. The very fact of easily detectable orthologs between distantly related species implies strong functional constraint and slow evolutionary rates without explicit rate estimation. The ability to detect orthologs in distantly-related genomes is one measure of conservation; the more distantly related the species where an ortholog is detectable the greater the conservation of the protein.

Figure 1.

Evolutionary analyses of protein interaction networks. Circles and black lines indicate proteins and protein-protein interactions (PPIs), respectively. Proteins of interest are black or gray and their PPI partners are white. Proteins may differ in their connectivity (A) or clustering (B) in the PPI network. The PPI partners of paralogs (C) and orthologs (D) are often conserved, and the relative physical location of genes whose protein products interact is often conserved over evolutionary time (E). (A) The black protein is said to have a greater connectivity because the number of PPIs is larger than that of gray one. (B) Black proteins are clustered more tightly than gray ones. (C) Black proteins are derived from a gene duplication. Duplicated pairs (paralogs) often have the same PPI partners, although there may be loss or gain of PPI partners during evolution. If a gene has a self-interaction before gene duplication, the duplicated pair may interact with each other after gene duplication. (D) Broken lines indicate orthology between human and yeast genes. The black and gray proteins interact in human and in yeast. (E) Rectangles and gray lines indicate genes and genomes, respectively. Broken lines indicate orthology between human and dog. The black and gray genes are located close together in the human genome and their protein products interact. Interacting gene clusters such as this in the human genome are likely to be conserved in other vertebrate genomes