From Protein Structure to Function with Bioinformatics.pdf

180 N.J. Burgoyne and R.M. Jackson7.5 Protein-Protein InterfacesThe genomic era has made great leaps forward in the understanding of the quantityand expression of genes in the genome. But the products of these genes are far lesswell understood, especially when viewed as a whole. There exist experimentalmethods to determine networks of protein-protein interactions, and whole areas ofbioinformatics dedicated to their analysis. The subject of this section is howeverlimited to the analysis and prediction of individual protein interfaces using theirsurface properties. Specifically, we focus on those interactions that occur throughassociation in which the proteins involved are independently stable in solution.These are known as non-obligate protein-protein interactions and form transientcomplexes, as opposed to the proteins that can only exist in oligomeric states (obligateinteractions).7.5.1 Properties of Protein-Protein InterfacesFrom observing large numbers of interactions we can learn statistical trends thattypically define a protein-protein interface. It is worth noting that for all of theproperties mentioned here there will be examples of complexes which have valuesthat are far higher than the average, and others that are far lower. A protein interfacebetween two globular proteins is typically a flat circular patch of protein surface,where the size is usually directly proportional to the size of the proteins (Jones andThornton 1996). Small monomers, like superoxide dismutase, can have interfaceareas as small as 700 Å 2 while the much larger tetrameric catalase monomers eachhave 10,500 Å 2 of interface (Janin et al. 1988). An average size interface is 800 Å 2per monomer, which is about 10% of the surface of a typical globular protein (Janinand Chothia 1990; Jones and Thornton 1995).Typically 55% of the interface surface is non-polar, 25% is polar and the remaining20% is contributed by charged atoms (Janin and Chothia 1990). This makes an averageprotein interface less hydrophobic than the protein interior but more so than the restof the protein surface. In addition, the interface is usually less charged than the rest ofthe protein surface. To form stable complexes, the interacting proteins must satisfy theburied polar and charged groups that lie within the interface. Indeed it is the complementaritybetween such groups that is thought to be the source of the specificity ininteractions (Chothia and Janin 1975). Complementarity is also high between hydrogenbonding groups, with 80% of such groups in either interface being satisfied by theinteracting interface (Xu et al. 1997). An average protein interface will have one satisfiedhydrogen-bonding group per 80 Å 2 of buried surface (Lo Conte et al. 1999). Acharged group in an interface will not normally be counter balanced by interacting witha group of opposite charge, but will instead be surrounded by other complementarypolar groups from the interacting molecule (Lo Conte et al. 1999).