If you can't be with the one you love – Can you make a model of it?

Abstract

One of the many techniques used in the Protein Structure Initiative to obtain structures of target proteins has been to express a large number of orthologues and attempt crystallization on all of them. Thus, it is often the case that the PSI may solve the structure of a protein that is a relative of the one under study in the laboratory. At the Fox Chase Cancer Center, we have developed and used methods for protein structure prediction on large number of targets under study by investigators. The purposes of molecular modeling are highly varied – from understanding sequence/structure/function relationships to ligand docking for lead drug design. Often, molecular modeling is used to design specific mutations to disrupt particular protein‐protein interactions while leaving others intact. We have developed databases and software to model proteins in biologically relevant states, including homooligomers and complexes of different proteins. This requires the ability to accurately identify biologically relevant states within protein crystals. We have found that the public databases of “biological units” are inconsistent, and that comparison of protein‐protein interfaces in all available template structure crystals is often required for identifying the best template oligomeric structure. This work was funded by the NIH, grants P20 GM76222 (PSI2) and R01 GM73784.