Using structural bioinformatics to model and design membrane proteins

Cells require membrane proteins for a wide spectrum of critical functions. Transmembrane proteins enable cells to communicate with its environment, catalysis, ion transport and scaffolding. The functional roles of membrane proteins are specified by their sequence composition and precise three dimensional folding. The exact mechanisms driving folding of membrane proteins is still not fully understood. Further, the association between membrane proteins occurs with pinpoint specificity. For example, there exists common sequence features within families of transmembrane receptors, yet there is little cross talk between families. Therefore, we ask how membrane proteins dial in their specificity and what factors are responsible for adoption of native structure. Advancements in membrane protein structure determination methods has been followed by a sharp increase in three dimensional structures. Structural bioinformatics has been utilized effectively to study water soluble proteins. The field is now entering an era where structural bioinformatics can be applied to modeling membrane proteins without structure and engineering novel membrane proteins. The transmembrane domains of membrane proteins were first categorized structurally. From this analysis, we are able to describe the ways in which membrane proteins fold and associate. We further derived sequence profiles for the commonly occurring structural motifs, enabling us to investigate the role of amino acids within the bilayer. Utilizing these tools, a transmembrane structural model was constructed of principle cell surface receptors (integrins). The structural model enabled understanding of possible mechanisms used to signal and to propose a novel membrane protein packing motif. In addition, novel scoring functions for membrane proteins were developed and applied to modeling membrane proteins. We derived the first all-atom membrane statistical potential and introduced the usage of exposed volume. These potentials allowed modeling of complex interactions in membrane proteins, such as salt bridges. To understand the geometric preferences of salt bridges, we surveyed a structural database. We learned about large biases in salt bridge orienations that will be useful in modeling and design. Lastly, we combine these structural bioinformatic efforts, enabling us to model membrane proteins in ways which were previously unaccessible