Abstract
The surfaces of proteins are richly populated with a complex mixture of non-polar, polar and charged amino acid residues. The spatial arrangement of these residues dictates the property of the surface itself. A protein structure is energetically a highly frustrated entity. Although all protein molecules of a given kind share the fold motif, they may differ in loop conformation, side chain arrangements etc. The Protein Databank has many examples of single or multiple residues where the electron density maps are very weak. We have analysed the 25% homologous subset of all protein 3D structures. We have linked this information with homologous proteins for which the sequence, but not the 3D structure was known. We assumed that the 3D structural fold was identical to the protein in the 25% homologous subset. By binning the protein surface into 10 bins with respect to the extent of solvent exposure, we compiled a large database of information concerning the packing of amino acid residues in protein surfaces. This database can now be queried with respect to ,e.g., positive or negative amino acid pair correlations at a given solvent exposure depth. Very distinct packing rules emerge from this analysis. We envision that this information can be interfaced into protein structure prediction algorithms. The current status for this holy grail of molecular biology is that whereas secondary structures can be predicted with good accuracy, the packing of secondary structures into 3D structures cannot.

The 3D structure that has been experimentally determined, are defined only in one discrete spot in an effectively infinite physical chemical space. There is an assumption that the determined 3D structure will be applicable under all conditions where the protein is functional - but no proof for this essential assumption exists.

We have investigated a particularly challenging system consisting of a triglyceride lipase (Cutinase from Fusarium solanii) and its interaction with its substrate using a wide range of techniques, including 600 MHz NMR, ATR-FTIR, thermal scan CD and Steady State Fluorescence as well as quartz crystal microbalance and total internal reflectance fluorescence. We have investigated both the native and mutant lipase with respect to its behaviour under various physical chemical conditions. We have proposed a mechanism for the enzymatic function based on pH dependent electrostatics, and investigated the protein surface for special arrangements of amino acids. We found that Valine, Leucine and Isoleucine were overrepresented in the vicinity of the active site. The most exiting observation we have made is that the presence of the triglyceride substrate significantly increases the H/D exchange rates for many peptide NH. This strongly indicates that the 3D structure or the structural dynamics is altered because of the presence of the substrate. This raises a wide range of important questions in regard to how and when we can assume that the 3D structural information can be used as a basis for functional assessments.