The research group Klebe studies various aspects of protein-ligand interactions. One focus is based on computer simulations. Via the comparison of structural data and the docking and the simulation of interaction patterns it is attempted to establish correlations among the molecular properties of the considered binding partners. As a result structure-activity and structure-function-relationships are found which help to predict the properties of molecules in mutual recognition processes and guide the experimentalists to focus on the most important interactions. Computer methods help to select from huge small molecule compound libraries of the most promising candidates that fit into the binding pocket of a macromolecule. Through interactions with the macromolecule they can modulate the function of the latter and redirect its properties. The computer methods also serve the purpose to discover putative binding partners in a cell, apart from the actually targeted one. This approach helps to provide first ideas about undesired side effects and unwanted cross-reactivity of drug molecules.
In the context of synthetic microbiology the group collaborates closely with the research groups Hüllermeier and Freisleben in computer sciences. Goal is the prediction of protein function on the basis of the spatial structure. The function of proteins is intimately coupled with the communication of molecules performed through the formation of intermolecular contacts. This step requires mutual recognition. As a result of the transient binding of molecules enzymatic transformations in metabolism or transmission of information in signal transduction can occur. Subsequently, important changes in an organism can be triggered or signal transduction can be transmitted or readjusted.
Molecular recognition nearly exclusively involves transient binding of the interacting partners via their highly structured surface regions. These regions are usually classified as binding pockets. Similarities in these pockets suggest functional relationships. They also allow conclusions on putative redundancies across proteins with respect to their function. If a particular protein should be inhibited in a disease situation it is of utmost importance to know whether a related protein exists that can supersede and replace the function of the inhibited one. In synthetic cells, it is important to know the minimal configuration of non-redundant proteins. Also here the comparative analysis of proteins with respect to functional predictions can help.
To perform the computational comparison in an efficient way algorithms have been developed and are further enhanced to detect automatically binding pockets, to classify them in terms of physicochemical properties and to make them available for mutual comparisons. The query pocket of a protein of interest can be compared against all pockets found in crystallographically determined proteins. Functionally related proteins are detected independent of any fold or sequence similarity and can indicate possible cross-reactivity. At present, the entire space of binding pockets is compared. Any given structuring of this space is analyzed and will hopefully indicate putative clustering of proteins with similar function.
Up to now structural data of proteins and their ligands were extracted from crystallographically determined complexes in the PDB. For this purpose the research group has developed database systems such as ReLibase, Cavbase, Secbase, or Waterbase. The group is also active in determining crystal structures and more than 360 entries in the PDB, containing at present about 83,000 structures, have been contributed. In SYNMIKRO the group helps to maintain a state-of-the-art infrastructure for structural biology. Beyond crystallographically determined structures it is planned to incorporate model-built complexes to expand the scope of the approach. More information about the research topics of the group can be found on the web http://pc1664.pharmazie.uni-marburg.de/research.