Fragmentscreening an der CDPME-Synthetase (IspD): Neue Leitstrukturen für ein Enzym in der Malaria und Tuberkulose-Therapie

Summary Background In the last two decades the approach of fragment-based drug design (FBDD) has evolved as an important tool for the generation of lead compounds. This assessment is confirmed by the approval of the first drug discovered by this approach, Vemurafenib, in 2011 by the FDA. This success motivates to face the newly posed challenges of FBDD in the context of fragment hit detection. These challenges mainly involve the problem that conventional bioassay-based screening methods are often inappropriate to detect weakly binding fragment. Therefore, to obtain optimal results from fragment screening the optimal choice of the best method and the selection of a suited strategy is essential for the success. Goals In this thesis strategies for the rapid and cost effective identification of high-quality fragment hits have been developed. The first strategy is a fragment screening cascade. It coprises a screening-, a validation- and a characterization phase to facilitate fast and effective X-ray independent screenings. This is especially suited for protein targets for which no crystallization conditions have been found. The second strategy is based entirely on crystallography to render hit identification more effective. To advance in this field, the work flow of crystallization has been optimized and innovative methods for co-crystallization and crystal soaking have been developed. Results 1. Fragment Screening Cascade: A screening cascade was developed that, at the first stage, aims at the identification of primary hits using TSA and enzyme assays. As model protein IspD was used. The second phase consists of a ligand-based NMR (WATERLOGSY) discrimination of specific or unspecific fragment-protein-binding modes in order to exclude or validate primary hits respectively. In the third phase a protein-based NMR (TROSY) is carried out for the characterization of binding modes by means of reference spectrum comparison. In the first phase, 27 primary hits could be identified successfully and in the second phase this amount could be reduced to two hits. Fragment 134 was excluded owing to its non-competitive binding. In the third phase even the binding of fragment 005 could not be confirmed using TROSY-NMR. These experiments show that the cascade is an effective, less expensive and time-saving strategy for hit identification. Nevertheless, the cascade must be tested with more model proteins. The rather disappointing results of the TROSY-NMR are most likely due to the fact that the IspD-binding pocket is polar and charged. In contrast, the fragments of the applied library were less polar and only weakly charged. 2. Sole crystallographic fragment screening: A systematic approach to crystallography has been developed using microseeding and detailed adjustment by means of DVR/DV. This enables a fast improvement of crystallization conditions and the optimization of working conditions for other temperatures. The obtained IspD-crystals showed very good diffraction power and reproducibility. Furthermore, the “Baking-Method” has been developed, which enables, for the first time, efficient screening by means of co-crystallization. This is achieved because the conditions for co-crystallization must not be optimized individually for every fragment separately. Beyond this, the method enables the immobilization of complete fragment libraries in microtiter plates. In order to improve the process of soaking, the ISOSOAK-method has been developed. It offers the possibility to screen for optimal soaking conditions and facilitates soaking at very high concentrations of fragments. Furthermore, the UNISOAK-method has been developed which allows soaking under standard conditions or soaking of many different crystals simultaneously (tested for seven crystal at the same time). The diffraction power of soaked crystals in comparison to the uncomplexed proteins turned out to be at least of equal quality. Studies of the AG Klebe showed that different methods used for screening yield only sparse amounts of overlapping results. This calls for a sole crystallographic approach, since this facilitates the detailed characterization of binding modes. The methods developed here offer the required tools to implement a very effective, time-saving and cost-effective work flow. Particularly, in the field of soaking considerable progress has been made. Furthermore, cocktail screenings, which are problematic regarding data interpretation and reliability, become obsolete, since fragment soaking can now be carried out fast and individually. Due to these improvement fragment screenings come into reach also of small and medium-sized companies.