Table of Contents:
Structure-based drug design is a widely used, successful method in modern drug discovery. To obtain a potent inhibitor an initial lead compound is stepwise optimized to inhibit the target protein. One prerequisite is the knowledge of the binding mode which is normally provided by an X-ray crystal structure. The availability of an initial complex structure is often the time limiting step.
The main focus of this work lies on the crystallographic binding mode determination of different ligands. All together 30 protein-ligand complex structures are presented which were determined in the course of two main projects.
The first project deals with pepsin-like aspartic proteinases. These proteins play an important role in several severe diseases such as malaria (plasmepsins), Alzheimer’s disease (β secretase) and hypertension (renin). Endothiapepsin served as a model system to investigate the binding mode of several novel ligands. This way the binding mode of inhibitors based on the Gewald reaction, inhibitors with an azepin scaffold and pyrrolidin based inhibitors could be resolved. During this process some interesting details could be observed. In case of the inhibitors based on the Gewald reaction three distinct binding modes were obtained in which the catalytic water molecule plays three different roles: Once it is displaced, once it mediates an interaction towards the ligand as an H-bond acceptor and once as an H-bond donor. The complex structure containing an azepin derivative shows, besides the ligand interacting with catalytic aspartates, two bound fragments of this ligand in the binding pocket. This way almost the entire volume of the binding pocket is occupied which reveals interesting hints on preferred interactions. In the complex structure with the pyrrolidin derivative a disordered wide-open flap with a partly open flap pocket could be observed. This structure therefore provides insights on the flexibility of this protein family. Furthermore, ITC measurements of this complex assured a double deprotonated catalytic dyade.
Additionally, complex structures with the clinically applied HIV-protease inhibitors ritonavir and saquinavir were obtained. With respect to ritonavir also a crystal structure with the secreted aspartic proteinase 2 (SAP2) succeeded. While the binding mode of ritonavir is equivalent with the one found in the HIV-protease, saquinavir shows an unusual addressing of the catalytic dyade.
Another focus of this project is the fragment-based drug design. Benzamidine and a hydrazine derivative were used as examples to show that small low-affinity ligands can be successfully crystallized. In a second step an inhouse library containing 364 fragments was tested against Endothiapepsin. This led to 55 assay hits. Out of the 55 assay hits for eleven compounds a crystal structure was obtained. The fragments show diverse binding modes and cover almost the entire volume of the binding pocket. Furthermore, this study suggests a critical view on the “rule of three” which is frequently used in the design of fragment libraries. Especially the threshold on the maximal number of H-bond acceptors seems to be too strict.
The second project deals with the optimization of a small unspecific probe molecule into a potent inhibitor. The protein kinase A (PKA) is used as an easy to handle model system. Based on phenol, and following several design cycles an inhibitor with an affinity in the low nanomolar range could be synthesized. To achieve this, the approach of a fragment based drug design was combined with the methods of a de-novo design. This study demonstrates how experimental and computer-based methods effective complement each other. Additionally for the first time a rather closed conformation of the glycine rich loop was observed. This provides a valuable starting point for further studies on the dynamic of this protein
In summary, the presented crystal structures provide different starting points for further drug design. In case of the pepsin-like aspartic proteinases many novel possibilities to address the catalytic dyad were revealed. Furthermore, the structure showing an open flap in endothiapepsin as well as the one showing a rather closed glycine rich loop in PKA give a first glimpse on the dynamics of these proteins. New methodical aspects are contributed through the conduction of fragment based screenings as well as a probe-based drug design. Last but not least the diverse structures are supportive, especially in combination with computer based and further biochemical methods, to better understand the nature of protein-ligand interactions.