Table of Contents:
In recent years, high-throughput screenings of large libraries of chemical compounds for the discovery of new biologically active substances have become increasingly important for the industrial research. The compounds of those libraries show a high diversity to cover a large chemical space. One disadvantage is, that these compounds mostly have similar molecular weights compared to approved drugs. Since different effects of the individual substituents occur, they cannot be considered separately from each other, so that the effect of the substituents and thus the further optimisation can only be estimated with difficulty. In the fragment-based approach, on the other hand, small molecules are found which can subsequently be adapted to the target protein. Fragments adopt perfect orientations to their target protein, because they only have a few numbers of possible interaction sites and no overlayed effects influence these findings.
An example of the consecutive expansion of a fragment to a ligand with nanomolar affinity to PKA is given in chapter 7. Previous studies already showed some possible drug candidates with high affinities to PKA. However, these compounds were only obtained as racemic mixtures, so that the measured affinities do not represent the final values of the individual compounds. Therefore, a stereoselective synthesis was done to obtain the R- and S-enantiomers separated from each other. Surprisingly only the S-configured compound was found in the different crystal structures of the protein-ligand complexes. This indicates a contamination of the supposedly stereoselectively prepared R-configured compound with the S-configured one, which seems to be more potent. Further studies did not resolve, if this could be due to a partial racemisation favoured by the applied synthesis conditions. Finally, from the X-ray crystallographic results, the S-configured β-amino acid derivative is supposed to be the compound with the highest affinity to PKA.
Another expansion of fragments was performed in another project (chapter 6). Here, fragments were chosen from an initial in silico screen of a fragment library and expanded follow-up compounds were studied. The fragment library represents an alternative to conventional libraries, as the fragments are derived from natural products. The aim was to find new lead structures for protein kinase A. In a FRET-based assay nine out of 22 follow-up compounds were found to have an inhibitory effect on PKA. However, crystallographic attempts to accommodate the ligands in the active pocket of the PKA using the soaking method have failed. One reason for this could be the steric requirement of the molecules, which prevents diffusion through the solvent channels. Therefore, cocrystallisation experiments were started, which should enable the formation of the protein-ligand complex. For each ligand one cocrystallisation set was prepared, from which crystals could be obtained in seven cases. X-ray crystallographic measurements and the analysis of the data resulted in one hit. Furthermore, adjustments of the crystallisation conditions can still be made to obtain more crystals of the protein-ligand complexes. So far, one protein-ligand complex could be found from an X-ray crystallographic experiment.
In the other projects, fragments series were studied to get different informations. For example, in chapter 2, the characteristics of the ATP-binding pocket of protein kinase A could be described by studying the interactions between the chosen fragments with the hinge region of the kinase. This study shows that a carboxylic acid and carboxamide group as well as an amidino group in the uncharged state accomplish the H-bonding pattern of the hinge region. However, this requires the molecule to be equipped with a suitable electron-pushing or pulling group to ensure the protonation state needed. In order to predict such electronic properties, elaborate quantum mechanical calculations are required, which are not yet routinely feasible. In the meantime, the results of this project can be used in addition to existing concepts to serve as a guide for the expansion of initial lead structures.
Chapter 3 compares the concept of cocrystallisation with that of soaking. For this, protein-ligand complexes with different ligands were compared by both methods. Different orientations of the structures obtained by both methods were found. For ligands with more degrees of freedom, larger differences were observed. Thus, soaking appears to be well suited for fragments, because of its time- and cost-effectiveness. Compounds with more degrees of freedom should be cocrystallized to present the more reliable picture.
X-ray crystallographic screenings of libraries by soaking is very time-consuming. This is especially the case, if the evaluation and interpretation of the data is not fully automated. A method to indicate the binding of chemical compounds to the hinge region of a PKA-crystal is developed and validated in chapter 4. One advantage to see fragment binding by simple means is, that the recording of X-ray crystallographic data can be redundant, if the compounds to be studied actually fails to bind. The applied method is based on a fluorescent reporter ligand, which is first of all populated in the binding pocket of the PKA crystal via soaking. From this time on, the crystal shows a fluorescence emission. By displacement of the reporter ligand, the fluorescence emission should be extinguished. Thus, in the absence of the fluorescence the protein-fragment complex structure to be studied should be measured by X-ray crystallography. In this work, aminofasudil was found to be a suitable reporter ligand and could be displaced by staurosporine and isoquinoline. In the experiment with isoquinoline the crystal unfortunately still emitted fluorescence. Nevertheless, the fluorescence spectrum changed, which could also be used as an indicator for the successful binding of the fragment/ligand of interest.
The fragments studied in chapter 5 are based on the natural substrate of PKA (ATP) and an approved drug (Fasudil). The aim was to characterise the binding of these fragments at PKA by isothermal titration calorimetry and X-ray crystallography and to look for possible differences. The smallest compound in this series, isoquinoline, surprisingly shows a high affinity to the target protein. Compared to Fasudil, this molecule binds in an alternative binding pose, even though isoquinoline is a motife in Fasudil. This short example shows, that a development of a drug, based on the crystal structure with isoquinoline, might have resulted in an alternative expansion strategy. On the basis of these findings, an extension in the 6-position instead of the chosen 5-position might have been chosen. In this chapter it is also shown that a bivalent binding motife at the hinge region does not lead to a gain in affinity. This motife is mainly present in the binding event of the weaker binding fragments.