The design and development of a drug is complex and usually extends over many years. Different approaches can facilitate lead optimization. This work focuses on specific aspects that could lead to advances in drug discovery.
After an introduction to drug design and the two studied target proteins, the first section of this thesis (Chapter 2) deals with the analysis and understanding of the function of transient binding pockets and the resulting perspective to improve the optimization of a lead structure. For this project, the already well-studied specificity pocket of human aldose reductase (ALR-2) was chosen. This enzyme is considered a key player of the polyol pathway. When glycolysis, an important degradation pathway of energy metabolism, is saturated, e.g., due to increased glucose levels in diabetics, the mechanism of the polyol pathway is activated. As a result, ALR-2 is stimulated to convert D-glucose to D-sorbitol. A series of inhibitors with low steric demand and functional groups of different electronic nature at the terminal aromatic moiety or terminal substituent, as well as their structural and thermodynamic properties, were studied in detail by isothermal titration calorimetry (ITC) and X-ray crystallography. In addition, the electrostatic potentials and charge distribution at the terminal aromatic groups of the inhibitors and their effects on binding to the transient pocket were analyzed by electronic surface potential (ESP) calculations. These analyses confirmed the previously established theory that terminal aromatic systems of the inhibitors with an electron-deficient aromatic group can trigger the opening of the specificity pocket and lead to preferential π-interaction with the electronrich indole moiety of W111. The results also indicate that a certain volume of inhibitor seems to be a prerequisite for pocket opening, since too small substituents lead to more complex binding positions with increased residual mobility. It was further shown that a shift in pH between pH 5 and 8 has no effect on the binding position of the inhibitors in the crystal with respect to the opening of the specificity pocket. This allows a comparison between thermodynamic and crystallographic data obtained at different pH values.
The following chapter of this thesis (Chapter 3) focuses on the strategy of preorganization used in latestage drug design for ligand optimization without having to resort to a completely different scaffold. Here, the S1 pocket of the also well-studied serine protease thrombin, whose conversion of fibrinogen to fibrin plays an important role in blood clotting, served as a model. In this work, we first present a series of congeneric thrombin ligands with a variety of functional groups that trigger preorganization prior to binding. The resulting structural fixation of a ligand in its bioactive conformation, either by fixation of the bound conformation in a suitable ring system or by fixation via intramolecular hydrogen bonds (H-bonds), has positive effects on affinity for entropic reasons. Fixation in solution and complex formation were characterized by crystallography, ITC, and molecular dynamics (MD) simulations. Clearly, these preorganizational modifications do not affect the overall binding mode, provided that the required bound conformation of the molecule with the archetypal binding mode is well reproduced by the modified molecules in the bound state. Thus, the most important interactions are preserved. However, the results of this study also show that the presence of an additional intramolecular contact, preferably in the form of a dominantly populated conformer, is equally important to achieve the expected affinity enhancement. At worst, the modified scaffold in solution adopts a new, strongly preferred conformation, and the preorganization effect, expected to enhance the affinity for the target, is lost. Based on these results, it can be shown that the thermodynamics of preorganization are largely governed by enthalpy rather than entropy. Furthermore, in this work, an important salt bridge is shielded by actively reducing its surface exposure, resulting in an improved enthalpic binding profile.
Part three of this thesis (Chapter 4) discusses the selectivity-determining features in the S1 pockets of the serine protease thrombin, the previously described blood clotting factor, and the related enzyme trypsin, which is responsible for digestion. Human trypsin is a proteolytic enzyme produced by the duodenum whose function is to degrade larger proteins into smaller components in the intestine. Thus, it represents one of the most important digestive enzymes. In these studies, a series of ligands were evaluated for their selectivity toward both peptidases using X-ray and neutron crystallography along with ITC. The results revealed that the local geometry of the two S1 pockets is highly conserved. However, thrombin cleaves peptide chains only after arginine, whereas trypsin cleaves after lysine and arginine. Thrombin has a Na+ binding site near D189 which is not present in trypsin. This suggests that simple steric features cannot explain the selectivity difference. Although the E192Q exchange at the edge of the S1 pocket has very little effect on the steric and dynamic properties of ligand binding and the local geometry of the two serine proteases, the different partitioning into enthalpy and entropy contributions is a clear indication of a given difference. The analyses of this study suggest that E192, together with the thrombin-specific sodium ion, contribute to generate an electrostatic gradient in the S1 pocket. This feature is not present in trypsin and is therefore important, together with the differences in solvation patterns in the S1 pocket, for the selectivity of both enzymes. The observation of protonation effects induced in this context is the first evidence for significant charge attenuation at the carboxylate group of D189 in thrombin compared to trypsin. From these studies, it appears that this phenomenon is one of the most important selectivity-determining features between the two proteases and thus influences and controls the other selectivity-distinguishing features that are less obvious at first glance, such as the differences in the solvation pattern and in the arrangement of water molecules in the two enzymes. In summary, it can be deduced from this work that it is not straightforward to assign the selectivity-determining features in the S1 pockets of thrombin and trypsin to a single dominant factor.
The last part of the thesis (Chapter 5) is an abbreviated subproject and deals with eight thrombin inhibitors that differ in the P1 residue. Due to the lack of time, only the crystal structures obtained by X-ray crystallography were used to analyze the binding behavior of the inhibitors. Their analysis shows that a P1 thiophene changes its binding behavior in the S1 pocket of thrombin once a halogen substituent is attached. The comparably electron-rich sulfur of a pure thiophene residue turns toward the inhibitor's own P2 and P3 carbonyl groups. In contrast, the sulfur of a P1-thiophene, which is rather electron-poor due to the halogen substituents, can serve as an H-bond acceptor and thus interact water-mediated with the π-system of Y228. However, a halogen atom in a P1-furan is not capable of recruiting such a water. Due to the shallower angle of oxygen compared to sulfur and the resulting change in geometry, the distance between the halogen and the ring of Y228 is optimized for a direct halogen-π-interaction. While a methoxy substituent in the para-position of a P1 6-ring does not exhibit a particularly strong interaction, the same substituent in the metha-position can also interact directly with Y228. However, compared to other P1 6-rings, this ring occupies a slightly different position. A hydroxy group in the ortho-position forms a water-mediated interaction with the π-system of Y228. Due to lack of time and the late takeover of data analysis by a former PhD student, additional thermodynamic and kinetic measurements of the binding process were not possible in the present work. Therefore, no definitive conclusions on the affinity and selectivity of thrombin can be derived based on this chapter.