Table of Contents:
Recent hardware development increase the computing power, in consequence many biological and chemical processes can now be successfully modelled in a way which was not to imagine 20 years ago. Examples of such processes are molecular dynamics studies of large biomolecules, the prediction of free energy of binding for protein-ligand complexes, investigations of reaction paths in enzymes, to mention only a few. One issue which is still unresolved concerns the accurate estimation of protonation states in protein-ligand complexes. In this thesis, we present the development of a novel charge assignment procedure named PEOE_PB (Partial Equalisation of Orbital Electronegativities - optimized for Poisson-Boltzmann calculations), which represents a method for the assignment of atomic partial charges. It works reliably with both proteins and small organic molecules using a consistent approach. Such charges are a key parameter in Poisson-Boltzmann (PB) calculations, which are a well-established method for pKa calculations of proteins. The development of the PEOE PB charges is necessary, because no generic procedure exists to perform PB-based pKa calculations of protein-ligand complexes. The development of the PEOE PB charges is described in section 2. We adapted the PEOE formalism to optimally predict solvation free energies of small organic molecules. Modifications were performed in a heuristic manner, i.e. purely result-oriented. The changes considered exclusively parameter a of the PEOE polynomial (see section 2.2.1). In our optimization protocol, we first attempted to reproduce best solvation free energies of the polar amino acids (r2 is 0.94, RMSD is 0.84) and continued with a dataset of 80 small organic molecules (r2 is 0.78, RMSD is 1.57). The latter step underlines the generic character of our PEOE PB charges. Subsequently, we performed calculations on a dataset of nine (apo) proteins with 132 experimentally determined pKa values and obtained an overall RMSD of 0.88. The dielectric constant of the protein is set to 20. Active site residues with highly shifted pKa values are given for two enzymes of the dataset. For these, the following issues were indicated: * The value of the dielectric constant has to be lowered from 20 to 4, which can be partially explained by the degree of burial of the binding pocket. * The orientation of the tyrosine OH group was observed to have a remarkable influence on the pKa value of an active site residue (a glutamate with a highly increased pKa value). This fact emphasizes that the position of polar hydrogens can be crucial. In the final step of our PEOE PB validation study, we performed pKa calculations for three protein-ligand complexes, where experiment revealed a proton transfer. All our calculations agree with the experimental findings. In section 3, we describe pKa calculations for a series of ligands binding to the serine proteases trypsin and thrombin. For the studied complexes, we previously performed an extensive ITC and crystallographic study and were able to identify protonation changes for four complexes . However, since ITC measures only the overall proton exchange, it does not provide structural insights into the functional groups involved in the proton transfer. By using Poisson-Boltzmann calculations based on our PEOE PB charges, we compute pKa values for all complexes from our previous work in order to reveal the residues with altered protonation states. The results indicate that His57, a member of the catalytic triad, is responsible for the most relevant pKa shifts resulting in the experimentally detected protonation changes. This Â®nding is in contrast to our previous assumption that the observed protonation changes occur at the carboxylic group of the ligands. The newly detected proton acceptor is used for a revised factorization of the ITC data, which is necessary in cases where the protonation inventory changes upon complexation. pKa values of complexes showing no change of protonation in the ITC experiment are reliably predicted in most cases, whereas predictions of strongly coupled systems remain problematic. Such coupled systems appear if two (or more) strongly interacting titratable groups are close to each other. The HIV protease (HIVP) is a prominent example of successful structure-based drug design, and it is a well-studied system for which, as experimental evidence shows, protonation changes in the active site occur upon ligand binding (section 4). In the apo enzyme, the catalytic dyad consisting of two aspartates is in the mono-protonated state. This protonation state can be altered by ligands bearing a cyclic urea moiety. Our PEOE PB charge model reliably suggests the experimentally determined protonation states in the active site of HIVP. Furthermore, we perform pKa calculations for two HIVP complexes with novel types of inhibitors developed and synthesized in our group [115, 114]. For these complexes, no experimental knowledge on the protonation states is given. For one of the compounds, containing a central pyrrolidine ring, the calculations predict that both catalytic aspartates should be deprotonated upon ligand binding (Table 4.12 and 4.13). Such a protonation pattern has yet not been observed in any HIVP complex. Similarly to the experimental trypsin/thrombin study, a combined crystallographic and thermodynamic investigation of the ligand complexation process of human aldose reductase (hAR) was performed in our group . The ITC measurements detected a proton transfer induced by the ligand binding process. Our calculations suggest an active site tyrosine residue (Tyr48) as proton acceptor, which is equivalent to a deprotonated tyrosine in the holo enzyme. This is in agreement with the ITC results for the Tyr48Phe mutant. Good quantitative agreement with the ITC experiment is obtained for the hAR complexes with inhibitors bearing a carboxylic head group. In contrast, the calculations cannot reliably predict the properties of inhibitors with a cyclic hydantoin moiety. A possible explanation for the deÂ®ciency is the fact that the ligand shows a strong electrostatic interaction with the active site tyrosine and lysine residues. Furthermore, its pKa value in aqueous solutions falls next tothe physiological pH range which makes the system very sensitive to the actual pKa shifts. One limiting factor for the large-scale application of the pKa calculation methodology presented here concerns the ligand processing. For this purpose, we included the PEOE_PB algorithm in the PDB2PQR program. This tool serves as the input file generator for the Poisson-Boltzmann solver APBS. Besides the fully flexible ligand consideration within the PDB2PQR framework, substructure matching has been enabled for the ligand. With this technique, it is possible to automatically detect titratable groups of the ligand and assign pKa values. These pKa values originate from a database which has designed built and currently contains 348 molecules with experimentally determined pKa values.