Search

In this thesis, we investigated the conformational behavior of small molecules and peptides, and the influence it has on their properties. An overview of the computational methods which can be used to study thermodynamic and kinetic properties of biomolecular systems, and of the issues which arise when comparing calculated and experimental data, is presented in Chapter 1. The time-dependent evolution of a system can be studied on an atomic level by molecular dynamics (MD) simulations. In order to properly describe the conformational behavior of a molecule in an MD simulation, an accurate sampling of the potential energy surface is required. For this, multiple simulations can be performed starting from diverse structures, which were generated using e.g. enhanced-sampling methods. The kinetic information from multiple unconnected MD trajectories can then be retrieved by Markov state models (MSMs). In MSMs, the conformational space of a system is represented by kinetically metastable sets, and the interconversion rates between these sets are described by implied timescales. In Chapter 2, MD simulations and MSMs in water and chloroform were used to investigate the membrane permeability of cyclosporine A (CsA). CsA is a well known example of a cyclic peptide, which can passively diffuse through membranes, and is used as an anti-inflammatory drug. We sampled the transitions between “open” and “closed” conformations of CsA to understand the interconversion processes that facilitate membrane permeability. The state-of-the-art membrane permeability hypothesis states that only a low-dielectric (“closed”) conformation of a cyclic peptide can cross the apolar interior of a membrane. The MSMs in water and chloroform revealed the existence of two different kinds of “congruent” conformational states, i.e. metastable sets present in both environments. This finding led to an extended hypothesis of membrane permeability, with potentially more than one congruent conformation accessible in water, which can facilitate passive diffusion through a membrane. As the interconversions between the conformations of CsA present in polar and apolar environments proved to be essential for its membrane permeability, in Chapter 3 we investigated if the use of experimental structural information in MD simulations can induce interconversions between “open” and “closed” conformations of CsA. For this, nuclear Overhauser enhancement (NOE) distance restraints were derived from NMR measurements in chloroform. Although the structures present in the ensembles obtained in the restrained simulations resulted in favorable agreement with experimental data, the simulations starting from an “open” conformation in chloroform did not lead to interconversion to the “closed” conformation observed experimentally in chloroform. In Chapter 4, we investigated kinetic properties of cyclosporine E (CsE), a synthetic derivative of CsA. CsA and CsE are an example of a “permeability cliff”, where a small difference in structure (missing methylation of one backbone amide) leads to a significant change in membrane permeability, as the permeability of CsE is one order of magnitude lower than of CsA. The most striking difference between the MSMs of CsA and CsE were the interconversion timescales between the metastable sets. This indicates that the lower membrane permeability of CsE is a result of slower interconversion rates between the open and “congruent” conformations in water. In Chapter 5, we studied a series of six cyclic decapeptides with a constant backbone methylation pattern, but differently substituted side chains. Analogously to the methodology applied in Chapters 2 and 4, MD simulations were performed in water and chloroform. MSMs were constructed using core-set models, which led to a more reliable representation of the kinetically stable conformations. The membrane permeability was found to correlate with the population of “congruent” conformations in the MSMs in water. Chapter 6 describes the investigation of the stability of covalently cross-linked collagen triple helices. The conformational behavior of linker and wild-type collagens was investigated with MD simulations at 300 K and 400 K. Our results indicate that cross-linking increases the conformational stability of triple helices compared to the non-linked variants. The stereochemistry of the linked residues as well as the location of linkers proved to be important factors affecting the stability of the triple helical conformation. Chapter 7 addresses the problem of a proper partial charge estimation for small molecules. All MD simulations of peptides described in previous Chapters were performed using a force field developed for biomolecules. Small molecules, however, require individual parameterization strategies. Bonded and van der Waals parameters are usually taken from biomolecular force fields using a matching strategy, while partial charges are derived from quantum-mechanical (QM) calculations. In Chapter 7, we investigated the impact of the specific structures used to derive the partial charges. The effect of the partial charges obtained from these structures was quantified by comparing calculated and experimental hydration free energies. The results show that the choice of the input geometry for which the charges are derived has a significant impact on the obtained values and the resulting properties of the molecule. Chapter 8 discusses possible directions of research to extend and validate studies presented in this thesis, especially to understand membrane permeability of cyclic peptides in order to facilitate their rational design. Additionally, alternative approaches to obtain starting conformations, which sample the conformational space more efficiently, are proposed.