Tools to study molecules in high electric fields

This thesis describes the application of the quantum mechanics/molecular mechanics (QM/MM) Car-Parrinello methodology to two systems of high biological interest: (a) The potassium channel KcsA from Streptomyces lividans and (b) The DNA repair enzyme EndoIV of E.Coli. bacteria. The two studies provide new insights into key biological mechanisms such as the ion selectivity of cell membranes and the DNA repair system. Special emphasize was put on the importance of polarization effects in these systems. Most classical simulations performed today are based on non-polarizable force fields which only include electronic polarization implicitly. This thesis work shows that non-polarizable potential functions do not describe accurately the permeation in ion channels in all cases. Indeed, most current biomolecular packages were developed for the study of globular proteins and have only recently been applied to the study of ion channels. The importance of ion/protein induced polarization effects in the KcsA channel was demonstrated using different computational approaches: first, the fluctuations in the electronic density were studied (Chapter 3). Second, the quantum mechanical electrostatic potential inside the channel was compared to the corresponding result of popular MM force fields (Chapter 4). Third, the coordination geometry of the K+ and Na+ ions inside the channel was investigated (Chapter 5). It was found that the binding geometry of Na+ observed in classical simulations differs significantly from the results of ab initio simulations. In Chapter 6, the protonation state of the Glu71 and Asp80 residues is investigated. While errors due to a neglect of induced polarization effects amount to typically ∼0.1e/per atom, errors in the attribution of a protonation state cause an error of 1e. Thus a correct description of protonation states of the Glu71 and Asp80 residues is crucial. These residues located in the vicinity of the conductive pore in the KcsA K+ channel were found to share dynamically a proton, which may directly influence ion conduction. The proton exchange was found to be faster (sub-picosecond) than the ion translocation (nanosecond). It is important to stress that this mechanism could not have been seen by a classical study. The influence of Glu-71 on the ionic conduction was subsequently confirmed by structural and electrophysiological experiments (Cordero-Morales, Cuello et al. 2006; Cordero-Morales, Cuello et al. 2006), which indicate that Glu71 acts as a voltage sensor regulating the gating of the channel. Chapter 7 deals with an investigation of the catalytic efficiency of the DNA repair enzyme Endonuclease IV. The enzyme catalyses the hydrolysis of DNA at the phosphodiester bond of apurinic/apyrimidinic (AP) sites. The computer model highlighted the importance of electrostatic interactions in the stabilization of the phosphorane transition state. A revised catalytic mechanism is proposed for the phosphohydrolysis which involves a proton transfer. Our results also suggest that the reaction rate is not controlled by the chemical step, which is in agreement with experimental results.