Residual density validation and the structure of Labyrinthopeptin A2

Abstract

The present work is divided into two main parts. The first part deals with the analysis and validation of the residual density. For the quantification and analysis the concept of the fractal dimension is employed. For this purpose several new descriptors were developed and implemented into a program called "jnk2RDA''. For testing the influence of the experimental resolution on the descriptors ideal simulated data were created to exclude the influence of any unknown sources of error. It could be shown that the fractal dimension of the residual density is strongly dependent on the resolution of the data. For simulating a more "realistic'' case Gaussian noise was added to the ideal data. It was seen that the fractal dimension of the zero residual density is relatively independent on the amount of noise whereas the flatness (maximum and minimum residual density values) decreases with increasing noise.Several parameters describing the electron density were varied from their refined values for simulated and for experimental data to investigate their influence on the residual density. It could be shown that each parameter has its individual influence on the distribution of the residual density and on its fractal dimension distribution and that the shape of the fractal dimension distribution can help to identify which parameter could be set to a more appropriate value. It was seen that the manipulation of parameters for experimental data does not necessarily show the expected results as the sources of error cannot always be identified and controlled.The progression of a multipole refinement starting from the Independent Atom Model to the final Multipole Model was monitored with the residual density descriptors and it could be shown that the parameters which contribute most to the improved description of the electron density are the monopole and multipole population parameters. An investigation of the influence of extinction correction on the residual density showed that the refinement of an even relatively small extinction parameter can improve the residual density remarkably. This improvement can be very large compared to any model improvement even including what can be achieved by refining the multipole population parameters.It was shown that the size and resolution of the residual density grid has an enormous impact on the residual density descriptors. The optimum grid has a resolution that is in a certain relation to the experimental resolution and which is proportional to the cell axes. For properly chosen values the computing effort and memory demand are minimized and the obtained information is maximized.The residual density descriptors were applied to verify the correct refinement on data of a disordered structure with multipole methods. It was shown that the residual density did not only improve in the molecular region when the disorder is taken into account but also for the whole unit cell.A new method was developed which allows for the correction of negative intensity observations that can occur for several reasons in an X-ray crystallography measurement. This new algorithm is a very effective way to correct the data and to make it possible that also negative observed intensities are included in the refinement, which is especially important for high resolution data. No such treatment is existing yet for small molecule crystallography.The second part of this work presents a new lantibiotic (called labyrinthopeptin A2) consisting of 18 amino acids, which shows new structural features. This cyclic peptide contains the amino acid lanthionine, two unusual cis peptide bonds and new intramolecular links. The Cα atoms of two alanine residues are bonded to the Cβ atoms of two neighbored alanine residues. This leads to quaternary substituted Cα atoms which is an uncommon structural motif. Despite the small rings in A2 consisting of only four amino acids the dihedral angles lie all in allowed regions in the Ramachandran plot. The peptide can formally be divided into two segments each containing two rings, the main chains of which are structurally very similar, which was shown by an overlay of the corresponding atoms.