Computational infrared spectroscopy : Calculation of the amide I absorption of proteins

Abstract: Infrared spectroscopy is an important technique that allows to retrieve structural information from the analysis of absorption spectra. The main application of infrared spectroscopy within life science is the study of the amide I band, which is correlated with protein backbone conformation and, consequently, with the secondary structure of proteins. However, band assignment and interpretation of the infrared spectra is not straightforward.Therefore, several simulation methods were developed to guide the interpretation of experimental amide I spectra. In this thesis, one of these methods is a normal mode analysis, which is based on the evaluation of the intrinsic vibration of the amide groups and the interactions between them. The calculation considers several effects: transition dipole coupling, nearest neighbor interaction, the local environment effect and the effect of hydrogen bond. From the normal mode analysis, it is possible to obtain the simulated infrared spectrum and the contribution of each amide group to a specific spectral range of the spectrum.The aim of this thesis and of the included publications is to explain this approach, to improve it and to show its potential. Results from simulations were compared with experimental data for different proteins of interest: amyloid-β oligomers and β-helix proteins. Simulated and experimental infrared spectra showed similar bands. Simulations also provided additional conclusions: they confirmed the random mixing of amyloid-β peptides in oligomers; they suggested that amyloid-β peptides contribute at least two strands in the structure of the oligomers; they revealed that the high wavenumber band, typical of antiparallel β-sheets, can be caused by other secondary structures, but not by parallel β-sheets. In addition, to verify and to improve the accuracy of this approach, simulation results were also put in a direct comparison with results from density functional theory calculations. From this comparison, a new optimal set of parameters for the calculations is suggested.