Kai Jochen Kohlhoff
In Nuclear Magnetic Resonance spectroscopy (NMR), chemical shifts provide readily measurable and highly sensitive probes of molecular structure. In structural biology they have thus often been exploited for the refinement of structures of proteins derived from other NMR observables.
In the first part of this thesis, a fast and accurate new tool for the prediction of chemical shifts from three-dimensional protein structures is presented. This predictor, CamShift, is primarily based on the information derived from interatomic distances. An assessment of the accuracy of CamShift is provided by the measurement of the root mean square deviation between calculated and observed chemical shifts, which yields values of 0.53 ppm, 0.29 ppm, 3.08 ppm, 1.18 ppm, 1.43 ppm, and 1.16 ppm for 1HN, 1Hα, 15N, 13Cβ, 13Cα, and 13C’, respectively. These results are comparable with those achieved by the most accurate chemical shift predictors currently available, which are, however, based on much more computationally expensive procedures.
In the second part of the thesis, we present a strategy (CamShift-MD) based on CamShift for using chemical shifts as restraints in molecular dynamics simulations to enable the determination of the structures of proteins.
In perspective, we discuss how the CamShift approach has the potential to solve structures for which traditional X-ray crystallography or NMR spectroscopy methods are not suitable as for example the highly dynamical non-native states that play an important role in the behavior of proteins in the cellular environment. CamShift restraints can also be used in molecular dynamics simulations of protein ensembles by comparing chemical shift values predicted for each structure separately and averaged over the ensemble with those from experiments.
|Institution||University of Cambridge|