Modeling membrane protein structure through site-directed ESR spectroscopy

Site-directed spin labeling (SDSL) electron spin resonance (ESR) spectroscopy is a
relatively new biophysical tool for obtaining structural information about proteins. This
thesis presents a novel approach, based on powerful spectral analysis techniques (multicomponent
spectral simulations and evolutionary optimizations of ESR spectra) and
modeling of the protein structure by calculating the restrictions of the conformational space
of the attached spin label.
First, the feasibility of the ESR spectral analysis was enhanced by speeding-up the
spectrum optimization and by automation of the analysis routines to enable the handling of
large sets of spectroscopic data (e.g., for the joint analysis of SDSL-ESR spectra from
multiple sites of a spin-labeled protein). According to the testing examples a speed-up
factor of 5-7 was achieved.
Secondly, SDSL-ESR was used to study the topology of the long N-terminal domain
of the photosynthetic light-harvesting complex CP29. Wild-type protein containing a single
cysteine at position 108 and nine single cysteine mutants were produced, allowing to label
different parts of the domain with a nitroxide spin label. In all cases the apoproteins were
either solubilized in detergent, or they were reconstituted with their native pigments in
vitro. The spin label ESR spectra were analyzed in terms of a multi-component spectral
simulation approach. These results permit to trace the structural organization of the long Nterminal
domain of CP29 leading to a structural model for its N-terminal domain.
Thirdly, we proposed a novel way to translate the local structural constraints gained
by SDSL-ESR data into a low-resolution structure of a protein by simulating the
restrictions of the local conformational spaces of the spin label attached at different protein
sites along the primary structure of the membrane-embedded protein. The proposed
structural model takes into account the restricting effect of the protein backbone, amino
acid side chains and lipid environment. We tested the sensitivity of this approach for
artificial oligopeptides and then for membrane-embedded M13 major coat protein
decorated with a limited number of strategically placed spin labels by employing highthroughput
site-directed mutagenesis. We found a reasonably good agreement of the
simulated and the experimental data taking a protein conformation close to an α-helix.
Finally, by using an optimization algorithm we optimized the parameters of the
protein-lipid model by improving the fit of the simulation data to the experimental
conformational space data. The outcome of the optimization was a family of best-fit
structures of membrane-embedded M13 protein, which not only agree with the available
SDSL-ESR data, but also was consistent with a recent model based on site-directed
fluorescence labeling.
Therefore, the present method provides a challenging starting point for the
development of a powerful methodology for the protein structure characterization, an
alternative approach to conventional techniques.