Multifrequency Electron Spin Resonance Spectra of a Spin-Labeled Protein Calculated from Molecular Dynamics Simulations

Abstract

Multifrequency electron spin resonance (ESR) spectra provide a wealth of structural and dynamic information about the local environment of the spin label, and indirectly, about the protein to which they are attached. Relating the features of the observed spectra to the underlying molecular motions and interactions is, however, challenging. To make progress toward a rigorous interpretation of ESR spectra, we perform extensive molecular dynamics (MD) simulations of fully solvated T4 Lysozyme, labeled with the spin label MTSSL at positions 72 and 131. These two sites have been the object of numerous experimental studies, and are generally considered as prototypical solvent-exposed sites on the surfaces of α-helices. To extend the time window afforded by the MD simulations, stochastic Markov models reflecting the dynamics of the spin label side chains in terms of their rotameric states are constructed from the trajectories. The calculated multifrequency ESR spectra are in very good agreement with experiment for three different magnetic field strengths without adjusting any parameters. During the trajectories, the spin labels interconvert among a fairly large number of conformations, and display a propensity to form interactions with protein residues other than their nearest neighbors along the helix. The detailed picture of the spin label emerging from the MD simulations provides useful insight into the molecular origins of the available spectroscopic and crystallographic data.

Conformational dynamics and distribution of nitroxide spin labels

“…The same approach was then applied to residues 72 and 131 in T4 lysozyme [49]. At site 131, spectra simulated for the p and m states of 3 were nearly identical and a good agreement of spectra obtained at 22°C…”

“…As such data cannot be directly obtained by experiment, all assumptions involved in these computations have to be considered carefully before conclusions can be drawn. First, transitions between the p and m states of 3 are too infrequent in MD simulations at ambient temperature to obtain good statistics [49,52]. Frequent transitions are observed at a simulation temperature of 600 K [47];…”

“…Transitions between the p and m states of 3 are particularly slow and are observed only rarely in MD trajectories up to 100 ns at ambient temperature [47][48][49]. Sampling problems were also encountered in an MD study where distance distributions between spin labels were directly taken from MD trajectories and compared to experimental distributions [50].…”

“…For some of the trajectories, good agreement between simulation and experiment was observed, while other trajectories were discarded because the agreement was substantially worse. The enhanced 'umbrella' sampling method was used to estimate the m:p ratio for 3 with an uncertainty of about 4% in MD computations for residues in T4 lysozyme labeled with MTSL [49]. The ratio was found to be 27:73 for residue 72, and 55:45 for residue 131, indicating that subtle differences in the environment at these two surface-exposed helical sites can shift this ratio significantly.…”

“…In MD simulations using current standard force fields, transitions between different states of 1 and 2 are somewhat slower (a few nanoseconds to tens of nanoseconds [49]) than those between different states of 4 and 5. The frequency of transitions between 5 states depends on the other dihedral angles, in particular on 4, and can be very high for some conformations due to an almost flat 5 potential [51].…”