Biomedical Engineering Reference
In-Depth Information
nanoseconds, and millisecond trajectories—the appropriate timescale for
averaging of RDCs—are still not generally available. 77 Long MD simulations
also only provide isolated trajectories and do not avoid the problem of
sampling phase space in a statistically meaningful way.
A popular alternative to performing long simulations is therefore to adapt
the potential energy force field to implement time- or ensemble-averaged
restraints, that force a multiple copy molecular ensemble to reproduce
conformationally averaged RDCs. 50,78-80 The application of such restrained
MD approaches to the study of the same ubiquitin and GB3 data, resulted in
the identification of a limited number of sites along the protein backbone that
required higher numbers of copies of the protein in order to reproduce all of
the experimental RDCs. 51,53 A more recent study using the SECONDA-
identified subset of RDCs as restraints resulted in an ensemble of structures
that was shown to span the conformational space of a number of complexes of
ubiquitin. This led to further discussion about the role played by intrinsic
dynamics in the formation of native protein-protein interactions and more
specifically concerning the importance of conformational selection as a driving
force. 50 While restrained MD methods are efficient for identifying conforma-
tional ensembles in agreement with experimental data, it is also clear that the
addition of a pseudo-potential to the physical force field may perturb the
simulated dynamics, especially the timescales, in a non-predictable way. One
additional drawback of ensemble-averaging against experimental observables
is the tendency to mix restraints that are averaged over different timescales. If
no knowledge of the timescale of characteristic correlation times of the
individual interactions is available, their interpretation in terms of interatomic
distances may lead to incorrect amplitudes or modes. MD simulations
performed in the absence of conformational restraints has therefore also been
extensively investigated. Here the problem is simply the development of
accurate force fields, and the ability to simulate enough sufficiently long
trajectories, or to overcome the sampling problem using enhanced sampling
procedures, as described below.
8.4.2.4 Accelerated Molecular Dynamics Combined with RDCs
In order to sample conformational space effectively, the accelerated molecular
dynamics (AMD) approach was employed in a restraint-free analysis that
aimed at providing a self-consistent structural dynamic representation of
protein conformational sampling. 81 AMD does not use an experimental
pseudo-potential, and is therefore restraint-free. Acceleration is achieved by
scaling the potential energy landscape by a constant factor (a), for all terms
below a given threshold (E boost ), thereby enhancing the escape rate between
low-energy conformational sub-states. On increasing the level of acceleration,
the simulation probes more conformational space. Re-weighting of trajectories
with respect to free-energy obtains a canonical Boltzmann distribution, and a
series of short standard MD simulations are seeded from this distribution. 62
Search WWH ::




Custom Search