Biomedical Engineering Reference
In-Depth Information
molecular dynamics (MD) simulations
30,32
or corresponding to an exhaustive
set of allowed conformations.
71,140
The ability to construct dynamic ensembles using RDCs relies on being able
to compute RDCs for a given candidate conformer on the basis of its structure.
This, in turn, requires a means for determining the overall tensor of the
molecule. Domain elongation provides a simple solution to this otherwise
potentially intractable problem, given that the overall tensor of a non-
elongated RNA molecule may vary from molecule to molecule in a manner
that is difficult to measure experimentally or predict computationally.
129
In
elongated RNA molecules, the overall tensor of the RNA can be determined
by analysing RDCs measured in the elongated helix. Because the elongated
helix dominates the overall structure, internal motions in different parts of the
RNA molecule are less likely to modulate the overall tensor. Thus, the overall
tensor determined for the elongated helix can be used to predict RDCs for any
arbitrary structure and time-averaged RDCs can be determined for a given
MD trajectory or a candidate ensemble of conformations.
129
Although there are sparse examples of using RDCs to probe DNA
dynamics, one of the earliest RDC ensembles was constructed for DNA.
139
Clore and co-workers performed structure refinement on the model Dickerson
dodecamer against X-ray and NMR structures, X-ray scattering, CSA, and
RDC data. The incorporation of P-H39 RDC and CSA data fitted poorly to
existing structures, and only fitted well when a four-state ensemble was
allowed, demonstrating anisotropic motion within the DNA backbone. The
derived ensemble showed significant deviations from idealised B-form
geometry, with large amplitude tilt and propeller twist motions (9-18u and
15-30u, respectively).
139
Another approach for constructing ensembles uses RDCs to guide the
selection of RNA conformers from a pool containing thousands of
conformers.
141-144
First, the agreement between experimentally measured
RDCs and values computed from the entire pool of conformations, such as an
MD trajectory, is evaluated. For example, in the case of HIV-1 TAR, the
measured RDCs agreed poorly with those computed from an 80 nanosecond
MD simulation (RMSD 5 15.1 Hz compared to experimental error of y4
Hz).
16
This disagreement may reflect deficiencies in the force field, but it may
also reflect lack of convergence, given that the RDC timescale sensitivity
extends well beyond 80 nanoseconds into the millisecond time regime. To
construct an ensemble describing the experimental data, a 'Sample and Select'
method was implemented, operating as follows [Figure 9.4(C)].
141
Sub-
ensembles with increasing size are constructed in an attempt to find the
smallest member ensemble (N) satisfying the measured RDCs. Here, N
conformers are randomly selected from the pool and the agreement between
measured and predicted RDCs is computed. Next, one of the chosen
conformers is replaced randomly with another conformer from the pool, and
the agreement with measured RDCs is re-examined and the newly selected
conformer is either accepted or rejected based on the metropolis criteria
