Chemistry Reference
In-Depth Information
when multiple RDCs per peptide plane, or RDCs from more than one type of
alignment medium, are available [
335
,
339
].
When structures of secondary structure elements or globular domains are
known, it is possible to use RDCs to calculate the relative orientation of these
structural units. First demonstrated for multi-domain proteins [
344
,
345
], a similar
strategy has since been used to determine average protein structures for small
membrane proteins in the absence of other long-range restraints, using TM helices
as the rigid, or tightly restrained bodies [
137
,
318
,
346
]. However, for each
structural unit there will be four equivalent orientations that can be defined with
respect to a right-handed alignment frame (assuming that the alignment tensor is
not axially symmetric) [
347
]. Hence it is necessary to eliminate this degeneracy
through the measurement of two or more sets of RDCs under different aligning
conditions [
348
,
349
] or by the recognition of structural restraints that can rule out
extraneous solutions [
318
,
347
,
350
,
351
].
A related application for RDCs has also been described based on the sequence-
dependent pattern of RDCs along a helical structure, called a “dipolar wave” by the
Opella group [
317
,
352
,
353
]. The magnitude and periodicity of the dipolar wave
depends on the orientation of the helix, and can allow irregularities in helix
structure to be identified. Most commonly, dipolar waves have been used to help
determine the location of helices in a protein sequence, allowing these structural
elements to be more rigidly restrained over the course of a structure calculation
[
57
,
159
,
323
,
354
]. This is particularly useful for larger helical membrane proteins,
since
-helices are not well defined by the NOEs available in sparsely protonated
samples [
262
].
Aside from helices, RDCs have a more general capacity to recognize related
structures. This approach has been used to identify homologous structures in the
protein databank [
355
], which could then be used as a starting point for RDC-based
refinement to the final fold [
356
]. Having relatively few representatives in the PDB,
a more useful application for membrane proteins is the identification of structurally
homologous protein fragments from the structure database [
357
] or the structure
modeling program Rosetta [
358
] to generate an initial model for refinement. This
approach works well for smaller proteins having around four measured couplings
per peptide plane, although ambiguities can arise with multiple structures being
equally compatible with the RDC data for some fragments. Hence for larger
proteins, it is necessary to supplement RDC data with distance restraints provided
by NOEs [
358
] or paramagnetic relaxation enhancement (PRE) data [
41
,
351
]to
resolve these ambiguities.
The power of this fragment-replacement approach was recently demonstrated
for the 6-TM helix UCP2 protein, for which an average of 2.2 RDCs per residue
could be acquired (Fig.
6d
)[
41
]. These RDCs were used to determine local and
secondary structures using a PDB-based fragment database. This resulted in 15
continuous segments of UCP2 with local backbone structures that could be highly-
restrained during structure calculation. RDC input at this stage helped to determine
relative segment orientations, and PREs provided information on their relative
positions. Although no structural information on side chains are obtained in this
a
