Chemistry Reference
In-Depth Information
of broadening will be underestimated if the labeling reaction is incomplete. In
addition, the diamagnetic reference spectrummust not contain any residual paramag-
netic contributions. These can be significant when water-soluble reducing agents such
as ascorbic acid are used to reduce MTSL, particularly since the membrane-mimetic
environment may shield the radical center. This was observed in the case of OmpA,
with EPR spectra suggesting that only ~90% of the paramagnetic center was reduced,
in spite of attempts to achieve higher levels of completion with a wide range of
reducing agents [
367
]. To avoid this problem it is possible to prepare a separate
sample labeled with a diamagnetic analog of MTSL that contains an acetyl group in
place of the oxygen radical, allowing a purely diamagnetic reference spectrum to be
acquired [
367
]. Similarly, the reference spectrum for PRE measurements with metal
chelating tags is usually obtained with a diamagnetic ion (e.g., Ca
2+
,Mg
2+
)being
bound in place of the paramagnetic metal [
327
,
329
,
330
].
If care is taken to ensure that samples are completely labeled and that an
appropriate diamagnetic reference is used, then it is possible to generate accurate
distance restraints from these PRE measurements. For example, residues showing
paramagnetic:diamagnetic peak intensity ratios (
I
para
/
I
dia
) of 15-85% in MTSL-
labeled OmpA had PRE-based distance measurements that were within 2
˚
of the
corresponding distances determined from the crystal structure [
367
]. This allows
relatively tight restraint bounds to be applied during structure calculations, with
2
˚
for the 15-24
˚
distance range, and an upper limit of 15
˚
being applied for
residues showing intensity ratios under 15%. It is also possible to impose lower
distance limits of 20-22
˚
85%), although
these did not improve the quality of the OmpA structure [
367
]. This was the
approach used in the structure determination of the 4-TM helix DsbB, which
along with RDCs and a handful of long-range NOEs, generated a highly precise
ensemble (pairwise backbone rmsd of 0.80
˚
)[
39
]. A comparable level of precision
was also achieved for Rv1761, which used slightly looser bounds of
for residues showing small PREs (
>
3
˚
, since the
narrower bounds gave rise to distance violations [
159
]. More conservative bounds
(e.g.
4
˚
[
67
,
351
,
360
] or greater [
41
]) or NOE-like assignments into distance
bins (i.e., strong, medium weak PREs) have also been used [
88
,
368
], although this
does reduce the impact of the PRE on the structure [
367
].
An important consideration when converting PREs into distances is the influence
of spin-label and backbone dynamics. Specifically, the
r
6
dependence causes the
measured PRE for a particular
1
H atom to be dominated by conformers having
shorter distances to the spin label, even if this state is infrequently sampled [
363
].
Consequently, PREs involving flexible regions of a protein cannot be accurately
converted into a single average distance. Since dynamic regions of the protein can
be identified with standard backbone relaxation measurements, PREs involving
these residues should be discarded or only used with very loose restraint bounds.
Less straightforward is the treatment of spin labels that are bound to the protein
via a flexible linker, where exchange between various rotameric states can occur.
To account for the different contributions to the PRE that can arise from this type of
exchange, an ensemble approach was developed that refines the structures against
PREs in place of PRE-derived distances [
363
], improving the accuracy of structures
