Chemistry Reference
In-Depth Information
acquisition [
40
,
145
]. Specifically, many of the 3D spectra were acquired using non-
uniform sampling in all indirectly detected dimensions [
300
]. In this method, data is
only acquired for a subset of the incremented evolution periods, with more frequent
sampling in regions of short evolution times that have larger signal-to-noise ratios
[
301
]. The emphasis on data acquisition in parts of the time domain having the
greatest signal intensity leads to increases in sensitivity per unit time over conven-
tional uniformly sampled experiments [
302
]. Maximum entropy reconstruction was
used to process this sparsely sampled dataset, producing multidimensional spectra
that retained the resolution of the conventionally sampled experiment in spite of the
reduced number of indirect points [
303
,
304
]. In the case of pSRII, the signal-to-
noise ratio was further enhanced by adding multiple datasets acquired on one or
more samples [
40
], a useful approach when sample stability limits the time that can
be used to acquire a single dataset.
Non-uniform sampling (NUS) also opens the door to experiments of increased
dimensionality that might normally require several months of acquisition time to
obtain adequate resolution in the indirectly detected domains [
302
,
305
]. This
strategy was used to address the problem of chemical shift degeneracy in methyl
NOESY data for the 283-residue VDAC
-barrel (Fig.
6d
)[
73
]. Four-dimensional
spectra were acquired on this sample to correlate
1
H shifts with the shift of the
directly attached heteroatom on both sides of the NOE interaction [
73
,
306
]. These
spectra were processed using multidimensional decomposition (MDD) [
307
,
308
],
an approach that is particularly well-suited for the accurate reproduction of signal
intensities in spectra having a wide range of peak intensities, as is seen in NOESY
spectra [
309
]. Moreover, significant sensitivity increases for weak peaks were
realized by combining multiple datasets during MDD. This technique, called
coupled MDD, was used to process a
13
C-
13
C separated 4D [
1
H-
1
H] NUS-
NOESY by co-processing with a
1
H-
13
C HMQC template spectrum [
310
]. Coupled
MDD allowed NOEs involving the 15% of VDAC methyl groups that did not
appear in the MDD-processed spectrum to be detected and assigned, many of which
were critical for determination of the global fold.
Overall, the application of non-linear sampling and other random sampling
techniques to membrane proteins is still in its early days, and does not reflect the
diversity of sparse sampling techniques that have been developed to date (reviewed
in [
302
,
305
]). Nonetheless these studies demonstrate the significant advantages
that can be gained by applying these data acquisition strategies to challenging
membrane protein samples, which should motivate an increasing number of
research groups to adopt this technology.
b
5.3 Global Folds in the Absence of Long-Range NOEs
Even when sample labeling and NMR experiments are optimized to minimize
unfavorable relaxation processes and increase sensitivity, it is not unusual for
difficulties in side chain and NOE assignment to preclude an NOE-driven approach
