Biomedical Engineering Reference
In-Depth Information
exploited by TALOS+
91
to yield angular restraints or for selection of accurate
local structure (fragments) by screening a library of protein fragments against
chemical shift data (CS-ROSETTA, CHESHIRE).
22,23
The full set of
backbone chemical shifts is somewhat redundant; CS-ROSETTA calculations
where up to two types of chemical shifts are omitted yield identical resultsas
for the full set of chemical shifts. Adequate results are even obtained if only C
a
and C
b
or only H
N
, N and C
a
chemical shifts are used.
92
The order of
importance of each atom type for de novo calculations can be ranked as C
a
y
C
b
. H
a
y C . N y H
N
.
92
Ligand binding causes chemical shift perturbations (CSPs) of interface
residues. Obtaining chemical shifts in bound and unbound structures yields
highly ambiguous and low-resolution distance restraints that are used in
HADDOCK to guide docking calculations.
93,94
4.3.5.1 Advantages
Chemical shift assignments are used to assign peaks to specific atoms in the
biomolecule and are, thus, a prerequisite to interpret any of the other NMR
data (NOE, RDC, PRE etc.) as structural restraints. Accordingly, this data is
the first to measure and assign and thus readily available. Relaxation
dispersion experiments allow the determination of chemical shifts of otherwise
inaccessible lowly populated (hidden) conformational states if they are in
conformational exchange on the millisecond timescale.
7,8,95
Motional averaging for PRE and NOE data goes with r
26
such that short
distances of minor conformations in fast exchange can dominate results
leading to structural distortions. In the presence of multiple conformations in
fast exchange the measured chemical shift is a weighted sum of the chemical
shifts of the individual conformations. Thus, minor conformations will not
dominate the result, rendering chemical shift-derived restraints less prone to
distorting the structure.
4.3.5.2 Disadvantages
It is challenging to interpret chemical shift data structurally. A prerequisite for
such an interpretation is a fast and reliable method to back-calculate chemical
shifts from structural models. Currently such methods obtain chemical shifts
with ca. 2.5 ppm and 0.5 ppm error on the N and HN chemical shift,
respectively.
84,86
The SHIFTX2 method uses homology information to reach
significantly lower values,
84
but since the improvement stems from copying
chemical shifts from homologous proteins this reduction in prediction error
will not impact on structure calculations. To put the 2.5 ppm error into
perspective, note that for proteins undergoing conformational changes, such as
Hsp70
96
or PKA,
97,98
the largest change of chemical shift between different
conformational states just about reaches the 2-3 ppm range.
96-98
