Biomedical Engineering Reference
In-Depth Information
CS-ROSETTA has been found to dramatically improve convergence on the
correct structure for proteins up to 170 residues in size, while a further
refinement step that includes information from backbone
1
H
N
-
1
H
N
NOE
restraints has had success with proteins containing up to 200 residues.
139
This
approach could be extended further by supplying additional sparse data, such
as PREs or NOEs between selectively protonated methyl groups, but systems
that contain a mixture of structured and flexible regions will continue to be
problematic.
3.9 Predicting Quaternary Structure From Chemical
Shifts
The environmental changes that occur on formation of an interface make
chemical shift perturbations highly sensitive probes of protein-protein
assembly.
160,161
A typical experiment for monitoring the formation of a
complex is to follow the chemical shifts of resonances in the [
1
H,
15
N]-HSQC
spectrum of a uniformly
15
N-labelled protein as an unlabelled interaction
partner is titrated into the sample. Shift changes at
1
H
N
sites are mainly
determined by changes in hydrogen bonding, whereas
15
N sites are
predominantly affected by conformational alterations, so for these backbone
nuclei, shift differences between the bound and unbound states are regularly
combined into a single parameter Dd
comb
. There is no clear consensus as to
how this quantity should be calculated, although it is commonly expressed as
either a vector length:
Dd
comb
~
X
n
i~1
h
Þ
2
.
n
i
1
=
2
ð
w
i
Dd
i
,
ð
3
:
6
Þ
or a Hamming distance:
Dd
comb
~
X
n
i~1
n
o.
n,
j
w
i
Dd
i
j
ð
3
:
7
Þ
where Dd
i
is the chemical shift difference in ppm for nucleus i out of a total n
atom types under consideration (typically selected from the
1
H
N
,
15
N,
13
C
a
and
13
C9 resonances of a residue) and w
i
is a weighting factor.
162
The weighting
factors w
i
reflect the different sensitivities of atom types to structural change
and have been expressed in terms of ratios of: the magnetogyric ratio, c
i
,of
nucleus i with respect to that of nucleus 1 (i.e., 268 6 10
6
rad s
-1
T
-1
for
1
H
N
and 27 6 10
6
rad s
-1
T
-1
for
15
N);
163
expected atom-specific chemical shift
ranges (e.g., 5.5 ppm for
1
H
N
and 32 ppm for
15
N);
164
or average variances of
shifts deposited in the BMRB for different atom types (e.g., 0.66 ppm for
1
H
N
and 4.3 ppm for
15
N).
165
After determining chemical shift perturbation (CSP) values for resonances
from the labelled protein, the next step is to select which are significant and
