Biomedical Engineering Reference
In-Depth Information
calculated in a straightforward fashion, since residual interactions with
neighbouring residues and transient hydrogen bonding, ring-current shift
and solvent effects come built in as part of the package.
45
For example, by
comparing
13
C
a
shifts in the presence and absence of 5 M urea intrinsic
referencing was shown to provide a sensitive probe of residual secondary
structure in the pH 2.3 acid-denatured state of an acyl-coenzyme A binding
protein.
46
More often, polypeptides cannot be fully denatured, so approximations to a
reference state are required. Two parallel approaches for compiling libraries of
reference shifts have evolved: experiments on short peptides or statistical
analysis of chemical shift databases. In solution, a short peptide samples a wide
range of conformations and is therefore proposed to mimic a time- and
ensemble-averaged random coil state. Experiments on pentapeptide Ac-
GGXGG-NH
2
and hexapeptide Ac-GGXYGG-NH
2
samples,
47,48
protected
with acetyl and amide groups at their N- and C-termini to avoid end charge
effects, have been carried out under controlled conditions to monitor the
effects of pH and temperature,
49
the chemical denaturants urea
47,50
and
guanidinium hydrochloride,
51
and co-solvents such as acetonitrile,
52
trifluor-
oethanol,
53
dimethyl sulphoxide, chloroform and methanol,
54
on
1
H
a
,
13
C
a
,
13
C
b
,
13
C9,
1
H
N
and
15
N chemical shifts.
13
C random coil shifts vary little with
pH for most residue types except for Asp, Glu and His, which can change their
side-chain protonation states under conditions typically encountered in protein
NMR (pH 4 to 8).
49
Another way of obtaining an average over multiple conformations is to
compare the native state chemical shifts of different proteins studied under a
variety of experimental conditions.
55
Initial attempts used secondary structure
identification software to classify each residue in the RefDB database as helix,
strand or coil and then determined mean values and standard deviations for the
shifts of coil residues.
19
A refined reference set can be accessed via the CamCoils
server, derived from analysis of a larger collection of chemical shift and
coordinate data coupled with more stringent identification of flexible loop
regions in globular proteins.
56
Alternatively, the ncIDP library discards
information about folded proteins altogether, instead compiling a random coil
shift set from a database of 14 polypeptides independently shown to be
unstructured.
57
Database approaches are less likely to contain specific artefacts
that could result from a particular choice of peptide model, but typically collate
measurements from a range of experimental conditions and so might miss subtle
effects, e.g., those related to the charge state of titrating side-chains.
49,58
Database analysis and peptide experiments both indicate that random coil
chemical shifts are affected by the sequence of neighbouring amino acid
residues.
48,56,57,59,60
Schwarzinger and co-workers adjusted for these local
effects by comparing shift measurements for Ac-GGXGG-NH
2
pentapeptides
with those for Ac-GGGGG-NH
2
, assuming that the sequence-dependent
correction factors for i ¡ 1 and i ¡ 2 sites obtained for glycine would be the
same
types.
61
for
other
residue
Nitrogen-15
shifts
were
shown
to
be
