Biomedical Engineering Reference
In-Depth Information
13
C and
15
N nuclei are 25.1449530 MHz and 10.1329118 MHz, corresponding
to
signals
from
the
reference
compounds
DSS
and
liquid
ammonia,
respectively.
14
Since 2010, it has been mandatory for research groups depositing protein
solution structures in the worldwide Protein Data Bank (PDB) to place
accompanying chemical shift assignments and referencing information in the
BioMagResBank (BMRB) data repository.
21,22
At the close of 2011, the
BMRB contained in excess of 4 800 000 assigned isotropic chemical shifts from
more than 7000 proteins and peptides, along with associated meta-data,
including details of pH, temperature, co-solutes, magnetic field strength,
distance and dihedral angle restraints, residual dipolar coupling (RDC)
measurements, etc.
23,24
The BMRB is an open-access archival database: the
entries are systematically validated,
25
but cannot be guaranteed to be error-free
because changes can only be made with the consent of depositors. Accordingly,
recent surveys suggest that 20-35% of BMRB entries may contain mis-
referenced
1
H
N
,
13
Cor
15
N chemical shift data, while others possess incorrect
atom names or syntax.
19,26-28
It is therefore not surprising that some
researchers have compiled custom databases in which all accepted entries
employ common referencing criteria.
19,29-33
The most established of these is
the Re-referenced Protein Chemical Shift Database (RefDB), curated by the
Wishart
group
and
containing
data
for
more
than
2100
proteins,
all
computationally re-referenced to DSS.
19
Strategies for re-referencing previously acquired chemical shift assignments
are either based on intrinsic properties of the data or deduced from the 3D
structure coordinates of the target protein. The LACS method devised by
Wang and Markley uses correlations between the secondary chemical shifts of
1
H
a
,
13
C
a
,
13
C
b
,
13
C9 and
15
N sites and a reference-independent variable,
Dd(
13
C
b
)-Dd(
13
C
a
), to derive offset corrections for each nucleus type.
26,34
The
CheckShift,
35
PANAV,
27
PSSI
36
and SSP
37
approaches rely on chemical shift
or sequence-based algorithms to predict the location of elements of secondary
structure in the target protein and then iteratively adjust
13
Cand
15
N offsets
until the calculated secondary shifts match reference values. In contrast, the
SHIFTCOR procedure employed to prepare data for inclusion in the RefDB
database aims to minimise the deviation between experimental
1
H,
13
C and
15
N
shifts and values that have been back-predicted from the protein structure.
19
Finally, VASCO makes offset corrections by obtaining secondary structure
and solvent accessibility information from the structure of the target protein so
that the shifts of nuclei in different environments can be compared with
reference distributions harvested from statistical analysis of a large set of
chemical shift and coordinate data.
33
Databases compiled from re-referenced
assignment lists are more likely to yield consistent results when interrogated
for correlations between chemical shifts and protein structure.
Distributions of experimental chemical shifts provide useful information for
assigning protein resonances. For example, the lysine
1
H shifts displayed in
Figure 3.2 can guide users towards accurate side-chain assignments by helping
