96
B. BERSCH et al.
3.1.1
Chemical Shift and 3D Structure
From the early days of NMR, spectroscopists made a number of attempts at
understanding chemical shifts in proteins in terms of 3D structure. In parallel to
empirical methods discussed later, theoretical approaches have been developed
which steadily increase in speed and accuracy (Oldfield 1995). At present, ab
initio calculations on a complete protein cannot be carried out in an acceptable
amount of time. However, on fragments of two or three amino-acids, excellent
correlation for the , 'Ij1, Xl dependence have been found between predicted and
experimental shifts. So far, these methods are much better at predicting chemical
shift variation than absolute shielding; this is not a severe limitation as experimental data are frequently reported with respect to a reference or a random coil
chemical shift. It is expected that in the near future chemical shift theory will be
able to directly provide additional structural restraints for structure calculations
(Beger and Bolton 1997).
3.1.2
Qualitative Interpretations (Ligand Binding and Chemical Shift Mapping)
Even in absence of detailed understanding of the chemical shift origin, it can be
used qualitatively to monitor the interaction of two molecules. The simplest and
most popular technique is called "chemical shift mapping", wherein changes in
chemical shifts in molecule A are recorded during titration with molecule B.
From the titration curve obtained, a binding constant can be estimated (as for
any other spectroscopic technique) but when resonance assignments are available for molecule A, the location of resonances that undergo perturbation can be
mapped onto its 3D structure. Before interpreting these effects as the putative
binding site, it is preferable to use an independent probe (nOe for instance) to
confirm that no major conformational transition occurs in molecule A due to the
interaction. Otherwise, long range effects can be erroneously interpreted as for
example multiple binding sites. Numerous examples of this technique are available in the literature: they involve protein-protein interactions (see for instance:
Rajagopal et al1997, McKay et al1998), RNA-protein interactions (Lee et a11997)
or peptide bound to protein (Qin et al1995).
3.1.3
Secondary Chemical Shifts
The secondary structure of a protein can often be predicted on the basis of chemical shift information. For the 20 natural amino-acids, the chemical shifts of the
backbone and side-chain nuclei have been characterized for random coil conformation. When a protein folds into its native conformation, a deviation from
random-coil chemical shift, often referred to as secondary chemical shift, is
observed. Spera and Bax (1991) and Wishart et al (1992) have reported empirical
rules for IH and !3e spins. For instance, the H U chemical shift in amino acids
experiences an upfield shift (with respect to the random coil value) in a-helices
B. BERSCH et al.
3.1.1
Chemical Shift and 3D Structure
From the early days of NMR, spectroscopists made a number of attempts at
understanding chemical shifts in proteins in terms of 3D structure. In parallel to
empirical methods discussed later, theoretical approaches have been developed
which steadily increase in speed and accuracy (Oldfield 1995). At present, ab
initio calculations on a complete protein cannot be carried out in an acceptable
amount of time. However, on fragments of two or three amino-acids, excellent
correlation for the , 'Ij1, Xl dependence have been found between predicted and
experimental shifts. So far, these methods are much better at predicting chemical
shift variation than absolute shielding; this is not a severe limitation as experimental data are frequently reported with respect to a reference or a random coil
chemical shift. It is expected that in the near future chemical shift theory will be
able to directly provide additional structural restraints for structure calculations
(Beger and Bolton 1997).
3.1.2
Qualitative Interpretations (Ligand Binding and Chemical Shift Mapping)
Even in absence of detailed understanding of the chemical shift origin, it can be
used qualitatively to monitor the interaction of two molecules. The simplest and
most popular technique is called "chemical shift mapping", wherein changes in
chemical shifts in molecule A are recorded during titration with molecule B.
From the titration curve obtained, a binding constant can be estimated (as for
any other spectroscopic technique) but when resonance assignments are available for molecule A, the location of resonances that undergo perturbation can be
mapped onto its 3D structure. Before interpreting these effects as the putative
binding site, it is preferable to use an independent probe (nOe for instance) to
confirm that no major conformational transition occurs in molecule A due to the
interaction. Otherwise, long range effects can be erroneously interpreted as for
example multiple binding sites. Numerous examples of this technique are available in the literature: they involve protein-protein interactions (see for instance:
Rajagopal et al1997, McKay et al1998), RNA-protein interactions (Lee et a11997)
or peptide bound to protein (Qin et al1995).
3.1.3
Secondary Chemical Shifts
The secondary structure of a protein can often be predicted on the basis of chemical shift information. For the 20 natural amino-acids, the chemical shifts of the
backbone and side-chain nuclei have been characterized for random coil conformation. When a protein folds into its native conformation, a deviation from
random-coil chemical shift, often referred to as secondary chemical shift, is
observed. Spera and Bax (1991) and Wishart et al (1992) have reported empirical
rules for IH and !3e spins. For instance, the H U chemical shift in amino acids
experiences an upfield shift (with respect to the random coil value) in a-helices