Protein Structure and Dynamics by NMR in Solution
99
3.3.1
Nuclear Overhauser Effect
The nuclear Overhauser effect (nOe) is due to the exchange of magnetization
between two spins (cross-relaxation) and can be detected as the change in the
intensity of an NMR resonance (I) when another spin (S) is irradiated. A requirement for the nOe is that the spin-lattice relaxation of nucleus I is governed by
dipole-dipole relaxation with S. Apart from its ability to enhance signals of low
sensitivity nuclei eSN, l3C ... ), nOe is widely used to obtain information about
internuclear distances, essentially between IH. The nOe has both an interproton
distance dependence (lfr 6 -only short distances < 6 A are in practice detectable)
and an effective correlation time dependence (Neuhaus and Willamson 1989).
For the determination of internuclear distances, the latter dependence is often
neglected by assuming that it arises from the tumbling of a rigid, approximately
spherical molecule. Various complexities in interpretation of nOe data support its
rather qualitative use: J-couplings give rise to artefacts, multiple transfers, known
as spin diffusion, (I ~ M ~ S) occur at longer mixing times, conformational flexibility can lead to averaging of nOe and also, a proton which oscillates between
two sites, may show incompatible nOe with very distant (> 10 A) spins. As a
result, for protein structure determinations, spectroscopists do not longer try to
quantify each nOe precisely, but aim at identifying as many cross-peaks as possible.
On proteins, nOe are measured not by saturation of individual signals but by
means of a NOESY 2D experiment: a cross-peak is the evidence that two spins
have exchanged magnetization during a mixing period of typically 50-150 ms.
For larger lSN/ l3 C labeled proteins, the NOESY scheme can be combined with an
IH_lSN/l3C correlation experiment to edit the heteronuclear frequency in a third
dimension (Marion et al1989) and thus reduce spectral overlap.
Isotope labeling also turns out to be useful for structure determination of
dimers where it is intrinsically impossible to distinguish intermolecular nOe
from intramolecular ones. Although distance violations due to incorrect assignment may probably become patent during the structure refinement, it is more
convenient to rely on experimental discrimination by using asymmetric labeling
(Folkers et al. 1993). On a 1:1 mixture containing unlabelled and ISN/l3C-Iabeled
protein, nOe at the dimer interface can be discriminated by heteronuclear filtered
or edited experiments (see Handel and Domaille 1996 for an application).
3.3.2
Heteronuclear Relaxation Rate Constants
The function of a protein depends on its ability to explore excited states and is
hence intimately coupled to flexibility. A complete description of a structure of a
protein will require information on how the structure changes with time. Long
before labeled proteins became available, l3C and ISN relaxation experiments
were devised to study the dynamics of small molecules. Investigations of protein
dynamics most commonly measure the longitudinal (TI) and transverse (T2)
relaxation time constants as well as the steady state eH}X nOe (or cross-
99
3.3.1
Nuclear Overhauser Effect
The nuclear Overhauser effect (nOe) is due to the exchange of magnetization
between two spins (cross-relaxation) and can be detected as the change in the
intensity of an NMR resonance (I) when another spin (S) is irradiated. A requirement for the nOe is that the spin-lattice relaxation of nucleus I is governed by
dipole-dipole relaxation with S. Apart from its ability to enhance signals of low
sensitivity nuclei eSN, l3C ... ), nOe is widely used to obtain information about
internuclear distances, essentially between IH. The nOe has both an interproton
distance dependence (lfr 6 -only short distances < 6 A are in practice detectable)
and an effective correlation time dependence (Neuhaus and Willamson 1989).
For the determination of internuclear distances, the latter dependence is often
neglected by assuming that it arises from the tumbling of a rigid, approximately
spherical molecule. Various complexities in interpretation of nOe data support its
rather qualitative use: J-couplings give rise to artefacts, multiple transfers, known
as spin diffusion, (I ~ M ~ S) occur at longer mixing times, conformational flexibility can lead to averaging of nOe and also, a proton which oscillates between
two sites, may show incompatible nOe with very distant (> 10 A) spins. As a
result, for protein structure determinations, spectroscopists do not longer try to
quantify each nOe precisely, but aim at identifying as many cross-peaks as possible.
On proteins, nOe are measured not by saturation of individual signals but by
means of a NOESY 2D experiment: a cross-peak is the evidence that two spins
have exchanged magnetization during a mixing period of typically 50-150 ms.
For larger lSN/ l3 C labeled proteins, the NOESY scheme can be combined with an
IH_lSN/l3C correlation experiment to edit the heteronuclear frequency in a third
dimension (Marion et al1989) and thus reduce spectral overlap.
Isotope labeling also turns out to be useful for structure determination of
dimers where it is intrinsically impossible to distinguish intermolecular nOe
from intramolecular ones. Although distance violations due to incorrect assignment may probably become patent during the structure refinement, it is more
convenient to rely on experimental discrimination by using asymmetric labeling
(Folkers et al. 1993). On a 1:1 mixture containing unlabelled and ISN/l3C-Iabeled
protein, nOe at the dimer interface can be discriminated by heteronuclear filtered
or edited experiments (see Handel and Domaille 1996 for an application).
3.3.2
Heteronuclear Relaxation Rate Constants
The function of a protein depends on its ability to explore excited states and is
hence intimately coupled to flexibility. A complete description of a structure of a
protein will require information on how the structure changes with time. Long
before labeled proteins became available, l3C and ISN relaxation experiments
were devised to study the dynamics of small molecules. Investigations of protein
dynamics most commonly measure the longitudinal (TI) and transverse (T2)
relaxation time constants as well as the steady state eH}X nOe (or cross-