Chemistry Reference
In-Depth Information
increasing rotational correlation time of the protein. As a consequence, the experiment will
becomemore sensitivewith increasingmolecular weight of the target protein. The saturation
of the protein will then be transferred, via intermolecular
1
H-
1
H cross-relaxation, to weakly
binding ligands at the ligand-protein interface. After the ligand has dissociated, the long
spin-lattice relaxation times (
T
1
) of the free ligand protons ensure that it is possible to
detect the transferred saturation as an attenuation of the ligand signals. During the time of
the saturation, more unsaturated ligand molecules will bind and dissociate, leading to an
increasing population of saturated ligands. The saturated spectrum is then subtracted from
a spectrum obtained under nonsaturating conditions to obtain an STD spectrum showing
only the signals from compounds binding reversibly to the target protein; see Figure 4.3
for an example. This simplicity of the resulting spectrum is a very attractive feature of
the STD experiment. In practice, the pulse sequence is written in such a way that the
subtraction is performed automatically in every other scan, i.e. the individual spectra are
never observed.
The STD experiment allows for very high ligand:protein ratios to be used, often as high
as 50 or 100. Thus, the protein concentration can be kept very low, in fragment screening it
is possible to use less than 1 M target protein in favorable cases. The protein signals are
not visible in the STD spectrum due to the low protein concentration and/or the presence of
a spinlock in the pulse sequence. The sensitivity of the method depends critically on how
efficiently the protein resonances have been saturated.
[
92
]
The duration of the saturating
selective pulse train is typically 1-4 s, where the longer saturation times in this interval
can be used for smaller proteins. The upper limit of the saturation time for an efficient
saturation transfer is determined by the
T
1
of the ligand in the bound state. Aspectral region
suitable for efficient saturation is the protein methyl groups which usually has a maximum
intensity close to 0.7 ppm. In order to avoid direct saturation of ligands, the saturation
is usually applied further upfield where most target proteins have upfield shifted methyl
resonances. The STD experiment is preferably performed in samples containing 100%
D
2
O. This makes optimal water suppression less crucial and minimizes exchange-mediated
saturation leakage via dipole-dipole interactions between saturated protein protons and
hydration water molecules.
[
93, 94
]
The intensity of the STD signals contains information on the ligand affinity and it should
in principle be possible to rank fragment hits directly with respect to affinity. There are,
however, several caveats to be aware of and caution should be exercised when interpreting
data. First, the individual STD peak heights should be normalized to the corresponding
peak heights in a reference spectrum. This could either be the spectrum obtained under
nonsaturating conditions (collected as part of the STD experiment) or a T
2
-filtered
1
H1D
spectrum of the same sample, using the same length of the spinlock and interscan delay as in
the STD experiment. The parameters in the STD experiments, such as saturation frequency,
saturation time and relaxation delay, should be identical for all samples. For the normalized
STD response to be directly related to the relative affinity of different ligands to the target
protein, there are a number of assumptions that should be fulfilled: the contribution to the
STD signals should come from binding to a single site only, the exchange must be fast
(should be fulfilled for weakly binding fragments) and the
T
1
values for the proton signals
to be compared must be similar. The last condition is important since the STD responses
are highly dependent on the
T
1
value for the observed proton
[
94, 95
]
so that, for example, a
smaller
T
1
value would result in a smaller STD response.
