Chemistry Reference
In-Depth Information
These studies lay the groundwork for applying NOE matching to large proteins of high
therapeutic interest.
We have recently undertaken NMR studies of several kinase-inhibitor complexes.
Neither the kinase (which we subsequently refer to as 'kinaseX') nor the exact chem-
ical structures of these inhibitors can be revealed at this time. The inhibitors all contain a
heterocyclic core that is expected to bind to the 'hinge' region of kinaseX by accepting a
hydrogen bond from a backbone HN, a basic aliphatic moiety that is expected to bind in
the general location where the ribose and phosphates of ADP/ATP bind, and an aromatic
substituent linked to the heterocyclic core. High-sensitivity standard 2D
1
H-
1
H NOESY
and TOCSY spectra were obtained in an inhibitor complexed to a uniformly
2
H-labeled
kinaseX using
2
H-labeled buffer components, as predicted previously.
[
42
]
These spectra
afforded the
1
H NMR assignments for kinaseX inhibitors and allowed the identification of
several intermolecular NOE contacts as outlined below.
TheADPbinding pocket of kinaseXcontains two valines and three leucines.We produced
a sample of kinaseX that incorporated [
1
H]Leu into an otherwise fully deuterated protein
and a second sample that incorporated [
1
H]Val into an otherwise fully deuterated protein.
{To prevent unwanted labeling of an amino acid in the biosynthetic pathway of desired
amino acid, we supplement the growth media during induction with the undesired amino
acid(s) that are
2
H-labeled; e.g. [
2
H]Val was added to samples incorporating [
1
H]Leu.}
NOESY spectra of an inhibitor ('kinaseX inhibitor 1') in complex with kinaseX were
recorded at 15 °C using these kinaseX samples (Figure 5.19). The heterocyclic core of
kinaseX inhibitor 1 has aromatic
1
H resonances at 8.91 and 6.44 ppm and the aromatic
substituent has aromatic
1
H resonances at 7.01 and 6.78 ppm. The inhibitor also has
aliphatic
1
H resonances at 1.82 and 1.48 ppm. The heterocyclic core aromatic reson-
ances give rise to NOEs of varying intensities to at least two Leu residues, whereas the
aromatic substituent yields only one very weak (tentative) NOE involving a leucine at
F
1
7.01 ppm (Figure 5.19A). The heterocyclic core has intense NOEs to
valine resonances at 1.65 and 1.41 ppm and weak NOEs to a valine resonance at 2.10 ppm
(Figure 5.19B). The resonances of the aromatic substituent at 7.01 and 6.78 ppm give rise
to medium- and strong-intensity NOEs, respectively, involving a Val resonance at 0.14 ppm
(Figure 5.19B).
KinaseX samples that are residue type-specifically labeled with [
1
H]Thr, [
1
H]Lys and
[
1
H]Met have been also produced and NOEs between these residues and inhibitors have
been observed (data not shown). Since there is one threonine, one lysine and one methionine
in the ADP binding pocket of kinaseX, sequence-specific assignments for these residues
can be obtained directly by the observation of protein-inhibitor NOEs.
A cautionary note must be provided for using peaks from type-specifically labeled
samples and merging peak lists from different spectra. An implicit assumption made when
calibrating peaks from
uniformly
labeled samples is that the strongest NOE cross peaks
correspond to distances approximating the van der Waals radii and the weakest NOE cross
peaks correspond to distances in the range 5-5.5 Å. This assumption no longer holds true for
spectra acquired in type-specific labeled proteins. In such spectra, distances corresponding
to the strongest observed NOEs can be in excess of 5 Å and those for the weakest observed
NOEs can be in excess of 9 Å. Hence it is essential for the NOEs from type-specifically
labeled samples to be properly scaled before translating them into distances. This can
be achieved by comparison with intra-ligand NOEs (2D NOEs or double reverse filtered
=
0.82 ppm,
F
2
=
