Chemistry Reference
In-Depth Information
and decoy poses. To determine whether the 3D X-filtered NOESY experiment contains
enough information to enable NOE matching to identify the correct binding pose, we ran
tests using simulated data derived from a CDK2/
4
complex, an FKBP-12/
5
complex and a
peptide deformylase (PDF)/
6
complex. The compounds used for these test cases are shown
in Scheme 5.2.
N
O
O
Cl
N
N
HO
O
N
N
NH
O
OH
4
5
6
Scheme 5.2
Structures of compounds used for test cases.
5.5.1 CDK2
Compound
4
is anATP-mimic that binds in the active site of the catalytic domain of CDK2
kinase. The crystal structure of CDK2 in complex with
4
(1ckp
26
) served as the 'target pose'
for the NOE matching simulations. To generate the required input files for NOE matching,
a list of CDK2/
4
NOEs was derived from distances calculated using the CDK2/
4
complex
using a distance cutoff of 5 Å and using the BMRB average chemical shifts for the simulated
'experimental' chemical shifts. The simulated NOE list for this complex contained a total
of 69 peaks, which were clustered into 43 protein
1
H
13
C groups. Trial binding poses were
generated with
Poser
. The compound binding site was defined as the active site of the
protein; the 'posing box' was expanded by1Åinallcoordinate axes. For each of the 13
ligand conformers (generated with Omega), over 808 million poses were generated and
evaluated by
Poser
, with 10 579 poses being retained. The RMSDs of the retained trial
poses to the target pose ranged from 0.69 to 7.85 Å. The NOE matching protocol was run
using BMRB-predicted chemical shifts.
The results obtained from applying NOE matching to CDK2/
4
are shown in Figure 5.6.
The pose with the minimum COST value has an RMSD of 0.74 Å to the target pose. The
pose with the closest RMSD to the target pose itself ranks 14 out of 10 579 poses.
5.5.2 FKBP-12
Compound
5
is one of numerous fragments identified in an in-house NMR screen of
FKBP-12. The solution structure of FKBP-12 in complex with
5
was determined by
restrained simulated annealing
[
27
]
using data derived from standard three-dimensional NMR
techniques. The average structure of the resultant ensemble of NMR structures was calcu-
lated and subjected to unrestrained energy minimization; this structure served as the 'target
pose' for the NOE matching simulations. The experimentally determined resonance assign-
ments were used for the ligand resonance assignments. To generate the required input files
for NOE matching, a list of FKBP-12/
5
NOEs was derived from distances calculated using
the FKBP-12/
5
complex using a distance cutoff of 5.0 Å (similar to what we observed in
our experimental NOESY spectrum) and using the BMRB
21
average chemical shifts for
