Chemistry Reference
In-Depth Information
9000
8000
7000
6000
5000
4000
3000
0
1
2
3
RMSD to Target
4
5
6
Figure 5.10
COST versus the RMSD (Å) to the target pose for PDF/
6
. The predicted protein
proton chemical shifts were set to the values determined using SHIFTX.
SHIFTX chemical shift prediction provided a small improvement in the NOE matching
scoring. In general, for lower COST poses, poses that were more dissimilar to the target
pose (that is, decoy poses) were now scored with a higher relative COST than poses that
were more similar to the target pose. For example, the COST of poses with RMSDs of 0.96,
1.12, 2.35 and 4.09 Å from the target pose increased by 5.7, 5.4, 7.6 and 8.5%, respectively.
By improving the COST difference between correct and decoy poses, incorporation of
SHIFTX-predicted shifts into theNOEmatching protocol leads to higher confidence results.
We expect that, in general, as chemical shift prediction tools improve, so will the results
obtained from NOE matching. Rapid empirical predictions of ligand-induced chemical
shift changes for protein resonances could benefit NOE matching (e.g. see ref. 31). Also,
accurate quantum mechanics-based predictions of binding-induced chemical shift changes
for ligand
[
32
]
and/or protein resonances could be applied to evaluate further a small number
of selected poses.
5.5.6 Bcl-x
L
with Experimental Data
Recently, a potent inhibitor of Bcl-x
L
was discovered after applying the 'SAR by NMR'
technique to an initial hit from an NMR-based screen.
[
14
]
We applied NOE matching to one
of the commercially available initial hits [4-fluorobiphenyl-4-carboxylic acid 7,
K
d
=
300
±
30 μM].
O
F
OH
7
Compound 7
The deposited coordinates of the NMR structure of this fragment in complex with Bcl-
x
L
(PDB entry 1YSG)
[
14
]
were used as the target pose. Trial poses were generated using
