Chemistry Reference
In-Depth Information
Br
Br
N
Cl
O
Cl
N
O
N
O
H
O
Cl
Cl
O
1
2
3
Scheme 5.1
Structures of compounds used for test cases.
5.3.1 mFABP
In the case of mFABP bound to compound
1
(Scheme 5.1), the 'target pose' was derived
from a high-resolution NMR ensemble (Supporting Information for ref. 15). A total of 461
trial binding poses were generated, with RMSDs to the target pose ranging from 0.20 to
6.66 Å. The experimental 3D X-filtered NOESY spectrum (
τ
m
=
60 ms) for this complex
yielded 140 peaks, which were clustered into 54 protein
1
H
13
C groups. (Although not
discussed further here, we have shown that NOE matching performed well on this system
when up to 80% of the peaks were deleted; see Table S2 in the Supporting Information
for ref. 15).
The NOE matching protocol contains adjustable parameters that define weights for
the observed and predicted NOE intensities (or for functions of differences between
the observed and predicted intensities) and parameters that define weights for functions
of the differences between the observed and predicted chemical shifts. Another para-
meter (the standard deviation multiplier, SDM) is a scaling factor which modulates the
tolerances associated with the predicted chemical shifts. For the tests reported, the weight-
ing parameters were held constant and the SDM parameter was coarsely varied. For
mFABP/
1
using BMRB-predicted chemical shifts, an SDM value of 0.50 was found to
be near optimal (SDM values of 0.25, 0.50 and 0.75 gave very similar results, in general).
Some of the results obtained from applying NOE matching to this system using SDM
=
0.50 are shown in Figure 5.3. The correlation coefficient (
r
) between the COST and
RMSD is 0.912.
The pose with the minimum COST value has an RMSD of 0.75 Å to the target pose. The
target pose itself ranks 14 out of 462 poses. In addition, a majority of the experimentally
identified
1
H
13
C groups are correctly assigned for low COST poses. For mFABP/
1
, NOE
matching worked very well using no protein NMR assignments and the simplest chemical
shift prediction scheme (predictions set to BMRB average shifts).
5.3.2 LFA-1
NOE matching was applied to two different complexes involving LFA-1 (see Scheme
5.1, compounds
2
and
3
). For LFA-1/
2
, an X-ray structure of the complex (Supporting
Information for ref. 15; S. Sheriff, unpublished work) was used as the target pose. A total
