Chemistry Reference
In-Depth Information
preferentially bound YndB to infer the natural ligand. OMEGA [
94
] was used to
generate a database of ~10,000,000 conformers from the lipid library. The program
FRED was then used to dock the lipid conformer library to YndB. FRED [
111
] used
rigid docking based on shape complementarity and a consensus scoring system to
rank the ligands. The relative enrichment for each lipid class was calculated at
different thresholds. Only one lipid category, the polyketides, had a positive relative
enrichment, where all of the polyketides identified belonged to the flavonoid class
of lipids. Within the flavonoids, three subclasses emerged as favorable hits from
the virtual screen, where chalcones/hydroxychalcones, flavanones, and flavones/
flavonols accounted for 44.9%, 28.6%, and 14.3% of the top 50 hits, respectively.
trans
-Chalcone, flavanone, flavone, and flavonol were selected to represent each
class. The compounds were titrated into YndB to confirm binding and to measure
K
D
. The titrations were followed using a series of 2D
1
H-
15
N HSQC NMR
experiments, where CSPs were measured to calculate
K
D
s (Fig.
2
).
trans
-Chalcone
(
K
D
<
1
m
M), flavanone (
K
D
32
3
m
M), flavone (
K
D
62
9
m
M), and flavonol
(
K
D
86
M) were all shown to bind YndB in the same ligand binding site with
K
D
s that mimicked the virtual screen ranking. Chalcones and flavonoids have not
been identified among the natural products of
Bacillus
organisms, but are important
precursors to plant antibiotics. The screening results are consistent with the symbi-
otic relationship between
B. subtilis
and plants.
B. subtilis
YndB is proposed to be
part of a stress-response network that senses chalcone-like molecules during a
plant's response to a pathogen infection. The stress-response may induce
B. subtilis
sporulation or the production of antibiotics to assist
16
m
in combating the plant
pathogens.
4.2 Rapid Protein-Ligand Structure Determination
A protein-ligand complex is instrumental to a structure-based approach to drug
discovery. A new protein-ligand structure is required for each iteration of the lead
modification process, until the compound has been evolved into a drug candidate.
As a result, rapid protein-ligand structure determination benefits the drug discovery
process. There are several methods that utilize NMR CSPs from a protein-ligand
binding interaction with molecular docking to generate a corresponding co-structure.
Some recent techniques include the McCoy and Wyss method [
157
], LIGDOCK
[
158
], NMRScore [
159
], AutoDockFilter [
121
], QCSP-Steered Docking [
160
], and
HADDOCK [
44
]. Basically, the CSPs are used to guide the docking process
qualitatively and then to steer or filter the docking quantitatively. The docked
model is validated by an agreement with the experimental CSPs.
AutoDockFilter (ADF) utilizes a post-filtering approach for rapidly
(~35-45 min) generating a co-structure. First, CSPs from the 2D
1
H-
15
N HSQC
spectrum are mapped onto the protein surface to define the AutoDock 4.0 3D search
grid. A 100 docked ligand poses are generated within the CSP defined search grid.
Second, the CSPs are used to filter the ligand conformers and select the best pose
