LuxR-dependent quorum sensing: Computer aided discovery of new
inhibitors structurally unrelated to N-acylhomoserine lactones
Laurent Soulère
*
, Mohamad Sabbah, Fanny Fontaine, Yves Queneau, Alain Doutheau
INSA Lyon, Institut de Chimie et Biochimie Moléculaires et Supramoléculaires, Laboratoire de Chimie Organique, Bât J. Verne, 20 av A. Einstein, 69621 Villeurbanne Cedex, France
CNRS, UMR 5246 ICBMS, Université Lyon 1, INSA-Lyon, CPE-Lyon, Bât CPE, 43 bd du 11 novembre 1918, 69622 Villeurbanne Cedex, France
article info
Article history:
Received 11 May 2010
Revised 11 June 2010
Accepted 12 June 2010
Available online 17 June 2010
Keywords:
Virtual screening
Molecular docking
Quorum sensing
Antagonist
LuxR
AHL
abstract
A virtual screening, involving flexible docking sequences within the LuxR, TraR and LasR binding sites,
was used as a structural binding sites similarity filter to specifically target conserved residues in the pro-
teins of the LuxR family (namely Tyr62, Trp66, Tyr70, Asp79, Trp94 for LuxR). This docking-based screen-
ing, employing a genetic algorithm, was performed on a 2344 chemical compounds library, together with
empirical binding free energy (
D
G
bind
) calculations. Docking results were analysed, and the compounds
detected with reproducible low
D
G
bind
values or identified as being in the top 120 for most of the docking
sequences, were selected as hits candidates which interact with conserved residues. Biological evaluation
with LuxR-dependent quorum sensing led to the discovery of some new inhibitors, namely tamoxifen,
sertraline, pimethixene, terfenadine, fendiline and calmidazolium. Notably, calmidazolium was identified
as one of the most potent AHL-structurally unrelated inhibitors of LuxR-dependent quorum sensing, with
an IC
50
value of 7.0 ± 0.2
l
M.
Ó 2010 Elsevier Ltd. All rights reserved.
Bacteria are able to communicate with each other using a specific
system called ‘Quorum Sensing’ (QS). This cell to cell communica-
tion system is based on the correlation between population size
and the concentration of small diffusible chemical messengers
called autoinducers. When a critical concentration is reached, target
genes encoding for phenotypes such as bioluminescence, biofilm
formation or virulence are expressed allowing bacteria to adapt
their behaviour to their environment.
1–6
Consequently, QS is now
targeted for the design of potential antibacterial drugs.
7–10
In Gram
negative bacteria, quorum sensing involves N-acyl homoserine lac-
tones (AHLs) as autoinducers, associated with transcriptional regu-
lators belonging to proteins of the LuxR family. Most studies
describing the discovery of LuxR-protein antagonists are based on
the rational design of AHL analogues or on biological screenings of
compound libraries.
11
To date, in silico screening of ligand dat-
abases, which has become a key methodology in drug discov-
ery,
12–15
has been only moderately employed to discover new QS
inhibitors,
16–18
especially using LuxR-type proteins. In 2008, Zeng
et al. reported the virtual screening of active compounds from Tradi-
tional Chinese Medicines with antibacterial activity using docking
simulations and the TraR protein.
19
More recently, Yang et al.
described an interesting study involving docking-based virtual
screening, within the binding site of LasR, using structural alignment
with the natural ligand, 3-oxo-C
12
-HSL which led to the discovery
of new antagonists, namely salicylic acid, nifuroxazide, and
chlorzoxazone.
20
As a part of our research programme, aimed at discovering
potential modulators of the AHLs-dependent transcriptional regu-
lators,
21–23
we describe here a virtual screening using a genetic
algorithm and involving flexible docking sequences within the
LuxR, TraR and LasR binding sites. This protocol was used as a
structural binding sites similarity filter to specifically target con-
served residues in the LuxR-proteins. Indeed, as shown in other
studies,
24,25
agonistic or antagonistic activities are related to the
interactions of ligands with the conserved residues (namely
Tyr62, Trp66, Tyr70, Asp79, Trp94 for LuxR). Thus, four docking se-
quences for each protein were performed then analysed in order to
select compounds that target specifically these residues due to
structural homology.
The docking-based virtual screening, employing a genetic algo-
rithm (GADock)
26–28
which randomly generates conformations
within a binding site, was performed with 2344 biologically rele-
vant compounds from the Chembank subset chemical library of
Ligand.Info Meta-Database.
14,29,30
Chembank was, in fact, created
for the discovery of small molecules that modulate specific biolog-
ical pathways using a chemical genetics approach.
31,32
Among the
2344 structures, 2289 compounds
33
were docked within a binding
site and ranked, according to the binding free energy calculations
(
D
G
bind
),
28
resulting in one docking sequence. Subsequently, with
reference to four docking sequences within each binding site of
LuxR, TraR and LasR (12 sequences, 27468 docking experiments),
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.06.081
* Corresponding author. Address: Laboratoire de Chimie Organique, INSA-Lyon,
20 avenue Albert Einstein, Villeurbanne F-69621, France. Tel.: +33 4 72 43 83 42;
fax: +33 4 72 43 88 96.
E-mail address: laurent.soulere@insa-lyon.fr (L. Soulère).
Bioorganic & Medicinal Chemistry Letters 20 (2010) 4355–4358
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl