Discovery of new Gram-negative antivirulence drugs: Structure and properties of novel E. coli WaaC inhibitors

Article abstract of DOI:10.1016/j.bmcl.2008.05.117

Heptosyltransferases such as WaaC represent promising and attractive targets for the discovery of new Gram-negative antibacterial drugs based on antivirulence mechanisms. We report herein our approach to the identification of the first micromolar inhibitors of WaaC and the preliminary SAR generated from this family of 2-aryl-5-methyl-4-(5-aryl-furan-2-yl-methylene)-2,4-dihydro-pyrazol-3-on es identified by virtual screening.

Full text of DOI:10.1016/j.bmcl.2008.05.117

Bioorganic & Medicinal Chemistry Letters 18 (2008) 4022–4026
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of new Gram-negative antivirulence drugs: Structure
and properties of novel E. coli WaaC inhibitors
F. Moreau a, N. Desroy a, J. M. Genevard a, V. Vongsouthi a, V. Gerusz a, G. Le Fralliec a, C. Oliveira a, S. Floquet a,
a
b
b,
A. Denis a, , S. Escaich , K. Wolf , M. Busemann , A. Aschenbrenner b
*
a MUTABILIS SA, 102 Avenue Gaston Roussel, 93230 Romainville, France
b 4SCAG, Planegg-Martinsried, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Heptosyltransferases such as WaaC represent promising and attractive targets for the discovery of new
Gram-negative antibacterial drugs based on antivirulence mechanisms. We report herein our approach
to the identification of the first micromolar inhibitors of WaaC and the preliminary SAR generated from
this family of 2-aryl-5-methyl-4-(5-aryl-furan-2-yl-methylene)-2,4-dihydro-pyrazol-3-ones identified
by virtual screening.
Received 12 May 2008
Revised 30 May 2008
Accepted 31 May 2008
Available online 5 June 2008
Ó 2008 Elsevier Ltd. All rights reserved.
Keywords:
Antivirulence drugs
Anti-infective
WaaC
Glycosyltranferases
E. coli
Virtual screening
The global emergence of resistance to antimicrobial agents is
increasingly limiting the effectiveness of current drugs, and there
is a medical and social need for new anti-infective therapies, par-
ticularly for the treatment of nosocomial infections and the life
threatening conditions due to septicemia.
concept is based on interfering with the host/pathogen interactions
and is different from the classical antibiotics in two key ways. First,
rather than targeting genes that are essential for basic metabolism
in vitro, antivirulence focuses on mechanisms that are essential for
host/pathogen interactions especially those allowing the bacteria
to escape the innate immune system of the host. Secondly, because
the antivirulence approach targets specifically the invasive bacte-
ria (i.e., activity only against bacteria that are causing pathogenesis
via dissemination in the host) there is no growth inhibition of bac-
teria constitutive of the normal flora such as gut or skin. This ap-
proach by saving bacterial growth in natural sites preserves the
efficiency of antibiotics while also decreasing the selection of resis-
tant bacteria, the selective pressure being limited to the compart-
ments where innate immunity is active.
Lipopolysaccharide (LPS, Scheme 1) is a key element of the out-
er membrane of Gram-negative bacteria. It constitutes the outer
leaflet of the outer membrane and is essential for cell survival5
although an LPS-deficient N. meningitidis mutant as well as an
engineered E. coli mutant expressing only the Lipid A instead of
the Kdo2–Lipid A have recently been described.6,7 LPS biosynthesis
occurs in the cytoplasm, the heptose molecules being sequentially
Bacterial resistance to antibiotics is a serious public health
problem.1–3 Two factors contribute to this phenomenon: the emer-
gence of resistant bacteria consequential to the selection induced
by massive use of antibiotics in human and veterinary medicine,
and diffusion of resistant bacteria in various ecosystems.
Although resistant Gram-positive bacteria such as MRSA have
become the most frequently isolated pathogens among patients
with nosocomial infections, the prevalence of resistant Gram-neg-
ative bacteria such as the ESBL Enterobacteriaceae (e.g., E. coli,
K. pneumoniae) and the multidrug-resistant (MDR) P. aeruginosa
and A. baumanii has also been increasing, the majority of these
Gram-negative pathogens being now resistant to several com-
monly prescribed antibiotics. In order to treat infections due to
MDR bacteria, new molecules with new mechanisms of action
are urgently needed.
One innovative therapeutic concept that has been recently
introduced to find new antimicrobial agents is antivirulence.4 This
added to the Kdo2–Lipid A module (Kdo: 3-deoxy-D-manno-oct-2-
ulosonic acid) by two heptosyltransferases called WaaC (Scheme 2),
WaaF and WaaQ if a third heptose molecule is present.8 Gram-neg-
ative bacteria that lack heptose display the deep-rough phenotype9
and show a reduction in outer membrane protein content, an
* Corresponding author. Tel.: +33 (0)1 57 14 05 22; fax: +33 (0)1 57 14 05 24.
E-mail address: alexis.denis@mutabilis.fr (A. Denis).
Present address: Bayer Schering Pharma AG, Berlin, Germany.
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.05.117

F. Moreau et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4022–4026
4023
for the discovery of new antibacterial drugs based on antivirulence
mechanisms.
We report herein our approach to the identification of WaaC
inhibitors and the preliminary SAR generated from a family of 2-
aryl-5-methyl-4-(5-aryl-furan-2-yl-methylene)-2,4-dihydro-pyra-
zol-3-ones identified by virtual screening and further optimized.
First, a biochemical assay was set up using a synthetic ADP-
L-
glycero-b-
D
-manno-heptose sugar donor13 and a commercial Re-
LPS (from S. minnesota) as the acceptor substrate dissolved in
TritonX100 micelles. WaaC activity was monitored by a coupled
assay involving pyruvate kinase (PK) and luciferase or lactate
dehydrogenase as coupling enzymes to detect the product ADP.
On the other hand we have used the crystal structure of WaaC
solved at 1.9 Å resolution in complex with ADP and a substrate
analogue14 to identify the active site and binding cavities of the en-
zyme to carry out a virtual screening. In particular, a small hydro-
philic and charged cavity very close to the activated heptose was
identified as a potential binding site for the acceptor Kdo of the
LPS substrate. Using this model, 5 million of commercially avail-
able compounds were electronically/virtually screened by target-
ing the acceptor binding site.15 Two hundred thirty-seven hits
were selected according to docking score, Lipinski violations,
chemical stability and diversity, and tested. One compound, 1a
(Table 1) was identified as a potent inhibitor of WaaC, with an
IC50 of 2 lM. According to the generated model of the Kdo site,
the phenyl furyl moiety of compound 1a binds the protein partly
out of the small Kdo pocket while intruding deeply the second phe-
nyl bearing a carboxylate group in the hydrophilic pocket to make
an electrostatic contact with the Arg 120 or Arg 143 and non polar
contacts with Ile 287, His 139 and Ala 140 (Fig. 1).
Scheme 1. Schematic representation of E. coli LPS (Hep, heptose).
increased sensitivity towards detergents or hydrophobic antibiot-
ics and are more susceptible to phagocytosis by macrophages.10
In addition, a N. meningitidis deep-rough mutant has been
shown to present an attenuated virulence in an infant rat model11
and we have recently demonstrated that the inner core LPS which
constitutes the outer leaflet of Gram-negative bacteria outer mem-
brane is necessary to protect E. coli from complement killing in vi-
tro and for virulence in vivo but is dispensable for gut
colonization.12
A series of analogues of 1a was then synthesized to explore the
structure–activity relationship and the mechanism of action.
The 2-aryl-5-methyl-4-(5-aryl-furan-2-yl-methylene)-2,4-dihydro-
pyrazol-3-one analogues 1a–n were readily synthesized by condensa-
tion of 2-aryl-5-methyl-2,4-dihydro-pyrazol-3-one derivatives 2 with
5-aryl-2-furaldehyde derivatives 3 (Scheme 3).16
The 2-aryl-5-methyl-2,4-dihydro-pyrazol-3-ones 2 were either
commercially available or prepared in high yields by refluxing
the corresponding aryl-hydrazines with ethyl acetoacetate in
acetic acid.17,18 5-Aryl-furaldehydes 3 were either commercially
available or prepared by Suzuki coupling of aryl boronic acids with
5-bromo-furaldehyde in moderate to high yields (21–90%). In some
The heptosyltransferases such as WaaC which are conserved in
Gram-negative bacteria, represent promising and attractive targets
Scheme 2. Addition of the first heptose molecule to a 3-deoxy-D-manno-oct-2-ulosonic acid residue of the Kdo2–Lipid A molecule catalyzed by WaaC.

4024
F. Moreau et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4022–4026
as determined by NOESY NMR analysis of a mixture of isomers.20
Table 1
Inhibition of WaaC by compounds 1a–1n
Compounds 1a–n were thus tested as a mixture of isomers.
Acrylamide derivatives 4 were also synthesized as analogues of
1a (Scheme 4). Compounds 4 were prepared from meta- and para-
amino-benzoates by reductive amination with formaldehyde or
acetone (30–70%) followed by acylation21 to afford phosphonates
5 (85–99%). Horner–Wadsworth–Emmons olefination of 5 with
furaldehydes 3 using Masamune–Roush conditions22 (40–86%)
and saponification (50–95%) led to the E isomers of acrylamide
derivatives 4.
N
N
R1
O
O
R2
The SAR generated was in clear agreement with the docking
model and the expected mechanism of action (Table 1). The car-
boxylic acid of the aryl pyrazolone moiety was shown to be essen-
tial for the activity, as it can be placed in the Kdo pocket in close
polar interaction with the Arg 120 (m-CO2H, compounds 1a–1f)
or Arg 143 residues (p-CO2H, compounds 1g–1j). Compounds 1l–
1n lacking this functionality displayed no activity. The most potent
compounds 1a, 1b, 1f and 1g all have two CO2H groups with at
least one in meta position to enable a close electrostatic interaction
with Arg 120. In the case where R1 was m-CO2H (compounds 1a–
1f), R2 in meta or para position could afford a suitable interaction
with Arg 120 (Fig. 1). However when R1 and R2 were p-CO2H (1h),
the p-CO2H group of R2 was shifted away by up to 1.5 Å and was no
longer able to interact with Arg 120 (Fig. 2), resulting in a 20-fold
loss of activity. Therefore when R1 was p-CO2H, the conformation
of the molecule required R2 to be in meta position to properly
interact with Arg 120 as exemplified by 1g.
a
Compound
R1
R2
IC50 (lM)
1a
1b
1c
1d
1e
1f
1g
1h
1i
1j
1k
1l
1m
1n
m-CO2H
m-CO2H
m-CO2H
m-CO2H
m-CO2H
m-CO2H
p-CO2H
p-CO2H
p-CO2H
p-CO2H
m-CO2H
None
m-CO2H
2
1
31
49
15
3
m-CO2H, p-Cl
m-Me, p-CO2H
m-CO2Me
m-COMe
p-CO2H
m-CO2H, p-Cl
p-CO2H
m-CONH2
None
p-SO2NH2
m-CO2H, p-Cl
m-CO2H
3.1
51
18
55
>100
>100
>100
>100
m-CO2Me
m-Me
m-CO2H
a
Values are means of three experiments.
cases, anhydrous conditions were preferred in order to avoid
saponification of the ester functionality.
Reaction of 2 with 3 in refluxing acetic acid afforded the desired
2-aryl-5-methyl-4-(5-aryl-furan-2-yl-methylene)-2,4-dihydro-pyr-
azol-3-ones 1a–n in moderate to good yields (26–96%) as a mixture
of E/Z isomers.19
The aromatic ring linked to the furan moiety seems to be impor-
tant for the positioning of a second acceptor group, the order of po-
tency being CO2H (1a) > COMe (1e) = CONH2 (1i) > CO2Me (1d).
The introduction of a sulfonamide group (1k) was deleterious for
the activity.
Although separation of the isomers was possible by preparative
HPLC, each isolated isomer rapidly equilibrated to give back the
original mixture of isomers. The Z isomer was the major product
While the introduction of a lipophilic substituent in the meta
position (1c) had a detrimental effect, in the para position (1b) it
afforded a gain in potency.
Figure 1. (A) Structure of F-ADP-heptose bound to WaaC. (B) Identification of acceptor Re-LPS site and docking of Kdo. (C) Docking of 1a in the Re-LPS site and Kdo pocket. (D)
Interactions of 1a (blue) and 1f (green) with key residues of WaaC.

F. Moreau et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4022–4026
4025
Scheme 3. Reagents and conditions: (a) acetic acid, reflux, 3 h; (b) Pd(PPh3)4, Na2CO3 aqueous 2 M, toluene, methanol, reflux, 3 h; or Pd(PPh3)4, CsF, toluene, methanol, 60 °C,
overnight; (c) acetic acid, sodium acetate, reflux, 3 h.
Scheme 4. Reagents and conditions: (a) formaldehyde 37% aqueous or acetone,
NaBH3CN 1 M in THF, acetic acid, THF, rt, overnight; (b) (diethoxy)phosphorylacetic
acid, 2-chloro-1-methyl-pyridinium iodide, TEA, dichloromethane, rt, overnight; (c)
LiCl, DIPEA, acetonitrile, rt, overnight; (d) LiOH, THF, water, rt, overnight.
Scheme 5. Fragments and analogues of compounds 1a–1n.
Figure 3. Overlay of 1a with 4.
lene-pyrazolone arrangement was essential for the best fit of our
inhibitors in the Kdo pocket of the WaaC enzyme.
Mechanistic studies carried out with compound 1g demon-
strated that the inhibitors of the series were competitive with
the Re-LPS substrate in agreement with the in silico binding site
close to the Kdo site (Fig. 4). The inhibition by 1g does not depend
on WaaC concentration, nor on the order of addition of enzyme,
substrates and inhibitor. It is not affected by the concentration of
the donor substrate ADP-heptose. The inhibition kinetics are linear.
Extra-intestinal pathogenic E. coli (ExPEC) represent the most
common cause of Gram-negative sepsis. Because WaaC defective
mutant of the pathogenic E. coli K1 strain are significantly less vir-
ulent in mice compared to the wild type strain, the heptosyltrans-
ferase WaaC is a good target for antibacterial drugs based on the
new antivirulence concept.
Figure 2. Lack of interactions of 1h (blue) with Arg 120 in contrast to 1g (green).
Heterocyclic parts 2a and 7 of the best compound 1b were inac-
tive, which demonstrated that the full length molecule was neces-
sary for inhibiting the enzyme (Scheme 5). Analogues 823 and 4
were also explored but displayed no activity, although acrylamide
derivatives 4 positioned the two essential phenyl carboxylate
groups similarly to 1a in molecular modeling experiments
(Fig. 3). Therefore we finally assessed that the furanyl-methy-
We have identified for the first time small molecules able to in-
hibit this new class of antimicrobial target.

4026
F. Moreau et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4022–4026
References and notes
[1g]=0µM
70
60
50
40
30
20
10
0
1. Croft, A. C.; D’Antoni, A. V.; Terzulli, S. L. Med. Sci. Monit. 2007, 13, RA 103.
2. Alanis, A. J. Arch. Med. Res. 2005, 36, 697.
3. Payne, D.; Tomasz, A. Curr. Opin. Microbiol. 2004, 7, 435.
4. (a) Marra, A. Drugs R. D. 2006, 7, 1; (b) Talbot, G. H. Expert. Rev. Anti Infect. Ther.
2008, 6, 39; (c) Alksne, L. Expert Opin. Investig. Drug 2002, 11, 1149; (d) Lee, Y.;
Almqvist, F.; Hultgren, J. Curr. Opin. Pharmacol. 2003, 3, 513.
5. Gronow, S.; Brade, H. J. Endotoxin Res. 2001, 7, 3.
6. Steeghs, L.; de Cock, H.; Evers, E.; Zomer, B.; Tommassen, J.; Van der Ley, P.
EMBO J. 2001, 20, 6937.
[1g]=1.2µM
[1g]=3.7µM
[1g]=11µM
[1g]=33µM
7. Meredith, T. C.; Aggarwal, P.; Mamat, U.; Lindner, B.; Woodard, R. W. ACS Chem.
Biol. 2006, 1, 33.
8. Reeves, P. R.; Hobbs, M.; Valvano, M. A.; Skurnik, M.; Whitfield, C.; Coplin, D.; Kido,
N.; Klena, J.; Maskell, D.; Raetz, C. R.; Rick, P. D. Trends Microbiol. 1996, 4, 495.
9. Rietschel, E. T.; Kirikae, T.; Schade, F. U.; Mamat, U.; Schmidt, G.; Loppnow, H.;
Ulmer, A. J.; Zahringer, U.; Seydel, U.; Di Padova, F. FASEB J. 1994, 8, 217.
10. Coleman, W. G.; Leive, L. J. Bacteriol. 1979, 139, 899.
11. Stojiljkovic, I.; Hwa, V.; Larson, J.; Lin, L.; So, M.; Nassif, X. FEMS Microbiol. Lett.
1997, 151, 41.
12. Vongsouthi, V.; Metais, A.; Floquet S.; Escaich, S. Abstracts N° B-798, 44th
Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC),
October 30–November 2, 2004; Washington, DC.
13. Escaich, S.; Vincent, S.; Moreau, F. European Patent 1669365, 2006.
14. Grizot, S.; Salem, M.; Vongsouthi, V.; Durand, D.; Moreau, F.; Dohi, H.; Vincent,
S.; Escaich, S.; Ducruix, A. J. Mol. Biol. 2006, 363, 383.
15. The virtual high-throughput screening was carried out using the technology
4SCanÒ Seifert, H. J.; Wolf, K.; Vitt, D. Biosilico 2003, 1, 143.
0
0.4
0.8
1.2
1/[Re-LPS] (µM-1)
-10
16. Kalluraya, B.; John, J. K.; Holla, B. S. J. Indian Chem. Soc. 1995, 72, 99.
17. Duffy, K. J.; Darcy, M. G.; Delorme, E.; Dillon, S. B.; Eppley, D. F.; Erickson-Miller, C.;
Giampa, L.; Hopson, C. B.; Huang, Y.; Keenan, R. M.; Lamb, P.; Leong, L.; Liu, N.;
Miller, S. G.; Price, A. T.; Rosen, J.; Shah, R.; Shaw, T. N.; Smith, H.; Stark, K. C.; Tian,
S.-S.; Tyree, C.; Wiggall, K. J.; Zhang, L.; Luengo, J. I. J. Med. Chem. 2001, 22, 3730.
18. DeRuiter, J.; Carter, D. A.; Arledge, W. S.; Sullivan, P. J. Heterocycl. Chem. 1987,
24, 149.
Figure 4. Competition between 1g and Re-LPS: double reciprocal plot showing
intersecting lines on y-axis at 1/Vmax
.
19. Data for 1a obtained as a mixture of E/Z isomers: 1H NMR (DMSO-d6, 400 MHz)
d 2.38 (s, 3H, major isomer), 2.78 (s, 3H, minor isomer), 7.56–7.60 (m, 2H of
both isomers and 1H of minor isomer), 7.66–7.70 (m, 1H, both isomers), 7.76
(d, J = 8.0 Hz, 1H, both isomers), 7.83 (d, J = 4.0 Hz, 1H, minor isomer), 7.86 (s,
1H, major isomer), 8.02 (d, J = 8.0 Hz, 1H, major isomer), 8.15 (d, J = 8.0 Hz, 1H,
minor isomer), 8.21 (d, J = 8.0 Hz, 2H, both isomers), 8.47 (s, 1H, minor isomer),
8.50 (s, 1H, major isomer), 8.52 (s, 1H, minor isomer), 8.56 (s, 1H, major
isomer), 8.74 (d, J = 4.0 Hz, 1H, major isomer). ESI-MS m/z: 417 (M+H)+.
20. For instance the ratio (Z:E) for 1a is (3.5:1) and for 1g (3.4:1). NOESY NMR
analysis (DMSO-d6, 400 MHz) for 1a displayed interaction between the methyl
on the pyrazolone (2.38 ppm) and the olefinic proton (7.86 ppm) in the case of
The crystal structure of WaaC alone or in complex with ADP
was used to identify the active site and binding cavities and to
carry out a virtual screening of commercially available libraries.
We have shown that the activity of WaaC can be easily assayed
by using
a synthetic ADP-heptose substrate. The 2-aryl-5-
methyl-4-(5-aryl-furan-2-yl-methylene)-2,4-dihydro-pyrazol-3-
one 1a was discovered as a micromolar inhibitor and a series of
analogues was synthesized to explore the SAR. All the synthetic
modifications as well as the docking model and biochemical
experiments are in agreement with a binding close to the accep-
tor site of the Re-LPS, likely in the Kdo pocket for this new class
of inhibitors of WaaC.
The level of inhibition of 1b and the preliminary SAR clearly
demonstrate that this family constitutes a promising lead series
for further optimization as antivirulence drug.
the
Z isomer, and interaction between the methyl on the pyrazolone
(2.78 ppm) and a proton of the phenyl (8.15 ppm) linked to the furan ring
for the E isomer. Similar observations were made for 1g.
21. Minami, T.; Isonaka, T.; Okada, Y.; Ichikawa, J. J. Org. Chem. 1993, 58, 7009.
22. Blanchette, M. A.; Choy, W.; Davis, J. T.; Essenfeld, A. P.; Masamune, S.; Roush,
W. R.; Sakai, T. Tetrahedron Lett. 1984, 25, 2183.
23. Compound 8 was prepared by condensation of aryl-pyrazolone 2a with 3-
formyl benzoic acid according to the same procedure as in Scheme 2.