Systematic synthesis of inhibitors of the two first enzymes of the bacterial heptose biosynthetic pathway: Towards antivirulence molecules targeting lipopolysaccharide biosynthesis

Article abstract of DOI:10.1002/chem.201100396

L-Heptoses (L-glycero-D-manno-heptopyranoses) are constituents of the inner core of lipolysaccharide (LPS), a molecule playing key roles in the mortality of many infectious diseases as well as in the virulence of many human pathogens. The inhibition of the first enzymes of the bacterial heptose biosynthetic pathway is an almost unexplored field to date although it appears to be a very novel way for the development of antivirulence drugs. We report the synthesis of a series of D-glycero-D-manno-heptopyranose 7-phosphate (H7P) analogues and their inhibition properties against the isomerase GmhA and the the kinase HldE, the two first enzymes of the bacterial heptose biosynthetic pathway. The heptose structures have been modified at the 1-, 2-, 6- and 7-positions to probe the importance of the key structural features of H7P that allow a tight binding to the target enzymes; H7P being the product of GmhA and the substrate of HldE, the second objective was to find structures that could simultaneously inhibit both enzymes. We found that GmhA and HldE were extremely sensitive to structural modifications at the 6- and 7- positions of the heptose scaffold. To our surprise, the epimeric analogue of H7P displaying a D-glucopyranose configuration was found to be the best inhibitor of both enzymes but also the only molecule of this series that could inhibit GmhA (IC₅₀=34 μM) and HldE (IC₅₀=9.4 μM) in the low micromolar range. Noteworthy, this study describes the first inhibitors of GmhA ever reported, and paves the way to the design of a second generation of molecules targeting the bacterial virulence. Copyright

Full text of DOI:10.1002/chem.201100396

FULL PAPER
DOI: 10.1002/chem.201100396
Systematic Synthesis of Inhibitors of the Two First Enzymes of the Bacterial
Heptose Biosynthetic Pathway: Towards Antivirulence Molecules Targeting
Lipopolysaccharide Biosynthesis
Maxime Durka,[a] Abdellatif Tikad,[a] Rꢀgis Pꢀrion,[a] Michael Bosco,[a]
Mounir Andaloussi,[a] Stꢀphanie Floquet,[b] Elodie Malacain,[b] FranÅois Moreau,[b]
Mayalen Oxoby,[b] Vincent Gerusz,[b] and Stꢀphane P. Vincent*[a]
Abstract: l-Heptoses (l-glycero-d-
manno-heptopyranoses) are constitu-
ents of the inner core of lipolysacchar-
ide (LPS), a molecule playing key roles
in the mortality of many infectious dis-
eases as well as in the virulence of
many human pathogens. The inhibition
of the first enzymes of the bacterial
heptose biosynthetic pathway is an
almost unexplored field to date al-
though it appears to be a very novel
way for the development of antiviru-
lence drugs. We report the synthesis of
a series of d-glycero-d-manno-hepto-
pyranose 7-phosphate (H7P) analogues
and their inhibition properties against
the isomerase GmhA and the the
kinase HldE, the two first enzymes of
the bacterial heptose biosynthetic path-
way. The heptose structures have been
modified at the 1-, 2-, 6- and 7-posi-
tions to probe the importance of the
key structural features of H7P that
allow a tight binding to the target en-
zymes; H7P being the product of
GmhA and the substrate of HldE, the
second objective was to find structures
that could simultaneously inhibit both
enzymes. We found that GmhA and
HldE were extremely sensitive to struc-
tural modifications at the 6- and 7- po-
sitions of the heptose scaffold. To our
surprise, the epimeric analogue of H7P
displaying a d-glucopyranose configu-
ration was found to be the best inhibi-
tor of both enzymes but also the only
molecule of this series that could inhib-
it GmhA (IC50 =34 mm) and HldE
(IC50 =9.4 mm) in the low micromolar
range. Noteworthy, this study describes
the first inhibitors of GmhA ever re-
ported, and paves the way to the
design of a second generation of mole-
cules targeting the bacterial virulence.
Keywords: bactericides · cell wall ·
heptose · inhibition · virulence
Introduction
(Kdo) and two molecules of l-glycero-a-d-manno-heptose
(heptose), and the outer core is composed of hexoses.[4]
Lipid A and one Kdo is the minimal structure for maintain-
ing cell viability.[2] Gram-negative bacteria that lack heptose
display the deep-rough phenotype[1] and show a dramatically
increased sensitivity towards detergents or hydrophobic an-
tibiotics and are much more susceptible to phagocytosis by
macrophages,[5] as well as to the bactericidal effect of the
host.[6] Therefore, the inhibition of the heptose biosynthetic
pathway can be seen as a novel approach to develop new
weapons to combat bacterial pathogens: instead of targeting
the central metabolism or the cell wall construction, this ap-
proach consists in attenuating or even abolishing the viru-
lence of the microorganism without the need to kill it.
The main polysaccharide present at the surface of gram-neg-
ative bacteria is lipopolysaccharide (LPS). This complex
molecule plays key roles in the mortality of many infectious
diseases as well as in the virulence of numerous human
pathogens.[1] As cell wall glycolipids, LPS insures a protec-
tion against hydrophobic molecules and participate to the
bacterial cell integrity.[2]
LPS is an amphipathic molecule that can be decomposed
into three main substructures: lipid A, the oligosaccharide
core and the O-antigen.[3] The oligosaccharide core can be
divided into two parts: the inner core is formed at least of
one molecule of 3-deoxy-a-d-manno-oct-2-ulosonic acid
Indeed, antivirulence has emerged as an alternative che-
motherapeutic strategy to the discovery of new bactericides
or bacteriostatic molecules.[7–9] With this concept in mind,
we have recently reported on the inhibition of heptosyl-
transferase WaaC as a novel approach to inhibit the bacteri-
al cell wall resistance to innate immune response.[10,11] More-
over, we just disclosed the synthesis and the inhibition prop-
erties of glycosylated fullerenes designed to inhibit the ad-
hesion of uropathogenic E. coli strains to bladder cells.[12]
Thus, the inhibition of the heptose biosynthesis appears to
be a very novel field for the development of antivirulence
[a] Dr. M. Durka, Dr. A. Tikad, Dr. R. Pꢀrion, Dr. M. Bosco,
Dr. M. Andaloussi, Prof. S. P. Vincent
Chemistry Department, University of Namur
rue de Bruxelles 61, 5000 Namur (Belgium)
Fax : (+32)81-72-45-17
E-mail: stephane.vincent@fundp.ac.be
[b] S. Floquet, E. Malacain, Dr. F. Moreau, Dr. M. Oxoby, Dr. V. Gerusz
Mutabilis S.A., 102, route de Noisy
93230 Romainville (France)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201100396.
Chem. Eur. J. 2011, 17, 11305 – 11313
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
11305

drugs. Interestingly, the inhibition of this pathway has been
relatively unexplored compared, for instance, to the pepti-
doglycan biosynthesis. This can be explained by the fact that
i) the enzymes involved in the heptose biosynthesis have
been cloned and characterized only recently, and ii) their
substrates are not commercially available and their synthesis
is not straightforward.
The bacterial heptoses are constructed from sedoheptu-
lose-7-phosphate 1, which originates from the central metab-
olism (Scheme 1).[13] A keto–aldose isomerase GmhA trans-
Recently, Tanner et al. have led a detailed mechanistic
study on the HldD-catalyzed epimerization step.[17–22] Crystal
structures of HldD have been obtained and provided impor-
tant information about its catalytic pocket.[18,23] Despite this
mechanistic and structural knowledge, no potent inhibitor of
HldD has been described to date.
Surprisingly, the inhibition of the first enzymes of the hep-
tose biosynthetic pathway is an almost unexplored field, al-
though its importance had already been acknowledged by
Paulsen et al. in 1994.[24] They indeed described the synthesis
of a series of d-glycero-d-manno-heptosides modified at the
anomeric position but the biological evaluation has, unfortu-
nately, never been published. In 2006, Wright et al. devel-
oped an in vitro screen of bacterial lipopolysaccharide bio-
synthetic enzymes that allowed the identification of an in-
hibitor of HldE.[25] Later on, some of us identified inhibitors
of HldE from a high-throughput screening that allowed a
structure activity relationship studies on heterocyclic struc-
tures.[26] Noteworthy, several crystal structures of GmhB[27]
and GmhA have been obtained, some of them in complex
with its substrate that allowed a mapping of the interactions
between the sugar-phosphate and the enzyme.[28–30] Howev-
er, to the best of our knowledge, there is no GmhA inhibitor
reported to date.
Thus, there is no literature report on the relative affinities
(i.e., the hierarchy of interactions) of all the functional
groups present in the heptose skeleton with the first en-
zymes of this therapeutically relevant biosynthetic pathway.
Therefore, we defined, as a first objective, to determine the
tolerance of structural modifications of the heptose phos-
phates as ligands (substrates or inhibitors) of these enzymes,
especially the two first ones. Moreover, H7P 2 being the
product of GmhA and the substrate of HldE, our second
objective was to find structures
forms
1 into d-glycero-d-manno-heptose 7-phosphate 2
(coined H7P or Heptose-7-Phosphate for simplification in
this paper) which is then phosphorylated by the kinase
HldE. After hydrolysis of the terminal phosphate of 3, inter-
mediate 4 is transformed into nucleotide-sugar 5. Interest-
ingly, bacterial strains such as E. coli use the same HldE
enzyme for two non-consecutive steps. A regioselective d to
l epimerization is then catalyzed by HldD, yielding ADP-l-
heptose 6, the donor substrate of heptosyltransferases
(WaaC, WaaF and WaaQ).
To date, the most studied enzymes of this pathway are the
epimerase HldD and heptosyltransferase WaaC. In 2000,
Kosma et al. described the first synthesis of the two anomers
of ADP-l-heptose 6,[14] thus demonstrating the anomeric
configuration of 6. Later, the same team disclosed the syn-
thesis of C-glycosidic analogues[15] of 6 for which no inhibi-
tion data are available to date. In 2008, our group has de-
scribed the synthesis and the inhibition properties of the 2-
fluoro analogue of 6.[10] Fortunately, we were able to obtain
a 3D structure of WaaC in complex with this inhibitor.[11]
Based to a virtual screening and a structure–activity rela-
tionship studies, Moreau et al. have been able to develop
low-micromolar inhibitors of heptosyltransferase WaaC.[16]
that could simultaneously inhib-
it both enzymes.
To answer to these important
questions, we report here the
synthesis and the inhibition
evaluation of a series of hep-
tose-7-phosphate
analogues.
The heptose structures have
been modified at the 1-, 2-, 6-
and 7-positions and assayed
against GmhA and HldE.
Strategically, we decided that
our first synthetic targets
should be analogues of the key
structure heptose-7-phosphate 2
for two important reasons.
First, the initial step of the hep-
tose biosynthesis being in fact
an equilibrium between 1 and
2,[28] H7P is thus substrate of the
two first enzymes of this path-
way. As we will show below,
molecules mimicking 2 can thus
Scheme 1. Bacterial heptose biosynthetic pathway.
11306
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 11305 – 11313

Antivirulence Molecules Targeting Lipopolysaccharide Biosynthesis
FULL PAPER
be inhibitors of both enzymes. Second, sedoheptulose-7-
phosphate (1) being also a product of the central metabo-
lism of eukaryotes, glycomimetics of this biosynthetic inter-
mediate may lead to non-selective inhibition profiles. In
contrast, mannoheptosides such as compounds 2–6 are only
found in prokaryotes.[4]
To probe the importance of the key structural features of
H7P, we designed and synthesized a series of analogues of 2
depicted in Figure 1. For the carbohydrate substructure, the
two main stereochemical characterics of heptoside 2 are its
configurations at C2 (d-manno) and C6 (d-glycero) that con-
fers to this biosynthetic intermediate its sugar identity. Com-
pounds 7 (d-glycero-d-manno), 8 (l-glycero-d-manno) and 9
(d-glycero-d-gluco) were thus designed to address the ques-
tion of defining the extent of which modifications at these
positions hamper the binding process or at least decrease
the affinity of the heptosidic analogues towards the en-
zymes. To prevent these molecules to be substrates of HldE,
an anomeric a-methyl group was installed on each of these
molecules.
Results and Discussion
Several synthetic procedures have been developed for the
construction of l-heptoses, either through a homologation
of a mannoside,[31,32] or by reconstitution of the heptose skel-
eton through a central asymmetric aldol reaction.[33] For the
synthesis of fluoro-heptosides, we also explored the stereo-
selective epoxide formation followed by a cesium acetate
opening.[10] The two strategies that are the most commonly
used to date are a d-selective dihydroxylation of alkene 22
(Scheme 2, pathway D)[34] followed, if l-heptosides are re-
quired, by a Mitsunobu inversion,[35–37] and the l-selective
addition of Grignard reagents (pathway L).[38–46]
To probe the importance of the 7-phosphate group in
terms of interactions and binding strength with the enzymes,
a small series of phosphate mimics was also synthesized: car-
boxylates 10 and 11, phosphonates 12 and 13, phosphora-
mide 14, sulfonamide 15 and triazole 16.
To evaluate the impact of the anomeric a-methyl group
on the binding process, lactols 11 and 13 were designed to
be directly compared to 10 and 12, respectively.
Overall, this series comprises five methyl glycosides (com-
pounds 7–10, 12) and five free anomeric sugars (compounds
11, 13–16) that will enable us to scan the possibilities of
structural variations on the d-glycero-d-manno-pyranose 7-
phosphate 2 central structure.
Scheme 2. General synthetic scheme.
General synthesis of d-glycero-d-mannoheptosides: For the
synthesis of all analogues of 2 with the d-glycero-d-manno-
heptose scaffold, we followed a general synthetic pathway
depicted in Scheme 3. This strategy relies on the known
pathway described in the literature for the construction of
d-heptosides (see Scheme 2).[47,48] The anomeric thiophenyl
functionality was selected for its robustness and because it
allows both the introduction of a-methyl or a chemoselec-
tive hydrolysis to provide lactols. Starting alkene 22 was ob-
tained by a Swern oxidation of alcohol 18 directly followed
by a Wittig olefination (Scheme 2).
The latter was dihydroxylated in 74% yield with a 4:1 d/l
ratio. The pure d-glycero diol 25 could be separated at this
stage by careful silica gel chromatography and was trans-
formed into the desired 6-benzylated heptoside 26 through a
three-step sequence of protective group manipulations that
required a single purification final step. Intermediate 26, the
central knot of our divergent synthetic approach could be
produced in 11 steps in a robust multigram scale with an
overall yield of 27% from d-mannose. Phosphorylation of
26 under standard Mitsunobu conditions yielded 7-phos-
phate 27 in 90% yield. This intermediate could easily be de-
Figure 1. Targeted H7P analogues.
Chem. Eur. J. 2011, 17, 11305 – 11313
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11307

S. P. Vincent et al.
regarding modifications at the
2-position of the heptose skele-
ton. In fact, for the synthesis of
ADP-2-fluoroheptose that we
synthesized as a potent inhibi-
tor
of
heptosyltransferase
WaaC, we had already studied
the homologation of methyl a-
glucosides and demonstrated
the d and l configurations of
the heptosides obtained either
by dihydroxylation of 34 (see
Scheme 4) or by nucleophilic
epoxide opening.[10] The se-
quence silylation/benzylation/
desilylation produced alcohol
36 in 60% yield over three
steps. A Mitsunobu phosphory-
lation followed by hydrogenoly-
sis furnished the desired gluco-
heptoside 9 in 97% yield for
the two last steps.
Synthesis of 8, the l-glycero-d-
manno epimeric analogue of in-
hibitor 7: With the same ration-
ale than for the preceding mol-
ecule, we developed the synthe-
sis of the epimer at the 6-posi-
tion of the mannoheptose core
structure. Due to its extremely
Scheme 3. Synthesis of H7P analogues with a d-glycero-d-manno-heptoside core structure: i) 10% K2OsO4-
(H2O)2, K3Fe(CN)6, K2CO3, tBuOH/H2O 1:1, 08C!RT; ii) a) TIPSCl, imidazole, THF, 08C!RT, then
ii) b) BnBr, NaH, DMF, 08C!RT, then ii) c) TBAF, THF, 08C!RT; iii) PPh3, (BnO)2PO2H, NEt3, DEAD
40% in toluene, THF, RT; iv) NBS, acetone/H2O, ꢀ158C; v) Pd/C 10%, H2, AcOEt/EtOH/H2O 3:5:2, RT;
vi) MeI, Ag2O, DMF, RT; vii) MsCl, DMAP, Py; viii) NaN3, TBABr, DMF, 908C; ix) (BnO)3P, toluene, 908C;
x) a) PPh3, toluene, then x) b)MsCl, then x) c) NaHCO3aq; xi) 2-butyn-1,4-diol, 1108C; xii) H2, 20% Pd(OH)2,
MeOH.
high
l-diastereoselectivity
(Scheme 1, pathway L), the
most appealing strategy for us
was the sequence developed by
van Boom and co-work-
ers[39,40,49] which consists in the
protected to give the natural occuring compound 2,[22,48] that
was used as substrate of HldE for the inhibition assays. In
parallel, the thioacetal 27 was easily transformed into
methyl heptoside 28 under standard glycosidation conditions
and hydrogenolyzed to furnish 7, the first key analogue of
H7P 2.
addition of the silylated Grignard reagent PhMe2SiCH2MgCl
followed by a Fleming–Tamao oxidation (Scheme 5).[50–52]
A two-step azidation protocol furnished 7-azide 29 in
80% yield from 26. A Huisgen cycloaddition involving this
second key intermediate 29 and a symmetrical butynediol
yielded triazole 16 after the two usual deprotection steps.
On the other hand, the azide functionality of 29 could be
chemoselectively reduced into an intermediate primary
amine that was transformed either into sulfonamide 31 with
mesyl chloride or phosphoramide 30 using dibenzyl chloro-
phosphate. The optimized deprotection protocol thus pro-
vided two additional heptosides 14 and 15.
Scheme 4. Synthesis of the d-glycero-d-gluco analogue 9: i) a) (COCl)2,
DMSO, NEt3, CH2Cl2, ꢀ788C ! 08C, then i) b) Ph3PCH3Br, nBuLi
(2.5m), THF, ꢀ788C!RT; ii) K2OsO
4ACTHNUGTREN(NUNG H2O)210%, NMO, H2O/acetone
1:1, 08C; iii) a) TIPSCl, Im, THF, 08C ! RT, then iii) b) BnBr, NaH,
DMF, 08C ! RT, then iii) c) TBAF, THF, RT; iv) PPh3, (BnO)2PO2H,
NEt3, DEAD 40% in toluene, THF, RT; v) 10% Pd/C, H2, AcOEt/
EtOH/H2O 3:5:2, RT.
Synthesis of the d-glycero-d-gluco epimeric analogue of in-
hibitor 7: As mentioned above, we considered that it was
important to probe the tolerance of the targeted enzymes
11308
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 11305 – 11313

Antivirulence Molecules Targeting Lipopolysaccharide Biosynthesis
FULL PAPER
Scheme 5. Synthesis of the l-glycero-d-manno analogue 8: i) a) COCl)2,
DMSO, NEt3, CH2Cl2, ꢀ788C ! 08C, i) b) PhSi(Me)2-CH2-MgCl, THF,
08C, then i) c) BnBr, NaH, DMF, 08C ! RT; ii) Hg
ACHTUNGTERNUN(NG OTFA)2, AcOH/
Ac2O 1:1, AcOK, 40% AcOOH in AcOH, 108C ! RT; iii) PPh3,
(BnO)2PO2H, NEt3, 40% DEAD in toluene, THF, RT; iv) 10% Pd/C,
H2, AcOEt/EtOH/H2O 3:5:2, RT.
Indeed, this reaction gave the l-adduct in 75% yield with a
d/l diastereoselectivity superior to 14:1 as reported in the
literature. In our strategy, this sequence was also attractive
because we could achieve the direct benzylation of the inter-
mediate alcohol and obtain l-heptoside 21 ready for regiose-
lective functionalizations at the 7-position without the need
to manipulate protective groups to differentiate the 6- and
the 7-position of heptosides. For the Fleming–Tamao step,
we could reproduce the yields reported in the literature on a
50 mg scale (>60%) but we faced, however, major difficul-
ties when we scaled-up this oxidation. Side reactions such as
partial debenzylation very significantly decreased the yields
(<30%). After inspection of the recent literature data, we
found that Ley and co-workers had performed Fleming–
Tamao oxidations on a very complex and sensitive structure
exploiting a milder procedure based on the use of mercury
salts.[53] Inspired by this seminal work, we developed a novel
procedure for the Fleming–Tamao oxidation of intermediate
21 under almost neutral conditions. The yields were always
in the range of 80–90% on a multigram scale. The resulting
l-heptoside 38 could thus be engaged in a Mitsunobu phos-
phorylation. The subsequent hydrogenolysis yielded the de-
sired final l-stereoisomer 8. The same molecule had been al-
ready prepared from another Grignard reagent and a differ-
ent sequence.[54,55]
Scheme 6. Synthesis of octoses and phosphonates analogues of H7P 2:
i) P
(Ph3)=CH-CO2Me, CH2Cl2, 08C
!
RT; ii) 10 % K2OsO4ACHTUNGTRENNUNG(H2O)2,
NMO, H2O/acetone 1:2, 08C ! RT; iii) 10% Pd/C, H2, AcOEt/EtOH/
H2O 3:5:2, RT; iv) LiOH, H2O, RT; v) LAH, THF, 08C ! RT; vi) Ac2O,
Py, DMAP, RT; vii) Ac2O, AcOH, H2SO4, 08C; viii) MeONa, MeOH,RT;
ix) CH2(P
08C ! RT.
G
deacetylation followed by LiOH mediated methyl ester hy-
drolysis gave the best yield of target molecule 11 and was
preferred as the direct total saponification of intermediate
42.
A similar sequence could be applied from dibenzyl phos-
phonate 23. The same syn-dihydroxylation gave diol 44 in
60% yield that could be deprotected either directly to fur-
nish 12 or after the acetolysis sequence to provide lactol 13
in quantitative (one step) and 53% yield (three steps) from
44, respectively.
Enzymatic assays: The IC50 values of compounds 7 to 16
were measured following the procedure we already de-
scribed.[26] Among the two catalytic activities of the enzyme
HldE, we naturally assayed the heptosides against the
kinase activity.
An analysis of the IC50 values reported in Table 1 could
lead us to draw clear conclusions about the effect of modifi-
cations of the heptose-7-phosphate structure and thus al-
lowed a “mapping” of the interactions with these two bacte-
rial enzymes. Compound 7, the closest analogue of H7P 2
displayed a good inhibition level only against the kinase
HldE. The inhibition profile of 7 against GmhA strongly
suggests that mimicking the product of the first step of the
heptose biosynthetic pathway is not the proper strategy to
inhibit it. Then, we could demonstrate that the inversion of
the configuration at C6 dramatically affects the affinity of
the heptoside towards both enzymes. The l-heptose-phos-
phate 8 is a poor inhibitor of GmhA and does not display
any inhibition on the kinase activity of HldE.
Synthesis of octosides and phosphonates analogues of 2:
The synthesis of the targeted octosides (Scheme 6) began
with the Horner–Wadsworth–Emmons olefination of alde-
hyde 19 giving the expected (E) unsaturated ester 24 in
94% yield. A similar sequence has been followed in the lit-
erature for generating analogues of mannose-6-phos-
phates.[56] The subsequent syn-dihydroxylation yielded the
(6S,7R)-isomer as expected from the Kishiꢂs rule.[57] The two
configurations were ascertained by X-ray crystallography of
peracetate 41 obtained after reduction of the ester.[58]
A
careful acetolysis of 40 allowed the deprotection of the
anomeric acetal and was directly followed by a debenzyla-
tion to give peracetate 42 in 84%. A sequence of Zemplꢀn
Chem. Eur. J. 2011, 17, 11305 – 11313
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11309

S. P. Vincent et al.
Table 1. Inhibition data.[a]
Compounds
Isomerase GmhA IC50 [mm]
Kinase HldE IC50 [mm]
7
8
9
>100
679
34
16
>1000
9.4
10
11
12
13
14
15
16
>1000
239
438
>1000
>1000
>1000
>1000[b]
>1000[c]
>1000
>1000
47
1096
>1000
756
[a] All measurements were repeated
3
to
6
times, then averaged.
[b] 35% Inhibition at a concentration of 1000 mm. [c] 31% Inhibition at
1000 mm.
For the inhibition of isomerase GmhA, the presence of
the anomeric a-methyl group affects the binding strength of
the inhibitors. The relative IC50 values of carboxylates 10
and 11, as well as phosphonates 12 and 13, clearly show that
the removal of the anomeric methyl group significantly en-
hances the binding affinities (almost 10 times for 13). Such
an effect cannot be discussed for the kinase activity of
HldE, since the methyl glycoside 9 displayed a low micro-
molar inhibition profile while the other analogues were not
inhibitors (with or without anomeric methyl groups).
Scheme 7. Putative enzymatic isomerization of 2.
only molecule of this series that could inhibit GmhA (IC50 =
34 mm) and HldE (IC50 =9.4 mm) in the low micromolar
range. This result is in deep contrast with the inhibition data
discussed above for manno-heptoside 8, another epimeric
analogue of H7P. Regarding isomerase GmhA, this surpris-
ing result might be rationalized by the fact that this enzyme
indeed transforms the 2-position by catalyzing a 1,2-ketol-
to-aldol isomerization. Although a mechanism involving a
hydride shift has not been totally ruled out, a double tauto-
merization involving an intermediate 1,2-enediol seems
likely, as proposed by Junop[28] and Harmer[29] (see
Scheme 7).
Thus, the conformation(s) adopted by the carbohydrate
scaffold at the transition state(s) may place the 2-hydroxyl
in an intermediate position between a pure “manno” (axial)
or “gluco” (equatorial) positions. The slightly better inhibi-
tion level observed against HldE for the d-gluco derivative
9 compared to the d-manno analogue 7 is more difficult to
rationalize. To study the binding mode of 9, we performed a
Lineweaver–Burk experiment (Figure 2) that led us to con-
firm that this gluco-heptose was indeed a competitor of the
manno-heptose substrate 2. This experiment also gave us a
Ki of 5.2 mm for 9, a value in agreement with its IC50.
As for the epimer at C-6, the two enzymes were extreme-
ly sensitive to modifications of their substrate at C-7 despite
the fact that in both cases, the catalytic transformation
occurs at the anomeric position. This lack of tolerance to
modification of the phosphate moiety was especially strong
with the kinase HldE for which IC50ꢂs were always above
1 mm. Phosphate mimics 14, 15 and 16 displayed weak inhib-
ition profiles of GmhA, close to 1 mm. Phosphoramides such
as 14 might be labile in aqueous solution and degrade under
the assay conditions, thus causing a lack of inhibition. Sur-
prisingly, carboxylate 11 displayed a significantly lower IC50
than phosphoramide 14. This result could not be really an-
ticipated from the simple examination of the structures of
It is tempting to rationalize our results in light with the
crystallographic structure of GmhA. To date eight structures
of the isomerase orthologues of 5 bacterial species are re-
ported in the literature. The structures of GmhA are de-
scribed in apo, substrate and product-bound forms. It has
been suggested that GmhA can adopt two distinct confor-
mations during isomerization through reorganization of the
quaternary structure (open and closed). It was suggested[29]
that the closed conformation of B. pseudomalleiꢂs GmhA is
catalytically relevant for the development of therapeutics.
Unfortunately, for E. coliꢂs GmhA, only the open form was
crystallized in apo or substrate-bound forms. Since our in-
hibition study has been performed on E. coliꢂs GmhA, we
will only refer here to the E. coli structure.[28] The latter has
been cocrystallized only with sedoheptulose-7-phosphate 1
(see Scheme 1 for the pyranose form). Very interestingly,
this sugar is present in an acyclic ketone form (represented
in Scheme 7) suggesting that the enzyme stabilizes the hep-
tose in a linear tautomeric form. This stabilization is consis-
tent with the two possible isomerization mechanisms, espe-
cially the one involving an enediol intermediate depicted in
Scheme 7. This stabilization might also explain why lactols
11 and 13, that can adopt an acyclic form within GmhA cat-
alytic pocket, display significantly lower IC50 values than
their corresponding methyl pyranosides 10 and 12 that are
forced to bind the enzyme as a 6-membered ring.
Very interestingly, the two enzymes were found much
more permissive to modifications at the 2-position of the
carbohydrate scaffold. Thus, low-micromolar inhibition
levels were measured for the gluco analogue 9 of manno-
heptose H7P 2. To our surprise, this epimeric analogue of
H7P was the best inhibitor of both enzymes but also the
11310
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 11305 – 11313

Antivirulence Molecules Targeting Lipopolysaccharide Biosynthesis
FULL PAPER
worthy, this study describes the first inhibitors of GmhA
ever reported.
From this systematic study and the recently published 3D-
structures of GmhA, we are confident that the structural
bases of the development of powerful inhibitors of this im-
portant biosynthetic pathway will be found and will ulti-
mately lead to the discovery of novel therapeutic strategies
to fight infectious diseases. Inhibitors of bacterial heptose
synthesis are expected to prevent full LPS development in
Gram negative bacteria, inducing a high sensitivity to the
host complement and preventing or inhibiting bacterial in-
fection. Small molecules inhibitors of heptose synthesis may
therefore provide a novel way to treat bloodstream infec-
tions caused by pathogenic Gram-negative bacteria, without
affecting the commensal flora and with less selective pres-
sure than conventional antibacterial agents.
Figure 2. Competition between HldE substrate H7P (KM =0.095 mm) and
~
&
*
inhibitor 9 (Ki =5.2 mm; *: 33mm, : 11mm, : 3.7mm, : 0) for HldE
(Lineweaver–Burk plot).
Experimental Section
these molecules compared to 2. To our delight, phosphonate
13 displayed a low micromolar IC50 against GmhA, a result
that was rather unexpected given the poor inhibition of all
other phosphate mimics.
An examination of E. coli GmhA structures shows that a
collection of four serine and one threonine residues consti-
tute the phosphate binding pockets.[28,29] Moreover, site-di-
rected mutagenesis of the threonine residue led to an inac-
tive enzyme.[28,29] These residues are structurally conserved
and form tight contacts with the substrate 1. Thus, even
slight structural modifications of the phosphate moiety of
substrate analogues might give repulsive interactions and
dramatically decrease the binding affinity of the inhibitors.
General techniques: All reactions were carried out under an argon at-
mosphere. Yields refer to chromatographically and spectroscopically ho-
mogeneous materials. Reagents and chemicals were purchased from
Sigma–Aldrich and Acros at ACS grade and were used without purifica-
tion. All reactions were performed using purified and dried solvents: tet-
rahydrofuran (THF) was refluxed over sodium/benzophenone, dichloro-
methane (CH2Cl2), triethylamine (NEt3), and pyridine were refluxed over
calcium hydride (CaH2). All reactions were monitored by thin-layer chro-
matography (TLC) carried out on Merck aluminum roll silica gel 60-
F254 using UV light and a molybdate/sufuric acid solution as staining re-
agent. 1H, 13C, and 31P NMR spectra were recorded on a JEOL (JNM
EX-400) or a Bruker (AMX-400) spectrometer at 208C. All compounds
were characterized by 1H, 13C, and 31P NMR as well as by 1H,1H and
1H,13C correlation experiment when necessary. The following abbrevia-
tions were used to describe the multiplicities: s=singlet, d=doublet, t=
triplet, m=multiplet, br=broad, brs=broad singlet. The numbering of
the protons and carbons is analogous to the proton numbers resulting
from the name of the compound. Aromatic, benzyl, acetyl and methyl
(carbons and protons) are respectively labeled with “arom”, “Bn”, “Ac”
and “Me” subscript, quaternary carbons are indicated with a “q” sub-
script. Chemical shifts (d) are reported in ppm and referenced indirectly
to TMS via the solvent (or residual solvent) signals. Merck silica gel (60,
particle size 0.040–0.063 mm) was employed for flash column chromatog-
raphy and preparative thin layer chromatography using technically sol-
vent distilled prior to use as eluting solvents. LC-MS measurement were
performed on an Agilent 6200 series TOF mass spectrometer using an
Agilent 1200 series LC system. Purifications of final molecules were real-
ized by semi-preparative HPLC using a Waters Delta prep 4000 chroma-
tography system equipped with a RP-C18 column (Agilent). Compounds
17,[59] 18,[60] 21,[39] were prepared from known procedures.
Conclusion
In summary, we have reported a systematic study in which
we synthesized and evaluated the binding affinities of a
series of Heptose-7-phosphate analogues (H7P, d-glycero-d-
manno-pyranose), an important intermediate in the LPS
biosynthesis. Derivatizations of a central scaffold allowed
the preparation of heptosides with structural modifications
at the 1-, 2-, 6- and 7-positions. The inhibition profile of the
whole inhibitor family towards the two first enzymes of the
heptose biosynthetic pathway clearly indicated that the two
enzymes are extremely sensitive to structural modifications
of the heptose scaffold at the 6- and 7-positions. Interesting-
ly, we could also show that both enzymes tolerated an epi-
merization at the 2-position and maintain low-micromolar
inhibition levels with gluco-heptose analogues.
Overall, this study also showed that the two first enzymes
of the heptose biosynthetic pathway do not show the same
level of tolerance regarding the structural modifications of
the H7P scaffold. Indeed, the isomerase GmhA was more
permissive than the kinase HldE which only allowed struc-
tural changes at the 2-position of the heptose moiety. Note-
Inhibition of the enzymatic activity of GmhA (luminescent assay): The
assay buffer “AB” contained 50 mm Hepes pH7.5, 1 mm MnCl2, 25 mm
KCl, 0.012% Triton-X100 and 1 mm dithiothreitol (DTT) and 0.1 mm
Myelin basic protein (MBP). The following components were added in a
white polystyrene Costar plate up to a final volume of 30 mL: 10 mL in-
hibitor dissolved in DMSO/water 50/50, and 20 mL GmhA of E. coli in
AB. After 30 min of pre-incubation at room temperature, 30 mL of Sub-
strates mix in AB were added in each well to a final volume of 60 mL.
This reaction mixture was then composed of 2 nm GmhA, 3 mm sedohep-
tulose-7-phosphate (Sigma), 3 mm ATP (Sigma) and 50 nm HldE of E.
coli in assay buffer. After 30 min of incubation at room temperature,
100 mL of the revelation mix were added to a final volume of 160 mL, in-
cluding the following constituents at the respective final concentrations:
10000 light unitsmLꢀ1 luciferase (Sigma), 20 mm d-luciferin (Sigma),
Chem. Eur. J. 2011, 17, 11305 – 11313
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11311

S. P. Vincent et al.
100 mm N-acetylcysteamine (Aldrich). Luminescence intensity was imme-
diately measured on Luminoskan (Thermofischer) and converted into in-
hibition percentages. For IC50 determinations, the inhibitor was tested at
6 to 10 different concentrations, and the related inhibitions were fitted to
a classical langmuir equilibrium model using XLFIT (IDBS).
[16] F. Moreau, N. Desroy, J. M. Genevard, V. Vongsouthi, V. Gerusz, G.
Le Fralliec, C. Oliveira, S. Floquet, A. Denis, S. Escaich, K. Wolf,
M. Busemann, A. Aschenbrenner, Bioorg. Med. Chem. Lett. 2008,
18, 4022–4026.
[17] A. Mayer, M. E. Tanner, Biochemistry 2007, 46, 6149–6155.
[18] T. Kowatz, J. P. Morrison, M. E. Tanner, J. H. Naismith, Protein Sci.
2010, 19, 1337–1343.
Inhibition of the enzymatic activity of HldE (luminescent assay on kinase
activity): The assay buffer “AB” contained 50 mm Hepes pH7.5, 1 mm
MnCl2, 25 mm KCl, 0.012% Triton-X100 and 1 mm dithiothreitol (DTT)
and 0.1 mm Myelin basic protein (MBP). The following components were
added in a white polystyrene Costar plate up to a final volume of 30 mL:
10 mL inhibitor dissolved in DMSO/water 50:50, and 20 mL HldE of E.
coli in AB. After 30 min of pre-incubation at room temperature, 30 mL of
substrate mix in AB were added in each well to a final volume of 60 mL.
This reaction mixture was then composed of 3 nm HldE, 0.2 mm of hep-
tose-7-phosphate H7P 2 and 0.2 mm ATP (Sigma) in assay buffer. After
30 min of incubation at room temperature, 200 mL of the revelation mix
were added to a final volume of 260 mL, including the following constitu-
ents at the respective final concentrations: 5000 light unitsmLꢀ1 lucifer-
ase (Sigma), 30 mm d-luciferin (Sigma), 100 mm N-acetylcysteamine (Al-
drich). Luminescence intensity was immediately measured on Luminos-
kan (Thermofischer) and converted into inhibition percentages. For IC50
determinations, the inhibitor was tested at 6 to 10 different concentra-
tions, and the related inhibitions were fitted to a classical Langmuir equi-
librium model using XLFIT (IDBS).
[19] J. P. Morrison, J. A. Read, W. G. Coleman, M. E. Tanner, Biochem-
istry 2005, 44, 5907–5915.
[20] J. P. Morrison, M. E. Tanner, Biochemistry 2007, 46, 3916–3924.
[21] J. A. Read, R. A. Ahmed, J. P. Morrison, W. G. Coleman, M. E.
Tanner, J. Am. Chem. Soc. 2004, 126, 8878–8879.
[22] J. A. Read, R. A. Ahmed, M. E. Tanner, Org. Lett. 2005, 7, 2457–
2460.
[23] A. M. Deacon, Y. S. Ni, W. G. Coleman, S. E. Ealick, Structure 2000,
8, 453–462.
[24] H. Paulsen, M. Pries, J. P. Lorentzen, Liebigs Ann. Chem. 1994,
389–397.
[25] G. P. De Leon, N. H. Elowe, K. P. Koteva, M. A. Valvano, G. D.
Wright, Chem. Biol. 2006, 13, 437–441.
[26] N. Desroy, F. Moreau, S. Briet, G. Le Fralliec, S. Floquet, L. Durant,
V. Vongsouthi, V. Gerusz, A. Denis, S. Escaich, Bioorg. Med. Chem.
2009, 17, 1276–1289.
[27] H. H. Nguyen, L. B. Wang, H. Huang, E. Peisach, D. Dunaway-Ma-
riano, K. N. Allen, Biochemistry 2010, 49, 1082–1092.
[28] P. L. Taylor, K. M. Blakely, G. P. De Leon, J. R. Walker, F. McAr-
thur, E. Evdokimova, K. Zhang, M. A. Valvano, G. D. Wright, M. S.
Junop, J. Biol. Chem. 2007, 283, 2835–2845.
Please see the Supporting Information for synthetic experimental details
and analytical data of all compounds.
[29] N. J. Harmer, J. Mol. Biol. 2010, 400, 379–392.
[30] J. Seetharaman, K. R. Rajashankar, V. Solorzano, R. Kniewel, C. D.
Lima, J. B. Bonanno, S. K. Burley, S. Swaminathan, Proteins Struct.
Funct. Bioinf. 2006, 63, 1092–1096.
Acknowledgements
[31] J. Hannson, S. Oscarson, Curr.Org. Chem. 2000, 4, 535–564.
[32] S. Oscarson, Carbohydr. Polym. 2001, 44, 305–311.
[33] T. Ohara, A. Adibekian, D. Esposito, P. Stallforth, P. H. Seeberger,
Chem. Commun. 2010, 46, 4106–4108.
This work was funded by FNRS (FRFC grant 2.4.625.08.F and “Crꢀdit au
chercheur” 1.5.167.10) and a Marie Curie RTN grant (2005-019561 “Dy-
namic”). We are also grateful to Mutabilis S.A. for their financial and
technical supports.
[34] N. C. R. van Straten, N. M. A. J. Kriek, C. M. Timmers, S. C. M. Wig-
chert, G. A. van der Marel, J. H. van Boom, J. Carbohydr. Chem.
1997, 16, 947–966.
[35] D. Crich, A. Banerjee, Org. Lett. 2005, 7, 1395–1398.
[36] D. Crich, A. Banerjee, J. Am. Chem. Soc. 2006, 128, 8078–8086.
[37] L. Dong, J. M. Roosenberg, M. J. Miller, J. Am. Chem. Soc. 2002,
124, 15001–15005.
[38] H. Kubo, K. Ishii, H. Koshino, K. Toubetto, K. Naruchi, R. Yamasa-
ki, Eur. J. Org. Chem. 2004, 1202–1213.
[39] G. J. P. H. Boons, M. Overhand, G. A. van der Marel, J. H. van
Boom, Carbohydr. Res. 1989, 192, c1–c4.
[1] E. T. Rietschel, T. Kirikae, F. U. Schade, U. Mamat, G. Schmidt, H.
Loppnow, FASEB J. 1994, 8, 217–225.
[2] S. Gronow, H. Brade, J. Endotoxin Res. 2001, 7, 3–23.
[3] A. Silipo, A. Molinaro, T. Ierano, A. De Soyza, L. Sturiale, D. Gar-
ozzo, C. Aldridge, P. A. Corris, C. M. A. Khan, R. Lanzetta, M. Par-
rilli, Chem. Eur. J. 2007, 13, 3501–3511.
[4] P. Kosma, Curr. Org. Chem. 2008, 12, 1021–1039.
[5] W. G. Coleman, L. Leive, J. Bacteriol. 1979, 139, 899–910.
[6] C. R. H. Raetz, C. Whitfield, Annu. Rev. Biochem. 2002, 71, 635–
700.
[7] S. Escaich, Curr. Opin. Chem. Biol. 2008, 12, 400–408.
[8] M. A. Valvano, P. Messner, P. Kosma, Microbiology 2002, 148,
1979–1989.
[40] G. J. P. H. Boons, G. A. van der Marel, J. T. Poolman, J. H. van
Boom, Recl. Trav. Chim. Pays-Bas 1989, 108, 339–343.
[41] M. Dasser, F. Chretien, Y. Chapleur, J. Chem. Soc. Perkin Trans. 1
1990, 3091–3094.
[42] K. Dziewiszek, A. Banaszek, A. Zamojski, Tetrahedron Lett. 1987,
28, 1569–1572.
[9] L. Cegelski, G. R. Marshall, G. R. Eldridge, S. J. Hultgren, Nat. Rev.
Microbiol. 2008, 6, 17–27.
[43] M. Kim, B. Grzeszczyk, A. Zamojski, Tetrahedron Lett. 2000, 43,
9313–9337.
[44] C. Bernlind, S. Bennett, S. Oscarson, Tetrahedron: Asymmetry 2000,
11, 481–492.
[45] C. Bernlind, S. Oscarson, J. Org. Chem. 1998, 63, 7780–7788.
[46] M. Jørgensen, E. H. Iversen, R. Madsen, J. Org. Chem. 2001, 66,
4625–4629.
[47] J. S. Brimacombe, A. K. M. S. Kabir, Carbohydr. Res. 1988, 179, 21–
30.
[10] H. Dohi, R. Pꢀrion, M. Durka, M. Bosco, Y. Rouꢀ, F. Moreau, S.
Grizot, A. Ducruix, S. Escaich, S. P. Vincent, Chem. Eur. J. 2008, 14,
9530–9539.
[11] S. Grizot, M. Salem, V. Vongsouthi, L. Durand, F. Moreau, H. Dohi,
S. Vincent, S. Escaich, A. Ducruix, J. Mol. Biol. 2006, 363, 383–394.
[12] M. Durka, K. Buffet, J. Iehl, M. Holler, J.-F. Nierengarten, J. Tagan-
na, J. Bouckaert, S. P. Vincent, Chem. Commun. 2011, 47, 1321–
1323.
[13] B. Kneidinger, C. Marolda, M. Graninger, A. Zamyatina, F. McAr-
thur, P. Kosma, M. A. Valvano, P. Messner, J. Bacteriol. 2002, 184,
363–369.
[48] H. Gꢃzlek, A. Graziani, P. Kosma, Carbohydr. Res. 2005, 340, 2808–
2811.
[49] G. J. P. H. Boons, M. Overhand, G. A. van der Marel, J. H. van
Boom, Angew. Chem. 1989, 101, 1538–1539; Angew. Chem. Int. Ed.
Engl. 1989, 28, 1504–1506.
[14] A. Zamyatina, S. Gronow, C. Oertelt, M. Puchberger, H. Brade, P.
Kosma, Angew. Chem. 2000, 112, 4322–4325; Angew. Chem. Int. Ed.
2000, 39, 4150–4153.
[15] A. Graziani, H. Amer, A. Zamyatina, A. Hofinger, P. Kosma, Tetra-
hedron: Asymmetry 2007, 18, 115–122.
[50] I. Fleming, P. E. J. Sanderson, Tetrahedron Lett. 1987, 28, 4229–
4232.
11312
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 11305 – 11313

Antivirulence Molecules Targeting Lipopolysaccharide Biosynthesis
FULL PAPER
[51] G. J. P. H. Boons, C. J. J. Elie, G. A. van der Marel, J. H. van Boom,
Tetrahedron Lett. 1990, 31, 2197–2200.
[55] B. Grzeszczyk, O. Holst, A. Zamojski, Carbohydr. Res. 1996, 290, 1–
15.
[56] S. Vidal, C. Vidil, A. Morere, M. Garcia, J. L. Montero, Eur. J. Org.
Chem. 2000, 3433–3437.
[52] G. J. P. H. Boons, G. A. Vandermarel, J. H. van Boom, Tetrahedron
Lett. 1989, 30, 229–232.
[57] J. K. Cha, W. J. Christ, Y. Kishi, Tetrahedron 1984, 40, 2247–2255.
[58] M. Durka, J. Wouters, S. P. Vincent, unpublished results.
[59] H. B. Borꢀn, K. Eklind, P. J. Garegg, B. Lindberg, A. Pilotti, Acta
Chem. Scand. 1972, 26, 4143.
[60] Z. Szurmai, L. Baaltoni, A. Liptak, Carbohydr. Res. 1994, 254, 301–
309.
[61] P. C. B. Page, M. J. McKenzie, J. A. Gallagher, J. Org. Chem. 2001,
66, 3704–3708.
[62] B. Grzeszczyk, O. Holst, S. Mꢃller-Loennies, A. Zamojski, Carbo-
hydr. Res. 1998, 307, 55–67.
[53] S. V. Ley, A. Abad-Somovill, J. C. Anderson, C. Ayats, R. Bꢄnteli,
E. Beckmann, A. Boyer, M. G. Brasca, A. Brice, H. B. Broughton,
B. J. Burke, E. Clenator, D. Craig, A. A. Denholm, R. M. Denton, T.
Durand-Reville, L. B. Gobbi, M. Gçbel, B. L. Gray, R. B. Gross-
mann, C. E. Gutteridge, N. Hahn, S. L. Harding, D. C. Jennens, L.
Jennens, P. J. Lovell, H. J. Lovell, M. L. de la Puente, H. C. Kolb, W.-
J. Koot, S. L. Maslen, C. F. McCusker, A. Mattes, A. R. Pape, A.
Pinto, D. Santafianos, J. S. Scott, S. C. Smith, A. Q. Somers, C. D.
Spilling, F. Stelzer, P. L. Toogood, R. M. Turner, G. E. Veitch, A.
Wood, C. Zumbrunn, Chem. Eur. J. 2008, 14, 10683–10704.
[54] A. Stewart, C. Bernlind, A. Martin, S. Oscarson, J. C. Richards,
E. K. H. Schweda, Carbohydr. Res. 1998, 313, 193–202.
Received: February 4, 2011
Revised: May 26, 2011
Published online: August 26, 2011
Chem. Eur. J. 2011, 17, 11305 – 11313
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11313