Inhibition of lipid A stimulated activation of human dendritic cells and macrophages by amino and hydroxylamino monosaccharides

Article abstract of DOI:10.1002/anie.200604932

(Figure Presented) Innovative mimics of lipid A were synthesized and shown to efficiently inhibit lipid A induced cytokine production in human dendritic cells and macrophages. These compounds, which consist of a D-glucose unit functionalized with an amine

Full text of DOI:10.1002/anie.200604932

Communications
DOI: 10.1002/anie.200604932
Drug Discovery
Inhibition of Lipid A Stimulated Activation of Human Dendritic Cells
and Macrophages by Amino and Hydroxylamino Monosaccharides
Francesco Peri,* Francesca Granucci, Barbara Costa, Ivan Zanoni, Chiara Marinzi, and
Francesco Nicotra
Sepsis and septic shock are serious clinical
syndromes linked with high mortality
rates[1] and caused by the presence of
circulating microbial antigens in the blood
of affected patients.[2] The development of
septicaemia is often associated with a
systemic inflammatory response to bacte-
rial lipopolysaccharides (LPSs). The toxic-
ity of LPSs lies in the membrane-anchoring
moiety named lipid A.[3] Lipid A is the
lipophilic portion of bacterial LPSs. Its
general chemical structure consists of a b-
(1!6)-linked disaccharide core with two
phosphate esters (at positions C1 and C4’)
and an R-3-hydroxy fatty acid at ester and
amide linkages; some of these fatty acids
are acylated at the 3-hydroxy group
(Scheme 1A).
Lipid A units of different bacterial
origin differ in their chemical structure
(mainly in the ramification of lipophylic
chains) and display a variety of biological
activities. There are highly toxic forms (for
example, hexa-acylated lipid A from
Escherichia coli or Salmonella species)
that stimulate massive cytokine production,
less toxic forms, and even variants that have
antagonistic properties towards cytokine
stimulation. Biophysical studies on the
Scheme 1. Structures of A) Escherichia coli lipid A, B) the synthetic antagonist E 5564, C) the
monosaccharide mimic GLA-58, and D) a synthetic disaccharide antagonist. Glucose
three-dimensional conformation adopted
by different lipid A structures have
revealed that under physiological condi-
derivatives 1, 2, and 3.
tions the most active (toxic) forms have a conical shape,
whereas a lack of activity or antagonistic properties are
associated with a cylindrical structure.[4] The different 3D
structures of lipid A are thought to influence the geometry of
the complex with the cellular target, the Toll-like receptor 4
(TLR4), and therefore the intensity and the nature of the
biological activity of the LPS.
Great effort has been devoted to the synthesis of lipid A
mimics, which have shown activity as agonists or antagonists
in the TLR4-activation process. The phosphorylated disac-
charide core plays a fundamental role in the biological
activity; it is therefore present in the majority of lipid A
synthetic mimics, as in the case of compound E 5564
(Scheme 1B). This potent lipid A antagonist developed by
the Eisai Group was found to be a good antisepsis drug in
phase II clinical trials.[5] Very few compounds are known that
lack the disaccharide structure but maintain activity as lipid A
agonists or antagonists.[6] Monosaccharide lipid A analogues,
including compound GLA-58 (Scheme 1C), preserved LPS-
mimetic activity despite the absence of the second, non-
reducing monosaccharide moiety.[7] A series of reducing N-
acylated glucosylamines with a phosphate group at C4 and
linear or branched 3-hydroxytetradecanoic acid ester and
amide chains at C2 and C3 were recognized by murine
[*] Prof. F. Peri, Prof. F. Granucci, Dr. B. Costa, Dr. I. Zanoni,
Dr. C. Marinzi, Prof. F. Nicotra
Department of Biotechnology and Biosciences
University of Milano–Bicocca
Piazza della Scienza, 2, 20126 Milano (Italy)
Fax : (+39)02-6448-3565
E-mail: francesco.peri@unimib.it
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
3308
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 3308 –3312

Angewandte
Chemie
macrophages as both LPS agonists and antag-
onists. Furthermore, compound GLA-58 was
shown to be active as an LPS antagonist in
human cells. Interestingly, it was found
recently that some substances with chemical
structures totally unrelated to that of lipid A,
such as taxol, flavolipin,[8] and vitamin D,[9]
have LPS-mimetic activity towards TLR4.
Herein we present the synthesis and unex-
pected biological activity of the glucose-
derived amine 1, ammonium species 2, and
hydroxylamine
3 (Scheme 1). These com-
pounds, unlike other lipid A mimics synthe-
sized so far, have a monosaccharide structure
and are devoid of phosphate esters. In a
previous report, we showed that a nonphos-
phorylated
N(OMe)-linked
disaccharide
(Scheme 1D) inhibits lipid A induced cytokine
production in MT2 macrophages in a dose-
dependent manner.[10]
Compound 3 was first obtained serendip-
itously when we attempted the chemoselective
synthesis of this disaccharide by the direct
condensation of two unprotected monosac-
charides. Unexpectedly, the reaction of the
methoxyamino monosaccharide 10 (Scheme 2)
with traces of dihydrofuran (DHF) present in
the solvent (THF) led to the formation of the
tetrahydro adduct 3 in very low quantities. The
ability of 3 to inhibit lipid A activity was tested
Scheme 3. Synthesis of monosaccharides 1–3: a) anisaldehyde dimethylacetal, CSA,
DMF, 94%; b) C14H29Br, NaH, DMF, 74%; c) LiAlH4, AlCl3, CH2Cl2/Et2O, 86%; d) Dess–
Martin periodinane, CH2Cl2; e) cyclopentylamine, NaBH3CN, AcOH, CH2Cl2, MeOH,
75% (two steps); f) TFA, CH2Cl2, 80%; g) CH3I, Na2CO3, DMF, 94%; h) O-methylhydrox-
ylamine hydrochloride, pyridine, 65% from 6; i) NaBH3CN, glacial AcOH, 92%; j) TFA,
CH2Cl2, 90%; k) DHF, PPTS, CH2Cl2, 65%. DMF=N,N-dimethylformamide, TFA=tri-
on dendritic cells (DCs) and macrophages fluoroacetic acid.
furnished 5 (74% yield), the benzylidene ring in which was
opened regioselectively by treatment with LiAlH4 and AlCl3
to give 6 in 86% yield.[10] The oxidation of 6 with Dess–Martin
periodinane gave the corresponding aldehyde 7, which was
converted directly into 8 by reductive amination with cyclo-
pentylamine and NaBH3CN (75% yield over two steps).
Acidic cleavage of the p-methoxybenzyl (PMB) group at C4
of 8 gave 1 in 80% yield; the secondary cyclopentylamine at
C6 was converted into the quaternary ammonium ion by
treatment with methyl iodide and sodium carbonate to afford
compound 2 in 94% yield. Alternatively, aldehyde 7 was
treated with O-methylhydroxylamine hydrochloride in pyri-
dine to give the corresponding C6 methyloxime in 65% yield
from 6. The methyloxime was then reduced to the methylhy-
droxylamine 9 by treatment with sodium cyanoborohydride in
glacial acetic acid (92% yield). The PMB ether was cleaved
with TFA/CH2Cl2 to afford 10 (90% yield), which was finally
treated with DHF in the presence of the acid catalyst
pyridinium p-toluenesulfonate (PPTS) to give the target
monosaccharide 3 in 65% yield.
Scheme 2. Unexpected reaction of the monosaccharide 10 with dihy-
drofuran impurities present in THF.
(MFs). Modest activity was found, unfortunately accompa-
nied by significant chemical instability, particularly under
acidic conditions, owing to the presence of an N,O-acetal
group.[11] We then developed a synthetic route for the gram-
scale synthesis of compounds 1 and 2, two chemically stable
analogues of 3 (Scheme 3). Compound 1 has a cyclopentyl-
amine group linked to C6, and compound 2 is the corre-
sponding dimethyl ammonium salt, with a permanent positive
charge on the nitrogen atom.
The synthesis of all compounds started with the con-
version of commercially available methyl a-d-glucopyrano-
side into the methyl 4,6-O-(4-methoxybenzylidene)-a-d-glu-
copyranoside 4 in 94% yield by treatment with anisaldehyde
dimethylacetal and camphorsulfonic acid (CSA). The alkyla-
tion of 4 with tetradecyl bromide in the presence of NaH
The biological activities of compounds 1–3 were inves-
tigated by analyzing their ability to interfere with lipid A
induced DC and MF activation. These two special classes of
leukocytes are involved directly in boosting and controlling
inflammatory responses and, following encounter with micro-
bial products, such as LPSs, produce large amounts of TNFa,
Angew. Chem. Int. Ed. 2007, 46, 3308 –3312
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3309

Communications
Figure 1. Antagonistic activity of monosaccharides 1–3 towards BMDC, BMF, and D1 cells. TNFa production was measured in the supernatants
after 24 h. From right to left: lipid A (0.5 mm): cells treated with lipid A alone; DMSO: cells incubated in the presence of the complete medium
plus DMSO; inhibitor (50 mm): cells treated only with monosaccharides 1–3; other columns: lipid A plus compounds 1–3 in increasing doses. The
data represent the mean and standard deviation for triplicate wells. The results are representative for three independent experiments. *, p<0.05;
**, p<0.01; ***, p<0.001, where p is the statistical probability value.
one of the principal mediators of inflammation and septic
shock. In particular, DCs are distributed in tissues that
interface with the external environment, such as the skin, gut,
and lungs, where they can perform a sentinel function for
incoming pathogens, and have the capacity to recruit and
activate cells of the innate immune system.[12] After microbial
uptake has occurred, DCs undergo a complex process of
maturation with the sequential acquisition of different
immune-regulatory functions. Once the maturation process
initiated by direct exposure to TLR ligands has been
completed, DCs become efficient activators of acquired
immune responses. In this context, LPS activation of TLR4
is one of the most potent stimuli for DC priming.[13]
Because of the pivotal role of early activated DCs and
MFs in directing inflammatory reactions, we tested whether
the two compounds could antagonize the stimulatory activity
of lipid A on bone-marrow-derived macrophages (BMFs)
and DCs (BMDCs) by interfering with their capacity to
induce TNFa production. As a source of DCs, we also used
D1 cells, a DC line that is dependent on the long-term growth
factor GM-CSF (granulocyte macrophage colony-stimulating
factor).[14] BMF, BMDC, and D1 cells were stimulated with
lipid A (0.5 mm) after preexposure to compounds 1–3, and the
amounts of TNFa released in the supernatant were measured
by ELISA. As shown in Figure 1, monosaccharides 1 and 2
interfered with lipid A activity in a dose-dependent manner,
whereas compound 3 showed consistently lower activity.
Monosaccharide 10, which lacks the tetrahydrofuran ring, was
totally inactive. In general, monosaccharides 1–3 did not show
any direct inflammatory function, as they are not capable of
stimulating cytokine production. Compound 2 was more
potent than 1, which was in turn a better inhibitor than 3.
Moreover, compound 2 had the highest solubility in the
aqueous media used in our tests. For these reasons, amines 1
and 2 seemed the best candidates for drug development,
whereas the chemical instability, sparing solubility, and lower
potency of hydroxylamine 3 render it a less interesting lead.
The activity of the most promising lead, compound 2, was
then analyzed in more detail by monitoring the production of
a second inflammatory cytokine, IL-1b. IL-1b and TNFa
synthesis was inhibited in BMF and D1 cells, and the
inhibitory effect was proportional to the concentration of
the inhibitor. To evaluate the selectivity for TLR4, TNFa
production was investigated in the presence of compound 2 in
response to stimulation by the CpG motif of bacterial DNA,
which is recognized by TLR9,[15] and the tripalmitoyl cysteine
(Pam3Cys-SK4) moiety, which is specific for TLR2.[16] Cyto-
kine production was not inhibited by 2 in either the TLR9- or
the TLR2-mediated inflammatory cascade (Figure 2), thus
indicating a relevant level of selectivity.
We then investigated the cytotoxic potential of the
monosaccharides. Compounds 1–3 are composed of a polar
part (the sugar moiety) linked to lipophilic chains, and can
therefore act as detergents and kill cells by generating pores
in the plasma membranes. The toxicity of compounds 1–3 was
investigated with the PI test, the apoptotic potential with the
annexinV test. Neither for the DCs nor for the MFs was an
appreciable percentage of dead cells found after incubation
for 48 h with 2 at concentrations between 10 and 100 mm
(Figure 3).
Further direct evidence that the activity of compound 2 is
selective for the TLR4 receptor was obtained with experi-
3310
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 3308 –3312

Angewandte
Chemie
counteract significantly the
effect of lipid A in TLR4-
transfected cells, whereas it
was inactive at the maximal
dose of 50 mm in influencing
the effect of CpG on TLR9-
transfected
(Figure 4).
HEK 293
The biological data pre-
sented allow some prelimi-
nary considerations on the
structure–activity relation-
ship of this new class of
lipid A antagonists. The
presence of
a five-mem-
Figure 2. Selectivity of the monosaccharide 2 for TLR4. A) The antagonistic activity of compound 2 was
investigated by testing its ability to interfere with IL-1b production 24 h after stimulation with lipid A. The
quantity of IL-1b was measured in the supernatants by ELISA. Lipid A (0.5 mm): cells treated with lipid A
alone; NT: cells incubated in the presence of the complete medium; DMSO: cells incubated in the presence
of the complete medium plus DMSO; inhibitor (50 mm): cells treated only with the monosaccharide 2; other
columns: lipid A plus compound 2 in increasing doses. The data represent the mean and standard deviation
for triplicate wells. The results are representative for three independent experiments. B) TNFa production
measured by ELISA 24 h after stimulation with CpG or PAM3Cys in the presence of compound 2. NT: cells
incubated in the presence of the complete medium plus DMSO; other columns: cells treated with CpG or
Pam3Cys-SK4 in the presence or absence of the inhibitor 2 (50 mm). *, p<0.05; **, p<0.01; ***, p<0.001.
bered ring attached to the
nitrogen atom at C6 appears
to be fundamental to the
activity of these com-
pounds; compound 10,
which lacks this ring, is
completely inactive. The
permanent positive charge
on the nitrogen atom of 2
has a positive effect on both
activity and solubility.
ments on a TLR4- and TLR9-transfected HEK 293 cell
system.[17] In these cells, TLR-dependent activation of NF-kB
(nuclear factor kB) can be measured readily after stimulation.
The activation of the TLR4 signaling pathway leads to NF-kB
nuclear translocation and inflammatory cytokine produc-
tion.[18] TLR4- and TLR9-transfected HEK 293 cells were
incubated with lipid A (0.5 mm) and CpG (0.5 mm), respec-
tively, after preexposure to compound 2, and the activation of
NF-kB was measured 2 h later. Monosaccharide 2 was able to
Monosaccharides 1–3 and the other lipid A mimics
described to date display remarkable structural diversity.
Therefore, the experimental evidence for the interaction of
these monosaccharides with TLR4 could open interesting
perspectives in both drug discovery and the characterization
Figure 4. Antagonistic activity of monosaccharide 2. HEK 293 cells
transfected stably with the human TLR4A or TLR9 gene were treated
with lipid A alone or lipid A in the presence of monosaccharide 2, and
with CpG alone or CpG in the presence of monosaccharide 2,
respectively. NF-kBactivation was evaluated in the nuclear extract 2 h
later, and is expressed as optical density (OD) at 450 nm. NT:
nontreated cells. The data represent the mean and standard error for
duplicate wells. The results are representative for three independent
experiments. *, p<0.001 versus NT cells; 8, p<0.001 versus lipid A
alone (ANOVA (analysis of variance); Tukey test).
Figure 3. Compound 2 does not induce the death of MFs or DCs.
Mouse DCs and MFs were either left untreated or treated for 24 h
with compound 2 at the indicated concentrations. Cells were then
analyzed by FACS for the presence of annexinV and propidium iodide
(PI) double-positive cells. A) percentage of annexinV and PI double-
positive cells; B) percentage of annexinV and PI double-negative cells.
Angew. Chem. Int. Ed. 2007, 46, 3308 –3312
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3311

Communications
pholipid dimer connected through an acyclic backbone: E. Lien,
J. C. Chow, L. D. Hawkins, P. D. McGuinness, K. Miyake, T.
Espevik, F. Gusovsky, D. T. Golenbock, J. Biol. Chem. 2001, 276,
1873 – 1880.
of TLR4–ligand molecular recognition and binding. We are
currently testing the activity of 2 and of molecules with similar
structures in different biological contexts. Recent publica-
tions outlined the central role of TLR4 in a number of
important pathological events, such as the modulation of a
series of immunological and allergic reactions[19] and pain
transmission.[20] We are therefore convinced that compound 2
and other molecules capable of binding TLR4 could be used
as antisepsis drugs as well as leads for the development of
pain-relief and antiallergic drugs with innovative structures.
[7] a) M. Matsuura, M. Kiso, A. Hasegawa, Infect. Immun. 1999,
6282 – 6292; b) R. Tamai, Y. Asai, M. Hashimoto, K. Fukase, S.
Kusumoto, H. Ispida, M. Kiso, T. Ogawa, Immunology 2003, 110,
66 – 72.
[8] K. Kawasaki, K. Gomi, Y. Kawai, M. Shiozaki, M. Nishijima, J.
Endotoxin Res. 2003, 9, 301 – 307.
[9] O. Equils, Y. Maiki, A. M. Shapiro, K. Michelsen, D. Lu, J.
Adams, S. Jordan, Clin. Exp. Immunol. 2006, 143, 58 – 64.
[10] F. Peri, C. Marinzi, M. Barath, F. Granucci, M. Urbano, F.
Nicotra, Bioorg. Med. Chem. 2006, 14, 190 – 199.
[11] The half-life of 3 is about 12 h at pH 4.5, as assessed by HPLC
kinetic analysis.
[12] L. Zitvogel, J. Exp. Med. 2002, 195, 9F – 14F.
[13] F. Granucci, C. Vizzardelli, N. Pavelka, S. Feau, M. Persico, E.
Virzi, M. Rescigno, G. Moro, P. Ricciardi-Castagnoli, Nat.
Immunol. 2001, 2, 882 – 888.
[14] C. Winzler, P. Rovere, M. Rescigno, F. Granucci, G. Penna, L.
Adorini, V. S. Zimmermann, J. Davoust, P. Ricciardi-Castagnoli,
J. Exp. Med. 1997, 185, 317 – 327.
[15] H. Hemmi, O. Takeuchi, T. Kawai, T. Kaisho, S. Sato, H. Sanjo,
M. Matsumoto, K. Hoshino, H. Wagner, K. Takeda, S. Akira,
Nature 2000, 408, 740 – 745.
[16] S. Agrawal, A. Agrawal, B. Doughty, A. Gerwitz, J. Blenis, T.
Van Dyke, B. Pulendran, J. Immunol. 2003, 171, 4984 – 4989.
[17] J. C. Chow, D. W. Young, D. T. Golenbock, W. J. Christ, F.
Gusovsky, J. Biol. Chem. 1999, 274, 10689 – 10692.
[18] K. Takeda, T. Kaisho, S. Akira, Annu. Rev. Immunol. 2003, 21,
335 – 376.
Received: December 6, 2006
Published online: March 23, 2007
Keywords: biological activity · carbohydrates · drug design ·
.
immunology · medicinal chemistry
[1] a) J. Cohen, Nature 2002, 420, 885 – 891; b) D. C. Angus, Crit.
Care Med. 2001, 29, 1303 – 1310.
[2] a) R. C. Bone, Chest 1991, 100, 802 – 808; b) J. E. Parrillo, N.
Engl. J. Med. 1993, 328, 1471 – 1477.
[3] a) O. Westphal, O. Lüderitz, Angew. Chem. 1954, 66, 407 – 417;
b) E. T. Rietschel, T. Kirikae, F. U. Schade, U. Mamat, G.
Schmidt, H. Loppnow, A. J. Ulmer, U. Zähringer, U. Seydel, F.
Di Padova, M. Schreier, H. Brade, FASEB J. 1994, 8, 217 – 225.
[4] a) U. Seydel, M. Oikawa, K. Fukase, S. Kusumoto, K. Branden-
burg, Eur. J. Biochem. 2000, 267, 3032 – 3039; b) M. G. Netea, M.
van Deuren, B. J. Kullberg, J.-M. Cavaillon, J. W. M. van der
Meer, Trends Immunol. 2002, 23, 135 – 139.
[19] R. F. Tsuji, K. Hoshino, Y. Noro, N. M. Tsuji, T. Kurokawa, T.
Masuda, S. Akira, B. Nowak, Clin. Exp. Allergy 2003, 33, 249 –
258.
[20] F. Y. Tanga, N. Nutile-McMenemy, J. A. DeLeo, Proc. Natl.
Acad. Sci. USA 2005, 102, 5856 – 5861.
[5] a) T. Glück, S. M. Opal, Drugs 2004, 64, 837 – 859; b) W. J.
Christ, L. D. Hawkins, D. Lynn, M. D. Lewis, Y. Kishi, Carbo-
hydr.-Based Drug Discovery, 2003, 1, 341 – 355.
[6] An interesting example of a lipid A agonist without a disacchar-
ide structure is compound ER-112022, which contains a phos-
3312
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 3308 –3312