Synthesis of diacylated trehalose sulfates: Candidates for a tuberculosis vaccine

Article abstract of DOI:10.1002/anie.200803835

Able antigen analogues: Analogues of the mycobacterial sulfoglycolipid antigen 1 isolated from Mycobacterium tuberculosis have been prepared by a short and general route. Some derivatives showed interesting immunogenic properties and activated cytotoxic T

Full text of DOI:10.1002/anie.200803835

Communications
DOI: 10.1002/anie.200803835
Vaccines
Synthesis of Diacylated Trehalose Sulfates: Candidates for a
Tuberculosis Vaccine**
Julie Guiard, Anthony Collmann, Martine Gilleron, Lucia Mori, Gennaro De Libero,
Jacques Prandi,* and Germain Puzo
Tuberculosis remains a major world-wide health problem and
results in the loss of almost 2 million lives annually, with the
majority of deaths occurring in the developing world.[1]
Despite the discovery of active antibiotics in the 1960s and
the production of the Bacillus Calmette–Guꢀrin (BCG)
vaccine in the early 20th century, tuberculosis is still not
under control. The increasing incidence of human immuno-
deficiency virus (HIV) epidemics and the emergence of
multidrug-resistant strains of Mycobacterium tuberculosis, the
causative agent of tuberculosis, impair the eradication of the
disease. New therapeutic approaches and the development of
new vaccines to fight tuberculosis are thus urgently needed.
Recently, a diacylated sulfoglycolipid, acyl2SGL (1),[2] was
characterized and identified as a new mycobacterial antigen
able to stimulate populations of CD1b-restricted human
T lymphocytes during infection with M. tuberculosis. Further-
more, these acyl2SGL-specific activated T cells were shown
to: 1) release interferon-g (IFN-g), 2) recognize M. tuber-
culosis infected antigen-presenting cells, and 3) kill intra-
cellular mycobacteria in vitro.[2] In light of these properties,
sulfoglycolipid 1 seemed to be a promising candidate for the
development of a new tuberculosis vaccine. Compound 1 is
specific to the M. tuberculosis species and is not present in
M. bovis BCG, the species used in the current vaccine. The
structure of acyl2SGL encompasses an a,a-d-trehalose core,
which is esterified at the 2-position with a palmitic (or stearic)
acid, esterified at the 3-position with a hydroxyphthioceranoic
acid, and O-sulfated at the 2’-position. Hydroxyphthiocer-
anoic acids are a family of complex dextrorotatory fatty acids
specific to the Mycobacterium genus, which contain a hydroxy
group and methyl groups arranged in a 2,4,6 pattern.[3] All
methyl-substituted stereocenters are of the l series,[3] whereas
the configuration of the hydroxy-substituted carbon atom has
not been assigned.
Compound 1 was isolated in tiny amounts (about
1 mglꢀ1) from cultures of M. tuberculosis. Its low availability
limits its further development as a potential tuberculosis
vaccine. We therefore devised a synthetic route to this class of
diacylated sulfated trehalose compounds and hypothesized
that the hydroxyphthioceranoic acid, which is not readily
available from natural sources, might be replaced by simpler
fatty acids. Herein we report the preparation of various
sulfoglycolipid (SGL) analogues of the natural compound
acyl2SGL (1) in which the hydroxyphthioceranoic acid has
been replaced by less complex acids. Some of these analogues
were able, like the natural product, to activate the acyl2SGL-
specific T-cell clone with the production of (IFN-g). Of the
utmost importance, it was found that small modifications to
the structure of the hydroxyphthioceranoic acid substituent
can modulate the immunogenicity of the analogues. Recent
interest in the synthesis of mycobacterial sulfoglycolipids has
led to the preparation of a tetraacylated trehalose sulfate.[4,5]
The elaboration of the trisubstituted a,a-d-trehalose core
of the sulfoglycolipids on the basis of the pioneering synthetic
studies of Goren and co-workers, and Baer and Wu was
straightforward.[3,6,7] The known compound 4,6,4’,6’-dibenzy-
lidene a,a-d-trehalose (2)[8] was acylated selectively at the 2-
position (or 2’-position) by using palmitoyl chloride (or
another acyl chloride) in pyridine to give a trehalose
derivative 3 in 45% yield (Scheme 1). This direct acylation
reaction avoided the dibutylstannylene procedure,[6] which
involves toxic tin derivatives and which was found to be
difficult to carry out on a larger scale. The use of the
bifunctional reagent 1,3-dichloro-1,1,3,3-tetraisopropyldisi-
loxane (TIPSCl2) enabled the selective protection of the 2’-
and 3’-positions of 3 in one step and gave alcohol 4 in good
yield (60%). Owing to the steric crowding around the alcohol
functionality of 4, the nucleophilicity of the oxygen atom was
low, and the esterification of compound 4 with fatty-acid
derivatives proved to be difficult. The best yields of 5 (up to
[*] J. Guiard, Dr. M. Gilleron, Dr. J. Prandi, Dr. G. Puzo
Dꢀpartement des Mꢀcanismes Molꢀculaires des Infections
Mycobactꢀriennes, Institut de Pharmacologie et de Biologie
Structurale du CNRS, UMR 5089, Universitꢀ de Toulouse III
205 route de Narbonne, 31077 Toulouse Cedex (France)
Fax: (+33)5-61-17-59-94
E-mail: Jacques.Prandi@ipbs.fr
A. Collmann, Dr. L. Mori, Prof. Dr. G. De Libero
Experimental Immunology, Department of Biomedicine
University Hospital Basel
Hebelstrasse 20, 4031 Basel (Switzerland)
[**] Participating laboratories were funded by the 6th European Union
Framework Program (TB-VAC program, LSHP-CT-2003-503367).
M.G., J.P., and G.P. received funding from the Centre National de la
Recherche Scientifique (CNRS).
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200803835.
9734
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 9734 –9738

Angewandte
Chemie
lengths from C16 to C30, showed that they were unable to
activate T cells specific for the acyl2SGL complexes (data not
shown). As the only difference between these compounds and
the natural antigen is the nature of the 3-O-acyl group, these
data emphasized the crucial role of the hydroxyphthiocer-
anoyl group for CD1b presentation, T-cell-receptor (TCR)
engagement, and T-lymphocyte activation. We therefore
prepared polymethyl chiral fatty acids and incorporated
them into the synthesis of the sulfoglycolipids. Hydroxy-
phthioceranoic and phthioceranoic acids belong to the
deoxypolypropionate family of natural products. Numerous
approaches to the stereoselective construction of the 1,3,5…-
polymethyl pattern on the basis of enolate alkylation,[9–11]
conjugate addition,[12–14] or other methods[15–17] have been
reported. Recently, the first synthesis of phthioceranoic acid
was completed by using an iterative catalytic asymmetric 1,4-
conjugate addition of a Grignard reagent to an unsaturated
thioester.[18] As the synthesis of these polymethyl fatty acids is
linear and use of a cycle of reactions, the number of steps
involved, the reaction times, and the number of chromato-
graphic purification steps are important parameters for the
efficiency of the synthetic route. For these reasons, we chose
the asymmetric alkylation of hydrazone enolates derived
from (S)-1-amino-2-methoxymethylpyrrolidine (SAMP) with
iodides, a method developed by Enders and co-workers,[19,20]
as the whole cycle of four reactions can be carried out in less
than three days with only two chromatographic steps
(Scheme 2).
Scheme 1. General synthesis of sulfoglycolipids 9–26: a) R1COCl, pyri-
dine, DMAP, room temperature; b) TIPSCl2 (1.2 equiv), pyridine, room
temperature, 48 h; c) R2COCl (2.0 equiv), pyridine, DMAP, or R2COOH
(2.0 equiv), DCC (2.0 equiv), toluene, microwave irradiation; d) 1m
nBu4NF in THF, room temperature, 24 h; e) SO3·pyridine (1.5 equiv),
DMF, room temperature, 24 h; f) 1.7% aqueous H2SO4/CHCl3/MeOH,
room temperature, 24 h. R1 and R2 groups are shown in Table 1.
DCC=dicyclohexylcarbodiimide, DMAP=4-dimethylaminopyridine,
TIPS=tetraisopropyldisiloxane.
68%) were observed when the free fatty acid was activated
with dicyclohexylcarbodiimide in toluene under microwave
irradiation, but this method precluded the recycling of the
fatty acid. When precious chiral polymethyl fatty acids were
used, despite the lower acylation yields (26–54%), the
reaction was carried out more classically in pyridine after
activation of the acid by conversion into the acid chloride
through treatment with oxalyl chloride. The 2’- and 3’-
positions of diester 5 were unmasked with 1m tetrabutylam-
monium fluoride in THF to afford diol 6 in high yield
(> 90%), provided that the slightly basic commercial solution
had been neutralized to pH 6.5 (pH paper) with trifluoro-
acetic acid before use. Without this precaution, some scram-
bling of the acyl groups on the two glucose units of the
trehalose was observed. Selective 2’-O-sulfation of diol 6 with
the SO3·pyridine complex in anhydrous N,N-dimethylforma-
mide (DMF) at room temperature gave a mixture of 2’-O- and
3’-O-monosulfated trehalose derivatives. The observed regio-
selectivity was 5:1 in favor of the desired 2’-O-sulfate, and the
regioisomers were separated readily by chromatography on
silica gel. Compound 7 was isolated in 40–50% yield. The
synthesis of the SGLs was completed by removal of the
benzylidene protecting groups under carefully controlled
acidic conditions (chloroform/methanol/1.7% aqueous
H2SO4 60:40:8 (v/v/v), room temperature), and the final
products were isolated in almost quantitative yield. All
compounds in Table 1 were prepared by this route.
Scheme 2. General synthesis of polymethylated fatty acids: a) A(n)
(1.0 equiv), PPh3 (1.2 equiv), I2 (1.3 equiv), imidazole (3.1 equiv),
toluene, reflux, 1 h; b) 8 (2.0 equiv), LDA (2.0 equiv), THF, 08C, 1 h,
then B(n) (1.0 equiv), ꢀ50!ꢀ258C, overnight; c) 4m HCl, petroleum
ether, room temperature, 3 h; d) Jones reagent, acetone, 08C, 30 min;
e) (carbethoxyethylidene)triphenylphosphorane (1.2 equiv), CH2Cl2,
room temperature, overnight, then 3m KOH, H2O/EtOH, reflux;
f) BH3·THF complex (1.7 equiv), THF, 08C!RT, overnight. LDA=
lithium diisopropylamide.
The starting material used was the long-chain alcohol A(1),
which was obtained in 98% yield after the reduction with
[21]
BH3·THF of (2S)-2-methyloctadecanoic acid E(1) in THF.
The treatment of alcohol A(1) under conditions developed by
Garegg and Samuelsson (PPh3, imidazole, I2 in toluene)[22]
gave iodide B(1) in 93% yield. This iodide was treated with the
Preliminary testing of the antigenic properties of sulfo-
glycolipids 9–11, which contain linear fatty acids with chain
Angew. Chem. Int. Ed. 2008, 47, 9734 –9738
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
9735

Communications
Table 1: Synthetic analogues of 1.
Product
R1CO
R2CO
1
palmitoyl/stearoyl
9
10
11
palmitoyl
palmitoyl
palmitoyl
palmitoyl
triacontanoyl
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
palmitoyl
palmitoyl
palmitoyl
palmitoyl
palmitoyl
palmitoyl
palmitoyl
palmitoyl
palmitoyl
octanoyl
tetracosanoyl
palmitoyl
palmitoyl
palmitoyl
palmitoyl
lithium enolate of SAMP propanal hydrazone (8)[23] in THF
to give the 2S alkylated product C(2) in good yield (72%).
Owing to the very low solubility of the long-chain iodides B(n)
at ꢀ788C in THF, alkylation reactions had to be performed at
ꢀ508C with warming to ꢀ258C, and erosion of the optical
purity of the products occurred to a small extent. The
diastereomeric ratio for each alkylation was 94:6. The key
aldehyde D(2) was obtained almost quantitatively after acidic
hydrolysis of hydrazone C(2) in a petroleum ether/aqueous
HCl biphasic mixture at room temperature. This aldehyde
could then be oxidized quantitatively with the Jones
reagent[24] to the saturated acid E(2) or submitted to a Wittig
reaction with (carbethoxyethylidene)triphenylphosphorane
followed by saponification to give the unsaturated acid F(3)
9736 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 9734 –9738

Angewandte
Chemie
in 80–85% yield. Aldehyde D(2) could also be reduced with
the BH3·THF complex in THF to alcohol A(2) in quantitative
yield and the whole cycle repeated with comparable overall
yield for the introduction of additional methyl groups.
This procedure is short and efficient (only four steps and
60% overall yield for each cycle) and can be scaled up easily.
All polymethyl fatty acids prepared by this route are
presented in Table 1 (third column). They were used for the
synthesis of the corresponding sulfoglycolipids 13–26 accord-
ing to the general procedure described herein.
Sulfoglycolipids 9–26 were tested for their ability to
activate the acyl2SGL-specific T cells by following the
reported procedure.[2] Human dendritic cells that express
the CD1b protein were incubated with varying amounts of the
antigen, and the T-lymphocyte clone was added. The amount
of IFN-g released by the T cells was quantified, and dose-
dependent curves were obtained (Figure 1a). As also
groups was observed when a chiral polymethylated a,b-
unsaturated fatty acid was used in place of the hydroxy-
phthioceranoic acid moiety (see compounds 16–20). How-
ever, with these unsaturated acids, there was no further
improvement of the antigenic activity upon the introduction
of a further methyl group once four methyl groups had been
incorporated in the fatty-acid chain (compounds 19 and 20).
Other structural variations on the SGLs gave more insight
into the parameters which are crucial for the antigenic activity
of these analogues. The length and the respective positions of
the fatty acyl chains at the 2- and 3-positions of the trehalose
unit were found to be very important. Shortening of the 2-O-
acyl chain of the SGL from a palmitoyl (in 19) to an octanoyl
group gave compound 21, the potency of which was greatly
diminished with respect to that of 19. An even more
pronounced effect on the activity was observed when the 3-
O-acyl group was shortened from a trimethyltetracosanoyl
chain to a trimethyloctanoyl chain (compare the results for 18
and 25). Lengthening of the 2-O-acyl chain from a palmitoyl
(in 19) to a tetracosanoyl group (in 22) was also deleterious to
the T-cell-stimulatory activity. The immunogenicity was also
lost when the two acyl groups in 17 were permutated, as in 23.
Finally, the introduction of further unsaturation at the g,d-
position of the 3-O-fatty acid (in 26) and the use of a
methylated fatty acid with an R stereocenter (in 24) gave
inactive products.
A general and convenient synthesis of 2,3-di-O-acyl-2’-O-
sulfate trehalose derivatives has been developed. This route
enables the preparation of numerous analogues of the natural
M. tuberculosis antigen sulfoglycolipid 1, with the incorpo-
ration of polymethylated chiral fatty acids in place of the
complex hydroxyphthioceranoic acid moiety. Although none
of the synthetic compounds were as potent as the natural
sulfoglycolipid 1, some of these sulfoglycolipids showed
promising T-lymphocyte-activation properties. All active
analogues have a saturated or monounsaturated polymethy-
lated fatty acid with stereocenters of the S configuration at
the 3-position of the trehalose core. These synthetic sulfogly-
colipids are available in large amounts, and their protective
effects against tuberculosis infection might be investigated
further. The synthesis of analogues with other variations of
the fatty-acid structure is underway in an attempt to gain a
better understanding of the factors that influence their
interaction with CD1 proteins[25,26] and their recognition by
the TCR.
Figure 1. IFN-g release by the T-cell clone. a) Response curves for
~
*
^
compounds 14, 15, and 18–20: ꢁ 14, 15, & 18, 19, 20.
b) Relative intensity of IFN-g release for analogues 13–26 compared to
that for 1, for a fixed antigen concentration of 3.3 mgmLꢀ1
.
observed with the natural product, the maximum activity
was usually observed for an antigen concentration of around
3 mgmLꢀ1. More importantly, it was found that the level of T-
lymphocyte activation was highly dependent on the length,
the position, and the structure of the fatty-acid residues
connected to the trehalose core. When the hydroxyphthio-
ceranoic acid moiety was replaced by linear fatty acids
(products 9–11, data not shown) or with saturated fatty acids
containing one or two methyl groups (compounds 12 and 13),
no T-cell activation was observed (Figure 1b). When this
fatty-acid chain was substituted with additional methyl groups
(compounds 14 and 15), a steady increase in the immunoge-
nicity of the sulfolipids was observed. A similar relationship
between the immunogenicity and the number of methyl
Received: August 4, 2008
Published online: November 3, 2008
Keywords: antigens · fatty acids · glycolipids · tuberculosis ·
.
vaccines
[1] http://www.who.int/tb/en/.
[2] M. Gilleron, S. Stenger, Z. Mazorra, F. Wittke, S. Mariotti, G.
Bꢁhmer, J. Prandi, L. Mori, G. Puzo, G. De Libero, J. Exp. Med.
2004, 199, 649 – 659.
[3] M. B. Goren, O. Brokl, B. C. Das, E. Lederer, Biochemistry 1971,
10, 72 – 81.
Angew. Chem. Int. Ed. 2008, 47, 9734 –9738
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
9737

Communications
[4] F. Lin, L. H. van Halbeek, C. R. Bertozzi, Carbohydr. Res. 2007,
[16] E.-i. Negishi, Z. Tan, B. Liang, T. Novak, Proc. Natl. Acad. Sci.
USA 2004, 101, 5782 – 5787.
342, 2014 – 2030.
[5] C. D. Leigh, C. R. Bertozzi, J. Org. Chem. 2008, 73, 1008 – 1017.
[6] H. H. Baer, X. Wu, Carbohydr. Res. 1993, 238, 215 – 230.
[7] A. Liav, M. B. Goren, Carbohydr. Res. 1984, 127, 211 – 216.
[8] H. H. Baer, B. Radatus, Carbohydr. Res. 1984, 128, 165 – 174.
[9] A. Abiko, S. Masamune, Tetrahedron Lett. 1996, 37, 1081 – 1084.
[10] A. G. Myers, B. H. Yang, H. Chen, L. McKinstry, D. J. Kopecky,
J. L. Gleason, J. Am. Chem. Soc. 1997, 119, 6496 – 6511.
[11] M. O. Duffey, A. LeTiran, J. P. Morken, J. Am. Chem. Soc. 2003,
125, 1458 – 1459.
[17] S. Hanessian, S. Giroux, V. Mascitti, Synthesis 2006, 1057 – 1076.
[18] B. ter Horst, B. L. Feringa, A. J. Minnaard, Org. Lett. 2007, 9,
3013 – 3015.
[19] A. A. Birkbeck, D. Enders, Tetrahedron Lett. 1998, 39, 7823 –
7826.
[20] D. Enders, J. Tiebes, N. De Kimpe, M. Keppens, C. Stevens, G.
Smagghe, O. Betz, J. Org. Chem. 1993, 58, 4881 – 4884.
[21] G. S. Besra, D. E. Minnikin, P. R. Wheeler, C. Ratledge, Chem.
Phys. Lipids 1993, 66, 23 – 34.
[12] D. R. Williams, A. L. Nold, R. J. Mullins, J. Org. Chem. 2004, 69,
5374 – 5382.
[22] P. J. Garegg, B. Samuelsson, J. Chem. Soc. Chem. Commun. 1979,
978 – 980.
[13] R. Des Mazery, M. Pullez, F. Lopez, S. R. Harutyunyan, A. J.
Minnaard, B. L. Feringa, J. Am. Chem. Soc. 2005, 127, 9966 –
9967.
[23] D. Enders, H. Eichenauer, Tetrahedron Lett. 1977, 18, 191 – 194.
[24] K. Bowden, I. M. Heilbron, E. R. H. Jones, B. C. L. Weedon, J.
Chem. Soc. 1946, 39 – 45.
[14] B. ter Horst, B. L. Feringa, A. J. Minnaard, Chem. Commun.
2007, 489 – 491.
[15] T. Novak, Z. Tan, B. Liang, E.-i. Negishi, J. Am. Chem. Soc. 2005,
127, 2838 – 2839.
[25] G. De Libero, L. Mori, Trends Immunol. 2006, 27, 485 – 492.
[26] A. de Jong, E. Casas Arce, T.-Y. Cheng, R. P. van Summeren,
B. L. Feringa, V. Dudkin, D. Crich, I. Matsunaga, A. J. Minnaard,
D. B. Moody, Chem. Biol. 2007, 14, 1232 – 1242.
9738
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 9734 –9738