Stereoselective synthesis and immunogenic activity of the C-analogue of sulfatide

Article abstract of DOI:10.1021/ol061100y

The C-sulfatide 1b was synthesized through a [2,3]-Wittig sigmatropic rearrangement and a Horner-Wadsworth-Emmons olefination as the key steps. The C-analogue 1b is less immunogenic than natural sulfatide 1a, but induces a preferential secretion of the proinflammatory cytokine IFN-γ.

Full text of DOI:10.1021/ol061100y

ORGANIC
LETTERS
2006
Vol. 8, No. 15
3255-3258
Stereoselective Synthesis and
Immunogenic Activity of the C-Analogue
of Sulfatide
Emilia Modica,† Federica Compostella,*,† Diego Colombo,† Laura Franchini,†
Marco Cavallari,‡ Lucia Mori,‡ Gennaro De Libero,‡ Luigi Panza,§ and
Fiamma Ronchetti†
Dipartimento di Chimica, Biochimica e Biotecnologie per la Medicina, UniVersita` di
Milano, Via Saldini 50, 20133-Milano, Italy, Experimental Immunology, Department of
Research, Basel UniVersity Hospital, Hebelstrasse 20, 4031-Basel, Switzerland, and
Dipartimento di Scienze Chimiche, Alimentari, Farmaceutiche e Farmacologiche,
UniVersita` del Piemonte Orientale, Via BoVio 6, 28100-NoVara, Italy
federica.compostella@unimi.it
Received May 4, 2006
ABSTRACT
The C-sulfatide 1b was synthesized through a [2,3]-Wittig sigmatropic rearrangement and a Horner
−Wadsworth−Emmons olefination as the
key steps. The C-analogue 1b is less immunogenic than natural sulfatide 1a, but induces a preferential secretion of the proinflammatory
cytokine IFN-
γ
.
CD1 antigen-presenting molecules present different types of
self and foreign lipid antigens to T lymphocytes which
recognize the CD1-lipid complexes on the surface of antigen-
presenting cells.1 CD1-restricted T cells exert a wide range
of effector functions. They secrete a large variety of cytokines
upon T cell receptor (TCR) triggering, including IFN-γ and
IL-4, and participate in both proinflammatory and anti-
inflammatory immune responses.2
The role of T cells reacting against self-glycosphingolipids
in autoimmune diseases, such as in Multiple Sclerosis, has
been clarified by both in vitro and ex vivo data.3 Recent
studies have demonstrated the importance in their immuno-
genicity of both the lipid and the sugar structures. For
example, modifications in the fatty acids of GM1 ganglioside
or of phospholipids may profoundly hinder the response of
specific T cells.4 Also the structure of the sphingosine chain
contributes to immunogenicity. These modifications have
been associated with modifications in the structures of CD1-
lipid complexes which interact with the TCR or with
differences in the half-life of the immunogenic complexes.5
Finally, the structure of the sugar moiety is also very
important, mostly because the TCR establishes cognate
interactions with parts of the hydrophilic moiety of glyco-
lipids. Therefore, the position of the hydroxyl groups as well
as the type of glycosidic bonds play critical roles in
immunogenicity. Interestingly, modifications in the glyco-
† Universita` di Milano.
‡ Basel University Hospital.
§ Universita` del Piemonte Orientale.
(1) (a) De Libero, G.; Mori, L. Nat. ReV. Immunol. 2005, 5, 485-496.
(b) Moody, D. B.; Zajonc, D. M.; Wilson, I. A. Nat. ReV. Immunol. 2005,
5, 387-399.
(2) De Libero, G.; Donda, A.; Gober, H. J.; Manolova, V.; Mazorra, Z.;
Shamshiev, A.; Mori, L. Neurochem. Res. 2002, 27, 675-685.
(3) Shamshiev, A.; Donda, A.; Carena, I.; Mori, L.; Kappos, L.; De
Libero, G. Eur. J. Immunol. 1999, 29, 1667-1675
(4) Shamshiev, A.; Donda, A.; Prigozy, T. I.; Mori, L.; Chigorno, V.;
Benedict, C. A.; Kappos, L.; Sonnino, S.; Kronenberg, M.; De Libero, G.
Immunity 2000, 13, 255-264.
(5) (a) Miyamoto, K.; Miyake, S.; Yamamura, T. Nature 2001 413, 531-
534. (b) Ndonye, R M.; Izmirian, D. P.; Dunn, M. F.; Yu, K. O. A.; Porcelli,
S. A.; Khurana, A.; Kronenberg, M.; Richardson, S. K.; Howell, A. R. J.
Org. Chem. 2005, 70, 10260-10270
10.1021/ol061100y CCC: $33.50
© 2006 American Chemical Society
Published on Web 06/28/2006

sidic bond may also confer upon the ligand the capacity to
preferentially stimulate proinflammatory functions in the
responding T cell population. Indeed, a C-glycoside analogue
of R-galactosylceramide (R-GalCer) enhances proinflamma-
tory T cell response as compared to natural R-GalCer.6
Whether this is a general phenomenon occurring also with
other types of glycolipid antigens and lipid-specific T cell
populations is not known.
Scheme 1. Preparation of the â-C-Glycoside 6
We have shown that sulfatide (1a), a â-D-galactosyl-
ceramide sulfated at position 3 of galactose, is one of the
mammalian endogenously derived self-antigens presented by
CD1 proteins.7 Sulfatide 1a is a natural ligand for all human
CD1 family members and is presented by all CD1 molecules
to specific T cells. Furthermore, the response to sulfatide is
very frequent in Multiple Sclerosis patients (our unpublished
results) and also occurs in mice during experimental allergic
encephalomyelitis (EAE), a model of autoimmune brain
disease.8
To investigate whether the substitution of the anomeric
oxygen with a methylene affects the proinflammatory T cell
responses also with sulfatide as the model antigen, we
prepared 1b, the C-glycoside isosteric analogue of sulfatide
1a. This modification might confer novel binding conforma-
tions in the active site of CD1 molecules or might influence
the binding capacity to CD1.9,10
Here we describe the synthesis of the C-sulfatide 1b and
show that this analogue is less stimulatory than sulfatide 1a.
The easy and highly stereoselective synthesis of the sulfatide
analogue 1b is based on the preparation of the â-C-galactosyl
ceramide 2, previously described by Dondoni et al. in an
alternative synthetic approach,11 followed by regioselective
sulfation of galactose.
The key features of our approach to the skeleton of â-C-
GalCer 2 are a [2,3]-Wittig sigmatropic rearrangement for
the construction of the â-C-glycoside12 and a Horner-
Wadsworth-Emmons olefination for the installation of the
carbon unsaturated chain.
glucosidic compound.12b The dianionic [2,3]-Wittig re-
arrangement of 4 was carried out by treatment with excess
LDA at -78 °C, followed by reaction of the crude with
diazomethane to afford the R-hydroxy ester 5 as a single
isomer, as clearly shown by the single set of peaks observed
in the 1H and 13C NMR spectra. The Z geometry of the newly
formed double bond was established by a NOE experiment,
which showed an effect of 5% on the H-2 atom of galactose
on irradiation of the vinylic hydrogen. Due to its acidic
lability, compound 5 was promptly hydrogenated with Pt(C)
in methanol to give compound 6 in high yield. The
configuration of the new stereocenter at position 1 of
galactose, generated after double bond reduction, was found
1
to be â through the H NMR coupling constant of 9 Hz
between H-1 and H-2 of galactose. Moreover, compound 6
was used to establish the absolute configuration of the
stereocenter at C-2 previously formed in the [2,3]-Wittig
1
rearrangement. The analysis of the H NMR spectra of its
(R)- and (S)-MTPA esters14 (see the Supporting Information)
allowed us to unequivocally assign the absolute configuration
at C-2 as S and confirmed the stereoisomeric purity and the
complete transmission of stereochemical information to this
new stereocenter during the rearrangement.
The stereochemistry at C-2 of compound 6 requires the
introduction of a masked amino functionality at that position
with retention of configuration. Therefore we decided to
employ a Mitsunobu reaction with methanesulfonic acid as
nucleophile on compound 6 (Scheme 2), followed by reaction
with sodium azide. Mitsunobu mesylation was performed
on compound 6 under the condition reported by Lawless et
al.15 followed by treatment with sodium azide in DMF, which
Starting from known 2,3,4,6-tetra-O-benzyl-1-C-vinyl-R-
D-galactopyranose 3,13 galactoside 4 was obtained (Scheme
1) exclusively as the R-anomer in 90% yield by using a
protocol reported in the literature for the corresponding
(11) Dondoni, A.; Perrone, D.; Turturici, E. J. Org. Chem. 1999, 64,
5557-5564.
(6) (a) Schmieg, J.; Yang, G.; Franck, R. W.; Tsuji, M. J. Exp. Med.
2003, 198, 1631-1641. (b) Yang, G.; Schmieg, J.; Tsuji, M.; Franck, R.
W. Angew. Chem., Int. Ed. 2004, 43, 3818-3822. (c) Chen, G.; Schmieg,
J.; Tsuji, M.; Franck, R. W. Org Lett. 2004, 6, 4077-80.
(7) Shamshiev, A.; Gober, H. J.; Donda, A.; Mazorra, Z.; Mori, L.; De
Libero, G. J. Exp. Med. 2002, 195, 1013-1021.
(12) (a) Tomooka, K.; Nakamura, Y.; Takeshi, N. Synlett 1995, 4, 321-
322. (b) Lay, L.; Meldal, M.; Nicotra, F.; Panza, L.; Russo, G. Chem.
Commun. 1997, 15, 1469-1470.
(13) Li, X.; Ohtake, H.; Takahashi, H.; Ikegami, S. Tetrahedron 2001,
57, 4297-4309.
(8) Jahng, A.; Maricic, I.; Aguilera, C.; Cardell, S.; Halder. R. C.; Kumar,
V. J. Exp. Med. 2004, 199, 947-957.
(14) (a) Dale, J. A.; Mosher, H. S. J. Am. Chem. Soc. 1973, 95, 512-
519. (b) Devijver, C.; Salmoun, M.; Daloze, D.; Braekman, J. C.; De Weerdt,
W. H.; De Kluijver M. J.; Gomez, R. J. Nat. Prod. 2000, 63, 978-980. (c)
Chun, J.; Byun, H.-S.; Arthur, G.; Bittman, R. J. Org. Chem. 2003, 68,
355-39.
(9) Espinosa, J. F.; Montero, E.; Vian, A.; Garcia, J. L.; Dietrich, H.;
Schmidt, R. R.; Martin-Lomas, M.; Imberty, A.; Can˜ada, F. J.; Jimenez-
Barbero, J. J. Am. Chem. Soc. 1998, 120, 1309-1318.
(10) Postema, M. H. D.; Piper, J. L.; Betts, R. L. Synlett 2005, 9, 1345-
1358.
(15) Lawless, L. J.; Blackburn, A. G.; Ayling, A. J.; Perez-Payan, M.
N.; Davis, A. P. J. Chem. Soc., Perkin Trans. 1 2001, 11, 1329-1341.
3256
Org. Lett., Vol. 8, No. 15, 2006

reduction of the azide did not proceed efficiently, instead
reduction with hydrogen sulfide in pyridine/water followed
by reaction of the crude amine with tert-butoxycarbonyl-
anhydride (Boc2O) in DCM/pyridine afforded the N-protected
R-amino ester 9 in 90% yield. Compound 9 was then
converted into the ketophosphonate 10 in 80% yield by
treatment with 6 equiv of lithium dimethyl methylphos-
phonate in THF at -78 °C.
Scheme 2. Introduction of the Azido Moiety
The HWE reaction of 10 with tetradecanal in the presence
of anhydrous potassium carbonate16,18 in acetonitrile exclu-
sively afforded the E-enone 11 in 71% yield, as indicated
by the coupling constant of 15.5 Hz between the vinylic
protons. The carbonyl group of compound 11 was then
subjected to hydride reduction: the optimal reagent was
found to be DIBAL-H (3 equiv in THF at -78 °C),19 which
afforded a mixture of two diastereoisomers in a 95:5 ratio.20
afforded azide 8a in high yield as a single diastereoisomer.
The net retention of stereochemistry during this functional
group interconversion was confirmed by comparing the
physical data of compound 8a with those of its C-2 epimer
8b. Compound 8b was derived from alcohol 6 by mesylation
under standard conditions followed by displacement with
tetrabutylammonium azide and showed a different set of
NMR signals and the same ESI-MS molecular ion.
The next planned key steps for the construction of the
sphingoind skeleton were a Horner-Wadsworth-Emmons
(HWE)16 olefination to introduce the trans double bond and
the aliphatic chain, followed by the stereoselective reduction
of the keto group of the enone (Scheme 3).
The main isomer 12 was obtained pure after column
chromatography and the absolute configuration of the new
stereocenter at C-4 of the sphingoid base was unequivocally
assigned as R by the preparation of the oxazolidinone 13:
11,17,21 NOE analysis of 13 gave a 5% enhancement between
H-3 and H-4 that is consistent with a cis relationship between
the two protons, also confirmed by the 8 Hz coupling
constant value. The assignment and the optical purity of the
product were corroborated by preparation of the (R)- and
(S)-MTPA esters of 12 and analysis of their 1H NMR spectra
(see the Supporting Information).
The last steps of the synthesis (Scheme 4) were the
installation of the desired acyl chain,22 the removal of the
Scheme 3. Preparation of the â-C-Galactosylsphingosine 12
Scheme 4. Synthesis of the C-Sulfatide 1b
benzyl groups, and the introduction of the sulfate group at
position 3 of galactose. Removal of the tert-butoxycarbonyl
(16) Koskinen, A. M. P.; Koskinen, P. M. Synlett 1993, 7, 501-502.
(17) Shikata, K.; Azuma, H.; Tachibana, T.; Ogino, K. Tetrahedron 2002,
58, 5803-5809.
First, reduction of the azide and protection of the amino
group as N-Boc derivative were necessary to avoid decom-
position of the R-azido ester in basic conditions and
considering that the steric hindrance of this bulky amino
protecting group should enhance the selectivity of the
reduction of the keto group on the HWE product.17 Staudinger
(18) Mikolajczyk, M.; Balczewski, P. Synthesis 1987, 659-661.
(19) Reduction of 10 with L-Selectride gave a 1:1 mixture of the two
diastereoisomers.
(20) The dr was deduced by integration of the C-5 signals at δ 5.40
(95%) for the major isomer and δ 5.45 (5%) for the minor isomer in the 1H
NMR spectrum of the crude mixture.
Org. Lett., Vol. 8, No. 15, 2006
3257

group of 12 with HCl in dioxane, followed by treatment of
the crude amine with nervonic acid and EDCI in DCM gave
the tetra-O-benzylgalactosyl ceramide analogue 14 in 70%
yield. Debenzylation with sodium in liquid ammonia fol-
lowed by exhaustive acetylation of the crude afforded 15 in
50% yield. Finally, compound 15 was submitted to Zemple´n
deacetylation to yield the â-C-galactosyl ceramide 2, and
the sulfate group at position 3 of galactose was introduced
according to Flitsch et al.23 affording the target compound
1b in 90% yield over two steps.
In conclusion, we have developed an efficient and facile
synthesis of the sulfatide analogue 1b. Worthy of note is
the very high stereoselectivity observed in each synthetic
step, which allowed us to obtain the proper â-stereochemistry
at the C-anomeric position and to construct the sphingosine
skeleton with the correct stereochemistries at positions 3 and
4.
C-Sulfatide 1b is less stimulatory than sulfatide 1a. These
findings are opposite to those obtained with R-C-galactosyl-
ceramide which stimulates natural killer T cells more
effciently than R-O-galactosylceramide,6,24 thus indicating
that modifications of the glycosidic bond do not always result
to the same functional effect.
The reduced potency of C-sulfatide 1b might be the
consequence of reduced CD1a loading in early endosomes
or on the plasma membrane. This effect might also be
ascribed to different populations of conformations around
the C-glycosidic bond as compared to those around the
O-glycosidic one. Therefore, a sufficient number of stimu-
latory CD1a complexes might be generated only with large
doses of C-sulfatide 1b.
The biological activity of the synthetic sulfatide analogue
1b was compared to the synthetic22 but naturally occurring
sulfatide 1a by using a T-cell antigen presentation assay.
The CD1a-restricted human T cell clones K34B9.1 and
K34A6.2 were selected for these studies because both react
to natural sulfatide and release the proinflammatory cytokine
IFN-γ.7 Results (Figure 1) show that C-sulfatide 1b has lower
The reduced efficacy indicates that sulfatide 1b forms a
complex with CD1a that is slightly different from the
complex formed by sulfatide with an O-glycosidic bond. This
difference might derive from altered hydrogen bond interac-
tions between 3-O-sulfogalactose and Arg73, Arg76, Glu154,
and Ser77 of CD1a, which fix the position of the hydrophilic
part of sulfatide.25 Moreover, although not explicitely
indicated in ref 25 it cannot be excluded that the anomeric
oxygen is involved in hydrogen bonds with Arg73 and
Thr158 of CD1a. Such modifications might change the
strength of interaction with the TCR of the two cell clones
tested in this study and thus lead to reduced stimulatory
capacity of sulfatide 1b. Whether this modified interaction
also leads to different signaling in the TCR, which in turn
induces preferential IFN-γ release, will be the basis of further
investigations.
Figure 1. Glycolipid immunogenicity is influenced by the C-
glycosidic bond. C1R cells expressing human CD1a were pre-
incubated with sulfatide 1a (filled circles) or with C-sulfatide 1b
(open circles) at the doses indicated, before addition of sulfatide-
specific CD1a-restricted T cell clone K34B9.1. IFN-γ (A) or IL-4
(B) released were measured by ELISA. Data are expressed as mean
pg/mL ( SD of duplicates. Similar results were obtained with the
K34A6.2 T cell clone (see the Supporting Information).
Acknowledgment. This work was supported by MIUR-
Italy (COFIN 2004, FIRB 2001), Swiss National Foundation
grant No. 3100 A0-109918/1, and the Swiss Multiple
Sclerosis Society.
potency and efficacy than sulfatide 1a in inducing release
of both IFN-γ and IL-4. Moreover, the C-sulfatide 1b doses
necessary to induce the same cytokines amounts as those
induced by sulfatide 1a were 5 times larger for IFN-γ and
almost 10 times larger for IL-4. This indicates that C-sulfatide
1b may induce a modest but preferential secretion of the
proinflammatory IFN-γ.
Supporting Information Available: Experimental pro-
cedures and full spectroscopic data for all compounds. This
material is available free of charge via the Internet at
http://pubs.acs.org.
OL061100Y
(23) Guilbert, B.; Davis, N. J.; Pearce, M.; Aplin, R. T.; Flitsch, S. L.
Tetrahedron: Asymmetry 1994, 5, 2163-2178.
(24) Berkers, C. R.; Ovaa, H. Trends Pharmacol. Sci. 2005, 26,0.252-
257.
(25) Zajonc, D. M.; Elsliger, M. A.; Teyton, L.; Wilson, I. A. Nat.
Immunol. 2003, 4, 808-15.
(21) Benedetti, F.; Miertus, S.; Norbedo, S.; Tossi, A.; Zlatoidzky, P. J.
Org. Chem. 1997, 62, 9348-9353.
(22) Compostella, F.; Franchini, L.; De Libero, G.; Palmisano, G.;
Ronchetti, F.; Panza, L. Tetrahedron 2002, 58, 8703-8708.
3258
Org. Lett., Vol. 8, No. 15, 2006