Toward libraries of biotinylated chondroitin sulfate analogues: From synthesis to in vivo studies

Article abstract of DOI:10.1002/chem.201202173

Chondroitin sulfate-E (CS-E) oligosaccharidic analogues (di to hexa) were prepared from lactose. In these compounds, the 2-acetamido group was replaced by a hydroxyl group. This modification speeded up the synthesis, and large oligosaccharides were constructed in a few steps from a lactose-originated block. The protecting groups used were as follows; Fmoc for hydroxyl groups to be glycosylated, allyl group for anomeric position protection, and trichoroacetimidate leaving groups were used to prepare up to octasaccharides. We took advantage of the presence of allyl group to develop a click biotinylation, through its transformation into a 3-azido-2-hydroxyl propyl group in two steps (epoxidation and sodium azide epoxide opening). The biotinylating agent was a water-soluble propargylated and biotinylated triethylene glycol (PEG). By using surface plasmon resonance (SPR), it was shown that the di-, tetra-, and hexasaccharides display a binding affinity and selectivity toward HSF/GSF and CXCL12 similar to that of CS-E. A parallel study confirmed their mimicry of natural compounds, based on the hexasaccharide interaction with Otx2, a homeodomain protein involved in brain maturation, thus validating our simplification approach to synthesize bioactive GAG. Copyright

Full text of DOI:10.1002/chem.201202173

FULL PAPER
DOI: 10.1002/chem.201202173
Toward Libraries of Biotinylated Chondroitin Sulfate Analogues:
From Synthesis to In Vivo Studies
Guillaume Despras,[a, b, c] Clꢀmence Bernard,[d] Anthony Perrot,[d] Laurent Cattiaux,[a, b, c]
Alain Prochiantz,[d] Hugues Lortat-Jacob,[e] and Jean-Maurice Mallet*[a, b, c]
Abstract: Chondroitin sulfate-E (CS-
E) oligosaccharidic analogues (di to
hexa) were prepared from lactose. In
these compounds, the 2-acetamido
pare up to octasaccharides. We took
advantage of the presence of allyl
group to develop a click biotinylation,
through its transformation into a 3-
azido-2-hydroxyl propyl group in two
steps (epoxidation and sodium azide
epoxide opening). The biotinylating
agent was a water-soluble propargylat-
ed and biotinylated triethylene glycol
(PEG). By using surface plasmon reso-
nance (SPR), it was shown that the di-,
tetra-, and hexasaccharides display a
binding affinity and selectivity toward
HSF/GSF and CXCL12 similar to that
of CS-E. A parallel study confirmed
their mimicry of natural compounds,
based on the hexasaccharide interac-
tion with Otx2, a homeodomain pro-
tein involved in brain maturation, thus
validating our simplification approach
to synthesize bioactive GAG.
group was replaced by
a hydroxyl
group. This modification speeded up
the synthesis, and large oligosacchar-
ides were constructed in a few steps
from a lactose-originated block. The
protecting groups used were as follows;
Keywords: allyl groups · biotinyla-
tion · click chemistry · lactose · pro-
tecting groups · structure–activity
relationships
Fmoc for hydroxyl groups to be gly
ACHTUNGTNERcNUNG os-
ACHTUNGTRENNUNGylated, allyl group for anomeric posi-
tion protection, and trichoroacetimi-
date leaving groups were used to pre-
Introduction
chains of various lengths (10 to 200 repeating units). The
sulfate groups carried by the disaccharidic portions can be
located on different positions and each sulfation pattern de-
fines a subclass of CS.
The CS chains are linked to carrier proteins through a tet-
rasaccharidic moiety covalently bound to an l-serine residue
to form structures called proteoglycans (PGs). Mainly found
in the extracellular matrix of a wide variety of tissues but
also on cell surfaces, the PGs present two degrees of struc-
tural heterogeneity. On one hand, each carrier protein could
link several kinds of GAGs, additionally to CS. On the
other hand, CS chains are non-regular polymers with varia-
tion on the sulfation pattern.
CS proteoglycans participate in important biological proc-
esses such as regulation of cells behaviors[1–3] (growth, prolif-
eration, and migration), development and repair of the cen-
tral nervous system (CNS),[4–10] or cellular recognition.[11–13]
These critical biological activities seem to be governed by
the finely tuned repartition of the negative charges borne by
the sulfate and carboxylate groups.[14] The sulfation patterns
expressed all along the CS chains would encode information
decipherable by positively charged domains of protein re-
ceptors. However, very little is known about the role of the
sulfation code and the structure–activity relationship was
not well studied, due the the lack of pure and chemically de-
fined large CS fragments.
Chondroitin sulfates (CS) are important polysaccharides be-
longing to the family of glycosaminoglycans (GAGs). These
linear and anionic polymeric chains are constituted by a rep-
etition of O-sulfated disaccharidic units. The disaccharidic
block contains a d-galactosamine 1!4-linked to a d-glucur-
onic acid and the disaccharides are 1!3-linked to form
[a] Dr. G. Despras, L. Cattiaux, Dr. J.-M. Mallet
UPMC Paris 06, UMR 7203
Laboratoire des BioMolꢀcules
Universitꢀ P. et M. Curie, 4 Place Jussieu
75005 Paris (France)
E-mail: Jean-Maurice.Mallet@ens.fr
[b] Dr. G. Despras, L. Cattiaux, Dr. J.-M. Mallet
CNRS, UMR 7203 (France)
[c] Dr. G. Despras, L. Cattiaux, Dr. J.-M. Mallet
ENS, UMR 7203, Dꢀpartement de Chimie
Ecole Normale Supꢀrieure, 24
Rue Lhomond 75005 Paris (France)
[d] C. Bernard, A. Perrot, Prof. A. Prochiantz
Collꢁge de France, Center for Interdisciplinary Research
in Biology (CIRB), Paris, 75005 (France)
CNRS UMR 7241/INSERM U 1050
Labex Memolife, PSL-Research University (France)
[e] Dr. H. Lortat-Jacob
Institut de Biologie Structurale, UMR 5075, CNRS, CEA
Universitꢀ Joseph Fourier, 41 rue Horowitz
38027 Grenoble (France)
The purification of homogeneous depolymerized products
is not realistic beyond tetrasaccharides. The synthesis of
large fragments is very complicated, because it implies gal-
actosamine, and these oligosaccharides have been prepared
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201202173.
Chem. Eur. J. 2013, 19, 531 – 540
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
531

deacetylated in Zemplen condi-
tions, followed by the selective
preparation of 4’,6’-O-benzyli-
dene using benzaldehyde and
zinc chloride. The remaining 6-
primary position was then sily-
lated to give 7 (65% yield from
6).
The first critical step was the
selective protection of the 3’
position of compound 7. It
could be achieved (72% yield)
Figure 1. CS-E (top) and analogues (bottom).
after long sequences of reactions.[9,15–22] We propose here to
prepare very close analogues of CS-E in a rapid manner
(Figure 1) and to explore structure–activity relationships. As
with the heparin series,[23] we only modified an uncharged
residue: the 2-NHAc group was replaced by a 2-OH; the
sulfation and charge patterns being completely preserved.
Similar group exchanges were already successfully described
for gangliosides and related compounds.[24] This slight modi-
fication has halved the chemical steps needed for CS-E syn-
thesis.
Figure 2. Common building block.
To allow easy biological evaluation (i.e., protein binding),
CS-E analogues were prepared in their biotinylated forms
(di 3, tetra 4, and hexa 5) and in their unconjugated forms
(tetra 1 and hexa 2) (Figure 1). Their synthesis can be easily
and rapidly achieved from the abundant and cheap natural
disaccharide lactose.
at room temperature by using FmocCl and sym-collidine to
generate a hindered acylating agent in situ.[26] Benzoylation
of 8 in mild conditions[27] (CH2Cl2/pyridine (3:1), low tem-
perature) gave 9 (81% yield). The use of pure pyridine as
solvent was found to be detrimental because of in situ parti-
al removal of the Fmoc group and formation of 3’-O-ben-
zoate.
Result and Discussion
The second critical step was the selective deprotection of
the silyl group in the presence of OFmoc; this was achieved
using TMSOTf at low temperature[28] to produce compound
10 (84%). This probably occurred through the formation of
These chondroitin sulfate analogues were prepared by “oli-
gomerization” of
a
key disaccharide lactose block
(Figure 2). The synthetic plan is
depicted on Scheme 1.
Benzylidene groups were
used to protect positions 4’,6’ to
be sulfated and benzoate was
used for “permanent” and par-
ticipating alcohol function pro-
tecting groups.
A selectively
cleavable group is needed on
the 3’ position for chain elonga-
tion and an allyl group is used
to block position 1 and also as
a linker precursor in the bio
ACHTUNGTRENNUNGylation reaction.
ACHTUNGTNERtNUNG in-
Synthesis
of
disaccharidic
blocks: The synthesis started
from peracetylated allyl b-lac-
tose 6[25] accessible in a large
scale (50 g) from d-lactose
(Scheme 2). Compound 6 was Scheme 1. Synthetic plan.
532
www.chemeurj.org
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 531 – 540

Libraries of Biotinylated Chondroitin Sulfate Analogues
FULL PAPER
Lower chemical shifts were ex-
pected in the a configuration
d=95–97 ppm.[31] Yields were
within the range of 60–70%,
and comparable to published
yields for similar reactions
using disarmed uronate donors.
Final steps: sulfation and sapo-
nification: Unconjugated (allyl)
CS analogues were prepared
from tetra- and hexasaccharide
(compounds 16 and 18, respec-
tively) by using the following
sequence: a) benzylidene re-
moval, b) sulfation of free OH,
and c) saponification of ester
groups. Although Fmoc is
stable in benzylidene removal
conditions, it is cleaved during
Scheme 2. Reagents and conditions: a) NaOMe, MeOH; b) PhCHO, ZnCl2; c) TBDMSCl, pyridine, 65% (3
steps); d) FmocCl, sym-collidine, CH2Cl2, 72%; e) BzCl, CH2Cl2/pyridine, 08C to RT, 81%; f) TMSOTf,
CH2Cl2, ꢀ788C, 84%; g) TEMPO, BAIB, CH2Cl2/water; h) BnBr, NaHCO3, DMF, 82% in 2 steps; i) Pd-
ACHTUNGTRENNUNG(PPh3)4, ZnCl2, Bu3SnH, THF, RT, 89%; j) CCl3CN, NaH, CH2Cl2, 08C to RT, 74%; k) Et3N, dichloromethane
(DCM), RT, 90%.
a labile trimethylsilyl ether in
position 6, which was readily
hydrolyzed in the work up.[26]
Oxidation using 2,2,6,6-tet-
A
N-oxide
iodoben-
E
N
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
terification reaction of the
crude carboxylic acid gave 11
(82% yield).
To obtain the donor block,
compound 11 was first deally-
lated by using Bu3SnH and Pd-
ACHTUNGTRENNUNG(PPh3)4 and ZnCl2 as cata-
lysts.[29] The resulting hemiace-
tal 12 (89% yield) was convert-
ed into the anomeric trichloro-
ACHTUNGTRENNUNGacetimidate 13 by treatment
with trichloroacetonitrile and
sodium hydride. Finally, the ac-
ceptor block 14 was obtained
after the cleavage of the Fmoc
group on compound 11.
Scheme 3. Construction of oligosaccharides backbone through iterative glycosylation/deprotection sequences.
Oligosaccharides synthesis: With these building blocks in
hand, iterative glycosylation reactions were undertaken by
using TMSOTf as a promoter, in the presence of molecular
sieves in dichloromethane at room temperature (Scheme 3).
Rather high quantities of promoter (0.3 equiv) and “high”
temperatures (208C) were used to avoid the formation of
the orthoester. After removing the Fmoc group, the oligo-
merisation process was continued up to the octasaccharide
19 (Scheme 3).
the sulfation reaction. Consequently, to avoid the presence
of an extra sulfate, it was replaced by an acetyl group. Ben-
zylidene acetals were removed by using aqueous TFA, and
the free hydroxyls were sulfated by using SO3·Me3N in
1
DMF (Scheme 4). The large shifts in H and 13C signals in
the NMR spectrum confirmed the positions of the sulfates
(Table 1).
The last step is the saponification of ester groups by two
different procedures. First, a two-step deprotection was
used, involving a saponification under mild conditions with
LiOH/H2O2 followed by treatment with aqueous NaOH[18]
b Selectivity was ascertained in the NMR spectra, in par-
ticular 13C NMR C1 signals between d=100–102 ppm.[30]
Chem. Eur. J. 2013, 19, 531 – 540
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
533

J.-M. Mallet et al.
Click biotinylation: Biotinylation is a convenient way to re-
producibly immobilize hydrophilic compounds on surfaces.
Biotin binds to streptavidin with a very high affinity
(1015 mꢀ1). We have, in the past, used biotin and biotin sul-
fone for the immblization of oligosaccharides onto micro-
plate chips and beads.[34] Unfortunately, this method involves
a carboxylic function at the end of the linker carried by the
carbohydrate moiety, which is not compatible with ben-
zoates and benzyl uronic esters. Jacquinet et al.[35,36] recently
described the biotinylation of sulfated and unsulfated CS
Scheme 4. Final steps for non-biotinylated oligosaccharides. Reagents
and conditions: a) Ac2O, pyridine; b) TFA, water, THF/CH2Cl2, sonica-
tion; c) SO3·NEt3, DMF, 608C; d) LiOH, H2O2, THF/water, 08C to RT
then NaOH (4m), MeOH, 08C to RT; e) 4m NaOH, MeOH, 08C to RT.
ꢀ
ꢀ
oligosaccharides by using a NHZ CH2 CH2 linker.
Here, we wanted to use directly the allyl group and its
double bond to introduce a biotin moiety. A first
possible strategy is to add cysteamine on the allyl
and then to use an amide-coupling reaction group.
1
Table 1. Chemical shifts H and 13C of galactose protons before/after sulfation.
H4Gal
H6Gal
H5Gal
C4Gal
C6Gal
C5Gal
This well-known photo-induced addition of thiols
onto double bounds has been used in sugar re-
search.[37] Another strategy is the ozonolysis of the
double bound to give a reactive aldehyde and to
use it for reductive amination.[38,39] These methods
are not fully convenient because they use large
amounts of reagents and/or have limited yields due
20
22
21
23
4.02–3.71
4.72, 4.70
4.02–3.65
4.74–4.65
3.30–3.03
3.81–3.64
3.29–3.00
4.20–3.59
3.30–3.03
3.89–3.64
3.29–3.00
3.91–3.59
69.30, 68.20
77.27, 73.21
68.25–68.16
77.26–72.94
62.15, 61.81
67.96
62.10–61.83
68.23–67.89
74.68, 74.01
74.38, 74.28
74.68–74.03
74.31
(Scheme 5). These conditions are “traditionally” employed
in GAG synthesis to avoid the b-elimination of the 4-O-gly-
cosidic bond of the glucuronic ester moiety. Nevertheless,
Martin-Lomas et al.[32] reported a direct saponification using
KOH at room temperature in the case of heparin-like oligo-
saccharide synthesis. Thus, we decided to apply a similar ap-
proach for the deprotection of CS analogues and tetra- and
hexasaccharide 22 and 23 were treated with aqueous NaOH
in methanol to give the final compounds 1 and 2. The NMR
spectra are in accordance with the structure. Further struc-
tural confirmation was then undertaken by using mass spec-
trometry. MALDI or electrospray mass spectrometry of the
sodium salt is generally deceptive due to poor volatility and
concomitant extensive fragmentation. Recently, this has
been improved with a new method from the group of Vas-
seur[33]; the principle is to use, instead of sodium salt, a large
hydrophobic pyrenemethyl guanidinium counterion (pmg).
Adducts between sulfated oligosaccharide and pmg are very
stable and can be analyzed by MALDI MS (see the Sup-
porting Information).
to solubility and low reactivity problems. We now propose
an alternative approach based on the Huisgen[40] reaction.
The use of the conditions described by the groups of Sharp-
less[41] and Meldal[42] (copper-catalyzed click chemistry)
allows one to perform the reaction in various solvents using
an equimolar ratio of reagents. The triazole linker is stable
and not protonated until pH<2.[43] Click chemistry has al-
ready been used in glycochemistry to prepare conjugates,
arrays,[44] cyclodextrine analogues,[45] and multivalent lectin
ligands.[46] This reaction was used successfully to conjugate a
CS-D (from natural sources) to PEG or BSA with a good
yield.[47] Some examples of “click biotinylation” have been
reported in the case of S-Sialosides[48] and CD22 ligands.[49]
We chose to insert the azide function on the carbohydrate
moiety and the alkyne group on the biotin derivative. Thus,
the critical step is the clean introduction of an azide on the
allyl chain. Nifantiev et al.[50] described the conversion of an
allyl group into a 2-azido ethyl on methyl urinate-bearing
oligosaccharides. Despite good overall yields (64–85%), this
sequence is tedious (7 steps), and the uronate function is
not compatible with the re-
quired reduction step of an al-
dehyde function. Consequently,
we decided to prepare the
azido compounds through an
electrophilic epoxide, which
could be obtained by the oxida-
tion of the allylic double bond
(Scheme 5). The synthesis only
involves two steps, which are
compatible with the uronic
ester.
Two strategies were envi-
sioned for the coupling: bio
ACHTUNGTRENNUNGtin-
Scheme 5. Allyl to azido propyl conversion and alternatives for biotinylation.
ACHTUNGTNERyNUNG lation before or after the de-
534
www.chemeurj.org
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 531 – 540

Libraries of Biotinylated Chondroitin Sulfate Analogues
FULL PAPER
protection/sulfation sequence (Scheme 5) by using a cop-
per(II)/ascorbate catalytic system.
The reaction of NaN3/NH4Cl[51] on glycidyl epoxide is re-
gioselective and only gives 3-azido 2-hydroxy propyl. In con-
trast, the epoxidation of the allyl glycoside by using peracid
usually gives a 1:1 mixture of diastereomers. The presence
of two epimeric OH should not be critical for biological ac-
tivity as long as this position is not sulfated.
Compounds 15 and 17 were treated with meta-chloroper-
benzoic acid (m-CPBA) in dichloromethane at RT to give
the intermediate epoxides (Scheme 6). The crude com-
Scheme 7. Clickable-biotin derivatives.
30 in good yields. An acetal cleavage/sulfatation sequence
gave the protected biotinylated compounds 31 and 32
(Scheme 8).
The disaccharide 33 was deprotected following the two-
step mild saponification (LiOH/H2O2) to give compound 3
(Scheme 9). Analysis by using mass spectrometry showed
that the sulfide function of the biotin moiety was oxidized
into a sulfone group. This side reaction is not detrimental;
biotin sulfone is known to be strongly recognized by
avidin.[54] Although biotin sulfone is compatible with biolog-
ical tests, we choose to saponify the other compounds using
the direct procedure to avoid the possible uncomplete oxi-
dation of the biotin leading to a mixture of sulfoxide and
sulfone. Thus, the protected biotinylated tetrasaccharide 34
Scheme 6. Reagents and conditions: a) m-CPBA, CH2Cl2; b) NaN3,
NH4Cl, THF/EtOH/water, heated to reflux; c) DBU (cat.), Ac2O, pyri-
dine, 63% after 3 steps.
pounds were reacted with NaN3/NH4Cl in THF/ethanol/
water (2:2:1) to give the 3-azido 2-hydroxy propyl deriva-
tive. The acetylation in the presence of catalytic DBU in
pyridine removed the Fmoc group (not fully compatible
with the epoxide opening) and protected two alcohol func-
tions that were not to be sulfat-
ed. The disaccharide 24 and tet-
rasaccharide 25 were obtained
in 63 and 70% yields, respec-
tively. The NMR signal of ace-
tate was used to confirm the
presence of a 1:1 mixture of the
two epimers.
Two biotin units were envi-
sioned; the simplest compound
used is the known biotin N-
propargyl amide 26;[52] the use
of which was disappointing in
Scheme 8. Reagents and conditions: a) 28, Cu
sonication; c) Me3N·SO3, DMF, 608C.
ACHTUGNRTNE(NUGN OAc)2/ascorbate, CH2Cl2/water; b) TFA, water, THF/CH2Cl2,
preliminary experiments be-
cause it is poorly soluble in or-
ganic solvent and in water, and
gave a poor yield in the cou-
pling reaction. Compound 28, a
more versatile PEGylated ver-
sion of 26, was preferred: one
important feature is that 28 is
water soluble. This compound
was obtained from the alkyne-
amine 27[53] and biotin-NHS
(Scheme 7).
The protected disaccharide
24 and tetrasaccharide 25 were
then converted into the biotin-
Scheme 9. Reagents and conditions: a) LiOH, H2O2, THF/water, 08C to RT then NaOH (4m), MeOH, 08C to
conjugated compounds 29 and RT; b) NaOH (4m), MeOH, 08C to RT.
Chem. Eur. J. 2013, 19, 531 – 540
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
535

J.-M. Mallet et al.
was treated with NaOH (4m) in MeOH to give the final CS
tetrasaccharidic analogue 4 (Scheme 9).
minal and highly basic extension, interacts both with HS
and CS/DS. Altogether, HGF/SF, CXCL12a and CXCL12g
represent interesting models to investigate the activity of
the CS analogues prepared here, regarding both their nature
(CS versus HS) and their degree of polymerisation (di- to
hexasaccharide).
The yields are not as good as those in the allyl series,
mostly because of the increased polarities and purification
difficulties due to the presence of the biotin. An alternative
was explored, that is, to delay the introduction of the biotin
(Scheme 10). The functionalization with the azide group di-
To analyze the protein binding capacities of the di-, tetra-
and hexasaccharides we first
adopted a relevant system in
which the oligosaccharides were
coupled, owing to their biotin
group, to a solid phase (strepta-
vidin-activated
sensorchip),
thus mimicking the presenta-
tion of CSPG at the cell sur-
face, whereas the proteins of in-
terest were injected in the fluid
phase. Injection of a range of
concentration of HGF/SF (up
to 5.5 nm), a known CS binding
protein[59], over the CS-E-like
di-, tetra- or hexasaccharide
surfaces, gave a series of sensor-
grams (Figure 3a–c), which
Scheme 10. Reagents and conditions: a) m-CPBA, CH2Cl2; b) NaN3, NH4Cl, THF/EtOH/water, heated to
reflux; c) DBU (cat.), Ac2O, pyridine; d) TFA, water, THF/CH2Cl2; e) Me3N·SO3, DMF, 608C ; f) NaOH,
MeOH, 08C to RT; g) compound 28, CuSO4/ascorbate, water.
rectly followed by benzylidene acetals cleavage was carried
out in good overall yields (60%). After sulfation/saponifica-
tion, the deprotected oligosaccharides were coupled to the
biotin derivative 28 in water. Very good yields were ob-
tained for this last “click biotinylation”(90% for 4, 75% for
5).
show that the hexasaccharide 5 displays a strong HGF/SF-
binding capacity, whereas the tetra- or disaccharide are only
poorly active or completely inactive. To investigate whether
these materials are specific CS mimetics (rather than the
polyanions of more general GAG mimetics) we made use of
the CXCL12 chemokine, of which two isoforms CXCL12a
and CXCL12g bind to heparan sulfate, whereas the
CXCL12g only recognizes CS.[60] As shown in Figure 3d–i,
we found that CXCL12a did not bind to any part of the oli-
gosaccharide, whereas CXCL12g strongly interacted with
the hexasaccharide 5, and albeit to a much lesser extent,
with the tetrasaccharide 4.
Next, to investigate the binding of these oligosaccharides
to protein in solution, a competition assay was designed in
which CXCL12g was co-incubated with the non biotinylated
CS-E-like tetra- or hexasaccharide or (for comparison) with
heparin-derived hexasaccharide and injected over a chon-
droitine sulfate-functionalized sensorchip. Similar to the
direct binding assay, competition was observed for both the
heparin and the CS hexasaccharides, but not for the tetra-
saccharide (Figure 4) . Interestingly, this experiment showed
that the hexasaccharide 2 is almost as active as the heparin
derived oligosaccharide.
Biological Evaluation
Analysis of chemokine-oligosaccharide interaction by SPR:
To investigate the activities of these CS-E-like oligosacchar-
ides, surface plasmon resonance (SPR)-based binding and
competition assays were performed with three model pro-
teins, namely HGF/SF, CXCL12a and CXCL12g. Hepato-
cyte growth factor/scatter factor (HGF/SF) is a multifunc-
tional plasminogen-related protein that induces growth, mo-
tility, and differentiation of epithelial, endothelial, and
neural cells through activation of its tyrosine kinase receptor
MET.[55] HGF/SF also interacts with HS and CS/DS
chains[56] and it has been demonstrated that efficient activa-
tion of the MET receptor requires HGF/SF engagement
with GAGs to form a signalling ternary complex. The mini-
mal size of the oligosaccharide that binds HGF/SF with high
affinity is a tetrasaccharide from heparan sulfate but a hexa-
saccharide from dermatan sulfate.[57] CXCL12 (or SDF for
stromal cell-derived factor) is also a GAG binding protein
that occurs in different isoforms. It belongs to the chemo-
kine family that promotes directional kinesis of cells
through the establishment of concentration gradients that
are believed to be maintained within tissues through GAGs
binding.[58] CXCL12a exclusively binds to HS, whereas
CXCL12g, a CXCL12 isoform characterized by a long C-ter-
Analysis of Otx2-oligosaccharide interaction: Homeopro-
teins define a family of transcription factors with important
developmental and physiological properties. It was shown
recently that in addition to regulating transcription in a cell
autonomous way, these transcription factors also have a
non-cell autonomous signalling activity.[61] A particularly in-
teresting case is provided by the homeoprotein Otx2 that is
found in parvalbumin-expressing GABAergic interneurons
(PV-cells) in the visual cortex. The presence of Otx2 in PV-
536
www.chemeurj.org
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 531 – 540

Libraries of Biotinylated Chondroitin Sulfate Analogues
FULL PAPER
Figure 3. SPR analysis of HGF/SF and CXCL12 binding to di-, tetra-, and hexasaccharides. Overlay of sensorgram showing binding of HGF/SF (a,b,c),
CXCL12g (d,e,f) or CXCL12a (g,h,i) to hexa- (a,d,g), tetra- (b,e,h) or di- (c,f,i) saccharides. In each panel, the protein concentrations used were (from
top to bottom) 5.5–3.7–2.5–1.6–1.1, and 0.73 nm for HGF/SF, 13.3–8.9–5.9–3.9–2.6, and 1.7 nm for CXCL12g and 150–100–66.7–44.4–29.6, and 19.7 nm for
CXCL12a.
means that closing one eye during this period leads to a
strong decrease of its visual acuity. This does not happen if
the eye is closed before P20 or after P40.
In two precedent studies[62,63] it was shown that Otx2 in-
fused directly inside the visual cortex is primarily internal-
ized by PV-cells. This specific recognition is mediated by the
recognition of Otx2 by glycosaminoglycans (GAGs) present
within the perineuronal nets (PNNs) enwrapping the PV-
cells. The preferential transfer of endogenous Otx2 into PV-
cells is lost upon injection of chondroitinase ABC, suggest-
ing that chondroitin sulfate (CS) participates in the recogni-
tion of Otx2. Within the Otx2 sequence was identified a
short domain that has the structure of a GAG-binding
domain (RKQRRERTTFTRAQL) and participates in PNN
recognition. By using isothermal calorimetry, it was demon-
strated that among the CS family, disulfated CS (CS-D and
CS-E) have the highest affinities for the GAG-binding
domain of Otx2.[57]
Figure 4. Binding of soluble heparin and CS-E analogues to CXCL12g.
Oligosaccharides: hexasaccharide (white), heparin hexasaccharide
2
(black) or tetrasaccharide 1 (grey) from 0 to 20 mm were incubated with
CXCL12g (10 nm) and injected over a chondroitin sulfate activated sur-
face. The binding responses (mean of two experiments for the hexasac-
charide) were recorded and plotted against the oligosaccharide concen-
tration.
This is why we have investigated if Otx2 recognizes the
CS-E-like hexasaccharide. To that end, we used the gel shift
assay, an assay allowing one to test whether a protein binds
to a biotinylated DNA probe in the presence of competing
GAGs. In the present case, the biotinylated version of the
hexasaccharide (5) was used as a probe. As illustrated in
Figure 5a, the incubation of the hexasaccharide 5 with Otx2
led to a shift in its migration. Blotting the membrane with
an antibody against Otx2 showed that the shift is really due
to the interaction of the hexasaccharide with Otx2 (Fig-
cells does not reflect its synthesis by this cell type but its
import from an extracellular source.[10] Otx2 accumulates in
PV-cells starting at day 20 (P20) of the mouse postnatal life
and continuing throughout life. This accumulation opens at
P20 and closes at P40 a period of physiological plasticity,
called the critical period (CP) for binocular vision. This
Chem. Eur. J. 2013, 19, 531 – 540
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
537

J.-M. Mallet et al.
Accordingly, Otx2 internalization is antagonized by the infu-
sion of the GAG-binding domain and severely decreased
following the injection of chondroitinase ABC into the
visual cortex .[63] This led us to investigate if the hexasac-
charide 2 could modify the amount of endogenous Otx2 in-
ternalized by PV-cells following its infusion into the adult
mouse visual cortex for three days. As illustrated in Fig-
ure 6a and b and quantified in Figure 6e, a marked decrease
in the number of Otx2-positive cells was observed in cortical
supragranular layers of infused hemispheres, compared to
the untreated hemispheres (at 3 days: (77.3ꢁ13.9)% of con-
trol; at 7 days: (53.1ꢁ0.8)% of control). Thus, Otx2 locali-
zation in the visual cortex was disrupted by the presence of
2. In addition, blocking Otx2 transfer induced a significant
loss of the staining for WFA, a lectin that binds with high af-
finity to the PNNs (Figure 6c–e, at 3 days: (87ꢁ10.9)%; at
7 days: (72.2ꢁ1.3)%). Indeed, the persistent accumulation
of Otx2 is needed to maintain the PNNs around the PV-
cells in the visual cortex .[62] PV expression was not modified
significantly (at 3 days: (100.3ꢁ6.5)%; at 7 days: (107.3ꢁ
11.5)%), demonstrating an absence of cell death (Figure 6c–
e).
All in all, these results demonstrate that the hexasacchar-
ide 2 can be used as an inhibitor of Otx2 in vitro binding to
gene promoters and of its in vivo internalization. Given that
Otx2 internalization beyond critical period closure is neces-
sary to maintain a non-plastic state, it will be interesting to
test if, as is the case for the GAG-binding domain,[57] the
CS-E hexasaccharide analogue can be used to reactivate
plasticity in the adult.
Figure 5. Otx2 binds to hexasaccharides: a,b) Gel shift assay showing that
Otx2 binds to hexasaccharide 5. Biotinylated hexasaccharide is revealed
by streptavidin-HRP (a) and Otx2 protein by anti-Otx2 antibody (b).
c) Gel shift assay using a biotinylated DNA probe showing the competi-
tion ability for Otx2 binding of commercial chondroitin sulfates and of
the hexasaccharide 2. d) Quantification of the DNA shift intensity, com-
pared to control (probe + Otx2).
Conclusion
The efficient synthetic approach described here allowed us
to obtain CS-E oligosaccharides analogues from the cheap
and abundant lactose disaccharide. The easy access to highly
differentiated lactoside blocks lead to the rapid construction
of structurally complex oligosaccharides (up to octasacchar-
ide) in good yields and stereoselectivity. The building block
17 has proved to be very versatile and has enabled us to di-
rectly use the allyl-protected oligosaccharides for the cou-
pling reaction with biotin. This biotinylation was based on
the copper-catalyzed Huisgen cycloaddition and was success-
fully applied at two different stages: on final sulfated oligo-
saccharides or on protected intermediates. This attractive
ure 5b). This demonstrates that Otx2 binds the CS-E-like
hexasaccharide.
The GAG-binding domain of Otx2 corresponds to the
first helix of its DNA-binding domain (homeodomain) plus
the RK doublet upstream of the helix. We thus formed the
hypothesis that a CS recognized by Otx2 should antagonize
its binding to specific promoter sequences. Using the gel
shift assay with an IRBP1 DNA probe that binds specifically
to Otx2, we tested the competing activities of different com-
mercial sugars. At 4 pmol, CS-E almost entirely displaced
the transcription factor from its cognate binding site (Fig-
ure 5c–d). CS-D was slightly less active and CS-A and CS-C
were devoid of competing activity. Strikingly, the synthetic
hexasaccharide 2 presented the highest competing activity,
suggesting that cell surface and DNA recognition involve
similar domains of Otx2, in particular the first helix of the
homeodomain.
“click biotinylation” method can be extended to other al
ated substrates.
The preliminary biological investigations showed good ac-
ACHTUNGTRENNUNGlyl-
AHCTUNGTRENNUNG
tivities of the designed analogues with two different GAG
protein receptors. Moreover, the specific interaction with
HGF/SF (regarding size dependency) and CXCL12g versus
CXCL12a (regarding HS/CS binding) demonstrated the
ability of the analogues to mimic CS fragments, which vali-
dates the proof of concept. The biological properties of CS
analogues will be further explored through studies with CS-
specific proteins. It is worth noting that the CS disaccharide
As already mentioned, the preferential internalization of
endogenous Otx2 by PV-cells is mediated by its specific
binding to complex sugars present at the surface of the cells.
538
www.chemeurj.org
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 531 – 540

Libraries of Biotinylated Chondroitin Sulfate Analogues
FULL PAPER
homeoprotein recognition and
transfer. It is of note that the
hexasaccharide competes with
specific DNA probes for Otx2
binding. This means that the
same domain, encompassing
the 1st helix of the homeodo-
main, is involved in DNA and
CS-E recognition. It is thus pos-
sible to envisage the use of syn-
thetic GAGs composing the
“sugar code” to antagonize the
transcriptional activity of dis-
tinct homeoproteins.
The success of the nitrogen-
to-oxygen replacement ap-
proach and the preliminary in-
teresting biological results re-
ported here have opened the
way toward the preparation of
other CS analogues (of a differ-
ent series) but also other GAG
analogues, such as dermatan
sulfate (DS) or keratane sulfate
(KS). These compounds could
be extensively used for various
applications such as structure–
activity studies, understanding
of complex biological processes,
or medicinal chemistry.
Figure 6. Otx2 binding to hexasaccharide interferes with Otx2 transfer in vivo: a,b) Hexasaccharide 2 infused
for 3 days in adult V1 decreases the number of Otx2-stained cells. c,d) Hexasaccharide 2 infusion decreases
the number of WFA-stained cells, but not that of PV-positive cells. Arrows indicate PV-cells that are not sur-
rounded by WFA when staining. e) Percentage of Otx2, WFA, and PV-positive cells in the supragranular
layers of V1b (infused compared to contralateral, untreated side), 3 days or 7 days post infusion. Scale bar=
100 mm.
unit E has been associated with the formation of osteosarco-
ma cell line tumor nodules in the liver and with metastasis
of Lewis lung carcinoma cells, presumably through binding
to a number of growth factors including VEGF, FGF, and
HGF/SF.[63] There is also strong evidence that over-expres-
sion or the HGF/SF-MET system contributes to growth, in-
vasion, and metastasis in a large variety of tumors.[64] There-
fore interfering with the HGF/SF-MET interactions, with
the CS-E analogues might be of interest in such pathologies.
It is also worth noting that amongst the many interleukins
that bind to heparin, some of them have been shown to in-
teract with chondroitin sulfate, including interleukins-4,[65]
7,[66] and 8.[67] The analogues described in this study could
thus be of interest in the context of the immune response.
The CS-E-like hexasaccharide binds to the homeoprotein
Otx2. Our in vivo experiments showed that its infusion into
the adult visual cortex can block the transfer of Otx2 into
PV-cells, suggesting that it competes with endogenous
GAGs involved in Otx2 recognition. The permanent transfer
and capture of Otx2 by PV-cells is necessary to maintain a
consolidated non-plastic state of the adult visual cortex.[56,57]
The hexasaccharide 2 could then be used to open windows
of plasticity in the adult cortex. Other homeoproteins that
can be transferred between cells possess a GAG-binding
domain, raising the intriguing possibility of a sugar code for
Acknowledgements
The CXCL12g cDNA was a kind gift from Dr. F. Arenzana-Seisdedos
(Pasteur Institut). Cꢀdric Laguri is acknowledged for the purification of
CXCL12 and Gꢀrard Bolbach for recording the MALDI spectra. We
thank the Agence Nationale pour la Recherche (ANR-NT05-4 41967,
ANR-10-BLAN-141-01 and ANR-11-BLAN-069467) and the Global Re-
search Laboratory Program (2009-00424) from the Korean Ministry of
Education for financial support.
[1] S. Mizuguchi, T. Uyama, H. Kitagawa, K. H. Nomura, K. Djima, K.
Gengyo-Ando, S. Mitani, S. Sugahara, K. Nomura, Nature 2003, 423,
443–448.
[2] A. R. Im, Y. Park, Y. Shik Kim, Biol. Pharm. Bull. 2010, 33, 1268–
1273.
[3] W. W. Du, B. B. Yang, T. A. Shatseva, B. L. Yang, Z. Deng, PLoS
ONE 2010, 5, e13828.
[4] D. M. Snow, D. A. Steindler, J. Silver, Dev. Biol. 1990, 138, 359–376.
[5] P. A. Brittis, D. R. Canning, J. Silver, Science 1992, 255, 733–736.
[6] C. L. Dou, J. M. Levine, J. Neurosci. 1995, 15, 8053–8066.
[7] T. Pizzorusso, P. Medini, N. Berardi, S. Chierzi, J. W. Fawcett, L.
Maffei, Science 2002, 298, 1248–1251.
[8] K. E. Rhodes, J. W. Fawcett, J. Anat. 2004, 204, 33–48.
[9] S. E. Tully, R. Mabon, C. I. Gama, S. M. Tsai, X. W. Liu, L. C.
Hsieh-Wilson, J. Am. Chem. Soc. 2004, 126, 7736–7737.
[10] S. Sugiyama, A. A. Di Nardo, S. Aizawa, I. Matsuo, M. Volovitch,
A. Prochiantz, T. K. Hensch, Cell 2008, 134, 508–520.
[11] M. Jalkanen, M. Jalkanen, J. Cell Biol. 1992, 116, 817–825.
Chem. Eur. J. 2013, 19, 531 – 540
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
539

J.-M. Mallet et al.
[12] A. Alkhalil, R. N. Achur, M. Valiyaveettil, C. F. Ockenhouse, D. C.
Gowda, J. Biol. Chem. 2000, 275, 40357–40364.
[13] G. A. Bentley, B. Gamain, Nat. Struct. Mol. Biol. 2008, 15, 895–897.
[14] C. I. Gama, S. E. Tully, N. Sotogaku, P. M. Clark, M. Rawat, N. Vai-
dehi, W. A. Goddard, A. Nishi, L. C. Hsieh-Wilson, Nat. Chem.
Biol. 2006, 2, 467–473.
[15] J. C. Jacquinet, Carbohydr. Res. 1990, 199, 153–181.
[16] K. W. Neumann, J. Tamura, T. Ogawa, Bioorg. Med. Chem. 1995, 3,
1637–1650.
[17] J. Tamura, K. W. Neumann, S. Kurono, T. Ogawa, Carbohydr. Res.
1997, 305, 43–63.
[43] J.-L. M. Abboud, C. Foces-Foces, R. Notario, R. E. Trifonov, A. P.
Volovodenko, V. A. Ostrovskii, I Alkorta, J. Elguero, Eur. J. Org.
Chem. 2001, 3013–3024.
[44] F. Fazio, M. C. Bryan, O. Blixt, J. C. Paulson, C. H. Wong, J. Am.
Chem. Soc. 2002, 124, 14397–14402.
[45] K. D. Bodine, D. Y. Gin, M. S. Gin, J. Am. Chem. Soc. 2004, 126,
1638–1639.
[46] S. Sen Gupta, J. Kuzelka, P. Singh, W. G. Lewis, M. Manchester,
M. G. Finn, Bioconjugate Chem. 2005, 16, 1572–1579.
[47] M. Yamaguchi, K. Kojima, N. Hayashi, I. Kakizaki, A. Kon, K. Ta-
kagaki, Tetrahedron Lett. 2006, 47, 7455–7458.
[48] R. R. Kale, H. Mukundan, D. N. Price, J. F. Harris, D. M. Lewallen,
B. I. Swanson, J. G. Schmidt, S. S. Lyer, J. Am. Chem. Soc. 2008, 130,
8169.
[18] F. Bꢀlot, J. C. Jacquinet, Carbohydr. Res. 2000, 326, 88–97.
[19] J. Tamura, M. Tokuyoshi, Biosci. Biotechnol. Biochem. 2004, 68,
2436–2443.
[49] H. H. M. Abdu-Allah, K. Watanabe, K. Hayashizaki, Y. Iwayama,
H. Takematsu, Y. Kozutsumi, T. Tsubata, H. Ishida, M. Kiso, Tetra-
hedron Lett. 2009, 50, 4488–4491.
[50] A. V. Kornilov, A. A. Sherman, L. O. Kononov, A. S. Shashkov,
N. E. Nifant’ev, Carbohydr. Res. 2000, 329, 717–730.
[51] C. Chiappe, P. Crotti, E. Menichetti, M. Pineschi, Tetrahedron:
Asymmetry 1998, 9, 4079–4088.
[52] W. I. Yu, G. Guizzunti, T. L. Foley, M. D. Burkart, J. J. La Clair, J.
Nat. Prod. 2010, 73, 1659–1666.
[53] A. Natarajan, W. J. Du, C. Y. Xiong, G. L. DeNardo, S. J. DeNardo,
J. Gervay-Hague, Chem. Commun. 2007, 695–697.
[54] J. Zemplen, D. M. Mock, J. Nutr. 1999, 129, 494S–497S.
[55] C. Birchmeier, W. Birchmeier, E. Gherardi, G. F. Vande Woude,
Nat. Rev. Mol. Cell Biol. 2003, 4, 915–925.
[56] S. Sarrazin, M. Lyon, J. A. Deakin, M. Guerrini, P. Lassalle, M. Del-
ehedde, H. Lortat-Jacob, Glycobiology 2010, 20, 1380–1388.
[57] M. Lyon, J. A. Deakin, D. Lietha, E. Gherardi, J. T. Gallagher, J.
Biol. Chem. 2004, 279, 43560–43567.
[58] H. Lortat-Jacob, Curr. Opin. Struct. Biol. 2009, 19, 543–548.
[59] K. R. Catlow, J. A. Deakin, Z. Wei, M. Delehedde, D. G. Fernig, E.
Gherardi, J. T. Gallagher, M. S. Pavao, M. Lyon, J. Biol. Chem. 2007,
283, 5235–5248.
[60] C. Laguri, R. Sadir, P. Rueda, F. Baleux, P. Gans, F. Arenzana-Seis-
dedos, H. Lortat-Jacob, PLoS One 2007, 2, e1110.
[61] A. Joliot, A. Prochiantz, Nat. Cell Biol. 2004, 6, 189–196.
[62] M. Beurdeley, J. Spatazza, H. C. Lee, S. Sugiyama, C. Bernard,
A. A. Di Nardo, T. K. Hensch, A. Prochiantz, J. Neurosci. 2012, 32,
9429–9437.
[63] S. Basappa, S. Murugan, K. N. Sugahara, C. M. Lee, G. B. ten Dam,
T. H. van Kuppevelt, M. Miyasaka, S. Yamada, K. Sugahara, Glyco-
biology 2009, 19, 735–742.
[64] E. Gherardi, W. Birchmeier, C. Birchmeier, G. Van de Woude, Nat.
Rev. Cancer 2012, 12, 89–103.
[65] H. Lortat-Jacob, P. Garonne, J. Banchereau, J. A. Grimaud, Cyto-
kine 1997, 9, 101–105.
[20] C. Lopin, J. C. Jacquinet, Angew. Chem. 2006, 118, 2636–2640;
Angew. Chem. Int. Ed. 2006, 45, 2574–2578.
[21] J. Tamura, Y. Nakada, K. Taniguchi, M. Yamane, Carbohydr. Res.
2008, 343, 39–47.
[22] J. C. Jacquinet, C. Lopin-Bon, A. Vibert, Chem. Eur. J. 2009, 15,
9579–9595.
[23] M. Petitou, C. A. A. van Boeckel, Angew. Chem. 2004, 116, 3180–
3196; Angew. Chem. Int. Ed. 2004, 43, 3118–3133.
[24] L. Bastide, B. Priem, S. Fort, Carbohydr. Res. 2011, 346, 348–351.
[25] G. Zheng, A. Graham, M. Shibata, J. R. Missert, A. R. Oseroff, T. J.
Dougherty, R. K. Pandey, J. Org. Chem. 2001, 66, 8709–8716.
[26] K. Ishihara, H. Kurihara, H. Yamamoto, J. Org. Chem. 1993, 58,
3791–3793.
[27] X. M. Zhu, S. Kawatkar, Y. Rao, G. J. Boons, J. Am. Chem. Soc.
2006, 128, 11948–11957.
[28] V. Bou, J. Vilarrasa, Tetrahedron Lett. 1990, 31, 567–568.
[29] P. Pasetto, R. W. Franck, J. Org. Chem. 2003, 68, 8042–8060.
[30] A. Vibert, PhD Thesis, Universitꢀ d’Orlꢀans (F), 2009.
[31] S. Rat, D. Mathiron, P. Michaud, J. Kovensky, A. Wadouachi, Tetra-
hedron 2007, 63, 12424–12428.
[32] J. L. de Paz, J. Angulo, J. M. Lassaletta, P. M. Nieto, M. Redondo-
Horcajo, R. M. Lozano, G. Gimenez-Gallego, M. Martin-Lomas,
ChemBioChem 2001, 2, 673–685.
[33] K. Ohara, J. C. Jacquinet, D. Jouanneau, W. Helbert, M. Smietana,
J. J. Vasseur, J. Am. Soc. Mass Spectrom. 2009, 20, 131–137.
[34] M. Collot, B. Sendid, A. Fievez, C. Savaux, A. Standaert-Vitse, M.
Tabouret, A. S. Drucbert, P. M. Danze, D. Poulain, J. M. Mallet, J.
Med. Chem. 2008, 51, 6201–6210.
[35] A. Vibert, C. Lopin-Bon, J. C. Jacquinet, Chem. Eur. J. 2009, 15,
9561–9578.
[36] A. Vibert, C. Lopin-Bon, J. C. Jacquinet, Eur. J. Org. Chem. 2011,
4183–4204.
[37] S. Fort, L. Birikaki, M. P. Dubois, T. Antoine, E. Samain, H. Dri-
guez, Chem. Commun. 2005, 2558–2560.
[38] M. A. Bernstein, L. D. Hall, Carbohydr. Res. 1980, 78, C1–C3.
[39] S. M. Rele, S. S. Iyer, E. L. Chaikof, Tetrahedron Lett. 2007, 48,
5055–5060.
[40] R. Huisgen, Angew. Chem. 1963, 75, 604–637.
[41] V. V. Rostovtsev, L. G. Green, V. V. Fokin, K. B. Sharpless, Angew.
Chem. 2002, 114, 2708–2711; Angew. Chem. Int. Ed. 2002, 41, 2596–
2599.
[66] D. Clarke, O. Katoh, R. V. Gibbs, S. D. Griffiths, M. Y. Gordon, Cy-
tokine 1995, 7, 325–330.
[67] C. W. Frevert, M. G. Kinsella, C. Vathanaprida, R. B. Goodman,
D. G. Baskin, A. Proudfoot, T. N. Wells, T. N. Wight, T. R. Martin,
Am. J. Respir. Cell Mol. Biol. 2003, 28, 464–472.
Received: June 19, 2012
Revised: August 28, 2012
Published online: November 14, 2012
[42] C. W. Tornoe, C. Christensen, M. Meldal, J. Org. Chem. 2002, 67,
3057–3064.
540
www.chemeurj.org
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 531 – 540