Double affinity amplification of galectin-ligand interactions through arginine-arene interactions: Synthetic, thermodynamic, and computational studies with aromatic diamido thiodigalactosides

Article abstract of DOI:10.1002/chem.200701932

A series of aromatic monoor diamido-thiodigalactoside derivatives were synthesized and studied as ligands for galectin-1, -3, -7, -8N terminal domain, and -9N terminal domain. The affinity determination in vitro with competitive fluorescence-polarization

Full text of DOI:10.1002/chem.200701932

FULL PAPER
DOI: 10.1002/chem.200701932
Double Affinity Amplification of Galectin–Ligand Interactions through
Arginine–Arene Interactions: Synthetic, Thermodynamic, and
Computational Studies with Aromatic Diamido Thiodigalactosides
Ian Cumpstey,[a] Emma Salomonsson,[b] Anders Sundin,[a] Hakon Leffler,[b] and
Ulf J. Nilsson*[a]
Abstract: A series of aromatic mono-
or diamido-thiodigalactoside deriva-
tives were synthesized and studied as
ligands for galectin-1, -3, -7, -8N termi-
nal domain, and -9N terminal domain.
The affinity determination in vitro with
competitive fluorescence-polarization
experiments and thermodynamic analy-
sis by isothermal microcalorimetry pro-
vided a coherent picture of structural
requirements for arginine–arene inter-
actions in galectin–ligand binding.
Computational studies were employed
to explain binding preferences for the
different galectins. Galectin-3 formed
two almost ideal arene–arginine stack-
ing interactions according to computer
modeling and also had the highest af-
finity for the diamido-thiodigalacto-
sides (Kd below 50 nm). Site-directed
mutagenesis of galectin-3 arginines in-
volved in binding corroborated the im-
portance of their interaction with the
aromatic
diamido-thiodigalactosides.
Furthermore, the arginine mutants re-
vealed distinct differences between
free, flexible, and solvent-exposed argi-
nine side chains and tightly ion-paired
arginine side chains in interactions with
aromatic systems.
Keywords: carbohydrates · galectin ·
inhibitors · thioglycosides
Introduction
fragment showed anticancer effects in a mouse model.[8] Ga-
lectin-1 has immunomodulatory effects,[9,10] may promote
cancer, and molecules interacting with galectin-1 may inhibit
cancer growth.[11] Also other galectins have been implicated
in cancer[12–14] and are involved in inflammation.[15]
The galectins are a family of soluble proteins that are de-
fined by their affinity for b-galactosides and a conserved se-
quence motif[1,2] with approximately 15 known in humans.
Galectin-3 is a multifunctional protein,[3] but experiments in
cell cultures and with null mutant mice clearly indicate rate-
limiting roles in inflammatory conditions,[4–6] making it a po-
tential target for the development of inhibitors as potential
anti-inflammatory agents. Galectin-3 has also been suggest-
ed to play a role in cancer,[7] and an inhibitory galectin-3
The mechanisms by which galectins influence inflamma-
tion and cancer include modulation of apoptosis, cell adhe-
sion, angiogenesis, growth-factor signaling, and, possibly as a
result, cell differentiation. Most of these effects appear to
require the carbohydrate-binding activity of the galectins,
making their carbohydrate-binding site a suitable target for
inhibitor design. As galectins may form noncovalent di- or
multimers of carbohydrate-recognition domains (CRDs) or
have two CRDs within the same peptide chain, their binding
to glycoconjugates results in cross-linking and lattice forma-
tion.[16] This in turn is thought to be their mechanism of
action, resulting in signaling or regulation of receptor target-
ing or residence time at the cell surface.[17–20] For example,
biosynthetic control of membrane-bound protein glycoforms
has been demonstrated to modulate galectin-1 binding and
T-cell apoptosis.[9] Moreover, it was recently demonstrated
that intracellular glycan-dependent clustering involving ga-
lectin-3 plays a decisive role in orchestrating raft-independ-
ent apical sorting of proteins.[21]
[a] Dr. I. Cumpstey, A. Sundin, Dr. U. J. Nilsson
Organic Chemistry
Lund University
POB 124, 22100 Lund (Sweden)
Fax : (+46)46-222-8218
E-mail: ulf.nilsson@organic.lu.se
[b] E. Salomonsson, Dr. H. Leffler
Section MIG, Department of Laboratory Medicine
Lund University
Sçlvegatan 23, 22362 Lund (Sweden)
Supporting information for this article is available on the WWW
under http://www.chemistry.org or from the author.
Chem. Eur. J. 2008, 14, 4233 – 4245
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4233

The above observations all point towards clinical implica-
tions of modulating galectin–carbohydrate binding activities.
Galectin-3C, the cancer-inhibiting fragment of galectin-3
mentioned above, competes for the same carbohydrate li-
gands, but is incapable of forming aggregates or lattices, and
hence, inhibits the function of intact galectin-3.[8] The fact
that galectin-3C had no toxic effects in mice together with
the fact that galectin-3 null mice are healthy under animal
house conditions,[22] suggests that it should be possible to
find a therapeutically favorable degree of inhibition of ga-
lectin-3 carbohydrate-binding activity without serious side
effects. For this purpose, it is desirable that selective mono-
valent low-molecular-weight inhibitors could be synthesized
and such molecules could also be of use as biomolecular
tools to investigate the biological mechanisms of the galec-
tins.[23]
We recently reported that large increases in affinity for
galectin-3 could be achieved by substitution of the 3’-posi-
tion of N-acetyllactosamine by various aromatic amide moi-
eties.[24,25] An X-ray crystal structure of one of these ligands
bound into the binding site of galectin-3 showed that to
form the complex, an arginine residue (Arg144) had moved
3.5 Š to form a face-to-face interaction between its guanidi-
nium functionality and the aromatic ring of the amide sub-
stituent on N-acetyllactosamine (LacNAc)[25] (Scheme 1a).
Galectin-3 has another arginine residue in the vicinity of the
bound disaccharide. This arginine appeared to be an attrac-
tive option to target with a second arene–arginine interac-
tion and thus form even more strongly binding inhibitors.
However, rather than use LacNAc as a skeleton to build on,
we decided to investigate the simpler and non-natural hy-
drolytically stable disaccharide thiodigalactoside. Thiodiga-
lactoside has been shown to bind to galectins with a similar
affinity to lactose or LacNAc. In addition, X-ray crystal
structures were recently published of a toad galectin in com-
plex with LacNAc and with thiodigalactoside; the two galac-
tose residues were bound identically in the b-galactoside
binding site, whereas the second galactose of thiodigalacto-
side was bound in the same place as the GlcNAc of
LacNAc, with identical hydrogen bonding networks between
the disaccharides and the protein[26] (Scheme 1b). It seems
logical to assume that derivatization of thiodigalactoside at
the 3-position of the galactose in the b-galactoside binding
site will give an aromatic amide that can interact with galec-
tin arginine side chains, for example, Arg144 in galectin-3.
However, the 3-position of the second galactose residue is
positioned such that similar derivatization at this position
will give an amide directed over another arginine side chain,
for example, Arg186 in galectin-3 (Scheme 1c).
Following on from earlier studies,[27] we report herein de-
tails of the synthesis of thiodigalactoside derivatives that
bear two identical amides at the two 3-positions (i.e. C2-sym-
metrical compounds). In addition, we report thiodigalacto-
sides having two different amides at the 3- and 3’-positions,
or a single amide at the 3-position of one of the galactose
residues, which provided the opportunity to undertake a sys-
tematic study to assess the importance of the presence of
two aromatic amides and to try to optimize interaction with
each of the two relevant arginine residues. We also disclose
the results of fluorescence-polarization binding studies of
the thiodigalactoside derivatives to galectins-1, -3, -7, -8N,
and -9N, as well as microcalorimetry studies of the binding
to galectin-3 (C-terminal domain). Furthermore, computa-
tional modeling of the interaction of the thiodigalactoside
derivatives with galectins corroborated the hypothesis, and
binding studies with galectin-3 mutants in which Arg144 and
Arg186 were mutated to serine provided a picture of the rel-
ative importance of the two arginines in the structurally dif-
ferent interactions with aromatic amides.
Results and Discussion
Chemical synthesis: Initial attempts to synthesize 3-amide-
derivatized thiodigalactosides were based on a symmetrical
synthesis of the disaccharide in which both glycosidic bonds
are formed in a single reaction step by using two equivalents
of the glycosyl bromide and sodium sulfide as a nucleophile.
The synthesis of thiodigalactoside has been published by
using procedures based on this concept and phase-transfer
conditions.[28] To keep the common intermediate as long as
possible in the synthesis before diversification and synthesis
of different amides, we initially attempted to carry out a
sodium sulfide mediated glycosylation by using 2,4,6-tri-O-
acetyl-3-azido-3-deoxy-d-galactosyl bromide (6).[29] Unfortu-
Scheme 1. a) Galectin-3 complex with a LacNAc 3’-benzamido derivative
that has high affinity for galectin-3 owing to interactions between the ar-
omatic moiety and Arg144. b) Thiodigalactoside with hydroxy groups
that form key interactions with galectin-1 (shown in bold). c) 3,3’-Diami-
do-thiodigalactoside derivatives proposed to form double interactions
with arginine side chains of galectin-3 (Arg144 and Arg186).
4234
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 4233 – 4245

Galectin–Ligand Interactions
FULL PAPER
nately, but not unexpectedly, this reaction produced an in-
tractable mixture of products, presumably owing to reduc-
tion of the azide functionality. Instead, an earlier diversifica-
tion was necessary for the synthesis of symmetrical dia-
mides. Thus, galacto azide 1[29] (Scheme 2) was reduced by
catalytic hydrogenation, and the resulting amine was acylat-
ed with the respective aromatic acyl chlorides to give the
amides 2a–e (Scheme 3). Bromination of 2a–e was carried
out by using HBr in AcOH, and the crude bromides 3a–e
were subjected to treatment with dried sodium sulfide in
acetonitrile to give the thiodigalactosides 4a–e in moderate
yields. For the amides with more electron-rich aromatic sys-
tems, leaving the bromination reaction for longer resulted in
the formation of side products. This was particularly pro-
nounced for the 2-naphthamide derivative 3b, which could
help to explain the low yield in the dimerization reaction. In
contrast, the electron-poor para-nitro derivative 3e was
formed very cleanly. Deprotection of the thiodigalactoside
derivatives 4a–e was carried out by using either an amine or
sodium methoxide in methanol to give the desired symmet-
rical diamides 5a–e.
Even though the azide functionality in 6 was intolerant of
sodium sulfide, it has been shown that an azide can survive
similar reactions in which other sulfur nucleophiles are pres-
ent. Thus, for the synthesis of unsymmetrical amides, we de-
cided to try to substitute the azido bromide 6 with an
anomeric thiol 7. Treatment of the azido galactosyl bromide
6 with galacto thiol 7 under the same conditions gave the de-
sired monoazido thiodigalactoside derivative 8. Catalytic hy-
drogenolysis of the monoazido disaccharide 8 and subse-
quent acylation with aromatic acyl chlorides gave the mono-
amides 9a–c, which were then deacetylated to give the de-
protected monoamide-substituted thiodigalactosides 10a–c
(Scheme 3).
To synthesize unsymmetrical thiodigalactoside diamides,
that is, derivatives bearing different amides at the 3- and 3’-
positions, we planned to again use the azido galactosyl bro-
mide 6 as the electrophile, however, a galacto thiol with an
aromatic amide already in place at the 3-position (12) was
used as a nucleophile this time. Thus, the dimethoxybenza-
mide-substituted galactose derivative 2c was brominated as
above to give 3c, and an anomeric thioacetyl group was in-
troduced by using potassium thioacetate to give 11
(Scheme 4). Selective hydrolysis of the thioacetate was ach-
ieved by using sodium methoxide at ꢀ408C, and the crude
anomeric thiol 12 was treated with the azido bromide 6
under the same conditions as described earlier for the un-
symmetrical glycosylations to give the desired disaccharide
13 as the only identified product. Catalytic hydrogenolysis
and subsequent acylation with 2-naphthoyl chloride and 1-
naphthoyl chloride gave the unsymmetrical diamides 14 and
15, respectively, which were deacetylated to give the depro-
tected unsymmetrical diamide-substituted thiodigalactosides
16 and 17 (Scheme 5).
Scheme 2. a) 1. H2, Pd/C, EtOH, HCl (Et2O); 2. RCOCl, DCM, pyridine,
DMAP, 40–91% over two steps. b) HBr/AcOH, Ac2O, DCM. c) Na2S
(dried), MeCN, 4-Š molecular sieves. d) NaOMe, MeOH. e) MeNH2,
H2O, 7–27% over three steps. DCM=dichloromethane, DMAP=4-di-
methylaminopyridine.
Binding of thiodigalactoside amides to galectin-1, -3, -7,
-8N, and -9N: The set of five symmetrical diamides 5a–e,
three monoamides 10a–c, and two unsymmetrical diamides
16 and 17 were evaluated for binding to galectin-1, -3, -7,
-8N, and -9N by using a fluorescence polarization assay
(Table 1).[31,32] Data for aromatic LacNAc C3’-amides 18–
23[25] and lactose C2 esters 24–27[30] are included for refer-
ence as the they address the specific interactions of Arg144
and Arg186 from galectin-3 and the corresponding residues
in galectin-1, -7, -8N, and -9N. The unsubstituted thiodiga-
lactoside 28 was also included for reference, as were the
Scheme 3. a) K2CO3, MeCN, RT, 72%. b) 1. H2, EtOH, Pd/C, HCl
(Et2O); 2. RCOCl, DCM, pyridine; 9a, 43%; 9b, 40%; 9c, 66%.
c) MeNH2, H2O; 10a, 95%; 10b, 66%; 10c, 60%.
Chem. Eur. J. 2008, 14, 4233 – 4245
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4235

U. J. Nilsson et al.
Scheme 5. Generic structures of compounds evaluated against galectin-1,
-3, -7, -8N, and -9N.
Scheme 4. a) HBr (33% in AcOH), DCM. b) KSAc, MeCN, 72% from
2c. c) NaOMe (1 equiv), MeOH. d) 6, K2CO3, MeCN, 39% from 1.
e) 1. H2, Pd/C, EtOH, HCl; 2. 1-naphthoyl chloride, DCM, py. f) 1. H2,
Pd/C, EtOH, HCl; 2. 2-naphthoyl chloride, DCM, py. g) MeNH2 H2O, 16,
67% from 13, 17, 50% from 13.
thamide 5b (0.16 mm)>10b (1.8 mm)>28 (49 mm), and 3,5-di-
methoxybenzamide 5c (0.046 mm)>10c (1.1 mm)>28
(49 mm)). The question then emerged as to whether the two
arginine side chains, Arg144 and Arg186, display different
preferences for aromatic structures. To answer this question,
the inclusion of the known LacNAc 3’-amides 18–23 and the
lactose 2-esters 24–27 in addition to the unsymmetrical thio-
digalactosides 16 and 17 in the analysis is important. This is
because the unsymmetrical nature of these two classes of
compounds places their aromatic moieties in a position that
allows them to interact with only Arg144 or Arg186, respec-
tively. Detailed insight into the preferences of Arg144 and
Arg186 for different aromatic structures could provide guid-
ance for the design of aromatic amides that are individually
optimized to interact with either of the arginine residue.
The aromatic moieties of the LacNAc amides, 3’-amides 18–
23, interact with Arg144 and a preference for the 3,5-dime-
thoxy and 4-nitrobenzamides 21 and 22 and the 2-naphtha-
mide 20 is observed. The 1-naphthamide 19 is almost one
order of magnitude worse than the 2-naphthamide 20. In
contrast, the 1-naphthoate 25 is twice as potent as the 2-
naphthoate 26, which suggests that Arg186 prefers the 1-
naphthyl group, as the aromatic moieties of the lactose 2-
esters 24–27 are bound to interact with Arg186. The 3,5-di-
methoxybenzoate 27 binds well to galectin-3 and the 3,5-di-
methoxyphenyl group can thus be concluded to interact well
with both Arg144 and Arg186. The latter observation helps
to explain why, although both unsymmetrical diamides 16
methyl glycosides of LacNAc 29 and lactose 30 (Scheme 5
and Table 1).
Galectin-1 showed some increased affinity for 5a–d over
thiodigalactoside 28, presumably owing to formation of one
arginine–arene interaction. This galectin lacks an arginine in
close proximity to galactose C3 of the complexed lactose or
LacNAc, which is reflected by the observation that aromatic
LacNAc C3’ amides 18–23 show limited affinity enhance-
ment over the parent LacNAc 29. The aromatic C2 lactose
esters 24–27 display significantly improved affinity over the
parent lactoside 30 owing to arginine–arene interactions
with Arg74,[30] suggesting that the interaction of the amido
groups of 5a–d with the galectin-1 Arg74 group is responsi-
ble for their improved affinity.
In the case of galectin-3, the increase in affinity of the
amide-substituted derivatives over the parent thiodigalacto-
side 28 is particularly impressive, with increases as great as
several hundred-fold for the best inhibitors. In addition, a
progressive increase in affinity is observed for galectin-3
along each of the homologous series of unsubstituted, mono-
substituted, and disubstituted thiodigalactosides studied (i.e.
benzamide 5a (1.7 mm)>10a (3.2 mm)>28 (49 mm), 2-naph-
4236
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 4233 – 4245

Galectin–Ligand Interactions
FULL PAPER
Table 1. Affinity constants (mm) of thiodigalactoside derivatives and reference compounds for galectins-1, -3,
In galectins-7 and -9N, both
of the relevant arginine resi-
dues are present (i.e. Arg31 and
Arg79 for galectin-7 and Arg44
and Arg87 for galectin-9N),
and an increase in affinity is
seen for the amide-substituted
thiodigalactosides over the
parent compound 28. Notably,
the increases in affinity were
not as big for these galectins as
they were for galectin-3. In ad-
dition, it was not the same com-
pounds that had the highest af-
finity for all the different galec-
tins. This means that the affini-
ties of the thiodigalactosides for
the galectins are governed by
the local environment around
the bound ligands, where, for
example, steric and electronic
factors can be very different be-
tween the different galectins.
For example, the higher affinity
of the mono-(3,5-dimethoxy-
benzamide) derivative 10c than
its bis-substituted analogue 5c
for both galectin-7 and -9N may
arise from a steric clash by the
-7, -8N, and -9N.[a]
Galectin
-1
-3
-7
-8N
-9N
symmetrical diamido-thiodigalactosides
5a
5b
5c
5d
5e
R=phenyl
R=2-naphthyl
R=3,5-dimethoxyphenyl
R=3-methoxyphenyl
R=4-nitrophenyl
35ꢁ22
1.7ꢁ0.4[b]
4.3ꢁ3.5 high
7.0ꢁ1.5
9.6ꢁ7.2
4.7ꢁ1.4
0.16ꢁ0.04[b]
0.046ꢁ0.020[b]
0.050ꢁ0.012[b]
0.049ꢁ0.029[b]
1.7ꢁ0.2
17ꢁ0.2
2.8ꢁ1.1
4.0ꢁ1.0
>100
0.73ꢁ0.14
0.9ꢁ0.3
high
ꢂ2
ꢂ100
1.8ꢁ0.4
n.d.[c]
>100
0.68ꢁ0.16
monoamido-thiodigalactosides
10a R=phenyl
10b R=2-naphthyl
n.d.[c]
n.d.[c]
n.d.[c]
3.2ꢁ0.2
1.8ꢁ0.4
1.1ꢁ0.1
17ꢁ5
350ꢁ17
2.1ꢁ0.8
9.8ꢁ4.4 30ꢁ5
0.42ꢁ0.18
0.69ꢁ0.07
10c
R=3,5-dimethoxyphenyl
10ꢁ2
72ꢁ12
unsymmetrical diamido-thiodigalactosides
16
17
R=2-naphthyl
R=1-naphthyl
n.d.[c]
n.d.[c]
0.069ꢁ0.13
0.052ꢁ0.021
3.5ꢁ1.4 high
1.0ꢁ0.8 high
0.48ꢁ0.15
0.63ꢁ0.02
LacNAc 3’-amides
18
19
20
21
22
23
R=phenyl
R=1-naphthyl
R=2-naphthyl
22ꢁ6
6.7[25]
4.4[25]
0.48[25]
1.1[25]
0.95[25]
2.5[25]
41ꢁ1
high
ꢂ60
25ꢁ8
7.6ꢁ3.3 high
>100
high
11ꢁ4
ꢂ100
ꢂ100
28ꢁ6
37ꢁ5
high
high
high
high
R=3,5-dimethoxyphenyl
R=4-nitrophenyl
R=3- methoxyphenyl
30ꢁ16
8.8ꢁ4.4
67ꢁ22
>100
47ꢁ6
>100
lactose esters[30]
24
25
26
27
R=phenyl
R=1-naphthyl
R=2-naphthyl
R=3,5-dimethoxyphenyl
4.4
14
8.7
21
7.8
2.5
5.2
2.2
145
>1000
530
740
40
54
124
86
1.6
2.7
14
>1000
reference compounds
28
29
30
thiodigalactoside
Me b-LacNAc[30]
Me b-Lac[30]
24ꢁ11
49ꢁ11[b]
160ꢁ18
550
110
61ꢁ17
1000
62
38ꢁ8
490
23
65
59
187
160
[a] Determined by a fluorescence polarization assay. The location of R is shown in Scheme 5. Galectin-1, -7,
and -9N were studied at 08C, whereas galectin-3 and -8N were studied at ambient temperature. [b] Average of
14 experiments provided basically the same Kd values but with improved data statistics as compared with pre-
viously reported data.[31] [c] Not determined.
second
3,5-dimethoxybenza-
mide moiety, which is avoided
in the case of galectin-3.
and 17 bind strongly to galectin-3, the 2-naphthamide 16
binds more tightly. Thus, compounds 16 and 17 presumably
bind galectin-3 in a manner that allows their naphthamides
to interact with Arg144 and the 3,5-dimethoxybenzamide to
interact with Arg186. Computational modeling, as discussed
below, corroborated this hypothesis.
Galectin-7 and -9N show considerable affinity enhance-
ments in response to the presence of one or two aromatic
amides on thiodigalactoside. The case of galectin-8N is, how-
ever, in striking contrast. Here, all but one of the amide-sub-
stituted derivatives have lower affinity than the parent com-
pound, and the one compound with an increased affinity
(10c) only binds two times better than unsubstituted thiodi-
galactoside 28. In interpreting this result, it is useful to com-
pare the amino acid sequences of the different galectins
(Figure 1).
Figure 1. Partial sequence alignment of galectin-1, -3, -7, -8N, and -9N.
The residues corresponding to Arg144 and Arg186 in galectin-3 are high-
lighted.
Clearly, in galectin-8N, there is no arginine residue that
corresponds to the Arg186 of galectin-3 according to se-
quence alignment. This fact, coupled with the low affinity of
the LacNAc amide derivatives 18–23 for galectin-8N, sup-
ports our theory that the diamido-thiodigalactoside deriva-
tives 5a–e bind to galectin-3 with double arene–arginine in-
teractions (with Arg144 and Arg186).
As the aromatic structures chosen had already been
shown to produce derivatives with a high affinity for galec-
tin-3 through interaction with Arg144, albeit on a LacNAc
skeleton, it is perhaps not surprising that the resulting thio-
digalactoside amides had the highest affinity for galectin-3.
It is possible, therefore, that exploration of other analogues
with aromatic amides or other structures would yield com-
Chem. Eur. J. 2008, 14, 4233 – 4245
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4237

U. J. Nilsson et al.
pounds with higher affinities and selectivities for galectin-7
and -9N.
culations were performed. Several starting conformations of
the ligands with respect to amide conformation and confor-
mations of substituents on the aromatic amides were used as
the input in the calculations. All minimizations converged to
low-energy complex structures and the complexes with best
inhibitor of each of the galectins are depicted in Figure 2. In
each case, one of the thiodigalactoside galactoside residues
is bound in the conserved galactose site through stacking of
its b face onto a tryptophan side chain and with HO4 and
HO6 deeply buried and hydrogen bonded. An ion-pairing
networkof arginine groups and carboxylate groups, which
results in an extended flat p-electron system, interacts, in
each of the galectins, except galectin-8N, with one aromatic
Computational studies of thiodigalactosides in complex with
galectins: To obtain a deeper insight into the structural fea-
tures of diamido-thiodigalactosides binding to galectins,
computational studies were performed. Starting from pub-
lished X-ray crystal structures of galectin-1, -7, and -9N in
complex with lactose,[33–35] galectin-3 in complex with a C3’-
amido-derivatised LacNAc-based inhibitor,[25] and a homolo-
gy model of galectin-8N in complex with LacNAc,[36] the dif-
ferent ligand structures 5a–e, 10a–e, 16, and 17 were built
into the galectin binding sites and energy-minimization cal-
Figure 2. Modeled low-energy structures of galectins in complex with their respective best diamido-thiodigalactoside inhibitors. The side chains directly
involved in complex formation with the inhibitors are depicted by transparent surfaces. a) Galectin-1 in complex with 5d; b) galectin-3 in complex with
5c; c) galectin-7 in complex with 5b; d) galectin-8N in complex with 5d; e) galectin-9N in complex with 5e.
4238
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 4233 – 4245

Galectin–Ligand Interactions
FULL PAPER
amide residue of the inhibitors with varying degrees of com-
plementarity. This is in analogy with the observation that
lactose 2-O-benzoates[30] were efficient inhibitors of galec-
tin-1 and -3 and were tolerated by galectin-7 and -9N owing
to stacking of the aromatic esters onto the ion-pair network
of arginine and carboxylate groups. The remaining aromatic
amide residues of the inhibitors are left to interact with vari-
able and nonconserved structural elements near C3 of the
buried galactose residue.
Detailed analysis of the binding to each of the galectins:
In galectin-1, one aromatic amide of the amido-thiodigalac-
tosides 5a–e, 10a–c, 16, and 17 stacks onto an arginine
(Arg73) side chain in an networkof ion-pairing groups of
arginine and carboxylate groups (Figure 2a). The corre-
sponding aromatic amide oxygen group points away from
the protein surface, thus avoiding repulsive van der Waals
contacts with the Glu71 carboxylate. The remaining aromat-
ic amide fills a cavity close to His52, however, only with
moderate shape and electronic complementarity.
In the case of the tightest-binding galectin, galectin-3, all
energy-minimized complexes supported the hypothesis of
two simultaneous affinity-enhancing arginine–arene interac-
tions (Figure 2b). Indeed, the complexes between galectin-3
and amido-thiodigalactosides 5a–e, 10a–c, 16, and 17 all dis-
played high surface complementarities. The lowest-energy
complexes had one aromatic moiety that stacked effectively
onto Arg186, which was involved in an arginine–carboxylate
ion-pairing network. As has been seen in the galectin-1 com-
plexes, the corresponding aromatic amide oxygen groups
point away from the protein surface in all cases, thus avoid-
ing repulsive van der Waals contacts with a glutamate
(Glu184) carboxylate group. The remaining aromatic moiety
shows optimal complementarity with a cavity formed be-
tween the Arg144 side chain and the protein backbone.
Within this cavity, Arg144 stacks in a face-to-face manner
with the aromatic amide stacked in the same way. This has
been observed in a published complex with an aromatic C3
amide of LacNAc.[25] The low-energy complexes with the un-
symmetrical amides 16 and 17 both had the 3,5-dimethoxy-
benzamide moiety stacked onto Arg186. Thus, the naphtha-
mides of 16 and 17 presumably interact with Arg144, which
may explain why the 2-naphthamide 16 binds more tightly;
2-naphthamides are preferred over 1-naphthamides by
Arg144.[25]
One aromatic moiety of the amido-thiodigalactosides 5a–
e, 10a–c, 16, and 17 also stacks face-to-face onto an arginine
(Arg74) of the arginine–carboxylate ion-pair networkof ga-
lectin-7 (Figure 2c). However, as a result of this arene–
Arg74 stacking, the corresponding amide oxygen atom will
end up in an unfavorable position close to the carboxylate
group of either Glu58 or Glu72. The remaining aromatic
moiety of the bis-amido-thiodigalactosides shows compara-
tively poor complementarity with galectin-7 and does not
stackwith Arg31 (corresponding to Arg144 in galectin-3),
but instead interacts edge-to-face with His33. The reason for
this is probably that Arg31 is interacting with Asp55 and
Arg53,[34] thus preventing it from rearranging to stackwith
the aromatic amide of the inhibitor. The corresponding
Arg144 of galectin-3 does not form such distinct interactions
with other amino acids and thus has a lower barrier of rear-
rangement and stacking with the aromatic amide of the in-
hibitor.
An arginine residue corresponding to Arg86 in the argi-
nine–carboxylate networkof galectin-3 is absent in galectin-
8N (mentioned above and shown in Figure 1). However, the
modeling studies with the galectin-8N homology model sug-
gested that one aromatic group of the amido-thiodigalacto-
sides 5a–e, 10a–c, 16, and 17 instead stacked face-to-face
onto another arginine side chain, Arg60, albeit with moder-
ate surface complementarity (Figure 2d). Furthermore, in
such a low-energy conformation, the corresponding amide
oxygen is placed in an unfavorable position close to Glu77,
which is similar to the observation for galectin-7 discussed
above and also in accordance with the observation that lac-
tose 2-O-benzoates bind poorly[30] to this galectin. The re-
maining aromatic amide of the inhibitors 5a–e, 10a–c, 16,
and 17 appears not to be able to interact favorably with ga-
lectin-8N as it shows low shape complementarity and does
not reach the Arg33 side chain (corresponding to Arg144 in
galectin-3). Taken together, none of the aromatic amido
moieties of the amido-thiodigalactosides 5a–e, 10a–c, 16, 17,
the reference LacNAc amides 18–23, or lactose esters 24–27
are involved in favorable interactions with galectin-8N.
The ion-paring networkof the ligand-binding site of ga-
lectin-9N is different in that it has a tyrosine residue that
corresponds to a carboxylate side chain in the other galec-
tins. Nevertheless, one aromatic moiety of the amido-thiodi-
galactosides 5a–e, 10a–c, 16, and 17 stacks face-to-face onto
an arginine group (Arg87) of the ion-pair networkof galec-
tin-9N (Figure 2e). This interaction provides some affinity
amplification, which is corroborated by the observation that
lactose esters 24–27 display somewhat improved binding
over lactose.[30] A possible explanation for the moderate af-
finity amplifications is that the corresponding amide oxygen
is bound to come in contact with a glutamate (Glu85),
which is a structural feature that is analogous to the calcu-
lated complexes with galectin-7 and -8N. The remaining aro-
matic amide moieties of the amido-thiodigalactosides 5a–e,
16, and 17 provide some affinity amplification as well, which
is reflected in the results with the corresponding LacNAc
amides 18–23. This relatively limited effect may be attribut-
ed to the lackof interaction with Arg44 (corresponding to
Arg144 in galectin-3). According to the calculation and the
published structure,[35] Arg44 of galectin-9N preferentially
takes part in interactions with Glu67 and Arg65, a structural
feature that is similar to that observed in galectin-7. Never-
theless, the moderate individual affinity amplifications by
the two aromatic amides of 5a–e, 16, and 17 result in a sig-
nificantly improved total affinity amplification over the
parent thiodigalactoside 28.
Thermodynamic characterization of galectin-3 complexes
with microcalorimetry: To investigate the thermodynamic
characteristics of the interactions, we decided to carry out
Chem. Eur. J. 2008, 14, 4233 – 4245
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4239

U. J. Nilsson et al.
an isothermal calorimetry experiment with the galectin that
was most efficiently inhibited by the thiodigalactoside
amides. Thus, the binding of the five C2-symmetrical dia-
mides 5a–e, the three monoamides 10a–c, and thiodigalacto-
side 28 to galectin-3C (C-terminal) was measured by using
microcalorimetry (Figure 3 and Table 2). Galectin-3C was
used to avoid interfering self-aggregation through the N-ter-
lish good contact simultaneously with both arginine residues
(Arg144 and Arg186), which results in a loss in DH. This
could explain the rather marginal increase in affinity on
going from a mono- to a diamide in this series compared
with the much bigger increases for 10b!5b and 10c!5c.
Binding of the thiodigalactosides to galectin-3 arginine mu-
tants: Two arginine mutants of galectin-3, R144S and
R186S,[30] were evaluated for binding to the thiodigalacto-
side amides 5a–e, 16, and 17 to obtain information regarding
the relative importance of the two arginine–arene interac-
tions (Table 3). Differential scanning calorimetry with the
two mutants showed that they are as stable as the wild type,
Table 3. Dissociation constants (mm) at 48C of thiodigalactoside deriva-
tives and reference compounds for wild-type galectins-3 and R144S and
R186S mutants.[a]
Wt
R144S
R186S
symmetrical diamido-thiodigalactosides
5a R=phenyl
0.62ꢁ0.006
1.4ꢁ0.2
>500
5b R=2-naphthoyl
5c R=3,5- dimethoxy-
phenyl
0.044ꢁ0.020 0.059ꢁ0.027
0.027ꢁ0.009 0.032ꢁ0.010
2.6ꢁ0.8
4.6ꢁ0.4
5d R=3-methoxyphenyl
5e R=4-nitrophenyl
unsymmetrical diamido-thiodigalactosides
0.078ꢁ0.017 0.076ꢁ0.028
0.012ꢁ0.002 0.030ꢁ0.005
15ꢁ2.9
2.8ꢁ0.2
Figure 3. Isothermal titration calorimetry of galectin-3 with 5e.
16 R=2-naphthyl
17 R=1-naphthyl
LacNAc 3’-amides
18 R=phenyl
19 R=1-naphthyl
20 R=2-naphthyl
21 R=3,5- dimethoxy-
phenyl
0.027ꢁ0.014 0.021ꢁ0.003
4.3ꢁ1.4
12ꢁ2.4
0.061ꢁ0.026 0.079ꢁ0.015
Table 2. Isothermal microcalorimetry data for 5a–e, 10a–c, and 28 bind-
ing to galectin-3C.
1.7ꢁ0.2
4.2ꢁ0.7
7.6ꢁ2.3
1.4ꢁ0.1
1.1ꢁ0.3
n.d.[b]
n.d.[b]
n.d.[b]
n.d.[b]
ꢀDH
ꢀTDS
n
Kd
[mm]
0.94ꢁ0.16
0.24ꢁ0.05
0.38ꢁ0.08
[kcalmolꢀ1
]
[kcalmolꢀ1
]
28
5a
8.3ꢁ0.4
9.6ꢁ0.06
10.9ꢁ0.1
15.7ꢁ0.1
10.2ꢁ0.04
12.8ꢁ0.2
9.4ꢁ0.05
15.9ꢁ0.1
13.4ꢁ0.2
2.9
1.5
3.7
5.2
2.3
2.5
1.8
6.7
3.5
0.93
1.04
0.94
0.93
0.96
1.00
1.03
0.93
1.03
118ꢁ3
1.1ꢁ0.1
22 R=4-nitrophenyl
23 R=3-methoxyphenyl
lactose esters[30]
0.16ꢁ0.03
0.67ꢁ0.14
0.66ꢁ0.10
58ꢁ7.9
n.d.[b]
n.d.[b]
10a
5b
10b
5c
10c
5d
5e
5.9ꢁ0.2
0.020ꢁ0.003
1.8ꢁ0.1
24 R=phenyl
1.3ꢁ0.1
n.d.[b]
n.d.[b]
n.d.[b]
n.d.[b]
high
high
0.029ꢁ0.005
2.7ꢁ0.1
25 R=1-naphthyl
26 R=2-naphthyl
27 R=3,5- dimethoxy-
phenyl
0.28ꢁ0.02
0.97ꢁ0.47
0.39ꢁ0.06
480ꢁ54
0.19ꢁ0.01
0.055ꢁ0.006
high
reference compounds
28 thiodigalactoside
29 Me b-LacNAc[30]
30 Me b-Lac[30]
9.5ꢁ2.2
21ꢁ5.0
38ꢁ6.8
18ꢁ3.3
26ꢁ4.3
54ꢁ11
186ꢁ51
1400ꢁ600
146ꢁ24
minal, which was observed with intact galectin-3.[37] In gen-
eral, the affinity constants obtained by microcalorimetry cor-
respond well to those obtained in the fluorescence polariza-
tion assay. Increases in enthalpic contributions by the aro-
matic amide moieties were responsible for the affinity am-
plifications observed. Partial entropy compensations
occurred for the tighter-binding diamides 5b, 5d, and 5e,
which suggests that desolvation effects are significant.
Interestingly, stepwise addition of amides in the 2-naph-
thoyl (28, 10b, 5b) and 3,5-dimethoxybenzoyl (28, 10c, 5c)
series gave an increase in DH for each added amide, which
can be interpreted as a stepwise addition of arginine–arene
interactions between galectin and ligand. In the unsubstitut-
ed benzamide series (28, 10a, 5a), however, there is an in-
crease in DH between 28 and 10a, but a decrease upon the
addition of the second benzamide to form 5a. It is possible
that the benzamide moiety is too small to be able to estab-
[a] R is defined in Scheme 5. [b] Not determined.
which suggests that no major conformational rearrange-
ments occur upon Arg–Ser mutation.[38] Furthermore, the
stability of the mutants, as determined by differential scan-
ning calorimetry, was increased by the presence of lactose to
the same extent as the wild type.
The conclusion that no major conformational changes
occur upon R144S mutation is supported by the observa-
tions that affinities of the compounds (28–30) not interacting
with Arg144 are barely affected. The effect of mutating
Arg144 to a serine had surprisingly small effect on the affin-
ity for compounds interacting with this residue. However, a
closer lookat a complex between a diamido-thiodigalacto-
side and galectin-3 (Figure 4a) shows that aromatic amides
4240
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 4233 – 4245

Galectin–Ligand Interactions
FULL PAPER
with a favorable enthalpy may be counterbalanced by an en-
tropic penalty in which Arg144 conformational freedom be-
comes more restricted.
In contrast, the R186S mutant shows greatly reduced af-
finity for all ligands. Again, a closer lookat the galectin-3
structure provides a possible explanation. The Arg186 side
chain forms ion pairs with Glu165 and Glu184 and is situat-
ed between one aromatic amide of 5a–e and the protein
(Figure 4a). Replacing Arg186 with the smaller serine
leaves a void space to be filled and, as a result, the ion pair
with Glu165 is disrupted (Figure 4c). This leads to a much
poorer surface complementarity between the ligand aromat-
ic moiety and the protein and significantly altered solvation.
In addition, the charge of Glu165 is not neutralized by an
arginine side chain and is located close to and possibly
repels the p system of the aromatic moiety. Affinities for the
unsubstituted references 28–30 are also greatly reduced, in-
dicating that this mutation also affects interactions with the
carbohydrate core structure. This is not surprising as one hy-
droxy group of the disaccharides 28–30 (one galactose HO2
in 28, GlcNAc HO3 in 29, and Glc HO3 in 30) bridges the
Arg162–Glu185 ion pair that is part of the planar ion-pair
network[30] created by Arg162, Glu165, Arg186, and Glu185.
However, a major conformational change in R186S com-
pared to the wild type is unlikely as the R186S mutant, in
addition to having the same melting temperature in differ-
ential-scanning calorimetry experiments, display the same
recognition pattern towards natural ligands that are extend-
ed at galactose O3 with Arg144-interacting mono- or oligo-
saccharide structures.[38]
Conclusions
In conclusion, we have synthesized C2-symmetrical and un-
symmetrical thiodigalactoside derivatives bearing one or
two aromatic amide substituents at the 3-positions. The dia-
mides 5a–e and 16,17 showed greatly enhanced affinity for
galectin-3 over the parent thiodigalactoside 28, as measured
by fluorescence polarization and isothermal microcalorime-
try, whereas the monosubstituted compounds 10a–c showed
intermediate affinity for galectin-3. This result can be ex-
plained as being due to a double arene–arginine interaction,
that is, interaction of the two aromatic amide moieties with
Arg144 and Arg186. Mutant studies and computer modeling
have confirmed that this is likely to be the case. Moreover,
fluorescence polarization assays showed that binding to ga-
lectins-7 and -9N, both of which also have arginine residues
in the relevant positions, was also increased by the derivati-
zation, albeit to a lesser extent than for galectin-3 and with
different compounds showing the highest affinities. Galec-
tin-1 showed only modestly improved affinity for the di-
Figure 4. Modeled low-energy structures of compound 5a in complex
with a) wild type, b) R144S mutant, and c) R186S mutant galectin-3. A
void volume between Glu165 and Glu184 and below one benzamide is
indicated with an arrow in (c).
interacting with Arg144 are placed in between the Arg144
guanidinium ion and the protein surface. Hence, removal of
the Arg144 side chain results in a complex in which the aro-
matic moiety of the ligand retains good surface complemen-
tarity with the protein surface (Figure 4b). Hence, the affini-
ty enhancement achieved from interactions between the
ligand aromatic amides and the free and relatively flexible
Arg144 appears to be driven not only by the Arg144–arene
interaction itself, but also to a large extent by desolvation
effects of the Arg144 side chain and favorable van der
Waals contact between the aromatic amide and the protein
surface. Furthermore, the limited contribution to the free
energy of a direct Arg144 guanidinium–arene interaction
A
nine residue close to galactose C3 of the bound ligands (cor-
responding to, for example, Arg144 in galectin-3). Galectin-
8N, on the other hand, is missing the arginine corresponding
to Arg186 and, in this case, derivatization with aromatic
Chem. Eur. J. 2008, 14, 4233 – 4245
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4241

U. J. Nilsson et al.
7.88 ppm (at, J=7.1 Hz, 4H; Ar-H); 13C NMR (100.6 MHz, CD3OD): d=
58.8 (d, C3), 62.9 (t, C6), 69.2, 69.4 (2”d, C2, C4), 81.7 (d, C5), 86.4 (d,
C1), 128.6, 129.5, 132.7 (3”d, Ar-CH), 135.9 (s, Ar-C), 170.6 ppm (s, C=
O); FAB+ (m/z): 587 ([M+Na]+, 100%); HRMS (m/z): calcd for
C26H32O10N2SNa [MNa+] 587.1675; found 587.1676.
amides resulted in dramatic loss of affinity, or, in one case, a
marginal gain.
The thiodigalactoside diamides were shown herein to be
the best carbohydrate-based galectin-3, -7, and -9N inhibi-
tors that have been reported to date and clearly demon-
strate the success of our strategy of targeting arginine resi-
dues with aromatic substituents to increase ligand affinity
for proteins. The galectin inhibition efficiency of the thiodi-
galactoside diamides makes them promising leads for the
development of galectin-targeting drugs. Indeed, the pro-
tein-binding affinity demonstrated herein was recently con-
ferred in a clinically relevant fibrosis model,[39] thus further
suggesting the potential of the compounds as galectin-target-
ing leads. Finally, the studies significantly contribute to our
knowledge about fundamental aspects of arginine–arene in-
teractions and provide a basis for successfully taking advant-
age of the relatively poorly exploited arginine–arene inter-
actions in drug development.
Bis[3-deoxy-3-(2-naphthamido)-b-d-galactopyranosyl]sulfane
(5b):
1H NMR (400 MHz, CDCl3:CD3OD, 1:1): d=3.67–3.76 (m, 4H; 5-H, 6-
H), 3.83 (dd, J5,6’ =6.8 Hz, J6,6’ =1.2 Hz, 2H; 6’-H), 3.93 (at, J=10.0 Hz,
2H; 2-H), 4.08 (d, J3,4 =3.0 Hz, 2H; 4-H), 4.22 (dd, J2,3 =10.3 Hz, 2H; 3-
H), 4.78 (d, J1,2 =9.7 Hz, 2H; 1-H), 7.51–7.56 (m, 4H; Ar-H), 7.86–7.95
(m, 8H; Ar-H), 8.41 ppm (s, 2H; Ar-H); FAB+ (m/z): 687 ([M+Na]+,
100%); HRMS (m/z): calcd for C34H36O10N2SNa [MNa+] 687.1988;
found 687.1987.
Bis[3-deoxy-3-(3-methoxybenzamido)-b-d-galactopyranosyl]sulfane (5d):
1H NMR (400 MHz, [D6]DMSO): d=3.46–3.63 (br m, 12H; OH-2, OH-
4, OH-6, H-5, H-6, H-6’), 3.70 (at, J=10.0 Hz, 2H; H-2), 3.81 (s, 6H;
OCH3), 3.86 (d, J3,4 =2.9 Hz, 2H; 4-H), 3.95 (m, 2H; 3-H), 4.72 (d, J1,2
=
9.7 Hz, 2H; 1-H), 7.08 (dd, J=2.2 Hz, J=8.1 Hz, 2H; Ar-H), 7.37 (at,
J=7.9 Hz, 2H; Ar-H), 7.46 (d, J=2.2 Hz, 2H; Ar-H), 7.48 (d, J=7.8 Hz,
2H; Ar-H), 8.07 ppm (d, JNH,3 =7.9 Hz, 2H; NH); FAB+ (m/z): 647
([M+Na]+, 100), 625 ([M+H]+, 12%); HRMS (m/z): calcd for
C28H36O12N2SNa [MNa+] 647.1887; found 647.1888.
Bis[3-deoxy-3-(4-nitrobenzamido)-b-d-galactopyranosyl]sulfane
(5e):
1H NMR (400 MHz, [D6]DMSO): d=3.51–3.52 (m, 6H; 5-H, 6-H, 6’-H),
3.72 (atd, J=10.0 Hz, JOH,2 =6.1 Hz, 2H; 2-H), 3.89 (m, 2H; 4-H), 3.97
(ddd, J2,3 =10.5 Hz, J3,4 =3.0 Hz, JNH,3 =7.9 Hz, 2H; 3-H), 4.61 (at, J=
5.5 Hz, 2H; 6-OH), 4.73 (d, J1,2 =9.8 Hz, 2H; 1-H), 4.96–4.98 (m, 4H; 2-
OH, 4-OH), 8.15, 8.32 (2”d, J=8.9 Hz, 8H; Ar-H), 8.56 ppm (d, 2H;
NH); FAB+ (m/z): 677 ([M+Na]+, 100%); HRMS (m/z): calcd for
C26H30O14N4SNa [MNa+] 677.1377; found 677.1367.
Experimental Section
General methods: Proton nuclear magnetic resonance (1H) spectra were
recorded on a Bruker DRX 400 (400 MHz) or a Bruker ARX 300
(300 MHz) spectrometer; multiplicities are quoted as singlet (s), doublet
(d), doublet of doublets (dd), triplet (t), apparent triplet (at) or apparent
triplet of doublets (atd). Carbon nuclear magnetic resonance (13C) spec-
tra were recorded on a Bruker DRX 400 (100.6 MHz) spectrometer.
Spectra were assigned using COSY, HMQC, and DEPT experiments. All
chemical shifts are quoted on the d scale in parts per million (ppm).
Low- and high-resolution (HRMS) fast atom bombardment mass spectra
were recorded by using a JEOL SX-120 instrument. Optical rotations
were measured on a Perkin–Elmer 341 polarimeter with a path length of
1 dm; concentrations are given in grams per 100 mL. TLC was carried
2,3,4,6-Tetra-O-acetyl-b-d-galactopyranosyl 2,4,6-tri-O-acetyl-3-azido-3-
deoxy-1-thio-b-d-galactopyranoside (8): 2,3,4,6-Tetra-O-acetyl-1-thio-b-d-
galactopyranose (23 mg, 0.063 mmol) and 2,4,6-tri-O-acetyl-3-azido-3-
deoxy-b-d-galactopyranosyl bromide[29] (25 mg, 0.063 mmol) were dis-
solved in acetonitrile (2 mL) and potassium carbonate (18 mg,
0.13 mmol) was added. The reaction mixture was stirred at room temper-
ature under N2 for 2.5 h after which time TLC (4:5 heptane/ethyl ace-
tate) showed the formation of a major product (Rf =0.2) and very little
remaining of either starting material (Rf =0.4 and 0.7). The reaction mix-
ture was diluted with dichloromethane (30 mL) and washed with H2SO4
out on MerckKieselgel sheets that were precoated with 60F
silica.
254
Plates were developed by using 10% sulfuric acid. Flash column chroma-
tography was carried out on silica (Matrex, 60 Š, 35–70 mm, Grace
Amicon). Dichloromethane was distilled from calcium hydride immedi-
ately before use. Acetonitrile was distilled from calcium hydride and
stored over 4-Š molecular sieves. All other reagents and solvents were
used as supplied. Isothermal microcalorimetry was preformed as de-
scribed.[25] Fluorescence polarization experiments with rat galectin-1 and
human galectin-3, -7, -8N, and -9N were performed as described.[31,32] Kd
values were calculated as averages of 4 to 25 single-point measurements.
The deprotected symmetrical diamides 5a–e were prepared as de-
scribed.[27]
(30 mL of
a 10% aqueous solution). The organic phase was dried
(MgSO4), filtered, and concentrated in vacuo. The residue was purified
by flash column chromatography (1:2 heptane/ethyl acetate) to give 8
(26 mg, 61%) as a colorless oil; [a]2D2 =ꢀ14 (c=0.5 in CHCl3); 1H NMR
(400 MHz, CDCl3): d=1.99, 2.06, 2.06, 2.08, 2.14, 2.17, 2.18 (7”s, 21H;
7”CH3), 3.64 (dd, J2,3 =10.0 Hz, J3,4 =3.3 Hz, 1H; 3b-H), 3.85–3.91 (m,
2H; 5a-H, 5b-H), 4.07–4.14 (m, 3H; 6a-H, 6b-H, 6’b-H), 4.19 (dd, J5,6’
=
6.4 Hz, J6,6’ =11.3 Hz, 1H; 6’a-H), 4.79 (d, J1,2 =10.0 Hz, 1H; 1b-H), 4.80
(d, J1,2 =10.1 Hz, 1H; 1a-H), 5.05 (dd, J2,3 =10.0 Hz, J3,4 =3.4 Hz, 1H; 3a-
H), 5.17 (at, J=10.0 Hz, 1H; 2b-H), 5.22 (at, J=10.0 Hz, 1H; 2a-H), 5.44
(d, 1H; 4a-H), 5.47 ppm (d, 1H; 4b-H); 13C NMR (100.6 MHz, CDCl3):
d=20.7, 20.8, 20.8, 20.9, 20.9, 21.0 (6”q, 7”C(O)CH3), 61.4, 61.7 (2”t,
C6a, C6b), 63.0 (d, C3b), 67.2, 67.4, 67.8, 68.6 (4”d, C2a, C4a, C2b, C4b),
72.0 (d, C3a), 74.8, 75.6 (2”d, C5a, C5b), 81.4, 81.5 (2”d, C1a, C1b), 169.6,
169.6, 170.1, 170.2, 170.3, 170.5, 170.5 ppm (7”s, 7”C=O); FAB+ (m/z):
700 ([M+Na]+, 100%); HRMS (m/z): calcd for C26H35O16N3SNa [MNa+]
700.1636; found 700.1635.
Bis[3-deoxy-3-(3,5-dimethoxybenzamido)-b-d-galactopyranosyl]sulfane
(5c): 1H NMR (400 MHz, [D6]DMSO): d=3.48–3.51 (m, 6H; H-5, H-6,
6’-H), 3.69 (atd, J=10.0, JOH,2 =6.3 Hz, 2H; 2-H), 3.79 (s, 12H; 2”
OCH3), 3.85 (d, J3,4 =2.9, JOH,4 =5.5 Hz, 2H; 4-H), 3.94 (ddd, J2,3 =10.4,
J
NH,3 =7.9 Hz, 2H; 3-H), 4.60 (at, J=5.6 Hz, 2H; 6-OH), 4.72 (d, J1,2 =
9.6 Hz, 2H; 1-H), 4.87 (d, 2H; 4-OH), 4.93 (d, 2H; 2-OH), 6.63 (t, J=
2.2 Hz, 2H; Ar-H), 7.07 (d, 2H; Ar-H), 8.07 ppm (d, 2H; NH); FAB+
(m/z): 707 ([M+Na]+, 100%); HRMS (m/z): calcd for C30H40O14N2SNa
[MNa+] 707.2098; found 707.2095.
Bis[3-deoxy-3-(benzamido)-b-d-galactopyranosyl]sulfane (5a): 1H NMR
(300 MHz, D2O): d=3.70–3.91 (m, 8H; 2-H, 5-H, 6-H, 6’-H), 4.07 (d,
Typical procedure for the synthesis of 9a–c: Compound
8 (5 mg,
7.4 mmol) was dissolved in ethanol (2 mL) and palladium (10% on
carbon, 5 mg) and HCl (0.4 mL of a 1m solution in diethyl ether) were
added. The mixture was degassed and stirred at room temperature under
a hydrogen atmosphere. After 20 min, the mixture was filtered through
Celite and concentrated in vacuo. The residue was dissolved in dichloro-
methane (2 mL) and pyridine (0.5 mL). The acid chloride (24 mmol) was
added and the mixture was stirred at room temperature. After 2 h and
40 min, TLC (1:2 heptane/ethyl acetate) showed the presence of a single
carbohydrate product (Rf =0.3). The reaction mixture was diluted with
dichloromethane (20 mL) and washed with H2SO4 (20 mL of a 10%
J
3,4 =3.0 Hz, 2H; 4-H), 4.23 (dd, J2,3 =10.3 Hz, 2H; 3-H), 4.99 (d, J1,2 =
9.9 Hz, 2H; 1-H), 7.51 (at, J=7.4 Hz, 4H; Ar-H), 7.61 (at, J=7.4 Hz,
2H; Ar-H), 7.78 ppm (at, J=7.2 Hz, 4H; Ar-H); 1H NMR (400 MHz,
CD3OD): d=3.67–3.74 (m, 4H; 5-H, 6-H), 3.80 (dd, J5,6’ =6.6, J6,6’
=
10.5 Hz, 2H; 6’-H), 3.89 (at, J=10.0 Hz, 2H; 2-H), 4.03 (d, J3,4 =2.9 Hz,
2H; 4-H), 4.17 (dd, J2,3 =10.3 Hz, 2H; 3-H), 4.83 (d, J1,2 =9.7 Hz, 2H; 1-
H), 7.47 (at, J=7.4 Hz, 4H; Ar-H), 7.54 (at, J=7.4 Hz, 2H; Ar-H),
4242
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 4233 – 4245

Galectin–Ligand Interactions
FULL PAPER
aqueous solution) and NaHCO3 (20 mL of a saturated aqueous solution),
dried over MgSO4, filtered, and then concentrated in vacuo. The residue
was purified by flash column chromatography to give the following
amides:
(d, C4a), 70.3 (d, C2a), 74.6 (d, C3a), 79.8, 80.4 (2”d, C5a, C5b), 84.2 (d,
C1a), 85.0 (d, C1b), 127.9, 129.4, 132.9 ppm (3”d, Ar-CH); FAB+ (m/z):
484 ([M+Na]+, 100%); HRMS (m/z): calcd for C19H27O10NSNa [MNa+]
484.1253; found 484.1248.
2,3,4,6-Tetra-O-acetyl-b-d-galactopyranosyl 2,4,6-tri-O-acetyl-3-deoxy-3-
benzamido-1-thio-b-d-galactopyranoside (9a): Yield: 43% (colorless oil);
1H NMR (400 MHz, CDCl3): d=1.99, 2.07, 2.08, 2.08, 2.15, 2.18 (6”s,
21H; 7”CH3), 3.92 (at, J=6.7 Hz, 1H; 5a-H), 3.99 (at, J=6.4 Hz, 1H;
b-d-Galactopyranosyl 3-deoxy-3-(2-naphthamido)-1-thio-b-d-galactopyra-
noside (10b): Yield: 66% (colorless oil); 1H NMR (400 MHz, D2O): d=
3.61–3.92 (m, 9H; 2a-H, 3a-H, 5a-H, 6a-H, 6’a-H, 2b-H, 5b-H, 6b-H, 6’b-H),
3.99 (d, J3,4 =3.1 Hz, 1H; 4a-H), 4.12 (d, J3,4 =3.0 Hz, 1H; 4b-H), 4.29 (dd,
5b-H), 4.08–4.22 (m, 4H; 6a-H, 6’a-H, 6b-H, 6’b-H), 4.51 (ddd, J2,3
10.7 Hz, J3,4 =3.1 Hz, JNH,3 =7.7 Hz, 1H; 3b-H), 4.84 (d, J1,2 =10.1 Hz, 1H;
1a-H), 4.94 (d, J1,2 =10.1 Hz, 1H; 1b-H), 5.06 (dd, J2,3 =10.0 Hz, J3,4
=
J
2,3 =10.5 Hz, 1H; 3b-H), 4.84 (d(obs), 1H; 1a-H), 4.99 (d, J1,2 =9.8 Hz,
G
1H; 1b-H), 7.63–7.69 (m, 2H; Ar-H), 7.85 (dd, J=1.5 Hz, J=8.5 Hz, 1H;
Ar-H), 8.80–8.07 (m, 3H; Ar-H), 8.39 ppm (s, 1H; Ar-H); 13C NMR
(75 MHz, CD3OD): d=58.9 (d, C3b), 62.9 (t, C6a, C6b), 69.2, 69.4, 70.7,
71.8 (4”d, C2a, C4a, C2b, C4b), 76.3 (d, C3a), 81.0, 81.6 (2”d, C5a, C5b),
85.4, 86.5 (2”d, C1a, C1b), 125.2, 127.8, 128.8, 129.0, 129.2, 130.0 (6”d,
Ar-CH), 133.1, 134.1, 136.3 ppm (3”s, Ar-C); FAB+ (m/z): 534
([M+Na]+, 100%); HRMS (m/z): calcd for C23H29O10NSNa [MNa+]
534.1410; found 534.1428.
=
3.4 Hz, 1H; 3a-H), 5.14 (at, J=10.3 Hz, 1H; 2b-H), 5.28 (at, J=10.0 Hz,
1H; 2a-H), 5.45 (d, 1H; 4a-H), 5.58 (d, 1H; 4b-H), 6.47 (d, 1H; NH), 7.42
(at, J=7.5 Hz, 2H; Ar-H), 7.52 (t, J 7.4 Hz, 1H; Ar-H), 7.65 ppm (d, J=
7.2 Hz, 2H; Ar-H); 13C NMR (100.6 MHz, CDCl3): d=20.7, 20.9, 20.9,
20.9, 21.0, 21.1 (6”q, 7”C(O)CH3), 54.3 (d, C3b), 61.7, 62.0 (2”t, C6a,
C6b), 67.4, 67.7 (2”d, C2a, C4a), 68.7, 68.9 (2”d, C2b, C4b), 72.0 (d, C3a),
75.1 (d, C5a), 76.1 (d, C5b), 81.4 (d, C1a, C1b), 127.0, 129.0, 132.2 (3”d,
Ar-CH), 133.4 (s, Ar-C), 167.4, 169.7, 169.9, 170.2, 170.3, 170.6, 172.0 ppm
(7”s, C=O); FAB+ (m/z): 778 (M+Na+, 100%); HRMS FAB+ (m/z):
calcd for C33H41O17NSNa [MNa+] 778.1993; found 778.1987.
b-d-Galactopyranosyl 3-deoxy-3-(3,5-dimethoxybenzamido)-1-thio-b-d-
galactopyranoside (10c): Yield: 60% (colorless oil); 1H NMR (300 MHz,
CD3OD): d=3.49 (dd, J=3.2 Hz, J=9.2 Hz; 1H), 3.56–3.88 (m, 9H),
3.82 (s, 6H; 2”OCH3), 4.00 (d, J3,4 =2.9 Hz, 1H; 4a-H), 4.13 (dd, J2,3
10.3 Hz, J3,4 =3.0 Hz, 1H; 3b-H), 4.67 (d, J1,2 =9.7 Hz, 1H; 1a-H), 4.80 (d,
1,2 =9.8 Hz, 1H; 1b-H), 6.64 (t, J=2.2 Hz, 1H; Ar-H), 7.03 ppm (d, 2H;
=
2,3,4,6-Tetra-O-acetyl-b-d-galactopyranosyl 2,4,6-tri-O-acetyl-3-deoxy-3-
(2-naphthamido)-1-thio-b-d-galactopyranoside (9b): Yield: 40% (color-
less oil); 1H NMR (400 MHz, CDCl3): d=1.99, 2.08, 2.08, 2.09, 2.16, 2.19
(6”s, 21H; 7”CH3), 3.94 (at, J=6.5 Hz, 1H; 5a-H), 4.02 (at, J=6.4 Hz,
J
Ar-H); 13C NMR (75 MHz, CD3OD): d=56.0 (q, 2”OCH3), 58.8 (d,
C3b), 62.9 (t, C6a, C6b), 69.2, 69.3, 70.7, 71.7 (4”d, C2a, C4a, C2b, C4b),
76.3 (d, C3a), 81.0, 81.6 (2”d, C5a, C5b), 85.3, 86.4 (2”d, C1a, C1b), 104.5,
106.5 (2”d, Ar-CH), 137.9, 162.4 ppm (2”s, Ar-C); FAB+ (m/z): 544
([M+Na]+, 100%); HRMS (m/z): calcd for C21H31O12NSNa [MNa+]
544.1465; found 544.1463.
1H; 5b-H), 4.09–4.24 (m, 4H; 6a-H, 6’a-H, 6b-H, 6’b-H), 4.57 (ddd, J2,3
10.7 Hz, J3,4 =3.1 Hz, JNH,3 =7.6 Hz, 1H; 3b-H), 4.85 (d, J1,2 =10.1 Hz, 1H;
1a-H), 4.97 (d, J1,2 =10.1 Hz, 1H; 1b-H), 5.07 (dd, J2,3 =9.9 Hz, J3,4
=
=
3.4 Hz, 1H; 3a-H), 5.19 (at, J=10.2 Hz, 1H; 2b-H), 5.30 (at, J=10.0 Hz,
1H; 2a-H), 5.46 (d, 1H; 4a-H), 5.63 (d, 1H; 4b-H), 6.62 (d, 1H; NH),
7.53–7.60 (m, 2H; Ar-H), 7.70 (dd, J=1.7 Hz, J=8.6 Hz, 1H; Ar-H),
7.86–7.94 (m, 3H; Ar-H), 8.20 ppm (s, 1H; Ar-H); 13C NMR (100.6 MHz,
CDCl3): d=20.7, 20.9, 21.0, 21.0, 21.1 (5”q, 7”CH3), 54.4 (d, C3b), 61.7,
62.0 (2”t, C6a, C6b), 67.4, 67.7 (2”d, C2a, C4a), 68.7, 69.0 (2”d, C2b,
C4b), 72.0 (d, C3a), 75.1 (d, C5a), 76.1 (d, C5b), 81.4, 81.4 (2”d, C1a, C1b),
123.3, 127.1, 127.9, 128.1, 128.9, 129.3 (6”d, 7”Ar-CH), 130.6, 132.7,
135.1 (3”s, 3”Ar-C), 167.5, 169.8, 169.9, 170.2, 170.3, 170.6, 172.1 ppm
(7”s, C=O); FAB+ (m/z): 828 ([M+Na]+, 100%); HRMS (m/z): calcd
for C37H43O17NSNa [MNa+] 828.2149; found 828.2142.
1-(S)-Acetyl-2,4,6-Tri-O-acetyl-3-deoxy-3-(3,5-dimethoxybenzamido)-1-
thio-b-d-galactopyranose (11): Compound 2c (180 mg, 0.35 mmol) was
dissolved in dichloromethane (4 mL) that had been dried over 4-Š mo-
lecular sieves. Acetic anhydride (67 mL, 0.71 mmol) and HBr (0.8 mL of a
33% solution in AcOH) were added and the mixture was stirred under
N2 at room temperature. After 2.5 h, the reaction mixture was diluted
with dichloromethane (50 mL) and poured into ice water (50 mL). The
organic phase was washed with saturated NaHCO3 (50 mL), dried over
MgSO4, filtered, and concentrated in vacuo to give crude 3c. The crude
3c was immediately dissolved in distilled acetonitrile (2 mL). Potassium
thioacetate (40 mg, 0.35 mmol) was added and the mixture was stirred
under N2 at room temperature. After 1 h and 40 min, TLC (1:1 heptane/
ethyl acetate) showed the presence of a major product (Rf =0.2). The re-
action mixture was diluted with dichloromethane (50 mL) and washed
with saturated NaHCO3 (50 mL), dried over MgSO4, filtered, and con-
centrated in vacuo. The residue was purified by flash column chromatog-
raphy (1:1 heptane/ethyl acetate) to give 11 (133 mg, 72%) as a white
solid; [a]2D1 =+78.6 (c=1.0 in CHCl3); 1H NMR (400 MHz, CDCl3): d=
2.03 (s, 6H; 2”OC(O)CH3), 2.12 (s, 3H; OC(O)CH3), 2.39 (s, 3H;
SC(O)CH3), 3.78 (s, 6H; 2”OCH3), 4.01 (m, 1H; 5-H), 4.07–4.15 (m,
2H; 6-H, 6’-H), 4.58 (dd, J2,3 =10.9 Hz, J3,4 =3.2 Hz, JNH,3 =8.1 Hz, 1H; 3-
H), 5.16 (at, J=10.4 Hz, 1H; H-2), 5.39 (d, J1,2 =10.2 Hz, 1H; 1-H), 5.54
(d, 1H; 4-H), 6.38 (d, 1H; NH), 6.53 (t, J=2.2 Hz, 1H; Ar-H), 6.73 ppm
(d, 1H; Ar-H); 13C NMR (100.6 MHz, CDCl3): d=20.8, 20.9 (2”q, 3”
OC(O)CH3), 31.0 (q, SC(O)CH3), 54.0 (d, C3), 61.7 (t, C6), 67.7 (d, C2),
68.8 (d, C4), 76.2 (d, C5), 81.0 (d, C1), 104.1, 104.9 (2”d, Ar-CH), 135.6,
161.0 (2”s, Ar-C), 167.2, 169.9, 170.5, 171.6 (4”s, 3”OC=O, NC=O),
191.9 ppm (s, SC=O); FAB+ (m/z): 550 ([M+Na]+, 80%); HRMS (m/z):
calcd for C23H29O11NSNa [MNa+] 550.1359; found 550.1362.
2,3,4,6-Tetra-O-acetyl-b-d-galactopyranosyl 2,4,6-tri-O-acetyl-3-deoxy-3-
(3,5-dimethoxybenzamido)-1-thio-b-d-galactopyranoside (9c): Yield:
66% (colorless oil); 1H NMR (400 MHz, CDCl3): d=1.99, 2.07, 2.08,
2.08, 2.16, 2.18 (6”s, 21H; 7”C(O)CH3), 3.81 (s, 6H; 2”OCH3), 3.92
(at, J=6.7 Hz, 1H; 5a-H), 3.99 (at, J=6.6 Hz, 1H; 5b-H), 4.08–4.23 (m,
4H; 6a-H, 6’a-H, 6b-H, 6’b-H), 4.48 (ddd, J2,3 =10.7 Hz, J3,4 =3.0 Hz, JNH,3
=
7.7 Hz, 1H; 3b-H), 4.83 (d, J1,2 =10.1 Hz, 1H; 1a-H), 4.93 (d, J1,2 =9.9 Hz,
1H; 1b-H), 5.06 (dd, J2,3 =10.0 Hz, J3,4 =3.4 Hz, 1H; 3a-H), 5.12 (at, J=
10.3 Hz, 1H; 2b-H), 5.28 (at, J=10.0 Hz, 1H; 2a-H), 5.45 (d, 1H; 4a-H),
5.56 (d, 1H; 4b-H), 6.42 (d, 1H; NH), 6.58 (d, J=2.2 Hz, 1H; Ar-H),
6.78 ppm (d, 2H; Ar-H); 13C NMR (100.6 MHz, CDCl3): d=20.7, 20.9,
20.9, 20.9, 21.1 (5”q, 7”C(O)CH3), 54.3 (d, C3b), 55.7 (q, 2”OCH3),
61.7, 62.0 (2”t, C6a, C6b), 67.4, 67.6 (2”d, C2a, C4a), 68.7, 68.9 (2”d,
C2b, C4b), 72.0 (d, C3a), 75.1 (d, C5a), 76.0 (d, C5b), 81.4 (d, C1a, C1b),
104.3, 104.9 (2”d, 7”Ar-CH), 161.1 (s, Ar-C), 169.9 ppm (s, C=O);
FAB + (m/z): 838 ([M+Na]+, 100%); HRMS (m/z): calcd for
C35H45O19NSNa [MNa+] 838.2204; found 838.2200.
Typical procedure for the synthesis of 10a–c:
b-d-Galactopyranosyl 3-deoxy-3-benzamido-1-thio-b-d-galactopyranoside
(10a): Compound 9a (5 mg, 7.0 mmol) was dissolved in methanolic meth-
ylamine (40%, 2 mL). After 15 h, the reaction mixture was diluted with
methanol, concentrated in vacuo, and purified on RP-HPLC (C18 H2O/
2’,4’,6’-Tri-O-acetyl-3’-azido-3’-deoxy-b-d-galactopyranosyl
2,4,6-tri-O-
acetyl-3-(3,5-dimethoxybenzamido)-3-deoxy-1-thio-b-d-galactopyranoside
(13): Compound 11 (131 mg, 0.25 mmol) was dissolved in methanol
(15 mL) and cooled to ꢀ408C. Freshly prepared sodium methoxide
(0.24 mmol in 0.810 mL methanol) was transferred to the sugar solution
and the reaction mixture was stirred at ꢀ408C. After 25 min, TLC (1:1
heptane/ethyl acetate) showed very little remaining starting material
(Rf =0.2) and the presence of a major product (Rf =0.15). The reaction
was quenched by the addition of Duolite C436 and stirred until the pH
value reached 7. The mixture was filtered and concentrated in vacuo to
1
MeCN) to give 10a (2.9 mg, 95%) as a colorless oil; H NMR (400 MHz,
D2O): d=3.61–3.90 (m, 9H; 2a-H, 3a-H, 5a-H, 6a-H, 6’a-H, 2b-H, 5b-H, H-
6b, H-6’b), 3.99 (d, J3,4 =3.2 Hz, 1H; H-4a), 4.08 (d, J3,4 =3.0 Hz, 1H; 4b-
H), 4.23 (dd, J2,3 =10.3 Hz, 1H; 3b-H), 4.84 (m, 1H; 1a-H), 4.97 (d, J1,2
=
9.9 Hz, 1H; 1b-H), 7.53 (at, J=7.6 Hz, 2H; Ar-H), 7.63 (t, J=7.5 Hz,
1H; Ar-H), 7.80 ppm (d, J=7.2 Hz, 2H; Ar-H); 13C NMR (100.6 MHz,
D2O): d=57.4 (d, C3b), 61.9 (t, C6a, C6b), 68.2, 68.3 (2”d, C2b, C4b), 69.5
Chem. Eur. J. 2008, 14, 4233 – 4245
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4243

U. J. Nilsson et al.
give the crude thiol 12. The crude 12 was dissolved in acetonitrile (8 mL)
and the bromide 6 (109 mg, 0.28 mmol) and potassium carbonate (67 mg,
0.48 mmol) were added. The reaction mixture was stirred at room tem-
perature for 3 h and 20 min, after which time TLC analysis (1:2 heptane/
ethyl acetate) showed the formation of a major product (Rf =0.05) and
the absence of the thiol 12 (Rf =0.15) as well as remaining excess bro-
mide 6 (Rf =0.7). The reaction mixture was diluted with dichloromethane
(50 mL) and washed with H2SO4 (50 mL of a 10% aqueous solution).
The organic phase was dried (MgSO4), filtered, and concentrated in
vacuo. The residue was purified by flash column chromatography (2:5
10.3 Hz, J3,4 =2.9 Hz, 1H; 3y-H), 4.02 (dd, J2,3 =10.3 Hz, 1H; 3x-H), 4.72
(d, J1,2 =9.6 Hz, 1H; 1y-H), 4.74 (d, J1,2 =9.6 Hz, 1H; 1x-H), 6.61 (t, J=
2.2 Hz, 1H; dimethoxyphenyl-Ar-H), 7.03 (d, 2H; dimethoxyphenyl-Ar-
H), 7.58–7.61 (m, 2H; naphthyl-H), 7.93–8.01 (m, 4H; naphthyl-H),
8.48 ppm (s, 1H; naphthyl-H); 13C NMR (100.6 MHz, [D6]DMSO with a
drop of D2O): d=56.0 (q, 2”OCH3), 57.7 (d, C3a, C3b), 60.9, 61.1 (2”t,
C6a, C6b), 67.1, 67.2, 67.6, 67.8 (4”d, C2a, C4a, C2b, C4b), 80.1 (d, C5a,
C5b), 84.2, (d, C1a, C1b), 103.7, 105.9 (2”d, dimethoxyphenyl-Ar-CH),
124.9, 127.4, 128.2, 128.3, 129.3 (5”d, naphthyl-CH), 132.2, 132.6, 134.7,
137.1, 160.8 (5”s, Ar-C), 166.9, 167.4 ppm (2”s, 2”C=O); FAB+ (m/z):
697 ([M+Na]+, 100%; HRMS FAB+ (m/z): calcd for C32H38O12N2SNa
[MNa+] 697.2043; found 697.2056.
heptane/ethyl acetate) to give 13 (78 mg, 39%) as
a colorless oil;
1H NMR (400 MHz, CDCl3): d=2.07, 2.08, 2.09, 2.14, 2.16, 2.19 (6”s,
18H; 6”C(O)CH3), 3.65 (dd, J2,3 =10.1 Hz, J3,4 =3.3 Hz, 1H; 3a-H), 3.81
(s, 6H; 2”OCH3), 3.87 (at, J=6.7 Hz, 1H; 5a-H), 3.97 (at, J=6.4 Hz,
Molecular modeling: Molecular modeling was performed with the
MMFFs force field and the GB/SA solvation model for water implement-
ed in MacroModel (version 9.1, Schrçdinger, LLC, New York, 2005). All
backbone torsions were selected for random variation. Starting confor-
mations were built from the published crystal structures of galectin-1, -7,
and -9N in complex with lactose,[33–35] of galectin-3 in complex with a C3’-
amido-derivatised LacNAc-based inhibitor,[25] and a homology model of
galectin-8N in complex with LacNAc.[36] Starting conformations of the
amides were positioned in the two possible orientations. Complexes with
naphthoates were minimized starting from both of the two alternative
planar conformations and arene substituent starting conformations were
systematically varied. The minimizations converged in all cases and the
binding modes and overall structures of the minimized complexes closely
resembled that of the crystal structures used for building starting confor-
mations.
1H; 5b-H), 4.09–4.15 (m, 4H; 6a-H, 6’a-H, 6b-H, 6’b-H), 4.48 (ddd, J2,3
=
10.6 Hz, J3,4 =3.0 Hz, JNH,3 =7.7 Hz, 1H; 3b-H), 4.82 (d, J1,2 =10.0 Hz, 1H;
1a-H), 4.93 (d, J1,2 =10.0 Hz, 1H; 1b-H), 5.10 (at, J=10.2 Hz, 1H; 2b-H),
5.22 (at, J=10.0 Hz, 1H; 2a-H), 5.48 (d, 1H; 4a-H), 5.56 (d, 1H; 4b-H),
6.40 (d, 1H; NH), 6.57 (t, J=2.2 Hz, 1H; Ar-H), 6.77 ppm (d, 2H; Ar-
H); 13C NMR (100.6 MHz, CDCl3): d=20.8, 20.8, 20.9, 21.0 (4”q, 6”
C(O)CH3), 54.1 (d, C3b), 55.6 (q, 2”OCH3), 61.8 (t, C6a, C6b), 63.0 (d,
C3a), 67.9 (d, C4a), 68.6, 68.7, 68.8 (3”d, C2a, C2b, C4b), 75.8, 75.9 (2”d,
C5a, C5b), 81.2, 81.4 (2”d, C1a, C1b), 104.1, 104.9 (2”d, Ar-CH), 135.5,
161.0 (2”s, Ar-C), 167.3, 169.6, 169.9, 170.0, 170.5, 170.5, 171.8 ppm (7”s,
7”C=O); FAB+ (m/z): 821 ([M+Na]+, 100%); HRMS (m/z): calcd for
C33H42O17N4SNa [MNa+] 821.2163; found 821.2173.
3’-Deoxy-3-(2-naphthamido)-b-d-galactopyranosyl 3-deoxy-3-(3,5-dime-
thoxybenzamido)-1-thio-b-d-galactopyranoside (16) and 3’-deoxy-3-(1-
naphthamido)-b-d-galactopyranosyl 3-deoxy-3-(3,5-dimethoxybenzami-
do)-1-thio-b-d-galactopyranoside
(17):
Compound
13
(28 mg,
0.035 mmol) was dissolved in ethanol (4 mL) and palladium (10% on
carbon, 20 mg), acetic acid (10 mL), and HCl (0.8 mL of a 1m solution in
diethyl ether) were added. The mixture was degassed and stirred at room
temperature under a hydrogen atmosphere. After 1 h and 15 min, the
mixture was filtered through Celite and concentrated in vacuo to give the
intermediate amine. Half of the intermediate amine was dissolved in di-
chloromethane (2 mL) and pyridine (0.5 mL). 1-Naphthoyl chloride
(17 mL, 0.12 mmol) was added, and the mixture was stirred at room tem-
perature. After 2 h, ethyl acetate (1 mL) was added and the reaction mix-
ture was concentrated in vacuo to give the crude 15. Methylamine
(1.5 mL of a 40% aqueous solution) was added and the mixture was
stirred further at room temperature. After 12 h, TLC analysis (5:1 chloro-
form/methanol) showed the presence of a single carbohydrate product
(Rf =0.2). The reaction mixture was diluted with methanol (1 mL) and
concentrated in vacuo. The residue was purified by flash column chroma-
tography (5:1 chloroform/methanol) to give 17 (6 mg, 50%) as a white
solid; 1H NMR (400 MHz, [D6]DMSO with a drop of D2O, the sugar resi-
dues are arbitrarily assigned as “x” and “y”): d=3.60–3.74 (m (obs), 8H;
2x-H, 5x-H, 6x-H, 6’x-H, 2y-H, 5y-H, 6y-H, 6’y-H), 3.77 (s, 6H; 2”OCH3)
3.83 (d, J3,4 =3.1 Hz, 1H; 4y-H), 3.94 (dd, J2,3 =10.3 Hz, 1H; 3y-H), 3.96
(d, J3,4 =3.1 Hz, 1H; 4x-H), 4.06 (dd, J2,3 =10.2 Hz, 1H; 3x-H), 4.72 (d,
Acknowledgements
This workwas supported by the Swedish Research Council and the pro-
grams “Glycoconjugates in Biological Systems” and “Chemistry for Life
Sciences” were sponsored by the Swedish Strategic Research Foundation.
[1] S. H. Barondes, V. Castronovo, D. N. W. Cooper, R. D. Cummings,
K. Drickamer, T. Feizi, M. A. Gitt, J. Hirabayashi, C. Hughes, K.
Kasai, H. Leffler, F. T. Liu, R. Lotan, A. M. Mecurio, M. Monsigny,
S. Pillai, F. Poirer, A. Raz, P. W. J. Rigby, J. M. Rini, J. L. Wang, Cell
1994, 76, 597–598.
[2] H. Leffler, S. Carlsson, M. Hedlund, Y. Qian, Glycoconjugate J.
2004, 19, 433–440.
[3] J. Dumic, S. Dabelic, M. Flçgel, Biochim. Biophys. Acta Gen. Subj.
2006, 1760, 616–635.
[4] F. T. Liu, D. K. Hsu, Drug News Perspect. 2007, 20, 455–460.
[5] N. E. Rhaleb, S. Pokharel, P. Harding, S. Rasoul, H. Peng, O. A. Car-
retero, Am. J. Pathol. 2008, 172, 247–255.
[6] N. C. Henderson, A. C. Mackinnon, S. L. Farnworth, T. Kipari, C.
Haslett, J. P. Iredale, F. T. Liu, J. Hughes, T. Sethi, Am. J. Pathol.
2008, 172, 288–298.
[7] F.-T. Liu, G. A. Rabinovich, Nat. Rev. Cancer 2005, 5, 29–41.
[8] C. M. John, H. Leffler, B. Kahl-Knutsson, I. Svensson, G. A. Jarvis,
Clin. Cancer Res. 2003, 9, 2374–2383.
[9] M. A. Toscano, G. A. Bianco, J. M. Ilarregui, D. O. Croci, J. Cor-
reale, J. D. Hernandez, N. W. Zwirner, F. Poirer, E. M. Riley, L.
Baum, G. A. Rabinovich, Nat. Immun. 2007, 825–834.
[10] M. Salatino, D. O. Croci, G. A. Bianco, J. M. Ilarregui, M. A. Tosca-
no, G. A. Rabinovich, Expert Opin. Biol. Ther. 2008, 8, 45–57.
[11] V. L. Thijssen, F. Poirer, L. G. Baum, A. W. Griffioen, Blood 2007,
110, 2819–2827.
J
1,2 =9.6 Hz, 1H; 1y-H), 4.75 (d, J1,2 =9.6 Hz, 1H; 1x-H), 6.62 (t, J=
2.2 Hz, 1H; dimethoxyphenyl-Ar-H), 7.03 (d, 2H; dimethoxyphenyl-Ar-
H), 7.51–7.56 (m, 3H; naphthyl-H), 7.70 (d, J=7.1 Hz, 1H; naphthyl-H),
7.95 (dd, J=3.3 Hz, J=6.4 Hz, 1H; naphthyl-H), 7.99 (d, J=8.3 Hz, 1H;
naphthyl-H), 8.26 ppm (dd, J=3.4 Hz, J=6.2 Hz, 1H; naphthyl-H);
13C NMR (100.6 MHz, [D6]DMSO with a drop of D2O): d=55.8 (q, 2”
OCH3), 57.3, 57.6 (2”d, C3a, C3b), 60.7 (t, C6a, C6b), 66.8, 67.0, 67.5, 67.8
(4”d, C2a, C4a, C2b, C4b), 79.9, 80.0 (2”d, C5a, C5b), 84.1, 84.1, (2”d,
C1a, C1b), 103.5, 105.8 (2”d, dimethoxyphenyl-Ar-CH), 125.3, 125.9,
126.5, 127.0, 128.5, 130.1 (6”d, naphthyl-CH), 133.4, 135.0, 137.1, 160.6
(4”s, Ar-C), 166.6, 169.1 ppm (2”s, 2”C=O); FAB+ (m/z): 697
([M+Na]+, 100%); HRMS (m/z): calcd for C32H38O12N2SNa [MNa+]
697.2043; found 697.2048. The other half of the intermediate amine was
treated similarly by using 2-naphthoyl chloride (22 mg, 0.12 mmol) to
give 16 (8 mg, 67%) as a white solid; 1H NMR (400 MHz, [D6]DMSO
with a drop of D2O, the sugar residues are arbitrarily assigned as “x” and
“y”) 3.66–3.84 (m (obs), 15H; 2x-H, 5x-H, 6x-H, 6’x-H, 2y-H, 4y-H, 5y-H,
[12] I. Camby, M. Le Mercier, F. Lefranc, R. Kiss, Glycobiology 2006, 16,
137R–157R.
[13] H. Lahm, S. Andre, A. Hoeflich, H. Kaltner, H. C. Siebert, B.
Sordat, C.-W. von der Lieth, E. Wolf, H.-J. Gabius, Glycoconjugate
J. 2003, 20, 227–238.
6y-H, 6’y-H, 2”OCH3) 3.88 (d, J3,4 =3.0 Hz, 1H; 4x-H), 3.94 (dd, J2,3
=
[14] S. Saussez, R. Kiss, Cell. Mol. Life Sci. 2006, 63, 686–697.
4244
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 4233 – 4245

Galectin–Ligand Interactions
FULL PAPER
[15] G. A. Rabinovich, F.-T. Liu, A. Anderson, Scand. J. Immunol. 2007,
66, 143–158.
[30] I. Cumpstey, E. Salomonsson, A. Sundin, H. Leffler, U. J. Nilsson,
ChemBioChem 2007, 8, 1389–1398.
[31] I. Cumpstey, S. Carlsson, H. Leffler, U. J. Nilsson, Org. Biomol.
Chem. 2005, 3, 1922–1932.
[16] J. Nieminen, A. Kuno, J. Hirabayashi, S. Sato, J. Biol. Chem. 2006,
282, 1374–1383.
[32] P. Sçrme, B. Kahl-Knutsson, M. Huflejt, U. J. Nilsson, H. Leffler,
Anal. Biochem. 2004, 334, 36–47.
[17] G. A. Rabinovich, M. A. Toscano, S. S. Jackson, G. R. Vasta, Curr.
Opin. Struct. Biol. 2007, 17, 513–520.
[33] M. F. Lꢁpez-Lucendo, D. Solꢂs, S. Andrꢃ, J. Hirabayashi, K.-i. Kasai,
H. Kaltner, H.-J. Gabius, A. Romero, J. Mol. Biol. 2004, 343, 957–
970.
[34] D. D. Leonidas, E. H. Vatzaki, H. Vorum, J. E. Celis, P. Madsen,
K. R. Acharya, Biochemistry 1998, 37, 13930–13940.
[35] M. Nagae, N. Nishi, S. Nakamura-Tsuruta, J. Hirabayashi, S. Watak-
suki, R. Kato, J. Mol. Biol. 2007, 375, 119–135.
[36] S. Carlsson, C. T. ꢄberg, M. C. Carlsson, A. Sundin, U. J. Nilsson, D.
Smith, R. D. Cummings, J. Almkvist, A. Karlsson, H. Leffler, Glyco-
biology 2007, 17, 663–676.
[37] N. Ahmad, H.-J. Gabius, S. Andre, H. Kaltner, S. Sabesan, R. Roy,
B. Liu, F. Macaluso, C. F. Brewer, J. Biol. Chem. 2004, 279, 10841–
10847.
[38] E. Salomonsson, M. C. Carlsson, R. Hendus-Altenburger, C. T.
ꢄberg, R. Nilsson, E. Nordberg-Karlsson, V. Johansson, A. Karls-
son, U. J. Nilsson, H. Leffler, 2008, unpublished work.
[39] A. C. MacKinnon, S. L. Farnworth, N. C. Henderson, P. S. Hodkin-
son, T. Kipari, H. Leffler, U. J. Nilsson, C. Haslett, J. Hughes, T.
Sethi, J. Immunol. 2008, 180, 3650–2658.
[18] M. Demetriou, M. Granovsky, S. Quaggin, J. W. Dennis, Nature
2001, 409, 733–739.
[19] E. A. Partridge, C. Le Roy, G. M. Di Guglielmo, J. Pawling, P.
Cheung, M. Granovsky, I. R. Nabi, J. L. Wrana, J. W. Dennis, Sci-
ence 2004, 306, 120–124.
[20] K. S. Lau, E. A. Partridge, A. Grigorian, C. I. Silvescu, V. N. Rein-
hold, M. Demetriou, J. W. Dennis, Cell 2007, 129, 123–134.
[21] D. Delacour, C. Greb, A. Koch, E. Salomonsson, H. Leffler, A.
Le Bivic, R. Jacob, Traffic 2007, 8, 379–388.
[22] F. Poirer, Biochem. Soc. Symp. 2002, 69, 95–103.
[23] R. J. Pieters, ChemBioChem 2006, 7, 721–728.
[24] P. Sçrme, Y. Qian, P.-G. Nyholm, H. Leffler, U. J. Nilsson, ChemBio-
Chem 2002, 3, 183–189.
[25] P. Sçrme, P. Arnoux, B. Kahl-Knutsson, H. Leffler, J. M. Rini, U. J.
Nilsson, J. Am. Chem. Soc. 2005, 127, 1737–1743.
[26] M. A. Bianchet, H. Ahmed, G. R. Vasta, L. M. Amzel, Proteins
2002, 40, 378–388.
[27] I. Cumpstey, A. Sundin, H. Leffler, U. J. Nilsson, Angew. Chem.
2005, 117, 5240–5242; Angew. Chem. Int. Ed. 2005, 44, 5110–5112.
[28] J. Bogusiak, W. Szeja, Carbohydr. Res. 1985, 141, 165–167.
[29] T. L. Lowary, O. Hindsgaul, Carbohydr. Res. 1994, 251, 33–67.
Received: December 6, 2007
Published online: March 25, 2008
Chem. Eur. J. 2008, 14, 4233 – 4245
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4245