Regioselective 3-O-Substitution of Unprotected Thiodigalactosides: Direct Route to Galectin Inhibitors

Article abstract of DOI:10.1002/chem.202002084

The synthesis of tailored bioactive carbohydrates usually comprises challenging (de)protection steps, which lowers synthetic yields and increases time demands. We present here a regioselective single-step introduction of benzylic substituents at 3-hydroxy groups of β-d-galactopyranosyl-(1→1)-thio-β-d-galactopyranoside (TDG) employing dibutyltin oxide in good yields. These glycomimetics act as inhibitors of galectins—human lectins, which are biomedically attractive targets for therapeutic inhibition in, for example, cancerogenesis. The affinity of the prepared glycomimetics to galectin-1 and galectin-3 was studied in enzyme-linked immunosorbent (ELISA)-type assays and their potential to inhibit galectin binding on the cell surface was shown. We used our original in vivo biotinylated galectin constructs for easy detection by flow cytometry. The results of the biological experiments were compared with data from molecular modeling with both galectins. The present work reveals a facile and elegant synthetic route for the preparation of TDG-derived glycomimetics that exhibit differing selectivity and affinity to galectins depending on the choice of 3-O-substitution.

Full text of DOI:10.1002/chem.202002084

Accepted Article
Accepted Article
Title: Regioselective 3-O-Substitution of Unprotected
Thiodigalactosides: Direct Route to Galectin Inhibitors
Authors: Pavla Bojarová, Tomáš Vašíček, Vojtěch Spiwok, Jakub
Červený, Lucie Petrásková, Ladislav Bumba, David Vrbata,
Helena Pelantová, and Vladimír Křen
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.202002084
Link to VoR: https://doi.org/10.1002/chem.202002084
01/2020

10.1002/chem.202002084
Chemistry - A European Journal
FULL PAPER
Regioselective 3-O-Substitution of Unprotected
Thiodigalactosides: Direct Route to Galectin Inhibitors
Tomáš Vašíček[a], Vojtěch Spiwok[b], Jakub Červený[a], Lucie Petrásková[a], Ladislav Bumba[a], David
Vrbata[a], Helena Pelantová[a], Vladimír Křen[a] and Pavla Bojarová*[a,c]
and their analogues were identified to bind galectins with affinities
in high micromolar range.[6,7] Simple lactose and N-
acetyllactosamine (LacNAc) are typical examples; higher affinities
were accomplished with more complex compounds of poly-
LacNAc type, especially in their multivalent presentation.[8]
However, the potential of natural carbohydrates for targeting
galectins in vivo is questionable since these compounds may be
prone to enzymatic hydrolysis by glycosidases and may be
recognized and bound by a number of other receptors, such as
the asialoglycoprotein receptor, which may considerably lower
their in vivo half-life.[9] An outstanding lead compound for
developing disaccharide-based galectin inhibitors is β-D-
galactopyranosyl-(11)-thio-β-D-galactopyranoside (TDG; 1) - a
polar, hydrolytically stable molecule exhibiting inhibitory potency
in a similar order like lactose or LacNAc with galectins. Nilsson
and coworkers[10] pioneered the approach to high-affinity small-
molecule galectin inhibitors based on TDG. Symmetrical small-
molecule inhibitors carrying 4-aryl-1,2,3-triazolyl (2),[11] 3-
benzamido (3),[12] 4-amido-1,2,3-triazolyl (4),[13] or 3-O-
coumarylmethyl (5)[14] moieties at the 3- and 3´-positions were
shown to reach even nanomolar affinities with enhanced
selectivity towards Gal-1 or Gal-3. Notably, thiodigalactoside 2
has recently been approved by FDA for the treatment of idiopathic
pulmonary fibrosis[15] and 3,3´-O-dicoumarylmethyl substituted
thiodigalactoside 5 was presented as the first potent and selective
inhibitor of Gal-3 in contrast to Gal-1.[14]
Abstract: The synthesis of tailored bioactive carbohydrates usually
comprises challenging (de)protection steps, lowering synthetic yields
and increasing time demands. We present here a regioselective
single-step introduction of benzylic substituents at 3-hydroxy groups
of β-D-galactopyranosyl-(11)-thio-β-D-galactopyranoside (TDG)
employing dibutyltin oxide in good yields. These glycomimetics act as
inhibitors of galectins - human lectins, which are biomedically
attractive targets for therapeutic inhibition in e.g. cancerogenesis. The
affinity of prepared glycomimetics to galectin-1 and galectin-3 was
studied in ELISA-type assay and their potential to inhibit galectin
binding on the cell surface was shown. There we used our original in
vivo biotinylated galectin constructs for easy detection by flow
cytometry. The results of biological experiments were compared with
data from molecular modeling with both galectins. The present work
reveals a facile and elegant synthetic route for the preparation of
TDG-derived glycomimetics that exhibit differing selectivity and
affinity to galectins depending on the choice of 3-O-substitution.
Introduction
A
regioselective introduction of synthetically diverse
moieties onto valuable carbohydrates derivatives still represents
a considerable hurdle in synthetic endeavors. Carbohydrates are
readily available natural compounds frequently used as
precursors of multi-purposeful target molecules with biological
activities. They are often studied in connection with their binding
to lectins – proteins that elicit biological response in vivo upon
recognizing complex carbohydrate structural patterns.[1] β-D-
Galactopyranosides and their derivatives are efficient ligands of
galectins - animal and human lectins with a characteristic
carbohydrate recognition domain (CRD) that play crucial roles in
vital physiological and pathophysiological processes.[2] Galectins,
especially the most abundant galectin-1 (Gal-1) and galectin-3
(Gal-3) are mainly studied for their participation in cancerogenesis,
metastatic processes,[3] inflammation,[4] fibrotization,[5] and related
disorders. Their potential as promising therapeutic targets is not
fully explored yet. A number of naturally occurring carbohydrates
[a]
T. Vašíček, J. Červený, Dr. L. Petrásková, Dr. L. Bumba, D. Vrbata,
Dr. H. Pelantová, prof. V.Křen, Dr. P.Bojarová
Institute of Microbiology, Academy of Sciences of the Czech
Republic, Vídeňská 1083, 14220 Prague 4, Czech Republic
E-mail: bojarova@biomed.cas.cz
[b]
[c]
Dr. V.Spiwok
Institute of Chemical Technology Prague, Technická 3, 16628
Prague 6, Czech Republic
Dr. P. Bojarová
Figure 1 Symmetrically 3,3’-O/N-disubstituted thiodigalactosides as galectin
inhibitors.[11-14]
Department of Health Care Disciplines and Population Protection,
Faculty of Biomedical Engineering, Czech Technical University in
Prague, Nám. Sítná 3105, CZ-272 01 Kladno, Czech Republic
The synthetic routes leading to these high-affinity galectin
inhibitors based on TDG lead structure are, however, not
straightforward. This fact is best exemplified by the substantial
number of required synthetic steps including complex protection-
deprotection strategies. For example, the synthetic route leading
Supporting information for this article is given through a link at the end of
the document.
This article is protected by copyright. All rights reserved.

10.1002/chem.202002084
Chemistry - A European Journal
FULL PAPER
to
symmetrical
3,3´-O-diazidothiogalactosides
employs
substitutions at the 3-O- and 3´-O-positions, via one-step
galactopyranosyl isothiuronium salt
a
as sulfur-bearing
a
stannylidene-mediated benzylations. These compounds can be
nucleophile and glycosyl bromide b as an electrophile (Scheme
1A).[10c] Another approach uses triisopropylsilyl thio-β-D-
galactoside (TIPS-galactoside) c as a masked thiol nucleophile
reacting with glycosyl bromide b as an electrophilic acceptor in
the presence of tetrabutylammonium fluoride (TBAF) (Scheme
1B).[16] Although the latter dimerization is high-yielding,
stereoselective, and runs at room temperature with full
conversions within minutes, nucleophile donor a has to be
prepared in seven steps from the commercially available
phenylthio-4,6-O-benzylidene-β-D-galactoside d in less than 14%
overall yield.
easily purified using a combination of anion exchange
chromatography in Li+ cycle and silica gel chromatography. This
procedure ensures a high compound purity required for biological
experiments.
We demonstrate that these 3,3´-O-disubstituted compounds
act as small-molecule inhibitors of galectins. Due to their
glycomimetic character, these compounds should be resistant to
enzymatic hydrolysis and feature a prolonged half-life in vivo
contrary to natural oligosaccharides. The affinities to Gal-1 and -
3 were determined in ELISA-type assay, revealing considerable
structural affinity relations within the tested set. The most
prospective application of galectin-inhibiting glycomimetics is
inhibition of galectin present in tumor microenvironment, which
should inhibit, e.g., masking of cancer cells against the host
immune system, their migration (related to metastatic formation),
etc. Therefore we decided to demonstrate the potency of the best
compounds to inhibit galectin binding on the surface of a model
cell line. The conclusions of inhibitory assays were supported by
in silico molecular modeling with CRDs of both galectins. The
prepared glycomimetics are model compounds paving the way to
readily available galectin inhibitors, available in a fair yield through
our elegant single-step synthetic procedure. Depending on the
type of aromatic substitution, the compounds exhibited varying
selectivity for Gal-3 compared to Gal-1, which may be further
exploited in a future development of selective galectin inhibitors.
Results and Discussion
Synthesis of 3,3´-O-disubstituted thiodigalactosides
In view of all the drawbacks of current methodologies
towards 3,3´-O-disubstituted thiodigalactosides, we searched the
literature for alternative strategies. We found that glycosides of D-
gluco, D-manno, D-galacto or L-fuco or D-rhamno configuration
were successfully benzylated via a respective stanylidene
intermediate at the C-3 position.[18] The work by Giordano and
Iadonisi showed that not only primary hydroxyls are prone to
regioselective O-benzylation but a slight change in reaction
conditions also delivers regioselective 3-O-substitution of
mannosides and galactosides.[19]
Therefore we concentrated on developing a simple elegant
methodology that would afford functional substituted
thiodigalactosides stemming from TDG as a starting building
block. Our optimized methodology presented herein gives rise to
the symmetrical 3,3´-O-disubstituted thiodigalactosides directly
from the commercially available TDG (1) in a single step, using
stannylidene-mediated regioselective benzylations with readily
available bromides (Scheme 2). The reaction of unprotected 1
with Bu2SnO in the presence of tetra-n-butyl ammonium bromide
(TBAB) and N,N-diisopropylethylamine (DIPEA) afforded the tin
intermediate,[20] which reacted with the respective bromide in one
pot reaction to yield the desired compounds 5-15. The reaction
temperature was kept at 82 °C to ensure proper conversion after
18 hours in most cases.
Scheme 1 Synthetic routes towards 3,3’-N/O-disubstituted thiodigalactosides
as galectin inhibitors. (A), reaction of galactopyranosyl isothiuronium salt a as a
sulfur-bearing nucleophile and glycosyl bromide b as an electrophile;[15] (B), S-
dimerization of triisopropylsilyl thio-β-D-galactoside (TIPS-galactoside) c as a
masked thiol nucleophile and glycosyl bromide b as an electrophilic acceptor
using fluoride;[16] (C), condensation of 3-O-propargyl thiouronium galactoside
donor g and 3-O-propargyl galactosyl bromide acceptor f in an S-glycosylation
reaction, followed by 1,3-dipolar cycloaddition reaction yielding di-
coumarylmethyl TDG 5 or its isosteres.[14]
The
preparation
of
3,3´-O-dicoumarylmethyl
thiodigalactoside (Scheme 1C) and, in principle, of its any 3,3´-O-
disubstituted isosteres, was accomplished by a rather lengthy 5-
step procedure starting from the previously published phenyl 3-O-
propynyl-1-thio-β-D-galactopyranoside e.[17] The acetylation of e
is followed by the formation of sensitive bromide f that affords
galactopyranosyl isothiuronium salt g. Then, in the excess of
triethylamine, 3,3´-O-dipropargyl-β-D-galactopyranosyl sulfane is
obtained as a product of condensation of the thiogalactoside
donor and the galactoside acceptor in an S-glycosylation reaction.
This di-O-propargyl derivative then undergoes a crucial 1,3-
dipolar cycloaddition reaction and, after deacetylation, it yields
variously substituted coumarylmethyl thiodigalactosides derived
from 5 in a yield ranging from 17% to 47%.[14] Besides the
thiodigalactoside diesters, which have been prepared in a 3-step
procedure,[10b] and the coumarylmethyl-substituted isosteres,[14]
other 3-O-substitutions of galactosides were not studied in the
literature.
Herein, we report a straightforward approach towards novel
symmetrical thiodigalactosides carrying various aromatic
This article is protected by copyright. All rights reserved.

10.1002/chem.202002084
Chemistry - A European Journal
FULL PAPER
iodide (TBAI), did not seem to influence the yields. The amount of
dibutyltin oxide (3 eq) and DIPEA (5 eq) used in these reactions
were kept constant. After optimization of the reaction conditions
with the pilot compound, 3,3´-O-dibenzyl substituted
thiodigalactoside 6 (36% isolated yield), we used this synthetic
method to prepare a series of isosteres comprising para-
substituted fluoro- (7), chloro- (8), nitro- (9), cyano- (14),
methyleneazide (10), and tert-butyl (15) substituted 3,3´-O-
dibenzyl TDGs as well as biphenyl (11) and naphthyl substituted
TDGs (12) and (13).
Scheme
2 (A) One-step synthetic route towards 3,3´-O-disubstituted
symmetrical thiodigalactosides 6-15 starting from TDG (1) using dibutyltin oxide
and the respective bromide. The structure of the stannylidene intermediate was
proposed on the basis of the literature.[20] (B), MW-assisted Sn-mediated
regioselective substitution of TDG (1) yielding glycomimetic 5; the respective 3-
(bromomethyl)-2H-1-benzopyran-2-one (coumarylmethyl bromide) 5a used in
the synthesis was prepared in two steps: (i), propanoic anhydride (1.5 eq),
sodium propionate (0.5 eq), triethylamine (0.5 eq), 12 h reflux, yield 76%; (ii),
N-bromosuccinimide (2 eq), 2,2´-azo-bis(2-methylbutyronitrile) (cat.), dioxane,
14 h reflux, yield 60%.
In pilot experiments, we performed tin-mediated substitution
of 1 using benzyl bromide in mild excess (2 eq), but only partial
conversion was detected, hence we increased the loading of
benzyl bromide to a larger excess (4 eq) to reach a higher
conversion of the starting material. To ensure the maximum
conversion, further reactions were carried out in large excess of
the benzylation reagent (8 eq). The selection of the phase transfer
catalyst, namely tetra-n-butyl ammonium bromide (TBAB) or
Figure 2 Symmetrical 3,3´-O-disubstituted thiodigalactosides 6-15.
The respective bromides were commercially available,
except for 4-(azidomethyl)phenyl bromide 10a, which was
prepared according to the literature[21] by monoazidation from
This article is protected by copyright. All rights reserved.

10.1002/chem.202002084
Chemistry - A European Journal
FULL PAPER
xylene. For full synthetic details and structural characterization,
see also the Supporting Information. The yields of obtained TDG-
derivatives (Table 1) are in large comparable, ranging from 16%
to 61% (compound 11). Most of the prepared compounds,
especially those carrying a halogen atom in para-benzylic position,
were rather hygroscopic, turning easily from white fluffy material
into yellowish oil. This could be prevented by storage at low
temperature or under argon atmosphere. Notably, our method
was not applicable to bromides that do not strictly possess
benzylic structure, due to no resonance stabilization effect of a
carbocation being formed, such as 3-(2-bromoethyl)indole, and 1-
bromo-3-phenylpropane.
Importantly, we also aimed to apply our stannylidene-
mediated synthetic strategy to the coumarylmethyl derivative 5
previously synthesized in the literature by a multistep procedure
outlined in Scheme 1.[14] The corresponding bromide required for
the substitution, 3-(bromomethyl)-2H-1-benzopyran-2-one 5a,
was prepared in a two-step reaction sequence (cycloaddition and
radical bromination) from salicylaldehyde in 46% overall yield
(Scheme 2B). In the first reaction attempts we always obtained a
complex mixture that did not show any evidence of a product.
Repeated attempts such as lowering (60 °C) or increasing the
temperature (90 °C) did not lead to any formation of the desired
product. Our hope then turned to microwave-assisted synthesis,
which, eventually, afforded the target 3,3´-O-dicoumarylmethyl
thiodigalactoside 5 in 44% yield; the reaction took only 15 min. In
comparison, the previously published synthetic route towards
compound 5 by Nilsson and coworkers[14] comprised six reaction
steps starting from the commercially available compound e
(Scheme 1C), in an overall yield of 41%. Notably, the present
synthetic procedure allowed us to dramatically shorten the
reaction time compared to the previous method[14] and it might
potentially also be useful for other complex 3,3´-substitutions.
After structural characterization and determination of purity
of prepared glycomimetics 5-15, their affinity to galectins was
determined in a competitive ELISA-type assay as described
further.
1 (TDG, control)
5 (coumarylmethyl)
6 (benzyl)
na [c]
39
36
37
34
36
16
61
30
59
82±16
79±16
1
7.2±2.3 (11)
50±10 (1.6)
92±18 (0.9)
31±10 (3)
13±3 (6)
2.2±0.8 (35)
5.5±1.2 (14)
8.5±2.3 (9)
8.5±3.1 (9)
4.4±1.7 (18)
12±2 (7)
3.2
9.0
10.8
3.6
3.0
3.8
6
7 (fluorobenzyl)
8 (chlorobenzyl)
9 (nitrobenzyl)
10 (azidomethylbenzyl)
11 (biphenylmethyl)
12 (naphthylmethyl)
45±6 (1.8)
500[d] (0.2)
15±4 (6)
82±3 (1)
3.5±1.2 (23)
11±4 (7)
4.2
1.2
13
13±5 (6)
(bromonaphthylmethyl)
14 (cyanobenzyl)
15 (t-butylbenzyl)
46
56
41±14 (2)
6.1±1.8 (13)
5.8 ± 1.9 (13)
6.7
9.4
54 ± 7 (1.5)
[a] Affinity was determined in a competitive ELISA-type assay with recombinant
Gal-1 or Gal-3. rp, relative potency, i.e. IC50 (TDG)/IC50 (glycomimetic) [b] Ratio
of IC50 (Gal-1)/ IC50 (Gal-3) shows how many times the affinity of the particular
glycomimetic is higher for Gal-3 than for Gal-1 under the assay conditions. [c] na,
not applicable. [d] Due to limited solubility, the inhibition curve of 11 was not fully
saturated with Gal-1; therefore only an estimate of IC50 is given.
The results in Table 1 show that the O-benzylic type
substitution of TDG is considerably more potent for inhibition of
Gal-3 than of Gal-1 with the exception of biphenylmethyl 11.
Compound 11, however, had quite a weak affinity to both tested
galectins. The introduction of O-benzylic substitution in TDG
resulted in an increase in affinity to Gal-3 of up to ca 23-times
(naphthylmethyl 12) compared to the unsubstituted TDG. Notably,
quite a strong inhibition was reached with non-substituted
aromatic rings and the introduction of additional moieties in the
para-position of the benzyl ring, irrespective of their size or
electronegativity (fluoro 7, chloro 8, cyano 14, t-butyl 15), did not
bring any significant affinity improvement. A slight positive
influence may be contributed only to the presence of nitro group
(9). Another noteworthy conclusion may be drawn from the results
acquired with coumarylmethyl derivative 5. Though the compound
was reported to be quite selective for Gal-3 compared to Gal-1
(91 nM vs 16 M, respectively[14] in the fluorescence polarization
assay, in our hands, the same compound showed a rather poor
selectivity, with IC50 values in low micromolar range for both Gal-
1 and Gal-3. This finding clearly demonstrates the strong impact
of the type of measurement on the found results. However,
coumarylmethyl derivative 5 still proved to be the strongest Gal-3
inhibitor of the set with IC50 of 2.2 M.
Inhibitory potency of glycomimetics towards Gal-1 and
Gal-3
To assess the inhibitory potential of glycomimetics 5-15 with
Gal-1 and Gal-3, we employed competitive ELISA-type inhibition
assay.[7] In this assay, the respective glycomimetic in solution
competed for galectin binding with standard immobilized
competitor asialofetuin (ASF). ASF is a glycoprotein carrying
three triantennary LacNAc-terminated N-glycans that interact
both with Gal-1 and Gal-3. Each galectin was incubated with
increasing concentrations of the glycomimetic, and the inhibition
of galectin binding to immobilized ASF was quantified
colorimetrically using anti-His antibody conjugated to horseradish
In the case of Gal-1, the best results were reached with
coumarylmethyl (5), nitrobenzyl (9), naphthylmethyl (12), and
peroxidase. Unsubstituted TDG (1) was used as a positive control. bromonaphtylmethyl (13) substitution (7-15 M, ca 6 to 15-times
Since some of the compounds were poorly water soluble, DMSO
co-solvent (2-10% v/v) was used to reach the saturating
concentration in the dose-response inhibition curves. These
concentrations were verified to have no effect on the IC50 values
of both galectins. The respective inhibition constants (IC50) were
calculated by non-linear regression of sigmoidal inhibition curves
(Figure 3) and they are shown in Table 1.
higher affinity compared to unsubstituted TDG). In contrast, the
introduction of benzyl (6) brought only a tiny improvement in
affinity compared to unsubstituted TDG (1.6-times higher affinity).
Interestingly, in the case of Gal-1, the para-benzylic substitution
seemed to have a more notable effect on affinity, roughly ranked
as nitro (best)  chloro ≈ cyano ≈ azidomethyl ≈ benzyl ≈ t-butyl 
fluoro (worst). Apparently, though in the previously reported
triazolyl derivatives like compound 2, fluoro substitution brought a
significant improvement in affinity, in the case of O-benzylic
substitution the opposite effect was found. For both Gal-1 and
Gal-3, the biphenylmethyl substituent 11 rather decreased
inhibitory potency of the respective TDG derivatives, more
considerably in the case of Gal-1 (more than 6-times worse
binding of 11 than of unsubstituted TDG). This information might
Table 1 Inhibitory potency and selectivity of glycomimetics 5-15 for Gal-1 and
Gal-3
IC50 [µM] (rp) [a]
Yield
[%]
Selectivity
for Gal-3 [b]
Compound
Gal-1
Gal-3
This article is protected by copyright. All rights reserved.

10.1002/chem.202002084
Chemistry - A European Journal
FULL PAPER
be used in future studies searching for selectivity for other
galectins than Gal-1 and -3.
Table 2 Inhibitory potency of glycomimetics 6, 9, and 12 towards Gal-1-AVI and
Gal-3-AVI on the surface of HEK 293 cells
Concerning the selectivity of prepared glycomimetics
between Gal-1 and Gal-3, compounds 6 (benzyl), 7 (fluorobenzyl),
and 15 (t-butylbenzyl) were the most selective, showing 9 to 11-
fold increased affinity to Gal-3 compared to Gal-1, followed by 14
(cyanobenzyl). Thus, we may conclude that both affinity and
selectivity in the present small molecule inhibitor series may be
steered by a careful choice of O-benzylic substitution at the C-3
position of TDG.
IC50 [µM] [a]
Selectivity for Gal-3 [b]
Compound
Gal-1
Gal-3
1 (TDG, control)
6 (benzyl)
146 ± 45
137 ± 54
34 ± 14
201 ± 48
68 ± 20
1.7 ± 0.4
0.3 ± 0.1
1.2 ± 0.6
2
81
9 (nitrobenzyl)
12 (naphtylmethyl)
113
167
[a] Affinity was determined in a binding inhibition assay with HEK 296 cells using
externally added recombinant Gal-1-AVI or Gal-3-AVI, with detection by flow
[b]
cytometry.
Ratio of IC50 (Gal-1)/ IC50 (Gal-3) shows how many times the
affinity of the particular glycomimetic is higher for Gal-3 than for Gal-1 under the
assay conditions.
The results of the inhibition assay on the cell surface (Table 2)
show that HEK 293 cells have a higher affinity to Gal-1 than to
Gal-3 as also visible on the unsaturated binding curve of Gal-3-
AVI to the cell surface in contrast to Gal-1-AVI (Figure S35 in the
Supporting Information). This trend, apparently given by the
heterogeneous glycan coating on the cell surface (different from
ASF in the standard ELISA-type assay) also reflected on the
binding of positive control TDG, which inhibited binding of Gal-3
ca twice stronger (i.e., with lower IC50) than of Gal-1 on the cell
surface whereas in the standard ELISA-type assay it inhibited
both galectins comparably (cf Table 1). Nevertheless, all three
tested glycomimetics showed a significantly increased selectivity
for Gal-3 when competing with the HEK 293 cell surface (4 to 8
times more selective than TDG). Other interesting information
may be extracted from the structure-affinity relations within this
mini-series. With Gal-3, we could observe the same trend as in
the standard ASF-based ELISA assay: the presence of O-
benzylic moiety strongly increased the glycomimetic inhibitory
potential compared to TDG. Whereas the IC50 of compounds 6
(benzyl) and 12 (naphtylmethyl) were quite comparable, a
surprisingly strong inhibition was found for the nitrobenzyl-
substituted compound 9, in the nanomolar range (IC50 = 0.3 ± 0.1
M).
As already mentioned, the inhibitory capacity of all tested
compounds was much lower for Gal-1 on the cell surface than for
Gal-3. Compound 9 also showed the strongest inhibition of Gal-1
of all three glycomimetics tested (IC50 = 34 ± 14 M). In contrast,
the other two candidates, 6 and 12, did not exhibit any improved
Gal-1 inhibition compared to positive control TDG. Thus,
compound 9 carrying the nitrobenzyl moiety can be proclaimed
the most promising glycomimetic in this series. It displayed a high
inhibitory potential towards both galectins in two independent
assays, and, moreover, good water solubility, which contributes
to its bioavailability.
Figure
3 Competitive inhibition of Gal-3 binding to ASF by selected
glycomimetics as determined by ELISA-type assay. (A), Dose-response
inhibition curves of binding of (A) Gal-1 or (B) Gal-3 by the best-performing
glycomimetics
(naphtylmethyl), and by TDG standard (1). Due to limited solubility, compound
5 was assayed in 10% DMSO/ buffer and compound 12 in 2% DMSO/ buffer.
5
(coumarylmethyl),
6
(benzyl),
9
(nitrobenzyl), 12
Inhibition of galectin binding on the cell surface
Selected glycomimetics with the highest inhibitory potential in
ELISA-type assays were further studied for inhibition of binding of
external galectin to the cell surface. In a living organism, inhibition
of binding of extracellular galectin to the cancer cell surface is one
molecular mechanism that may activate the signalization
processes at the cellular level. Moreover, by removing the
protecting galectin envelope on the surface, the cancer cells are
exposed to the recognition by the host immune system. We
selected the HEK 293 cell line on the basis of its high capacity to
bind externally added Gal-1 and Gal-3 to the cell surface (Figure
S35). For the competitive assay on the cell surface, we present
here a novel experimental design that uses our original AVI-
tagged galectin constructs, Gal-3-AVI[7] and Gal-1-AVI.
Direct galectin labelling is much more straightforward and
priceworthy than using labelled anti-galectin antibodies. However,
our own pilot attempts to directly fluorescently label Gal-3 led to
the galectin inactivation. Though chemical biotinylation of
galectins is described in the literature,[22] it generally has a low
yield and a bad reproducibility, often resulting in heterogeneous
biotinylation. Our original AVI-tagged galectin constructs were
verified to exhibit the same lectin activity as the original galectin
constructs by ELISA assay. The presence of the N-terminal AVI-
tag (GLNDIFEAQKIEWHE) in the protein construct enables
selective conjugation of a single biotin molecule to the lysine
residue in the tag. The covalent biotinylation selectively occurs
during galectin expression by the action of the co-expressed biotin
ligase enzyme (see also Figure S34 in the Supporting
Information). The well-defined biotin-AVI-tag enables specific
labeling of the galectin protein bound to the cell surface by
streptavidin-phycoerythrin conjugate, suitable for detection by
flow cytometry. The results of the binding inhibition assay on the
surface of HEK 293 cells with Gal-1-AVI and Gal-3-AVI and the
best performing glycomimetics 6 (benzyl), 9 (nitrobenzyl) and 12
(naphtylmethyl) are shown in Table 2 and Figure 4. For the
respective dose-dependent inhibition curves, see Figure 35B in
the Supporting Information. For the measurement of poorly water-
soluble compound 12, 2% DMSO was employed. It was shown to
have no significant effect on the IC50 values as tested with lactose,
N-acetyllactosamine, and TDG (1) standards.
This article is protected by copyright. All rights reserved.

10.1002/chem.202002084
Chemistry - A European Journal
FULL PAPER
O-benzylic substituents; their temporary unbinding, flipping, and
exchanges between A and B binding modes were observed
during the 100 ns simulations (Figure 5). This is also more
figuratively demonstrated in the respective animations (Table S13
in the Supporting Information containing respective URLs). The
evolution of simulations (illustrated as root-mean-square
deviation profiles) is shown in Figure S37 for Gal-1 and in Figures
S38a, b, c for Gal-3 in the Supporting Information. According to
the root-mean-square deviation profiles, the binding mode A
appeared to be more stable in the simulations of substituted
thiodigalactosides (cf Figures S38b and c), which indicates that A
is apparently the preferred binding mode. This is in agreement
with the previously reported experimental structure carrying TDG
with asymmetric substitution (Figure S36 in the Supporting
Information).[23] In contrast, the Glide scores from docking
indicated the binding mode B as preferential for 5, 12, and 13.
This discrepancy may be caused by the fact that docking usually
favors larger contact surface area in the binding pocket, which
rather corresponds to the binding mode B.
Generally, the dynamics was more intensive for benzyl
substituted TDG (6) compared to bicyclic moieties. There,
naphthylmethyl (12) was more stable in the binding site than
bromonaphthylmethyl (13). It can be concluded from the
simulations that 3-O-benzylic moieties interact dynamically with
the protein in various orientations between Arg144 and the 3-O-
benzylic moiety. The stated contribution of the 3-O-benzylic
moiety to binding free energies was expressed as a time-average
of contributions in different orientations.
Figure 4 Overlay of flow cytometry histograms of competitive inhibition of
binding of the in vivo biotinylated Gal-1-AVI or Gal-3-AVI constructs to the HEK
293 cell surface in the presence of various concentrations of the glycomimetics
6, 9, and 12. The binding of galectin constructs was examined by flow cytometry
upon staining the cells with a streptavidin-phycoerythrin conjugate. Molar
concentrations of individual competing glycomimetics are indicated for each
histogram. Non-treated cells are shown in grey. Data represent the results of
one representative experiment out of two independent experiments performed
in duplicates. The respective inhibition curves are shown in the Supporting
Information, Figure S35.
In silico docking and molecular dynamics
In order to rationalize the differences between affinities of
synthesized molecules, we carried out protein-ligand docking
calculations, molecular dynamics simulations and quantum
chemical calculations of interaction energies. Structure of Gal-3
was adopted from the complex with asymmetric TDG derivative
(for its structure see Figure S36, Supporting Information; PDB ID:
5OAX).[23] The structure of Gal-1 was adopted from the complex
with TDG (PDB ID: 3OYW).[24]
Previous crystallographic data[23] revealed that C-3
substituted TDG-derived ligands bind in two different binding
modes differing in the orientation of the side chain of Arg144. In
the first binding mode, one of the 3-O-benzylic moieties binds to
the surface of the guanidinium moiety of Arg144 (denoted as
structure A). In the other binding mode, the 3-O-benzylic moiety
is sandwiched between the guanidinium moiety of Arg144 and the
rest of the protein (denoted as structure B). Because the binding
modes of ligands prepared in this study were not known, we
docked them into structures A and B separately, and we selected
the better of the two binding modes (based on the docking score
and a visual inspection of structures). All ligands 5-15 were
docked in a good agreement with experimental structures of TDG
(1) and its respective derivatives in the parent structures of
5OAX[23] and 3OYW.[24] The results of docking (Glide scores) are
shown in Table S12. The results of docking explain a high affinity
of 5 to Gal-3 due to hydrogen bonds of both coumaryl moieties
with water molecules. The negative scores of 9 (nitrobenzyl) and
12 (naphthylmethyl) accord well with the high affinity of these
compounds to Gal-3.
Molecular dynamics simulations were carried out for
complexes of Gal-3 with TDG (1), and selected glycomimetics,
namely coumarylmethyl (5), benzyl (6), naphthylmethyl (12), and
bromonaphthylmethyl (13). In the case of Gal-1, complexes with
TDG (1) were compared with benzyl (6), and naphthylmethyl (12)
substituted thiodigalactosides. All simulations of complexes of
Gal-3 were done in parallel from the initial binding modes A and
B, except for TDG (1), which does not interact directly with
Arg144. The simulations revealed a strong dynamic nature of 3-
Figure 5 Overlays of snapshots of complexes of selected glycomimetic ligands
with Gal-1 and Gal-3 after 20, 40, 60, 80, and 100 ns of molecular dynamics
simulations. Simulations with Gal-3 started from the binding mode A. Galectin
binding site with selected residues is shown only in a single conformation (after
20 ns of complexes of 6 with Gal-1 or Gal-3).
In molecular dynamics simulations, the interactions of 3-O-
benzylic moieties of glycomimetic ligands with both galectins
appeared quite dynamic. Since the strength of protein-ligand
interactions was characterized by different binding modes (A, and
B in the case of Gal-3), we decided to look into other parameters
to shed more light on the binding mode. For this purpose we
carried out quantum chemical calculation of interaction energies
stabilizing complexes of Gal-3 Arg144 side chain with selected 3-
O-benzylic moieties of the glycomimetic ligands. These energies
represent the strongest possible interaction that these isolated
moieties can theoretically have in the optimum mutual
configuration in vacuum at the temperature of absolute zero. The
quantum chemical calculations reflect the impact of cation-π
interactions that are not fully addressed in the standard
calculations of docking scores (Glide) or molecular dynamics
This article is protected by copyright. All rights reserved.

10.1002/chem.202002084
Chemistry - A European Journal
FULL PAPER
simulations. Arg144 side chain was represented by
methylguanidinium. Selected glycomimetic ligands, namely
coumarylmethyl (5), benzyl (6), naphthylmethyl (12), and
bromonaphthylmethyl (13) substituted TDGs, were represented
as separate 3-O-benzylic moieties (R-O-CH3 where the methyl
group originated from C-3 position of the carbohydrate, with
respective hydrogen atoms added to obtain a model separate
structure). The corresponding galectin atoms were extracted from
the structure of Gal-3 (PDB ID: 5OAX).[23] The mutual distance of
terminal methyl groups of interacting partners was kept constant
(8.06 Å) to mimic the situation in the galectin-ligand complex. The
energy stabilizing the resulting complexes was calculated in
vacuum at the second order Møller–Plesset perturbation theory
and optimized.[25] The resulting interaction energies are depicted
in Figure 6.
inhibitory assay on the surface of model HEK 293 cells, its IC50
was in nanomolar range. In the latter assay we also demonstrated
the usefulness of our original well-defined in vivo biotinylated
galectin constructs for detection by flow cytometry.
In sum, the present work discloses the structure-affinity
relationship within eleven 3,3´-O-disubstituted thiodigalactosides
and it shows the compound potential as inhibitors of Gal-1 and
Gal-3 in two types of binding assays. We believe that the
described method of tin-mediated 3-O-substitution of deprotected
TDG may find broader use in the synthesis of novel galectin
inhibitors with a therapeutic potential.
Experimental Section
Materials
TLC was performed on precoated Merck silica gel DC-Alufolien Kieselgel
60 F254 plates. The spots were visualized by charring with 5% H2SO4 in
EtOH or, alternatively, with vanillin stain dip solution (10 g vanillin, 10 mL
H2SO4 in 200 mL 96% EtOH) and heating. β-D-Galactopyranosyl-(11)-
thio-β-D-galactopyranoside (TDG; 1) was purchased from Carbosynth
(Compton, UK). All benzylic bromides used for the regioselective
substitutions as well as dibutyltin oxide were used as delivered from
Sigma-Aldrich (Prague, CZ). N,N-Diisopropyl ethyl amine (DIPEA) was
purchased from Alfa Aesar, Thermo Fisher Scientific (Heysham, UK).
Propionaldehyde, sodium propionate, salicylaldehyde, and tetra-n-
butylammonium bromide (TBAB) from ACROS Organics. All solvents were
of analytical grade. Acetonitrile and methanol were from Macron (Poland),
dioxane from ACROS Organics, Thermo Fisher Scientific (Geel, Belgium).
If not stated otherwise, all other solvents and chemicals were from VWR
Chemicals (Stříbrná Skalice, CZ).
Figure 6 Quantum chemical interaction energies of 3-O-benzylic moieties of
glycomimetics 5, 6, 12, and 13 or unsubstituted TDG 1 (black) with Arg144
residue (grey). More negative interaction energies indicate stronger binding.
Distance between anchor carbons (8.06 Å) was constrained during the
geometry optimization.
General procedure for the synthesis of 3,3´-O-disubstituted
thiodigalactosides
The quantum chemical interaction energies are not direct
contributions to binding free energy because they do not include
desolvation and entropy penalties. Nevertheless, there was a
visible trend of interaction energies: 1 (weakest binding) < 6 ≈ 13
≈ 12 < 5 (strongest binding). This correlates very well with the
A mixture of the starting substrate β-D-galactopyranosyl-(11)-thio-β-D-
galactopyranoside (1 eq), dibutyltin oxide (3 eq) TBAB (1 eq) was weighed
in a round-bottomed flask, suspended in dry dioxane and kept under argon.
The temperature was slowly raised to 60 °C upon vigorous stirring and
then DIPEA (5 eq) was dropwise added to generate the tin intermediate.
The reaction was kept stirring at 80 °C for 30 min and then and the
respective bromide in excess (8 eq) was added. Reaction was left at 82 °C
until the partially dissolved reaction mixture turned from misty yellowish to
translucent dark orange. When full conversion, mostly after 24 h, was
indicated by TLC analysis, the flask was cooled down and volatiles were
removed by evaporation in vacuo. A crusty orange solid residue was
dissolved in CHCl3/MeOH (90/10), adsorbed to a pad of silica gel,
evaporated to dryness and loaded onto the top of a silica gel column
(CHCl3/MeOH, 98/2 gradient to 85/15). Flash chromatography afforded
pure product that, after evaporation, was dissolved in water and lyophilized.
The purity of products was determined by HPLC. The products were fully
characterized by NMR, and HRMS.
experimental
values.
The
bromo
moiety
in
the
bromonaphtylmethyl derivative 13 seems to influence the binding
rather marginally and the crucial affinity improvement is brought
by the naphtyl planar core.
Conclusions
The present results disclose a facile and good-yielding one-step
derivatization of a commercially available thiodigalactoside,
affording small molecule inhibitors of galectins with up to 60-times
increased affinity especially to Gal-3. The selectivity and affinity
to Gal-1 or Gal-3 can be steered by a careful choice of 3-O-
benzylic substitution. In accordance, the molecular dynamics
simulations insinuated a varying flexibility (and, consequently,
different strength of binding) of the glycomimetics depending on
the size and character of the aromatic substitution.
Though especially the bulkier aromatic substitutions may
cause limited water solubility (e.g., coumarylmethyl-substituted 5),
some of the prepared compounds combine the increased affinity
to galectins with good water solubility. This is especially the case
of compound 9, the nitrobenzyl-substituted thiodigalactoside. In
ELISA-type assay it showed 18-times increased selectivity to Gal-
3 compared to unsubstituted TDG, and, importantly, in the
Modified procedure for the synthesis of 3,3´-O-disubstituted
thiodigalactosides
A suspension of β-D-galactopyranosyl-(11)-thio-β-D-galactopyranoside
(1 eq) and dibutyltin oxide (3 eq) in dry dioxane was kept stirring under
argon. The temperature was slowly raised to 70 °C, then DIPEA (5 eq) was
dropwise added with few pellets of molecular sieves 4A and allowed to
react for 3 h. Then it was freeze-dried overnight, re-dissolved in dioxane
and respective bromide in excess (8 eq) was added. The reaction was left
at 90 °C in microwave reactor (Biotage® Initiator+, Biotage AB, Uppsala,
Sweden) until the reaction reached equilibrium (4-8h). The liquid phase
was removed by decantation and the precipitate was washed with CHCl3
(4-8 mL) four times. Combined organic phases were evaporated to
dryness. The residue was re-dissolved in MeOH and desalted (TBAB) by
acidic ion exchange resin in Li+ cycle prior to column chromatography. A
crusty orange solid residue was dissolved in CH2Cl2/MeOH (90/10),
adsorbed to a pad of silica gel, evaporated to dryness and loaded onto the
top of a silica gel column (CHCl2/MeOH/NH4OH, gradient 98/2/0 to
90/9.5/0.5). Flash chromatography afforded pure product that, after
This article is protected by copyright. All rights reserved.

10.1002/chem.202002084
Chemistry - A European Journal
FULL PAPER
evaporation, was dissolved in MeOH/water (1/3) and lyophilized. The
purity of products was determined by HPLC. The products were fully
characterized by NMR, and HRMS. This procedure was applied for the
synthesis of compounds 5, 12, 13, and 15.
v/v) and analyzed on Chromolith RP-18e column (100  3 mm, Merck,
Germany) equipped with Chromolith RP-18e guard column (5  4.6 mm).
Binary gradient elution was as follows: mobile phase A = 5% acetonitrile,
0.1% formic acid, mobile phase B = 80% acetonitrile, 0.1% formic acid;
gradient: 0-2 min 0% B, 2-7 min 0-90% B, 7-8 min 90% B, 8-11 min 90-
0% B, 11-14 min 0% B to equilibrate the column; flow rate 1.2 mL/min, 25
C; injection volume 1 L.
Compound 11 was dissolved in acetonitrile/ water (1:4, v/v) and
analyzed on Prontosil 125-5-Amino H column (5 m, 250  4.6 mm,
Bischoff chromatography, Germany). Isocratic elution was used: mobile
phase 24% acetonitrile, flow rate 1 mL/min, 28 C and the injection volume
was 1 L.
bis-{3-O-[(2H-1-Benzopyran-2-on-3-yl)methyl]-β-D-galactopyranosyl}
sulfane (5): modified reaction conditions: microwave irradiation 15 min,
140 °C; yield 73 mg (39%); white fluffy powder. NMR data are shown in
Table S1, Figures S1a and S1b; MS data in Figure S12; HPLC analysis in
Figure S23 in the Supporting Information. NMR data for the respective
bromide 5a are in Table 1a, Figures S1c and S1d. []D20 = +7.7 (c 0.181,
DMSO).
Compounds 12-13 were dissolved in acetonitrile/ water (4:1, v/v)
and analyzed on TSKgel Amide-80 column (5 m, 4.6  250 mm, Tosoh
Bioscience, JP) equipped with TSKgel Amide-80 guard column (3.2×15
mm). Binary gradient elution was used: mobile phase A = acetonitrile;
mobile phase B = deionized water. The gradient was as follows: 22 % B
for 0-7 min, 22-31 % B for 7-16 min, 31-22 % B for 16-17 min, and 22 % B
for 17-22 min for column equilibration. The flow rate was 1 mL min−1 at 25
C and the injection volume was 1 L.
bis-(3-O-Benzyl-β-D-galactopyranosyl) sulfane (6): reaction time 20 h;
yield 36 mg (36%); white fluffy powder. NMR data are shown in Table S2,
Figures S2a and S2b; MS data in Figure S13; HPLC analysis in Figure
S24 in the Supporting Information. []D20 = -6.0 (c 0.217, MeOH).
bis-[3-O-(4-Fluorobenzyl)-β-D-galactopyranosyl] sulfane (7): reaction
time 20 h; yield 60 mg (37%); white fluffy powder. NMR data are shown in
Table S3, Figures S3a and S3b; MS data in Figure S14; HPLC analysis in
Figure S25 in the Supporting Information. []D20 = 0 (c 0.218, MeOH).
Optical rotation
bis-[3-O-(4-Chlorobenzyl)-β-D-galactopyranosyl] sulfane (8): reaction
time 20 h; yield 44 mg (34%); white fluffy powder. NMR data are shown in
Table S4, Figures S4a and S4b; MS data in Figure S15; HPLC analysis in
Figure S26 in the Supporting Information. []D20 = +5.3 (c 0.226, MeOH).
Specific rotation of compounds 5-15 was measured with an Autopol IV
polarimeter (Rudolph Research Analytical, USA, 2001). The []D20 values
were recorded in MeOH; the concentration units are g mL-1.
Nuclear Magnetic Resonance (NMR) analysis
bis-[3-O-(4-Nitrobenzyl)-β-D-galactopyranosyl] sulfane (9): reaction
time 28 h; yield 63 mg (36%); white solid. NMR data are shown in Table
S5, Figures S5a and S5b; MS data in Figure S16; HPLC analysis in Figure
S27 in the Supporting Information. []D20 = +9.9 (c 0.262, MeOH).
NMR spectra of glycomimetics 5-15 and of bromides 5a and 10a were
measured by Bruker Avance III 400 MHz spectrometer (1H: 399.87 MHz;
13C: 100.55 MHz) for compounds 5a, 7, 8, 9, 12, 14, and 15, Bruker Avance
III 600 MHz spectrometer (1H: 600.23 MHz; 13C: 150.93 MHz) for
compounds 5, 6, 10, and 10a or by Bruker AVANCE III 700 MHz
spectrometer (1H: 700.13 MHz; 13C: 176.05 MHz) for compounds 11 and
13 in CD3OD (except the compound 5a measured in CDCl3 and 10a in
DMSO-d6) at 303 K. 1H NMR, 13C NMR, gCOSY, gHSQC, and gHMBC
experiments were performed using standard manufacturer’s software.
Spectra were referenced using the residual signal of solvent (CD3OD:
bis-[3-O-(4-Azidomethylbenzyl)-β-D-galactopyranosyl] sulfane (10):
reaction time 24 h; yield 5 mg (16%); white fluffy powder. NMR data are
shown in Table S6, Figures S6a and S6b; MS data in Figure S17; HPLC
analysis in Figure S28 in the Supporting Information. NMR data for the
respective bromide 10a are in Table 6a, Figures S6c and S6d, synthetic
20
procedure is shown in Scheme S1 in the Supporting Information. []D
-5.9 (c 0.085, MeOH).
=
H 3.306, C 49.05 ppm; CDCl3: H 7.263, C 77.01 ppm; DMSO: H 2.499,

C 39.46 ppm).
bis-{3-O-[4-(Biphenylmethyl)]-β-D-galactopyranosyl} sulfane (11):
reaction time 30 h; yield 81 mg (61%); white fluffy crust. NMR data are
shown in Table S7, Figures S7a and S7b; MS data in Figure S18; HPLC
analysis in Figure S29 in the Supporting Information. []D20 = +6.5 (c 0.183,
MeOH).
Mass spectrometry analysis
High resolution mass spectra were measured using LTQ Orbitrap XL
hybrid mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA)
equipped with an electrospray ion source. The mobile phase was
methanol/ water (4:1) flowing at 100 μL/min. The samples were dissolved
in methanol or methanol/ water and injected using a 5-μL loop into the
mobile phase. For the negative ion mode, spray voltage, capillary voltage,
tube lens voltage, and capillary temperature were 5.0 kV, -25 V, -125 V,
and 275 °C, respectively. For the positive ion mode, spray voltage,
capillary voltage, tube lens voltage, and capillary temperature were 5.0 kV,
9 V, 150 V, and 275 °C, respectively. The Orbitrap mass spectra were
recorded at the resolution of 100,000.
Mass spectra of compounds 13 and 15 were measured using
Shimadzu Prominence LC analytical system consisting of Shimadzu LC-
20AD binary HPLC pump, Shimadzu CTO-10AS column oven, Shimadzu
SIL-20ACHT cooling autosampler, Shimadzu CBM-20A system controller,
and Shimadzu SPD-20MA diode array detector; LC-MS-2020 mass
detector with single quadrupole equipped with an electrospray ion
source (Shimadzu, Kyoto, Japan). The mobile phase was acetonitrile
flowing at 300 μL/min. The samples were dissolved in
acetonitrile/ methanol (4:1); injection volume was 1 L. For the negative
and positive ion mode, spray voltage and capillary temperature were 4.5
kV, and 250 C, respectively.
bis-{3-O-[(Naphthalen-2-yl)methyl]-β-D-galactopyranosyl}
sulfane
(12): reaction time 20 h; yield 54 mg (30%); white fluffy powder. NMR data
are shown in Table S8, Figures S8a and S8b; MS data in Figure S19;
HPLC analysis in Figure S30 in the Supporting Information. []D20 = +2.6
(c 0.229, MeOH).
bis-{3-O-[(1-Bromonaphthalen-2-yl)methyl]-β-D-galactopyranosyl}
sulfane (13): reaction time 20 h; yield 123 mg (59%); white fluffy powder.
NMR data are shown in Table S9, Figures S9a and S9b; MS data in Figure
S20; HPLC analysis in Figure S31 in the Supporting Information. []D20 = -
-6.8 (c 0.222, DMSO).
bis-[3-O-(4-Cyanobenzyl)-β-D-galactopyranosyl] sulfane (14): reaction
time 24 h; yield 75 mg (46%); white fluffy powder. NMR data are shown in
Table S10, Figures S10a and S10b; MS data in Figure S21; HPLC analysis
in Figure S32 in the Supporting Information. []D20 = +7.0 (c 0.227, MeOH).
bis-{3-O-[4-(tert-Butyl)benzyl]-β-D-galactopyranosyl} sulfane (15):
reaction time 18 h; yield 100 mg (56%); white fluffy powder. NMR data are
shown in Table S11, Figures S11a and S11b; MS data in Figure S22;
HPLC analysis in Figure S33 in the Supporting Information. []D20 = +4.8
(c 0.251, MeOH).
Galectin production
The recombinant human His6-tagged Gal-1, Gal-3, and Gal-3-AVI protein
constructs were prepared and purified as described previously.[7,26] For the
preparation of Gal-1-AVI protein construct, we genetically introduced the
AVI-tag sequence (GLNDIFEAQKIEWHE) into the plasmid construct
encoding the His6-tagged Gal-1 protein (pETDuet1-Gal-1). A 331 bp DNA
fragment, amplified from the pETDuet1-Gal-1 vector by polymerase chain
HPLC analysis
HPLC analyses were performed on a Shimadzu Prominence LC analytical
system consisting of a Shimadzu DGU-20A3 degasser, Shimadzu LC-
20AD binary HPLC pump, a Shimadzu SIL-20ACHT cooling auto sampler,
a Shimadzu CTO-10AS column oven, a Shimadzu CBM-20A system
controller and a Shimadzu SPD-20MA diode array detector (Shimadzu,
JP). Compounds 5-10 and 14-15 were dissolved in acetonitrile/water (5:95,
reaction
(PCR)
using
a
pair
of
primers
(5‘-
TTTGGATCCGGGTCTGAACGACATCTTCGAGGCTCAGAAAATCGAA
TGGCACGAAGGTGGATCTGGTGGATCTCGCATGGCTTCTGGTCTG-
This article is protected by copyright. All rights reserved.

10.1002/chem.202002084
Chemistry - A European Journal
FULL PAPER
3’ and 5’-CTTGAATTCGTATCCATCTG-3’), was cut with BamHI and
EcoRI restriction enzymes, and cloned into the BamHI/EcoRI-cleaved
pETDuet1-Gal-1 vector. The in vivo biotinylated galectin constructs were
expressed in E. coli BL21 (DE3) cells transformed with the respective
plasmid construct and carrying the expression vector encoding biotin
ligase BirA. The cells were grown in mineral M9 medium supplemented
with glycerol (20 g/L), yeast extract (20 g/L), 150 mg/L ampicillin, and 10
mg/L chloramphenicol to an optical density of 0.6 at 600 nm upon induction
with 1 mM isopropyl 1-thio--D-galactopyranoside (IPTG) and additional
growth for 4 h at 37 °C in the presence of 50 M D-biotin (Sigma). The
bacterial cells were harvested by centrifugation (5,000 g, 20 min, 4 °C),
washed with PBS and disrupted by sonication. The cell lysate was
centrifuged at 20,000 g at 4 °C for 30 min and the supernatant was loaded
on Ni2+-Sepharose column (GE Healthcare). After washing with 20 mM
Na2HPO4/ 500 mM NaCl/ 50 mM imidazole, pH 7.4, the proteins were
eluted with buffer containing 20 mM Na2HPO4 (pH 7.4)/ 500 mM NaCl/ 500
mM imidazole. The purified proteins were dialyzed against 50 mM
stained with a streptavidin-phycoerythrin conjugate (1/250 dilution, 50 L;
eBiosciences, San Diego, CA, USA) for 30 min at 4 °C. After washing, cells
were resuspended in PBS/BSA buffer and analyzed by flow cytometry in
the presence of Hoechst 332581 stain (1 mg/mL). Two independent
experiments were performed in duplicate. Data were analyzed using the
FlowJo software (Tree Star, Ashland, OR, USA) and appropriate gating
was used to exclude cell aggregates and dead cells. Inhibitory capacities of
the glycomimetics were derived from a non-linear fitting of the data plotted
as the mean fluorescence intensity of phycoerythrine-stained cells versus
glycomimetic concentrations. It was shown on the respective standards
that 2% DMSO had no significant influence on the resulting IC50. Under
the conditions of the assay, lactose had IC50 = 350 M with Gal-1, and IC50
= 176 M with Gal-3; N-acetyllactosamine had IC50 = 221 M with Gal-1,
and IC50 = 46 M with Gal-3. The values for positive control TDG (1) are
given in Table 2.
Docking of compounds 5-15 into the binding site of Gal-1 and Gal-3
All docking studies were done using Glide program[27] of Schrödinger
package. Ligands and selected water molecules were retained from crystal
structures. Positions of hydrogen atoms of water molecules were
optimized by “Optimize waters” options. Complexes were then minimized.
Finally, ligand molecules were deleted. These sites were used for protein-
ligand docking by Glide with “Sample ring conformations” option off and
“Rewarding intramolecular hydrogen bonds” option on.
NaH2PO4/ 150 mM NaCl, pH 7.5) supplemented with
2 mM
ethylenediaminetetraacetic acid (EDTA) and stored at -20 °C. Protein
concentrations were determined by Bradford assay using calibration with
bovine serum albumin as a standard. The molecular weight of Gal-1-AVI
and Gal-3-AVI was calculated to be 18,894 and 30,550 Da, respectively,
according to the amino acid sequence. The homogeneity of the purified
galectin constructs was analyzed by SDS-PAGE (Figure S34, Supporting
Information.
Molecular dynamics simulations
All simulations were carried out in Gromacs 5.1.3.[28] Missing side chain of
Met113 was added using Modeller.[29] Protein was modelled using
AMBER99SB-ILDN force field.[30] Simulated systems contained one
monomer of galectin, one hand-docked molecule of a ligand, chloride or
sodium ions to neutralize the charge and 5,000-8,000 TIP3P water
molecules. Force field parameters of ligands were combined manually
from Glycam06j[31] and General AMBER Force Field (GAFF)[32] force fields.
Charges were calculated by AM1-BCC method.[33] Each system was
energy minimized and then equilibrated by a series of simulations with
harmonic restraints applied on nonhydrogen atoms of protein and ligand.
Namely, the system was equilibrated by 10 ps NVT (constant number of
particles, volume and temperature) simulations at 10, 50 and 100 K
followed by NVT, NPT (constant number of particles, pressure and
temperature) and another NVT simulations (100 ps each) and finally by 1
ns NVT simulation. In sum, 100 ns unrestrained production simulations
were carried out for each ligand.
Competitive ELISA assay
The affinity of galectins for glycomimetics 5-15 was determined by
competitive ELISA assay as reported previously.[7] Briefly, the F16
Maxisorp NUNC-Immuno Modules (Thermo Scientific, Hvidovre,
Denmark) were coated overnight with asialofetuin (ASF; Sigma-Aldrich,
Hamburg, Germany; 0.1 μM in PBS, 50 μL, 5 pmol/well). Then they were
blocked with bovine serum albumin (2% w/v) dissolved in phosphate-
buffered saline (PBS) for 1 h at room temperature. Afterwards, a mixture
of the inhibitor compound in varying concentrations together with
recombinant galectin (total volume 50 μL; 4.5 μM final galectin
concentration) in EPBS buffer (50 mM NaH2PO4/ 150 mM NaCl/ 2 mM
ethylendiaminetetraacetic acid, pH 7.5) was added and incubated for 2 h.
Extensive rinsing of wells after every step with PBS containing Tween 20
(0.05% v/v) was performed. Detection of bound galectin was achieved
using mouse monoclonal anti-His6-IgG1 antibody conjugated with
horseradish peroxidase (Roche Diagnostics, Mannheim, Germany) diluted
in PBS (1:1000, 50 μL, 1 h, room temperature). TMB One (Kem-En-Tec,
Taastrup, Denmark) substrate solution (50 μL) was used to initiate the
reaction with IgG-conjugated peroxidase. The reaction was stopped by
adding 3 M hydrochloric acid (50 μL). The binding signal of bound galectin
was quantified colorimetrically at 450 nm using Sunrise absorbance
microplate reader (Tecan Group Ltd, Männedorf. Switzerland). The
obtained data were analyzed by non-linear regression (dose response-
inhibition-variable slope) using GraphPad Prism 7 (GraphPad software,
San Diego, CA, USA). Since compounds 5, 11-13, and 15 exhibited limited
solubility in water, these compounds, along with TDG (1) and N-
acetyllactosamine (Carbosynth, UK) as respective standards, were
subjected to a slightly modified procedure using 2-10% DMSO in EPBS
buffer in the 2 h incubation step with galectins. It was shown on the
respective standards that DMSO in this concentration had no measurable
influence on the affinity to galectins. Under the conditions of the assay,
lactose had IC50 = 345 M with Gal-1, and IC50 = 137 M with Gal-3; N-
acetyllactosamine had IC50 = 78 M with Gal-1, and IC50 = 44 M with Gal-
3.
Quantum chemical calculations
Model systems were subjected to energy minimization at the MP2/6-
31+G(d,p) (models of compounds 1 and 6) or HF/6-31+G(d,p) (models of
compounds 5, 12, and 13) level of theory with constraints applied on the
selected distance. Then, interaction energy was calculated at the MP2/6-
31+G(d,p) level of theory[25] as a difference of the energy of the whole
system and both subsystems. Counterpoise method[34] was used to
address the basis set superposition error. All calculations were done in
Gaussian 09.[35]
Acknowledgements
Support from grant projects LTC18038, and LTC19038 by the
Ministry of Education, Youth and Sports of the Czech Republic is
gratefully acknowledged. L.B. acknowledges support from the
project LM2018133 by the Ministry of Education, Youth and
Sports of the Czech Republic. V.S. acknowledges support from
the projects LM2018140 and LM2015047 by the Ministry of
Education, Youth and Sports of the Czech Republic. We thank
prof. J. Cvačka for measuring mass spectra and the group of Dr.
S. Matějková for determining specific rotations.
Flow cytometry
The human embryonic kidney (HEK) 293 cells (ATCC, Rockville, MD,
USA) were cultivated in the high glucose Dulbecco's Modified Eagle
Medium (DMEM, Sigma, USA) at 37 °C under 5% CO2 atmosphere. For
inhibition essays, aliquots of serially diluted glycomimetics and TDG
(positive control) were mixed with Gal-1-AVI (final concentration 5 g/mL)
or Gal-3-AVI (final concentration 10 g/mL) and incubated on ice for 30
min before mixed with the HEK 293 cells (1×105 in 100 µL final volume) in
PBS/BSA buffer (50 mM NaH2PO4/ 150 mM NaCl/ 1% w/v bovine serum
albumin, pH 7.5) and incubated in 96-well Nunc™ MicroWell™ Plates with
Nunclon™ Delta Surface (Nunc, Roskilde, Denmark) for 60 min on ice. For
compound 12, 2% dimethylsulfoxide (DMSO) was present in the PBS/ BSA
buffer during the incubation step. The cells were washed by centrifugation
(200 g, 3 min, 4 °C) and the surface-bound biotinylated galectins were
Keywords: galectin • glycomimetic • inhibitors • molecular
modeling • thiodigalactoside
This article is protected by copyright. All rights reserved.

10.1002/chem.202002084
Chemistry - A European Journal
FULL PAPER
[31] K. N. Kirschner, A. B. Yongye, S. M. Tschampel, J. González‐Outeiriño, C.
R. Daniels, B. L. Foley, R. J. Woods, J. Comput. Chem. 2008, 29, 622-655.
[32] J. Wang, R. M. Wolf, J. W. Caldwell, P. A. Kollman, D. A. Case, J. Comput.
Chem. 2004, 25, 1157-1174.
References
[1] P. Bojarová, V. Křen, Biomater. Sci. 2016, 4, 1142–1160.
[2] D. Laaf, P. Bojarová, L. Elling, V. Křen, Trends Biotechnol. 2019, 37, 402–
415.
[33] A. Jakalian, D. B. Jack, C. I. Bayly, J. Comput. Chem. 2002, 23, 1623-1641.
[34] S. F. Boys, F. Bernardi, Mol. Phys. 1970, 19, 553-566.
[3] K. Wdowiak, T. Francuz, E. Gallego-Colon, N. Ruiz-Agamez, M. Kubeczko,
I. Grochoła, J. Wojnar, Int. J. Mol. Sci. 2018, 19, 210.
[35] Gaussian 09, Revision A.02, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G.
E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. A.
Petersson, H. Nakatsuji, X. Li, M. Caricato, A. Marenich, J. Bloino, B. G.
Janesko, R. Gomperts, B. Mennucci, H. P. Hratchian, J. V. Ortiz, A. F.
Izmaylov, J. L. Sonnenberg, D. Williams-Young, F. Ding, F. Lipparini, F.
Egidi, J. Goings, B. Peng, A. Petrone, T. Henderson, D. Ranasinghe, V. G.
Zakrzewski, J. Gao, N. Rega, G. Zheng, W. Liang, M. Hada, M. Ehara, K.
Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O.
Kitao, H. Nakai, T. Vreven, K. Throssell, J. A. Montgomery, Jr., J. E. Peralta,
F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N.
Staroverov, T. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A.
Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, J. M. Millam, M.
Klene, C. Adamo, R. Cammi, J. W. Ochterski, R. L. Martin, K. Morokuma,
O. Farkas, J. B. Foresman, and D. J. Fox, Gaussian, Inc., Wallingford CT,
2016.
[4] V. Sundblad, L. G. Morosi, J. R. Geffner, G. A. Rabinovich, Immunol. 2017,
199, 3721-3730.
[5] N. Suthahar, W. C. Meijers, H. H. W. Silljé, J. E. Ho, F.-T. Liu, R. A. de Boer,
Theranostics 2018, 8, 593-609.
[6] J. Hirabayashi, T. Hashidate, Y. Arata, N. Nishi, T. Nakamura, M.
Hirashima. T. Urashima, T. Oka, M. Futai, W. E.G. Muller, F. Yagi, K. Kasai,
Biochim. Biophys. Acta 2002, 1572, 232-254.
[7] L. Bumba, D. Laaf, V. Spiwok, L. Elling, V. Křen, P. Bojarová, Int. J. Mol.
Sci. 2018, 19, 372.
[8] a) D. Laaf, P. Bojarová, H. Pelantová, V. Křen, L. Elling, Bioconjug. Chem.
2017, 28, 2832–2840; b) D. Laaf, H. Steffens, H. Pelantová, P. Bojarová,
V. Křen, L. Elling, Adv. Synth. Catal. 2017, 359, 4015-4024.
[9] A. David, Isr. J. Chem. 2010, 50, 204-219.
[10]a) P. Sörme, P. Arnoux, B. Kahl-Knutsson, H. Leffler, J.M. Rini, U.J. Nilsson,
J. Am. Chem. Soc. 2005, 127, 1737-1743; b) T. Delaine, I. Cumpstey, L.
Ingrassia, M. Le Mercier, P. Okechukwu, H. Leffler. R. Kiss, U.J. Nilsson,
J. Med. Chem. 2008, 51, 8109-8114; c) U. Nilsson, WO2009/139719,
Nov.19, 2009.
[11] T. Delaine, P. Collins, A. MacKinnon, G. Sharma, J. Stegmayr, V.K. Rajput,
S. Mandal, I. Cumstey, A. Larumbe, B.A. Salameh, B. Kahl-Knutsson, H.
van Hattum, M. van Scherpenzeel, R.J. Pieters, T. Sethi, H. Schambye, S.
Oredsson, H. Leffler, H. Blanchard, U.J. Nilsson, ChemBiochem 2016, 17,
1759–1770.
[12] I. Cumpstey, A. Sundin, H. Leffler, U.J. Nilsson, Angew. Chem., Int. Ed.
2005, 44, 5110–5112.
[13] B.A Salameh, I. Cumpstey, A. Sundin, H. Leffler, U.J. Nilsson, Bioorg. Med.
Chem. 2010,18, 5367–5378.
[14] V.K. Rajput, A. MacKinnon, S. Mandal, P. Collins, H. Blanchard, H. Leffler,
T. Sethi, H. Schambye, B. Mukhopadhyay, U.J. Nilsson, J. Med. Chem.
2016, 59, 8141.
[15] T. J. Hsieh, H. Y. Lin, Z. Tu, T. C. Lin, S. C. Wu, Y. Y. Tseng, F. T. Liu, S.
T. Hsu, C. H. Lin, Sci. Rep. 2016, 6, 29457.
[16] S. Mandal, U.J. Nilsson, Org. Biomol. Chem. 2014, 12, 4816–4819.
[17] D. Giguere, R. Patnam, M. Bellefleur, C. St.-Pierre, S. Sato, R. Roy, Chem.
Commun. 2006, 22, 2379-2381.
[18] a) S. David, A. Thieffry, A. Veyrières, J. Chem. Soc., Perkin. Trans.1, 1981,
1796–1801; b) G. Yang, F. Kong, S. Zhou, Carbohydr. Res. 1991, 211,
179–182.
[19] a) M. Giordano, A. Iadonisi, Tetrahedron Lett. 2013, 54, 1550–1552; b) M.
Giordano, A. Iadonisi, J. Org. Chem. 2014, 79, 213–222.
[20] L. Ballell, J.A.F. Joosten, F.A. el Maate, R.M.J Liskamp, R.J. Pieters,
Tetrahedron Lett.., 2004, 45, 6685-6687.
[21] V.K. Zashiri, R. Hunter, P.J. Smith, D. Taylor, R. Summers, K. Kirk, R.E.
Martin, TJ. Egan, Eur. J. Med. Chem., 2011, 46, 1729-1742.
[22] G. N. Wang, S. André, H.-J. Gabius, P. V. Murphy, Org. Biomol. Chem.,
2012, 10, 6893-6907.
[23] K. Peterson, R. Kumar, O. Stenstrom, P. Verma, P. R. Verma, M.
Hakansson, B. Kahl-Knutsson, F. Zetterberg, H. Leffler, M. Akke, D. T.
Logan, U. J. Nilsson, J. Med. Chem. 2018, 61, 1164-1175.
[24] K. A. Stannard, P. M. Collins, K. Ito, E. M. Sullivan, S. A. Scott, E. Gabutero,
I. D. Grice, P. Low, U. J. Nilsson, H. Leffler, H. Blanchard, S. J. Ralph,
Cancer Lett. 2010, 299, 95-110.
[25] C. Møller, M. S. Plesset, Phys. Rev. 1934, 46, 618-622.
[26] M. R. Tavares, M. Bláhová, L. Sedláková, L. Elling, H. Pelantová, R.
Konefał, T. Etrych, V. Křen, P. Bojarová, P. Chytil, Biomacromolecules
2020 in print, doi: 10.1021/acs.biomac.9b01370.
[27] R. A. Friesner, R. B. Murphy, M. P. Repasky, L. L. Frye, J. R. Greenwood,
T. A. Halgren, P. C. Sanschagrin, D. T. Mainz, J. Med. Chem. 2006, 49,
6177-6196.
[28] M. J. Abraham, T. Murtola, R. Schulz, S. Páll, J. C. Smith, B. Hess, E.
Lindahl, SoftwareX 2015, 1-2, 19-25.
[29] A. Šali, T. L. Blundell, J. Mol. Biol. 1993, 234, 779-815.
[30] K. Lindorff-Larsen, S. Piana, K. Palmo, P. Maragakis, J. L. Klepeis, R. O.
Dror, D. E. Shaw, Proteins 2010, 78, 1950-1958.
This article is protected by copyright. All rights reserved.

10.1002/chem.202002084
Chemistry - A European Journal
FULL PAPER
Entry for the Table of Contents
FULL PAPER
Galectin-binding glycomimetics in one
step. Single-step tin-mediated synthesis
affords 3-O-substituted thiodigalactosides
with increased inhibitory potential to human
galectins, especially Gal-3. Through the
choice of aromatic substitution we may steer
Tomáš Vašíček, Vojtěch
Spiwok, Jakub Červený,
Lucie Petrásková, Ladislav
Bumba, Helena Pelantová,
David Vrbata, Vladimír
Křen, Pavla Bojarová*
water
solubility
and
selectivity
of
glycomimetics to galectins. We show the use
of in vivo biotinylated galectin constructs to
test galectin inhibition in cell culture. The
present method may be useful for preparing
therapeutically potent galectin inhibitors.
Page No. – Page No.
Regioselective
Substitution
Unprotected
3-O-
of
Thiodigalactosides:
Direct Route to Galectin
Inhibitors
This article is protected by copyright. All rights reserved.