Synthesis of the Thomsen-Friedenreich-antigen (TF-antigen) and binding of Galectin-3 to TF-antigen presenting neo-glycoproteins

Article abstract of DOI:10.1007/s10719-020-09926-y

The Thomsen-Friedenreich-antigen, Gal(β1–3)GalNAc(α1-O-Ser/Thr (TF-antigen), is presented on the surface of most human cancer cell types. Its interaction with galectin 1 and galectin 3 leads to tumor cell aggregation and promotes cancer metastasis and T-cell apoptosis in epithelial tissue. To further explore multivalent binding between the TF-antigen and galectin-3, the TF-antigen was enzymatically synthesized in high yields with GalNAc(α1-EG3-azide as the acceptor substrate by use of the glycosynthase BgaC/Glu233Gly. Subsequently, it was coupled to alkynyl-functionalized bovine serum albumin via a copper(I)-catalyzed alkyne-azide cycloaddition. This procedure yielded neo-glycoproteins with tunable glycan multivalency for binding studies. Glycan densities between 2 and 53 glycan residues per protein molecule were obtained by regulated alkynyl-modification of the lysine residues of BSA. The number of coupled glycans was quantified by sodium dodecyl sulfate polyacrylamide gel electrophoresis and a trinitrobenzene sulfonic acid assay. The binding efficiency of the neo-glycoproteins with human galectin-3 and the effect of multivalency was investigated and assessed using an enzyme-linked lectin assay. Immobilized neo-glycoproteins of all modification densities showed binding of Gal-3 with increasing glycan density. However, multivalent glycan presentation did not result in a higher binding affinity. In contrast, inhibition of Gal-3 binding to asialofetuin was effective. The relative inhibitory potency was increased by a factor of 142 for neo-glycoproteins displaying 10 glycans/protein in contrast to highly decorated inhibitors with only 2-fold increase. In summary, the functionality of BSA-based neo-glycoproteins presenting the TF-antigen as multivalent inhibitors for Gal-3 was demonstrated.

Full text of DOI:10.1007/s10719-020-09926-y

Glycoconjugate Journal
https://doi.org/10.1007/s10719-020-09926-y
ORIGINAL ARTICLE
Synthesis of the Thomsen-Friedenreich-antigen (TF-antigen)
and binding of Galectin-3 to TF-antigen
presenting neo-glycoproteins
Marius Hoffmann1 & Marc R. Hayes2 & Jörg Pietruszka2,3 & Lothar Elling1
Received: 20 January 2020 /Accepted: 13 April 2020
#
The Author(s) 2020
Abstract
The Thomsen-Friedenreich-antigen, Gal(β1–3)GalNAc(α1-O-Ser/Thr (TF-antigen), is presented on the surface of most human
cancer cell types. Its interaction with galectin 1 and galectin 3 leads to tumor cell aggregation and promotes cancer metastasis and
T-cell apoptosis in epithelial tissue. To further explore multivalent binding between the TF-antigen and galectin-3, the TF-antigen
was enzymatically synthesized in high yields with GalNAc(α1-EG3-azide as the acceptor substrate by use of the glycosynthase
BgaC/Glu233Gly. Subsequently, it was coupled to alkynyl-functionalized bovine serum albumin via a copper(I)-catalyzed
alkyne-azide cycloaddition. This procedure yielded neo-glycoproteins with tunable glycan multivalency for binding studies.
Glycan densities between 2 and 53 glycan residues per protein molecule were obtained by regulated alkynyl-modification of the
lysine residues of BSA. The number of coupled glycans was quantified by sodium dodecyl sulfate polyacrylamide gel electro-
phoresis and a trinitrobenzene sulfonic acid assay. The binding efficiency of the neo-glycoproteins with human galectin-3 and the
effect of multivalency was investigated and assessed using an enzyme-linked lectin assay. Immobilized neo-glycoproteins of all
modification densities showed binding of Gal-3 with increasing glycan density. However, multivalent glycan presentation did not
result in a higher binding affinity. In contrast, inhibition of Gal-3 binding to asialofetuin was effective. The relative inhibitory
potency was increased by a factor of 142 for neo-glycoproteins displaying 10 glycans/protein in contrast to highly decorated
inhibitors with only 2-fold increase. In summary, the functionality of BSA-based neo-glycoproteins presenting the TF-antigen as
multivalent inhibitors for Gal-3 was demonstrated.
.
.
.
.
Keywords Thomsen-Friedenreich-antigen Glycosynthase Neo-glycoproteins Multivalency Galectin-3
Electronic supplementary material The online version of this article
(https://doi.org/10.1007/s10719-020-09926-y) contains supplementary
Introduction
material, which is available to authorized users.
A carbohydrate structure of particular importance in O-glyco-
* Lothar Elling
L.Elling@biotec.rwth-aachen.de
sylation of mammalian glycoproteins is the Thomsen-
Friedenreich-antigen (TF-antigen; Gal(β1–3)GalNAc(α1-O-
Ser/Thr; core 1). Deriving from its precursor GalNAc(α-O-
Ser/Thr (Tn-antigen), the TF-antigen serves as scaffold for
longer and more complex mucin-type O-glycan structures
[1]. High levels of Tn- and TF-antigen occur in 70–90% of
all human carcinomas [1, 2]. They are mostly present on the
cell-surface bound glycoprotein mucin-1 (MUC1) and consid-
ered as “pancarcinoma antigens” [3].
The TF-antigen has been shown to be a potent ligand in both
galectin-1 (Gal-1) and galectin-3 (Gal-3) mediated cell interac-
tions with Gal-3 being the stronger binding partner [4–8]. Gal-3
is expressed on the outer cell membrane of endothelial cells as
well as carcinoma cells and binds to the TF-antigen presented
on MUC1 under static and flow conditions [9]. Circulating Gal-
Marius Hoffmann
m.hoffmann@biotec.rwth-aachen.de
Marc R. Hayes
m.hayes@fz-juelich.de
Jörg Pietruszka
j.pietruszka@fz-juelich.de
1
Laboratory for Biomaterials, Institute for Biotechnology and
Helmholtz-Institute for Biomedical Engineering, RWTH Aachen
University, Pauwelsstraße. 20, 52074 Aachen, Germany
2
3
Institute for Bioorganic Chemistry, Heinrich Heine University
Düsseldorf at Forschungszentrum Jülich, 52426 Jülich, Germany
Forschungszentrum Jülich, IBG-1: Biotechnology,
52426 Jülich, Germany

Glycoconj J
3 was shown to bind to MUC1 on the surface of cancer cells,
thereby polarizing the cell surface so that the exposure of adhe-
sion molecules leads to cellular aggregation and tumor forma-
tion [10]. Thus, the interaction between Gal-3 and the TF-
antigen plays a crucial role in homotypic aggregation of cancer
cells as well as in the initial adhesion and proliferation of car-
cinoma cells in the vascular endothelium [2, 3, 11]. Since these
mechanisms are among the first steps during cancer metastasis
and proliferation, small molecule antagonists with the potential
of preventing the binding of Gal-3 have been developed [12,
13]. In addition, multivalent synthetic ligands have been proven
as potent binders of Gal-3 with KD-values in the low nanomolar
range [14–16]. Especially, multivalent neo-glycoproteins
(NGPs) on the basis of bovine serum albumin (BSA) displaying
the LacdiNAc-LacNAc motif, GalNAc(β1–4)GlcNAc(β1–
3)Gal(β-GlcNAc, or thiodigalactosides were demonstrated as
effective ligands for Gal-3 with sub-nanomolar KD-values
[17–19]. The presentation of a single TF-antigen ligand on
synthetic glycopeptides was favorable to gain affinity for Gal-
3 binding when compared to a non-conjugated TF-antigen and
glycomimetics [20, 21]. Moreover, multivalent mucin-like pre-
sentation of the TF-antigen triggers Gal-3 binding as demon-
strated for a TF-antigen presenting anti-freeze protein from
codfish, which suppresses prostate cancer metastasis and T-
cell apoptosis in mice [22].
nitrophenyl β-D-galactopyranoside (pNP-Gal, 2) was pur-
chased from Carbosynth (Berkshire, United Kingdom). All
other chemicals, if not specifically mentioned, were obtained
from Carl Roth (Karlsruhe, Germany).
Synthesis of α-GalF (4)
The synthesis of the glycosyl donor α-D-glucopyranosyl fluo-
ride (α-GalF; 4) was carried out as previously described by
Henze et al. [24]. The peracetylated galactose (5 mmol, 2 g)
was dissolved at 0 °C in HF/pyridine solution (1 mL/mmol,
70% HF in pyridine). The reaction was stirred and monitored
by TLC. After completion, the reaction was diluted with
CH2Cl2 and dH2O. Excess fluoric acid was neutralized using
K2CO3 and the organic phase separated from the aqueous.
Pyridine was removed by extraction with saturated CuSO4-so-
lution and the organic phase dried over MgSO4 and filtered.
The solvent was removed under reduced pressure and the prod-
uct mixture purified by flash chromatography (45% α-anomer,
2.3 mmol, 0.8 g). Deacetylation occurred with 2 M ammonia in
methanol at 0 °C. After complete conversion the solvent was
evaporated under reduced pressure. α-GalF 4 was obtained
without further purification as colorless solid (quant., 0.4 g).
Expression and purification of recombinant enzymes
and galectin-3
However, so far and to the best of our knowledge, the effect
of multivalent presentation of the TF-antigen on BSA-based
NGPs for affinity studies with Gal-3 has not been investigated
yet.
The recombinant galactosidase and galactosynthase were
expressed and purified as described before [24, 25]. In brief,
E. coli BL21 (DE3) cells were transformed with the plasmid
pETDuet-1 carrying the recombinant gene for the galactosi-
dase BgaC or the galactosynthase BgaC/Glu233Gly, respec-
tively. Expression of the recombinant gene was conducted for
24 h in TB-medium at 25 °C after induction with 0.1 mM
IPTG. Purification of both enzymes was achieved by
immobilized metal-ion chromatography (IMAC).
Recombinant Gal-3 protein was produced and purified as
described previously [26]. Human Gal-3 was expressed with an
N-terminal His6-Tag in E. coli Rosetta (DE3) pLysS and purified
via IMAC. After purification, the lectin was stored at 4 °C in
phosphate buffered saline (PBS) containing 2 mM EDTA.
We herein report on the synthesis and tunable multivalent
presentation of Gal(β1,3)GalNAc(α1-EG3-azide (TF-antigen-
azide) on BSA-based NGPs and their binding interaction with
Gal-3 (Scheme 1). We compared the glycosidase BgaC from
Bacillus circulans and its glycosynthase mutant BgaC/
Glu233Gly [23] regarding the synthesis of the TF-antigen-
azide, which we subsequently coupled to alkynyl-
functionalized lysine residues of BSA via a copper(I)-catalyzed
alkyne-azide cycloaddition (CuAAC). Varying glycan densities
between 2 and 53 glycans per BSA were obtained and Gal-3
binding properties of the multivalent NGPs were evaluated by
an enzyme-linked lectin assay (ELLA) (Scheme 2).
Enzymatic activity of BgaC β-galactosidase
Materials and methods
Hydrolytic activity of the BgaC galactosidase was determined
as described previously [25]. 100 μL of appropriately diluted
enzyme solution were added to 900 μL of 4.4 mM 2 in 50 mM
citrate-Na2HPO4buffer (pH 6.0) and incubated for 5 min.
Samples of 100 μL were taken at different time points and
stopped by the addition of 200 μL of 200 mM Na2CO3.
Subsequently, the signal was measured at 405 nm.
Quantification was done via a p-nitrophenol (pNP, 3) standard
curve.
Materials
GalNAc(α1-(tri(ethylene glycol))-azide (GalNAc(α1-EG3-
azide, 1) was purchased from Sigma Aldrich (now Merck,
Darmstadt, Germany). Tris(3-hydroxypropyltriazolylmethyl)
amine (THPTA) for CuAAC was obtained from TCI
Chemicals (Eschborn, Germany). Propargyl-EG1-NHS (9) es-
ter was obtained from Broadpharm (San Diego, USA). p-

Glycoconj J
Scheme 1 a: Synthesis of the TF-antigen-azide 6 with BgaC and the
glycosynthase mutant BgaC/Glu233Gly. b: Synthesis of alkynyl-
functionalized BSA and coupling of 6 by click chemistry. The maximum
theoretical number of available amines (n = 60) is based on the sum of ε-
amino groups of lysines in the BSA sequence (59) and the N-terminus of
the protein (1). m: range of coupled glycans/BSA. BSA protein structure
was modified from PDB entry 3 V03 (rcsb.org) [33] using UCSF
Chimera [34]. 1: GalNAc(α1-EG3-azide; 2: p-nitrophenyl-β-D-
galactoside (pNP-Gal); 3: p-nitrophenol (pNP); 4: α-D-galactopyranosyl
fluoride (α-GalF); 5: Fluoride; 6: Gal(β1–3)GalNAc(α1-EG3-azide; 7:
Galactosyl-oligosaccharide (GAOS); 8: Bovine serum albumin (BSA); 9:
Propargyl-EG1-NHS-ester; 10: N-hydroxy-succinimide; 11: Alkynyl-
functionalized BSA; 12: TF-antigen-presenting neoglycoprotein.
Enzymatic activity and kinetic analysis
of BgaC/Glu233Gly galactosynthase
activity is defined as the amount of enzyme that converts
1 μmol acceptor substrate per minute.
For acceptor substrate kinetics, the concentration of 1 was
varied between 0 and 55 mM while the donor substrate 4 was
fixed at 25 mM. For the donor substrate kinetics, 4 was varied
while the acceptor substrate 1 was fixed at 5 mM. Values for
Enzymatic activity of the galactosynthase BgaC/Glu233Gly
was determined via a discontinuous HPLC assay. 10 mM 4 and
5 mM of 1 were added to 50 mM HEPES buffer (pH 7.2) and
pre-heated to 30 °C. Enzyme solution containing 600 μg of
protein was added to start the reaction. The total volume was
100 μL. Samples of 10 μL were inactivated at 95 °C for 5 min,
centrifuged and diluted with water (1:5). Reversed-phase
HPLC was conducted on a Multochrom 100–5 C18 column
(250 × 4 mm, CS Chromatographie, Langerwehe, Germany)
with an isocratic eluent consisting of 90% water and 10%
acetonitrile and a flow rate of 1 mL/min. One unit of enzyme
KM and vmax were calculated using SigmaPlot.
Synthesis of Gal(β1,3)GalNAc(α1-EG3-azide (6) using
BgaC galactosidase
Synthesis was conducted on a 1 mL scale. The solution
contained 10 mM 2 as well as 5 mM 1 in 50 mM Na2HPO4-
citrate buffer (pH 6.0). The final concentration of the enzyme
Scheme 2 Representation of the
two types of ELLAs used in this
study. a NGPs with different
glycan densities are immobilized
in the well of a microtiter plate.
Binding of Gal-3 is quantified via
the conversion of 3,3′,5,5′-
tetramethylbenzidine (TMB) by a
peroxidase (PO) coupled to an
anti-His6-antibody. b NGPs are
used as inhibitors in solution for
the binding of Gal-3 to
immobilized asialofetuin (ASF)

Glycoconj J
was 0.3 U/mL. Samples were taken at different time points
and the reaction was stopped at 95 °C for 5 min. After centri-
fugation and appropriate dilution, the supernatant was ana-
lyzed via HPLC. The flow rate was 1 mL/min. The ratio of
the two solvents in the eluate at given time points was as
follows: 90% H2O for 15 min, decrease to 40% H2O within
5 min, steady 40% H2O for 5 min, return to 90% H2O within
2 min. Peaks were detected at 205 nm and quantification was
carried out via external calibration curves.
The product of the galactosynthase reaction was dissolved
in D2O and analyzed via H-NMR spectroscopy. Due to the
1
low resolution and high overlap of the peaks in the acquired
spectrum, the product was peracetylated and analyzed via 1H-
and 13C-NMR using CDCl3 as the solvent (Fig. S8-S9).
Peracetylation was carried out as described by Steinmann
et al. [26]. Briefly, the enzymatic product was dissolved in
pyridine (1 mL/mmol), acetic anhydride (1.5 eq/OH-group)
added, and stirred over night at 25 °C. The reaction was dilut-
ed with ethyl acetate and pyridine removed by extraction with
saturated CuSO4-solution. The organic phase was dried over
MgSO4 and subsequently filtered. The solvent was then evap-
orated under reduced pressure.
Synthesis and purification of Gal(β1,3)
GalNAc(α1-EG3-azide (6) using galactosynthase
BgaC/Glu233Gly
Alkynyl-modification of BSA (8)
Semi-preparative synthesis was performed in a volume of
2 mL containing 10 mM 1 and 20 mM 2 in 50 mM HEPES
buffer (pH 7.2). The synthesis was monitored via HPLC as
described above. After the reaction was finished, the mixture
was heated to 95 °C for 5 min and subsequently centrifuged.
Remaining enzyme in the supernatant was removed by ultra-
centrifugation (Vivaspin® (Sartorius), 30 kDa MWCO) and
the volume was reduced via vacuum evaporation (Eppendorf
concentrator plus, Eppendorf, Germany). The product 6 was
isolated via preparative HPLC (Knauer Smartline) with col-
umn Multokrom 100–5 C18 (250 × 20 mm) by applying an
isocratic eluent consisting of 90% water and 10% acetonitrile
and a flow rate of 12.5 mL/min. The product fractions were
collected and the solvent was evaporated to dryness. The final
product was used for further analysis and subsequently dis-
solved in water.
Prior to modification, BSA 8 was delipidated via activate
charcoal treatment as described previously [27]. Depending
on the desired degree of alkynyl-modification, 16.7 μM
delipidated BSA was incubated with 0.2–20-fold molar ex-
cesses of Propargyl-EG1-NHS-ester 9 (Broadpharm, San
Diego, USA) for 1.5 h in 50 mM PBS (pH 7.4) at 25 °C.
NHS 10 and unreacted linker were removed via ultracentrifu-
gation (Vivaspin® 500 (Sartorius), 10,000 Da MWCO). The
degree of alkynyl-modification of 11 was determined via
SDS-PAGE (8% polyacrylamide gels, 25 mA constant cur-
rent) and the trinitrobenzene sulfonic acid (TNBSA)-assay
as reported previously [28]. Briefly, 50 μL of 12.5 μM 11 in
sodium tetraborate buffer (pH 9.0) were added to 50 μL of
7.5 mM TNBSA in the same buffer. The mixture was incu-
bated for 15 min at room temperature. Subsequently, the ab-
sorbance was measured at 420 nm. Each sample was analyzed
in triplicates and the standard deviation was calculated.
Quantification was achieved by comparing the signal of
alkynyl-modified BSA to unmodified BSA.
Mass spectrometry
Analysis of the molecular mass of the reaction products was
carried out via ESI-MS (Finnigan Surveyor MSQ Plus,
Thermo Scientific, needle voltage = 4 kV, temperature =
400 °C, cone voltage = 100 V, negative mode) via LC-MS-
measurement.
Coupling of TF-antigen-azide (6) to BSA-alkynyl
proteins (11) via CuAAC
Conjugation of glycan molecules with 11 was carried out via
CuAAC based on protocols described elsewhere [29, 30]. The
final concentrations of reagents (except for the buffer) varied
with the degree of alkynyl-functionalization based on the re-
sults of the TNBSA-assay. However, the molar ratios of the
substances were unchanged in the respective reactions and
amounted to Gal(β1,3)GalNAc(α1-EG3-azide 6/alkynyl
groups/Cu(I)/THPTA/ascorbate (1/0.5/2/10/40). For example,
1 mM of 6 was added to 0.5 mM alkynyl residues present on 11
in K2HPO4-buffer (final concentration 100 mM, pH 7.2). A
premixed solution of CuSO4 and THPTA was added to the
solution to a final concentration of 2 mM and 10 mM, respec-
tively. The addition of sodium ascorbate (final concentration
40 mM) initiated the reaction. The mixture was incubated for
Nuclear magnetic resonance
NMR spectra were recorded on Bruker Avance/DRX600
spectrometer (1H: 600 MHz, 13C: 151 MHz) in D2O and
CDCl3 at 30 °C. The residual solvent signals were used as
references (CDCl3: δH 7.26 ppm, δC 77.16 ppm; D2O: δH
4.79 ppm). Chemical shifts (δ) were reported in parts per
million (ppm) and the coupling constants (J) in Hertz (Hz).
The anomeric configuration of the glycosyl moieties was de-
termined by the coupling constant (J) between the anomeric 1-
H and the 2-H. The linkage of the glycosidic bond was deter-
mined by coupling signals between the 3′-C and 1″-C ob-
served via 1H-13C-HMBC NMR analysis.

Glycoconj J
3 h at 37 °C. For use in further experiments, neo-glycoproteins
12 were purified from the mixture via ultracentrifugation
(Vivaspin® 500, 10,000 Da MWCO). The coupling of glycan
molecules was assessed and quantified via SDS-PAGE (8%
polyacrylamide gels, 25 mA constant current).
configured pNP-HexNAc-glycosides over β-configured gly-
cosides as acceptor substrates.
In the present study, we first compared the synthesis of the
TF-antigen-EG3-azide 6 with the wild type galactosidase en-
zyme BgaC and the galactosynthase BgaC/Glu233Gly, re-
spectively, using the novel acceptor substrate GalNAc(α1-
EG3-azide 1 (Scheme 1A). Reactions with BgaC resulted in
a low synthetic yield (approx. 60% after 10 min reaction time)
with concomitant formation of a trisaccharide product as well
as galacto-oligosaccharides (GAOS, 7) (Fig. 1 and Fig. S1 in
supporting information). The identity of the side products was
proven by analysis of the respective peak masses by LC-ESI-
MS (Fig. S2-S4). After a reaction time of 10 min, the concen-
tration of product 6 decreased while concentration of the ac-
ceptor substrate 1 increased again until the initial reaction
conditions concerning these compounds were restored. The
donor substrate pNP-Gal was depleted after approx. 15 min.
Glycosynthases are known for their lack of hydrolytic ac-
tivity by replacing the nucleophilic amino acid in the reactive
center by a small non-nucleophilic residue [32]. We concluded
that BgaC/Glu233Gly is a suitable tool for the synthesis of the
TF-antigen-EG3-azide 3. Kinetic analysis revealed an appar-
ent KM,app of 0.22 mM and a vmax of 2.6 mU/mg for the donor
substrate 4 (Fig. 2). Regarding the acceptor substrate 1, satu-
ration of the fitted curve was not obtained with reasonable
substrate concentrations. Consequently, KM and vmax were
not determined in this study.
Enzyme-linked Lectin assay (ELLA)
For the binding assay (Scheme 2a), the respective neo-
glycoprotein 12 (50 μL, 0.1 μM) was immobilized overnight
in triplicates in wells of a 96-well microplate (MaxiSorp,
Nunc, Wiesbaden, Germany). 200 μL of 2% [w/v] BSA in
PBS was added and incubated for 1 h to block residual bind-
ing sites. Subsequently, 50 μL of Gal-3 in PBS supplemented
with 2 mM EDTA was added to each well in varying concen-
trations and incubated for 1 h. Anti-His6-Peroxidase (Roche,
Mannheim, Germany) was added for 1 h to detect bound His6-
tagged galectin. Quantification was achieved via the conver-
sion of 50 μL TMB One substrate by the immobilized perox-
idase to a blue solution. The reaction was stopped after 45 s by
the addition of 50 μL of 3 M HCl to yield a yellow solution.
The signal was read out at 450 nm. Between the steps, wells
were washed three times with 200 μL PBS supplemented with
0.05% [v/v] Tween-20 (PBS-T).
For the competitive inhibition assay (Scheme 2b), 50 μL of
0.1 μM (5 pmol) asialofetuin (ASF) per well were incubated in
triplicates overnight in the wells of a microplate (MaxiSorp,
Nunc, Wiesbaden, Germany). Subsequently, 200 μL of 2%
[w/v] BSA were added and incubated for 1 h to prevent un-
specific adsorption of proteins. 25 μL of inhibitor (lactose,
TF-antigen-azide 6 or NGP 12) in various concentrations
was added in triplicates, directly followed by 25 μL of Gal-3
(25 μM final concentration). Samples were incubated for 1 h.
Signal detection was conducted as mentioned above for the
binding assay. Between each step, wells were washed three
consecutive times with 200 μL PBS-T.
Galactosynthase BgaC/Glu233Gly was applied in the syn-
thesis of the TF-antigen-EG3-azide 6 (Scheme 1A) resulting
in 97% conversion of 1 after 24 h (Fig. 3). The formation of a
single product peak was observed by HPLC (Fig. S5). The
product peak was isolated by preparative RP-HPLC with a
final yield of 80% [mol/mol]. Analysis by LC-ESI-MS con-
firmed the calculated corresponding molecular ion mass of
539.3 g/mol for 6 (Fig. S6). The β1,3-linkage of 6 was veri-
fied via 1H-NMR and 13C-NMR (Fig. S7-S9).
Synthesis and analysis
Results
of TF-antigen-neo-glycoproteins (12)
Enzymatic synthesis of TF-antigen-EG3-azide (6)
The purified product 6 was used for the synthesis of BSA-
based neo-glycoproteins (NGPs 12) (Scheme 1B). First,
alkynyl-functionalization of the ε-amino groups of lysines of
BSA 8 by propargyl-EG1-NHS ester 9 was performed. BSA 8
consists of 583 amino acids with 59 incorporated lysine resi-
dues (UniProt accession number P02769). Including the N-
terminus of the protein, this sums up to 60 potential modifi-
cation sites. The degree of BSA alkynyl-functionalization was
controlled by varying the molar ratio between 9 and the amine
groups. The number of modified sites in 11 was determined by
quantification of the remaining unmodified lysine residues via
a TNBSA assay before and after treatment with 9. In addition,
reducing SDS-PAGE was performed to further assess the
In our previous studies, we reported on the generation of the
galactosynthase BgaC/Glu233Gly by rational design based on
the crystal structure of BgaC from B. circulans and its use in
combination with glycosyltransferases for the synthesis of
various type 1 and type 2 poly-LacNAc structures [23, 24].
The same enzyme variant has been shown to catalyze the
formation of 4-nitrophenyl α-D-2-N-galacto-N-biose
(pNP-αGNB, Gal(β1,3)GalNAc(α-pNP) using 4-
nitrophenyl α-D-2-N-acetylgalactosaminide (pNP-αGalNAc)
as the acceptor substrate [31]. The kinetic analysis in this
study suggested that the galactosynthase prefers α-

Glycoconj J
Fig. 1 Synthesis of the TF-
antigen-EG3-azide 6 with BgaC
(Scheme 1a). The synthesis reac-
tion was conducted with 10 mM
pNP-Gal 2 and 5 mM
GalNAc(α1-EG3-azide 1 in
50 mM Na2HPO4-citrate-buffer
(pH 6.0). pNP-Gal 2 was used as
the donor substrate. pNP 3 is a
side product of the reaction.
Analysis was carried out via
HPLC at a detection wavelength
of 205 nm. Quantification was
done by external standard curves
degree of alkynyl-modification. Subsequent coupling via
CuAAC with a two-fold molar amount of TF-antigen-EG3-
azide 6 over the alkynyl-labeled amino residues resulted in
NGPs with a varying number of coupled glycans. Molecular
weight shifts of NGPs were determined by the difference in
electrophoretic mobility before and after coupling (Fig. 4,
Table 1 and Fig. S10).
Figure 4a indicates that an increasing excess of 9 during the
coupling reaction leads to an increasing molecular weight of
alkynyl-modified BSA 11. This is also confirmed by the
TNBSA assay (Fig. 4c and Table 1). The numbers of
alkynyl-modified sites derived from both analytical methods
are in accordance when lower molar excesses of 9 are applied.
However, values vary significantly for samples treated with
more than a 4-fold molar excess of 9. The TNBSA assay
shows that the maximum of 60 sites per BSA molecule carry
the PEG-alkynyl moiety at a 4-fold molar excess of the linker
(Fig. 4c). This number does not increase when higher amounts
are used. However, SDS-PAGE analysis shows an alkynyl-
modification density of up to 114 alkynyl residues per BSA
molecule when the molar excess of 9 is increased to 20
(Table 1).
In a second step, the purified TF-antigen-azide disac-
charide 6 was coupled to 11 via CuAAC chemistry
(Scheme 1B). A molar ratio of 2:1 for azide and alkyne
functional groups was applied in each reaction. The num-
ber of alkynyl-carrying residues used for the calculations
was derived from the TNBSA assay for molar excess ratios
of 9 below 4:1 and from SDS-PAGE for molar excess ra-
tios above 4:1. SDS-PAGE analysis (Fig. 4c and Table 1)
of NGPs 12 indicated that the mass difference before and
after CuAAC in comparison to unmodified BSA increases
with increasing alkynyl modification of 11. Molecular
weight shifts were calculated using linear regression (Fig.
Fig. 2 Kinetic analysis of BgaC/Glu233Gly regarding the substrates α-
GalF 4 (a) and GalNAc(α1-EG3-azide 1 (b). (a) was fitted according to
the Michaelis-Menten equation (Eq. S1). Vmax: Not determined; KM:
Not determined. (b) was fitted according to the Hill equation (Eq. S2).
Vmax: 2.6 mU/mg; KM: 0.22 M. Each data point was generated in tripli-
cates. The standard deviation of the mean is provided in the form of error
bars

Glycoconj J
S10). Variable glycan densities between 2 and 53 glycans
per BSA molecule were obtained (Table 1).
Galectin-3 binding to immobilized TF-antigen
neo-glycoproteins (12)
Selected NGPs were immobilized in the wells of microplates
for determination of the binding affinity of human galectin-3
(Gal-3) in an enzyme-linked lectin assay (ELLA) (Scheme 2
and Fig. 5). The increase of binding signals resulting from the
binding of Gal-3 to immobilized NGPs with increasing glycan
densities while there is no binding signal for unmodified BSA.
NGPs with valencies below 8 glycans/BSA showed very
weak binding signals (Fig. 5a).
Fig. 3 Synthesis of the TF-antigen-azide 6 by BgaC/Glu233Gly
(Scheme 1a). Conversion of GalNAc(α1-EG3-azide 1 ( ) to the TF-
antigen-azide 6 ( ). The synthesis reaction was conducted in triplicates
on a 2 mL-scale with 20 mM οf 4 and 10 mM of 1 in 50 mM Na2HPO4-
citrate-buffer (pH 6.0). Standard deviations of the mean are provided as
error bars
However, increased binding of Gal-3 was observable to
NGPs with higher glycan densities starting from ≥8 glycans
Fig. 4 Results of the alkynyl-functionalization of amine residues of BSA
and subsequent coupling of TF-antigen-EG3-azide 6 to alkynyl-modified
BSA 11 via CuAAC chemistry (Scheme 1B). M: PageRuler Prestained
Protein Ladder. Each gel was loaded with 2 μg/lane of protein. a SDS-
PAGE of BSA samples after alkynyl-functionalization with the indicated
molar ratios of propargyl-EG1-NHS ester 9 and terminal amine residues.
The number of terminal amines consists of the number of ε-amines from
lysine residues (59) plus the N-terminus of the protein (1). The increase in
molecular weight (MW) compared to unmodified BSA is 0.111 kDa per
coupled propargyl-EG1-moiety. b SDS-PAGE of neo-glycoproteins after
CuAAC-catalyzed coupling of TF-antigen-EG3-azide 6 to the alkynyl-
modified BSA samples. The increase in MW compared to unmodified
BSA is 0.652 kDa per coupled glycan. c Comparison of the number of
alkynyl residues derived from the TNBSA assay and SDS-PAGE and
number of TF-antigens per BSA according to SDS-PAGE analysis. The
retardation factor (Rf) of the samples in the gel was determined and their
molecular weight was calculated by linear regression based on the protein
size standard (lane M) (Fig. S10). The number of attached propargyl-
EG1-linker and glycans was calculated by the differences in molecular
weights between unmodified BSA 8 and modified BSA samples 11 and
12, respectively. The TNBSA assay was conducted in triplicates.
Standard deviations of the mean are provided as error bars

Glycoconj J
Table 1 Number of alkynylated and glycosylated sites in the BSA
protein sequence according to analysis via TNBSA assay and SDS-PAGE
coupled glycan were determined at different Gal-3 concentra-
tions. For samples below 28 glycans/BSA, binding signals
were not observed at Gal-3 concentrations of 5 μM or lower
(Fig. 5). Therefore, NGPs with glycan densities higher than 28
glycans/BSA were compared in relation to the binding signal
of BSA28 (Fig. 6).
Propargyl-EG1-
NHS-ester 9/ termi- BSA
nal amine [mol/
mol]
Alkynes/
Alkynes/
BSA
(PAGE)
TF-Ag/
BSA
(PAGE)
Yield
CuAAC
[%]
(TNBSA)
[mol/mol]
[mol/mol] [mol/mol]
An increase of binding signals per glycan was observed for
a Gal-3 concentration of 10 μM, however, these numbers also
display high standard deviations. For other Gal-3 concentra-
tions binding signals reached approximately the same signal
level per glycan. We concluded that the NGPs show no sig-
nificant multivalency effect.
0
0
7
1.4
0.5
0
0
2
5
9
15
19
22
28
32
33
40
41
53
–
0.2
0.33
0.5
1
1.5
2
3
4
5
14
19
28
33
38
47
57
62
72
88
98
114
32
33
34
47
46
45
48
51
45
45
42
47
14 1.3
25 0.9
32 0.6
41 0.1
48 0.5
57 0.0
60 0.0
60 0.0
60 0.0
60 0.2
60 0.1
Competitive inhibition of Galectin-3 binding
by TF-antigen neo-glycoproteins (12)
7.5
10
20
The potential of monovalent TF-disaccharide 6 as well as TF-
antigen NGPs 12 to inhibit the Gal-3 interaction with ASF was
investigated. In order to provide enough sample material, synthe-
sis of NGPs 12 was performed on a larger scale (6 mL) with the
aim to synthesize samples with low, medium and high glycan
densities (Fig. 7 and Table S1). For this purpose, alkynyl func-
tionalities were introduced by using 2-, 10- and 20-fold molar
excesses of 9 during the coupling reaction. The resulting alkynyl
modification degrees of 10, 37 and 46 glycans per BSA were
comparable to the previous results on a smaller scale (Fig. 4 and
Table 1). Again, the available number of alkynyl residues defined
the final number of glycan residues per BSA molecule after
coupling of 9 to BSA. Yields regarding CuAAC ranged between
42 and 51%.
Yields of the click reaction were calculated on the basis of the number of
alkynes present according to TNBSA assay (for NHS-ester 9
molar excess <4) or SDS PAGE (for NHS-ester 9 molar excess ≥4).
The TNBSA-assay was conducted in triplicates. The standard deviation
of the mean is provided behind the calculated number of alkynyl groups
(Fig. 5b). Distinguishable binding signals were observed for a
Gal-3 concentration of ≥10 μM. The results confirm the accessi-
bility of the TF-antigen ligand on the immobilized NGPs for Gal-
3 binding. However, apart from ASF as a standard glycoprotein
for Gal-3 binding, binding curves for the applied Gal-3 concen-
trations did not reach saturation. Therefore, apparent KD-values
were not calculated.
Furthermore, potential effects of multivalent glycan display
were investigated. For this purpose, binding signals per
The resulting neo-glycoproteins presenting 10 (BSA10),
37 (BSA37) and 46 (BSA46) mol TF-antigen/mol BSA,
Fig. 5 Analysis of Gal-3 binding to immobilized TF-antigen NGPs with
glycan densities between 0 and 53 mol glycan / mol BSA in an enzyme-
linked lectin assay (ELLA). a: glycan densities between 2 and 8 mol TF-
antigen / mol BSA. b: glycan densities between 19 and 53 mol TF-
antigen / mol BSA. Sample designation indicates mol TF-antigen / mol
BSA. NGPs were immobilized in wells of a microplate (5 pmol/well) and
incubated with varying amounts of recombinant human Gal-3. Each sam-
ple was measured in triplicates. ASF served as a positive control.
Background for blank samples (no Gal-3) were subtracted from the bind-
ing signals. Final binding signal values are plotted for varying Gal-3
concentrations. All Gal-3 concentrations were analyzed in triplicates.
All curves were fitted using the software SigmaPlot

Glycoconj J
of 6 (Fig. S2) and the trisaccharide bearing an additional ga-
lactose (GalβGal(β1,3)GalNAc(α1-EG3-azide) (Fig. S4 and
S5). In contrast to previous reports, the formation of β1,6- or
β1,4-linked disaccharides resulting in distinct peaks can be
excluded by LC-ESI-MS analysis as none of the observed
peaks except for peak 3 revealed the mass of a Gal-GalNAc-
EG3-azide disaccharide [35, 36].
In contrast, the reaction of the galactosynthase shows the
formation of 6 as single product. The high product yield re-
sults from a lack in product hydrolysis [32]. In addition,
BgaC/Glu233Gly, unlike other reported glycosynthase mu-
tants [37], does not accept β1,3-linked galactosides as accep-
tor substrates [31]. Further reduction in yield due to the for-
mation of GAOS is therefore prevented. Thus, application of
the BgaC/Glu233Gly galactosynthase in the TF-antigen syn-
thesis resulted in a significant increase in product yield sim-
plifying product isolation.
Fig. 6 Relative binding signal per glycan for neo-glycoproteins with
different glycan densities at Gal-3 concentrations between 5 and 50 μM
respectively, were evaluated as competitive inhibitors for Gal-
3 binding to ASF using ELLA analysis (Fig. 8 and Table 2).
The monovalent disaccharides showed comparable IC50-
values in the mM-range with 1.51 mM and 1.56 mM for lactose
and TF-antigen-azide 6, respectively. Therefore, a high con-
centration between 100 μM – 10 mM must be applied to inhibit
the interaction between Gal-3 and the multivalent glycopro-
tein ASF. However, significantly lower IC50-values between
11 and 17 μM are observed for all tested neo-glycoproteins
(Table 2). The value of relative inhibitory potency (RIP) is a
benchmark for the relative strength of an inhibitor. The num-
ber of displayed TF-antigen glycans can be reduced to 10
glycans/mol BSA, which results in the highest RIP for
BSA10.
We conclude that the reaction of BgaC is difficult to control
and BgaC/Glu233Gly is the preferred enzyme to reach the
highest possible yield of 6.
Kinetic analysis of the synthesis reaction using
the glycosynthase BgaC/Glu233Gly
Kinetic analysis of BgaC/Glu233Gly revealed an apparent
KM,app of 0.22 mM and a vmax of 2.6 mU/mg regarding the
donor substrate αGalF 2 (Fig. 2). For comparison, a KM-value
of 1.0 mM was reported for 2 using pNP-αGalNAc as the
acceptor substrate [31]. Interestingly, KM,app and vmax could
not be determined for the novel acceptor substrate 1 as the
fitted curve did not show saturation within the investigated
substrate concentration range. In a previous study, KM values
between 0.71 mM and 0.41 mM were reported for the aryl-
glycosides pNP-αGlcNAc and pNP-αGalNAc, respectively
[31]. The difference in KM values may be due to a generally
higher affinity of the wild type glycosidase for aryl-substituted
glycosides compared to unmodified acceptor substrates [38].
It can be assumed that the same holds true for the respective
glycosynthase.
The RIP per glycan quantifies the effect of multivalency on
the affinity of Gal-3 towards the inhibitor. Here, BSA10 also
shows the highest number and therefore the greatest multiva-
lent effect.
Discussion
Synthesis of TF-antigen-EG3-azide (6)
Synthesis of TF-antigen-NGPs (12)
A comparison was made between synthesis of the TF-antigen
using the galactosidase BgaC and its glycosynthase equivalent
BgaC/Glu233Gly. Both enzymes were capable of product for-
mation with conversions of approx. 60% after 10 min and
97% after 24 h for BgaC and BgaC/Glu233Gly, respectively.
BgaC shows significant product hydrolysis and formation of
side products (Fig. 1) compared to the galactosynthase (Fig.
2). The transglycosylation product TF-antigen-EG3-azide 6 is
hydrolyzed and also utilized as an acceptor substrate for fur-
ther galactosylation yielding a trisaccharide product and fur-
ther galacto-oligosaccharides with additonal galactose moie-
ties (Scheme 1). Analysis via ESI-MS revealed the formation
Variation of the ratio of propargyl-EG1-NHS-ester 9 and
available amines of BSA 8 results in variable alkynyl-
modification of BSA. This is reflected by the analysis of the
coupling reaction of 9 with BSA 8 via TNBSA-assay and
SDS-PAGE. A higher number than 60 modified residues per
mol BSA are obtained when molar excesses of more than 4:1
of are applied. While the results of the TNBSA-assay show
that all 60 lysine residues of BSA react with 9, the apparent
molecular weights in SDS-PAGE analysis indicate that even
more than 60 molecules of 6 are attached to the protein when
higher excesses of 9 are used. The TNBSA reagent is known

Glycoconj J
Fig. 7 Results of the alkynyl-modification of terminal amine residues of
BSA 8 and subsequent coupling of TF-antigen-azide 6 to the acylated
sites via CuAAC chemistry on a 6 mL-scale. a SDS-PAGE of 11 after
alkynyl modification with the indicated molar ratios between Proparyl-
EG1-NHS ester 9 and terminal amine residues. The number of terminal
amines is put together by the number of ε-amines from lysine residues
(59) plus the N-terminus of the protein (1). The added molecular weight
(MW) to unmodified BSA of the coupled propargyl-EG1-moiety after
cleavage of the NHS-group 10 is 0.111 kDa. Each lane was loaded with
2 μg of protein. M: PageRuler Prestained Protein Ladder (ThermoFisher,
Waltham, USA). b SDS-PAGE of NGPs 12 after CuAAC-catalyzed cou-
pling of TF-antigen-azide 6 to 11. The added MW to unmodified BSA of
the coupled glycan is 0.652 kDa. Each lane was loaded with 2 μg of
protein. M: PageRuler Prestained Protein Ladder (ThermoFisher,
Waltham, USA). c Comparison of the number of alkynyl functionalized
residues derived from the TNBSA assay and SDS-PAGE and number of
TF-antigen per BSA according to SDS-PAGE analysis. The retardation
factor (Rf) of the samples in the gel was determined and their molecular
weight was calculated by linear regression based on the protein size
standard (lane M) (Fig. S11). The number of attached propargyl-EG1-
linker and glycans was calculated by the differences in molecular weights
between unmodified BSA 8 and modified BSA samples 11 and 12, re-
spectively, and is noted above the bars. The TNBSA-assay was conducted
in triplicates
to react exclusively with the ε-amines of lysine residues [39]
and is therefore limited to the quantification of available lysine
residues in the BSA sequence. In contrast, SDS-PAGE detects
any molecular weight shift independent from the site at which
the modification occurred. We hypothesize that 9 reacts with
additional sites in the BSA structure apart from the lysine
residues. Previous studies report the formation of ester bonds
with serine, tyrosine, and threonine residues of proteins and
small peptides [40, 41]. These residues constitute 13.5% of the
BSA sequence. The formation of ester linkages is further en-
hanced if a histidine residue is located in direct proximity to
one of these side chains [42]. The histidine-catalyzed O-acyl-
ation of side chains containing OH-groups is increased at low-
er pH-values between pH 6.7–7.8 as the low pKs-value of
His-side chains causes the imidazole ring to be deprotonated.
This, in turn, results in increased nucleophilicity and thus
higher reactivity towards NHS-activated esters. The N-acylat-
ed imidazole ring serves as an intermediate in the reaction and
the alkynyl-crosslinker is subsequently transferred onto direct-
ly adjacent hydroxyl-bearing side chains. This mechanism has
been applied before in the selective targeting of tyrosine res-
idues of bovine α-lactalbumin [43]. Moreover, it can be ex-
pected that the side chains of a globular protein are subject to
additional intramolecular interaction (e.g. hydrogen bonding)
which may have an effect on their nucleophilicity and thus can
potentially alter their reactivity towards acylation. Therefore,
it is important to put into consideration that at high excess
ratios of 9 additional sites of BSA 8 (Ser, Thr or Tyr) are

Glycoconj J
Fig. 8 Competitive inhibition assays of Gal-3 binding to ASF by mono-
valent TF-antigen-azide 3 and lactose (a) and BSA-based neo-glycopro-
teins decorated with 10, 37 and 46 TF-antigen/BSA (b). Binding signals
were detected using enzyme-linked lectin assay (ELLA). The residual
binding of 25 μM Gal-3 to 5 pmol/well of immobilized ASF after incu-
bation with varying concentrations of inhibitor is depicted. All inhibitor
concentrations were analyzed in triplicates
probably modified. These modifications are not detected with
the TNBSA assay. For lower ratios, the results obtained via
the TNBSA assay and SDS-PAGE are in accordance and yield
comparable numbers of modified side chains (Table 1). We
assume that the lysine residues are acetylated at lower NHS-
ester excess ratios.
per BSA seem possible due to the O-acylation of Tyr, Thr and
Ser within the BSA sequence.
To sum up, our approach enables the facile formation of
neo-glycoproteins combining enzymatic and chemical proto-
cols. Since the availability of 6 is the limiting factor in terms of
costs and effort, we attempted to use a reasonable excess of the
disaccharide during CuAAC, resulting in conversions of
around 50%. The resulting neo-glycoproteins contained be-
tween 2 and 53 glycans per molecule. Therefore, the presented
protocol enables the formation of tailor-made neo-glycopro-
teins with a desired glycan density.
Our data demonstrate that the number of attached glycans
can be controlled by the number of alkynyl-functionalized
amino acid residues and a constant azide/alkyne ratio. The
conversion of the click reactions with respect to the number
of alkynyl residues are constant between 42 and 51%, if the
number of alkynyl residues per BSA is at least 33 or higher
(Table 1). This is achieved by using a propargyl-EG1-NHS
ester 9/terminal amine-ratio of at least 1:1 or higher. The mod-
erate yields may be due to partially inaccessible alkynyl-
carrying side chains for the TF-antigen-azide 6. However,
previous studies on CuAAC-catalyzed labeling of human se-
rum albumin suggest that a higher excess of the glycan during
CuAAC might further increase the glycosylation degree to 60
glycans/BSA [44]. Our results suggest that with propargyl-
EG1-NHS ester 9 even glycan densities beyond 60 residues
Binding of Gal-3 to TF-antigen NGPs (12)
Our resultsshowthat bindingvaluesof Gal-3toimmobilized
NGPs displaying different numbers of glycans increased
with increasing glycan density (Fig. 5). However, in compar-
ison to ASF, the standard for Gal-3 binding analysis, binding
affinity remained low even at high occupation numbers. In
addition, an effect of multivalent presentation of the TF-
antigen was not observed (Fig. 6). In contrast, inhibition
studies showed that multivalent presentation of the TF-
antigen on BSA neoglycoproteins enhances the relative in-
hibitory potency by two orders of magnitude. The effect of
multivalency, however, was dependent on the degree of gly-
can decoration. BSA10, presenting 10 TF-antigen glycans,
exhibits a 142-fold higher relative inhibitory potency (RIP)
compared to monovalent TF-antigen. NGPs with a higher
glycan density both showed lower RIP-values (98 for
BSA37 and 92 for BSA46). The RIP value relative to the
glycan density reveals that multivalency effects occurred
most strongly for BSA10 (RIP/glycan of 14.2). Apparently,
decoration with only 10 glycans/BSA triggers a stronger
glyco-cluster effect [45] than a dense decoration with 37
(RIP/glycan of 2.6) or 46 (RIP/glycan of 2) glycans, respec-
tively. In comparison to our results depicted in Fig. 5, TF-
antigen NGPs are readily accessible for binding Gal-3 in
Table 2 Inhibition properties of TF-antigen-azide 3 and respective neo-
glycoproteins displaying different glycan densities
Ligand
ma
IC50 [mM]
RIPb
RIP/
Glycanc
Lactose
1
1
1.51 0.0584
1.56 0.0926
0.011 0.003
0.016 0.0004
0.017 0.0004
1
1
1
1
TF-antigen-azide (3)
BSA10
10
37
46
142
98
92
14.2
2.6
2
BSA37
BSA46
Lactose was utilized as a positive control
a m: Average number of glycans per ligand molecule; RIP: Relative
b
inhibitory potency compared to TF-antigen-azide, i.e.
c
IC50(unconjugated glycan)/IC50(neo-glycoprotein); RIP/Glycan:
Relative inhibitory potency per glycan compared to TF-antigen-azide,
i.e. RIP/m

Glycoconj J
solution whereas immobilization of TF-antigen NGPs re-
duces binding of Gal-3. These results are in accordance with
a previous study by Stowell et al., who found that no signif-
icant binding was detected when the TF-antigen was
immobilized on a glycan-array [46].
Conclusions
We here demonstrate the efficient enzymatic synthesis of
the EG3-azide-functionalized TF-antigen 3 and its cou-
pling to BSA via conventional click chemistry. The glycan
density of the resulting neo-glycoproteins is controllable
by the amount of propargyl-EG1-NHS ester 4 used during
alkynyl-functionalization of the protein. While monovalent
TF-antigen has proven to be a weak ligand for Gal-3, the
presented data shows that multivalent glycan presentation
does not lead to a higher binding affinity as previously
demonstrated for immobilized LacNAc-based neo-glyco-
proteins. This may be due to differences in the glycan pre-
sentation on BSA and the Gal-3 binding mode for the TF-
antigen. In contrast, the inhibitory potency of TF-antigen
NGPs for Gal-3 binding to ASF in solution was improved
by two orders of magnitude. The NGP with 10 TF-antigen
glycans depicted the highest relative inhibitory potential
per glycan, indicating that a minimal glycan density of
NGPs is favorable for inhibiting binding of Gal-3 to the
glycoprotein ASF.
Previous studies indicate that the monovalent attach-
ment of the TF-antigen to a MUC1-like peptide backbone
via an α1-linkage increases the affinity towards Gal-3 rel-
ative to the unconjugated glycan approximately 10-fold
[20, 47]. In addition, the multivalent display of the TF-
antigen on natural antifreeze-proteins from cod leads to a
significant glyco-cluster effect with KD-values approxi-
mately 6 orders of magnitude lower than for the unconju-
gated TF-disaccharide [22]. In contrast, multivalent pre-
sentation of the TF-antigen on glycan-arrays without link-
age to a peptide backbone did not result in a stronger bind-
ing [47]. Taken together, the coupling of the glycan to its
natural MUC1-scaffold or a similar peptide sequence
might be important for increasing its affinity towards
Gal-3. Gal-3 features a binding groove with the subsites
A-E. Subsite C and D bind the disaccharides lactose
(Gal(β1,4)Glc) or LacNAc type 2 (Gal(β1,4)GlcNAc),
whereas subsite A and B present a more flexible binding
pocket for additional interactions with e.g. poly-LacNAc
glycan chains [48–50]. We demonstrated sub-nanomolar
binding constants of the BSA-coupled tetrasaccharide
LacdiNAc-LacNAc due to glycan binding to the subsites
A-E of Gal-3 with additional binding interaction of the
terminal GalNAc residues in subsite A [17]. Therefore,
BSA as a protein scaffold for glycan presentation can be
regarded as functional for Gal-3 binding.
Monovalent disaccharides generally exhibit lower bind-
ing affinities for Gal-3 when compared to poly-LacNAc.
LacNAc type 2 is the best disaccharide ligand in compar-
ison to LacNAc type 1 (Gal(β1,3)GlcNAc) and the TF-
antigen [21, 48]. The TF-antigen disaccharide differs from
LacNAc type 2 with respect to the reducing sugar being
GalNAc instead of GlcNAc as well as the terminal Gal
being linked via a β1,3-linkage instead of a β1,4-linkage.
Structural analysis of Gal-3 binding the TF-antigen disac-
charide revealed that the GalNAc-moiety is responsible for
a unique binding pattern based on a Glu165-water-Arg186-
water-motif [8]. This binding pattern is not observed with
other glycan ligands of Gal-3. Binding of the Gal-moiety at
subsite C includes interaction with the same amino acids as
described for binding of LacNAc type 2 [8, 48]. Recently
reported binding constants of Gal-3 for TF-antigen (KD
47 μM) [8], LacNAc type 2 (KD 33 μM) and LacNAc type
1 (KD 93 μM) [51] suggest subtle differences. When com-
paring LacNAc type 1 and the TF-antigen, exchange of
GlcNAc with GalNAc in combination with the β1,3-link-
age apparently result in slightly better binding.
Acknowledgements Open Access funding provided by Projekt DEAL.
We acknowledge financial support by the DFG (Deutsche
Forschungsgemeinschaft) within the International Research Training
Group 1628, “Selectivity in Chemo- and Biocatalysis (SeleCa)”. BSA
graphics were created with UCSF Chimera, developed by the Resource
for Biocomputing, Visualization, and Informatics at the University of
California, San Francisco, with support from NIH P41-GM103311.
Authors’ contribution Marius Hoffmann and Lothar Elling designed the
study. Marius Hoffmann performed the experiments and calculated the
data. Marc R. Hayes performed the synthesis of α-GalF and NMR exper-
iments. Marius Hoffmann and Lothar Elling analyzed the data and drafted
the manuscript. Marc Hayes and Jörg Pietruszka critically reviewed the
manuscript. All authors read and approved of the final manuscript.
Compliance with ethical standards
Conflict of interest All authors declare that they have no conflict of
interest.
Ethical approval This article does not contain any studies with human
participants or animals performed by any of the authors.
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as
long as you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons licence, and indicate if
changes were made. The images or other third party material in this article
are included in the article's Creative Commons licence, unless indicated
otherwise in a credit line to the material. If material is not included in the
article's Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will
need to obtain permission directly from the copyright holder. To view a
copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Glycoconj J
15. Campo, V.L., Marchiori, M.F., Rodrigues, L.C., Dias-Baruffi, M.:
Synthetic glycoconjugates inhibitors of tumor-related galectin-3: an
update. Glycoconjugate J. 33(6), 853–876 (2016). https://doi.org/
10.1007/s10719-016-9721-z
16. Laaf, D., Bojarova, P., Elling, L., Kren, V.: Galectin-Carbohydrate
interactions in biomedicine and biotechnology. Trends Biotechnol.
37(4), 402–415 (2019). https://doi.org/10.1016/j.tibtech.2018.10.
001
References
1. Ju, T., Otto, V.I., Cummings, R.D.: The Tn Antigen—structural
simplicity and biological complexity. Angew. Chem. Int. Ed.
50(8), 1770–1791 (2011). https://doi.org/10.1002/anie.201002313
2. Yu, L.G.: The oncofetal Thomsen-Friedenreich carbohydrate anti-
gen in cancer progression. Glycoconjugate J. 24(8), 411–420
(2007). https://doi.org/10.1007/s10719-007-9034-3
17. Bumba, L., Laaf, D., Spiwok, V., Elling, L., Křen, V., Bojarová, P.:
Poly-N-Acetyllactosamine neo-glycoproteins as nanomolar ligands
of human galectin-3: Binding kinetics and modeling. Int. J. Mol.
Sci. 19(2), 372 (2018). https://doi.org/10.3390/ijms19020372
18. Laaf, D., Bojarová, P., Pelantová, H., Křen, V., Elling, L.: Tailored
multivalent neo-glycoproteins: synthesis, evaluation, and applica-
tion of a library of galectin-3-binding glycan ligands. Bioconj.
Chem. 28(11), 2832–2840 (2017). https://doi.org/10.1021/acs.
bioconjchem.7b00520
19. Zhang, H., Laaf, D., Elling, L., Pieters, R.J.: Thiodigalactoside-
bovine serum albumin conjugates as high-potency inhibitors of
galectin-3: an outstanding example of multivalent presentation of
small molecule inhibitors. Bioconj. Chem. 29(4), 1266–1275
(2018). https://doi.org/10.1021/acs.bioconjchem.8b00047
20. Rodriguez, M.C., Yegorova, S., Pitteloud, J.-P., Chavaroche, A.E.,
André, S., Ardá, A., Minond, D., Jiménez-Barbero, J., Gabius, H.-
J., Cudic, M.: Thermodynamic switch in binding of adhesion/
growth regulatory human galectin-3 to tumor-associated TF
Antigen (CD176) and MUC1 glycopeptides. Biochemistry.
54(29), 4462–4474 (2015). https://doi.org/10.1021/acs.biochem.
5b00555
21. Santarsia, S., Grosso, A.S., Trovão, F., Jiménez-Barbero, J.,
Carvalho, A.L., Nativi, C., Marcelo, F.: Molecular recognition of
a Thomsen–Friedenreich antigen mimetic targeting human
galectin-3. Chem Med Chem. 13(19), 2030–2036 (s). https://doi.
org/10.1002/cmdc.201800525
22. Guha, P., Kaptan, E., Bandyopadhyaya, G., Kaczanowska, S.,
Davila, E., Thompson, K., Martin, S.S., Kalvakolanu, D.V., Vasta,
G.R., Ahmed, H.: Cod glycopeptide with picomolar affinity to
galectin-3 suppresses T-cell apoptosis and prostate cancer metasta-
sis. Proc. Natl. Acad. Sci. U. S. A. 110(13), 5052–5057 (2013).
https://doi.org/10.1073/pnas.1202653110
3. Almogren, A., Abdullah, J., Ghapure, K., Ferguson, K., Glinsky,
V.V., Rittenhouse-Olson, K.: Anti-Thomsen-Friedenreich-Ag (anti-
TF-Ag) potential for cancer therapy. Front. Biosci., Scholar Ed. 4,
840–863 (2012). https://doi.org/10.2741/s304
4. Byrd, J.C., Bresalier, R.S.: Mucins and mucin binding proteins in
colorectal cancer. Cancer Metastasis Rev. 23(1/2), 77–99 (2004).
https://doi.org/10.1023/a:1025815113599
5. Jeschke, U., Karsten, U., Wiest, I., Schulze, S., Kuhn, C., Friese, K.,
Walzel, H.: Binding of galectin-1 (gal-1) to the Thomsen-
Friedenreich (TF) antigen on trophoblast cells and inhibition of
proliferation of trophoblast tumor cells in vitro by gal-1 or an
anti-TF antibody. Histochem. Cell Biol. 126(4), 437–444 (2006).
https://doi.org/10.1007/s00418-006-0178-1
6. Glinsky, V.V., Glinsky, G.V., Rittenhouse-Olson, K., Huflejt, M.E.,
Glinskii, O.V., Deutscher, S.L., Quinn, T.P.: The role of Thomsen-
Friedenreich antigen in adhesion of human breast and prostate can-
cer cells to the endothelium. Cancer Res. 61(12), 4851–4857 (2001)
7. Glinsky, V.V., Huflejt, M.E., Glinsky, G.V., Deutscher, S.L., Quinn,
T.P.: Effects of Thomsen-Friedenreich antigen-specific peptide P-
30 on beta-galactoside-mediated homotypic aggregation and adhe-
sion to the endothelium of MDA-MB-435 human breast carcinoma
cells. Cancer Res. 60(10), 2584–2588 (2000)
8. Bian, C.-F., Zhang, Y., Sun, H., Li, D.-F., Wang, D.-C.: Structural
basis for distinct binding properties of the human galectins to
Thomsen-Friedenreich antigen. PLoS One 6(9), e25007 (2011).
doi: 10.1371/journal.pone.0025007
9. Khaldoyanidi, S.K., Glinsky, V.V., Sikora, L., Glinskii, A.B.,
Mossine, V.V., Quinn, T.P., Glinsky, G.V., Sriramarao, P.: MDA-
MB-435 human breast carcinoma cell homo- and heterotypic adhe-
sion under flow conditions is mediated in part by Thomsen-
Friedenreich antigen-galectin-3 interactions. J. Biol. Chem.
278(6), 4127–4134 (2003). doi: 10.1074/jbc.M209590200
10. Zhao, Q.C., Guo, X.L., Nash, G.B., Stone, P.C., Hilkens, J.,
Rhodes, J.M., Yu, L.G.: Circulating galectin-3 promotes metastasis
by modifying MUC1 localization on cancer cell surface. Cancer
Res. 69(17), 6799–6806 (2009). https://doi.org/10.1158/0008-
5472.can-09-1096
23. Henze, M., Schmidtke, S., Hoffmann, N., Steffens, H., Pietruszka,
J., Elling, L.: Combination of glycosyltransferases and a
glycosynthase in sequential and one-pot reactions for the synthesis
of type 1 and type 2 N-acetyllactosamine oligomers. Chem Cat
Chem. 7(19), 3131–3139 (2015). https://doi.org/10.1002/cctc.
201500645
11. Zhao, Q.C., Barclay, M., Hilkens, J., Guo, X.L., Barrow, H.,
Rhodes, J.M., Yu, L.G.: Interaction between circulating galectin-3
and cancer-associated MUC1 enhances tumour cell homotypic ag-
gregation and prevents anoikis. Mol Cancer. 9(1), 154 (2010).
https://doi.org/10.1186/1476-4598-9-154
12. Oberg, C.T., Leffler, H., Nilsson, U.J.: Inhibition of galectins with
small molecules. Chimia 65(1–2), 18–23 (2011). doi: https://doi.
org/10.2533/chimia.2011.18
13. Delaine, T., Collins, P., MacKinnon, A., Sharma, G., Stegmayr, J.,
Rajput, V.K., Mandal, S., Cumpstey, I., Larumbe, A., Salameh,
B.A., Kahl-Knutsson, B., van Hattum, H., van Scherpenzeel, M.,
Pieters, R.J., Sethi, T., Schambye, H., Oredsson, S., Leffler, H.,
Blanchard, H., Nilsson, U.J.: Galectin-3-binding glycomimetics
that strongly reduce bleomycin-induced lung fibrosis and modulate
intracellular glycan recognition. ChemBioChem. 17(18), 1759–
1770 (2016). https://doi.org/10.1002/cbic.201600285
24. Henze, M., You, D.-J., Kamerke, C., Hoffmann, N., Angkawidjaja,
C., Ernst, S., Pietruszka, J., Kanaya, S., Elling, L.: Rational design
of a glycosynthase by the crystal structure of β-galactosidase from
Bacillus circulans (BgaC) and its use for the synthesis of N-
acetyllactosamine type 1 glycan structures. J. Biotechnol. 191(0),
78–85 (2014). https://doi.org/10.1016/j.jbiotec.2014.07.003
25. Kamerke, C., Pattky, M., Huhn, C., Elling, L.: Synthesis of
nucleotide-activated disaccharides with recombinant β3-
galactosidase C from Bacillus circulans. J. Mol. Catal. B: Enzym.
89, 73–81 (2013). https://doi.org/10.1016/j.molcatb.2012.12.014
26. Steinmann, A., Thimm, J., Matwiejuk, M., Thiem, J.: Formation of
homooligosaccharides using base-promoted glycosylation of un-
protected glycosyl fluorides. Macromolecules. 43(8), 3606–3612
(2010). https://doi.org/10.1021/ma100191d
27. Chen, R.F.: Removal of fatty acids from serum albumin by charcoal
treatment. J. Biol. Chem. 242(2), 173–181 (1967)
14. Cecioni, S., Imberty, A., Vidal, S.: Glycomimetics versus multiva-
lent glycoconjugates for the design of high affinity lectin ligands.
Chem. Rev. 115(1), 525–561 (2015). https://doi.org/10.1021/
cr500303t
28. Böcker, S., Laaf, D., Elling, L.: Galectin binding to neo-glycopro-
teins: LacDiNAc conjugated BSA as ligand for human galectin-3.
Biomolecules. 5(3), 1671–1696 (2015). https://doi.org/10.3390/
biom5031671

Glycoconj J
29. Hong, V., Presolski, S.I., Ma, C., Finn, M.G.: Analysis and optimi-
zation of copper-catalyzed azide-alkyne cycloaddition for
bioconjugation. Angew. Chem. Int. Ed. 48(52), 9879–9883
(2009). https://doi.org/10.1002/anie.200905087
30. Presolski, S.I., Hong, V.P., Finn, M.: Copper-catalyzed azide–
alkyne click chemistry for bioconjugation. Curr. Protoc. Chem.
Biol. 3(4), 153–162 (2011). https://doi.org/10.1002/
9780470559277.ch110148
31. Li, C., Kim, Y.-W.: Characterization of a galactosynthase derived
from Bacillus circulans β-Galactosidase: facile synthesis of D-
Lacto- and D-Galacto-N-bioside. Chem Bio Chem. 15, 522–526
(2014). https://doi.org/10.1002/cbic.201300699
32. Hayes, M., Pietruszka, J.: Synthesis of glycosides by
glycosynthases. Molecules. 22(9), 1434 (2017). https://doi.org/
10.3390/molecules22091434
33. Berman, H.M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T.N.,
Weissig, H., Shindyalov, I.N., Bourne, P.E.: The Protein Data Bank.
Nucleic Acids Res. 28(1), 235–242 (2000). https://doi.org/10.1093/
nar/28.1.235
34. Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S.,
Greenblatt, D.M., Meng, E.C., Ferrin, T.E.: UCSF Chimera—A
visualization system for exploratory research and analysis. J.
Comput. Chem. 25(13), 1605–1612 (2004). https://doi.org/10.
1002/jcc.20084
35. Naundorf, A., Caussette, M., Ajisaka, K.: Characterization of the
immobilized β-galactosidase C from Bacillus circulans and the
production of β(1,3)-linked disaccharides. Biosci., Biotechnol.,
Biochem. 62(7), 1313–1317 (1998). https://doi.org/10.1271/bbb.
62.1313
36. Kröger, L., Scudlo, A., Thiem, J.: Subsequent enzymatic
galactosylation and sialylation towards sialylated Thomsen–
Friedenreich antigen components. Adv. Synth. Catal. 348(10–11),
1217–1227 (2006). https://doi.org/10.1002/adsc.200606063
37. Cobucci-Ponzano, B., Moracci, M.: Glycosynthases as tools for the
production of glycan analogs of natural products. Nat. Prod. Rep.
29(6), 697–709 (2012). https://doi.org/10.1039/c2np20032e
38. Blanchard, J.E., Gal, L., He, S., Foisy, J., Warren, R.A.J., Withers,
S.G.: The identification of the catalytic nucleophiles of two β-
galactosidases from glycoside hydrolase family 35. Carbohydr.
Res. 333(1), 7–17 (2001). https://doi.org/10.1016/s0008-6215(01)
00108-2
propionate), DTSSP with peptides. J Am Soc Mass Spectr. 15(5), 736–
749 (2004). https://doi.org/10.1016/j.jasms.2004.01.011
42. Mädler, S., Bich, C., Touboul, D., Zenobi, R.: Chemical cross-
linking with NHS esters: a systematic study on amino acid reactiv-
ities. J. Mass Spectrom. 44(5), 694–706 (2009). https://doi.org/10.
1002/jms.1544
43. Kronman, M.J., Holmes, L.G., Robbins, F.M.: Inter- and intramo-
lecular interactions of α-Lactalbumin: X. effects of acylation of
tyrosyl and lysyl chains in molecular conformations. J. Biol.
Chem. 246(6), 1909–1921 (1971)
44. Wang, H., Huang, W., Orwenyo, J., Banerjee, A., Vasta, G.R.,
Wang, L.-X.: Design and synthesis of glycoprotein-based multiva-
lent glyco-ligands for influenza hemagglutinin and human galectin-
3. Biorg. Med. Chem. 21(7), 2037–2044 (2013). https://doi.org/10.
1016/j.bmc.2013.01.028
45. Lundquist, J.J., Toone, E.J.: The cluster glycoside effect. Chem.
Rev. 102(2), 555–578 (2002). https://doi.org/10.1021/cr000418f
46. Stowell, S.R., Arthur, C.M., Mehta, P., Slanina, K.A., Blixt, O.,
Leffler, H., Smith, D.F., Cummings, R.D.: Galectins-1, −2 and −3
exhibit differential recognition of sialylated glycans and blood
group antigens. J. Biol. Chem. 283(15), 10109–10123 (2008).
https://doi.org/10.1074/jbc.M709545200
47. Yongye, A.B., Calle, L., Arda, A., Jimenez-Barbero, J., Andre, S.,
Gabius, H.-J., Martinez-Mayorga, K., Cudic, M.: Molecular recog-
nition of the Thomsen-Friedenreich antigen-threonine conjugate by
adhesion/growth regulatory galectin-3: nuclear magnetic resonance
studies and molecular dynamics simulations. Biochemistry. 51(37),
7278–7289 (2012). https://doi.org/10.1021/bi300761s
48. Blanchard, H., Bum-Erdene, K., Hugo, M.W.: Inhibitors of
galectins and implications for structure-based design of galectin-
specific therapeutics. Aust. J. Chem. 67(12), 1763–1779 (2014).
https://doi.org/10.1071/CH14362
49. Collins, P.M., Bum-Erdene, K., Yu, X., Blanchard, H.: Galectin-3
interactions with glycosphingolipids. J. Mol. Biol. 426(7), 1439–
1451 (2014). https://doi.org/10.1016/j.jmb.2013.12.004
50. Diehl, C., Engström, O., Delaine, T., Hakansson, M., Genheden,
S., Modig, K., Leffler, H., Ryde, U., Nilsson, U.J., Akke, M.:
Protein flexibility and conformational entropy in ligand design
targeting the carbohydrate recognition domain of galectin-3. J.
Am. Chem. Soc. 132(41), 14577–14589 (2011). https://doi.org/
10.1021/ja105852y0002-7863
51. Hsieh, T.-J., Lin, H.-Y., Tu, Z., Huang, B.-S., Wu, S.-C., Lin, C.-H.:
Structural basis underlying the binding preference of human
galectins-1, −3 and −7 for Galβ1–3/4GlcNAc. PLoS One. 10(5),
e0125946 (2015). https://doi.org/10.1371/journal.pone.0125946
39. Cayot, P., Tainturier, G.: The Quantification of protein amino
groups by the trinitrobenzenesulfonic acid method: a reexamina-
tion. Anal. Biochem. 249(2), 184–200 (1997). https://doi.org/10.
1006/abio.1997.2161
40. Kalkhof, S., Sinz, A.: Chances and pitfalls of chemical cross-
linking with amine-reactive N-hydroxysuccinimide esters. Anal.
Bioanal. Chem. 392(1–2), 305–312 (2008). https://doi.org/10.
1007/s00216-008-2231-5
Publisher’s note Springer Nature remains neutral with regard to jurisdic-
tional claims in published maps and institutional affiliations.
41. Swaim, C.L., Smith, J.B., Smith, D.L.: Unexpected products from the
reaction of the synthetic cross-linker 3,3 '-dithiobis(sulfosuccinimidyl