Selective β-N-acetylhexosaminidase from Aspergillus versicolor—a tool for producing bioactive carbohydrates

Article abstract of DOI:10.1007/s00253-018-9534-z

β-N-Acetylhexosaminidases (EC 3.2.1.52) are typical of their dual activity encompassing both N-acetylglucosamine and N-acetylgalactosamine substrates. Here we present the isolation and characterization of a selective β-N-acetylhexosaminidase from the fung

Full text of DOI:10.1007/s00253-018-9534-z

Applied Microbiology and Biotechnology
https://doi.org/10.1007/s00253-018-9534-z
BIOTECHNOLOGICALLY RELEVANT ENZYMES AND PROTEINS
Selective β-N-acetylhexosaminidase from Aspergillus versicolor—a tool
for producing bioactive carbohydrates
Pavla Bojarová1 & Natallia Kulik2 & Kristýna Slámová1 & Martin Hubálek3 & Michael Kotik1 & Josef Cvačka3
Helena Pelantová1 & Vladimír Křen1
&
Received: 7 September 2018 /Revised: 5 November 2018 /Accepted: 17 November 2018
#
Springer-Verlag GmbH Germany, part of Springer Nature 2019
Abstract
β-N-Acetylhexosaminidases (EC 3.2.1.52) are typical of their dual activity encompassing both N-acetylglucosamine and N-
acetylgalactosamine substrates. Here we present the isolation and characterization of a selective β-N-acetylhexosaminidase from
the fungal strain of Aspergillus versicolor. The enzyme was recombinantly expressed in Pichia pastoris KM71H in a high yield
and purified in a single step using anion-exchange chromatography. Homologous molecular modeling of this enzyme identified
crucial differences in the enzyme active site that may be responsible for its high selectivity for N-acetylglucosamine substrates
compared to fungal β-N-acetylhexosaminidases from other sources. The enzyme was used in a sequential reaction together with a
mutant β-N-acetylhexosaminidase from Talaromyces flavus with an enhanced synthetic capability, affording a bioactive disac-
charide bearing an azido functional group. The azido function enabled an elegant multivalent presentation of this disaccharide on
an aromatic carrier. The resulting model glycoconjugate is applicable as a selective ligand of galectin-3 — a biomedically
attractive human lectin. These results highlight the importance of a general availability of robust and well-defined carbohy-
drate-active enzymes with tailored catalytic properties for biotechnological and biomedical applications.
.
.
.
.
.
Keywords Aspergillus versicolor β-N-Acetylhexosaminidase Glycosidase Homology modeling Heterologous expression
Pichia pastoris
Introduction
attractive ligands for targeting lectins — ubiquitous
carbohydrate-binding proteins that read the glycocode on sur-
faces of living cells. As a result, synthetic conjugates decorat-
ed with specific N-acetylhexosamine ligands (Bojarová et al.
2017; Laaf et al. 2017a, b) may serve as potent agents in
biomedical or biotechnological applications.
Oligosaccharides composed of N-acetylglucosamine and N-
acetylgalactosamine units are bioactive molecules of abundant
biological activities (Slámová and Bojarová 2017). They are
Glycosidases (O-glycoside hydrolases; EC 3.2.1.-) are des-
tined to the in vivo cleavage of oligo- and polysaccharides by
glycosyl transfer to water. They can form a glycosidic linkage if
a hydroxyl moiety of a carbohydrate acceptor acts as a more
efficient nucleophile than water. Such conditions can be
achieved by a variety of strategies including reduction of water
activity and use of glycosyl donors activated by a good leaving
group. Through a kinetically controlled transglycosylation in an
optimized setup, close-to-quantitative yields may be accom-
plished (Fialová et al. 2004). Enzymatic synthesis catalyzed
by glycosidases is a biocompatible elegant alternative to the
classical organic chemistry methods, which are burdened by a
number of protection and deprotection steps. It offers absolute
stereoselectivity, high regioselectivity, and glycosidic bond for-
mation in one step. In contrast to glycosyltransferases (EC 2.4)
Electronic supplementary material The online version of this article
(https://doi.org/10.1007/s00253-018-9534-z) contains supplementary
material, which is available to authorized users.
* Pavla Bojarová
bojarova@biomed.cas.cz
1
Laboratory of Biotransformation, Institute of Microbiology, Czech
Academy of Sciences, Vídeňská 1083, CZ-14220 Prague
4, Czech Republic
2
Laboratory of Structure and Function of Proteins, Institute of
Microbiology, Czech Academy of Sciences, Zámek 136,
CZ-37333 Nové Hrady, Czech Republic
3
Institute of Organic Chemistry and Biochemistry, Czech Academy of
Sciences, Flemingovo náměstí 2, CZ-16610 Prague
6, Czech Republic

Appl Microbiol Biotechnol
(Bojarová et al. 2013), glycosidases tolerate the presence of a
wide choice of functional groups in the substrate molecule
(Bojarová et al. 2008; Bojarová et al. 2011), and thus enable
to efficiently prepare tailored oligosaccharides as a part of a
chemoenzymatic synthetic pathway towards, e.g., bioactive
glycomimetics.
Materials and methods
Analytical methods
MS analysis of disaccharide 3 and conjugate 8
β-N-Acetylhexosaminidases (EC 3.2.1.52) are exo-
glycosidases (glycoside hydrolase family GH20, http://www.
cazy.org/) with a unique dual activity for cleaving both N-
acetylglucosamine (GlcNAc) and N-acetylgalactosamine
(GalNAc) carbohydrate moieties in a variety of substrates,
such as chitooligomers, glycosphingolipids, glycoproteins,
and other glycoconjugates (Slámová et al. 2010). Although
GlcNAc and GalNAc units differ only in the equatorial versus
axial hydroxyl at C-4, their chemical behavior and biological
implications are vastly different. The dual activity of β-N-
acetylhexosaminidases may be an obstacle in a selective
synthesis of defined N-acetylhexosamine oligosaccha-
rides, especially those carrying a combination of
GlcNAc and GalNAc units. Though there is a separate
family of O-GlcNAcases (GH84) (Gloster and Vocadlo
2010), these enzymes are specific for the cleavage of O-
GlcNAc residues from proteins and their affinity to ol-
igosaccharide substrates is very low. Moreover, these enzymes
are not applicable for synthetic purposes as they do not exhibit
any transglycosylation capabilities. Although specific glyco-
syltransferases exist for a selective introduction of GlcNAc
moieties (Slámová and Bojarová 2017), these enzymes lack
the substrate promiscuity and synthetic versatility of es-
pecially fungal glycosidases, which has been demon-
strated in numerous previous applications (Bojarová
and Křen 2009). In this article, we present the isolation
and characterization of a selective β-N-acetylhexosaminidase
from Aspergillus versicolor (AvHex). The enzyme was
recombinantly expressed in Pichia pastoris in a high yield
and purified by one-step anion-exchange chromatography.
The homology molecular model of this enzyme identified
the Gln315 amino acid residue in the active site as potentially
responsible for the enzyme high selectivity for GlcNAc sub-
strates compared to fungal β-N-acetylhexosaminidases from
other sources. The enzyme was used in a high-yielding
sequential reaction together with the transglycosylating
Tyr470His mutant of the β-N-acetylhexosaminidase
from Talaromyces flavus to yield a bioactive disaccha-
ride containing an azido functional group at C1. This
azido moiety enabled its multivalent presentation on a
simple aromatic carrier using click chemistry. The
resulting glycoconjugate is a selective ligand of
galectin-3, a human lectin with a high biomedical po-
tential. This study demonstrates the potential of robust
and selective carbohydrate-active enzymes with tailored
catalytic capabilities in biotechnological and biomedical
applications.
The electrospray mass spectra of disaccharide 3 and conjugate
8 were measured in the negative ion mode using the LTQ
Orbitrap XL hybrid mass spectrometer (Thermo Fisher
Scientific, Waltham, MA, USA). The mobile phase consisted
of methanol/water (4/1, v/v) at a flow rate of 100 μL/min. The
sample dissolved in methanol was injected into the mobile
phase flow using a 2-μL loop. Spray voltage, capillary volt-
age, tube lens voltage, and capillary temperature were 4.6 kV,
− 25 V, − 125 V, and 275 °C, respectively. The mass spectra
were internally calibrated using deprotonated palmitic acid as
lock mass.
MS analysis of compound 7
The EI spectrum of compound 7 was measured using orthog-
onal acceleration time-of-flight mass spectrometer GCT
Premier (Waters, Milford, MA, USA). The sample was dis-
solved in chloroform, loaded into a quartz cup of the direct
probe, and inserted into the ion source. The source tempera-
ture was 220 °C. Standard 70-eV spectra were recorded in the
positive ion mode. For exact mass measurement, the spectrum
was internally calibrated using perfluorotri-n-butylamine
(heptacosa).
NMR analysis
NMR analysis was performed on Bruker AVANCE III
700 MHz (disaccharide 3, compound 7) and 400 MHz (conju-
gate 8) spectrometers (Bruker BioSpin, Rheinstetten, Germany)
in D2O at 30 °C (3 and 8) or CDCl3 at 20 °C (7). Residual
signals of solvents (CDCl3: δH 7.263 ppm, δC 77.01 ppm; D2O:
δH 4.732 ppm) served as internal standards; the carbon spectra
in D2O were referenced to the signal of acetone (δc 30.50).
NMR experiments were acquired using standard manufac-
turers’ software TopSpin 3.5. Two-parameter double-exponen-
tial Lorentz-Gauss function was applied for 1H to improve res-
olution and line broadening (1 Hz) was applied to get better 13
C
signal-to-noise ratio. Individual spin systems in compounds
under study were assigned by COSYor 1d-TOCSY; the assign-
ment was transferred to carbons by HSQC. The HMBC exper-
iment enabled to assign quaternary carbons and join particular
substructures together.
MS analysis of intact protein
The purified AvHex protein in McIlvaine buffer pH 5.0 was
diluted 1:1 with matrix (saturated solution of synapinic acid

Appl Microbiol Biotechnol
dissolved in 50% acetonitrile/0.1% trifluoracetic acid/water)
and spotted onto the MALDI target. The linear time-of-
flight spectrum tuned to a mass range 20–50 kDa was
recorded on a MALDI TOF instrument (Ultraflextreme,
Bruker, Bremen, Germany).
synthetically (Generay, Shanghai, China) and cloned into the
yeast expression vector pPICZαA (Invitrogen, Life
Technologies, Carlsbad, CA, USA) downstream of the α-
factor-encoding DNA segment for extracellular protein
targeting and zeocin resistance using EcoRI and KpnI restric-
tion sites. The predicted signal propeptide sequence (residues
1–19) was not included in the cloned DNA segment (1770 base
pairs), whose sequence was codon-optimized for the produc-
tion in P. pastoris using OptigeneTM Codon Optimization
Analysis platform (Generay, Shanghai, China). It was deposited
in GenBank under the accession number MH937562. The con-
struct was linearized (15 μg) by restriction endonuclease SacI
(New England Biolabs, Ipswich, MA, USA), purified from an
agarose gel using a JetQuick Gel Extraction Spin Kit
(Genomed, Löhne, Germany) and used for electroporation of
P. pastoris KM71H competent cells (Invitrogen, Life
Technologies, Carlsbad, CA, USA) according to the manufac-
turer´s protocol (EasySelect Pichia Expression Kit; Invitrogen,
Life Technologies, Carlsbad, CA, USA).
Production of wild-type AvHex in A. versicolor
The β-N-acetylhexosaminidase producing fungal strain of
A. versicolor CCF 2491 (wild-type AvHex) originated from
the Culture Collection of Fungi (CCF), Department of Botany,
Charles University in Prague (CZ) (Weignerová et al. 2003).
The cultures were maintained on slants (g/L): mycological
agar, 20.0; bacto-peptone, 5.0; and malt extract, 35.0. The
peptone cultivation medium (M1) (Huňková et al. 1996)
contained the following (g/L): yeast extract, 0.5; mycological
peptone, 5.0; KH2PO4, 3.0; NH4H2PO4, 5.0; casein acid hy-
drolyzate, 7.5; pH was adjusted to 6.0. The peptone cultiva-
tion medium induced with chitooligomers (M2) contained the
following (g/L): yeast extract, 0.5; mycological peptone, 5.0;
KH2PO4, 3.0; NH4H2PO4, 5.0; chitin hydrolyzate, 7.5; pH
was adjusted to 6.0. The mineral cultivation medium induced
with GlcNAc (M3) (Slámová et al. 2012) contained the fol-
lowing (g/L): yeast extract, 0.5; KH2PO4, 3.0; NH4H2PO4,
5.0; (NH4)2SO4, 2.0; NaCl, 15.0; GlcNAc, 5.0 g/l); pH was
adjusted to 4.0. After sterilization, sterile solution of MgSO4·
7H2O (10% w/v) was added. Erlenmeyer flasks (500 mL) with
medium (100 mL) were inoculated with 0.5 mL of a spore
suspension in 0.1% (v/v) Tween 80 solution and cultivated on
a rotary shaker at 200 rpm and 28 °C for 11–12 days.
The resulting colonies were tested for AvHex production as
described below. Two colonies with the highest production of
AvHex in the highest purity as determined by SDS-PAGE
electrophoresis and activity measurements were cryopre-
served at − 80 °C in 15% (v/v) glycerol (OD600 = 50–100).
Production of recombinant AvHex in P. pastoris
Recombinant AvHex was produced essentially according to
the manufacturer’s instructions (EasySelect Pichia
Expression Kit; Invitrogen, Life Technologies), which also
shows the exact composition of cultivation media. For a
small-scale production of the recombinant enzyme, the obtain-
ed transformants were inoculated into 100 mL of BMGY me-
dium (Buffered Glycerol complex Medium) and incubated at
28 °C and 220 rpm overnight. Then the cells were collected by
centrifugation (5000 rpm, 10 min, 20 °C) and the pellet was
resuspended in 30 mL of BMMY medium (Buffered
Methanol complex Medium). To induce expression of
AvHex, methanol (0.5% v/v) was added every 24 h to the
cultures. The flasks were incubated on a rotary shaker at
28 °C and 220 rpm for 3 days. For the preparative production
of recombinant AvHex, 15 mL of YPD medium was inoculat-
ed with 100 μL of cryopreserved culture and incubated for 5 h
at 28 °C and 220 rpm. Then, 1 L of BMGH medium (Buffered
Minimal Glycerol medium) was inoculated with this
preculture and incubated overnight under the same conditions.
The cells were then collected and resuspended in 200 mL of
BMMH medium (Buffered Minimal Methanol medium) and
incubated at 28 °C on a rotary shaker with methanol addition
(0.5% v/v) every 24 h for three days.
The enzyme was purified from cultures grown in the
GlcNAc-induced mineral medium (M3). After 12 days of
cultivation, the mycelium was filtered off, and the medi-
um was dialyzed against 10 mM citrate-phosphate buffer
pH 3.5 for 4 h. Then, the medium (pH 3.5) was loaded
on a cation-exchange chromatography column (Fractogel
−
EMD SO3 ; Merck, Darmstadt, Germany) equilibrated
with 10 mM sodium citrate-phosphate buffer pH 3.5,
using an Äkta Purifier protein chromatography system
(Amersham Biosciences, Uppsala, Sweden). The enzyme
was eluted with a linear gradient (80 mL, 2 mL/min) of
0–1 M NaCl. The pure fractions (as verified by SDS-
PAGE) containing β-N-acetylhexosaminidase activity
were combined and concentrated by ultrafiltration in
McIlvaine buffer (50 mM citric acid/100 mM
Na2HPO4) at pH 5.0.
Gene cloning and expression
The putative gene of the β-N-acetylhexosaminidase-encoding
DNA (AvHex; corresponding to the hypothetical protein
ASPVEDRAFT_125959 (A. versicolor CBS 583.65,
GenBank: OJI97839.1; 806 amino acid residues) was prepared
Recombinant AvHex was purified from the culture super-
natant as follows. The biomass was removed by centrifugation
(5000 rpm, 10 min, 12 °C), the resulting supernatant was

Appl Microbiol Biotechnol
diluted four times with water, and its pH was adjusted to 6.5.
The enzyme was purified in a single step by anion-exchange
chromatography (Q-Sepharose Fast Flow, GE Healthcare,
Pittsburgh, PA, USA) on an Äkta Purifier protein chromatog-
raphy system (Amersham Biosciences, Uppsala, Sweden)
equilibrated with 10 mM sodium citrate-phosphate buffer pH
6.5. The enzyme was eluted with a linear gradient (60 mL,
2 mL/min) of 0–1 M NaCl. The pure fractions (as verified by
SDS-PAGE) containing β-N-acetylhexosaminidase activity
were combined and concentrated by ultrafiltration in
McIlvaine buffer (50 mM citric acid/100 mM Na2HPO4) at
pH 5.0.
in 7% polyacrylamide gel using 0.02 M HEPES/0.18 M gly-
cine buffer pH 8.0.
Protein deglycosylation
The wild-type and recombinant AvHex enzymes were N-de-
glycosylated by endoglycosidase Endo H (New England
Biolabs, Ipswich, MA, USA), using the manufacturer’s proto-
col. The buffer of the enzyme sample (containing 115 μg of
protein) was exchanged for the supplier’s Glyco-buffer (final
volume 120 μL). Endo H (30 μL, 15,000 U) was added and
the reaction mixture was incubated at 37 °C and 300 rpm
overnight. The deglycosylation resulted in a shift the apparent
molecular weights of the native and N-deglycosylated en-
zymes on SDS-PAGE gel.
Sequence determination by LC-MS/MS
Proteins in solution or in gel bands were reduced by dithiothre-
itol and alkylated by iodoacetamide prior to overnight digestion
with trypsin or chymotrypsin in 50 mM ammonium acetate
buffer pH 8.5. Digestion with pepsin was carried out under
acidic conditions (0.04 M HCl) for 2 h. Peptides in solution
were desalted on a C18 spin column and dried; peptides in gel
bands were extracted by acetone and dried in Speed-Vac con-
centrator. Resulting peptides were analyzed on an UltiMate
3000 RSLCnano system (Thermo Scientific, Waltham, MA,
USA) coupled to a TripleTOF 5600 mass spectrometer with a
NanoSpray III source (Sciex, Darmstadt, Germany). The pep-
tides were trapped and desalted with 2% acetonitrile in 0.1%
formic acid at a flow rate of 5 μL/min using an Acclaim
PepMap100 column (5 μm, 2 cm × 100 μm ID, Thermo
Scientific, Waltham, MA, USA). Eluted peptides were separat-
ed on an Acclaim PepMap100 analytical column (3 μm,
25 cm × 75 μm ID, Thermo Scientific, Waltham, MA, USA).
The 70-min elution gradient with a constant flow of 300 nL/
min was set to 5% of phase B (0.1% formic acid in 99.9%
acetonitrile; phase A, 0.1% formic acid) for the first 10 min;
then, gradient elution took place with an increasing concentra-
tion of acetonitrile. The TOF-MS mass range was set to 350–
1250 m/z; in MS/MS mode, the instrument acquired fragmen-
tation spectra with m/z in a range of 100–1600. Protein Pilot 4.5
(Sciex, Darmstadt, Germany) was used for protein identifica-
tion using the Uniprot database of available Aspergillus species
protein sequences and the putative AvHex sequence (hypothet-
ical protein ASPVEDRAFT_125959, BLAST) (Altschul et al.
1990).
Standard assay for enzyme activity
The β-N-acetylglucosaminidase activity of AvHex was
assayed in end-point experiments using p-nitrophenyl 2-
acetamido-2-deoxy-β-D-glucopyranoside (1; pNP-GlcNAc;
Sigma-Aldrich, St. Louis, MO, USA) as a substrate with a
starting concentration of 2 mM. After incubation of the reac-
tion mixture in McIlvaine buffer (50 mM citric acid/100 mM
Na2HPO4 pH 5.0) at 35 °C and 850 rpm for 10 min, the
liberated p-nitrophenol was determined spectrophotometrical-
ly at 420 nm under alkaline conditions (50 μL of the reaction
mixture was mixed with 1 mL of 0.1 M Na2CO3). One unit of
enzymatic activity was defined as the amount of enzyme re-
leasing 1 μmol of p-nitrophenol per minute under the above
conditions.
The effect of pH on AvHex activity was measured at 35 °C
as described above; Britton-Robinson buffer (0.04 M H3PO4,
0.04 M phenylacetic acid, 0.04 M H3BO3/0.2 M NaOH) was
used for the pH range of 2–10 with the respective blanks void
of enzyme. The temperature activity profile of AvHex was
determined as described above in McIlvaine buffer pH 5.0,
in the range of 25–80 °C.
Michaelis–Menten parameters of hydrolysis of pNP-
GlcNAc by AvHex were determined in a discontinuous kinetic
assay using pNP-GlcNAc as substrate at a concentration of
0.1–4.0 mM in McIlvaine buffer pH 5.0. At each substrate
concentration, the reaction mixture (400 μL) was incubated
at 25 °C and 850 rpm. The hydrolytic reaction lasted for ca
15 min. Aliquots (50 μL) were taken at regular time intervals
and added to 100 μL of 1 M Na2CO3 in the wells of a styrene
flat-bottom 96-well microtiter plate (Thermo Scientific,
Waltham, MA, USA). The absorbance at 420 nm was mea-
sured using the Spectra Max Plus spectrophotometer
(Molecular Devices, Sunnyvale, CA, USA). All data were
measured in duplicates. The initial reaction velocity for each
substrate concentration was determined by linear regression of
respective time points, and the kinetic constants (Km, kcat)
Protein characterization
Protein concentration was determined by the Bradford method
(Bradford 1976) using Protein Assay Dye Reagent Concentrate
(Bio-Rad, Watford, Hertfordshire, UK) calibrated for bovine
plasma γ-globulin (IgG, BioRad, Watford, Hertfordshire,
UK). The purity of AvHex was verified by SDS-PAGE in
10% polyacrylamide gel. Native electrophoresis was performed

Appl Microbiol Biotechnol
were determined from the initial velocities for each substrate
concentration using GraphPad Prism 7 (GraphPad Software,
La Jolla, CA, USA).
Sequential enzymatic synthesis
of 2-acetamido-2-deoxy-β-D-galactopyranosyl-(1 → 4)
-2-acetamido-2-deoxy-β-D-glucopyranosyl azide (3)
Homology modeling, docking, and molecular dynamics
For the first synthetic step of the sequential reaction, we used
the Tyr470His mutant of the β-N-acetylhexosaminidase
from T. flavus (TfHex) with an enhanced synthetic ca-
pability. The enzyme was prepared as described previ-
ously (Slámová et al. 2015), by extracellular expression in
P. pastoris KM71H (Invitrogen, Life Technologies,
Carlsbad, CA, USA) followed by a single purification step
of cation-exchange chromatography.
We used the sequence of the hypothetical protein
ASPVEDRAFT_125959 (GenBank database) in a BLAST
search (Altschul et al. 1990) to identify protein homologs.
Homology modeling was performed with Modeller 9.16 using
the crystal structure of the Aspergillus oryzae β-N-
acetylhexosaminidase (AoHex) as a template (Sali and
Blundell 1993). Sequence alignment was constructed with T-
coffee (Notredame et al. 2000) and manually modified. We
employed several models for refinement (Raval et al. 2012)
with YASARA (Krieger et al. 2002) using the YASARA force
field. The most stable model was obtained from the refinement
of the glycosylated dimeric model with pNP-GlcNAc substrate
in the active site using constrained molecular dynamics simu-
lation followed by a short (1 ns) free simulation of the system.
Molprobity was used for validation of model geometry: 95.8%
(548/572) residues of the best model were in favored areas of
the Ramachandran plot and just 3 of 4375 covalent bonds had
unfavorable geometry; all clashes were resolved (Davis et al.
2007). The averaged knowledge-based energy residues and Z-
score (− 11.07 for 572 aa) estimated by ProSA were within
favorable values (Wiederstein and Sippl 2007).
pNP-GalNAc (1; 41 mg, 0.12 mmol), acceptor 2 (118 mg,
0.48 mmol), and the Tyr470His mutant of TfHex (1.4 U,
12.6 mg, 450 μL) were incubated in a mixture of acetonitrile
(10% v/v) and McIlvaine buffer pH 5.0 (total reaction volume
2.4 mL) at 45 °C and 1000 rpm. The reaction was monitored
by HPLC and TLC (propan-2-ol/water/NH4OH aq., 7:2:1).
After 6.5 h, the reaction was stopped by boiling for 2 min,
and, upon cooling down to 35 °C, the AvHex enzyme was
added (4.9 U, 0.12 mg, 30 μL) and the reaction was incubated
at 35 °C and 850 rpm for 7 h to cleave 2-acetamido-2-
deoxy-β-D-glucopyranosyl-(1 → 4)-2-acetamido-2-deoxy-β-
D-glucopyranosyl azide (4) that originated in the
transglycosylation reaction as an unwanted by-product. The
reaction was monitored as above and then stopped by boiling
for 2 min. After cooling down to room temperature, the reac-
tion mixture was centrifuged (13,500 rpm; 10 min) and loaded
onto a Biogel P-2 column (2.6 × 100 cm, Bio-Rad, Watford,
Hertfordshire, UK ) using water as a mobile phase at an elu-
tion rate of 7.5 mL/h. Pure acceptor 2 was partially recovered
(28 mg). The fractions containing the product were combined
and lyophilized; the title compound 3 was obtained as a white
solid (25 mg, 0.056 mmol, 47% isolated yield). HRMS (ESI−)
found m/z 448.16819 (expected 448.16852 for [M−H]−,
C16H26N5O10). For the respective NMR and MS data, see
the Supplementary Material (Table S1, Figs. S1, S2).
GlyProt (Bohne-Lang and von der Lieth 2005) and
NGlycPred server (Chuang et al. 2012) were used for predicting
the protein glycosylation, modeled in YASARA based on the
glycosylation of the AoHex structure (pdb: 5OAR).
The initial orientation of substrates, docking, molecular
dynamics simulation, and complex analysis was performed
with YASARA (Krieger et al. 2002; Kulik et al. 2015) using
YAMBER 2 force field. Carbohydrate parameters were used
from Glycam force field incorporated in YASARA; non-
standard residues were parameterized with AutoSMILES al-
gorithm (Jakalian et al. 2002). Representative binding scores
were estimated by Glide from Schrödinger package (Friesner
et al. 2006) using values averaged over substrate–enzyme
complexes. The Glide XP score (Friesner et al. 2006)
approximates the ligand binding free energy and in-
cludes an electrostatic term, terms for van der Waals
interactions, hydrogen bonds, π-stacking and π-cation
interactions, contributions of hydrophobic interactions, strain
energy (unfavorable contacts), and penalty upon desolvation
of the substrate or protein.
1,2,3-tris(Alynyloxy)benzene (7)
The title compound 7 was prepared according to a published
procedure, with some modifications (Xie et al. 2008). To a
solution of pyrogallol (6; 1.26 g, 10 mmol) in dry DMF
(dimethylformamide, 30 mL), anhydrous K2CO3 (5.52 g,
40 mmol) was added and the solution was heated at 50 °C
for 1 h under argon. 3-Bromo-1-propyne (4.31 mL, 40 mmol)
was added dropwise and the reaction ran for 2 days at 65 °C,
monitored by TLC (ether/hexane, 1:1). After cooling down,
the reaction mixture was poured into water (300 mL), then
chloroform (150 mL) was added, the phases were separated
and the water phase was extracted with chlorofom (3 ×
100 mL). The combined organic phases were concentrated
in vacuo and repeatedly coevaporated with water to remove
2-Acetamido-2-deoxy-β-D-glucopyranosyl azide (2)
The title compound 2 was prepared from GlcNAc based on
the procedures described previously (Bojarová et al. 2017).
The 1H and 13C NMR data were consistent with the structure.

Appl Microbiol Biotechnol
the traces of DMF. The crude compound 7 was purified by
column chromatography on silica gel using ether/hexane, 1:3,
as a mobile phase. The title compound 7 was isolated as a
yellowish solid (1.67 g, 7.0 mmol, 70% yield). HRMS (EI+)
found m/z 240.0785 (expected m/z 240.0786 for M+·,
C15H12O3). For the respective NMR and MS data, see the
Supplementary Material (Table S2, Figs. S3, S4).
containing Tween 20 (0.05% v/v) was performed. Detection of
bound galectin-3 was achieved using anti-His6-IgG1 antibody
from mouse 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 re-
action was stopped by adding 3 M hydrochloric acid (50 μL).
The binding signal of bound galectin-3 was quantified color-
imetrically at 450 nm using Sunrise absorbance microplate
reader (Tecan Group Ltd, Männedorf. Switzerland). The ob-
tained data were analyzed by non-linear regression (dose
response-inhibition-variable slope) using GraphPad Prism 7
(GraphPad software, San Diego, CA, USA).
1,2,3-tris-[1-(2-Acetamido-2-deoxy-β-D-galactopyranosyl-
(1 → 4)-2-acetamido-2-deoxy-β-D-glucopyranosyl)
-1H-1,2,3-triazol-4-yl)-2-ethyloxy]benzene (8)
Compound 7 (5 mg, 0.021 mmol), sodium ascorbate (4.1 mg,
0.021 mmol), and disaccharide 3 (29 mg, 0.065 mmol) were
dissolved in a mixture of water (300 μL) and methanol
(735 μL). The reaction mixture was bubbled with argon and
a solution of CuSO4·5H2O (5.2 mg, 0.021 mmol) in water
(60 μL) was added. The reaction mixture was bubbled with
argon again and left overnight at room temperature. After
complete conversion was detected by TLC (propan-2-ol/
H2O/NH4OH aq, 7:2:1; 3 times developed), PBS buffer con-
taining 5% w/v EDTA (700 μL) was added. After 30 min, the
reaction mixture was concentrated to one third of its volume to
remove methanol and loaded onto a Biogel P2 column (2.6 ×
100 cm, Bio-Rad, Watford, Hertfordshire, UK) with water as a
mobile phase, at an elution rate of 7.3 mL/h. The fractions
containing product 8 were combined and lyophilized. The title
compound 8 was isolated as a white solid (17.5 mg,
0.011 mmol, 53%). HRMS (ESI−) found m/z 1586.59778 (ex-
pected m/z 1586.59874 for [M−H]−; C63H92N15O33). For the
respective NMR and MS data, see the Supplementary
Material (Table S3, Figs. S5, S6).
Results
Wild-type AvHex production in A. versicolor
and purification
AvHex was identified as the most selective enzyme for N-
acetylglucosamine (GlcNAc) substrates in our library of β-
N-acetylhexosaminidases (˃ 100 enzymes) from filamentous
fungi on the basis of our previous work (Weignerová et al.
2003). Its GlcNAcase/GalNAcase activity ratio ranges be-
tween ca 9 and 11 depending on the particular production
batch. β-N-Acetylhexosaminidases are produced extracellu-
larly by their native hosts into the cultivation medium and
their production may be enhanced in the presence of various
inducers. We compared three cultivation media and identified
M3, mineral medium induced with submersed GlcNAc (5 g/
L) (Slámová et al. 2012), as optimum for wild-type enzyme
production (ca 0.4 U/ml; 11 U/mg protein). The time depen-
dence of the secretion profile was similar to other fungi
(Ryšlavá et al. 2011), with the maximum β-N-
acetylglucosaminidase activity found after 10–12 days of cul-
tivation at 28 °C (Fig. 1a).
The enzyme was purified to apparent homogeneity as de-
termined by SDS-PAGE in a single chromatography step —
cation-exchange chromatography at pH 3.5, preceded by di-
alysis against the loading chromatography buffer to reduce the
medium conductivity. The overall purification yield reached
28% and the specific activity of purified AvHex was 79 U/mg
protein (Table 1). In total, an average of 0.13 mg protein could
be isolated from 100 mL of cultivation medium. A small pro-
tein with a molecular weight of ca 14 kDa co-purified with the
enzyme could be found in heavily overloaded samples (Fig.
1b, lanes 1, 2). On the basis of the previous research on fungal
β-N-acetylhexosaminidases (Plíhal et al. 2007; Škerlová et al.
2018) and the sequence alignment (Fig. 2), it was assigned to
be the AvHex propeptide (aa 20–97). The size of the catalytic
domain was estimated to be 66 kDa using SDS-PAGE; its N-
Competitive ELISA assay
Recombinant galectin-3 used in the assay was prepared as
described previously (Bumba et al. 2018). The affinity of
galectin-3 for conjugate 8 was determined by competitive
ELISA assay essentially as reported previously (Laaf et al.
2017b; Bumba et al. 2018). Briefly, the F16 Maxisorp
NUNC-Immuno Modules (Thermo Scientific, Hvidovre,
Denmark) were coated overnight with asialofetuin (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-3 (total volume 50 μL; 6.5 μM
final galectin-3 concentration) in EPBS buffer (50 mM
NaH2PO4, 150 mM NaCl, 2 mM ethylendiaminetetraacetic
acid, pH 7.5) was added and incubated for 2 h. In a control
assay, this step was performed in the presence of 0.025% v/v
Tween. Extensive rinsing of wells after every step with PBS

Appl Microbiol Biotechnol
Fig. 1 a Time course of secretion of AvHex in various cultivation media
(M1–M3). The best production was achieved in M3 medium (yeast
extract, 0.5; KH2PO4, 3.0; NH4H2PO4, 5.0; (NH4)2SO4, 2.0; NaCl, 15.0;
GlcNAc, 5.0 g/L). b Purification and N-deglycosylation of AvHex as
analyzed by SDS-PAGE electrophoresis. Lane M, molecular mass markers
consisting of phosphorylase b from rabbit muscle (97 kDa), bovine serum
albumin (66 kDa), chicken egg white ovalbumin (45 kDa), carbonic
anhydrase from bovine erythrocytes (30 kDa), trypsin inhibitor from soya
beans (20.1 kDa), and α-lactalbumin from bovine milk (14.4 kDa); lane 1,
wild-type AvHex (66 kDa) with propeptide (14 kDa) purified by cation-
exchange chromatography; lane 2, wild-type AvHex after N-deglycosyla-
tion (58 kDa) with Endo H (29 kDa) and Endo H; lane 3, Endo H. The
position of the putative propeptide co-purifying with the catalytic subunit is
indicated with an arrow
glycosylation amounted to ca 8 kDa. This agrees well with the
previously published data on related enzymes (Plíhal et al.
2007; Ryšlavá et al. 2011; Slámová et al. 2012) and with the
theoretical calculated MW of the AvHex catalytic domain of
56,872.29 kDa. In native electrophoresis, the AvHex protein
appears as a single band of MW roughly 140 kDa but this is
just an approximate value influenced by protein shape and
charge distribution. This is in agreement with the native form
of this enzyme being a dimer with each monomer containing a
glycoside hydrolase family 20 catalytic domain (http://www.
cazy.org/) and a propeptide (a total of 160 kDa). The purified
and concentrated enzyme was stable for ca 6 months when
stored at 4 °C in McIlvaine buffer pH 5.0 conserved with 0.
03% (w/v) NaN3 (< 20% loss of activity).
2011), and TfHex (T. flavus, GenBank: AY091636; Kulik
et al. 2015) with 69, 66, and 61 identities over complete aligned
sequences, respectively. Additionally, high sequence identi-
ties were observed in fungal β-N-acetylglucosaminidases
(Table S4, Supplementary Material). In contrast, identi-
ties of less than 31% were determined in bacterial β-N-
acetylhexosaminidase-encoding sequences (Fig. 2).
As we can see from Fig. 2, active-site amino acid residues
in AvHex are conserved except for Gln315 located in the loop
HL1. This residue is quite frequently found as glutamate
in β-N-acetylhexosaminidases. In contrast, in the
GlcNAc-selective chitobiase from Serratia marcescens
(SmChit; 1QBB), this residue is present as glutamine
(Gln494). Gln is also found in a number of β-N-
acetylglucosaminidases and other chitinases as shown in a
multiple sequence alignment with fungal enzymes (Table S4,
Supplementary Material).
Structural alignment and sequencing of the wild-type
AvHex
In order to compare the putative amino acid sequence with
the sequence of the purified wild-type AvHex, peptide mapping
was performed using various proteases (pepsin, trypsin, chy-
motrypsin). The enzyme was sequenced in both the native and
deglycosylated forms, isolated from the SDS-PAGE gel or in
solution. The N-terminal sequence of 19 amino acids (Fig. 3, in
bold black letters) was not found in the wild-type AvHex by any
The complete amino acid sequence of AvHex
(ASPVEDRAFT_125959; GenBank: OJI97839.1) has more
than 65% identity to many known fungal β-N-
acetylhexosaminidases (Fig. 2). Namely, to AoHex (A. oryzae,
GenBank: AY091636; Škerlová et al. 2018), PoHex
(Penicillium oxalicum, GenBank: EU189026; (Ryšlavá et al.
Table 1 Purification of AvHex from the culture medium of A. versicolor CCF 2491 and P. pastoris (800 mL)
Host
Purification step
Total protein (mg)
Total activity (U)
Specific activity (U/mg)
Purification factor (fold)
Yield (%)
A. versicolor
Culture medium
Dialysis
26.5
19.6
1.05
55.8
18.4
297.1
237.0
83.4
11.2
12.1
79.4
36.8
45.7
0
100
80
1.1
7.1
0
Cation exchange
Culture medium
Anion exchange
28
P. pastoris
2000
1680
100
84
4.2

Appl Microbiol Biotechnol
sequencing protocols. Therefore, we tentatively assigned it to
the A. versicolor signal sequence similar to other fungal β-N-
acetylhexosaminidases (Slámová et al. 2012; Plíhal et al. 2007).
Fig. 2 Multiple sequence alignment of AvHex catalytic domain (“Query”)
„
with solved crystal structures of catalytic domains of β-N-
acetylhexosaminidases from various sources performed with Esprint 3.0
(Robert and Gouet 2014) and Jalview (Clamp et al. 2004). Active-site
amino acids are marked by black dots and labeled according to AvHex
numeration. Identities of the aligned sequences are as follows: AoHex
(pdb: 5OAR/chain B)—73% (99% coverage), β-N-acetylhexosaminidase
from insect Ostrinia furnacalis (pdb: 3NSM/chain A)—31% (76%
coverage), human lysosomal β-hexosaminidase isoform B (pdb:
1NOW/chain A)—29% (75% coverage), β-N-acetylhexosaminidase from
bacteria Streptomyces coelicolor (pdb: 4C7D)—26% (68% coverage),
SmChit (pdb: 1QBB)—24% (68% coverage). Residues with identities
higher than 30% for a particular site are colored by ClustalX coloring
scheme, which groups residues according to similar properties by selected
color: G—orange; P—yellow; positively charged K, R—red; Y, H—
Molecular cloning, recombinant expression,
and purification of AvHex
Since the found protein amino acid sequence of the wild-type
AvHex and the translated amino acid sequence of the hypo-
thetical protein ASPVEDRAFT_125959 were in good accor-
dance, we used this sequence for the construction of the syn-
thetic gene of AvHex. The gene was synthesized commercial-
ly, and the respective nucleotide sequence was optimized for
the expression in P. pastoris (GenBank: MH937562; see Fig.
S7, Supplementary Material). The AvHex sequence contained
a single open reading frame of 1770 base pairs encoding 589
amino acids, comprising the propeptide and the catalytic do-
main (Slámová et al. 2012, Plíhal et al. 2007).
cyan; negatively charged D, E—magenta; polar Q, N, S, T—green;
hydrophobic L, V, M, A, I, W, C, F—
. Secondary structure
blue
elements are based on the 5OAR structure and defined by symbols
and letters as α-alpha helix or β-beta sheet. The long loops HL1 and
HL2 conserved in fungal β-N-acetylhexosaminidases are labeled;
their ends are marked by red lines
The pPICZαA-based construct enabled AvHex to be secreted
to the minimal cultivation medium using the P. pastoris KM71H
strain, which previously proved to be an efficient host for the
production of β-N-acetylhexosaminidases (Slámová et al. 2012,
2015). Upon induction by methanol, the highest activity (2.5 U/
mL, 37 U/mg) was found after 3–4 days of cultivation. The
recombinant AvHex was purified from the medium in a single
step by anion-exchange chromatography at pH 6.5, which was
selected as the second option since the recombinant enzyme
tended to bind irreversibly to the Fractogel resin used for the
wild-type counterpart, resulting in purification yields ≤ 10%.
By the selected method, though, we were able to isolate an
average of 2.3 mg of recombinant AvHex from 100 mL of cul-
tivation medium with a specific activity of 46 U/mg (overall
purification yield 84%; Table 1). Thus, we significantly increased
and simplified the enzyme production compared to the tedious
production in fungi (the cultivation time reduced by half, the
enzyme production enhanced 20 times). The recombinant
AvHex appeared as a single band of ca 140 kDa on native elec-
trophoresis (Fig. 4b), identical to the wild-type enzyme.
Electrophoresis under reducing conditions (SDS-PAGE; Fig.
4a) showed two closely positioned bands (78 kDa and 64 kDa
as determined by MALDI-MS; Fig. S8, Supplementary
Material). The N-deglycosylation by EndoH resulted in a pro-
nounced shift of the AvHex protein bands on SDS-PAGE gel by
ca 6 kDa, corresponding to the values of 72 kDa and 58 kDa,
respectively, determined by MALDI-MS (Fig. S8,
Supplementary Material). LC-MS/MS analysis of the protein
bands confirmed that both protein bands correspond to AvHex.
Based on MALDI-MS data, we suggest that the catalytic do-
mains of the recombinant AvHex enzyme are produced with
two different patterns of O-glycosylation, which have no mea-
surable influence on AvHex catalytic properties. Again, the
AvHex propeptide (ca 14 kDa) was visible as a weak band on a
SDS-PAGE gel (Fig. 4b). The native active AvHex is composed
of two monomers, each containing two polypeptide chains (cat-
alytic domain and propeptide), and it has a molecular weight of
ca 170 kDa as calculated from MALDI-MS results, using an
average mass of both catalytic domains for simplification. The
recombinant enzyme was perfectly stable under long-term stor-
age at 4 °C in McIlvaine buffer pH 5.0.
Biochemical properties of wild-type and recombinant
AvHex
The biochemical parameters of both the wild-type and recom-
binant AvHex were characterized and compared to determine
the effect of the host organism on the properties of the enzyme.
Like other related β-N-acetylhexosaminidases (Ettrich et al.
2007; Ryšlavá et al. 2011), AvHex was found to be a glycopro-
tein as also predicted by the presence of the glycosylation sites
in its amino acid sequence (Fig. S9, Supplementary Material).
Using the GlyProt (Bohne-Lang and von der Lieth 2005) and
NGlycPred server (Chuang et al. 2012), four potential N-gly-
cosylation sites were identified in the catalytic domain of
AvHex. Two of them (Asn361 and Asn508) correspond to the
confirmed glycosylated sites in AoHex. The NGlycPred server
predicted lower probability for the two remaining sites (Asn533
and Asn603) to be N-glycosylated in AvHex (probability
52–51%). This is in accordance with the situation in
AoHex where these respective two residues are not gly-
cosylated either (one of them is buried inside the pro-
tein, and the other carries a proline in the glycosylation
motif). N-Deglycosylation of both the wild-type and the re-
combinant enzymes (see also Fig. 4a) had no influence on
their activities nor on GlcNAcase selectivities.
Using pNP-GlcNAc as a standard substrate, we determined
the pH and temperature optima of the hydrolytic reaction cata-
lyzed by AvHex (Fig. 4c, d). Both the wild-type and

Appl Microbiol Biotechnol

Appl Microbiol Biotechnol
recombinant AvHex displayed similar pH and temperature pro-
files. pH optimum was found to be in a broad plateau between
pH 3.0 and 6.0 (Fig. 4c); the wild-type enzyme displayed a
slightly higher activity up to pH 8.0, in contrast to the recom-
binant enzyme, which may be the impact of differences in
glycosylation. Temperature activity profiles were almost
identical, with the half-maximum and the highest enzyme
activities at ca 40 and 55 °C, respectively (Fig. 4d). The
enzyme was found relatively unstable at 50 °C (within 1 h
its activity declined to ca 6%; Fig. S11, Supplementary
Material). Therefore, 35 °C was selected as an optimum
operating temperature for the enzymatic reactions, at which
the enzyme showed perfect stability. Lower reaction tem-
peratures are also recommendable in order to prevent spon-
taneous decomposition of sensitive compounds.
Homology modeling and docking
For the sequence alignment used for model building, see Fig. S9
(Supplementary Material). The most stable protein structure is
shown in Fig. 5. Both the propeptide and the loop HL2 showed
reorientation already after 2 ns of refinement of mono-
meric structure while with dimer structures the initial
conformation was preserved. This underlines the impor-
tance of dimer formation for maintaining the correct
enzyme conformation during molecular dynamics and the
positive impact of N-glycosylation in stabilizing the protein
structure close to the active site.
The interdomain interactions after refinement are rather
strong and comprise 23–24 hydrogen bonds. Loop HL1 ami-
no acid residues participate just in one interdomain hydrogen
bond interaction (Trp303 of one monomer and Asp599 of the
other monomer). In contrast, the contribution of interdomain
residues to loop HL2 stabilization appears as significant; its
amino acid residues form 12–13 hydrogen bonds with the
residues of the other domain.
The modeled structure of AvHex consists of a propeptide
(aa 19–92) followed by the catalytic domain, which includes a
small zincin-like domain (aa 110 to approx.187) and a C-
terminal (α/β)8 TIM-barrel (aa 188–608). The protein stretch
containing aa 98–108 (PSSSVAAKAKR) corresponds to the
segment cleaved during the activation process in the case of
AoHex. This part could not be modeled properly because,
presumably, the propeptide orientation changes after cleavage
(Škerlová et al. 2018). Active-site amino acids include
Asp353 and Glu354 (catalytic residues), hydrophobic
Trp405 and Trp427, Trp525, and other residues forming hy-
drogen bonds with the substrate, i.e., Tyr453, Gln315,
Arg201, Glu527, Asp455, and Trp490 (Fig. 6b, c).
In the refined model, the optimized orientation of the
side chain of Gln315, the only non-conserved active-site
residue, slightly differs from the position of the analogous
residue Gln494 in SmChit (Fig. 5d). This may be due to a
different orientation of HL1 in fungal β-N-
acetylhexosaminidases (e.g., AoHex) compared to
SmChit—the distance between C-α atoms of the respective
residues Glu307 (AoHex) and Gln494 (SmChit (Tews et al.
1996)) is 0.86 Å. In fungal β-N-acetylhexosaminidases
(Kulik et al. 2015; Škerlová et al. 2018), the analogous
The Michaelis–Menten parameters of hydrolysis of pNP-
GlcNAc substrate were determined both for wild-type and re-
combinant AvHex as follows. For the wild-type AvHex, kcat
=
35.3 1.7 s−1, Km = 0.22 0.06 mM, and kcat/Km =
160.5 s−1 mM−1. For the recombinant AvHex, kcat = 8.13
0.26 s−1, Km = 0.18 0.03 mM, and kcat/Km = 45.2 s−1 mM−1.
The values of the Michaelis–Menten constants (Km) of both
enzymes were similar and in good agreement with the corre-
sponding parameters published for other fungal β-N-
acetylhexosaminidases (Plíhal et al. 2007; Ryšlavá et al. 2011;
Slámová et al. 2012). Based on the kcat/Km values, the wild-type
enzyme is a more efficient catalyst than its recombinant counter-
part, which was also apparent from the values of specific activity
(Table 1). This may be brought about by, e.g., differences in
association of the catalytic domain with propeptide, which may
depend on a particular cultivation and may slightly vary from
batch to batch.
Additionally, sample transglycosylation reactions with pNP-
GlcNAc substrate were performed with both wild-type and re-
combinant AvHex, showing a similar reaction progress as mon-
itored by TLC (data not shown). Obviously, AvHex is able to
perform transglycosylation reactions like many fungal β-N-
acetylhexosaminidases (Bojarová and Křen 2011). In sum, we
may conclude that the recombinant AvHex produced in
P. pastoris retained the properties of the wild-type fungal en-
zyme, and thus it is an equal synthetic alternative that
offers the additional advantage of a high-yielding and
elegant production.
Fig 3 Sequence coverage of the hypothetical protein ASPVEDRAFT_
125959 (GenBank: OJI97839.1) identified in the genome of A. versicolor
CBS 583.65 by BLAST search in the GenBank Database as determined
by LC-MS/MS analysis of the wild-type AvHex. Peptide color code: grey,
< 95% peptide identification confidence; green, ≥ 95% peptide
identification confidence. The predicted N-terminal signal sequence of
19 amino acids is marked in bold black letters

Appl Microbiol Biotechnol
Fig. 4 a Purification and deglycosylation of recombinant AvHex as
analyzed by SDS-PAGE. Lane M, molecular mass markers consisting
of phosphorylase b from rabbit muscle (97 kDa), bovine serum albumin
(66 kDa), chicken egg white ovalbumin (45 kDa), carbonic anhydrase
from bovine erythrocytes (30 kDa), trypsin inhibitor from soya beans
(20.1 kDa), and α-lactalbumin from bovine milk (14.4 kDa); lane 1,
recombinant AvHex (two subunits of 78 and 64 kDa) with propeptide
(14 kDa) purified by anion-exchange chromatography; lane 2, recombi-
nant AvHex after N-deglycosylation (two subunits of 72 and 58 kDa) with
propeptide (14 kDa) and Endo H (29 kDa) and Endo H; lane 3, Endo H.
The position of the putative propeptide co-purifying with the catalytic
subunit is indicated with an arrow. b Native electrophoresis of AvHex.
Lane M, molecular mass markers consisting of porcine thyroid thyroglob-
ulin (669 kDa), equine spleen ferritin (440 kDa), bovine liver catalase
(232 kDa), bovine heart lactate dehydrogenase (140 kDa), and bovine
serum albumin (66 kDa); lane 1, wild-type purified AvHex; lane 2, re-
combinant purified AvHex. c pH optimum of wild-type and recombinant
AvHex. d Temperature optimum of wild-type and recombinant AvHex
Glu residue directly interacts with the carbohydrate sub-
strate at the −1 position; in TfHex, it stabilizes the interaction
with pNP-GalNAc substrate compared to GlcNAc by addi-
tional hydrogen bonding. Therefore, we hypothesize that the
non-conserved residue Gln315 in AvHex may contribute to its
exceptional N-acetylglucosaminidase selectivity compared to
other fungal β-N-acetylhexosaminidases that are quite pro-
miscuous to both GlcNAc and GalNAc substrates.
In the equilibrated models, we observed that all docked
GlcNAc substrates, i.e., pNP-GlcNAc, GlcNAc-oxazoline (in-
termediate of substrate-assisted catalysis) (Slámová and
Bojarová 2017), and GlcNAc monosaccharide, have a lower
binding score than the respective GalNAc ligands, which cor-
responds to a more favorable interaction of GlcNAc with
AvHex than GalNAc. It is worthwhile to note that the hydrolytic
product GlcNAc, which gets easily released from the active
site, exhibits a similar binding score to pNP-GalNAc. This
indicates that pNP-GalNAc has only weak binding interactions
with the active-site residues. It is caused by the loss of hydrogen
bond interactions with Arg201 and Glu527. On the other hand,
a lower binding score for pNP-GlcNAc corresponds to strong
binding interactions, which are brought about by a higher
Molecular docking
The refined model structure of AvHex was used for analyzing
the enzyme–substrate interactions by docking and molecular
dynamics simulation. The results are shown in Table 2.

Appl Microbiol Biotechnol
Fig. 5 Molecular model of AvHex. a Protein dimer (cartoon representation)
with pNP-GlcNAc substrate (stick representation) in the active site.
Propeptide of monomer 1 is in red, catalytic domain of monomer 1 in
magenta, propeptide of monomer 2 in blue, and catalytic domain of
monomer 2 in cyan. b, c Active-site amino acid residues with docked
pNP-GlcNAc after refinement (two different views). Trp residues forming
Trp hydrophobic cavity for substrate binding are shown separately in panel c
for clarity. Hydrogen bonds are depicted as yellow dotted lines. d Overlay of
+ 1 binding pocket segments in fungi (AvHex in cyan, SmChit in green, and
β-N-acetylhexosaminidase from O. furnacalis in magenta)
number of hydrogen bonds in the substrate–enzyme complex
(namely, Arg201, Gln315, and Glu527, see Fig. 6).
To analyze the contribution of Gln315 to substrate binding,
we calculated the Glide score for selected variants at the position
of Gln315 (see Table S5, Supplementary Material). The binding
score for pNP-GlcNAc bound in the active site of modeled
Gln315Gly and Gln315Glu mutants significantly increased
compared to the wild-type enzyme, which indicates the impor-
tance of Gln315 residue for the high affinity to pNP-GlcNAc. In
contrast, in the case of pNP-GalNAc, the differences in binding
affinity between the Gln315Gly and Gln315Glu mutants and
the wild type were less significant.
Fig. 6 a pNP-GlcNAc in the
active site of AvHex after 10 ns of
molecular dynamics simulation.
Hydrogen bonds are depicted as
yellow dotted lines. b pNP-
GalNAc in the active site of
AvHex after 10 ns of molecular
dynamics simulation. c Number
of hydrogen bonds over time
formed with bound pNP-GlcNAc
and pNP-GalNAc during the
molecular dynamics simulation. d
Time course of hydrogen bond
formation between the substrate
and Glu315; a value of 1 indicates
the presence of a hydrogen bond;
0 indicates the absence of a
hydrogen bond

Appl Microbiol Biotechnol
Table 2 Averaged Glide XP binding scores for GlcNAc and GalNAc
substrates in AvHex
acetylhexosaminidases and β-N-acetylglucosaminidases con-
taining Gln in the position corresponding to Gln315 in AvHex,
Tyr is found in the position corresponding to Tyr461 (see
Table S4, Supplementary Material). The hydroxyl hydrogen of
Tyr461 makes a hydrogen bond with the carboxyl oxygen of
Gln315 and it may form a water-mediated hydrogen bond with
the C-4 oxygen of GlcNAc-type substrate (equatorial hydroxyl),
which may result in additional stabilization of this configuration.
Substrate
Glide score (kJ/mol)a
pNP-GlcNAc
pNP-GalNAc
GlcNAc-oxazolineb
GalNAc-oxazolineb
GlcNAc
− 36.8
− 31.7
− 40.0
− 35.9
− 30.6
− 27.5
GalNAc
Application of AvHex in a sequential synthesis
of a functionalized disaccharide
a Glide score after molecular dynamics simulation for equilibrated
substrate–enzyme complex. Glide score represents the free energy of
binding. The lower score, the stronger binding
b For the systematic names and structures of oxazolines, see Scheme S1,
Supplementary Material
We employed the selective AvHex enzyme in a sequential en-
zymatic synthesis of a bioactive disaccharide, β-D-
GalNAc-(1 → 4)-β-D-GlcNAc-N3 that carries an azido func-
tion at C-1 (Scheme 1a, compound 3), using β-D-GlcNAc azide
(2) as an acceptor (Fialová et al. 2005).
The azido-terminated disaccharide 3 is quite difficult to pre-
pare using glycosyltransferases (Bojarová et al. 2013), which
are sensitive to changes in both donor and acceptor substrate
molecules; other linkers at C-1 are much more synthetically
feasible (Bumba et al. 2018). This disaccharide may be pre-
pared using site-directed mutants of TfHex (i.e., Tyr470His),
which exhibit increased transglycosylation potential (Slámová
et al. 2015). However, due to their dual affinity to both GalNAc
and GlcNAc substrates combined with their broad substrate
specificity, this reaction is complicated by the formation of
Further, molecular dynamics simulations with the
Gln315Glu mutant showed that pNP-GalNAc lost its hydrogen
bond with Arg201, which was compensated by the formation
of a new hydrogen bond between the substrate C-4 hydroxyl
and Glu315; a similar situation was previously observed in
TfHex, which is a promiscuous β-N-acetylhexosaminidase
with a GlcNAcase/GalNAcase ratio of 1/1.2. Therefore, we
may suppose that the Gln315Glu mutation decreases the differ-
ence in binding affinities to GalNAc and GlcNAc substrates.
A notable issue is the presence of Tyr461 close to Gln315. In
ca 85% of sequences of homologous fungal β-N-
Scheme 1 a Sequential synthesis of functionalized disaccharide 3 employing Tyr470His TfHex and AvHex. b Synthesis of trivalent glycoconjugate 8
using functionalized disaccharide 3 and trivalent carrier 7 by Cu(I)-catalyzed azide-alkyne cycloaddition

Appl Microbiol Biotechnol
Table 3 Inhibition constants for
glycoconjugate 8 and standard
carbohydrates with galectin-3 de-
termined by ELISA
Inhibitor
IC50 (μM)
Relative potency
Relative potency per glycan
Lactose (Galβ4Glc)
GalNAcβ4GlcNAc
Glycoconjugate 8
153 24
55 11
0.36
1
0.36
1
9.6 3.2
5.7
1.9
chitobiosyl azide (β-D-GlcNAc-(1 → 4)-β-D-GlcNAc-N3; 4)
as an unwanted by-product. This compound originates in an
enzymatic side reaction by processing the GlcNAc azide accep-
tor (2) and it may be contained in the ratio of up to 1:4 to the
desired product 3. Glycosyl azides have been shown to act as
glycosyl donors with many glycosidases (Bojarová et al. 2007).
Naturally, the separation of both disaccharides differing only in
one axial-equatorial position is practically impossible by con-
ventional methods such as HPLC or gel chromatography
(Bojarová et al. 2008). As a result, the synthetic reaction occurs
with a lower isolated yield and the product may have lower
purity. The employment of a sequential reaction presented here-
in elegantly solves this synthetic problem since AvHex effi-
ciently cleaves the unwanted chitobiosyl azide 4. Though
AvHex has some minor GalNAcase activity (GlcNAcase/
GalNAcase ratio of ca 9–11), we did not observe any
cleavage of our desired product 3 during the sequential
reaction. In the present reaction setup, we were able to
increase the yield of the desired disaccharide 3 by ca 30%,
reaching a 47% overall isolated yield. The structure of disac-
charide 3 was confirmed by HRMS and 1H and 13C NMR.
T h o u g h w e t e s t e d o t h e r e s t a b l i s h e d β - N -
acetylglucosaminidases for the selective hydrolysis of 4, e.g.,
that from Bacteroides thetaiotamicron or recombinant human
O-GlcNAcase, neither of the tested enzymes exhibited affinity
to chitobiose-based oligosaccharides, apparently due to
their selectivity for GlcNAc bound to a protein scaffold.
Therefore, we consider AvHex a very useful enzyme for
biotechnological applications involving tailored carbohy-
drate synthesis.
competitive ELISA-type assay using asialofetuin as a standard
immobilized competitor (Laaf et al. 2017b) and compared to
the free monovalent GalNAcβ4GlcNAc. The results are sum-
marized in Table 3 and the respective dose-dependent inhibition
curves are shown in Fig. S12 (Supplementary Material). We
can see that the presentation of the selective epitope on the
aromatic carrier via the triazole linker enhanced the conjugate
affinity 5.7-fold compared to the monovalent disaccharide, and
this mode of presentation has increased the relative inhibitory
potency per glycan 1.9-fold. This interaction is purely specific
and is not caused by potential non-specific hydrophobic forces
as demonstrated in a control assay in the presence of 0.025%
v/v Tween, giving the same IC50 values. In sum, conjugate 8 is a
good leading example for the construction of glycomimetics
based on aromatic scaffolds.
Discussion
GalNAc-terminated oligosaccharides are excellent ligands of
galectin-3, a soluble monomeric human lectin of chimera type
with a high biomedical relevance. Contrary to analogous epi-
topes terminated with β-D-galactopyranoside, they exhibit ex-
clusive selectivity for galectin-3 compared to a similarly fre-
quent galectin-1 (Šimonová et al. 2014), which is crucial for
potential biomedical applications (Bojarová and Křen 2016).
N-Acetylhexosamine carbohydrates with various functional
groups in the molecule are advantageously synthesized using
robust and stable glycosidases with a broad substrate specific-
ity. If an azido moiety is used for conjugation to a carrier by
click chemistry, the formed triazol supports binding to
galectin-3 by additional interactions with its binding site
(Bojarová et al. 2018). Moreover, Since the enzymatic glyco-
sylation of multivalent precursors by both glycosidases and
glycosyltransferases does not proceed easily and mostly af-
fords partially glycosylated by-products (Drozdová et al.
2011), a straightforward conjugation of functionalized oligo-
saccharides by click chemistry is a very attractive solution.
Nevertheless, azido-functionalized GalNAc-terminated epi-
topes are quite challenging to prepare by glycosyltransferases
due to their limited tolerance to structural modifications in the
donor and acceptor molecule. Therefore, a high-yielding syn-
thetic route to the selective azido-bearing GalNAcβ4GlcNAc
epitope, using a combination of a previously reported
Tyr470His variant of TfHex (Slámová et al. 2015) and a novel
highly selective AvHex, is an attractive option for the design of
Preparation of bioactive conjugate 8 by click
chemistry and its binding to galectin-3
The functionalized GalNAcβ4GlcNAc epitope 3 was conjugat-
ed to a simple aromatic carrier 7 based on pyrogallol (6), car-
rying three propargyl moieties (Scheme 1b). Compound 7 was
prepared according to a published procedure (Xie et al. 2008).
The conjugation with carbohydrate 3 was performed by Cu(I)-
catalyzed azide-alkyne cycloaddition in water/methanol mix-
ture to yield glycoconjugate 8. Though similar conjugates have
been tested with galectins (Wang et al. 2012), glycoconjugate 8
is the first dendrimer-like multivalent glycoconjugate that
carries the GalNAcβ4GlcNAc epitope selective for galectin-
3. Compound 8 was isolated in 53% yield by gel chromatogra-
phy; its binding affinity to galectin-3 was determined in a

Appl Microbiol Biotechnol
galectin-3 inhibitors. The novel trivalent glycoconjugate 8
shows good affinity to galectin-3 and as such may serve as a
leading structure for the construction of advanced multivalent
selective galectin-3 inhibitors.
Compliance with ethical standards This article does not con-
tain any studies with human participants or animals performed by any of
the authors.
Conflict of interest The authors declare that they have no conflict of
Since the dual activity towards both GlcNAc and
GalNAc substrates is an intrinsic characteristic of β-N-
acetylhexosaminidases, the selectivity of the present
AvHex is an exceptional catalytic feature. Notably, other
GlcNAc-selective enzymes known, such as GH84 O-
GlcNAcase (Macauley et al. 2005), are not applicable
for the processing of N-acetylhexosamine oligosaccha-
rides, nor do they show any transglycosylation capabil-
ities. Both of these properties, together with a broad pH
optimum, a good tolerance of functionalized carbohydrate
substrates and excellent long-term storage stability, predestine
AvHex for biotechological applications.
Molecular modeling analysis and comparison to the
existing β-N-acetylhexosaminidase counterparts, in partic-
ular AoHex, helped to decode the relations in AvHex binding
site and brought a possible explanation for its exceptional
selectivity. Residue Gln315, possibly in combination with
Tyr461, was identified as a probable source of the ex-
ceptional selectivity of AvHex to substrates in GlcNAc
configuration. It follows from the extensive alignment
study of the existing homologous enzymes as well as
from the computational modeling of mutants that this
particular combination of amino acid residues may influ-
ence the GlcNAcase/GalNAcase activity ratio in β-N-
acetylhexosaminidases. Recombinant expression in
P. pastoris, recently identified as an optimum heterolo-
gous producer of fungal β-N-acetylhexosaminidases
(Slámová et al. 2012), enabled to enhance the production
of AvHex ca 20-fold whereas its biocatalytic properties
and selectivity were maintained. In conclusion, the re-
combinant AvHex presented in this work is a highly use-
ful tool for a range of biotechnological applications, es-
pecially involving the synthesis and/or processing of N-
acetylhexosamine oligosaccharides that comprise a com-
bination of both β-GlcNAc and β-GalNAc residues. An
example of such carbohydrates is a selective ligand of
galectins presented herein.
interest.
Publisher’s Note Springer Nature remains neutral with regard to juris-
dictional claims in published maps and institutional affiliations.
References
Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local
alignment search tool. J Mol Biol 215:403–410. https://doi.org/10.
1016/S0022-2836(05)80360-2
Bohne-Lang A, von der Lieth CW (2005) GlyProt: in silico glycosylation
of proteins. Nucleic Acids Res 33:W214–W219. https://doi.org/10.
1093/nar/gki385
Bojarová P, Křen V (2009) Glycosidases: a key to tailored carbohydrates.
Trends Biotechnol 27:199–209. https://doi.org/10.1016/j.tibtech.
2008.12.003
Bojarová P, Křen V (2011) Glycosidases in carbohydrate synthesis: when
organic chemistry falls short. Chimia 65:65–70. https://doi.org/10.
2533/chimia.2011.65
Bojarová P, Křen V (2016) Sugared biomaterial binding lectins: achieve-
ments and perspectives. Biomat Sci 4:1142–1160. https://doi.org/10.
1039/c6bm00088f
Bojarová P, Petrásková L, Ferrandi E, Monti D, Pelantová H, Kuzma M,
Simerská P, Křen V (2007) Glycosyl azides – an alternative way to
disaccharides. Adv Synth Catal 349:1514–1520
Bojarová P, Křenek K, Kuzma M, Petrásková L, Bezouška K, Namdjou
D-J, Elling L, Křen V (2008) N-Acetylhexosamine triad in one mol-
ecule: chemoenzymatic introduction of 2-acetamido-2-deoxy-β-D-
galactopyranosyluronic acid residue into a complex oligosaccharide.
J Mol Catal B Enzym 50:69–73. https://doi.org/10.1016/j.molcatb.
2007.09.002
Bojarová P, Slámová K, Křenek K, Gažák R, Kulik N, Ettrich R,
Pelantová H, Kuzma M, Riva S, Adámek D, Bezouška K, Křen V
(2011) Charged hexosaminides as new substrates for β-N-
scetylhexosaminidase-catalyzed synthesis of immunomodulatory
disaccharides. Adv Synth Catal 353:2409–2420. https://doi.org/10.
1002/adsc.201100371
Bojarová P, Rosencrantz RR, Elling L, Křen V (2013) Enzymatic glyco-
sylation of multivalent scaffolds. Chem Soc Rev 42:4774–4797.
https://doi.org/10.1039/c2cs35395d
Bojarová P, Chytil P, Mikulová B, Bumba L, Konefal R, Pelantová H,
Krejzová J, Slámová K, Petrásková L, Kotrchová L, Cvačka J,
Etrych T, Křen V (2017) Glycan-decorated HPMA copolymers as
high-affinity lectin ligands. Polym Chem 8:2647–2658. https://doi.
org/10.1039/C7PY00271H
Bojarová P, Tavares MR, Laaf D, Bumba L, Petrásková L, Konefał R,
Bláhová M, Pelantová H, Elling L, Etrych T, Chytil P, Křen V
(2018) Biocompatible glyconanomaterials based on HPMA-
copolymer for specific targeting of galectin-3. J Nanobiotechnol
16:73. https://doi.org/10.1186/s12951-018-0399-1
Acknowledgments P. B. and V. K. acknowledge support by mobility
projects no. LTC18038 and LTC18041 (MEYS, the Ministry of
Education, Youth and Sports of the Czech Republic). N. K. acknowledges
access to the computing and storage facilities provided by the CESNET
LM2015042 and the CERIT Scientific Cloud LM2015085 under the
program BProjects of Large Research, Development, and Innovations
Infrastructures.^
Bradford M (1976) A rapid and sensitive method for the quantitation of
microgram quantities of protein utilizing the principle of protein-dye
binding. Anal Biochem 72:248–254. https://doi.org/10.1016/0003-
2697(76)90527-3
Bumba L, Laaf D, Spiwok V, Elling L, Křen V, Bojarová P (2018) Poly-
N-acetyllactosamine neo-glycoproteins as nanomolar ligands of
Funding This study was funded by the Ministry of Education, Youth, and
Sports of the Czech Republic mobility projects nos. LTC18038 and
LTC18041, by CESNET (LM2015042), and by CERIT Scientific
Cloud (LM2015085) under the program BProjects of Large Research,
Development, and Innovations Infrastructures.^

Appl Microbiol Biotechnol
human galectin-3: binding kinetics and modeling. Int J Mol Sci 19:
372. https://doi.org/10.3390/ijms19020372
Chem 28:2832–2840. https://doi.org/10.1021/acs.bioconjchem.
7b00520
Chuang G-Y, Boyington JC, Joyce MG, Zhu J, Nabel GJ, Kwong PD,
Georgiev I (2012) Computational prediction of N-linked glycosyla-
tion incorporating structural properties and patterns. Bioinformatics
28:2249–2255. https://doi.org/10.1093/bioinformatics/bts426
Clamp M, Cuff J, Searle SM, Barton GJ (2004) The Jalview Java align-
ment editor. Bioinformatics 20:426–427. https://doi.org/10.1093/
bioinformatics/btg430
Davis IW, LeaverFay A, Chen VB, Block JN, Kapral GJ, Wang X,
Murray LW, Arendall WB 3rd, Snoeyink J, Richardson JS,
Richardson DC (2007) MolProbity: allatom contacts and structure
validation. Nucleic Acids Res 35:W375–W383. https://doi.org/10.
1093/nar/gkm216.
Drozdová A, Bojarová P, Křenek K, Weignerová L, Henßen B, Elling L,
Christensen H, Jensen HH, Pelantová H, Kuzma M, Bezouška K,
Krupová M, Adámek D, Slámová K, Křen V (2011) Enzymatic
synthesis of dimeric glycomimetic ligands of NK cell activation
receptors. Carbohydr Res 346:1599–1609. https://doi.org/10.1016/
j.carres.2011.04.043
Ettrich R, Kopecký V Jr, Hofbauerová K, Baumruk V, Novák P, Pompach
P, Man P, Plíhal O, Kutý M, Kulik N, Sklenář J, Ryšlavá H, Křen V,
Bezouška K (2007) Structure of the dimeric N-glycosylated form of
fungal β-N-acetylhexosaminidase revealed by computer modeling,
vibrational spectroscopy, and biochemical studies. BMC Struct Biol
7:32–45. https://doi.org/10.1186/1472-6807-7-32
Fialová P, Weignerová L, Rauvolfová J, Přikrylová V, Pišvejcová A,
Ettrich R, Kuzma M, Sedmera P, Křen V (2004) Hydrolytic and
transglycosylation reactions of N-acyl modified substrates catalysed
by β-N-acetylhexosaminidases. Tetrahedron 60:693–701. https://
doi.org/10.1016/j.tet.2003.10.111
Fialová P, Carmona AT, Robina I, Ettrich R, Sedmera P, Přikrylová V,
Petrásková-Hušáková L, Křen V (2005) Glycosyl azide - a novel
substrate for enzymatic transglycosylations. Tetrahedron Lett 46:
8715–8718. https://doi.org/10.1016/j.tetlet.2005.10.040
Friesner RA, Murphy RB, Repasky MP, Frye LL, Greenwood JR,
Halgren TA, Sanschagrin PC, Mainz DT (2006) Extra precision
Glide: docking and scoring incorporating a model of hydrophobic
enclosure for protein-ligand complexes. J Med Chem 49:6177–
6196. https://doi.org/10.1021/jm051256o
Gloster TM, Vocadlo DJ (2010) Mechanism, structure, and inhibition of
O-GlcNAc processing enzymes. Curr Signal Transduct Ther 5:74–
91. https://doi.org/10.2174/157436210790226537
Huňková Z, Křen V, Ščigelová M, Weignerová L, Scheel O, Thiem J
(1996) Induction of β-N-acetylhexosaminidase in Aspergillus
oryzae. Biotechnol Lett 18:725–730. https://doi.org/10.1007/
BF00130773
Jakalian A, Jack DB, Bayly CI (2002) Fast, efficient generation of high
quality atomic charges. AM1BCC model: II. Parameterization and
validation. J Comput Chem 23:1623–1641. https://doi.org/10.1002/
jcc.10128
Krieger E, Koraimann G, Vriend G (2002) Increasing the precision of
comparative models with YASARA NOVA a self parameterizing
force field. Proteins 47:393–402. https://doi.org/10.1002/prot.10104
Kulik N, Slámová K, Ettrich R, Křen V (2015) Computational study of β-
N-acetylhexosaminidase from Talaromyces flavus, a glycosidase
with high substrate flexibility. BMC Bioinformatics 16:1–15.
https://doi.org/10.1186/s12859-015-0465-8.
Laaf D, Bojarová P, Mikulová B, Pelantová H, Křen V, Elling L (2017a)
Two-step enzymatic synthesis of β-D-N-acetylgalactosamine-(1-4)-
D-N-acetylglucosamine (LacdiNAc) chitooligomers for deciphering
galectin binding behavior. Adv Synth Catal 359:2101–2108. https://
doi.org/10.1002/adsc.201700331
Macauley MS, Whitworth GE, Debowski AW, Chin D, Vocadlo DJ
(2005) O-GlcNAcase uses substrate-assisted catalysis; kinetic anal-
ysis and development of highly selective mechanism-inspired inhib-
itors. J Biol Chem 280:25313–25,322. https://doi.org/10.1074/jbc.
M413819200
Notredame C, Higgins DG, Heringa J (2000) T-coffee: a novel method for
fast and accurate multiple sequence alignment. J Mol Biol 302:205–
217. https://doi.org/10.1006/jmbi.2000.4042.
Plíhal O, Sklenář J, Hofbauerová K, Novák P, Man P, Pompach P,
Ryšlavá H, Charvátová-Pišvejcová A, Křen V, Bezouška K (2007)
Large propeptides of fungal β-N-acetylhexosaminidases are novel
enzyme regulators that must be processed intracellularly to control
activity, dimerization, and secretion into the extracellular environ-
ment. Biochemistry 46:2719–2734. https://doi.org/10.1021/
bi061828m
Raval A, Piana S, Eastwood MP, Dror RO, Shaw DE (2012) Refinement
of protein structure homology models via long, all-atom molecular
dynamics simulations. Proteins 80:2071–2079. https://doi.org/10.
1002/prot.24098.
Robert X, Gouet P (2014) Deciphering key features in protein structures
with the new ENDscript server. Nucleic Acids Res 42:W320–
W324. https://doi.org/10.1093/nar/gku316
Ryšlavá H, Kalendová A, Doubnerová V, Skočdopol P, Kumar V,
Kukačka Z, Pompach P, Vaněk O, Slámová K, Bojarová P, Kulík
N, Ettrich R, Křen V, Bezouška K (2011) Enzymatic characteriza-
tion and molecular modeling of an evolutionarily interesting fungal
β-N-acetylhexosaminidase. FEBS J 278:2469–2484. https://doi.org/
10.1111/j.1742-4658.2011.08173.x
Sali A, Blundell TL (1993) Comparative protein modelling by satisfac-
tion of spatial restraints. J Mol Biol 234:779–815. https://doi.org/10.
1006/jmbi.1993.1626
Šimonová A, Kupper CE, Böcker S, Müller A, Hofbauerová K,
Pelantová H, Elling L, Křen V, Bojarová P (2014) Chemo-
enzymatic synthesis of LacdiNAc dimers of varying length as novel
galectin ligands. J Mol Catal B Enzym 101:47–55. https://doi.org/
10.1016/j.molcatb.2013.12.018
Škerlová J, Bláha J, Pachl P, Hofbauerová K, Kukačka Z, Man P,
Pompach P, Novák P, Otwinowski Z, Brynda J, Vaněk O,
Řezáčová P (2018) Crystal structure of native β-N-
acetylhexosaminidase isolated from Aspergillus oryzae sheds light
onto its substrate specificity, high stability, and regulation by
propeptide. FEBS J 285:580–598. https://doi.org/10.1111/febs.
14360
Slámová K, Bojarová P (2017) Engineered N-acetylhexosamine-active
enzymes in glycoscience. Biochim Biophys Acta Gen Subj 1861:
2070–2087. https://doi.org/10.1016/j.bbagen.2017.03.019
Slámová K, Bojarová P, Petrásková L, Křen V (2010) β-N-
Acetylhexosaminidase: what‘s in a name...? Biotechnol Adv 28:
682–693. https://doi.org/10.1016/j.biotechadv.2010.04.004
Slámová K, Bojarová P, Gerstorferová D, Fliedrová B, Hofmeisterová J,
Fiala M, Pompach P, Křen V (2012) Sequencing, cloning and high-
yield expression of a fungal β-N-acetylhexosaminidase in Pichia
pastoris. Prot Express Purif 82:212–217. https://doi.org/10.1016/j.
pep.2012.01.004
Slámová K, Krejzová J, Marhol P, Kalachova L, Kulik N, Pelantová H,
Cvačka J, Křen V (2015) Synthesis of derivatized chitooligomers
using transglycosidases engineered from the fungal GH20 β-N-
acetylhexosaminidase. Adv Synth Catal 357:1941–1950. https://
doi.org/10.1002/adsc.201500075
Tews I, Perrakis A, Oppenheim A, Dauter Z, Wilson KS, Vorgias CE
(1996) Bacterial chitobiase structure provides insight into catalytic
mechanism and the basis of Tay-Sachs disease. Nat Struct Biol 3:
638–648. https://doi.org/10.1038/nsb0796-638
Laaf D, Bojarová P, Pelantová H, Křen V, Elling L (2017b) Tailored
multivalent neo-glycoproteins: synthesis, evaluation, and applica-
tion of a library of galectin-3-binding glycan ligands. Bioconjug

Appl Microbiol Biotechnol
Wang GN, André S, Gabius HJ, Murphy PV (2012) Bi- to tetravalent
glycoclusters: synthesis, structure-activity profiles as lectin inhibi-
tors and impact of combining both valency and headgroup tailoring
on selectivity. Org Biomol Chem 10:6893–6907. https://doi.org/10.
1039/c2ob25870f
Weignerová L, Vavrušková P, Pišvejcová A, Thiem J, Křen V (2003) Fungal
β-N-acetylhexosaminidases with high β-N-acetylgalactosaminidase
activity and their use for synthesis of β-GalNAc-containing oligosac-
charides. Carbohydr Res 338:1003–1008. https://doi.org/10.1016/
S0008-6215(03)00044-2
Wiederstein M, Sippl MJ (2007) ProSA-web: interactive web service for
the recognition of errors in three-dimensional structures of proteins.
Nucleic Acids Res 35:W407–W410. https://doi.org/10.1093/nar/
gkm290
Xie J, Hu L, Shi W, Deng X, Cao Z, Shen Q (2008) Synthesis and
characterization of hyperbranched polytriazole via an ‘A2 + B3’ ap-
proach based on click chemistry. Polym Int 57:965–974. https://doi.
org/10.1002/pi.2433