Multi-enzyme Cascades for the In Vitro Synthesis of Guanosine Diphosphate L-Fucose

Article abstract of DOI:10.1002/cctc.202001854

Recombinant Leloir glycosyltransferases can be exploited to synthesize a wide range of HMOs using in vitro biocatalytic reactions. However, high costs and unavailability of bulk amounts of most nucleotide sugars, such as guanosine diphosphate L-fucose (GDP-Fuc), are major obstacles for the efficient large-scale synthesis. Here, we report two novel multi-enzyme cascades for the synthesis of GDP-Fuc from readily available and low cost precursors. The first cascade was developed to produce GDP-Fuc from guanosine (Guo), fucose (Fuc), polyphosphate (PolyP_n) and catalytic amounts of adenine triphosphate (ATP). GDP-Fuc was produced with a final concentration of 7 mM (4.1 g/L) and a reaction yield of 68 % from Guo and Fuc within 48 h with a biocatalyst load of 0.34 g_enzyme/g_product. A second cascade, consisting of ten enzymes and eleven reactions was developed to carry out the synthesis from mannose (Man), Guo, PolyP_n, L-glutamine (L-Glu) and catalytic amounts of ATP, and nicotinamide adenine dinucleotide phosphate (NADPH). Utilizing this cascade, GDP-Fuc was produced with a final concentration of 7.6 mM (4.5 g/L) and a reaction yield of 72 % in a reaction time of 48 h with a biocatalyst load of 0.97 g_enzyme/g_product. Finally, a method for chromatographic purification of GDP-Fuc was established achieving product purities of 90.5 %.

Full text of DOI:10.1002/cctc.202001854

Full Papers
doi.org/10.1002/cctc.202001854
ChemCatChem
Multi-enzyme Cascades for the In Vitro Synthesis of
Guanosine Diphosphate L-Fucose
Reza Mahour,[a] Pavel A. Marichal-Gallardo,[b] Thomas F. T. Rexer,*[c] and Udo Reichl[d]
Recombinant Leloir glycosyltransferases can be exploited to
synthesize a wide range of HMOs using in vitro biocatalytic
reactions. However, high costs and unavailability of bulk
amounts of most nucleotide sugars, such as guanosine
diphosphate L-fucose (GDP-Fuc), are major obstacles for the
efficient large-scale synthesis. Here, we report two novel multi-
enzyme cascades for the synthesis of GDP-Fuc from readily
available and low cost precursors. The first cascade was
developed to produce GDP-Fuc from guanosine (Guo), fucose
(Fuc), polyphosphate (PolyPn) and catalytic amounts of adenine
from Guo and Fuc within 48 h with a biocatalyst load of
0.34 genzyme/gproduct. A second cascade, consisting of ten enzymes
and eleven reactions was developed to carry out the synthesis
from mannose (Man), Guo, PolyPn, L-glutamine (L-Glu) and
catalytic amounts of ATP, and nicotinamide adenine dinucleo-
tide phosphate (NADPH). Utilizing this cascade, GDP-Fuc was
produced with a final concentration of 7.6 mM (4.5 g/L) and a
reaction yield of 72% in a reaction time of 48 h with a
biocatalyst load of 0.97 genzyme/gproduct. Finally, a method for
chromatographic purification of GDP-Fuc was established
achieving product purities of 90.5%.
triphosphate (ATP). GDP-Fuc was produced with
a final
concentration of 7 mM (4.1 g/L) and a reaction yield of 68%
Introduction
bers of the Leloir glycosyltransferase class of enzymes. Fucosyl-
transferases catalyze the transfer of fucose (Fuc) from guanosine
diphosphate L-fucose (GDP-Fuc, the activated form of Fuc) to
an acceptor oligosaccharide or protein. While most HMOs are
produced to date by in vivo approaches, i.e., fermentation, the
potential of in vitro enzymatic synthesis has been demonstrated
Positive effects of functional oligosaccharides on human health
have been shown in various studies.[1,2] This concerns, in
particular, the diet with human milk oligosaccharides (HMOs)
from infancy to adulthood. For instance, it was shown that 2’-
fucosyllactose contributes to the cognitive development of
infants.[2] Consequently, there are ongoing efforts to include a
wide variety of HMOs in infant food formula.[3] Of the more than
200 known HMO structures, around 70% are fucosylated.[4]
Fucosylation is catalyzed in vivo by fucosyltransferases, mem-
to yield
oligosaccharides
a
wide range of simple and notably complex
using recombinant Leloir
glycosyltransferases.[5,6] This synthesis, however, requires high
cost sugar nucleotides such as GDP-Fuc (~30 E/mg) which
significantly hampers low-cost large-scale manufacturing.[5,7]
Therefore, developing scalable synthesis routes for sugar
nucleotides is an important step towards the viable enzymatic
production of functional oligosaccharides.
[a] R. Mahour
Department of Bioprocess Engineering
Max Planck Institute for Dynamics of Complex Technical Systems
Sandtorstrasse 1, 39106 Magdeburg (Germany)
[b] Dr. P. A. Marichal-Gallardo
Department of Bioprocess Engineering
Max Planck Institute for Dynamics of Complex Technical Systems
Sandtorstrasse 1, 39106 Magdeburg (Germany)
[c] Dr. T. F. T. Rexer
Department of Bioprocess Engineering
Max Planck Institute for Dynamics of Complex Technical Systems
Sandtorstrasse 1, 39106 Magdeburg (Germany)
E-mail: rexer@mpi-magdeburg.mpg.de
[d] Prof. Dr. U. Reichl
In vivo GDP-Fuc is either produced de novo by the
conversion of guanosine diphosphate mannose (GDP-Man) and
nicotinamide adenine dinucleotide phosphate (NADPH) as part
of the hexosamine pathway or directly from Fuc through the
salvage pathway.[8] In one of the first studies on its preparative
scale synthesis, 78 mg of GDP-Fuc were produced from 100 mg
of GDP-Man and 148 mg of NADPH.[9] By a combination of four
different permeabilized microbial cells, Koizumi et al. synthe-
sized GDP-Fuc from mannose (Man) and guanosine mono-
phosphate (GMP) with yields of 17% and 52%, respectively, to
a final GDP-Fuc concentration of 18.4 g/L.[10] After the discovery
and successful recombinant production of bifunctional L-
fucokinase/GDP-L-fucose pyrophosphorylase (FKP), GDP-Fuc
was synthesized from Fuc, adenine triphosphate (ATP), and
guanosine trisphosphate (GTP).[11–13] Recently, Wang et al.
carried out the synthesis from Man, ATP, NADPH, and GTP with
a final concentration of 178 mg/L and a yield of 14%.[11,14] In
addition to the in vitro enzymatic synthesis, the production of
GDP-Fuc by the fermentation of genetically engineered bacte-
rial strains has been investigated.[15]
Max Planck Institute for Dynamics of Complex Technical Systems
Sandtorstrasse 1, 39106 Magdeburg
and
Otto-von-Guericke-University Magdeburg
Chair of Bioprocess Engineering
Universitätsplatz 2, 39106 Magdeburg,
(Germany)
Supporting information for this article is available on the WWW under
https://doi.org/10.1002/cctc.202001854
© 2020 The Authors. ChemCatChem published by Wiley-VCH GmbH. This is
an open access article under the terms of the Creative Commons Attribution
Non-Commercial NoDerivs License, which permits use and distribution in
any medium, provided the original work is properly cited, the use is non-
commercial and no modifications or adaptations are made.
ChemCatChem 2021, 13, 1981–1989
1981
© 2020 The Authors. ChemCatChem published by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/cctc.202001854
ChemCatChem
A
key prerequisite for the development of scalable
DMSO was used to generate Guo stock solutions that were
used as substrates for all enzyme reactions resulting in a DMSO
concentration of 1% v/v.
enzymatic synthesis is the availability of inexpensive substrates
in bulk amounts. Here, we report the development of two
cascades for the synthesis GDP-Fuc from inexpensive precur-
sors. The first cascade consists of five enzymes and seven
reactions for the synthesis of GDP-Fuc from Fuc, guanosine
(Guo), polyphosphate (PolyPn), and catalytic amounts of ATP.
However, due to the relatively high costs of Fuc compared to
the other substrates, a second pathway was constructed using
mannose (Man) as the sugar source. Therefore, a cascade
consisting of ten enzymes and eleven reactions was designed
and established to synthesize GDP-Fuc from Man, Guo, PolyPn,
L-glutamine (L-Glu), and catalytic amounts of NADPH and ATP.
In both cascades, dimethyl sulfoxide (DMSO) was used to
solubilize the Guo and avoid hydrogel formation among
guanosine-containing molecules. Moreover, an ion exchange
chromatography protocol was established to purify GDP-Fuc
from the reaction mix to purities comparable to commercial
standards. To the best of the authors’ knowledge, both
cascades presented herein are described for the first time.
Selection of enzymes
A literature search using the BRENDA enzyme database was
carried out to identify and select enzymes for the construction
of the pathway.[17] The criteria employed for selection were
overlapping pH and temperature activity ranges. The enzymes
selected are detailed in Table 1. Moreover, the catalyzed
reactions and activity parameters are shown as reported in the
literature.
Synthesis of GDP-Fuc from Fuc and Guo
Pathway design: The constructed pathway for the synthesis of
GDP-Fuc is illustrated in Figure 1. The cascade contains five
enzymes and seven reactions (see Table 1 regarding the
enzymes used in this work). There are three ATP-dependent
kinase reactions. Consequently, an excess amount of ATP is
needed to facilitate the full conversion of Fuc, Guo, and GMP.
As ATP is relatively expensive compared to Guo, a regeneration
cycle of ATP from ADP was established by exploiting the
promiscuity of the polyphosphate kinase (PPK3) towards
diphosphate nucleotides and by using PolyPn as the phosphate
Results
Solubility of Guo
Guo exhibits a very low solubility (~1.82 mM) in water.[16] Pure
dimethyl sulfoxide (DMSO) was found to be able to dissolve
°
°
Guo up to 1 M at 75 C and 0.5 M at 25 C. Consequently, pure
Table 1. Enzymes used in this study and their reported activity range.[a]
°
Enzyme
EC
Reaction
pH
T [ C]
Co-fact.
Mg2+
Ref.
Cascade 1: Enzymes for the synthesis of GDP-Fuc from Fuc and Guo
FKP
2.7.1.52
2.7.7.30
Fuc þATP !Fuc1P þADP
7.5
25–37
[11, 12, 18]
GTP þFuc1P ! GDP À Fuc þPPi
PPK3
2.7.4.1
GDP þPolyPn ! GTP þPolyPnÀ1
5–11
30–40
Mg2+
[19, 20]
ADP þPolyPn ! ATP þPolyPnÀ1
GSK
2.7.4.73
2.7.4.8
3.6.1.1
Guo þATP ! GMP þADP
GMP þATP ! GDP þADP
PPi ! 2 Pi
7.2
7.4
7–8
32
Mg2+
Mg2+
Mg2+
[21]
[22]
[20]
GMPK
PPA
30
25–35
Cascade 2: Additional enzymes for synthesis of GDP-Fuc from Man and Guo
WCAG
GLDH
GLK
1.1.1.271
1.4.1.2
GDP4dehydro6deoxyMan þNADPH ! GDP À Fuc þNADPþ
LGlut þNADPþ ! NADPH þNH3 þAKG
Man þATP ! Man6P þADP
GTP þMan1P ! GDPMan þPPi
GDPMan ! GDP4dehydro6deoxyMan þH2O
Man6P ! Man1P
7.5
8
37
Mg2+
[14]
[23]
[24]
[24]
[14]
[24]
37
2.7.1.2
7–8
7–8
7.5
7–8
25–35
25–35
37
Mg2+
Mg2+
Mg2+
Mg2+
MANC
GMD
2.7.7.13
4.2.1.47
5.4.2.8
MANB
25–35
[a] Abbreviations are as follows: GSK, guanosine-inosine kinase; GMPK, guanylate kinase; PPK3, polyphosphate kinase; MANC, mannose-1-phosphate
guanylyltransferase; MANB, phosphomannomutase; GLK, glucokinase; PPA, inorganic diphosphatase; FKP, bifunctional fucokinase/L-fucose-1-P-
guanylyltransferase; GMD, GDP-mannose 4,6-dehydratase; WCAG, GDP-L-fucose synthase; GLDH, glutamate dehydrogenase.
ChemCatChem 2021, 13, 1981–1989
www.chemcatchem.org
1982
© 2020 The Authors. ChemCatChem published by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/cctc.202001854
ChemCatChem
Figure 1. Cascade of five enzymes and seven reactions for the synthesis of
GDP-Fuc from Fuc, Guo, PolyPn, and a catalytic amount of ATP. The names of
the enzymes are in red. Abbreviations are as follows: PPi, diphosphate; Pi,
phosphate. Promiscuity of PPK3 allows regeneration of ATP and GTP by only
one enzyme.
source. In addition, inorganic diphosphatase (PPA) was used to
hydrolyze diphosphate and drive the reaction towards GDP-Fuc.
Synthesis of GDP-Fuc from Fuc and Guo: All reactions in the
cascade were performed at pH 7.5. The time course of reaction
intermediates is shown in Figure 2. A steep decline of Guo and
a production of GMP, GDP, GTP, and GDP-Fuc was observed.
GDP-Fuc was produced to a final concentration of 7 mM (4.1 g/
L) equivalent to a yield of 68% after 48 h from Guo and Fuc,
respectively. The biocatalyst load was 0.34 genzyme/gproduct. The
formation of GDP-Fuc was confirmed by MALDI-TOF (see
supplementary file (SI); Figure SI 5). As ATP was 12-fold less
abundant than the stoichiometric amount required, the results
also confirm the efficient in situ regeneration of ATP from ADP
and PolyPn.
Synthesis of GDP-Fuc from Man and Guo
Synthesis of GDP-Fuc from GDP-Man: To avoid using compara-
tively high cost Fuc as substrate, a second cascade was
established to synthesize GDP-Fuc from Man. Man is around 40-
fold cheaper than Fuc. First, a two-enzyme cascade containing
GDP-mannose-4,6-dehydratase (GMD) and GDP-L-Fucose syn-
thase (WCAG) was established to study the conversion of GDP-
Man to GDP-Fuc (see Figure 3). It is known that GDP-Fuc has an
inhibitory effect on GMD, resulting in low conversion yields.[25,26]
With the goal of improving the conversion yield, the role of
pH was investigated in a broader range (pH 7.0, 7.5, 8.0, 8.5,
and 9.0; Figure 4). It was observed that virtually full conversion
of GDP-Man to GDP-Fuc was achieved at pH�8.0. In contrast,
at pH 7.0 and 7.5, even a longer incubation time did not result
in higher conversion yields (data not shown).
Figure 2. Time course of reactions using cascade 1 (see Figure 1) to
synthesize GDP-Fuc. The reaction mixture for synthesis of GDP-Fuc consisted
of 200 mM Tris-HCl (pH 7.5), 10 mM Fuc, 10 mM Guo (in DMSO), 2.5 mM ATP,
7.5 mM PolyPn, 45 mM MgCl2, and the following enzymes: GSK (0.22 μg/μL),
GMPK (0.78 μg/μL), PPK3 (0.05 μg/μL), FKP (0.31 μg/μL), and PPA (0.03 μg/μL)
°
in a final volume of 200 μL and 37 C. The final DMSO content of the reaction
matrix was 1% v/v. (a) Shows the consumption of Guo and production of
GDP-Fuc, (b) shows the reaction time courses of GMP, GDP, and GTP, (c)
shows the reaction time courses of ATP, ADP, as well as AMP. The production
of AMP is mainly due to chemical conversion of ADP to ATP during the
reaction. Experiments were performed in triplicates and error bars represent
the standard error. The values of standard errors are listed in Table SI 2.
ChemCatChem 2021, 13, 1981–1989
www.chemcatchem.org
1983
© 2020 The Authors. ChemCatChem published by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/cctc.202001854
ChemCatChem
Figure 3. Cascade of two enzymes for the synthesis of GDP-Fuc from GDP-
Man.
Figure 4. Conversion of GDP-Man to GDP-Fuc after 5 h (end of the reactions)
°
with a two-enzyme cascade. The experiments were carried out at 37 C.
Concentrations were as follows: 150 mM Tris-HCl, 10 mM MgCl2, 3–4 mM
GDP-Man, 4 mM NADPH, WCAG (0.45 μg/μL), and GMD (1.03 μg/μL) in a the
final volume of 33 μL. Experiments were carried out in triplicates and error
bars represent the standard error.
Pathway design for synthesis of GDP-Fuc from Man and
Guo: The second cascade was developed to synthesize GDP-Fuc
from Man, Guo, PolyPn, L-Glu, and catalytic amounts of NADPH
and ATP (see Figure 5). L-Glu was used as a substrate for the
in situ regeneration of the expensive NADPH. Beforehand, the
synthesis of GDP-Man from Man, ATP, and Guo using the first
part of the cascade was verified (see SI - Figure SI 2).
Synthesis of GDP-Fuc from Man and Guo: Since a pH over
8.0 was necessary to achieve high yields of GDP-Fuc from GDP-
Man, a pH of 8.5 was selected to carry out the cascade
reactions. The time course of the reaction intermediates is
shown in Figure 6. The reaction starts with immediate con-
version of Guo to GMP, and further into GDP and GTP. The
Figure 6. Time course of reaction products and intermediates for the
synthesis of GDP-Fuc from Man, Guo, PolyPn, L-Glu, and catalytic amounts of
°
ATP and NADPH. The experiments were carried out at 37 C and 200 mM
Tris-HCl (pH 8.5), 75 mM MgCl2, 10.5 mM Man, 10.5 mM Guo, 50 mM L-Glu,
1 mM NADPH, 5.5 mM ATP, 13.5 mM PolyPn, GSK (0.11 μg/μL), GMPK
(0.49 μg/μL), PPK3 (0.02 μg/μL), GLK (0.33 μg/μL), MANB/C (0.17 μg/μL),
WCAG (0.07 μg/μL), GMD (0.17 μg/μL), PPA (0.03 PPA), and 10 units of GLDH
(2.99 μg/μL) in a final volume of 200 μL. The final DMSO content of the
reaction matrix was 1% v/v. (a) shows the consumption of Guo, the
production of GDP-Man and its consumption for synthesis of GDP-Fuc, (b)
shows the production of GMP followed by its consumption for production of
GDP and GTP, (c) shows the concentration of ATP, ADP as well as AMP; The
production of AMP is mainly due to chemical conversion of ADP to ATP.
Experiments were carried out in triplicates and error bars represent the
standard error. The values of standard errors are listed in Table SI 3.
Figure 5. Cascade 2 for the synthesis of GDP-Fuc from Man, Guo, PolyPn, L-
Glu, and catalytic amounts of ATP and NADPH. Man6-P, mannose 6-
phosphate; Man-1P, mannose 1-phosphate; H2O, water; GDP-4-dehydro-6-
deoxy-Man, guanosine diphosphate-4-keto-6-deoxy -mannose; NH3, ammo-
nia; AKG, α-ketoglutaric acid.
ChemCatChem 2021, 13, 1981–1989
www.chemcatchem.org
1984
© 2020 The Authors. ChemCatChem published by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/cctc.202001854
ChemCatChem
higher concentration of GDP-Fuc with regard to GDP-Man
during the reaction time demonstrates the high catalytic
activity of GMD and WCAG. GDP-Fuc was produced with a
concentration of 7.6 mM (4.5 g/L) and a reaction yield of 72%
after 48 h with a biocatalyst load of 0.97 genzyme/gproduct. ATP and
NADPH were used 5.7-fold and 10.5-fold less than stoichiomet-
ric amounts, respectively. The production of GDP-Fuc was
verified by MALD-TOF mass spectrometry (see SI - Figure SI 6).
Furthermore, to understand the effect of pH on the overall
performance of the cascade, pH values of 7.0, 7.5, 8.0, 8.5, and
9.0 were screened and it was found out that pH 8.5 and 9.0
resulted in the highest conversion yield (see SI - Figure SI 4).
as the previously injected purified standard. The GDP-Fuc was
further confirmed by injecting the cascade product spiked with
the purified commercial standard (not shown).
Due to the early elution of GDP-Fuc in the gradient at a
conductivity close to that of the equilibration buffer, it was
decided to increase the pH from 7.4 to 8.0 in order to achieve a
higher product retention. Another set of experiments with the
purified commercial standard and a 10-fold increase of loaded
cascade product (0.5 mL) was performed at pH 8.0 (Figure 7B)
with a modified gradient (step 1:15% desorption buffer over
25 CV; step 2:100% desorption buffer over 5 CV). As observed,
a conductivity exceeding 11 mS/cm was needed to desorb
GDP-Fuc loaded at pH 8.0 compared to around 7.6 mS/cm at
pH 7.4. Despite the expected change of the desorption profile
due to the higher loading, GDP-Fuc was fully resolved in the
gradient.
Chromatographic purification of the GDP-Fuc cascade
product
The purity of GDP-Fuc in the loaded cascade product was
estimated to be around 25% based on the chromatographic UV
signal at 260 nm (Figure 7B, bottom panel). The purified GDP-
Fuc was collected and re-injected using the same method. In
stark contrast to the cascade product, the purity of GDP-Fuc
after ion exchange chromatography was estimated to be 90.5%
(Figure 7C) and comparable to that of the commercial standard
(91.8%).
As a first screening step, purified commercially available GDP-
Fuc was injected into a strong quaternary amine anion ex-
changer column (HiTrap Q HP) to assess its capture and
subsequent elution with a gradient.
An injection of 5 μL of GDP-Fuc (7.9 mM) using 50 mM Tris
pH 7.4 resulted in successful capture of GDP-Fuc. It was possible
to recover the product by using a gradient elution at around
6% of desorption buffer (50 mM Tris, 1.0 M NaCl, pH 7.4) with a
conductivity of 7.6 mS/cm at the peak apex (Figure 7A, top
panel).
Discussion
Injection of a GDP-Fuc cascade product (50 μL) using the
same setup showed impurities in the flow-through and GDP-
Fuc was collected as a first and fully resolved peak in the
gradient at the same retention time (Figure 7A, bottom panel)
The objective of this study was to develop two multi-enzyme
cascades to synthesize GDP-Fuc from inexpensive and readily
available precursors. Guo is used in the food industry as raw
Figure 7. Chromatographic purification of GDP-Fuc produced in a multi-enzyme cascade by bead-based anion exchange chromatography. A 1 mL column
packed with a strong quaternary amine anion exchanger and a bead size of 34 μm was used. Panel A depicts preliminary scouting experiments with a purified
commercial GDP-Fuc standard and the cascade product at pH 7.4. Panel B shows an optimized gradient at pH 8.0 and a 10-fold load increase. GDP-Fuc was
identified as the second peak in the gradient from the cascade product. Panel C shows the purified GDP-Fuc cascade product with a purity grade (90.5%)
similar to that of the commercial standard (91.8%).
ChemCatChem 2021, 13, 1981–1989
www.chemcatchem.org
1985
© 2020 The Authors. ChemCatChem published by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/cctc.202001854
ChemCatChem
material to produce GMP as a flavor enhancer and is available
in bulk amounts.[27] However, using Guo in aqueous solutions is
limited by its hydrophobic nature, which results in low
solubility. Guo has unique self-assembly properties due to
multiple hydrogen bonding interactions resulting in hydrogel
formation.[28] DMSO was found to be able to solubilize Guo and
avoid hydrogel formation in aqueous solutions. This is due to
interactions between the alcohol, amide, amine, and ether as
functional groups of Guo are unfavourable in DMSO.[29] The
good solubility of the Guo-containing compounds is illustrated
by closed mass balances at each of the reaction measurement
time points in our experiments (see Figure 2 and Figure 6).
DMSO is especially suitable as a solvent due to its low toxicity,
i.e., it is widely used in the field of biotechnology as an
extractant and solvent as well as in medicine as a therapeutic to
reduce detumescence and as a carrier substance for drug
delivery.[30]
For the production of GDP-Fuc, at first a cascade was
established successfully to start the synthesis from Fuc (~4 E/
g), Guo (~0.1 E/g), PolyPn (0.05 E/g), and catalytic amounts of
ATP (~0.4 E/g). As Fuc is relatively expensive compared to the
other substrates, a second cascade was developed to produce
GDP-Fuc from Man (~0.1 E/g). However, the benefit of lower
substrate costs comes at the costs of using ten enzymes instead
of five plus an additional co-factor regeneration loop.
To avoid the stoichiometric usage of high-cost co-substrates
ATP and NADPH, in situ regeneration cycles were implemented.
ATP and NADPH are widely used co-substrates for the
enzymatic in vitro synthesis of chemicals. Thus, multiple exam-
ples for their in situ regeneration have been developed.[31] In
both cascades established in this study, ATP recycling from
PolyPn was implemented building on our previous work.[20,32,33]
In our second cascade, a NADPH recycling loop using L-Glu as
substrate was implemented that allowed using NADPH 10.5-
fold less than the stochiometric amount. L-Glu is used in the
food industry as a flavor enhancer and thus, readily available at
low costs (~0.07 E/g).[34]
The costs for the production of purified protein using E. coli
were previously estimated to be around 1.1 E/g.[35] Accordingly,
here the cost contribution of enzymes is around 1.5 E per liter
of reaction for cascade 1 and 4.7 E for cascade 2. Consequently,
the cost of enzymes can be estimated to be around 0.03 E and
1.1 E per gram of GDP-Fuc, respectively. Thus, compared to the
costs of substrates used, the cost contribution of the enzymes
to the total synthesis costs is minor (see Table SI 4 and 5).
From the experimental data, no single rate-limiting steps
consumption of hydrogen ions are part of the reaction
mechanism.[26,37] Once hydrogen ions are involved in a reaction,
dissociation constants of ligands and enzymes are a function of
the pH value.[38] Therefore, a pH screening was carried out to
evaluate its effect on the performance of the two-enzyme
cascade (GMD and WCAG) to synthesize GDP-fuc from GDP-
Man and NADPH. Interestingly, it was found that at alkaline
conditions in the range pH 8.0–9.0, an almost quantitative
conversion of GDP-Man to GDP-Fuc can be obtained. This is
possibly due to a low binding of GDP-Fuc to GMD at alkaline
pH values.
The performance of the two developed cascades is
compared to previously published enzymatic cascades in
Table 2. Pfeiffer et al. achieved a high concentration of 50 g/L
GDP-Man, however, GTP was used as substrate.[39] The high
costs of GTP (~21 E/g) can significantly hamper large-scale
application. In stark contrast, the utilization of Guo (~0.1 E/g)
or even GMP (~0.1 E/g), as shown here, substantially reduces
substrate costs around 200-fold.
Koizumi et al. obtained the highest reported GDP-Fuc
concentration (18.4 g/L) using whole cell catalysis at a 15 L scale
by using Man and GMP (see Table 2).[10] However, drawbacks
are low synthesis yields (17% and 52% regarding Man and
GMP, respectively) and very high biocatalyst loads (215 g/L) in
the form of four different permeabilized microbial cells. More-
over, to avoid low conversions of GDP-Man to GDP-Fuc, the
reactions were carried out in two separate reaction vessels.
It was possible to purify the product of our GDP-Fuc
cascades by ion exchange chromatography. This method is one
of the most widely used purification techniques in industry
thanks to its robustness, scalability, and costs.[40] Here, it was
possible to fully resolve the GDP-Fuc product using a gradient
elution and its purity was increased from 25% to 90.5%,
matching that of
a commercial standard (91.8%). These
preliminary results demonstrate that GDP-Fuc produced in a
multi-enzyme cascade can be efficiently purified in an inex-
pensive manner with readily available industrial materials using
a simple method.
To further reduce costs prior to scaling up the GDP-Fuc
synthesis, an optimization of the amounts of enzymes, sub-
strates and co-substrates used should be carried out through a
more comprehensive screening of reaction conditions. To keep
experimental work to a minimum, a design of experiments
approach seems especially wellsuited for this purpose.
were identified. However, the turnover of Guo to GMP is Conclusion
completed after a maximum of 3 h in both cascades and thus,
the concentration of GSK can be reduced to some extent
without severely affecting the productivity in future reactions.
Previous work on the enzymatic one-pot production of
GDP-Fuc from GDP-Man showed low conversion yields due to
inhibition of GMD by GDP-Fuc.[10,14] It has been suggested that
GDP-Fuc is a competitive inhibitor of GMD by binding to the
active site.[25,26,36] Both enzymes, GMD and WCAG, belong to the
short-chain dehydrogenases/reductases family and during the
conversion of GDP-Man to GDP-Fuc, both liberation and
Fuc is an important building block for many types of
oligosaccharides such as HMOs. The Fuc donor for enzymatic
assembly of fucosylated oligosaccharides is the activated
nucleotide sugar GDP-Fuc. Until now, high costs of GDP-Fuc
have hampered the enzymatic in vitro synthesis of fucosylated
oligosaccharides beyond the milligram scale. In this work, two
multi-enzyme cascades were established to synthesize GDP-Fuc
from low-cost and readily available precursors. In both cascades,
Guo was used as a substrate by employing DMSO as a co-
ChemCatChem 2021, 13, 1981–1989
www.chemcatchem.org
1986
© 2020 The Authors. ChemCatChem published by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/cctc.202001854
ChemCatChem
ChemCatChem 2021, 13, 1981–1989
www.chemcatchem.org
1987
© 2020 The Authors. ChemCatChem published by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/cctc.202001854
ChemCatChem
GLDH (2.99 μg/μL) in a final volume of 200 μL. The final DMSO
content of the reaction matrix was 1% v/v.
solvent to avoid hydrogel formation in the reaction matrix. At
first, a cascade of five enzymes and seven reactions was
developed to synthesize GDP-Fuc from Guo, Fuc, PolyPn and
catalytic amounts of ATP. GDP-Fuc was produced with a final
concentration of 7 mM (4.1 g/L) and a reaction yield of 68%
from Guo and Fuc in a batch process of 48 h and a biocatalyst
load of 0.34 genzyme/gproduct. To further reduce the costs of
substrates, a second cascade that uses Man instead of Fuc as
the sugar precursor and consists of ten enzymes and eleven
reactions was developed. Here, GDP-Fuc was produced with a
final concentration of 7.6 mM (4.5 g/L) and a reaction yield of
72% from Guo, Man, PolyPn, L-Glu, and catalytic amounts of
Chromatographic purification of GDP-Fuc with a bead-based
anion exchanger. For the purification of the GDP-Fuc product from
the enzymatic cascades, an ÄKTA Pure 25 liquid chromatography
system was used at room temperature and controlled by the
software UNICORN v6.3 (Cytiva; Uppsala, Sweden). The UV
absorbance was monitored at 260 nm. The column used was a pre-
packed 1 mL strong anion exchanger (HiTrap Q HP; 29-0513-25;
Cytiva; Uppsala, Sweden) with a bead size of 34 μm. The column
was operated at 1–4 mL/min (155–624 cm/h). GDP-Fuc purification
was performed in bind-elute mode. In short, (A) Equilibration: after
a washing step of 10 column volumes (CV) with water, the column
was equilibrated with equilibration buffer (10 mM Tris, pH 8.0) for
10 CV at 3 mL/min. (B) Sample injection: the 0.2 μm clarified
cascade product was diluted at least 10-fold in equilibration buffer
and fed to the column at 2 mL/min. After sample injection, a wash
step (2 mL/min) followed with equilibration buffer for 10 CV or until
baseline UV absorbance was achieved. (C) Elution: the bound
compounds were eluted at 1 mL/min using a combination of
different steps and gradients, most notably a first step of 15%
desorption buffer (10 mM Tris, 1.0 M NaCl, pH 8.0) for 25 CV
followed by a second step to 100% desorption buffer for 5 CV. The
column was finally stripped with 2 M NaCl at 3 mL/min for 10 CV
and re-equilibrated for subsequent runs.
ATP and NADPH after 48 h with
a biocatalyst load of
0.97 genzyme/gproduct. Finally, a robust and scalable method for
GDP-Fuc purification by ion exchange chromatography was
established achieving purities equivalent to the quality of
commercial standards. In conclusion, the developed enzyme
cascades and the purification method enable the efficient
synthesis and purification of GDP-Fuc significantly below
current production costs.
Experimental Section
Material and methods. Chemicals and standard methods for
analytics are described in the supplementary file (SI). All prices are
the list prices as of October 2020 from Carbosynth Ltd (Compton,
United Kingdom) with the exception of PolyPn for which the list
price was taken from Merck (Darmstadt, Germany). Enzymes were
expressed in E. coli and purified using ion metal affinity chromatog-
raphy as detailed in SI.
Notes
RM and TFTR are the inventors of a pending patent on the
described topic.
Multi-enzyme experiments. All enzymatic reactions were per-
formed in 1.5 mL Eppendorf safe-lock tubes (Eppendorf, Germany)
Acknowledgements
°
at 37 C and 550 rpm in Eppendorf Thermomixer comfort incuba-
tors (Eppendorf, Germany). For reaction time course measurements,
The experimental assistance of Claire Telfer for preliminary
experiments is acknowledged. Valerian Grote is thanked for advice
on and help with MALDI-MS measurements. Furthermore, the
authors would like to thank Anna Schildbach and Markus Pietzsch
from the Martin Luther University of Halle-Wittenberg for
providing the plasmids for GSK, GLK, MANB/C, and PPA. The
authors would further like to thank Simon Boecker from the
research group of Analysis and Redesign of Biological Networks at
the MPI Magdeburg for fruitful discussions on fermentation and
Francesca Cascella from the Department of Phyiscal and Chemical
Foundation of Process Engineering at the MPI Magdeburg for
valuable discussions on the solubilization properties of DMSO.
Open access funding enabled and organized by Projekt DEAL.
°
aliquots were taken and quenched at 90 C in MiliQ water for 3 min.
All initial concentrations given below are calculated from weighted
samples and might slightly deviate from measured initial concen-
trations.
First cascade starting from Fuc and Guo: The reaction mixture for
synthesis of GDP-Fuc contained 200 mM Tris-HCl (pH 7.5), 10 mM
Fuc, 10 mM Guo (in DMSO), 2.5 mM ATP, 7.5 mM PolyPn, 45 mM
MgCl2, and the following enzymes: GSK (0.22 μg/μL), GMPK
(0.78 μg/μL), PPK3 (0.05 μg/μL), FKP (0.31 μg/μL), and PPA (0.03 μg/
μL) in a final volume of 200 μL. The final DMSO content of the
reaction matrix was 1% v/v.
Second cascade starting from Man and Guo-Production of GDP-Fuc
from GDP-Man: The reaction mixtures contained 150 mM Tris-HCl
(various pH values), 10 mM MgCl2, 3–4 mM GDP-Man, 4 mM
NADPH, WCAG (0.45 μg/μL), and GMD (1.03 μg/μL) in a final volume
of 33 μL. To evaluate the role of the pH value on the conversion of Conflict of Interest
GDP-Man to GDP-Fuc, the following pH values were tested: 7.0, 7.5,
8.0, 8.5, 9.0.
The authors declare no conflict of interest.
Second cascade starting from Man and Guo-Production of GDP-Fuc
from Man and Guo with in situ regeneration of NADPH: The cascade
reactions contained 200 mM Tris-HCl (pH 8.5), 75 mM MgCl2,
10.5 mM Man, 10.5 mM Guo, 50 mM L-Glu, 1 mM NADPH, 5.5 mM
ATP, 13.5 mM PolyPn, GSK (0.11 μg/μL), GMPK (0.49 μg/μL), PPK3
(0.02 μg/μL), GLK (0.33 μg/μL), MANB/C (0.17 μg/μL), WCAG
(0.07 μg/μL), GMD (0.17 μg/μL), PPA (0.03 μg/μL), and 10 units of
Keywords: Multi-enzyme
nucleotides · GDP-Fuc · ATP regeneration · NADPH regeneration
cascade
reactions
·
sugar
[1] a) L. Bode, Glycobiology 2012, 22, 1147–1162; b) S. S. Comstock, S. M.
Donovan, Prebiotics and Probiotics in Human Milk (Eds.: M. K. McGuire,
ChemCatChem 2021, 13, 1981–1989
www.chemcatchem.org
1988
© 2020 The Authors. ChemCatChem published by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/cctc.202001854
ChemCatChem
M. A. McGuire, L. Bode), Academic Press, San Diego, 2017, pp. 223–248;
[22] G. Hible, L. Renault, F. Schaeffer, P. Christova, A. Z. Radulescu, C. Evrin,
A.-M. Gilles, J. Cherfils, J. Mol. Biol. 2005, 352, 1044–1059.
[23] J. R. Treberg, M. E. Brosnan, J. T. Brosnan, Mol. Cell. Biochem. 2010, 344,
253–259.
c) A. L. Morrow, Y. Yu, Prebiotics and Probiotics in Human Milk (Eds.: M. K.
McGuire, M. A. McGuire, L. Bode), Academic Press, San Diego, 2017, pp.
207–222.
[2] P. K. Berger, J. F. Plows, R. B. Jones, T. L. Alderete, C. Yonemitsu, M.
Poulsen, J. H. Ryoo, B. S. Peterson, L. Bode, M. I. Goran, PLoS One 2020,
15, e0228323.
[3] a) K. Bych, M. H. Mikš, T. Johanson, M. J. Hederos, L. K. Vigsnæs, P.
Becker, Curr. Opin. Biotechnol. 2019, 56, 130–137; b) K. Bych, Mik, Curr.
Opin. Biotechnol. 2019, 56, 130–137; c) L. Bode, S. Campbell, R. Furneaux,
J. Beauprez, A. Muscroft-Taylor, 2017, 251–293.
[24] T. F. T. Rexer, A. Schildbach, J. Klapproth, A. Schierhorn, R. Mahour, M.
Pietzsch, E. Rapp, U. Reichl, Biotechnol. Bioeng. 2018, 115, 192–205.
[25] a) R. H. Kornfeld, Biochimica et Biophysica Acta (BBA)-General Subjects
1966, 117, 79–87; b) L. Sturla, A. Bisso, D. Zanardi, U. A. Benatti, FEBS
Lett. 1997, 412, 126–130.
[26] J. R. Somoza, S. Menon, H. Schmidt, D. Joseph-McCarthy, A. Dessen,
M. L. Stahl, W. S. Somers, F. X. Sullivan, Structure 2000, 8, 123–135.
[27] K. I. Miyagawa, H. Kimura, K. Nakahama, M. Kikuchi, M. Doi, S. Akiyama,
Y. Nakao, Biotechnology. (N. Y.) 1986, 4, 225–228.
[4] M. Faijes, M. Castejón-Vilatersana, C. Val-Cid, A. Planas, Biotechnol. Adv.
2019, 37, 667–697.
[5] B. Nidetzky, A. Gutmann, C. Zhong, ACS Catal. 2018, 6283–6300.
[6] a) Z. Xiao, Y. Guo, Y. Liu, L. Li, Q. Zhang, L. Wen, X. Wang, S. M.
Kondengaden, Z. Wu, J. Zhou, X. Cao, X. Li, C. Ma, P. G. Wang, J. Org.
Chem. 2016, 81, 5851–5865; b) C.-H. Wong, T.-I. Tsai, C.-Y. Wu, 2016; c) L.
Mestrom, M. Przypis, D. Kowalczykiewicz, A. Pollender, A. Kumpf, S. R.
Marsden, I. Bento, A. B. Jarzbski, K. Szymaska, A. Chruciel, Int. J. Mol. Sci.
2019, 20, 5263; d) A. R. Prudden, L. Liu, C. J. Capicciotti, M. A. Wolfert, S.
Wang, Z. Gao, L. Meng, K. W. Moremen, G.-J. Boons, Proc. Natl. Acad. Sci.
USA 2017, 114, 6954–6959; e) T. Li, L. Liu, N. Wei, J.-Y. Yang, D. G.
Chapla, K. W. Moremen, G.-J. Boons, Nat. Chem. 2019, 11, 229–236.
[7] L. Bode, S. Campbell, R. Furneaux, J. Beauprez, A. Muscroft-Taylor,
Prebiotics and Probiotics in Human Milk (Eds.: M. K. McGuire, M. A.
McGuire, L. Bode), Academic Press, San Diego, 2017, pp. 251–293.
[8] S. Chang, B. Duerr, G. Serif, J. Biol. Chem. 1988, 263, 1693–1697.
[9] C. Albermann, J. Distler, W. Piepersberg, Glycobiology 2000, 10, 875–881.
[10] S. Koizumi, T. Endo, K. Tabata, H. Nagano, J. Ohnishi, A. Ozaki, J. Ind.
Microbiol. Biotechnol. 2000, 25, 213–217.
[28] T. Bhattacharyya, P. Saha, J. Dash, ACS Omega 2018, 3, 2230–2241.
[29] C. A. Hunter, Angew. Chem. Int. Ed. 2004, 43, 5310–5324; Angew. Chem.
2004, 116, 5424–5439.
[30] a) E. M. Kaiser, R. D. Beard, C. R. Hauser, J. Organomet. Chem. 1973, 59,
53–64; b) P. M. Osterberg, J. K. Niemeier, C. J. Welch, J. M. Hawkins, J. R.
Martinelli, T. E. Johnson, T. W. Root, S. S. Stahl, Org. Process Res. Dev.
2015, 19, 1537–1543.
[31] a) T. Shi, P. Han, C. You, Y.-H. P. J. Zhang, Synthetic and Systems
Biotechnology 2018, 3, 186–195; b) E. E. Ferrandi, D. Monti, S. Riva,
Cascade Biocatalysis: Integrating Stereoselective and Environmentally
Friendly Reactions, Vol. 9783527335220, 2014, pp. 23–42; c) W. Hummel,
H. Gröger, J. Biotechnol. 2014, 191, 22–31; d) H. Taniguchi, K. Okano, K.
Honda, Synthetic and Systems Biotechnology 2017, 2, 65–74.
[32] T. F. T. Rexer, A. Schildbach, J. Klapproth, A. Schierhorn, R. Mahour, M.
Pietzsch, E. Rapp, U. Reichl, Biotechnol. Bioeng. 2018, 115, 192–205.
[33] T. F. T. Rexer, L. Wenzel, M. Hoffmann, S. Tischlik, C. Bergmann, V. Grote,
S. Boecker, K. Bettenbrock, A. Schildbach, R. Kottler, R. Mahour, E. Rapp,
M. Pietzsch, U. Reichl, J. Biotechnol. 2020, 322, 54–65.
[11] M. J. Coyne, B. Reinap, M. M. Lee, L. E. Comstock, Science 2005, 307,
1778–1781.
[34] V. F. Wendisch, D. Eberhardt, M. Herbst, J. V. K. Jensen, 2016.
[35] P. Tufvesson, J. Lima-Ramos, M. Nordblad, J. M. Woodley, Org. Process
Res. Dev. 2011, 15, 266–274.
[36] F. X. Sullivan, R. Kumar, R. Kriz, M. Stahl, G.-Y. Xu, J. Rouse, X.-j. Chang, A.
Boodhoo, B. Potvin, D. A. Cumming, J. Biol. Chem. 1998, 273, 8193–8202.
[37] B. Persson, Y. Kallberg, J. E. Bray, E. Bruford, S. L. Dellaporta, A. D. Favia,
R. G. Duarte, H. Jrnvall, K. L. Kavanagh, N. Kedishvili, Chem.-Biol. Interact.
2009, 178, 94–98.
[12] G. Zhao, W. Guan, L. Cai, P. G. Wang, Nat. Protoc. 2010, 5, 636–646.
[13] J. B. McArthur, H. Yu, X. Chen, ACS Catal. 2019, 9, 10721–10726; H. Yu, Y.
Li, Z. Wu, L. Li, J. Zeng, C. Zhao, Y. Wu, N. Tasnima, J. Wang, H. Liu,
Chem. Commun. 2017, 53, 11012–11015.
[14] W. Wang, F. Zhang, Y. Wen, Y. Hu, Y. Yuan, M. Wei, Y. Zhou, AMB Express
2019, 9, 1–8.
[15] a) W.-H. Lee, Y.-W. Chin, N. S. Han, M.-D. Kim, J.-H. Seo, Appl. Microbiol.
Biotechnol. 2011, 91, 967; b) L. Li, S.-A. Kim, J. E. Heo, T.-J. Kim, J.-H. Seo,
N. S. Han, J. Biotechnol. 2017, 264, 1–7; c) S.-G. Byun, M.-D. Kim, W.-H.
Lee, K.-J. Lee, N. S. Han, J.-H. Seo, Appl. Microbiol. Biotechnol. 2007, 74,
768–775; d) W.-H. Lee, N.-S. Han, Y.-C. Park, J.-H. Seo, Bioresour. Technol.
2009, 100, 6143–6148.
[38] R. A. Alberty, J. Phys. Chem. B 2000, 104, 9929–9934.
[39] M. Pfeiffer, D. Bulfon, H. Weber, B. Nidetzky, Adv. Synth. Catal. 2016, 358,
3809–3816.
[40] H. Schmidt-Traub, M. Schulte, A. Seidel-Morgenstern, Prep. Chromatogr.,
Wiley Online Library, 2012.
[16] a) T. T. Herskovits, J. J. Bowen, Biochemistry 1974, 13, 5474–5483; b) T. T.
Herskovits, J. P. Harrington, Biochemistry 1972, 11, 4800–4811; c) H.-J.
Hinz, Thermodynamic data for biochemistry and biotechnology, Springer
Science \& Business Media, 2012.
[41] S. Fey, L. Elling, U. Kragl, Carbohydr. Res. 1997, 305, 475–481.
[42] L. Li, Y. Liu, Y. Wan, Y. Li, X. Chen, W. Zhao, P. G. Wang, Org. Lett. 2013,
15, 5528–5530.
[43] G. Zhao, W. Guan, L. Cai, P. G. Wang, Nat. Protoc. 2010, 5, 636.
[17] a) I. Schomburg, L. Jeske, M. Ulbrich, S. Placzek, A. Chang, D.
Schomburg, J. Biotechnol. 2017, 261, 194–206; b) I. Schomburg, A.
Chang, D. Schomburg, Nucleic Acids Res. 2002, 30, 47–49.
[18] H. Ohashi, C. Wahl, T. Ohashi, L. Elling, K. Fujiyama, Adv. Synth. Catal.
2017, 359, 4227–4234.
Manuscript received: November 17, 2020
Revised manuscript received: December 22, 2020
Accepted manuscript online: December 23, 2020
Version of record online: March 3, 2021
[19] J. Nahlka, Ptoprst, Organic \& biomolecular chemistry 2009, 7, 1778–
1780.
[20] R. Mahour, J. Klapproth, T. F. T. Rexer, A. Schildbach, S. Klamt, M.
Pietzsch, E. Rapp, U. Reichl, J. Biotechnol. 2018, 283, 120–129.
[21] H. Kawasaki, Y. Usuda, M. Shimaoka, T. Utagawa, Biosci. Biotechnol.
Biochem. 2000, 64, 2259–2261.
Correction added on March 9, 2021, after first online publication: Abbrevia-
tions of compounds in Table 1 were updated.
ChemCatChem 2021, 13, 1981–1989
www.chemcatchem.org
1989
© 2020 The Authors. ChemCatChem published by Wiley-VCH GmbH