Synthesis of Human Milk Oligosaccharides: Protein Engineering Strategies for Improved Enzymatic Transglycosylation

Abstract

Human milk oligosaccharides (HMOs) signify a unique group of oligosaccharides in breast milk, which is of major importance for infant health and development. The functional benefits of HMOs create an enormous impetus for biosynthetic production of HMOs for use as additives in infant formula and other products. HMO molecules can be synthesized chemically, via fermentation, and by enzymatic synthesis. This treatise discusses these different techniques, with particular focus on harnessing enzymes for controlled enzymatic synthesis of HMO molecules. In order to foster precise and high-yield enzymatic synthesis, several novel protein engineering approaches have been reported, mainly concerning changing glycoside hydrolases to catalyze relevant transglycosylations. The protein engineering strategies for these enzymes range from rationally modifying specific catalytic residues, over targeted subsite −1 mutations, to unique and novel transplantations of designed peptide sequences near the active site, so-called loop engineering. These strategies have proven useful to foster enhanced transglycosylation to promote different types of HMO synthesis reactions. The rationale of subsite −1 modification, acceptor binding site matching, and loop engineering, including changes that may alter the spatial arrangement of water in the enzyme active site region, may prove useful for novel enzyme-catalyzed carbohydrate design in general.

Full text

molecules
Review
Synthesis of Human Milk Oligosaccharides: Protein
Engineering Strategies for Improved
Enzymatic Transglycosylation
Birgitte Zeuner , David Teze , Jan Muschiol and Anne S. Meyer *
Protein Chemistry and Enzyme Technology, Department of Biotechnology and Biomedicine, Technical University
of Denmark, 2800 Kgs Lyngby, Denmark; bzeu@dtu.dk (B.Z.); davtez@dtu.dk (D.T.); jmus@dtu.dk (J.M.)
*Correspondence: asme@dtu.dk; Tel.: +45-45252600
Academic Editor: Ramón J. Estévez Cabanas
Received: 30 April 2019; Accepted: 26 May 2019; Published: 28 May 2019


Abstract:
Human milk oligosaccharides (HMOs) signify a unique group of oligosaccharides in breast
milk, which is of major importance for infant health and development. The functional benefits of
HMOs create an enormous impetus for biosynthetic production of HMOs for use as additives in
infant formula and other products. HMO molecules can be synthesized chemically, via fermentation,
and by enzymatic synthesis. This treatise discusses these different techniques, with particular focus
on harnessing enzymes for controlled enzymatic synthesis of HMO molecules. In order to foster
precise and high-yield enzymatic synthesis, several novel protein engineering approaches have been
reported, mainly concerning changing glycoside hydrolases to catalyze relevant transglycosylations.
The protein engineering strategies for these enzymes range from rationally modifying specific catalytic
residues, over targeted subsite
−
1 mutations, to unique and novel transplantations of designed
peptide sequences near the active site, so-called loop engineering. These strategies have proven
useful to foster enhanced transglycosylation to promote different types of HMO synthesis reactions.
The rationale of subsite
−
1 modification, acceptor binding site matching, and loop engineering,
including changes that may alter the spatial arrangement of water in the enzyme active site region,
may prove useful for novel enzyme-catalyzed carbohydrate design in general.
Keywords:
human milk oligosaccharides; transglycosylation; protein engineering; fucosidase;
sialidase; β-N-acetylhexosaminidase; transfucosylation; transsialylation; casein glycomacropeptide
1. Introduction
Human milk oligosaccharides (HMOs) denote a group of lactose-based carbohydrate structures
in human breast milk, which are considered to exert health benefits on the breast-fed baby via various
mechanisms. HMOs are present in human milk at concentrations of 5
−
15 g/L, which makes HMOs an
abundant component of human milk [
1
]. In contrast, the concentration and the variety of HMO-identical
structures are much lower in bovine milk, which is the basis of infant formula [
2
,
3
]. HMOs are critically
important for early development and infant health since they function as prebiotics and antimicrobial
agents in the gut of breastfed infants. They further protect the infant against pathogens by functioning
as soluble decoy receptors for pathogen adhesion as well as through a number of immunomodulating
effects [
1
,
4
]. In a recent study, purified HMOs have also been shown to exert beneficial effects in
adults [
5
], thus widening the potential applications and business opportunities of industrially produced
HMOs. No single HMO has all these effects alone, suggesting different roles for the over 150 different
HMO structures that have been identified in human milk [
6
,
7
]. All HMO structures are variations of
the specific HMO blueprint pattern composed of (up to) five different monosaccharides, always in
their pyranose form and in the same anomeric configuration:
β
-d-galactose (Gal),
β
-d-glucose (Glc),
Molecules 2019,24, 2033; doi:10.3390/molecules24112033 www.mdpi.com/journal/molecules

Molecules 2019,24, 2033 2 of 22
β
-d-N-acetyglucosamine (GlcNAc),
α
-l-fucose (Fuc), and the sialic acid
α
-d-N-acetylneuraminic acid
(Sia) (Figure 1). The reducing end of HMOs is lactose (Lac; Gal-
β
1,4-Glc). Lac can be elongated at the
Gal O-3 with lacto-N-biose (Gal-
β
1,3-GlcNAc), which prevents further elongation. The Gal moiety of
Lac can also be elongated by
β
-N-acetyllactosamine (LacNAc; Gal-
β
1,4-GlcNAc) units either at O-3 or
O-6, opening it up for further extensions. Finally, Gal, Glc, and GlcNAc residues can be fucosylated or
sialylated: Gal with Fuc at O-2 or O-3 as well as with Sia at O-3 and O-6, GlcNAc with Fuc at either
O-3 or O-4 and with Sia at O-6, and Glc only by Fuc at O-3 (Figure 1) [
1
]. In summary, five different
monosaccharides and 10 different types of linkages make up the entire plethora of HMO structures.
The specific blueprint patterns make it possible to achieve synthesis of a range of HMO structures
through targeting just a few different enzymatic activities either in vivo or in vitro.
Molecules 2019, 24, x 2 of 22
structures are variations of the specific HMO blueprint pattern composed of (up to) five different
monosaccharides, always in their pyranose form and in the same anomeric configuration: β-D-
galactose (Gal), β-D-glucose (Glc), β-D-N-acetyglucosamine (GlcNAc), α-L-fucose (Fuc), and the sialic
acid α-D-N-acetylneuraminic acid (Sia) (Figure 1). The reducing end of HMOs is lactose (Lac; Gal-
β1,4-Glc). Lac can be elongated at the Gal O-3 with lacto-N-biose (Gal-β1,3-GlcNAc), which prevents
further elongation. The Gal moiety of Lac can also be elongated by β-N-acetyllactosamine (LacNAc;
Gal-β1,4-GlcNAc) units either at O-3 or O-6, opening it up for further extensions. Finally, Gal, Glc,
and GlcNAc residues can be fucosylated or sialylated: Gal with Fuc at O-2 or O-3 as well as with Sia
at O-3 and O-6, GlcNAc with Fuc at either O-3 or O-4 and with Sia at O-6, and Glc only by Fuc at O-
3 (Figure 1) [1]. In summary, five different monosaccharides and 10 different types of linkages make
up the entire plethora of HMO structures. The specific blueprint patterns make it possible to achieve
synthesis of a range of HMO structures through targeting just a few different enzymatic activities
either in vivo or in vitro.
Figure 1. Human milk oligosaccharide (HMO) blueprint structure [1]. Gal: galactose, Glc: glucose,
GlcNAc: N-acetylglucosamine, Fuc: fucose, Sia: sialic acid (N-acetylneuraminic acid). Lactose is at the
reducing end of all HMO structures, which may be elongated with β-N-acetyllactosamine (LacNAc)
or lacto-N-biose units. Both lactose and elongated structures may be decorated with Fuc and/or Sia.
The colored shapes indicate the Symbol Nomenclature for Glycans (SNFG [8],
https://www.ncbi.nlm.nih.gov/glycans/snfg.html), which is commonly used for presenting the
numerous HMO structures.
While over 150 HMO structures exist, only 2′-fucosyllactose (2′-FL) and lacto-N-neotetraose
(LNnT) are currently commercially available for addition to infant formula [9]. Microbial engineering
work has recently made it possible to produce these two compounds in industrial scale by
fermentation of genetically modified Escherichia coli [9,10]. Several years of work on
commercialization and regulatory approval of synthesized HMOs has now paved the way for
expanding the HMO portfolio for future innovative food products beyond infant formula [9,10]. 2′-
FL was an obvious starting point for HMO production as it is the most abundant HMO [11,12] and
has a simple structure. In contrast, LNnT is less abundant both on its own and as an HMO core, and
is also present in human milk in lower levels than, e.g., lacto-N-tetraose (LNT) [11,12], but it appears
that LNnT is easier to synthesize in large scale and was therefore marketed first [9,13,14]. Indeed, for
more complex and larger structures, fermentation yields are often low [13,15].
Based on currently available in vitro and in vivo studies, it is likely that the putative synergistic
effect of numerous different HMOs can provide additional benefits in terms of health maintenance
and microbiota composition of infants and adults [4,5,16,17]. Thus, it is crucial to include as many
different HMO structures as possible in research of their bioactivity and health effects as well as in
food supplementation. Two recent studies have indicated that the more complex fucosylated and
sialylated HMOs had a larger antimicrobial effect on certain group B Streptococcus strains than
Figure 1.
Humanmilkoligosaccharide(HMO)blueprintstructure [
1
]. Gal: galactose, Glc: glucose, GlcNAc:
N-acetylglucosamine, Fuc: fucose, Sia: sialic acid (N-acetylneuraminic acid). Lactose is at the reducing end
of all HMO structures, which may beelongated with
β
-N-acetyllactosamine(LacNAc) or lacto-N-bioseunits.
Both lactose and elongated structures may be decorated with Fuc and/or Sia. The colored shapes indicate
the Symbol Nomenclature for Glycans (SNFG [
8
], https://www.ncbi.nlm.nih.gov/glycans/snfg.html), which
is commonly used for presenting the numerous HMO structures.
While over 150 HMO structures exist, only 2
0
-fucosyllactose (2
0
-FL) and lacto-N-neotetraose
(LNnT) are currently commercially available for addition to infant formula [
9
]. Microbial engineering
work has recently made it possible to produce these two compounds in industrial scale by fermentation
of genetically modified Escherichia coli [
9
,
10
]. Several years of work on commercialization and regulatory
approval of synthesized HMOs has now paved the way for expanding the HMO portfolio for future
innovative food products beyond infant formula [
9
,
10
]. 2
0
-FL was an obvious starting point for HMO
production as it is the most abundant HMO [
11
,
12
] and has a simple structure. In contrast, LNnT
is less abundant both on its own and as an HMO core, and is also present in human milk in lower
levels than, e.g., lacto-N-tetraose (LNT) [
11
,
12
], but it appears that LNnT is easier to synthesize in
large scale and was therefore marketed first [
9
,
13
,
14
]. Indeed, for more complex and larger structures,
fermentation yields are often low [13,15].
Based on currently available
in vitro
and
in vivo
studies, it is likely that the putative synergistic
effect of numerous different HMOs can provide additional benefits in terms of health maintenance
and microbiota composition of infants and adults [
4
,
5
,
16
,
17
]. Thus, it is crucial to include as many
different HMO structures as possible in research of their bioactivity and health effects as well as in food
supplementation. Two recent studies have indicated that the more complex fucosylated and sialylated
HMOs had a larger antimicrobial effect on certain group B Streptococcus strains than fucosylated or
sialylated lactose, and that the location and degree of fucosylation and sialylation play a key role in the
antimicrobial activity of HMOs [16,17].
Currently, many HMOs are not available in sufficient quantities and there is no efficient route to
their production. The status of enzymatic HMO production—both
in vitro
and in cell factories—was

Molecules 2019,24, 2033 3 of 22
recently thoroughly reviewed [
15
], and while great progress has been made over the past two
decades, the discrepancy between human milk composition and currently obtainable HMOs is evident,
providing a continued impetus to produce a wider span of true HMO structures by controlled
enzymatic synthesis. One way to accomplish production of larger and more complex HMOs is to
employ transglycosylation catalyzed by glycoside hydrolases (or glycosidases, GHs), possibly in
combination with use of fermentation-derived backbone structures such as LNnT or LNT as acceptor
substrates for the enzymatic glycosylation. This review summarizes the various HMO production
methods and focuses particularly on glycosidase-catalyzed transglycosylation, since this technology
appears to be the most promising to complement microbial cell factories (the current industrial
technology for production of a few HMO structures) in the quest to expand the industrial production
to cover the majority of the naturally occurring HMO portfolio. Glycosidases are important enzymes
for industrial-scale glycan synthesis due to their ability to use naturally occurring glycans, e.g., from
major agro-industrial side streams, as glycosyl donor substrates. An example of current industrial
use of glycosidases for transglycosylation is the production of prebiotic fructo-oligosaccharides (FOS;
the basic structure is a terminal Glc unit
α
1,2-linked to a linear chain of two or more
β
2,1-linked
fructose moieties) and galacto-oligosaccharides (GOS; a family of structures comprising two or more
Gal units linked by
β
-glycosidic bonds, often with a terminal Glc moiety) [
18
–
20
]. Today, both FOS
and GOS are added to infant formula, although they are not true HMOs and have been classified
as unnecessary in infant formula [
10
]. This review focuses on reports of enzymatic HMO synthesis
from abundant natural substrates. A major focal point of this review is different strategies to improve
enzymatic transglycosylation through rational protein engineering, even if the synthesized products
are non-HMO oligosaccharides. Protein engineering to attain enhanced transglycosylation is currently
manifesting itself as a valuable tool for improving enzyme performance in transglycosylation reactions
and a crucial means of obtaining feasible processes.
2. Routes to HMO Production Outside the Mammary Gland
While breast milk donation programs exist as part of health care services, especially for premature
infants, this natural source can by no means cover the demands of the industry seeking to add
HMOs to infant formula in order to supplement formula-fed infants with these beneficial components.
Instead, various technologies are in play in pursuit of a viable industrial process for HMO production:
fermentation of microbial cell factories, chemical synthesis, and a number of different enzymatic
in vitro reactions.
2.1. Microbial Cell Factories
In 2015 and 2016, the first two HMOs were marketed in the US and Europe in an intense
competition between American Abbott Laboratories, German Jennewein Biotechnologie GmbH, and
Danish Glycom A/S [
9
,
10
]. Following decades of research into metabolic engineering, 2
0
-FL and LNnT
were the first HMOs to become available in industrial scale from fermentation of engineered E. coli.
Both compounds have been registered as safe with the U.S. Food and Drug Administration (FDA) and
the European Food Safety Authority (EFSA) [
9
,
13
] and are now available in infant formula in more
than 30 countries. The research developments leading to this breakthrough as well as the challenges
faced were recently reviewed by authors from HMO-producing companies [
9
,
10
]. Fermentation titers
up to 180 g/L have been reported for 2
0
-FL [
9
]. Infant formula containing 2
0
-FL and LNnT was safe and
well-tolerated in a clinical trial with infants of up to 6 months; parents reported less morbidity upon
ingestion of 2
0
-FL and LNnT [
21
]. Beneficial effects on adult microbiota have also been reported [
5
].
Other HMO structures which can be produced in industrial scale by fermentation, but are not yet
marketed in infant formula, are LNT and difucosyllactose (DFL) [
9
]. Additionally, efforts have been
made to engineer Saccharomyces cerevisiae [
22
] or Lactococcus lactis [
23
]—organisms having the GRAS
(generally regarded as safe) label—to produce HMOs.

Molecules 2019,24, 2033 4 of 22
However, while metabolic engineering and fermentation technology has decisively improved
HMO production within the past two decades, the technology also faces limitations. For more
complex HMOs, fermentation titers are often low and/or the products remain intracellular [
13
,
15
].
For industrial production, extracellular products significantly ease purification [
9
]. In addition,
a suitable
β
1,6-N-acetylglucosaminyltransferase to produce branched HMOs is still missing [
13
].
Branched HMO structures comprise a large part of the natural HMO pool [
11
], but in fermentation
processes, they have only been obtained in low titers and mainly inside the cell as a side product
where the Neisseria meningitides
β
1,3-N-acetylglucosaminyltransferase catalyzed formation of the
β
1,6-linkage [
15
,
24
]. A human
β
1,6-N-acetylglucosaminyltransferase expressed in human embryonic
kidney cells (HEK293) has been employed for
in vitro
chemoenzymatic synthesis of branched HMOs [
25
]
(see Section 2.3), but its use for production by fermentation has yet to be reported.
2.2. Chemical Synthesis
More than 15 different HMO structures have been prepared by chemical synthesis [
13
,
26
].
Examples include lab-scale synthesis of lacto-N-fucopentaose (LNFP I) [
26
], gram-scale synthesis
of LNT [
27
], and a recent report of kilogram-scale synthesis of 2
0
-FL [
28
]. The main challenge in
chemical synthesis of HMOs is the high number of protecting group manipulations, which further
increases with chain length and branching. As a result, chemical synthesis is often time-consuming
and gives low product yields. Furthermore, chemical synthesis involves toxic reagents and usually
transition metal catalysts [
29
]. Consequently, chemical synthesis of complex carbohydrates is often
not a cost-efficient production method for large-scale synthesis of HMOs. Lowering the number
of required intermediates and product purification steps is a key to obtaining a feasible, scalable
process [
28
]. Although chemically synthesized 2
0
-FL and LNnT were originally registered for use as
novel ingredients for infant formula [
10
], the use of microbial cell factories is currently the only method
applied for industrial production of these HMOs, being a more economically feasible method [
9
].
Nevertheless, chemical synthesis remains an important tool for analytical purposes as well as for
generation of building blocks for chemoenzymatic synthesis methods. Indeed, to fully harvest the
potential of chemical synthesis for production of HMOs, it is now frequently combined with enzymatic
synthesis, particularly using regio- and stereospecific glycosyltransferases (see Section 2.3) [
25
,
30
–
32
].
Recently, a chemoenzymatic route using a glycosidase for synthesis of the HMO precursor structure
lacto-N-triose II (LNT2; GlcNAc-
β
1,3-Lac) was also reported [
33
], and the same approach was used for
synthesis of LNT [
34
]. Such chemoenzymatic approaches are particularly useful for generating large
libraries of HMO structures, which can be used for bioactivity studies [25,30,35].
2.3. Enzymatic Synthesis in Vitro
In humans, HMOs are synthesized by Leloir glycosyl transferases (GTs), i.e., GTs active on
sugar–nucleotide donor substrates. GTs are usually highly regio- and stereospecific, thus enabling
efficient and precise glycoside synthesis in a single reaction. As outlined in Section 2.1, these enzymes
are successfully utilized for HMO synthesis using microbial cell factories. However, the use of GTs
in vitro
is considered more difficult. Indeed, their requirement of sugar–nucleotide substrates requires
the setup of multienzyme cascade systems for nucleotide recycling to increase the process cost-efficiency.
Furthermore, GTs can be hard to express with adequate yields. These challenges and the current
opportunities for development of GTs into efficient biocatalysts were recently reviewed elsewhere [
36
].
For HMO synthesis, several examples of (one-pot) multienzyme cascade systems including GTs exist.
Sialyltransferases (SiaTs) of varying specificity have been employed for synthesis of sialylated HMOs
such as 3
0
-sialyllactose (3
0
-SL), 6
0
-sialyllactose (6
0
-SL), and disialyllacto-N-tetraose (DSLNT) [
37
–
39
].
The Korean company GeneChem uses SiaTs for synthesis of 3
0
-SL and 6
0
-SL in large scale and had
their 3
0
-SL GRAS-approved in 2018 [
9
,
37
,
40
,
41
]. Certain sialyltransferases are dual-activity enzymes,
which also exhibit transsialidase activity and thus accept non-nucleotide donors [
42
–
45
]. This feature
has, however, not been described for any other HMO-relevant GTs. Recently, a comprehensive library

Molecules 2019,24, 2033 5 of 22
of 60 different HMO structures was synthesized by a series of human GTs expressed in a mammalian
cell line [
25
]. Combined with chemical synthesis to equip the core lactose with a multifunctional
anomeric linker, a microarray was created and utilized for assessing protein binding [
25
]. Similarly, a
library of defined linear HMO structures was synthesized recently using a set of microbial and human
GTs [
46
]. These are excellent examples of how GTs are particularly strong for synthesis of clearly
defined HMO structures of varying length with different branching, fucosylation, and sialylation
patterns for analytical purposes and especially for bioactivity studies. However, such reactions are
currently not scalable to industrial production levels [10].
Glycoside hydrolases (GHs) present an alternative to the sugar–nucleotide dependent GTs
for catalysis of HMO synthesis. In contrast to GTs, GHs accept cheap and abundant substrates
and the enzymes are often robust and easy to express. The main challenge of using GHs for
transglycosylation is their inherent hydrolysis activity, usually of both substrates and products. While
several examples of naturally occurring transglycosidases exist [
47
], i.e., GHs which essentially do
not catalyze hydrolysis, the only HMO-relevant example is the transsialidase from Trypanosoma
cruzi, TcTS [
48
]. Consequently, our focus is turned to the GHs, which catalyze transglycosylation in
competition with hydrolysis. Reaction conditions can be optimized to favor transglycosylation, e.g.,
through increasing substrate concentration or by adjusting pH [
49
,
50
]. However, the strongest tool to
efficiently improve transglycosylation activity and/or diminish hydrolytic activity in GHs is protein
engineering. Enzymatic transglycosylation has shown great potential in HMO synthesis, especially
for sialylation or fucosylation of lactose or LNT [
42
,
48
,
51
–
57
], but also for synthesis of HMO core
structures [
58
–
61
]. This technology holds great potential to expand the current limited industrial-scale
HMO portfolio, but much of this potential relies on either the discovery of novel transglycosidases or
protein engineering of the enzymes for improved transglycosylation efficiency.
3. Glycosidase-Catalyzed Transglycosylation
Enzymatic transglycosylation is catalyzed by retaining glycoside hydrolases (GHs), i.e.,
glycosidases, which retain the configuration of the anomeric center of their products. The active site of
a GH is described by a subsite nomenclature [
62
]: subsites are labelled from
−
nto +n, where nis an
integer. While
−
nrepresents the nonreducing end of the glycoside recognized by the GH, +nrepresents
the reducing end. Catalytic cleavage takes place between the
−
1 and +1 subsites. For HMO synthesis,
exo-acting GHs with a single negative subsite (
−
1) are most common, but disaccharide-transferring
GHs such as lacto-N-biosidases [
61
] are also relevant, and in general, transglycosylation is not
limited to exo-acting enzymes [
47
]. To produce oligosaccharides through kinetically controlled
transglycosylation, the glycosyl moiety to be transferred to an acceptor substrate must be linked
with a glycosidic bond in the donor substrate (Figure 2) [
63
]. This donor glycosyl moiety is bound
in the negative subsite(s) and usually defines the names of the GHs which recognize it. With the
exception of most GlcNAc/GalNAc-processing enzymes, retaining GHs operate with the classical
Koshland double-displacement mechanism, which is a two-step reaction with at least two transition
states [
47
,
64
]. In the first step—the glycosylation step—a covalent glycosyl–enzyme intermediate is
formed upon binding of the donor glycosyl in the active and release of a leaving group; this takes place
via an oxocarbenium-like transition state. This intermediate has the opposite anomeric configuration
to that of substrate and product, and in the second step—the deglycosylation step—it undergoes
nucleophilic attack from either water or a glycosyl acceptor. Through the second transition state, this
nucleophilic attack results in either hydrolysis or transglycosylation, depending on the nature of the
nucleophile (Figure 2A). The retaining GH mechanism, its rate constants, and its importance in GH
engineering has been thoroughly described elsewhere [
47
]. The GH20
β
-N-acetylhexosaminidases
employ a substrate-assisted reaction mechanism, where the 2-acetamido group of the substrate acts as
an intramolecular nucleophile and the GlcNAc forms an oxazolinium ion intermediate rather than a
glycosyl–enzyme intermediate (Figure 2B) [65].

Molecules 2019,24, 2033 6 of 22
It is evident that transglycosylation takes place in competition with hydrolysis (Figure 2). The
balance between the transglycosylation rate (r
T
) and the hydrolysis rate (r
H
) is largely governed by
enzyme properties, which can by modified by protein engineering. However, transglycosylation can
also be favored through reaction conditions such as water activity, pH, temperature, and substrate
concentrations [
49
,
50
]. For HMO synthesis, the most common trick is the use of high acceptor
substrate concentration. A high acceptor-to-donor ratio (A:D) can make many retaining GHs
catalyze transglycosylation with moderate yields [
53
,
66
], but to work well at equimolar ratios,
protein engineering is often preferable to reach appreciable product yields [
55
,
67
]. Not only the donor
substrate, but also the transglycosylation product may be subject to hydrolysis catalyzed by the same
GH that catalyzed its formation (secondary hydrolysis; Figure 2). This leads to a transient product
maximum, and in a case where product hydrolysis is pronounced, tight reaction time control is essential.
Thus, successful protein engineering is tightly linked to reduction of hydrolytic activity.
Molecules 2019, 24, x 6 of 22
also be favored through reaction conditions such as water activity, pH, temperature, and substrate
concentrations [49,50]. For HMO synthesis, the most common trick is the use of high acceptor
substrate concentration. A high acceptor-to-donor ratio (A:D) can make many retaining GHs catalyze
transglycosylation with moderate yields [53,66], but to work well at equimolar ratios, protein
engineering is often preferable to reach appreciable product yields [55,67]. Not only the donor
substrate, but also the transglycosylation product may be subject to hydrolysis catalyzed by the same
GH that catalyzed its formation (secondary hydrolysis; Figure 2). This leads to a transient product
maximum, and in a case where product hydrolysis is pronounced, tight reaction time control is
essential. Thus, successful protein engineering is tightly linked to reduction of hydrolytic activity.
Figure 2. Reaction scheme sketches for glycosidase-catalyzed transglycosylation, which takes place
in competition with substrate hydrolysis [49,68]. (A) Classical Koshland double-displacement
mechanism exemplified by the α-L-fucosidase reaction: The intermediate, which is in the opposite
anomeric configuration compared to the substrate and product as per the double displacement
mechanism of retaining glycoside hydrolases (GHs), is attacked by a nucleophile. If this nucleophile
is water, (primary) hydrolysis occurs. If a glycosyl acceptor performs the nucleophilic attack,
transglycosylation occurs. (B) Substrate-assisted reaction mechanism of the GH20 β-N-
acetylhexosaminidases [65]. For both reaction mechanisms, the resulting glycosylated product may
Figure 2.
Reaction scheme sketches for glycosidase-catalyzed transglycosylation, which takes place in
competition with substrate hydrolysis [
49
,
68
]. (
A
) Classical Koshland double-displacement mechanism
exemplified by the
α
-l-fucosidase reaction: The intermediate, which is in the opposite anomeric
configuration compared to the substrate and product as per the double displacement mechanism of
retaining glycoside hydrolases (GHs), is attacked by a nucleophile. If this nucleophile is water, (primary)
hydrolysis occurs. If a glycosyl acceptor performs the nucleophilic attack, transglycosylation occurs. (
B
)
Substrate-assisted reaction mechanism of the GH20
β
-N-acetylhexosaminidases [
65
]. For both reaction
mechanisms, the resulting glycosylated product may also be subject to (secondary) hydrolysis catalyzed
by the same glycosidase. The balance between the transglycosylation rate (r
T
) and the hydrolytic rate
(r
H
) is governed by the reaction conditions as well as by enzyme properties, which can be altered
through protein engineering. Regioselectivity in the product formation may vary. In HMO synthesis,
R
1
and R
2
are glycosides, but for transglycosylation, in general, they can be other compounds, e.g.,
primary alcohols.

Molecules 2019,24, 2033 7 of 22
While GHs are stereospecific, as they produce products with defined anomeric configuration,
regioselectivity varies between enzymes and may depend on acceptor structure. Low regioselectivity
can be observed both as several different hydroxyl groups on the same monosaccharide moiety
of the acceptor and/or as hydroxyl groups from different monosaccharide moieties acting as
nucleophiles
[53,59,66,69]
. Enzyme regioselectivity must be harnessed when employing glycosidases
for HMO synthesis, either by choosing highly regioselective enzymes [
55
], through process design to
remove regioisomers either by specific enzymatic degradation [
61
,
70
] or by purification, or through
protein engineering for improved regioselectivity [70–73], which is beyond the scope of this review.
4. Improved Transglycosylation through Protein Engineering
Over the last decade, protein engineering has been established as a strong tool for improving the
transglycosylation efficiency of HMO-relevant glycosyl hydrolases [
54
,
55
,
59
,
67
,
73
–
77
]. We also review
a few approaches to improve transglycosylation efficiency that have not yet been reported for HMO
synthesis but could be relevant for the future of the field.
4.1. Glycosynthases
The most generic strategy to turn GHs into synthetic tools has so far been a mechanism-based
approach resulting in so-called glycosynthases; this approach was recently reviewed elsewhere [
78
].
It consists of mutating the catalytic nucleophile (or one of the catalytic residues, in the case
of substrate-assisted mechanisms) [
79
] and providing an activated donor of the same anomeric
configuration as the reaction intermediate, i.e., the opposite configuration of the desired product [
80
].
The requirement for activated donors (typically fluoride, sometimes azido or oxazoline saccharide
derivatives) limits their use in industrial scale. Regarding the glycosidic bonds to be formed in HMOs,
successful reports of galactosynthases [
81
] and fucosynthases [
82
–
85
] exist. For fucosynthases, an issue
is the fact that
β
-fluoride donors are much less stable than
α
-fluoride donors [
78
]. Instead, sufficient
substrate stability was achieved with a
β
-fucosyl azide donor substrate [
82
]. For glycosynthases which
accept GlcNAc donors, lessons learnt with chitinases [
86
,
87
] and endo-
β
-N-acetylglucosaminidases [
88
]
may be transferable to HMO structures. However, a recent study on a GH20
β
-N-acetylhexosaminidase
showed that the general strategy of glycosynthase engineering was not applicable [
89
]. Glycosynthase
mutants of a GH20 lacto-N-biosidase did not outperform the wild-type enzyme in terms of yield
or ratio between transglycosylation and hydrolysis. The only advantage of this glycosynthase was
the significantly lowered product hydrolysis rate [
34
]. No sialyl-transferring glycosynthase has been
reported. Notably, glycosynthase-catalyzed reactions lead to anionic byproducts such as azide and
fluoride that prevent their use in food production. Despite their high transglycosylation product yields,
glycosynthases may struggle with the same regioselectivity issues as encountered by their parent
glycosidases [
82
,
90
]. In some cases, equally good or better transglycosylation yields were obtained
with engineered glycosidases compared to the corresponding glycosynthases [67,75,91].
4.2. Rational Design
Unlike the glycosynthase strategy, alternative approaches for improved transglycosylation
efficiency maintain the GH activity on natural substrates. Three common approaches can be outlined:
(1) to modify the
−
1 subsite in order to reduce the transition states stabilization of the catalyzed
reactions (particularly hydrolysis); (2) to increase the affinity in the acceptor binding site(s), commonly
subsites +1 and +2; or (3) to disrupt binding of catalytic water [
50
]. While these are the goals, translating
them into identification of specific amino acid mutation targets is not necessarily straightforward.
In the following, available strategies are introduced.
The CAZy database divides GHs into families based on sequence similarity (www.cazy.org) [
92
].
Thus, for GHs within the same family, it may be possible to extrapolate successful mutations, although
this is not a given [
55
,
75
,
93
,
94
]. Indeed, mutations in the negative subsites are more likely to be
transferable than mutations in positive subsites, as the positive subsites may differ substantially within

Molecules 2019,24, 2033 8 of 22
each family. However, examples of transferability for positive subsite mutations do exist [
93
]. If working
with a well-studied family of enzymes where prior successful engineering studies and structural data
are available, rational design is possible. Certain GH families contain both hydrolases and natural
transglycosidases, which can be used as templates for rational engineering of the hydrolases [
47
,
74
].
Moreover, hydrophobic platforms are obvious targets in all subsites, as several studies have indicated
the importance of aromatic residues for increasing the ratio of transglycosylation over hydrolysis
(T/H)
[66,67,93,95–99]
. However, when aiming to conquer new territory, such templates or previous
successes are often not available. In addition, the changes that lead to increased transglycosylation
efficiency are often minute and hard to pinpoint from a structural point of view [
67
,
98
], or rationalization
of the previously obtained results may be unsuccessful
[67,95]
. In many cases, molecular modeling,
molecular dynamics, and quantum mechanics/molecular mechanics (QM/MM) simulations are
tentatively used to rationalize the obtained results for future predictions [
50
,
100
–
104
]. However,
the understanding of the structure–function relationships governing the transglycosylation/hydrolysis
balance in an enzyme is still so far from complete that generic in silico methods for predicting successful
mutations belong to the future. The currently available (semi)generic (semi)rational approaches to
engineer glycosidases for improved transglycosylation are outlined below.
4.2.1. Targeting Conserved Residues in Negative Subsite(s)
A stark difference in rates commonly discriminates hydrolases and natural transglycosidases, the
former being much faster [
99
,
105
,
106
]. Hence, a logical step to increase transglycosylation yields by
protein engineering could be to seek enzymes with decreased hydrolysis. Without any knowledge of
sequence or structure, one can use directed evolution where diversity is generated by error-prone PCR
or gene shuffling and screening for reduced hydrolysis. The screening task is large in directed evolution,
but it can be accomplished with direct colorimetric screening on petri dishes using the classical method
with X-Gal (5-bromo-4-chloro-3-indolyl
β
-d-galactopyranoside) or its analogues [
75
,
90
,
107
]. Taking
advantage of the fact that good transglycosidases have higher activity in the presence of an acceptor
substrate, a colorimetric screen measuring apparent rates of both hydrolysis and transglycosylation
directly on colonies have been developed [
108
]. Direct detection of transglycosylation products in
large libraries could also be done using biosensor strains expressing green fluorescent protein (GFP)
upon presence of specific HMO molecules [
109
], but this strategy has not yet been applied for enzyme
engineering purposes.
Outcomes of directed evolution have repeatedly singled out the
−
1 subsite as a common location
for mutations that drastically increased transglycosylation yields. The mutated residues also appeared
to be often conserved through evolution at the sequence level [
75
,
108
,
110
]. Consequently, targeting
conserved residues in subsite
−
1 was proposed as a semirational approach for increasing T/H. It was
hypothesized that such mutations decrease transition states (TS
‡
) stabilization, and that this will usually
affects the hydrolysis TS
‡
more than the transglycosylation one. A modification of TS
‡
stabilization of a
given reaction directly translates into a rate change of said reaction. The resulting mutant enzymes thus
have an increased T/H ratio, although often at the expense of catalytic efficiency [
67
]. The systematic
mutagenesis of conserved residues in the
−
1 subsite was applied to a GH1
β
-glycosidase, where the
mutation of any of seven first-shell residues led to improved transglycosidases (65–82% yields of
disaccharide synthesis versus 36% for the native enzyme) [
67
]. This approach was then refined on
a GH36
α
-galactosidase, where it was shown that second-shell conserved residues were important
targets, and that “conservative” mutations (e.g., Tyr into Phe) preserved higher overall activity than
Ala mutants and preserved or increased synthetic yields [
95
]. Another implementation, beyond
HMO synthesis, allowed high transglycosylation yields (80%) on a GH51
α
-arabinofuranosidase for a
combination of conserved residues from the
−
1 subsite and in silico screening for binding affinity in the
acceptor subsites [
111
]. Although not all mutations result in mutants with improved transglycosylation
capacity, this strategy dramatically lowers the screening effort compared to directed evolution or
site-saturation mutagenesis of active site residues. Besides reducing the sequence space to pinpoint

Molecules 2019,24, 2033 9 of 22
good candidates, an advantageous feature of conserved residues modification lies in its transferability:
once a particular mutation has been identified as beneficial for TG yields, it can be transposed on related
enzymes, i.e., enzymes within the same GH family or clan. This “mutation grafting” has been successful
in a number of cases [
112
,
113
], from which we would highlight the transfer of mutations identified
in the Thermotoga maritima GH29A
α
-fucosidase to a GH29B
α
-1,3/4-fucosidase from Bifidobacterium
longum subsp. infantis to synthesize fucosylated HMOs [55].
4.2.2. Loop Engineering
Another recently emerged strategy targets loops close to the active site, which are either inserted
or exchanged in order to shield the active site from water or to alter the water network inside it. The
strategy relies on structural data for loop identification. At best, crystal structures are available for
the targeted enzyme or at least within the same GH family in order to allow homology modeling.
Alternatively, secondary structure prediction tools could be used in combination with multiple
sequence alignment (MSA). Three different examples of loop engineering of HMO-relevant enzymes
exist; together, they outline two different strategies to identify relevant loops.
The human pathogen Trypanosoma cruzi expresses a native GH33 transsialidase, TcTS, which
efficiently catalyzes formation of
α
2,3-sialosides. Nonpathogenic Trypanosoma rangeli expresses a
GH33 sialidase, TrSA, which has 70% sequence identity to TcTS, but does not catalyze transsialylation.
Consequently, TcTS has been used several times as a template for engineering TrSA into an efficient
transsialidase [52,74,99]. One of these engineering studies used a rational strategy where only amino
acids up to 14 Å from the active were considered, particularly targeting amino acids in the sequence
alignment with large chemical differences, proposing that they relate to a low probability of random
evolutionary substitution [
52
]. With this approach, a seven-amino-acids-long loop almost 14 Å from
the active site was identified: exchanging these seven amino acids with the corresponding sequence
from TcTS introduced a net charge of +3. Since the loops aligned perfectly in the crystal structures of
TrSA and TcTS irrespective of ligand binding, it was hypothesized that the positive effect of this loop
exchange on the hydrolytic activity (4-fold decrease) was due to a reversal of the water network in the
active site. It was hypothesized that the low hydrolytic activity of TcTS was indeed due to disruption
of the water network by this charged loop, placing the catalytic water in an orientation unfavorable for
catalysis [52]. A similar phenomenon has been observed for a native GH31 α-transglucosylase [114].
The second example relies on the results obtained with two GH29B
α
1,3/4-l-fucosidases. While
CpAfc2 from Clostridium perfringens efficiently catalyzed transglycosylation, BbAfcB from Bifidobacterium
bifidum was hampered by high hydrolytic activity [
53
]. Structural alignment of their homology models
as well as sequence alignment revealed that all substrate-interacting residues and the main structural
features aligned well. However, CpAfc2 featured a loop approx. 13 Å from the ligand, which in BbAfcB
was further away and more open and disordered. Sequence alignment was poor in this region as the
two loops differed in length as well as in sequence [
54
]. It was hypothesized that replacement of the
loop sequence in BbAfcB with that of CpAfc2 would improve the transfucosylation ability of BbAfcB
through better shielding of the active site from the aqueous environment. Unlike the case for TcTS and
TrSA, these loops did obviously not align structurally, and great care was therefore taken to define the
starting point and end of the loops to be exchanged. Replacement of a 23-amino-acids-long loop from
BbAfcB with the corresponding 17-amino-acid loop of CpAfc2 resulted in almost complete quenching
of the hydrolytic activity on 3-FL, while the transfucosylation activity was lowered by only one order
of magnitude. As a result, the transfucosylation yield of the loop mutant was comparable to that of
CpAfc2 [54].
In order to extract a general strategy from these two examples, it is important to acknowledge the
need for a template enzyme from the same GH (sub)family which has a high transglycosylation activity,
be it a native transglycosidase or not. In both cases, pathogenic organisms provided these templates.
Indeed, pathogenic organisms are, in general, a promising source of transglycosidases, because they
use this activity as part of their camouflage strategy [
115
]. Having found a suitable template, the next

Molecules 2019,24, 2033 10 of 22
step is to identify major differences between the template enzyme and the enzyme to be engineered
through both sequence and structural alignments. Targeting areas with large chemical differences or
different loop sizes are obvious candidates.
Improving transglycosylation activity of a GH20
β
-N-acetylhexosaminidase for synthesis of the
HMO precursor LNT2 by loop engineering was carried out differently [
59
]. Since no natural GH20
transglycosidase is known, the structure comparison approach could not be used. As an alternative to
identification of beneficial loop sequences and structural placement of those, the bioinformatic tool
peptide pattern recognition (PPR) was used [
116
]. This technique basically performs grouping of a
set of protein sequences based on short conserved peptide sequences to identify novel enzymes with
potentially new activities [
117
]. The entire GH20 family as found in CAZy (approx. 3700 sequences)
was submitted to PPR analysis, which resulted in 34 different groups. Two previously described
β
-N-acetylhexosaminidases (HEX1 and HEX2) able to synthesize LNT2 from chitobiose and lactose
with very low yields [
58
] were found in the largest group, consisting of approx. 1000 sequences.
Detailed phylogenetic tree analysis of this group led to identification of four different five-amino-acid
stretches, which were present in GH20 enzymes closely related to HEX1 [
59
]. Some of these sequences
originated from pathogenic bacteria, e.g., the fish pathogen Renibacterium salmoninarum, and were
thus potentially natural transglycosidases. All additional loop sequences were charged, which was
described previously to increase transglycosylation activity [
52
,
74
]. Finally, preparation of homology
models of the loop-engineered variants revealed that the oxazoline intermediate was more shielded
from bulk water with the additional loop stretches present. The transglycosylation activity of eight
different loop variants was investigated, and three of them had an increased activity compared to the
wild type. The best variant showed a 9-fold higher overall transglycosylation yield [
59
]. However,
four different LNT2 isomers were produced, thus underlining that regioselectivity is probably still the
most challenging characteristic to be engineered.
No other examples of loop engineering with the aim of improving transglycosylation by
HMO-relevant GHs exist, but examples of loop exchange, deletion, and insertion to alter enzyme
specificities exist and may be used for inspiration [118–124].
4.3. Effect of Protein Engineering on Reaction Rates
In most cases of glycosidase engineering for improved transglycosylation yields, the increased
T/H is obtained through reduction of the hydrolysis rate rather than through an increase in the
transglycosylation rate [
50
,
52
,
54
,
74
,
90
]. In fact, the transglycosylation rate usually decreases as well,
albeit not as much as the hydrolytic rate [
50
,
54
,
74
]. Frequently, the general enzyme activity decreases
with the number of point mutations introduced [67,71,73,75,95,108].
5. Abundant Natural Substrates from Dairy and Agro-Industrial Side Streams
A requirement for obtaining a practicable process is the availability of cheap substrates. Microbial
cell factories convert simple sugars into HMOs (see Section 2.1), thus keeping the process cost-efficient.
For
in vitro
enzymatic processes, glycosidases hold the largest potential in terms of providing a low-cost
process because they accept cheap, naturally available substrates. While activated, synthetic donor
substrates often give higher HMO yields due to their good leaving groups, they are exclusively useful
for analytical purposes as well as for enzyme discovery and engineering [
49
,
125
]—not for industrial
processes. The use of lactose is obvious as it is highly abundant in the dairy industry and present as the
core of all HMOs (Figure 1). Identification of industrial side streams that are useful donor substrates
for HMO production mainly relies on the presence of the relevant monosaccharides (Fuc, Sia, GlcNAc,
and Gal) in a terminal position where the GH in question recognizes the unoccupied nonreducing end
as well as the linkage to the neighboring moiety. For development of an industrial HMO production
process, food-grade streams are preferred. Not all HMO building blocks are equally easy to find in
such a setting. This section provides a status on the synthesis of HMO structures from abundantly
available agro-industrial side streams.

Molecules 2019,24, 2033 11 of 22
The most developed process is the use of casein glycomacropeptide (CGMP) as a sialyl donor
(Table 1). CGMP is the 64-amino-acids-long C-terminal of
κ
-casein, which is released into the whey
upon chymosin action in cheese manufacturing. This soluble glycoprotein makes up approx. 20% of the
whey protein [
126
]. Whey is produced in massive quantities and CGMP is thus abundantly available
and of food-grade quality [
3
,
127
]. CGMP has a fairly defined glycosylation pattern, which includes
4–9% Sia in terminal positions [
69
,
128
]. The distribution between
α
2,3- and
α
2,6-linked Sia is almost
even: the percentage of
α
2,3-linked sialic acid is in the 50–59% range [
69
,
129
]. After use as a sialyl donor,
the desialylated CGMP is still useful as a protein supplement. Several examples of the use of CGMP
as a sialyl donor for synthesis of sialylated HMOs exist [
42
,
48
,
52
,
56
,
66
,
70
,
130
]. Among the enzymes
employed are the native T. cruzi transsialidase TcTS, engineered variants of the T. rangeli sialidase TrSA,
the dual-activity sialyltransferase from Pasteurella multocida, and several other microbial sialidases
(Table 1). Molar yields on the donor substrate—Sia bound in CGMP—generally ranged from 19–37%
for the enzymes specific for
α
2,3-sialosides (Table 1). Considering that only 50–59% of the Sia bound in
CGMP is available for these enzymes, especially the yields in the 30–40% range must be considered high
enough to hold industrial potential. For the dual-activity sialyltransferase from Pasteurella multocida,
which formed both 3
0
-SL and 6
0
-SL, the total yield was 53% [
70
] (Table 1). Membrane filtration setups
for up- and downstream processing have been suggested [
131
–
133
], and performance of the reaction in
an enzymatic membrane reactor increased the 3
0
-SL yield and biocatalytic productivity [
134
]. Indeed,
since both CGMP and Lac are abundantly available dairy streams, the industrial interest in their use
for enzymatic synthesis of sialylated oligosaccharides is evident [
130
,
135
], and direct application of
TcTS in milk and whey for 30-SL enrichment was patented 20 years ago [136].
Table 1.
Synthesis of true HMO structures where abundantly available natural substrates—or derivatives
of these—have been employed. Yields are given as molar yield based on the donor substrate (for casein
glycomacropeptide (CGMP), all available Sia moieties are considered, although not all are present with
a linkage accepted by the enzymes; see the main text). Abbreviations: A:D: molar donor-to-acceptor
ratio; n.d.: not determined.
Enzyme Donor Acceptor HMO
Product A:D Yield Ref.
Arthrobacter ureafaciens sialidase CGMP Lac n.d. ~45 5% [130]
Bifidobacterium infantis sialidase CGMP Lac n.d. ~45 1% [130]
Trypanosoma rangeli sialidase
(engineered) CGMP Lac 30-SL 44 31% [52]
Trypanosoma rangeli sialidase
(engineered, membrane reactor) CGMP Lac 30-SL ≤25 37% [134]
Pasteurella multocida sialidase
(sialyltransferase) CGMP Lac 60-SL, 30-SL 11 53% [42,70]
Haemophilus parasuis sialidase CGMP Lac 30-SL
(and isomer) 39 19%
(28%) [66]
Trypanosoma cruzi transsialidase CGMP Lac 30-SL 5 32% [48]
Trypanosoma cruzi transsialidase Fetuin Lac 30-SL 3 76% [137]
Bacteroides fragilis sialidase Hydrolyzed
colominic acid Lac 60-SL ~15 22% [57]
Fusarium graminearum fucosidase Citrus peel
xyloglucan Lac 20-FL 50 14% [53]
β-N-acetylhexosaminidases
(engineered, metagenomic) Chitobiose Lac LNT2
(and
isomers) 5 5% (30%) [59]
Bacillus circulans β-galactosidases Lac LNT2 LNnT 1 19% [61]
Another sialyl donor substrate is polysialic acid, a polymer of
α
2,8-linked Sia, which is also known
as colominic acid. Polysialic acid can be produced by fermentation of engineered E. coli and thus has
the potential to become a cheap and abundant substrate [
138
]. In a recent study, polysialic acid was

Molecules 2019,24, 2033 12 of 22
used as a sialyl donor for a sialidase from Bacteroides fragilis, which selectively catalyzed formation
of
α
2,6-sialosides: the highest yields of 6
0
-SL (22%; Table 1) were obtained when polysialic acid was
turned into oligosialic acid with a simple acid hydrolysis method ready for industrial application before
transglycosylation [
57
]. Colominic acid has also been used as a sialyl donor for other sialidases, albeit
with much lower yields [
49
]. Another abundant source of sialyl donors is slaughterhouse waste [
139
].
Bovine blood plasma glycoprotein (BPG) and porcine small intestinal mucin glycoprotein (PSMG) have
also been mentioned as sialyl donors for TcTS, but unlike bovine CGMP, they both contain a mixture of
two sialic acids, namely N-acetylneuraminic acid (Neu5Ac, the one found in HMOs and >99% of the
Sia in CGMP) and N-glycolylneuraminic acid (Neu5Gc) [
135
,
140
]. However, Neu5Gc from mammalian
dietary sources can be incorporated into human tissues, where it causes inflammation. Neu5Gc is
also believed to play a central role in cancer development, although it may only be carcinogenic in
combination with other factors [
141
–
143
]. Similarly, fetuin from fetal calf serum has been used as a
sialyl donor for synthesis of 30-SL with TcTS in a high-yield reaction (76%; Table 1) [137].
Chitin, a (
β
1,4)-GlcNAc polymer, is a structural component of fungal, algal, and yeast cell walls
and is present in the exoskeleton of crustaceans, mollusks, and insects [
144
,
145
]. Chitin is similar to
cellulose in terms of both structure and functionality in nature, recalcitrance, and abundance. Indeed,
its abundance in the ecosystem is measured in gigatons [
144
]. The main source of industrial chitin
is shellfish waste, which readily grants it food-grade status. Estimates of the amount of chitin-rich
shellfish waste are in the megaton range [
145
]. Chitin extraction from shellfish waste traditionally takes
place in two steps (demineralization and deproteinisation), which can be performed either chemically
or microbially [
146
]. However, chitin polymers are insoluble and therefore need depolymerization
before chitin can be utilized as a GlcNAc donor for HMO synthesis [
147
]. Generally, transglycosylation
is more efficient when using donor substrates with a low degree of polymerization [
53
,
57
]. Such
depolymerization could be accomplished by combining chitinases and chitin-active lytic polysaccharide
monooxygenases, as done by microorganisms able to degrade the recalcitrant polymer [
148
,
149
],
preferably as an optimized minimal enzyme cocktail suitable for industrial application [
150
]. Evidently,
the process from shellfish waste to HMOs is still in its infancy, but many of the required steps have
been studied separately, e.g., the use of chitobiose, a substrate which could be produced by chitin
depolymerization, as a donor substrate in the enzymatic production of the HMO core component
LNT2, albeit with yields below 10% [58,59] (Table 1).
Following generation of LNT2 or similar HMO precursor structures, a terminal Gal residue must
be added to yield a full HMO structure (Figure 1). This can be accomplished by
β
-galactosidases,
which are already widely used in industry for GOS production [
18
,
19
]. Using one of the most
efficient commercial
β
-galactosidase preparations—Biolacta, which contains several Bacillus circulans
β
-galactosidases [
151
]—a 19% yield of LNnT was obtained from a reaction with Lac and LNT2 [
61
]
(Table 1). A one-pot cascade reaction from Lac to LNnT was not accomplished due to low LNT2
yields [
60
].
β
-Galactosidase-catalyzed production of LNT from LNT2 has only been reported with a
o-nitrophenyl galactoside donor using a GH35
β
1,3-galactosidase from B. circulans [
61
]. An alternative
to this two-step approach is the use of lacto-N-biosidases, which transfer disaccharides, but so far, this
has only been accomplished with a p-nitrophenyl-activated donor substrate [
61
]. In summary, this
process can hardly compete with production of LNT and LNnT in microbial cell factories. However,
the sequential use of
β
-N-acetylhexosaminidases and
β
-galactosidases may provide a solution to
efficiently obtaining HMOs branched on the Gal moiety of Lac, a type of HMO synthesis reaction
which is currently not possible to obtain in microbial cell factories [13].
Suitable, natural Fuc donor substrates useful for HMO production are scarce. However, enzymatic
transfucosylation from fucosylated citrus peel xyloglucan to yield 2
0
-FL was recently reported [
53
]
(Table 1). Product yields were moderate (14% on the donor), and there is room for improvement
by protein engineering; by increasing substrate accessibility, e.g., by xyloglucanase treatment; or by
looking for other plant sources with higher Fuc content. Abundantly available sources of fucosylated
xyloglucan include citrus peel, berry press residues [
152
], and peanut shells [
153
,
154
]. The Fuc-rich

Molecules 2019,24, 2033 13 of 22
polymer fucoidan is found in brown seaweed, but the Fuc units are often sulfated [
155
]. While a
few fucoidan-degrading fucosidases have been described [
156
,
157
], no reports of transglycosylation
from fucoidan exists. However, the biomass resources are vast, and it is hypothesized that fucoidan
can be made accessible as a fucosyl donor through degradation by fucoidanase and sulfatase. Fuc is
also present in a terminal position on mucin, which can be isolated from slaughterhouse waste. The
α
1,3/4-l-fucosidase CpAfc2 from Clostridium perfringens, which has considerable transfucosylation
activity [
53
], is hydrolytically active on porcine gastric mucin [
158
]. However, no reports of
transfucosylation with mucin as a fucosyl donor exist. In conclusion, utilization of fucosylated
industrial side streams for HMO production is still in its infancy. Alternatively, simple fucosylated
compounds produced by microbial cell factories could serve as donor substrates as outlined below.
It is evident that only the use of CGMP and Lac is close to industrial application. Currently,
glycosidase-catalyzed synthesis using natural substrates cannot compete with microbial cell factories
or chemical synthesis for production of 2
0
FL at an industrial scale. However, glycosidase-catalyzed
transglycosylation may turn out to be a competitive process for more complex HMO structures,
because enzyme catalysis is not significantly hampered by small increases in substrate size. While the
yields in Table 1may appear moderate, they provide proof-of-concept of the technology, and protein
engineering clearly manifests itself as a valuable tool to provide sufficiently efficient biocatalysts.
From here, stepwise process optimization is required to develop a cost-efficient process, e.g., through
further enzyme engineering and recycling of enzyme and unreacted substrate. For natural polymeric
substrates, some upstream extraction work may be required to help increase the access of the enzyme
to the substrate or solubilize the reactive parts of the substrate and hence improve reaction rates and
yields. However, enzymatic reactions may require less downstream purification than production by
microbial cell factories. Enzymes can be produced on demand: today, industrial-scale expression of
recombinant enzymes is already a large business, and especially GHs are generally easy to express in
high yields and are robust in operation. Evidently, more work is required before glycosidase-catalyzed
synthesis of HMOs reaches the same industrial potential as production of GOS and FOS [
18
–
20
], but
the existence of the enzymatic GOS and FOS production processes provides evidence that enzymatic
transglycosylation can be a cost-efficient industrial process.
Current focus is on utilization of smaller HMO structures produced by fermentation or by
sialyltransferases as substrates for production of more complex HMOs using glycosidase-catalyzed
transglycosylation. Fucosyllactoses 2
0
-FL and 3-FL and sialyllactoses 3
0
-SL and 6
0
-SL, as well as
backbone structures LNnT and LNT, are among the few HMOs which are available in large or medium
scale [
9
,
159
]. For example, sialyllacto-N-tetraose c (LST c) was obtained from 6
0
-SL and LNnT in a
reaction catalyzed by the sialidase activity of the dual-activity sialyltransferase from Photobacterium
leiognathi after extensive protein engineering [
160
,
161
]. Another example is the use of the regiospecific
GH29B
α
1,3/4-l-fucosidases—either as a native enzyme from Clostridium perfringens or engineered
forms from Bifidobacterium bifidum or B. longum subsp. infantis—which led to appreciable yields of LNFP
II and LNFP III in reactions with 3-FL and LNT or LNnT, respectively [
54
,
55
]. Using an engineered
α
1,3/4-l-fucosidase with minimal hydrolytic activity [
54
], such a reaction would yield a mixture of
three HMOs (e.g., 3-FL, LNT, and LNFP II) and lactose, which are all relevant for infant formula
addition. Together with comprehensive industry-driven engineering studies on the B. longum subsp.
infantis fucosidase to yield other complex fucosylated HMOs [
162
], these examples indicate the interest
in this approach. Indeed, glycoside hydrolases hold the potential to be engineered to widen the current
industrial HMO portfolio significantly, using abundant natural substrates as well as simple HMOs
produced by other viable methods.
6. Conclusions and Perspectives
HMO production by microbial fermentation was shown to be a viable route 20 years ago [
9
]. After
decades of metabolic engineering of E. coli, from knock-out of genes encoding for substrate-degrading
enzymes to knock-in of genes encoding for relevant glycosyltransferases, alongside several other efforts

Molecules 2019,24, 2033 14 of 22
in expression regulation and transport, it is now possible to produce a handful of HMO structures
in industrial scale by fermentation of engineered E. coli [
9
]. However, certain important structures
are not possible to obtain by metabolic engineering and fermentation. For instance, the lack of a
suitable
β
-1,6-N-acetylglucosaminyltransferase blocks the possibility of obtaining branched HMO
compounds [
13
]. Glycosidase-catalyzed transglycosylation holds the potential of being implemented
as an alternative and promising technology which can facilitate diversification of the HMO structures
in order to access a more structurally complex and diverse HMO portfolio. Its success as an industrial
technology is largely dependent on successful engineering of glycosidases to exhibit significant
transglycosylation yields. The protein engineering strategies used to design HMO molecules may
prove useful for other enzymatic carbohydrate synthesis reactions in the future.
Author Contributions:
B.Z. and A.S.M. conceptualized the review. B.Z. drafted the manuscript, and B.Z., D.T,
J.M., and A.S.M. contributed to the manuscript writing. All authors approved the final version of the manuscript.
Funding:
This research was funded by DTU Bioengineering, Protein Chemistry and Enzyme Technology Division,
Technical University of Denmark. D.T. thanks the Novo Nordisk Foundation for a Postdoc Fellowship in
Biotechnology-based Synthesis and Production research (NNF17OC0025660).
Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the
study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to
publish the results.
References
1.
Bode, L. Human milk oligosaccharides: Every baby needs a sugar mama. Glycobiology
2012
,22, 1147–1162.
[CrossRef] [PubMed]
2.
Peterson, R.; Cheah, W.Y.; Grinyer, J.; Packer, N. Glycoconjugates in human milk: Protecting infants from
disease. Glycobiology 2013,23, 1425–1438. [CrossRef]
3.
de Moura Bell, J.M.L.N.; Cohen, J.L.; de Aquino, L.F.M.C.; Lee, H.; de Melo Silva, V.L.; Liu, Y.; Domizio, P.;
Barile, D. An Integrated Bioprocess to Recover Bovine Milk Oligosaccharides from Colostrum Whey Permeate.
J. Food Eng. 2018,216, 27–35. [CrossRef]
4.
Triantis, V.; Bode, L.; van Neerven, R.J.J. Immunological Effects of Human Milk Oligosaccharides. Front.
Pediatr. 2018,6, 190. [CrossRef] [PubMed]
5.
Elison, E.; Vigsnaes, L.K.; Rindom Krogsgaard, L.; Rasmussen, J.; Sørensen, N.; McConnell, B.; Hennet, T.;
Sommer, M.O.A.; Bytzer, P. Oral supplementation of healthy adults with 2
0
-O-fucosyllactose and
lacto-N-neotetraose is well tolerated and shifts the intestinal microbiota. Br. J. Nutr.
2016
,116, 1356–1368.
[CrossRef] [PubMed]
6.
Ninonuevo, M.R.; Ward, R.E.; Park, Y.; Clowers, B.H.; Killeen, K.; Lebrilla, C.B.; Grimm, R.; German, J.B.;
Yin, H.; Freeman, S.L.; et al. A Strategy for Annotating the Human Milk Glycome. J. Agr. Food Chem.
2006
,
54, 7471–7480. [CrossRef]
7.
Urashima, T.; Hirabayashi, J.; Sato, S.; Kobata, A. Human Milk Oligosaccharides as Essential Tools for Basic
and Application Studies on Galectins. Trends Glycosci. Glycotechnol. 2018,30, SE51–SE65. [CrossRef]
8.
Varki, A.; Cummings, R.D.; Aebi, M.; Packer, N.H.; Seeberger, P.H.; Esko, J.D.; Stanley, P.; Hart, G.; Darvill, A.;
Kinoshita, T.; et al. Symbol Nomenclature for Graphical Representations of Glycans. Glycobiology
2015
,25,
1323–1324. [CrossRef] [PubMed]
9.
Bych, K.; Mikš, M.H.; Johanson, T.; Hederos, M.J.; Vigsnæs, L.K.; Becker, P. Production of HMOs using
microbial hosts—From cell engineering to large scale production. Curr. Opin. Biotechnol.
2018
,56, 130–137.
[CrossRef]
10.
Bode, L.; Contractor, N.; Barile, D.; Pohl, N.; Prudden, A.R.; Boons, G.J.; Jin, Y.S.; Jennewein, S. Overcoming
the limited availability of human milk oligosaccharides: Challenges and opportunities for research and
application. Nutr. Rev. 2016,74, 635–644. [CrossRef] [PubMed]
11.
Thurl, S.; Munzert, M.; Boehm, G.; Matthews, C.; Stahl, B. Systematic review of the concentrations of
oligosaccharides in human milk. Nutr. Rev. 2017,75, 920–933. [CrossRef]

Molecules 2019,24, 2033 15 of 22
12.
Kunz, C.; Meyer, C.; Collado, M.C.; Geiger, L.; Garc
í
a-Mantrana, I.; Bertua-R
í
os, B.; Mart
í
nez-Costa, C.; Borsch, C.;
Rudloff, S. Influence of Gestational Age, Secretor, and Lewis Blood Group Status on the Oligosaccharide Content
of Human Milk. J. Pediatr. Gastroenterol. Nutr. 2017,64, 789–798. [CrossRef] [PubMed]
13.
Sprenger, G.A.; Baumgärtner, F.; Albermann, C. Production of human milk oligosaccharides by enzymatic
and whole-cell microbial biotransformations. J. Biotechnol. 2017,258, 79–91. [CrossRef]
14.
Baumgärtner, F.; Jurzitza, L.; Conrad, J.; Beifuss, U.; Sprenger, G.A.; Albermann, C. Synthesis of fucosylated
lacto-N-tetraose using whole-cell biotransformation. Bioorg. Med. Chem. 2015,23, 6799–6806. [CrossRef]
15.
Faijes, M.; Castej
ó
n-Vilatersana, M.; Val-Cid, C.; Planas, A. Enzymatic and cell factory approaches to the
production of human milk oligosaccharides. Biotechnol. Adv. 2019. [CrossRef]
16.
Craft, K.M.; Thomas, H.C.; Townsend, S.D. Interrogation of Human Milk Oligosaccharide Fucosylation
Patterns for Antimicrobial and Antibiofilm Trends in Group B Streptococcus.ACS Infect. Dis.
2018
,4,
1755–1765. [CrossRef]
17.
Craft, K.M.; Thomas, H.C.; Townsend, S.D. Sialylated variants of lacto-N-tetraose exhibit antimicrobial
activity against Group B Streptococcus.Org. Biomol. Chem. 2018,17, 1893–1900. [CrossRef]
18.
Gosling, A.; Stevens, G.W.; Barber, A.R.; Kentish, S.E.; Gras, S.L. Recent advances refining
galactooligosaccharide production from lactose. Food Chem. 2010,121, 307–318. [CrossRef]
19.
Torres, D.P.M.; Gonçalves, M.D.P.F.; Teixeira, J.A.; Rodrigues, L.R. Galacto-Oligosaccharides: Production,
properties, applications, and significance as prebiotics. Compr. Rev. Food Sci. Food Saf.
2010
,9, 438–454.
[CrossRef]
20.
Mano, M.C.R.; Neri-Numa, I.A.; da Silva, J.B.; Paulino, B.N.; Pessoa, M.G.; Pastore, G.M. Oligosaccharide
biotechnology: an approach of prebiotic revolution on the industry. Appl. Microbiol. Biotechnol.
2018
,102,
17–37. [CrossRef] [PubMed]
21.
Puccio, G.; Alliet, P.; Cajozzo, C.; Janssens, E.; Corsello, G.; Sprenger, N.; Wernimont, S.; Egli, D.; Gosoniu, L.;
Steenhout, P. Effects of infant formula with human milk oligosaccharides on growth and morbidity: A
randomized multicenter trial. J. Pediatr. Gastroenterol. Nutr. 2017,64, 624–631. [CrossRef]
22.
Liu, J.J.; Kwak, S.; Pathanibul, P.; Lee, J.W.; Yu, S.; Yun, E.J.; Lim, H.; Kim, K.H.; Jin, Y.S. Biosynthesis of a
Functional Human Milk Oligosaccharide, 2
0
-Fucosyllactose, and L-Fucose Using Engineered Saccharomyces
cerevisiae.ACS Synth. Biol. 2018,7, 2529–2536. [CrossRef]
23.
Li, L.; Kim, S.A.; Heo, J.E.; Kim, T.J.; Seo, J.H.; Han, N.S. One-pot synthesis of GDP-l-fucose by a four-enzyme
cascade expressed in Lactococcus lactis.J. Biotechnol. 2017,264, 1–7. [CrossRef]
24.
Priem, B.; Gilbert, M.; Wakarchuk, W.W.; Heyraud, A.; Samain, E. A new fermentation process allows
large-scale production of human milk oligosaccharides by metabolically engineered bacteria. Glycobiology
2002,12, 235–240. [CrossRef] [PubMed]
25.
Prudden, A.R.; Liu, L.; Capicciotti, C.J.; Wolfert, M.A.; Wang, S.; Gao, Z.; Meng, L.; Moremen, K.W.;
Boons, G.-J. Synthesis of asymmetrical multiantennary human milk oligosaccharides. Proc. Natl. Acad. Sci.
2017,114, 6954–6959. [CrossRef] [PubMed]
26.
Arboe Jennum, C.; Hauch Fenger, T.; Bruun, L.M.; Madsen, R. One-pot glycosylations in the synthesis of
human milk oligosaccharides. Eur. J. Org. Chem. 2014,2014, 3232–3241. [CrossRef]
27.
Craft, K.M.; Townsend, S.D. Synthesis of lacto-N-tetraose. Carbohydr. Res.
2017
,440–441, 43–50. [CrossRef]
[PubMed]
28.
Agoston, K.; Hederos, M.J.; Bajza, I.; Dekany, G. Kilogram scale chemical synthesis of 2
0
-fucosyllactose.
Carbohydr. Res. 2019,476, 71–77. [CrossRef]
29.
McKay, M.J.; Nguyen, H.M. Recent Advances in Transition Metal-Catalyzed Glycosylation. ACS Catal.
2012
,
2, 1563–1595. [CrossRef]
30.
Xiao, Z.; Guo, Y.; Liu, Y.; Li, L.; Zhang, Q.; Wen, L.; Wang, X.; Kondengaden, S.M.; Wu, Z.; Zhou, J.; et al.
Chemoenzymatic Synthesis of a Library of Human Milk Oligosaccharides. J. Org. Chem.
2016
,81, 5851–5865.
[CrossRef]
31.
Schmidt, D.; Thiem, J. Chemical synthesis using enzymatically generated building units for construction of
the human milk pentasaccharides sialyllacto-N-tetraose and sialyllacto-N-neotetraose epimer. Beilstein J.
Org. Chem. 2010,6, 1–7. [CrossRef]
32.
Yao, W.; Yan, J.; Chen, X.; Wang, F.; Cao, H. Chemoenzymatic synthesis of lacto-N-tetrasaccharide and sialyl
lacto-N-tetrasaccharides. Carbohydr. Res. 2015,401, 5–10. [CrossRef]

Molecules 2019,24, 2033 16 of 22
33.
Muschiol, J.; Meyer, A.S. A chemo-enzymatic approach for the synthesis of human milk oligosaccharide
backbone structures. J. Biosci. 2019,74, 85–89. [CrossRef] [PubMed]
34.
Schmölzer, K.; Weingarten, M.; Baldenius, K.; Nidetzky, B. Lacto-N-tetraose synthesis by wild-type and
glycosynthase variants of the
β
-N-hexosaminidase from Bifidobacterium bifidum.Org. Biomol. Chem.
2019
.
[CrossRef]
35.
Wen, L.; Edmunds, G.; Gibbons, C.; Zhang, J.; Gadi, M.R.; Zhu, H.; Fang, J.; Liu, X.; Kong, Y.; Wang, P.G.
Toward Automated Enzymatic Synthesis of Oligosaccharides. Chem. Rev. 2018,118, 8151–8187. [CrossRef]
36.
Nidetzky, B.; Gutmann, A.; Zhong, C. Leloir Glycosyltransferases as Biocatalysts for Chemical Production.
ACS Catal. 2018,8, 6283–6300. [CrossRef]
37.
Choi, Y.H.; Kim, J.H.; Park, J.H.; Lee, N.; Kim, D.H.; Jang, K.S.; Park, I.H.; Kim, B.G. Protein engineering of
α
2,3/2,6-sialyltransferase to improve the yield and productivity of
in vitro
sialyllactose synthesis. Glycobiology
2014,24, 159–169. [CrossRef]
38.
Yu, H.; Yan, X.; Autran, C.A.; Li, Y.; Etzold, S.; Latasiewicz, J.; Robertson, B.M.; Li, J.; Bode, L.; Chen, X.
Enzymatic and Chemoenzymatic Syntheses of Disialyl Glycans and Their Necrotizing Enterocolitis Preventing
Effects. J. Org. Chem. 2017,82, 13152–13160. [CrossRef]
39.
Weijers, C.A.G.M.; Franssen, M.C.R.; Visser, G.M. Glycosyltransferase-catalyzed synthesis of bioactive
oligosaccharides. Biotechnol. Adv. 2008,26, 436–456. [CrossRef]
40.
US Food and Drug Administration. 3
0
-Sialyllactose Sodium Salt. GRAS Notice GRN 000766. Available
online: https://www.accessdata.fda.gov/scripts/fdcc/index.cfm?set=GRASNotices&id=766 (accessed on 27
March 2019).
41.
Woo, J.S.; Kim, B.-G.; Kim, D.H.; Choi, Y.H.; Song, J.-K.; Kang, S.Y.; Seo, W.M.; Yang, J.Y.; Lee, S.M. Method
for preparing sialic acid derivative. US Patent US 9637768 B2, 2 May 2017.
42.
Guo, Y.; Jers, C.; Meyer, A.S.; Arnous, A.; Li, H.; Kirpekar, F.; Mikkelsen, J.D. A Pasteurella multocida
sialyltransferase displaying dual trans-sialidase activities for production of 3
0
-sialyl and 6
0
-sialyl glycans. J.
Biotechnol. 2014,170, 60–67. [CrossRef]
43.
Cheng, J.; Yu, H.; Lau, K.; Huang, S.; Chokhawala, H.A.; Li, Y.; Tiwari, V.K.; Chen, X. Multifunctionality
of Campylobacter jejuni sialyltransferase CstII: Characterization of GD3/GT3 oligosaccharide synthase, GD3
oligosaccharide sialidase, and trans-sialidase activities. Glycobiology 2008,18, 686–697. [CrossRef]
44.
Cheng, J.; Huang, S.; Yu, H.; Lau, K.; Chen, X. Trans-sialidase activity of Photobacterium damsela
α
2,6-sialyltransferase and its application in the synthesis of sialosides. Glycobiology
2010
,20, 260–268.
[CrossRef]
45.
Schmölzer, K.; Ribitsch, D.; Czabany, T.; Luley-goedl, C.; Kokot, D.; Lyskowski, A.; Zitzenbacher, S.;
Schwab, H.; Nidetzky, B. Characterization of a multifunctional
α
2,3-sialyltransferase from Pasteurella dagmatis.
Glycobiology 2013,23, 1293–1304. [CrossRef]
46.
Fischöder, T.; Cajic, S.; Reichl, U.; Rapp, E.; Elling, L. Enzymatic Cascade Synthesis Provides Novel Linear
Human Milk Oligosaccharides as Reference Standards for xCGE-LIF Based High-Throughput Analysis.
Biotechnol. J. 2018,14, 1800305. [CrossRef]
47.
Bissaro, B.; Monsan, P.; Faur
é
, R.; O’Donohue, M.J. Glycosynthesis in a waterworld: new insight into
the molecular basis of transglycosylation in retaining glycoside hydrolases. Biochem. J.
2015
,467, 17–35.
[CrossRef]
48.
Holck, J.; Larsen, D.M.; Michalak, M.; Li, H.; Kjærulff, L.; Kirpekar, F.; Gotfredsen, C.H.; Forssten, S.;
Ouwehand, A.C.; Mikkelsen, J.D.; et al. Enzyme catalyzed production of sialylated human milk
oligosaccharides and galactooligosaccharides by Trypanosoma cruzi trans-sialidase. N. Biotechnol.
2014
,
31, 156–165. [CrossRef]
49.
Zeuner, B.; Jers, C.; Mikkelsen, J.D.; Meyer, A.S. Methods for improving enzymatic trans-glycosylation for
synthesis of human milk oligosaccharide biomimetics. J. Agric. Food Chem. 2014,62, 9615–9631. [CrossRef]
50.
Lundemo, P.; Karlsson, E.N.; Adlercreutz, P. Eliminating hydrolytic activity without affecting the
transglycosylation of a GH1
β
-glucosidase. Appl. Microbiol. Biotechnol.
2017
,101, 1121–1131. [CrossRef]
[PubMed]
51.
Ajisaka, K.; Fujimoto, H.; Isomura, M. Regioselective transglycosylation in the synthesis of oligosaccharides:
comparison of
β
-galactosidases and sialidases of various origins. Carbohydr. Res.
1994
,259, 103–115.
[CrossRef]

Molecules 2019,24, 2033 17 of 22
52.
Jers, C.; Michalak, M.; Larsen, D.M.; Kepp, K.P.; Li, H.; Guo, Y.; Kirpekar, F.; Meyer, A.S.; Mikkelsen, J.D.
Rational design of a new Trypanosoma rangeli trans-sialidase for efficient sialylation of glycans. PLoS ONE
2014,9, e83902. [CrossRef] [PubMed]
53.
Zeuner, B.; Muschiol, J.; Holck, J.; Lezyk, M.; Gedde, M.R.; Jers, C.; Mikkelsen, J.D.; Meyer, A.S. Substrate
specificity and transfucosylation activity of GH29
α
-l-fucosidases for enzymatic production of human milk
oligosaccharides. N. Biotechnol. 2018,41, 34–45. [CrossRef]
54.
Zeuner, B.; Vuillemin, M.; Holck, J.; Muschiol, J.; Meyer, A.S. Loop engineering of an
α
-1,3/4-l-fucosidase for
improved synthesis of human milk oligosaccharides. Enzyme Microb. Technol. 2018,115, 37–44. [CrossRef]
55.
Saumonneau, A.; Champion, E.; Peltier-Pain, P.; Molnar-Gabor, D.; Hendrickx, J.; Tran, V.; Hederos, M.;
Dekany, G.; Tellier, C. Design of an
α
-l-transfucosidase for the synthesis of fucosylated HMOs. Glycobiology
2016,26, 261–269. [CrossRef]
56.
Zeuner, B.; Gonz
á
lez-Delgado, I.; Holck, J.; Morales, G.; L
ó
pez-Muñoz, M.-J.; Segura, Y.; Meyer, A.S.;
Dalgaard Mikkelsen, J. Characterization and immobilization of engineered sialidases from Trypanosoma
rangeli for transsialylation. AIMS Mol. Sci. 2017,4, 140–163. [CrossRef]
57.
Guo, L.; Chen, X.; Xu, L.; Xiao, M.; Lu, L. Enzymatic synthesis of 6
0
-sialyllactose, a dominant sialylated
human milk oligosaccharide, by a novel exo-
α
-sialidase from Bacteroides fragilis NCTC9343. Appl. Environ.
Microbiol. 2018,84, e00071-18. [CrossRef] [PubMed]
58.
Nyffenegger, C.; Nordvang, R.T.; Zeuner, B.; Ł˛e˙zyk, M.; Difilippo, E.; Logtenberg, M.J.; Schols, H.A.;
Meyer, A.S.; Mikkelsen, J.D. Backbone structures in human milk oligosaccharides: trans-glycosylation by
metagenomic β-N-acetylhexosaminidases. Appl. Microbiol. Biotechnol. 2015,99, 7997–8009. [CrossRef]
59.
Jamek, S.B.; Mikkelsen, J.D.; Busk, P.K.; Meyer, A.S.; Holck, J.; Zeuner, B.; Muschiol, J. Loop Protein
Engineering for Improved Transglycosylation Activity of a
β
-N-Acetylhexosaminidase. ChemBioChem
2018
,
19, 1858–1865. [CrossRef]
60.
Zeuner, B.; Nyffenegger, C.; Mikkelsen, J.D.; Meyer, A.S. Thermostable
β
-galactosidases for the synthesis of
human milk oligosaccharides. N. Biotechnol. 2016,33, 355–360. [CrossRef] [PubMed]
61.
Murata, T.; Inukai, T.; Suzuki, M.; Yamagashi, M.; Usui, T. Facile enzymatic conversion of lactose into
lacto-N-tetraose and lacto-N-neotetraose. Glycoconj. J. 1999,16, 189–195. [CrossRef]
62.
Davies, G.J.; Wilson, K.S.; Henrissat, B. Nomenclature for sugar-binding subsites in glycosyl hydrolases.
Biochem. J. 1997,321, 557–559. [CrossRef]
63.
Kasche, V. Mechanism and yields in enzyme catalyzed equilibrium and kinetically controlled synthesis
of
β
-lactam antibiotics, peptides and other condensation products. Enzyme Microb. Technol.
1986
,8, 4–16.
[CrossRef]
64.
Koshland, D.E. Stereochemistry and the mechanism of enzymatic reactions. Biol. Rev.
1953
,28, 416–436.
[CrossRef]
65.
Vocadlo, D.J.; Withers, S.G. Detailed comparative analysis of the catalytic mechanisms of
β
-N-acetylglucosaminidases from families 3 and 20 of glycoside hydrolases. Biochemistry
2005
,44,
12809–12818. [CrossRef]
66.
Nordvang, R.T.; Nyffenegger, C.; Holck, J.; Jers, C.; Zeuner, B.; Sundekilde, U.K.; Meyer, A.S.; Mikkelsen, J.D.
It all starts with a sandwich: Identification of sialidases with trans-glycosylation activity. PLoS ONE
2016
,11,
e0158434. [CrossRef]
67.
Teze, D.; Hendrickx, J.; Czjzek, M.; Ropartz, D.; Sanejouand, Y.-H.; Tran, V.; Tellier, C.; Dion, M. Semi-rational
approach for converting a GH1
β
-glycosidase into a
β
-transglycosidase. Protein Eng. Des. Sel.
2014
,27,
13–19. [CrossRef]
68.
Van Rantwijk, F.; Woudenberg-Van Oosterom, M.; Sheldon, R.A. Glycosidase-catalyzed synthesis of alkyl
glycosides. J. Mol. Catal. - B Enzym. 1999,6, 511–532. [CrossRef]
69.
Wilbrink, M.H.; Kate, G.A.; Van, S.S.; Sanders, P.; Sallomons, E.; Johannes, A.; Dijkhuizen, L.; Kamerling, J.P.
Galactosyl-lactose sialylation using Trypanosoma cruzi trans-sialidase as the biocatalyst and bovine
κ
-casein-derived glycomacropeptide asthe donor substrate. Appl. Environ. Microbiol.
2014
,80, 5984–5991.
[CrossRef] [PubMed]
70.
Guo, Y.; Jers, C.; Meyer, A.S.; Li, H.; Kirpekar, F.; Mikkelsen, J.D. Modulating the regioselectivity of a
Pasteurella multocida sialyltransferase for biocatalytic production of 3
0
- and 6
0
-sialyllactose. Enzyme Microb.
Technol. 2015,78, 54–62. [CrossRef]

Molecules 2019,24, 2033 18 of 22
71.
Choi, K.-W.; Park, K.; Jun, S.-Y.; Park, C.; Park, K.; Cha, J. Modulation of the Regioselectivity of a Thermotoga
neapolitana
β
-glucosidase by site-directed mutagenesis. J. Microbiol. Biotechnol
2008
,18, 901–907. [PubMed]
72.
Bobrov, K.S.; Borisova, A.S.; Eneyskaya, E.V.; Ivanen, D.R.; Shabalin, K.A.; Kulminskaya, A.A.; Rychkov, G.N.
Improvement of the Efficiency of Transglycosylation Catalyzed by
α
-Galactosidase from Thermotoga maritima
by Protein Engineering. Biochem. 2013,78, 1112–1123. [CrossRef]
73.
Talens-Perales, D.; Polaina, J.; Mar
í
n-Navarro, J. Structural Dissection of the Active Site of Thermotoga maritima
β
-Galactosidase Identifies Key Residues for Transglycosylating Activity. J. Agric. Food Chem.
2016
,64,
2917–2924. [CrossRef] [PubMed]
74.
Nyffenegger, C.; Nordvang, R.T.; Jers, C.; Meyer, A.S.; Mikkelsen, J.D. Design of Trypanosoma rangeli sialidase
mutants with improved trans-sialidase activity. PLoS ONE 2017,12, e0171585. [CrossRef] [PubMed]
75.
Feng, H.-Y.; Drone, J.; Hoffmann, L.; Tran, V.; Tellier, C.; Rabiller, C.; Dion, M. Converting a
β
-glycosidase
into a β-transglycosidase by directed evolution. J. Biol. Chem. 2005,280, 37088–37097. [CrossRef]
76.
Placier, G.; Watzlawick, H.; Rabiller, C.; Mattes, R. Evolved
β
-galactosidases from Geobacillus stearothermophilus
with improved transgalactosylation yield for galacto-oligosaccharide production. Appl. Environ. Microbiol.
2009,75, 6312–6321. [CrossRef]
77.
Zhang, H.-P.; Dong, Y.-N.; Chen, W.; Zhang, H.; Liu, X.-M.; Xia, Y.; Chen, H.-Q. Enhancement of the hydrolysis
activity of
β
-galactosidase from Geobacillus stearothermophilus by saturation mutagenesis. J. Dairy Sci.
2011
,
94, 1176–1184.
78. Hayes, M.R.; Pietruszka, J. Synthesis of glycosides by glycosynthases. Molecules 2017,22, 1434. [CrossRef]
79.
Giddens, J.P.; Lomino, J.V.; Amin, M.N.; Wang, L.X. Endo-F3 glycosynthase mutants enable chemoenzymatic
synthesis of core-fucosylated triantennary complex type glycopeptides and glycoproteins. J. Biol. Chem.
2016,291, 9356–9370. [CrossRef]
80.
Mackenzie, L.F.; Wang, Q.; Warren, R.A.J.; Withers, S.G. Glycosynthases: Mutant Glycosidases for
Oligosaccharide Synthesis. J. Am. Chem. Soc. 1998,120, 5583–5584. [CrossRef]
81.
Henze, M.; Schmidtke, S.; Hoffmann, N.; Steffens, H.; Pietruszka, J.; Elling, L. Combination of
Glycosyltransferases and a Glycosynthase in Sequential and One-Pot Reactions for the Synthesis of Type 1
and Type 2 N-Acetyllactosamine Oligomers. ChemCatChem 2015,7, 3131–3139. [CrossRef]
82.
Cobucci-Ponzano, B.; Conte, F.; Bedini, E.; Corsaro, M.M.; Parrilli, M.; Sulzenbacher, G.; Lipski, A.; Dal
Piaz, F.; Lepore, L.; Rossi, M.; et al.
β
-Glycosyl Azides as Substrates for
α
-Glycosynthases: Preparation of
Efficient α-l-Fucosynthases. Chem. Biol. 2009,16, 1097–1108. [CrossRef] [PubMed]
83.
Sakurama, H.; Fushinobu, S.; Hidaka, M.; Yoshida, E.; Honda, Y.; Ashida, H.; Kitaoka, M.; Kumagai, H.;
Yamamoto, K.; Katayama, T. 1,3-1,4-
α
-l-Fucosynthase that specifically introduces Lewis a/x antigens into
type-1/2 chains. J. Biol. Chem. 2012,287, 16709–16719. [CrossRef]
84.
Sugiyama, Y.; Gotoh, A.; Katoh, T.; Honda, Y.; Yoshida, E.; Kurihara, S.; Ashida, H.; Kumagai, H.;
Yamamoto, K.; Kitaoka, M.; et al. Introduction of H-antigens into oligosaccharides and sugar chains of
glycoproteins using highly efficient 1,2-α-l-fucosynthase. Glycobiology 2016,26, 1235–1247. [CrossRef]
85.
Wada, J.; Honda, Y.; Nagae, M.; Kato, R.; Wakatsuki, S.; Katayama, T.; Taniguchi, H.; Kumagai, H.; Kitaoka, M.;
Yamamoto, K. 1,2-
α
-l-Fucosynthase: A glycosynthase derived from an inverting
α
-glycosidase with an
unusual reaction mechanism. FEBS Lett. 2008,582, 3739–3743. [CrossRef] [PubMed]
86.
Ohnuma, T.; Fukuda, T.; Dozen, S.; Honda, Y.; Kitaoka, M.; Fukamizo, T. A glycosynthase derived from
an inverting GH19 chitinase from the moss Bryum coronatum.Biochem. J.
2012
,444, 437–443. [CrossRef]
[PubMed]
87.
Martinez, E.A.; Boer, H.; Koivula, A.; Samain, E.; Driguez, H.; Armand, S.; Cottaz, S. Engineering chitinases
for the synthesis of chitin oligosaccharides: Catalytic amino acid mutations convert the GH-18 family
glycoside hydrolases into transglycosylases. J. Mol. Catal. B Enzym. 2012,74, 89–96. [CrossRef]
88.
Umekawa, M.; Huang, W.; Li, B.; Fujita, K.; Ashida, H.; Wang, L.-X.; Yamamoto, K. Mutants of Mucor hiemalis
Endo-β-N-acetylglucosaminidase Show Enhanced Transglycosylation and Glycosynthase-like Activities. J.
Biol. Chem. 2008,283, 4469–4479. [CrossRef]
89.
Sl
á
mov
á
, K.; Kapešov
á
, J.; Kulik, N.; Kˇren, V. Transglycosylation activity of glycosynthase-type mutants
of
β
-N-acetylhexosaminidase from Talaromyces flavus. In Proceedings of the Carbohydrate Bioengineering
Meeting, Toulouse, France, 19–22 May 2019; p. 149.

Molecules 2019,24, 2033 19 of 22
90.
Osanjo, G.; Dion, M.; Drone, J.; Solleux, C.; Tran, V.; Rabiller, C.; Tellier, C. Directed evolution of the
α
-l-fucosidase from Thermotoga maritima into an
α
-l-transfucosidase. Biochemistry
2007
,46, 1022–1033.
[CrossRef] [PubMed]
91.
Teze, D.; Dion, M.; Daligault, F.; Tran, V.; Andr
é
-Miral, C.; Tellier, C. Alkoxyamino glycoside acceptors for
the regioselective synthesis of oligosaccharides using glycosynthases and transglycosidases. Bioorganic Med.
Chem. Lett. 2013,23, 448–451. [CrossRef]
92.
Lombard, V.; Golaconda Ramulu, H.; Drula, E.; Coutinho, P.M.; Henrissat, B. The carbohydrate-active
enzymes database (CAZy) in 2013. Nucleic Acids Res. 2014,42, D490-5. [CrossRef]
93.
Lundemo, P.; Adlercreutz, P.; Karlsson, E.N. Improved Transferase/Hydrolase Ratio through Rational Design
of a Family 1
β
-Glucosidase from Thermotoga neapolitana.Appl. Environ. Microbiol.
2013
,79, 3400–3405.
[CrossRef]
94.
Wu, Y.; Yuan, S.; Chen, S.; Wu, D.; Chen, J.; Wu, J. Enhancing the production of galacto-oligosaccharides by
mutagenesis of Sulfolobus solfataricus β-galactosidase. Food Chem. 2013,138, 1588–1595. [CrossRef]
95.
Teze, D.; Daligault, F.; Ferri
è
res, V.; Sanejouand, Y.H.; Tellier, C. Semi-rational approach for converting a
GH36 α-glycosidase into an α-transglycosidase. Glycobiology 2015,25, 420–427. [CrossRef]
96.
Armand, S.; Andrews, S.R.; Charnock, S.J.; Gilbert, H.J. Influence of the Aglycone Region of the Substrate
Binding Cleft of Pseudomonas Xylanase 10A on Catalysis. Biochemistry 2001,40, 7404–7409. [CrossRef]
97.
Hansson, T.; Kaper, T.; van der Oost, J.; de Vos, W.M.; Adlercreutz, P. Improved oligosaccharide synthesis by
protein engineering of
β
-glucosidase CelB from hyperthermophilic Pyrococcus furiosus.Biotechnol. Bioeng.
2001,73, 203–210. [CrossRef]
98.
Kelly, R.M.; Leemhuis, H.; Rozeboom, H.J.; van Oosterwijk, N.; Dijkstra, B.W.; Dijkhuizen, L. Elimination of
competing hydrolysis and coupling side reactions of a cyclodextrin glucanotransferase by directed evolution.
Biochem. J. 2008,413, 517–525. [CrossRef]
99.
Paris, G.; Ratier, L.; Amaya, M.F.; Nguyen, T.; Alzari, P.M.; Frasch, A.C.C. A sialidase mutant displaying
trans-sialidase activity. J. Mol. Biol. 2005,345, 923–934. [CrossRef]
100.
Tran, V.; Hoffmann, L.; Rabiller, C.; Tellier, C.; Dion, M. Rational design of a GH1
β
-glycosidase to prevent
self-condensation during the transglycosylation reaction. Protein Eng. Des. Sel. 2010,23, 43–49. [CrossRef]
101.
Mitchell, F.L.; Miles, S.M.; Neres, J.; Bichenkova, E.V.; Bryce, R.A. Tryptophan as a molecular shovel in
the glycosyl transfer activity of Trypanosoma cruzi trans-sialidase. Biophys. J.
2010
,98, 38–40. [CrossRef]
[PubMed]
102.
Pierdominici-Sottile, G.; Palma, J.; Roitberg, A.E. Free-energy computations identify the mutations required
to confer trans-sialidase activity into Trypanosoma rangeli sialidase. Proteins Struct. Funct. Bioinforma.
2014
,
82, 424–435. [CrossRef]
103.
Teze, D.; Hendrickx, J.; Dion, M.; Tellier, C.; Woods, V.L.; Tran, V.; Sanejouand, Y.-H. Conserved Water
Molecules in Family 1 Glycosidases: A DXMS and Molecular Dynamics Study. Biochemistry
2013
,52,
5900–5910. [CrossRef]
104.
Romero-T
é
llez, S.; Lluch, J.M.; Gonz
á
lez-Lafont,
À
.; Masgrau, L. Comparing Hydrolysis and
Transglycosylation Reactions Catalyzed by Thermus thermophilus
β
-Glycosidase. A Combined MD and
QM/MM Study. Front. Chem. 2019,7, 1–19. [CrossRef]
105.
Piens, K.; Faur
é
, R.; Sundqvist, G.; Baumann, M.J.; Saura-Valls, M.; Teeri, T.T.; Cottaz, S.; Planas, A.;
Driguez, H.; Brumer, H. Mechanism-based labeling defines the free energy change for formation of the
covalent glycosyl-enzyme intermediate in a xyloglucan endo-transglycosylase. J. Biol. Chem.
2008
,283,
21864–21872. [CrossRef]
106.
Raich, L.; Borodkin, V.; Fang, W.; Castro-L
ó
pez, J.; Van Aalten, D.M.F.; Hurtado-Guerrero, R.; Rovira, C. A
Trapped Covalent Intermediate of a Glycoside Hydrolase on the Pathway to Transglycosylation. Insights
from Experiments and Quantum Mechanics/Molecular Mechanics Simulations. J. Am. Chem. Soc.
2016
,138,
3325–3332. [CrossRef]
107.
Arab-Jaziri, F.; Bissaro, B.; Tellier, C.; Dion, M.; Faur
é
, R.; Donohue, M.J.O. Enhancing the chemoenzymatic
synthesis of arabinosylated xylo-oligosaccharides by GH51
α
-l-arabinofuranosidase. Carbohydr. Res.
2015
,
401, 64–72. [CrossRef]
108.
Kon
é
, F.M.T.; Le B
é
chec, M.; Sine, J.-P.; Dion, M.; Tellier, C. Digital screening methodology for the directed
evolution of transglycosidases. Protein Eng. Des. Sel. 2009,22, 37–44. [CrossRef] [PubMed]

Molecules 2019,24, 2033 20 of 22
109.
Enam, F.; Mansell, T.J. Linkage-Specific Detection and Metabolism of Human Milk Oligosaccharides in
Escherichia coli.Cell Chem. Biol. 2018,25, 1292–1303. [CrossRef]
110.
Hou, X.E.Z.; Ang, Y.W.; Eehan, E.J.M.; Hen, L.C. Mutation of a Conserved Tryptophan in the Chitin-Binding
Cleft of Serratia marcescens Chitinase A Enhances Transglycosylation. Biosci. Biotechnol. Biochem.
2006
,70,
243–251.
111.
Bissaro, B.; Durand, J.; Biarn
é
s, X.; Planas, A.; Monsan, P.; O’Donohue, M.J.; Faur
é
, R. Molecular Design
of Non-Leloir Furanose-Transferring Enzymes from an
α
-l-Arabinofuranosidase: A Rationale for the
Engineering of Evolved Transglycosylases. ACS Catal. 2015,5, 4598–4611. [CrossRef]
112.
Hassan, N.; Geiger, B.; Gandini, R.; Patel, B.K.C.; Kittl, R.; Haltrich, D.; Nguyen, T.H.; Divne, C.; Tan, T.C.
Engineering a thermostable Halothermothrix orenii
β
-glucosidase for improved galacto-oligosaccharide
synthesis. Appl. Microbiol. Biotechnol. 2016,100, 3533–3543. [CrossRef] [PubMed]
113.
Zhou, Y.; Gao, R.; Li, J.; Wang, Q.; Guo, Z.; Yang, J. Engineering T. naphthophila
β
-glucosidase for enhanced
synthesis of galactooligosaccharides by site-directed mutagenesis. Biochem. Eng. J. 2017,127, 1–8.
114.
Larsbrink, J.; Izumi, A.; Hemsworth, G.R.; Davies, G.J.; Brumer, H. Structural Enzymology of Cellvibrio
japonicus Agd31B Protein Reveals
α
-Transglucosylase Activity in Glycoside Hydrolase Family 31. J. Biol.
Chem. 2012,287, 43288–43299. [CrossRef] [PubMed]
115.
Amaya, M.F.; Watts, A.G.; Damager, I.; Wehenkel, A.; Nguyen, T.; Buschiazzo, A.; Frasch, A.C.; Withers, S.G.;
Alzari, P.M. Structural Insights into the Catalytic Mechanism of Trypanosoma cruzi trans-Sialidase. Structure
2004,12, 775–784. [CrossRef] [PubMed]
116.
Busk, P.K.; Lange, L. Function-Based Classification of Carbohydrate-Active Enzymes by Recognition of Short,
Conserved Peptide Motifs. Appl. Environ. Microbiol. 2013,79, 3380–3391. [CrossRef] [PubMed]
117.
Wilkens, C.; Busk, P.K.; Pilgaard, B.; Zhang, W.J.; Nielsen, K.L.; Nielsen, P.H. Diversity of microbial
carbohydrate-active enzymes in Danish anaerobic digesters fed with wastewater treatment sludge. Biotechnol.
Biofuels 2017,10, 1–14. [CrossRef] [PubMed]
118.
Ochoa-Leyva, A.; Sober
ó
n, X.; S
á
nchez, F.; Argüello, M.; Montero-Mor
á
n, G.; Saab-Rinc
ó
n, G. Protein Design
through Systematic Catalytic Loop Exchange in the (
β
/
α
)
8
Fold. J. Mol. Biol.
2009
,387, 949–964. [CrossRef]
[PubMed]
119.
Ochoa-Leyva, A.; Barona-G
ó
mez, F.; Saab-Rinc
ó
n, G.; Verdel-Aranda, K.; S
á
nchez, F.; Sober
ó
n, X. Exploring
the Structure–Function Loop Adaptability of a (
β
/
α
)
8
-Barrel Enzyme through Loop Swapping and Hinge
Variability. J. Mol. Biol. 2011,411, 143–157. [CrossRef] [PubMed]
120.
De Luca, F.; Benvenuti, M.; Carboni, F.; Pozzi, C.; Rossolini, G.M.; Mangani, S.; Docquier, J.-D. Evolution to
carbapenem-hydrolyzing activity in noncarbapenemase class D
β
-lactamase OXA-10 by rational protein
design. Proc. Natl. Acad. Sci. 2011,108, 18424–18429. [CrossRef] [PubMed]
121.
Schiano-di-Cola, C.; Røjel, N.; Jensen, K.; Kari, J.; Sørensen, T.H.; Borch, K.; Westh, P. Systematic deletions in
the cellobiohydrolase (CBH) Cel7A from the fungus Trichoderma reesei reveal flexible loops critical for CBH
activity. J. Biol. Chem. 2019,294, 1807–1815. [CrossRef] [PubMed]
122.
Borisova, A.S.; Eneyskaya, E.V.; Jana, S.; Badino, S.F.; Kari, J.; Amore, A.; Karlsson, M.; Hansson, H.;
Sandgren, M.; Himmel, M.E.; et al. Correlation of structure, function and protein dynamics in GH7
cellobiohydrolases from Trichoderma atroviride,T. reesei and T. harzianum.Biotechnol. Biofuels
2018
,11, 1–22.
[CrossRef]
123.
Hoque, M.A.; Zhang, Y.; Chen, L.; Yang, G.; Khatun, M.A.; Chen, H.; Hao, L.; Feng, Y. Stepwise Loop
Insertion Strategy for Active Site Remodeling to Generate Novel Enzyme Functions. ACS Chem. Biol. 2017,
12, 1188–1193. [CrossRef] [PubMed]
124.
Hayes, F.; Hallet, B.; Cao, Y. Insertion mutagenesis as a tool in the modification of protein function. Extended
substrate specificity conferred by pentapeptide insertions in the
Ω
-loop of TEM-1
β
-lactamase. J. Biol. Chem.
1997,272, 28833–28836. [CrossRef]
125.
Guzm
á
n-Rodr
í
guez, F.; Alatorre-Santamar
í
a, S.; G
ó
mez-Ruiz, L.; Rodr
í
guez-Serrano, G.; Garc
í
a-Garibay, M.;
Cruz-Guerrero, A. Employment of fucosidases for the synthesis of fucosylated oligosaccharides with
biological potential. Biotechnol. Appl. Biochem. 2019,66, 172–191. [CrossRef]
126.
Thomä-Worringer, C.; Sørensen, J.; L
ó
pez-Fandiño, R. Health effects and technological features of
caseinomacropeptide. Int. Dairy J. 2006,16, 1324–1333. [CrossRef]
127. Smithers, G.W. Whey and whey proteins-From “gutter-to-gold”. Int. Dairy J. 2008,18, 695–704. [CrossRef]

Molecules 2019,24, 2033 21 of 22
128.
Neelima; Sharma, R.; Rajput, Y.S.; Mann, B. Chemical and functional properties of glycomacropeptide (GMP)
and its role in the detection of cheese whey adulteration in milk: A review. Dairy Sci. Technol.
2013
,93, 21–43.
[CrossRef] [PubMed]
129.
Saito, T.; Itoh, T. Variations and Distributions of O-Glycosidically Linked Sugar Chains in Bovine
κ
-Casein. J.
Dairy Sci. 1992,75, 1768–1774. [CrossRef]
130.
McJarrow, P.; Garman, J.; Harvey, S.; Van Amelsfort, A. Dairy process and product. US Patent WO2003/049547
A2, 19 June 2003.
131.
Luo, J.; Nordvang, R.T.; Morthensen, S.T.; Zeuner, B.; Meyer, A.S.; Mikkelsen, J.D.; Pinelo, M. An integrated
membrane system for the biocatalytic production of 3
0
-sialyllactose from dairy by-products. Bioresour.
Technol. 2014,166, 9–16. [CrossRef] [PubMed]
132.
Luo, J.; Morthensen, S.T.; Meyer, A.S.; Pinelo, M. Filtration behavior of casein glycomacropeptide (CGMP) in
an enzymatic membrane reactor: Fouling control by membrane selection and threshold flux operation. J.
Memb. Sci. 2014,469, 127–139. [CrossRef]
133.
Nordvang, R.T.; Luo, J.; Zeuner, B.; Prior, R.; Andersen, M.F.; Mikkelsen, J.D.; Meyer, A.S.; Pinelo, M.
Separation of 3
0
-sialyllactose and lactose by nanofiltration: A trade-offbetween charge repulsion and pore
swelling induced by high pH. Sep. Purif. Technol. 2014,138, 77–83. [CrossRef]
134.
Zeuner, B.; Luo, J.; Nyffenegger, C.; Aumala, V.; Mikkelsen, J.D.; Meyer, A.S. Optimizing the biocatalytic
productivity of an engineered sialidase from Trypanosoma rangeli for 3
0
-sialyllactose production. Enzyme
Microb. Technol. 2014,55, 85–93. [CrossRef] [PubMed]
135.
Sallomons, E.; Wilbrink, M.H.; Sanders, P.; Kamerling, J.P.; Van Vuure, C.A.; Hage, J.A. Methods for providing
sialylated oligosaccharides. US Patent 9539270 B2, 10 January 2017.
136.
Pelletier, M.; Barker, W.A.; Hakes, D.J.; Zopf, D.A. Methods for producing sialyloligosaccharides in a dairy
source. US Patent 6323008 B1, 27 November 2001.
137.
Lee, S.G.; Shin, D.H.; Kim, B.G. Production of sialyloligosaccharides by trans-sialidase catalyzed reaction
using fetuin as a sialic acid donor. Enzyme Microb. Technol. 2002,31, 742–746. [CrossRef]
138.
Lin, B.X.; Qiao, Y.; Shi, B.; Tao, Y. Polysialic acid biosynthesis and production in Escherichia coli: current state
and perspectives. Appl. Microbiol. Biotechnol. 2016,100, 1–8. [CrossRef]
139.
Bah, C.S.F.; Bekhit, A.E.D.A.; Carne, A.; Mcconnell, M.A. Slaughterhouse blood: An emerging source of
bioactive compounds. Compr. Rev. Food Sci. Food Saf. 2013,12, 314–331. [CrossRef]
140.
Wilbrink, M.H.; Ten Kate, G.A.; Sanders, P.; Gerwig, G.J.; Van Leeuwen, S.S.; Sallomons, E.; Klarenbeek, B.;
Hage, J.A.; Van Vuure, C.A.; Dijkhuizen, L.; et al. Enzymatic Decoration of Prebiotic Galacto-oligosaccharides
(Vivinal GOS) with Sialic Acid Using Trypanosoma cruzi trans-Sialidase and Two Bovine Sialoglycoconjugates
as Donor Substrates. J. Agric. Food Chem. 2015,63, 5976–5984. [CrossRef]
141.
Dhar, C.; Sasmal, A.; Varki, A. From “Serum Sickness” to “Xenosialitis”: Past, Present, and Future Significance
of the Non-human Sialic Acid Neu5Gc. Front. Immunol. 2019,10, 807. [CrossRef]
142.
Pearce, O.M.T.; Läubli, H. Sialic acids in cancer biology and immunity. Glycobiology
2016
,26, 111–128.
[CrossRef] [PubMed]
143.
De Smet, S.; Vossen, E. Meat: The balance between nutrition and health. A review. Meat Sci.
2016
,120,
145–156. [CrossRef]
144.
Harish Prashanth, K.V.; Tharanathan, R.N. Chitin/chitosan: modifications and their unlimited application
potential-an overview. Trends Food Sci. Technol. 2007,18, 117–131. [CrossRef]
145.
Synowiecki, J.; Al-Khateeb, N.A. Production, Properties, and Some New Applications of Chitin and Its
Derivatives. Crit. Rev. Food Sci. Nutr. 2003,43, 145–171. [CrossRef]
146.
Arbia, W.; Arbia, L.; Adour, L.; Amrana, A. Chitin Extraction from Crustacean Shells by Biological Methods—A
review. Food Technol. Biotechnol. 2013,51, 12–25.
147.
Jamek, S.B.; Nyffenegger, C.; Muschiol, J.; Holck, J.; Meyer, A.S.; Mikkelsen, J.D. Characterization of two
novel bacterial type A exo-chitobiose hydrolases having C-terminal 5/12-type carbohydrate-binding modules.
Appl. Microbiol. Biotechnol. 2017,101, 4533–4546. [CrossRef]
148.
Vaaje-Kolstad, G.; Horn, S.J.; Sørlie, M.; Eijsink, V.G.H. The chitinolytic machinery of Serratia marcescens—A
model system for enzymatic degradation of recalcitrant polysaccharides. FEBS J.
2013
,280, 3028–3049.
[CrossRef] [PubMed]

Molecules 2019,24, 2033 22 of 22
149.
Nguyen-Thi, N.; Doucet, N. Combining chitinase C and N-acetylhexosaminidase from Streptomyces coelicolor
A3(2) provides an efficient way to synthesize N-acetylglucosamine from crystalline chitin. J. Biotechnol.
2016
,
220, 25–32. [CrossRef]
150.
Mekasha, S.; Toupalov
á
, H.; Linggadjaja, E.; Tolani, H.A.; Andˇera, L.; Arntzen, M.; Vaaje-Kolstad, G.;
Eijsink, V.G.H.; Agger, J.W. A novel analytical method for d-glucosamine quantification and its application
in the analysis of chitosan degradation by a minimal enzyme cocktail. Carbohydr. Res.
2016
,433, 18–24.
[CrossRef] [PubMed]
151.
Warmerdam, A.; Paudel, E.; Jia, W.; Boom, R.M.; Janssen, A.E.M. Characterization of
β
-Galactosidase
Isoforms from Bacillus circulans and Their Contribution to GOS Production. Appl. Biochem. Biotechnol.
2013
,
170, 340–358. [CrossRef]
152.
Hotchkiss, A.T.; Nuñez, A.; Strahan, G.D.; Chau, H.K.; White, A.K.; Marais, J.P.J.; Hom, K.; Vakkalanka, M.S.;
Di, R.; Yam, K.L.; et al. Cranberry Xyloglucan Structure and Inhibition of Escherichia coli Adhesion to
Epithelial Cells. J. Agric. Food Chem. 2015,63, 5622–5633. [CrossRef] [PubMed]
153.
Kato, Y.; Noro, O.; Azuma, Y. Analysis of the Oligosaccharides of Xyloglucan in Peanut Hulls (Studies on
Production of Fucose-containing Xyloglucan Oligosaccharide Part II). Nippon Shokuhin Kagaku Kogaku Kaishi
2000,47, 560–563. [CrossRef]
154.
Arumugam, N.; Biely, P.; Puchart, V.; Singh, S.; Pillai, S. Structure of peanut shell xylan and its conversion to
oligosaccharides. Process Biochem. 2018,72, 124–129. [CrossRef]
155.
Cao, H.T.T.; Mikkelsen, M.D.; Lezyk, M.J.; Bui, L.M.; Tran, V.T.T.; Silchenko, A.S.; Kusaykin, M.I.; Pham, T.D.;
Truong, B.H.; Holck, J.; et al. Novel Enzyme Actions for Sulphated Galactofucan Depolymerisation and a
New Engineering Strategy for Molecular Stabilisation of Fucoidan Degrading Enzymes. Mar. Drugs
2018
,16,
422. [CrossRef] [PubMed]
156.
Berteau, O.; McCort, I.; Goasdou
é
, N.; Tissot, B.; Daniel, R. Characterization of a new
α
-l-fucosidase isolated
from the marine mollusk Pecten maximus that catalyzes the hydrolysis of
α
-l-fucose from algal fucoidan
(Ascophyllum nodosum). Glycobiology 2002,12, 273–282. [CrossRef] [PubMed]
157. Perrella, N.N.; Withers, S.G.; Lopes, A.R. Identity and role of the non-conserved acid/base catalytic residue
in the GH29 fucosidase from the spider Nephilingis cruentata.Glycobiology
2018
,28, 925–932. [CrossRef]
[PubMed]
158.
Fan, S.; Zhang, H.; Chen, X.; Lu, L.; Xu, L.; Xiao, M. Cloning, characterization, and production of three
α
-l-fucosidases from Clostridium perfringens ATCC 13124. J. Basic Microbiol.
2016
,56, 347–357. [CrossRef]
[PubMed]
159.
Jung, S.M.; Park, Y.C.; Seo, J.H. Production of 3-Fucosyllactose in Engineered Escherichia coli with
α-1,3-Fucosyltransferase from Helicobacter pylori.Biotechnol. J. 2019,1800498, 1–7. [CrossRef]
160.
Champion, E.; McConnell, B.; Dekany, G. Ternary mixtures of 6
0
SL, LNnT and LST c. US Patent
US2018/0161350 A1, 14 June 2018.
161.
Vogel, A.; Schmiedel, R.; Champion, E.; Dekany, G. Mutated sialidases. US Patent US2018/0163185 A1, 14
June 2018.
162.
Champion, E.; Vogel, A.; Bartsch, S.; Dekany, G. Mutated fucosidase. US Patent WO2016/063261 A1, 28 April.
©
2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).