REVIEWS
Recent Advances in the Oligosaccharide Synthesis Promoted by
Catalytically Engineered Glycosidases
Giuseppe Perugino,a Beatrice Cobucci-Ponzano,a Mosè Rossi,a, b Marco Moraccia,*
a
b
Institute of Protein Biochemistry – Consiglio Nazionale delle Ricerche, Via P. Castellino 111, 80131, Naples, Italy
Fax ( þ 39)-081-6132277, e-mail m.moracci@ibp.cnr.it
Dipartimento di Biologia Strutturale e Funzionale, Università di Napoli “Federico II”, Via Cinthia 16,
80126, Naples, Italy
Received: February 2, 2005; Accepted: March 31, 2005
Abstract: In oligosaccharide synthesis the chemo-enzymatic approach is often chosen to obtain the desired glycosidic bond by exploiting the exquisite substrate specificity and stereoselectivity of glycosidases
and glycosyltransferases. In recent years the invention
of new enzymatic activities, named glycosynthases,
created a third route for the enzymatic synthesis of
oligosaccharides. Glycosynthases are mutant glycosidases, devoid of hydrolytic activity, that produce oligosaccharides with yields > 80%. The glycosynthase
technology is now well established and was greatly
improved recently. Several groups experimentally
proved that this methodology can be applied to several glycosidases and that the activity of existing glycosynthases can be significantly enhanced. In this review
we briefly describe the characteristics of these synthetic tools and show a survey of the most recent findings. Methods for the prediction of the glycosynthetic
potential of known glycosidases are also discussed.
1 Introduction
The challenges that are faced in the synthesis of carbohydrates involve the presence on the monosaccharides
of several functional groups and chiral centres. In these
regards, chemical synthesis offers exceptional flexibility
providing the means to generate any oligosaccharide,
oligosaccharide analogue, or glycoconjugate. Chemical
synthesis of oligosaccharides requires stereochemical
and regiochemical control in glycosidic linkage formation by sequential protection-deprotection strategies
of the functional groups present on the monosaccharides. These procedures often result in time-consuming
manipulations and low yields, hindering the efficient
production of oligosaccharides that is needed for biological testing.[1]
Enzymatic and chemo-enzymatic synthesis represent
alternative tools to obtain the desired glycosidic bond
by exploiting the exquisite substrate specificity and stereoselectivity of glycosidases and glycosyltransferases.[1,2]
For instance, the use of a multi-enzymatic system mimAdv. Synth. Catal. 2005, 347, 941 – 950
DOI: 10.1002/adsc.200505070
1
2
3
4
5
6
Introduction
Transformation of a Glycosidase into a Glycosynthase
Thioglycoligases and Thioglycosynthases
Molecular Bases of the Glycosynthase Mechanism
Strategies for the Improvement of the Glycosynthases Activity
How to Predict the Glycosynthetic Potential of a
Glycosidase
Keywords: chemical complementation; directed evolution; kinetics; protein engineering; reaction mechanisms; transglycosylation
icking the natural pathway in the Golgi apparatus allowed the synthesis of the (antithrombin III)-binding
pentasaccharide in only six steps, outrunning the traditional chemical route that consists in 60 steps.[3]
Enzymatic oligosaccharide synthesis is performed by
two main classes of catalysts: glycosidases and glycosyltransferases. Glycosidases are classified on the basis of
their amino acidic sequence into about 100 families
and 14 superfamilies or clans showing the same reaction
mechanism and 3D structure (http://afmb.cnrs-mrs.fr/
CAZY/index.html). Thanks to this valuable classification any new glycosidase can be assigned to an existing
family and the residues involved in the catalytic activity
or the 3D structure can be easily identified. The oligosaccharide synthesis promoted by glycosidases is accomplished either by displacing the equilibrium with a large
excess of acceptor (thermodynamically controlled synthesis) or by using activated glycosyl donors (kinetically
controlled transglycosylation) (Scheme 1A).[4] However, typically, retaining glycosidases give only modest synthetic yields (10 – 40%) precluding their exploitation in
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
941
Giuseppe Perugino et al.
REVIEWS
Giuseppe
Perugino
(1971) studied Biology
at the University of Naples “Federico II” where
he graduated with honours in 1997 by making
an experimental thesis
on enzymology in the Institute of Protein Biochemistry, CNR. He received his PhD in 2003
in biochemistry and molecular biology and spent
part of his training in the Department of Microbiology of the University of Wageningen (The Netherlands) under the supervision of Prof. W. de Vos.
He is involved since the late 1990s in the study of
glycosidases from hyperthermophiles and the chemo-enzymatic synthesis of oligosaccharides.
Beatrice Cobucci-Ponzano (1971) graduated
with honours in 1997 at
the University of Naples
“Federico II” in Biology
by making an experimental thesis in enzymology
in the Institute of Protein
Biochemistry, CNR. She
received her PhD in
2002 in biochemistry and
biophysics at the University of Padua (Italy) and
then she joined the Institute of Protein Biochemistry as staff researcher. Her research interests include the biochemistry and molecular biology of
extremophilic organisms and the enzymology of
glycosidases from hyperthermophiles.
large-scale synthesis (Scheme 1A). Glycosyltransferases are classified as belonging to the Leloir or the non-Leloir pathway on the basis of their substrate specificity.[5]
The former enzymes transfer the glycone moiety from a
nucleotidyl sugar substrate to an acceptor, while the enzymes of the non-Leloir pathway use glycosyl phosphates as donors. Despite their high specificity and almost quantitative yields, glycosyltransferases are hampered by their scarce availability and by the high costs
of the substrates.[1]
942
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Mosè Rossi (1938) graduated with honours in
1961 at the University of
Naples in Chemistry. In
1968 he was EMBO Fellow at MRC, Cambridge
(UK) – Department of
Molecular Biology in
Prof. J. I. Harris laboratory; in 1970 – 1972 he
was P. H. S. Fellow at
Stanford University in Prof. D. O. Woodwords
laboratory at the Department of Biological Sciences, Stanford, California (USA). He has been senior scientist in the Institute of Protein Biochemistry, CNR, since 1969 and director of the institute
from 1986 on and full professor of enzymology at
the University of Naples “Federico II” from 1981.
His primary research interests are in the field of
the application of enzymes in biotechnology. He
is one of the pioneers in the isolation and characterisation of enzymes and proteins from extremophilic organisms.
Marco Moracci (1962)
graduated in Biology at
the University of Naples
in 1987 with an experimental thesis on enzymology. He was post-doc
in the Department of
Chemistry of Cambridge
University (UK) under
the supervision of Prof.
A. R. Fersht. He joined the Institute of Protein Biochemistry, CNR, in 1992. He is senior researcher
in this institution since 2000 and he teaches enzymology at the University of Naples “Federico II”
since 2002. He is involved from late 1980s in the
study of extremophilic organisms and the structure-function of hyperstable proteins. He developed the glycosynthase approach on hyperthermophilic glycosidases.
A third route for the enzymatic synthesis of oligosaccharides appeared in the last decade with the invention
reported by Withers and colleagues in 1998 who described for the first time that a mutant of a b-glycosidase, devoid of hydrolytic activity, could produce oligosaccharides with yields > 80%. This new class of enzymes was
named glycosynthase.[6]
The glycosynthase approach is now well established
and has been described in detail in several reviews.[7]
Nevertheless, recent papers disclosed several aspects
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 941 – 950
Oligosaccharide Synthesis Promoted by Catalytically Engineered Glycosidases
of the glycosynthase mechanism and addressed the
question of how the synthetic potential of an existing
glycosynthase can be improved/expanded.
In this review we briefly describe the characteristics of
these novel synthetic tools and show a survey of the most
recent findings. Methods to predict the glycosynthetic
potential of known glycosidases are also discussed.
2 Transformation of a Glycosidase into a
Glycosynthase
The glycosynthase technology arose from mechanistic
studies performed in Withers laboratory on the b-glycosidase from Agrobacterium sp.[8] In particular, these authors showed that the modification of the catalytic nucleophile Glu358 into a residue unable to perform a nucleophilic attack (i.e., Ala or Gly) inactivates the enzyme, but small ions, like sodium azide, acting as external nucleophiles, restore the activity. The analysis of
the products obtained by reactivation enables the identification of the residue of interest (for a review of the
common methods for the analysis of the active site-residues in glycosidases, see [7a]).
A consequence of this modification is that the reaction
mechanism of the enzyme was changed from retaining
to inverting and it could use a glycosyl fluoride with opposite anomeric configuration to that of the normal substrate (a-glucosyl fluorides vs. b-d-glucosides).[8] Successively, Withers and colleagues showed that, in the
presence of a-glucosyl fluorides and suitable acceptors,
the nucleophile mutant synthesised oligosaccharides
acting as a glycosynthase.[6]
From this first report, the glycosynthase technology
was actively developed in other laboratories in the world
leading to a variety of novel activities including other
exo-glycosynthases, endo-glycosynthases, hyperthermophilic exo- and endo-glycosynthases, a-glycosynthases, mannosynthases, mannansynthases, and galactosynthases.[9]
The two main glycosynthase modes of action have
been classified into inverting and retaining on the base
of the anomeric configuration of donor substrate and
of the final products (Scheme 1B and 1C). The mechanism proposed for endo-glycosynthases, likewise that
of inverting exo-glycosynthases, employs a-oligosaccharidic fluorides as donors (Scheme 1D). Instead, retaining exo-glycosynthases synthesise oligosaccharides
by using activated donor-sugars with the same anomeric
configuration of the normal substrate (typically 2-nitrophenyl- (NP) or 2,4-dinitrophenyl-b-glucosides) in
the presence of external ions such as sodium formate
(Scheme 1C).[10] This external nucleophile mimics the
natural active site carboxylate promoting the formation
of a metastable glycosyl-formate intermediate, which
has been isolated in one case,[11] and the attack of an acceptor (Scheme 1C).
Adv. Synth. Catal. 2005, 347, 941 – 950
asc.wiley-vch.de
REVIEWS
A survey of the available glycosynthases is described
in Table 1 in which we make a comparative list of the
produced sugars and of the glycosidic bonds formed.
Details of the different enzymes and of the reaction conditions can be found in the papers describing these studies. The known glycosynthases include 12 enzymes from
7 different families (Table 1) and can produce a variety
of products at high yields, including mannose derivatives
that are particularly difficult to synthesize by following
the common chemical routes.[13] In addition, glycosynthases have been successfully used also for the solidphase synthesis of sugars and glycoconjugates.[14] Exoglycosynthase products include oligosaccharides ranging from di- to hexasaccharides. They can be new substrates for the study of glycosidases or can have several
applications; for instance, branched (1 – 3)-b- and (1 – 6)b-linked oligosaccharides, produced by the glycosynthase from Sulfolobus solfataricus, are the building
blocks of the b-1,3 – 1,6-glucans that are recognised as
elicitors of the defence response against pathogens in
plants.[15]
Endo-glycosynthases produce oligosaccharides and
polysaccharides depending on the reaction conditions.[9d,9h,16] One of these compounds, with a degree of
polymerization of up to 12 consists of alternating (1 –
3)-b- and (1 – 4)-b-glucosyl units, representing a novel
polysaccharide that forms platelets aggregating in
spherulites of diameters ranging from 5 to 50 mm.[16]
Thus, endo-glycosynthases appear to be valuable tools
for the production of novel biomaterials with specific
and controlled structures. Moreover, in the presence of
different polymers, these enzymes may direct the synthesis of composite materials with interesting biomedical applications.[17]
3 Thioglycoligases and Thioglycosynthases
A recent achievement in the rational design of improved
synthetic glycoside hydrolases is the preparation of thioglycoligases and thioglycosynthases (Scheme 2).[9l,18] In
thioglycoligases the acid/base catalyst of two retaining
b-glycosidases, namely the b-glucosidase from Agrobacterium and the b-mannosidase from Cellulomonas fimi,
was changed in the non-nucleophile residue alanine producing mutants whose activities were severely affected.[9l] In the presence of activated dinitrophenyl-b-glycosides the mutants were re-activated by the addition
to the reaction mixture of thio-sugar as acceptor. In
these compounds the SH group is more nucleophilic
than the OH group and does not require the assistance
of the base catalyst (Scheme 2A). The resulting disaccharides containing S-glycosidic linkages were obtained
with yields of 35 – 82%; consequently, the mutant enzymes were named thioglycoligases. These molecules
have considerable interest as stable glycoside analogues
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
943
Giuseppe Perugino et al.
REVIEWS
Table 1. List of available glycoside synthases.[a, b]
Organism
GH
Exo-glycosynthases
Agrobacterium
1
Sulfolobus solfataricus
1
Synthase
Products
Refs
b-glycosynthase
b-glycosynthase
b-glycosynthase
b-(1,3)-, b-(1,4)-di- and -trisaccharides
b-(1,4)-di-, tri- and -tetrasaccharides
b-(1,3)-, b-(1,4)- and b-(1,6)-2-NP-disaccharidesb-(1,3)and b-(1,6)- branched tri- and -tetrasaccharides
Glc-b-(1,3)-Glc-b-2-NP
Gal-b-(1,3)-Xyl-b-4P
b-(1,3)-, b-(1,4)- and b-(1,6)-disaccharidesb-(1,3)and b-(1,6)- branched tri- and -tetrasaccharides
b-(1,3)- and b-(1,4)-tri- to -hexasaccharides
b-(1,4)- and b-(1,6)-4-NP-disaccharides
Gal-b-(1,6)-Glc-b-4-NP
b-(1,3)-, b-(1,4)- and b-(1,6)-2-NP-disaccharidesb-(1,3)and b-(1,6)- branched tri- and tetrasaccharides
Glc-b-(1,3)-Glc-b-MU
Gal-b-(1,3)-Xyl-b-4P
b-(1,3)-2-NP-disaccharides
[6]
[9c]
[10a]
b-glycosynthase
b-glycosynthase
b-glycosynthase
Cellulomonas fimi
Schizosaccharomices pombe
Escherichia coli
Thermosphaera aggregans
Pyrococcus furiosus
2
31
2
1
1
Endo-glycosynthases
Bacillus licheniformis
16
Humicola insolens
Hordeum vulgare
7
17
Cellvibrio japonicum
Pyrococcus furiosus
Thioglycoligases
Agrobacterium
Cellulomonas fimi
[a]
[b]
b-mannosynthase
a-glucosynthase
b-galactosynthase
b-glycosynthase
b-glycosynthase
b-glycosynthase
b-glycosynthase
[9e]
[9f]
[9i]
[10b]
[12]
[12]
[10b]
26
16
b-glucansynthase
b-glucansynthase
b-glucansynthase
b-glucansynthase
b-glucansynthase
b-glucansynthase
b-mannansynthase
b-glucansynthase
Glc-b-(1,3)-Glc-b-(1,4)-Glc-b-MU
Glc-b-(1,4)-Glc-b-(1,3)-Glc-b-(1,4)-Glc-b-(1,4)-Glc-MU
Gal-b-(1,4)-Glc-b-(1,3)-Glc-b-(1,4)-Glc-b-(1,4)-Glc-MU
b-(1,4)-tri-, -tetrasaccharides
b-(1,3)-glucan
b-(1,3)-S-glucan
b-(1,4)-tri- to -heptasaccharides
Glc-b-(1,3)-Glc-b-(1,4)-Glc-b-MUGlc-b-(1,3)-Glc-b(1,3)-Glc-b-(1,3)-Glc-b-MU
[9b]
[9g]
[9g]
[9d]
[9h]
[9j]
[9k]
[9m]
1
2
b-thioglucoligase
b-thiomannoligase
Glc-b-S-(1,4)-Glc-b-4-NPGlc-b-S-(1,4)-Xyl-b-4-NP
Man-b-S-(1,4)-Glc-b-4-NPMan-b-S-(1,4)-Xyl-b-4-NP
[9l]
[9l]
Glycoside hydrolase family according to http://afmb.cnrs-mrs.fr/CAZY/index.htlm.
Abbreviations: 4P, 4-penten-1-yl-a-d-xylopyranoside; Gal, d-galactopyranoside; Glc, d-glucopyranoside;
Glc-b-(1,3)-Glc, d-glucosyl-b-(1,3)-d-glucopyranoside; Man, d-mannopyranoside; Xyl, d-xylopyranoside.
to be used as competitive inhibitors and ligands for the
formation of complexes for X-ray crystallography.[19]
More recently, Withers group demonstrated that a
thioglycoligase depleted also of the catalytic nucleophile is still catalytically competent.[18] The double mutant Glu171Ala/Glu358Gly of the b-glycosidase from
Agrobacterium incubated in the presence of a-glucosyl
fluoride and 4-nitrophenyl-4-thio-b-d-glucopyranoside
as donor and acceptor, respectively, produced the expected thio-linked 4-nitrophenyl disaccharide (Scheme
2B). Thus, the double mutant used in this way was
named thioglycosynthase. The authors reported that
one of the advantages of this new methodology over
the thioglycoligase described above is that the double
mutant is completely inactive (because of the absence
of the nucleophile residue) and does not hydrolyse the
donor. In addition, the thioglycosynthase uses glycosyl
fluoride donors that are more easily synthesised than dinitrophenyl glycosides.[18] However, the yields of the reaction did not increased significantly (~ 45 – 50%) and
944
[10a]
[12]
[10b]
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
the kcat was 100- to 400-fold lower than that measured
on the single mutants Glu171Ala (thioglycoligase) and
Glu358Gly (glycosynthase), respectively.[18] Presumably, the double mutant worked as a scaffold to position
the reagents and the reduced activity could be explained
by the lack of some unidentified chemical catalysis provided by the remaining carboxylate in the single mutants.
The catalytic competence of glycosidase mutants lacking the catalytic carboxylic groups was demonstrated
also by the group of Planas on the 1,3 – 1,4-b-glucanase
from B. licheniformis.[120] In this case, the activity of
the double mutant Glu134Ala/Glu138Ala was chemically rescued by sodium azide in the presence of the donor Glcb4Glcb3Glc-2,4DNP and it produced a-glycosyl
azide. No transglycosylation products were obtained in
this case.
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 941 – 950
Oligosaccharide Synthesis Promoted by Catalytically Engineered Glycosidases
REVIEWS
Scheme 1. Reaction mechanisms of glycosidases and glycosynthases. R ¼ R’ indicates the aglycone part of the sugars; X in the
schematic active site of the enzyme indicates the mutated nucleophile residue.
A crucial point for the exploitation of the synthetic potential of these new enzymes is the detailed knowledge
of their reaction mechanism; in this way their activity
can be enhanced and adapted to the synthesis of specific
products. These issues have been addressed recently in
different laboratories.
In inverting and retaining glycosynthases (Scheme 1B
and 1C) the enzymatic catalytic machinery consists of a
single carboxylic group that, in the parental glycosidas-
es, works as acid/base catalyst, and the products of the
reaction cannot be hydrolysed by the enzyme. However,
the two mechanisms differ substantially in the number
of steps of the reaction and in the catalytic function of
the carboxylic group in the active site. In inverting glycosynthases, only the de-glycosylation step is present, because the a-glycosyl-F mimics the covalent intermediate
observed in retaining glycosidases (Scheme 1B). Recently, Faijes and colleagues showed that in the inverting
endo-glycosynthase from B. licheniformis the carboxylic
group in the active site (Glu138) acts as a general base.[20]
This was experimentally demonstrated by analysing the
Adv. Synth. Catal. 2005, 347, 941 – 950
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
4 Molecular Bases of the Glycosynthase
Mechanism
asc.wiley-vch.de
945
Giuseppe Perugino et al.
REVIEWS
Scheme 2. Reaction mechanisms of thioglycoligase and thioglycosynthase. R ¼ R’ indicates the aglycone part of the sugars.
pH dependence of the glycosynthase activity, the kinetics of enzyme inactivation by a carboxyl-specific reagent, and the reactivation of a double mutant in which
both the nucleophile and the acid/base were changed in
non-nucleophilic residues (see also above). These studies revealed that, as expected for a general base catalysis,
the pKa of the Glu138, which is 7.0 in the wild-type enzyme (general acid catalysis), drop to 5.3 in the glycosynthase.
In retaining glycosynthases, likewise parental glycosidases (Scheme 1A), the reaction takes place through
two steps, a glycosylation and de-glycosylation, through
the formation of a metastable glycosyl-formate intermediate (Scheme 1C). In this case, we have recently
demonstrated that at acidic conditions the hyperthermophilic glycosynthases showed enhanced activity, possibly because the carboxylic group present in the active
site acts as a general acid catalyst.[10b] In particular, we
observed that the exo-glycosynthases from the hyperthermophilic archaea S. solfataricus, Pyrococcus furiosus, and Thermosphaera aggregans, showed an up to
17-fold increase in the reaction rate at pH 3.0 in
50 mM sodium formate if compared to the same enzymes assayed at pH 6.5 in 2 M sodium formate.[10b] Remarkably, the kcat values measured at acidic conditions
were similar to that of the corresponding wild-type enzymes assayed at optimal conditions and allowed the
synthesis of 4-methylumbelliferyl disaccharides.[12] We
proposed that this exceptional reactivation was the result of the restoration of the correct ionisation state of
the carboxylic group that in the active site acts as general
acid catalyst. In fact, removal of the catalytic nucleophile in retaining glycosidases causes a downward shift
in the pKa of the acid/base catalyst;[21] therefore, this
group is converted in the ionised form and it performs
946
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
the first step of the reaction less efficiently (Scheme 1C).
Conversely, in the presence of sodium formate buffer
pH 3.0 – 6.0 the Glu residue could adopt the protonated
form assisting the departure of the leaving group of the
donor.[10b] The high reaction rates observed suggest that
for retaining glycosynthases the first step of the reaction
(glycosylation of the enzyme) is rate limiting and that,
hence, they requires general acid catalysis for optimal
activity. It is worth noting that, at acidic conditions, we
could use less activated and commercially available donor substrates like 2-nitrophenyl and 4-nitrophenyl glucosides. This is a typical example of an application of hyperstable enzymes that can resist extreme operational
conditions like pH 3.0 and 65 8C.[10b,12]
The resolution of the 3D-structure of the endo-glycosynthase from Humicola insolens (Cel7B) cast new light
on the glycosynthase mechanism.[22] The enzyme shows
a long active-centre groove that can accommodate substrates of up to eight monosaccharides in which four subsites ( 2 to þ 2 by following the nomenclature reported
in [23]) significantly contribute to catalysis. The nucleophile of the reaction (Glu197) is located at the centre
of the active centre and in the glycosynthase variants
Glu197Ala/Ser no significant changes were observed
suggesting that the mutants maintain the same active
site structure of the wild type (Figure 1). The mutant
Glu197Ser, which is the most active glycosynthase variant, was crystallised with both lactose and cellobiose
and the resolution beyond 1.4 allowed detailed analysis of the interactions with the substrate. In the donor
subsites ( 2 and 1) cellobiose and lactose bind equally well as expected from the kinetic analysis of the enzyme (Figure 1).[22] More interestingly, in the acceptor
subsites ( þ 1 and þ 2) lactoside and cellobioside complexes are not equivalent since the former is slightly disasc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 941 – 950
Oligosaccharide Synthesis Promoted by Catalytically Engineered Glycosidases
placed by approximately 1.7 and binds in a sort of
þ 1.5 to þ 2.5 position (Figure 1). These explains why
lactosides are poor acceptors and are responsible for
substrate inhibition at high donor concentrations.[22]
One of the most interesting findings described in this
study is on the mode of action of the Glu197Ser mutant.
Ducros and colleagues found that the O4 of the acceptor
lies 4.4 from the C1 of the donor while the O3 group is
at only 3.3 ; this is quite surprising since the glycosynthase promotes the formation of b-1,4 bonds and products containing b-1,3 bonds have never been observed.[9d] Comparing the structure of the Cel7B H. insolens
glycosynthase with that of Fusarium oxysporum[24] the
authors suggest that the regioselectivity for 1,4 bond formation could be regulated not only by distance criteria,
but also by the angle of the nucleophilic attack. In fact,
the potential angle of attack of O3 is sub-optimal whereas the O4C1O1 angle is 1738, close to the optimal 1808
required for the nucleophilic attack (Figure 1).
5 Strategies for the Improvement of the
Glycosynthases Activity
The search for glycosynthases with improved activity is
of utmost importance not only because they act more
rapidly, but also because, as a direct consequence, they
can transfer a variety of donors to a much wider array
of acceptors on a useful timescale, making them considerably more versatile synthetic tools. For instance, we
showed that the hyperthermophilic retaining glycosynthases that had increased activity in acid conditions
(see above) widened their specificity towards galactoside and xyloside donors[10b] and accept a great deal of
acceptors.[12]
The nature of the amino acid substituting the nucleophile in glycosynthases has a significant effect on the activity of the enzyme. The non-nucleophilic residue has to
generate sufficient space to accommodate the axial
REVIEWS
anomeric fluoride of the donor in inverting glycosynthases and the a-formyl glucoside intermediate in retaining enzymes (Scheme 1B, 1C). In inverting glycosynthases, Gly[9f,9h,9k] and Ser mutations[9c,9e,9i] gave best
results. In particular, studies of directed evolution revealed that in the glycosynthase from Agrobacterium
the Gly, Ser, Ala, and Cys mutants have a ratio of rate
constants of 102 : 44 : 2 : 1, respectively.[25] It has been proposed that the increased glycosylation activity in the Ser
mutant originates from a stabilizing interaction between
the Ser hydroxy group and the departing fluorine.[9c]
More recently, the inspection of the 3D-structure of
the glycosynthase from H. insolens suggested that,
most probably, this occurs at the transition state.[22]
However, Gly mutants often resulted to be better synthases than the Ser mutants. In the glycosynthase from
C. japonicus this was tentatively explained with the observation that in the 3D-structure of the Gly mutant a
single water molecule replacing the nucleophile formed
a hydrogen bond to the O1 of the donor.[9k]
Similarly, Gly mutants were also found to be efficient
in retaining glycosynthases: in the enzyme from S. solfataricus the rate of hydrolysis of the 2,4-dinitrophenyl-bglucoside donor followed the ratio 59 : 9 : 1 for the Gly,
Ala, and Gln mutants.[9a]
Glycosynthases with improved activity have been recently obtained by directed evolution. This method consists in generating a large population of protein variants
that is screened for the desired function.[26] Recently,
Withers group reported that additional mutations
even increased the activity of this glycosynthase.[27] In
this case, mutations were introduced randomly in the sequence of the gene encoding for the Glu358Gly glycosynthase; then, by using an “on-plate” screening two
mutants were identified in the mutant library. This method is shown in Figure 2A: briefly, it consists in coupling
to the glycosynthase catalysed reaction an assay involving an endocellulase. The mutant library co-express
Cel5A, hence the glycosynthase activity can be moni-
Figure 1. Schematic view of the active site of H. insolens glycosynthase Cel7B. The binding sites and the ligands identified by
X-ray crystallography are shown as reported by Ducros and colleagues.[22] The sugar in the 2 to 1 sites indicates both lactose and cellobiose; lactose in the acceptor sites þ 1 to þ 2 is shown in grey.
Adv. Synth. Catal. 2005, 347, 941 – 950
asc.wiley-vch.de
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
947
Giuseppe Perugino et al.
REVIEWS
Figure 2. Method of selection of glycosynthase mutants with enhanced activity. (A) Scheme of the glycosynthase-cellulase coupled assay. The substrates of the glycosynthase are a-glucosyl F and 4-methylumbelliferyl-b-glucoside (MU-Glc) as donor and
acceptor, respectively. Mutant clones expressing an efficient glycosynthase promote the synthesis of MU-oligosaccharides that
are substrate of the b-1,4 glucanase from C. fimi (Cel5A) releasing the fluorescent methylumbelliferone; (B) yeast three-hybrid
assay exploited in the “chemical complementation” method (see text for the details of the two methods).
tored directly on a plate simply by adding the substrates
of the glycosynthase. With this straightforward technique the authors identified with two rounds of mutagenesis a multiple mutant (Ala19Thr/Glu358Gly/
Gln248Arg/Met407Val) showing a catalytic efficiency
27-fold higher than that of the parental glycosynthase
Glu358Gly. To understand the molecular reasons of
the improved activity of the new mutant the authors inspected a model generated using known structures of
homologous enzymes. From this analysis it was proposed that the mutations Ala19Thr and Met407Val improve the activity by contributing to the transition state
stabilisation and by increasing the fit in the active site of
the a-glucosyl fluoride substrate, respectively. Intriguingly, no straightforward interpretations could be proposed for Gln248Arg since the residue is located on
the protein surface. This study further confirms the usefulness of the directed evolution methods: the mutations
described could be hardly predicted by a rational design
approach.
Recently, a novel directed evolution approach, based
on the yeast three-hybrid system assay named “chemical
complementation”, was applied to the Cel7B glycosynthase from H. insolens [28] after application to other enzymes.[29] Briefly, the case described for the Cel7B glyco948
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
synthase utilises a three-hybrid assay based on two fusion proteins: the hormone binding domain of the glucocorticoid receptor (GR) connected to a transcription activation domain (AD), and a dihydrofolate reductase
(DHFR) fused to a DNA-binding domain (DBD) (Figure 2B). The association of AD to DBD allows the expression of the gene LEU2 producing the growth in
the absence of leucine. In addition, methotrexate
(Mtx) and dexamethasone (Dex) specifically bind to
the GR and DHFR proteins, respectively (Figure 2B).
Without glycosynthetic activity the fusion proteins
GR-AD and DHFR-DBD cannot dimerise, thus, the
gene LEU2 is not expressed and cannot complement
yeast cells that cannot grow. When the glycosynthase
catalyses in vivo the synthesis of the glycosidic bond between an Mtx-disaccharide-fluoride donor (Mtx-Lac-F)
and a Dex-disaccharide acceptor (Dex-Cel) the dimerisation of the fusion proteins occurs directing the expression of the LEU2 gene and the growth without leucine
(Figure 2B). The selection was used for the identification of Cel7B mutants with improved activity obtained
by directed evolution. In particular, saturation mutagenesis on the Glu197 site randomised to all 20 amino acids
and subsequent chemical complementation produced
three mutants Ala, Gly and Ser giving growth advantage
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 941 – 950
Oligosaccharide Synthesis Promoted by Catalytically Engineered Glycosidases
to yeast.[28] The most efficient evolved variant was
Glu197Ser in agreement with the results of Ducros
and colleagues described above.[22]
The mutation identified in the Cel7B glycosynthase
with the chemical complementation method was not
novel, however, it offers the advantage to analyse a
huge number of clones since it is based on a stringent selection (growth advantage vs. coupled assay). Practical
limits, which may affect the efficiency of the complementation, can be the design of acceptors and donors
suitable for the three-hybrid assay and the expression
in yeast of the glycosynthase of interest.
6 How to Predict the Gycosynthetic
Potential of a Glycosidase
From that stated in the previous paragraphs, the glycosynthase technology is now well established and it has
improved greatly in the last five years. Several groups
experimentally proved that the same methodology can
be applied to a great number of glycosidases and that
the activity of existing glycosynthases can be significantly improved by using chemistry[10b] or directed evolution.[27,28] Notwithstanding these issues, to obtain a new
glycosynthase is not trivial and several glycosidases mutated in the nucleophile did not show glycosynthase activity, i.e., the b-xylosidase from Thermosporum saccharolyticum, b-galactosidase from Bacillus circulans,[22] the
a-xylosidase (B. Di Lauro, B. C.-P. and M. M. unpublished results) and the a-l-fucosidase from S. solfataricus.[30]
Consequently, it is of utmost importance to try to understand what makes an efficient glycosynthase. This problem was addressed by Ducros and collegues by analysing
the effect on glycosidases of fluoro-glycosides mechanism-based inhibitors. These molecules, in which the hydroxy group at C2 is replaced with a fluorine atom, act as
covalent inhibitors of glycosides by trapping the intermediate 2-fluoroglycosyl enzyme.[31] The inhibited enzyme is often catalytically competent and the intermediate can be resolved by an acceptor glycoside or by water;
the rate constants of transfer and hydrolysis are ktrans and
kH2O, respectively.[22] Noticeably, with this method Blanchard and Withers identified a number of sugars able to
reactivate six different glycosidases and therefore acting
as efficient acceptors. [32]
Ducros and collegues proposed that glycosidases that
yield efficient glycosynthases have two reactivation
characteristics: first, a high rate constant of transfer to
an acceptor (they estimate ktrans > 10 2 min 1); second,
a better transferring activity to an acceptor rather than
to water (ktrans/kH2O > 20).[22] The rational observation
behind this prediction is that the 2-fluoro-glycosyl enzyme intermediate may resemble the a-glycosyl fluoride
bound to the inverting glycosynthases. By analysing the
kinetic behaviour of the parental glycosidases of existing
glycosynthases and of enzymes whose mutants did not
Adv. Synth. Catal. 2005, 347, 941 – 950
asc.wiley-vch.de
REVIEWS
show glycosynthase activity the authors found some support to this predictive method.
Interestingly, glycosidases which perform efficiently
transglycosylation reactions often did not act as glycosynthases upon mutation. This raises the question of
how the “natural” transglycosylating activity of a glycosidase is lost after mutation of the nucleophile. There is
no clear-cut answer to this question, however, the reactivity of the donors and the binding constants of the acceptors play an important role in both cases. Moreover,
as stated by Ducros and collegues, the acceptors can improve the glycosynthetic activity by stabilising the transition state and reducing the transfer activity to water
that is detrimental to the synthesis of the products.[22]
Therefore, the binding interactions of the acceptors in
the þ 1 site are important for the catalysis of glycosynthases.
These studies further confirm that a detailed kinetic
characterisation, including the use of mechanism-based
inhibitors, is essential prior to the preparation of a glycosynthase. The validity of the methods described needs to
be tested by analysing a larger number of glycosidases.
Acknowledgements
This work was supported by MIUR project “Folding di proteine:
laltra metà del codice genetico” RBAU015B47 006 and by the
project entitled “Nuove attività enzimatiche per la sintesi di oligosaccaridi di interesse biotecnologico” from Regione Campania L.R.N.5 del 28.03.2002. The IBP-CNR belongs to the Centro
Regionale di Competenza in Applicazioni Tecnologico-Industriali di Biomolecole e Biosistemi.
References
[1] P. Sears, C. H. Wong, Science. 2001, 291, 2344 – 2350.
[2] a) D. H Crout, G. Vic, Curr. Opin. Chem. Biol. 1998, 2,
98 – 111; b) C. R. Bertozzi, L. L. Kiessling, Science,
2001, 291, 2357 – 2364.
[3] B. Kuberan, M. Z. Lech, D. L. Beeler, Z. L. Wu, Nat. Biotechnol. 2003, 21, 1343 – 1346.
[4] a) D. E. Koshland Jr. Biol. Rev. 1953, 28, 416 – 436;
b) K. M. Koeller, C. H. Wong, Nat. Biotechnol. 2000, 8,
835 – 841.
[5] Y. Bourne, B. Henrissat, Curr. Opin. Struct. Biol. 2001,
11, 593 – 600.
[6] L. F. Mackenzie, Q. Wang, R. A. J. Warren, S. G. Withers,
J. Am. Chem. Soc. 1998, 120, 5583 – 5584.
[7] a) H. D. Ly, S. G. Withers, Annu. Rev. Biochem. 1999, 68,
487 – 522; b) G. Perugino, A. Trincone, M. Rossi, M. Moracci, Trends Biotechnol. 2004, 1, 31 – 37; c) M. Jahn, S. G.
Withers, Biocat. Biotrans. 2003, 21, 159 – 166; d) A. Planas, M. Faijes, Afinidad 2002, 59, 295 – 313; e) S. J. Williams, S. G. Withers, Austral. J. Chem. 2002, 55, 3 – 12;
f) D. L. Jakeman, S. G. Withers, Trends in Glycosci. Glycotech. 2002, 14, 13 – 25; g) G. J. Davies, S. J. Charnock, B.
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
949
Giuseppe Perugino et al.
REVIEWS
Henrissat, Trends in Glycosci. Glycotech. 2001, 13, 105 –
120; h) M. Moracci, A. Trincone, M. Rossi, J. Mol. Cat.
B: Enzymatic 2001, 11, 155 – 163; i) S. G. Withers, Can.
J. Chem. 1999, 77, 1 – 11.
[8] Q. Wang, R. W. Graham, D. Trimbur, R. A. J. Warren,
S. G. Withers, J. Am. Chem. Soc. 1994, 116, 11594 – 11595.
[9] a) M. Moracci, A. Trincone, G. Perugino, M. Ciaramella,
M. Rossi, Biochemistry 1998, 37, 17262 – 17270; b) C.
Malet, A. Planas, FEBS Lett. 1998, 440, 208 – 212; c) C.
Mayer, D. L. Zechel, S. P. Reid, R. A. J. Warren, S. G.
Withers, FEBS Lett. 2000, 466, 40 – 44; d) S. Fort, V. Boyer, L. Greffe, G. J. Davies, O. Moroz, L. Christiansen, M.
Schlein, S. Cottaz, H. Driguez, J. Am. Chem. Soc. 2000,
122, 5429 – 5437; e) O. Nashiru, D. L. Zechel, D. Stoll, T.
Mohammadzadeh, R. A. J. Warren, S. G. Withers, Angew. Chem. Int. Ed. 2001, 40, 417 – 420; f) M Okuyama,
H. Mori, K. Watanabe, A. Kimura, S. Chiba, Biosci. Biotechnol. Biochem. 2002, 66, 928 – 933; g) J. K. Fairweather, M. Faijes, H. Driguez, A. Planas, ChemBioChem
2002, 3, 866 – 873; h) M. Hrmova, T. Imai, S. J. Rutten,
J. K. Fairweather, L. Pelosi, V. Bulone, H. Rodriguez,
G. B. Fincher, J. Biol. Chem. 2002, 277, 30102 – 30111;
i) D. L. Jakeman, S. G. Withers, Can. J. Chem. 2002, 80,
866 – 870; j) J. K. Fairweather, M. Hrmova, S. J. Rutten,
G. B. Fincher, H. Driguez, Chemistry 2003, 9, 2603 –
2610; k) M. Jahn, D. Stoll, R. A. J. Warren, L. Szabò, P.
Singh, H. J. Gilbert, V. M. A. Ducros, G. J. Davies, S. G.
Withers, Chem. Commun. 2003, 12, 1327 – 1329; l) M.
Jahn, J. Marles, R. A. J. Warren, S. G. Withers, Angew.
Chem. Int. Ed. 2003, 42, 352 – 354; m) J. van Lieshout,
M. Faijes, J. Nieto, J. van der Oost, A. Planas, Archaea
2004, 1, 285 – 292.
[10] a) A. Trincone, G. Perugino, M. Rossi, M. Moracci, Bioorg. Med. Chem. Lett. 2000, 10, 365 – 368; b) G. Perugino,
A. Trincone, A. Giordano, J. van der Oost, T. Kaper, M.
Rossi, M. Moracci, Biochemistry 2003, 42, 8484 – 8493.
[11] J. L. Viladot, F. Canals, X. Batllori, A. Planas A. Biochem J. 2001, 355, 79 – 86.
[12] A. Trincone, A. Giordano, G. Perugino, M. Rossi, M.
Moracci, Bioorg. Med. Chem. Lett. 2003, 13, 4039 – 4042.
[13] a) J. Alais, S. David, Carbohydr. Res. 1990, 201, 69 – 77;
b) W. Gunther, H. Kunz, Carbohydr. Res. 1992, 228,
217 – 241.
[14] J. F. Tolborg, L. Petersen, K. J. Jensen, C. Mayer, D. L.
Jakeman, R. A. J. Warren, S. G. Withers, J. Org. Chem.
2002, 67, 4143 – 9.
[15] a) K. Skriver, F. L. Olsen, J. C. Rogers, J. Mundy, Proc.
Natl. Acad. Sci. USA 1991, 88, 7266 – 7270; b) B. Duvic,
K. Soderhall, K. Eur. J. Biochem. 1992, 207, 223 – 228;
950
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
c) F. Coté, M. G. Hahn, M. G. Plant Mol. Biol. 1994,
26, 1379 – 1411.
[16] M. Faijes, T. Imai, V. Bulone, A. Planas A. Biochem. J.
2004, 380, 635 – 641.
[17] A. G. Watts, S. G. Withers, Biochem. J. 2004, 380, 9 – 10.
[18] M. Jahn, H. Chen, J. Mullegger, J. Marles, R. A. J. Warren, S. G. Withers, Chem. Commun. 2004, 3, 274 – 275.
[19] a) M. Czjzek, M. Cicek, V. Zamboni, W. P. Burmeister,
D. R. Bevan, B. Henrissat, A. Esen, Biochem. J. 2001,
354, 37 – 46; b) M. Hrmova, J. N. Varghese, R. De Gori,
B. J. Smith, H. Driguez, G. B. Fincher, Structure 2001,
9, 1005 – 1016.
[20] M. Faijes, X. Perez, O. Perez, A. Planas, Biochemistry
2003, 42, 13304 – 13318.
[21] L. P. McIntosh, G. Hand, P. E. Johnson, M. D. Joshi, M.
Korner, L. A. Plesniak, L. Ziser, W. W. Wakarchuk,
S. G. Withers, Biochemistry 1996, 35, 9958 – 9966.
[22] V. M. Ducros, C. A. Tarling, D. L. Zechel, A. M. Brzozowski, T. P. Frandsen, I. von Ossowski, M. Schulein,
S. G. Withers, G. J. Davies, Chem. Biol. 2003, 10, 619 –
628.
[23] G. J. Davies, K. S. Wilson, B. Henrissat, Biochem. J. 1997,
321, 557 – 559.
[24] G. Sulzenbacher, H. Driguez, B. Henrissat, M. Schulein,
G. J. Davies, Biochemistry 1996, 35, 15280 – 15287.
[25] C. Mayer, D. L. Jakeman, M. Mah, G. Karjala, L. Gal,
R. A. Warren, S. G. Withers, Chem. Biol. 2001, 8, 437 –
443.
[26] a) E. T. Farinas, T. Bulter, F. H. Arnold, Curr. Opin. Biotechnol. 2001, 12, 545 – 551; b) J. A. Kolkman, W. P.
Stemmer, Nat. Biotechnol. 2001, 19, 423 – 428; c) H.
Tao, V. W. Cornish, Curr. Opin. Chem. Biol. 2002, 6,
858 – 864.
[27] Y. W. Kim, S. S. Lee, R. A. Warren, S. G. Withers, J. Biol.
Chem. 2004, 279, 42787 – 42793.
[28] H. Lin, H. Tao, V. W. Cornish, J. Am. Chem. Soc. 2004,
126, 15051 – 15059.
[29] a) K. Baker, C. Bleczinski, H. Lin, G. Salazar-Jimenez,
D. Sengupta, S. Krane, V. W. Cornish, Proc. Natl. Acad.
Sci. USA 2002, 99, 16537 – 16542. b) D. Sengupta, H.
Lin, S. D. Goldberg, J. J. Mahal, V. W. Cornish, Biochemistry 2004, 43, 3570 – 3581.
[30] B. Cobucci-Ponzano, A. Trincone, A. Giordano, M. Rossi, M. Moracci, Biochemistry 2003, 42, 9525 – 9531.
[31] S. J. Williams, S. G. Withers, Carbohydr Res. 2000, 327,
27 – 46.
[32] J. E. Blanchard, S. G. Withers, Chem. Biol. 2001, 8, 627 –
622.
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 941 – 950