De novo design of a trans-β-N-acetylglucosaminidase activity from a GH1 β-glycosidase by mechanism engineering

Article abstract of DOI:10.1093/glycob/cwu121

Glycoside hydrolases are particularly abundant in all areas of metabolism as they are involved in the degradation of natural polysaccharides and glycoconjugates. These enzymes are classified into 133 families (CAZy server, http://www.cazy.org) in which members of each family have a similar structure and catalytic mechanism. In order to understand better the structure/function relationships of these enzymes and their evolution and to develop new robust evolved glycosidases, we undertook to convert a Family 1 thermostable β-glycosidase into an exo-β-N-acetylglucosaminidase. This latter activity is totally absent in Family 1, while natural β-hexosaminidases belong to CAZy Families 3, 20 and 84. Using molecular modeling, we first showed that the docking of N-acetyl-d-glucosamine in the subsite -1 of the β-glycosidase from Thermus thermophilus (TtβGly) suggested several steric conflicts with active site amino-acids (N163, E338) induced by the N-acetyl group. Both N163A and N163D-E338G mutations induced significant N-acetylglucosaminidase activity in TtβGly. The double mutant N163D-E338G was also active on the bicyclic oxazoline substrate, suggesting that this mutated enzyme uses a catalytic mechanism involving a substrate-assisted catalysis with a noncovalent oxazoline intermediate, similar to the N-acetylglucosaminidases from Families 20 and 84. Furthermore, a very efficient trans-N-acetylglucosaminidase activity was observed when the double mutant was incubated in the presence of NAG-oxazoline as a donor and N-methyl-O-benzyl-N-(β-d-glucopyranosyl)-hydroxylamine as an acceptor. More generally, this work demonstrates that it is possible to exchange the specificities and catalytic mechanisms with minimal changes between phylogenetically distant protein structures.

Full text of DOI:10.1093/glycob/cwu121

Glycobiology, 2015, vol. 25, no. 4, 394–402
doi: 10.1093/glycob/cwu121
Advance Access Publication Date: 4 November 2014
Original Article
Original Article
De novo design of a trans-β-N-
acetylglucosaminidase activity from a GH1
β-glycosidase by mechanism engineering
Corinne André-Miral, Fankroma MT Koné2, Claude Solleux,
Cyrille Grandjean, Michel Dion, Vinh Tran, and Charles Tellier1
Faculté des Sciences et des Techniques, Unité de Fonctionnalité et Ingénierie des Protéines (UFIP), Université de
Nantes, UMR CNRS 6286, 2 rue de la Houssinière, BP 92208, Cedex 3 F-44322 Nantes, France
1To whom correspondence should be addressed: Tel: +33-2-51-12-55-33; Fax: +33-2-51-12-22; e-mail: charles.tellier@univ-
nantes.fr
2Present address: Université Nangui Abrogoua, 22 BP1297 Abidjan 22 (Côte d’Ivoire).
Received 24 September 2014; Revised 30 October 2014; Accepted 30 October 2014
Abstract
Glycoside hydrolases are particularly abundant in all areas of metabolism as they are involved in the
degradation of natural polysaccharides and glycoconjugates. These enzymes are classified into 133
families (CAZy server, http://www.cazy.org) in which members of each family have a similar structure
and catalytic mechanism. In order to understand better the structure/function relationships of these
enzymes and their evolution and to develop new robust evolved glycosidases, we undertook to con-
vert a Family 1 thermostable β-glycosidase into an exo-β-N-acetylglucosaminidase. This latter activ-
ity is totally absent in Family 1, while natural β-hexosaminidases belong to CAZy Families 3, 20 and
84. Using molecular modeling, we first showed that the docking of N-acetyl-D-glucosamine in the
subsite −1 of the β-glycosidase from Thermus thermophilus (TtβGly) suggested several steric con-
flicts with active site amino-acids (N163, E338) induced by the N-acetyl group. Both N163A and
N163D-E338G mutations induced significant N-acetylglucosaminidase activity in TtβGly. The double
mutant N163D-E338G was also active on the bicyclic oxazoline substrate, suggesting that this mu-
tated enzyme uses a catalytic mechanism involving a substrate-assisted catalysis with a noncovalent
oxazoline intermediate, similar to the N-acetylglucosaminidases from Families 20 and 84. Further-
more, a very efficient trans-N-acetylglucosaminidase activity was observed when the double mutant
was incubated in the presence of NAG-oxazoline as a donor and N-methyl-O-benzyl-N-(β-D-glucopyr-
anosyl)-hydroxylamine as an acceptor. More generally, this work demonstrates that it is possible to
exchange the specificities and catalytic mechanisms with minimal changes between phylogenetic-
ally distant protein structures.
Key words: β-glycosidase, mechanism engineering, N-acetylglucosaminidase, substrate-assisted catalysis, transglycosidase
Introduction
N-acetyl-hexosamines are found in many natural glycoconjugates
such as in oligosaccharides present as glycoproteins or glycolipids at
the cell surface (Rich and Withers 2009), in bacterial and fungal cell
walls as part of the peptidoglycan (Cheng et al. 2000; Vollmer 2008)
or chitin polymers (Eijsink et al. 2010), in blood group antigens (Liu
et al. 2007) and in human milk oligosaccharides (Bode and
Jantscher-Krenn 2012). Protein O-GlcNAcylation is also an essential
reversible post-translational modification in higher eukaryotes, which
© The Author 2014. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
394

Engineering trans-N-acetylglucosaminidase from a glucosidase
395
Scheme 1. The two different catalytic mechanisms for β-N-acetylhexoaminidases. (A) double displacement mechanism used by GH Family
(B) substrate-assisted mechanism of GH Families 20 and 84.
3
and
is involved in numerous cellular processes such as transcription, cell
cycle progression and signal transduction (Hart et al. 2011). Trun-
cated O-glycan chains that expose the terminal GlcNAc residue, iden-
tified as Tk antigen, have been identified as carcinoma-associated
antigens and as potential targets for immunotherapy (Meichenin
et al. 2000). As a result, the biotechnological potential of these mole-
cules is large but currently hampered by the difficulty of their chemical
synthesis. One of the enzymatic alternatives consists in using the syn-
thetic properties of glycosidases. These usually catalyze the hydrolysis
of glycoside bonds but can also synthesize glycosides, especially
through the transglycosylation reaction. Some of them naturally dis-
play interesting transfer properties but yields in oligosaccharides
remain generally moderate (Ogata et al. 2007; Slàmovà et al. 2010).
This approach has raised more interest by the development of ro-
bust evolved glycosidases that are far more efficient than native ones.
We recently applied directed evolution to the β-glycosidase of Ther-
mus thermophilus (TtβGly; CAZy GH1 family) (Feng et al. 2005;
Koné et al. 2009) and the α-L-fucosidase of Thermotoga maritima
(GH29) (Osanjo et al. 2007). The transglycosylation activities of
these glycosidases were considerably improved, with yields ranging
from 60 to 80% depending on the acceptors. Due to their broad
substrate specificity, the mutant transglycosidases from TtβGly can effi-
ciently transfer either fucosyl, glucosyl or galactosyl units to the nonre-
ductive end of oligosaccharides. However, TtβGly, like all the enzymes
belonging to Family 1, does not contain β-N-acetylglucosaminidase
activity or the corresponding transglycosylation activity, which prevents
its use in the grafting of O-GlcNAc residues. Natural β-hexoaminidases
belong to CAZy families GH3, GH20 and GH84, which differ in
their catalytic mechanisms (Vocadlo et al. 2000; Vocadlo and Withers,
2005). β-N-acetylglucosaminidases from Family 3 use a catalytic
mechanism similar to β-glycosidases from Family 1, which involves
the formation and breakdown of a covalent α-glycosyl enzyme inter-
mediate formed on an aspartate residue (Scheme 1A) (Slàmovà et al.
2010). On the other hand, β-N-acetylglucosaminidases from Families
20 and 84 utilize a substrate-assisted mechanism involving the trans-
ient formation of an enzyme-sequestered oxazoline or oxazolinium
ion intermediate (Scheme 1B) (Mark et al. 2001;Macauley et al.
2005). However, both mechanisms retain the configuration at the
anomeric center.
The aim of this work was to test the possibility of rationally design-
ing an exo-β-N-acetylglucosaminidase activity from TtβGly glycosi-
dase from Family 1 in order to understand how the active site
residues control the mechanism and the specificity of the reaction
and also possibly to use transglycosidase mutants for the synthesis
of O-GlcNAc conjugates. The outcome of this work demonstrates
that it is possible to evolve a β-glycosidase from Family 1 into func-
tions that are naturally found only in Families 3 and 20, and that
the choice of the mutations can dramatically affect the mechanism
of hydrolysis. These results are also exploited to design new
trans-N-acetylglucosaminidases.
Materials and methods
Strains and plasmids
Ampicillin-resistant (100 µg/mL) Escherichia. coli were selected on
Luria-Bertani (LB) agar plates. Expression of the TtβGly gene was per-
formed from the pBTac2 vector in the strain DH5α/XL1 of E. coli. The
expression plasmid containing the wild-type (WT) TtβGly gene
(1.3 kb), under the control of the Ptac promoter and between the
restriction sites EcoRI and PstI, was termed pBBGly.
Site-directed mutagenesis
β-glycosidase mutants were obtained by polymerase chain reaction
(PCR) with overlapping extension. Primers D (5′-CAATTAATCAT
CGGCTCG) and F (5′-AATCTTCTCTCATCCGCC) flank the gene be-
fore the Ptac promoter and after the PstI restriction site. Twenty pmol of
D or F primer was mixed with 20 pmol of mutagenic primers, 10 ng of
plasmid pBBGly (with or without mutations N163A, N163D or
E338G) in a 50-μl PCR. The reaction conditions were: 1X Dynazyme
buffer, 0.2 mmol/L of each dNTP and 1.5 mmol/L MgCl2 and 2.5 U
of polymerase Dynazyme Ext (Finnzymes). The sequences of mutagenic
primers were the following: N163A1 (5′-TCGCCACCCTGGCCGA
GCCCTGGTG) and N163A2 (5′- CACCAGGGCTCGGCCAGGG

396
C André-Miral et al.
TGGCGA) for the creation of the N163A mutation. The reaction was
thermocycled as follows: One cycle at 96°C, 5 min, then 30 cycles at 96°
C, 30; 55°C, 30 s; 72°C, 5 min. Mutagenized PCR products were di-
gested by the EcoRI and PstI restriction enzymes and cloned back
into the pBTac2 vector digested by the same enzymes. The 1.3 kb
DNA fragments encoding for mutant TtβGly enzymes were sequenced
in both forward and reverse directions.
in only nine structures in complex with enzymes. Due to the very rigid
bicyclic skeleton, any Cartesian coordinates can be used for in silico
substrate replacement on TtβGly.
For the sake of topological comparisons between the modified
catalytic context of 1UG6 (GH1) and that of other β-glucosaminidases
having O-GlcNAcase activity, two structures were selected belonging
to families GH84 and GH20. For GH84, only one complex with NGT
ligand was solved (PDB code 2CHN) involving a bacterial enzyme
(Dennis et al. 2006). Among several enzymes of the GH20 family,
that of Streptococcus gordonii (PDB code 2EPN) was selected due
to its best resolution (1.61 Å) (Langley et al. 2008).
For tridimensional (3D) structure superimpositions, the topologic-
al comparison between structures belonging to different glycoside
hydrolases (GH) families is difficult when focusing on a very specific
region (here, the available room delimiting the [−1] subsite). Further-
more, in the present case, although they share the same (β/α)8 folding,
they do not belong to the same clans according to CAZy classification
(GH-A, GH-K and alone for GH1, GH20 and GH84, respectively). As
suspected, preliminary superimpositions performed on inner β-sheets
gave acceptable overall shape comparisons but incorrect 3D analyses
at the subsite level. Finally, taking advantage of the inherent rigidity of
the NGT inhibitor (verified in all X-ray structures where this molecule
is present), the enzyme superimpositions were simply performed by
pair-fitting of heavy atoms of the bicyclic structure, concerning experi-
mental structures (2CHN, 2EPN) and a modeled one (1 double mu-
tant of 1UG6) as well. This pair-fitting protocol is an option of the
PyMOL graphic software used for molecular representation (The
PyMOL Molecular Graphics System, Version 1.5.0.4 Schrödinger,
LLC.).
Random mutagenesis
Random mutations were introduced by error-prone PCR (EPP) (Leung
et al. 1989). 20 pmol of primers D and F was mixed with 10 ng of plas-
mid in a 50-μL PCR. The reaction conditions were as follows: 1X Dy-
nazyme buffer, 10 mmol dATP and dGTP, 50 mmol dTTP and dCTP,
350 µmol MgCl2, 5 mmol MnCl2 and 1 unit of polymerase Dynazyme
Ext (Finnzymes). The reaction was thermocycled as follows: One cycle
at 96°C, 5 min, then 30 cycles at 96°C, 30; 50°C, 30 s; 72°C, 2 min
30 s; 70°C, 5 min. Mutagenized PCR products were digested by
EcoR1 and HindIII restriction enzymes and cloned back into the
pBTac2 vector. The plasmids EPP pECTorAN163A and EPP pECTorA-
TM and express mutant in
N163A,E338G were used to transform E. coli DH5αF’IQ
the periplasmic space using the Twin Arginine translocation (TAT) se-
cretion system (Silvestro et al. 1989). Transformed cells (≈6000) were
plated on a nitrocellulose membrane (0.45 µm, 21 × 21 cm2) lying on
LB agar supplemented with 100 µg/mL ampicillin and were grown at
37°C overnight. The nitrocellulose membrane bearing ∼6000 colonies
was transferred onto a new LB plate containing 5-bromo-4-chloro-
3-indolyl β-D-N-acetylglucosamine (X-GlcNAc, 0.5 mM) and isopro-
pylthiogalactoside (0.05 mM) and incubated at 37°C until colored
colonies appeared.
NAG-oxazoline synthesis
Purification of WT and mutant enzymes
A mixture of N-acetyl-D-glucosamine (1 g, 4.52 mmol), Ac2O (9 mL)
and pyridine (4.5 mL) was stirred at r.t. for 22 h. After diluting with
CH2Cl2 (20 mL), the resulting soln. was washed successively with sat.
aq. Na2CO3 soln. (2 × 10 mL), H2O, H2SO4 (3 M soln.), H2O, then
the organic layer was dried (MgSO4) and evaporated. Crude glucosa-
mine peracetate was used for the following experiment without
further purification. To a soln. of glucosamine peracetate (920 mg,
2.35 mmol) in 1,2-dichloroethane (30 mL), trimethylsilyl trifluo-
romethanesulfonate (0.5 mL, 2.76 mmol) was added dropwise, and
the mixture was heated at 50° for 5 h. (CH3)3N (1 mL) was added
and the mixture stirred for 10 min. After diluting with CH2Cl2
(30 mL), the mixture was washed with cold H2O, the organic
layer dried (MgSO4) and evaporated, and the residue submitted to col-
umn chromatography (Merck silica gel 60 (0.0040–0.0063 mm), 1%
Et3N in ethyl acetate/petroleum ether 9/1) to obtain 763 mg of
2-methyl-(1,2-dideoxy-3,4,6-tri-O-acetyl-D-glucopyrano)
Recombinant strains expressing the TtβGly mutant genes were grown
in 1 L of LB medium at 37°C overnight, centrifuged and resuspended
in 20 mL of lysis buffer (50 mM phosphate, 0.3 M NaCl, 10 mM
imidazole, pH 8). After sonication, extracts were heated for 20 min
at 70°C to precipitate most of the host proteins and centrifuged.
Then the 6-His-tagged proteins were purified by immobilized ion
metal-affinity chromatography: 2 L of Ni-NTA Superflow (Qiagen
France, F-91974 Courtaboeuf, France) was added to the supernatant
and stirred for 1 h at 4°C, then loaded onto a 10-mL column. The col-
umn was washed using 25 mL of washing buffer (50 mM phosphate,
0.3 M NaCl, 25 mM imidazole, pH 8), then 5 mL of elution buffer
was added (50 mM phosphate, 0.3 M NaCl, 250 mM imidazole,
pH 8). The purity of the protein (85–95%) was checked by sodium
dodecyl sulfate-polyacrylamide gel electrophoresis and protein chip
(Agilent).
(2,1-d)-oxazoline as a white solid (yield 75%). To a soln. of this oxazole
derivative (160 mg, 0.49 mmol) in CH3OH (5 mL) was added a solu-
tion of 5.3 M NaOCH3 in CH3OH (9 µL) at 0°, and the mixture was
stirred for 1 h. After the reaction had finished (1 h), the solvent was eva-
porated (quant). The white solid obtained was dissolved in borax buffer
(50 mM, pH 9.3) and stored as aliquots of 25 μL at −80°. 1H-nuclear
magnetic resonance (NMR) (D2O, 400 MHz) δH 2.05 (d, 3H, J = 0.87
Hz, CH3); 3.38 (ddd, 1H, J = 10.02, 4.64 and 2.86 Hz, H5); 3.60 (dd,
1H, J = 10.02 and 6.00 Hz, H4), 3.67 (dd, 1H, J = 12.00 and 4.64 Hz,
Molecular modeling
For the enzyme, the highly resolved structure of the native TtβGly
[PDB (Protein Data Bank, http://www.rcsb.org) code 1UG6] was
used. Due to the absence of any sugar substrate in its catalytic site,
the structure of the covalent intermediate between the β-glucosidase
of Bacillus polymyxa and β-D-glucopyranosyl fluoride (PDB code
1E4I) was used as a starting template to locate any pyranose ring in
the buried subsite [−1] because of the high similarity of these two en-
zymes at the catalytic core. This in silico replacement protocol, already
described in a previous paper (Feng et al. 2005), was used here to lo-
cate NAG-thiazoline (PDB ligand code NGT), a perfect structural ana-
log of NAG-oxazoline. As inhibitor alone, NAG-thiazoline was found
H6b); 3.81 (dd, 1H, J = 6.00 and 5.1 Hz, H3); 3.98 (dd, 1H, J = 12.00
and 2.86 Hz, H6a); 4.13 (m, 1H, H2); 6.08 (d, 1H, J = 7.6 Hz, H1);
13C-NMR (D2O, 100 MHz) δC 14.01 (CH3), 63.34 (C6), 65.01 (C5),
67.57 (C2), 68.41 (C3), 70.42 (C4), 99.41 (C1), 169.70 (CN).

Engineering trans-N-acetylglucosaminidase from a glucosidase
397
temperature of 150°C. The instrument was calibrated using MS/MS
fragments of a commercial xyloglucan infused at 5 µg/mL.
Kinetic studies
All kinetic studies were performed with a (Tecan France S.A.S.U.,
Lyon, France) or Labsystem microplate reader at 40°C in phosphate
buffer saline 1×. Initial reaction rates were calculated from the slope
of the zero-order plot of product concentration (pNP) against reaction
time, then the initial rates of reaction were plotted against substrate
concentration and the resulting curves were fitted to the hyperbolic
equation v = Vmax×[S]/(Km+[S]) using Origin 7.0 (OriginLab) to ob-
tain the kcat and Km parameters. When initial rates did not exhibit sat-
uration behavior at high substrate concentration, only kcat/Km
After purification, the product (N-methyl-O-benzyl-N-(β-2-D-
N-acetylglucopyranosyl(1→3)-β-D-glucopyranosyl)hydroxylamine)
was freeze-dried and analyzed by NMR. 1H-NMR (MeOD,
400 MHz) δH 1.92 (s, 3H, CH3); 2.65 (s, 3H, 3H7); 3.12–3.30
(m, 4H, H4, H5, H4′, H5′); 3.36–3.47 (m, 3H, H3′, H2, H3) 3.54–
3.61 (m, 3H, H2′, 2H6); 3.77 (dd, 1H, J = 12.08 and 2.16 Hz H6a’);
3.79 (dd, 1H, J = 12.04 and 1.96 Hz H6b’); 3.97 (d, 1H, J = 8.7 Hz,
H1); 4.64 (d, 1H, J = 8.4 Hz, H1′); 4.70 (m, 2H, 2 H8); 7.19–7.29
(m, 5H, 5 Har). 13C-NMR (MeOD, 100 MHz) δC 23.16 (CH3);
39.19 (C7); 57.98 (C6); 62.68 (C6′); 62.77 (C2′); 69.73 (C5); 71.48
(C2); 72.14 (C5′); 76.09 (C4); 76.61 (C8); 78.11 (C3′); 79.28 (C4′);
88.26 (C3); 95.42 (C1); 103.41 (C1′); 129.30, 129.48, 130.31
(5Car); 138.17 (C9); 174.43 (CN). High resolution-ESI-MS m/z
Calcd for C22H34N2O11 [M + H]+ 525.2060, found 525.2052.
(referred also as the catalytic efficiency) were determined. The kcat
/
Km parameter was evaluated at low substrate concentrations, when re-
action rate can be approximate to the equation: v = kcat/KM× [E] × [S].
kcat/KM is then an apparent second order kinetic constant, which is de-
termined by the initial slope of the curve V/[E] = f([S]).
Capillary electrophoresis measurements
Kinetics of NAG-oxazoline hydrolysis and transglycosylation activ-
ities with NAG-oxazoline as donor and N-methyl-O-benzyl-N-
(β-D-glucopyranosyl)-hydroxylamine as acceptor were determined
by capillary electrophoresis (Beckman P/ACE MDQ, Beckman Coul-
ter, France with an uncoated fused silica capillary, 50 cm in length and
75 μm in diameter) (Teze et al. 2013). Typical experimental conditions
were: 100 μL of medium containing 10 mM of 4-acetamidophenol
(used as an internal standard), 30 mM NAG-oxazoline, 16 mM
BnON(Me)-Glc and 100 μg of enzyme, which were incubated at
40°C in 50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid
(HEPES) buffer at pH 7.8 in the capillary electrophoresis apparatus
and analyzed every 20 min for 5 h. Separations were performed at
30 kV with 20 mM Borax pH 9.3 containing 50 mM sodium dodecyl
sulfate as running buffer. Donors, acceptors and products were
detected by ultra-violet (UV) absorbance at 200 nm and quantified
by comparison with 4-acetamidophenol as internal standard.
Results
Molecular modeling of pNPGlcNAc and NAG-oxazoline
in the −1 site of ttβGly
The substrate specificity of TtβGly is quite broad since this enzyme
is able to hydrolyze various β-glycosides such as β-glucosides,
β-fucosides and β-galactosides (Dion et al. 1999). However, it cannot
hydrolyze β-N-acetyl-glycosides such as pNPGlcNAc. This property
seems to be shared by all glycoside hydrolases of family GH1
(CAZy) (Lombard et al. 2014). Natural exo-N-acetylhexoaminidases
are found rather in GH Families 3, 20 and 84, which have the same
overall fold (β/α)8 as the glycoside hydrolases from Family 1, but be-
long to different clans. The structural feature that prevents GH1 en-
zymes from hydrolyzing N-acetylglucosamine conjugates is basically
the inappropriate room available in the subsite [−1] as revealed by
the in silico NAG-thiazoline and β-GlcNAc replacement (docking
protocol described in the “Materials and methods” section). In the
first case, docking revealed a strong steric conflict with Glu338 and,
to a much lesser extent, with Asn282 and Asn163 within the active
site (Figure 1A). In the second case, the most severe steric conflict
was due to Asn163 (Figure 1B).
The N-acetylhexoaminidases from Families 20 and 84 carry out
hydrolysis using a substrate-assisted catalysis, which goes through a
NAG-oxazolinium ion intermediate (Williams et al. 2002; Macauley
et al. 2005). As a result, NAG-oxazoline is a good mimic of this inter-
mediate. Docking the TtβGly active site with NAG-oxazoline showed
that binding within the −1 site was only possible with the E338G mu-
tant and not with the N163A mutant. On the other hand, the N163A
mutation allowed β-GlcNAc docking. Such steric hindrance may ex-
plain why N-acetylglucosides are not substrates for this enzyme.
Furthermore, multiple sequence alignment within Family 20 and
84 hexoaminidases has revealed that the acid/base catalytic residue
(E in Family 20 and D in Family 84) is always preceded in the sequence
by an aspartate, which is involved in hydrogen bonding of the NAG-
oxazolinium intermediate (Williams et al. 2002; Çetinbas et al. 2006).
These Asp-Glu or Asp-Asp pairs provide sequence signatures
of substrate-assisted catalysis within Families 20 and 84. In Family
1 glycoside hydrolases, the residue that precedes the acid–base catalyt-
ic residue is an Asn that, interestingly, corresponds to N163 in TtβGly.
It was thus tempting to postulate that N163D mutation could restore
an active site in the E338G mutant similar to that of GH Family 20.
Topological comparisons (Figure 2) between enzymes of families
GH1, GH20 and GH84 with modeled or experimentally observed
Transglycosylation product analysis
After the enzymatic reaction had finished, the crude product was puri-
fied by reverse phase high-pressure liquid chromatography on a C18
Nucleosil VP250/10 (Macherey-Nagel, France) (300 Å, 5 μm, 10 ×
250 mm) column, at a flow rate of 2.5 mL min−1, with Evaporating
Light Scattering Detector (Sedex 75, Sedere, France) and UV
(215 nm) detection. Gradient: 0% B for 5 min, 0–80% B over
25 min or 0% B for 5 min, 0–50% B over 20 min, 50% B for
10 min, respectively. Solvent system A: H2O; solvent system B:
meOH. Analysis of the 1H and 13C-NMR resonances and subsequent
structure assignments were carried out using standard 2D sequences
(Correlated Spectroscopy and Heteronuclear Multiple-Quantum Cor-
relation). The spectra were recorded with a Bruker AX400 spectrom-
eter operating at 400 MHz for 1H and at 100.6 MHz for 13C. Due to
their solubility, the spectra of the saccharides were recorded in D2O or
MeOD and the chemical shifts (in ppm) were quoted from the reson-
ance of methyl 3-(trimethylsilyl)-propansulfonate (DSS), which was
used as an internal reference. Mass measurements were performed
on an electrospray ionization (ESI)-Q-time-of-flight (TOF) using a
quadripole-TOF hybrid mass spectrometer (Micromass Q-TOF Ul-
tima Global mass spectrometer, Micromass Ltd., Manchester, UK).
Samples were diluted in MeOH to obtain a satisfactory intensity with-
out signal saturation. Samples were introduced at 5 µL/min and one
minute of acquisition was summed to produce the presented spectra.
Nitrogen was used as sheath gas (150 L/h). The MS analyses were car-
ried out using a typical needle voltage of 3 kV and a heated capillary

398
C André-Miral et al.
Fig. 1. In silico location of NAG-thiazoline (A) and β-GlcNAc (B) in the −1 subsite of TtβGly. Only enzyme residues (without hydrogen atoms) are shown as spheres.
‘Cartoon’ representations supporting these residues are in gray (PyMOL software). Regions of intolerable steric conflicts are circled in black.
Fig. 2. Topological comparisons between modeled (1UG6 mutant) or experimental complexes (2EPN and 2CHN) involving NAG-thiazoline. White opaque surfaces
represent the available room at the bottom of the catalytic site. The inhibitor (NGT) is shown as sticks. Enzyme residues stabilizing (or in contact) with the inhibitor
are shown as balls and sticks. Inner beta sheets of (β/α)8 folding are shown in ‘cartoon’ representation. Those from where residues run on are colored yellow.
NAG-thiazoline (E338G mutant of 1UG6, 2EPN and 2CHN, respect-
ively) gave relevant similarities in terms of [−1] subsite cavity shapes
the −1 subsite between GH1, GH20 and GH84 families
and functionally equivalent residues, as summarized in Table I. Re-
Table I. 3D superimposition of functionally equivalent residues at
GH1 (1UG6) GH20 (2EPN) GH84 (2CHN)
markably, these residues do not necessarily have the same topological
locations, which explains why in this case, the local superimposition
is much more appropriate than the overall one as would suggest
any sequence alignment. For instance, E392 (1UG6), D376 (2EPN)
and D344 (2CHN) (Table I) run on from beta sheets 8, 8 and 7 (for
GH1, GH20 and GH84, respectively) although they are involved in
substrate stabilization (hydrogen bonding of both O4 and O6 atoms
of the substrate).
Q18 [β1]
R95 [β1]
D223 [β4]
K166 [β2]
D242 [β4]
H bond O3
Catalytic residue
N163 [β4]
D163 [β4]
E164 [β4]
E338 [β7]
G338 [β7]
W385 [β8]
E392 [β8]
E224 [β4]
A334 [β7]
D243 [β4]
T310 [β6]
Catalytic residue
Oxazoline cavity
W374 [β8]
D376 [β8]
W377 [β7]
D344 [β7]
Π stacking
H bond O4 and O6
Analysis of the hydrolytic properties of TtβGly mutants
The mutations suggested by molecular modeling, N163A and N163D-
E338G, were incorporated into the gene of TtβGly. After expression in
E. coli, purified mutant enzymes were then tested for the hydrolysis of
pNPGlcNAc and pNPGlc in 100 mM phosphate buffer pH 7.0
(Table II). While WT enzyme exhibited no detectable hydrolytic activity
on pNPGlcNAc, both N163A and N163D-E338G mutants gave rise to
a low but significant activity toward this substrate. Saturation kinetics
were not observed for these mutants since substrate saturation was
prevented by the low solubility of pNPGlcNAc, and only kcat/Km para-
meters were determined (Table II). The appearance of a β-N-

Engineering trans-N-acetylglucosaminidase from a glucosidase
399
Table II. Kinetic parameters for wild-type (WT) and variant TtβGly with pNPGlc and pNPGlcNAc as substrate at 40°C, pH 6.5
Mutant
WT
Substrate
kcat (s−1
)
Km (mM)
kcat/ Km (mM−1 s−1
)
pNPGlucose
pNPGlcNAc
pNPGlucose
pNPGlcNAc
pNPGlucose
pNPGlcNAc
pNPGlucose
pNPGlcNAc
pNPGlucose
pNPGlcNAc
13.1 0.5
nda
2.51 0.04
nd
0.11 0.01
nd
4.0 0.2
nd
1.2 × 102
nd
N163A
0.62
1.2 × 10−4
Nd
E338G/N163A
nd
E338G/N163D
7.9 0.6 × 10−3
nd
15
nd
3
× 10−4
1.6 × 10−4
1.4 × 10−3
4.5 × 10−4
E338G/N163D/A211T/I216T
9.3 0.3 × 10−3
1.25 0.08 × 10−2
6.4 0.7
25.3 2.7
and, not detected.
Fig. 3. Kinetics of the hydrolysis of NAG-oxazoline (50 mM). Spontaneous
hydrolysis in phosphate buffer 100 mM pH 7.8 at 30°C (filled circle); With the
N163A mutant (1 mg/mL) (open circle) and with the N163D-E338G mutant
(1 mg/mL) (filled triangle). Inset: Initial velocities of the N163D-E338G TtβGly
mutant-catalyzed hydrolysis of NAG-oxazoline. Initial velocities were
corrected for the rate of the background reaction at each NAG-oxazoline
Fig. 4. Time-course of substrate consumption and product formation in the
presence of N163D-E338G TtβGly mutant (10 µg) with NAG-oxazoline as
donor (12 mM) and BnON(Me)-Glc (10 mM) as acceptor in 50 mM HEPES
buffer, pH 7.8 at 40°C. Key: NAG-oxazoline (filled square); BnON(Me)-Glc
(filled circle); BnON(Me)-GlcNAc (β1,3) Glc (filled triangle).
concentration. The
k
cat/Km value, estimated from the slope, was
.
6.2 0.5 × 10−4 mM−1 s−1
which slightly increased (3-fold) the kcat/Km compared with the origin-
al double mutant. However, the β-glucosidase specificity constant also
increased in this evolved mutant, which resulted in a promiscuous en-
zyme with similar catalytic activity toward pNPGlucose and
pNPGlcNAc.
acetylglucosaminidase activity in N163A or N163D-E338G mutants
was correlated with a strong decrease in the glucosidase activity.
Specificity constants kcat/Km for pNPGlucose showed a 190-fold and
2.2 × 105-fold reduction for N163A and N163D-E338G mutants,
respectively. In comparison, the double mutant N163A-E338G lost
both β-glucosidase and β-N-acetylglucosaminidase activities.
In order to improve the β-N-acetylglucosaminidase activity of the
N163A and N163D-E338G mutants, one cycle of directed evolution
was carried out by error-prone PCR. To facilitate the screening of the
N-acetylglucosaminidase activity directly on colonies, the mutant en-
zymes were expressed in the periplasmic space using the TAT secretion
system. The chromogenic X-GlcNAc substrate is not internalized in
the E. coli cytoplasm but can diffuse within the periplasm, so that
hydrolysis can be detected directly on colonies if the mutant enzymes
are present in this compartment. No improved enzymes were detected
from a library (≈6000 clones) derived from the N163A mutant, and
only one improved mutant was identified from the error-prone library
derived from the double N163D-E338G mutant. Gene sequencing re-
vealed that two additional mutations (A211T/I216T) were present,
NAG-oxazoline hydrolytic properties of ttβGly mutants
β-N-acetylglucosaminidases from Families 20 and 84 use a catalytic
mechanism involving a substrate-assisted catalysis, which goes
through a noncovalent bicyclic oxazoline intermediate (Scheme 1),
which can be easily synthesized (Shoda et al. 2002). We anticipated
that if one of the TtβGly mutants with β-N-acetylglucosaminidase ac-
tivity uses a substrate-assisted catalytic mechanism, it could also cata-
lyze NAG-oxazoline hydrolysis.
NAG-oxazoline is rapidly hydrolyzed to GlcNAc at acidic pH, but
hydrolyzes slowly at basic pH. Kinetics of NAG-oxazoline hydrolysis
were carried out at pH 7.8, where the enzyme retains most of its activ-
ity, while spontaneous hydrolysis of NAG-oxazoline is slow (t1/2 = 15
h at 30°C) (Figure 2). When the N163A mutant was added to the
NAG-oxazoline solution (50 mM), no significant change in the rate
of hydrolysis was observed compared with the background reaction

400
C André-Miral et al.
Scheme 2. Enzymatic transglycosylation with NAG-oxazoline as a donor.
(Figure 3). In contrast, the double mutant N163D-E338G significantly
increased the rate of NAG-oxazoline hydrolysis, demonstrating that
this compound is a substrate for this mutant. As no saturation was
Based on molecular modeling, the single N163A mutation solved
part of the steric conflict between the 2-acetamido group of GlcNAc
derivatives and resulted in an enzyme that showed a new hydrolytic
activity on pNPGlcNAc, but not on NAG-oxazoline. Furthermore,
this mutant is devoid of transglycosylation activity with NAG-oxazo-
line as a donor. Taken together, these results suggest that this new
N-acetylglucosaminidase activity operates via a double displacement
mechanism with a covalent glycosyl-enzyme intermediate, like
N-acetylglucosaminidases belonging to GH3. Accordingly, additional
mutation of the nucleophilic glutamate residue (E338) into a glycine
resulted in a complete loss of enzymatic activity.
observed up to 30 mM substrate concentration, only k /Km was deter-
cat
mined (6.2 0.5 × 10−4 mM−1 s−1). These results suggest that the
double mutant N163D-E338G uses a mechanism involving a
substrate-assisted catalysis. On the contrary, as the N163A mutant
did not catalyze the hydrolysis of NAG-oxazoline and the E338
nucleophile was not mutated, this suggests that this mutant utilizes
the double inversion mechanism with the formation of a covalent
intermediate.
From this inactive double mutant (N163A-E338G), a significant
N-acetylglucosaminidase activity was restored by replacing the
N163A mutation by a N163D mutation, while this double mutant
had lost the nucleophilic residue. This mutation was suggested by
the sequence signature of β-N-acetylglucosaminidases belonging to
GH20 and GH84 families, in which the two carboxylic active site
residues are adjacent in the protein sequence (Asp-Glu in Family 20
and Asp-Asp in Family 84). The first Asp residue is thought to form
a hydrogen bond with the nitrogen of the oxazoline ring, which stabi-
lizes the intermediate oxazoline ring in the substrate-assisted mechan-
ism (Scheme 1). The second carboxylate acts as a general acid–base
catalytic residue. We expected that, by introducing this particular tan-
dem Asp-Glu sequence within the active site of TtβGly, we could gen-
erate β-N-acetylglucosaminidase activity. Indeed, the N163D-E338G
TtβGly mutant, which possesses a tandem Asp163/Glu164 pair, ex-
hibited β-N-acetylglucosaminidase and trans-β-N-acetylglucosamini-
dase activities. The mechanism by which this new activity appeared
is likely to be a substrate-assisted one since first, this mutant acceler-
ated the hydrolysis of NAG-oxazoline and secondly, NAG-oxazoline
is a good donor for the transglycosylation reaction.
Interestingly, the transglycosidase activity of the N163D-E338G
TtβGly mutant was much higher than its hydrolytic activity, which
resulted in a mutant enzyme that could synthesize β-GlcNAc deriva-
tives at high yield (>90%). Until now, such a high yield of conversion
from oligosaccharide-oxazoline derivatives has only been observed
with endo-β-N-acetylglucosaminidases from Families 18, 56 and 85
(Shoda et al. 2002; Ochiai et al. 2009; Parsons et al. 2010). We recent-
ly demonstrated that the mutation of highly conserved residues within
the –1 site of glycosidases decreases dramatically their hydrolytic
activity, while their transglycosidase activity is less affected (Teze
et al. 2014). The present results also confirm this observation.
Attempts to improve the β-N-acetylglucosaminidase activity of the
TtβGly mutants by directed evolution resulted in a modest improve-
ment in the hydrolytic activity of the N163D-E338G double mutant
with the addition of two new mutations A211T and I216T. These mu-
tations are not in direct contact with the active site residues, so it is dif-
ficult to explain their effect. Furthermore, these four mutations
improved in a similar ratio (≈3-fold) the glucosidase activity and the
Transglycosidase activity of the N163D-E338G mutant
The ability of several transglycosylating endo-N-acetylglucosamini-
dases to catalyze the efficient transfer of sugar oxazolines to various
acceptors (Ochiai et al. 2009; Rich and Withers 2009; Parsons et al.
2010) prompted us to test the transglycosylation activity of the
N163D-E338G mutant. The enzymatic reaction was performed
using NAG-oxazoline as a donor and N-methyl-O-benzyl-N-(β-D-
glucopyranosyl)-hydroxylamine as an acceptor. This glucoconjugate
has previously been identified as a good transglycosylation acceptor
for TtβGly (Teze et al. 2013). A small excess of NAG-oxazoline
donor (molar ratio of donor-acceptor, 1.2:1) was chosen to compen-
sate for the hydrolysis (enzymatic and nonenzymatic) of the sugar
oxazoline during the time scale of the experiment. The reaction was
monitored by capillary electrophoresis until near complete consump-
tion of substrates (Figure 4). We observed an almost complete
consumption (>90%) of the acceptor to a single transglycosylation
product, which was identified by NMR after purification as N-me-
thyl-O-benzyl-N-(β-2-D-N-acetylglucopyranosyl
(1→3)-β-D-glucopyranosyl)hydroxylamine (Scheme 2). Such a high
yield of transglycosylation was possible since no self-condensation
of NAG-oxazoline was observed and no hydrolysis of the transglyco-
sylation product was detected after long-term incubation. The excess
of oxazoline was converted into GlcNAc by either spontaneous or
enzymatic hydrolysis.
In contrast, a similar reaction tested with the N163A mutant gave
no transglycosylation product and resulted in the complete conversion
of NAG-oxazoline into GlcNAc (data not shown).
Discussion
This work shows that a rational approach based on molecular model-
ing was successful in broadening the substrate specificity of a Family 1
glycosidase. This is the first time that an N-acetylglucosaminidase ac-
tivity has been created from a GH1 glycosidase. Furthermore, the re-
sults suggest that the catalytic mechanism can be controlled depending
on the mutations within the active site.

Engineering trans-N-acetylglucosaminidase from a glucosidase
401
Cheng O, Li H, Merdek K, Park JT. 2000. Molecular characterization of
the N-acetylglucosaminidase of Escherichia coli and its role in cell wall
recycling. J Bacteriology. 182:4836–4840.
Dennis RJ, Taylor EJ, Macauley MS, Stubbs KA, Turkenburg JP, Hart SJ,
Black GN, Vocadlo DJ, Davies GJ. 2006. Structure and mechanism of a
bacterial β-glucosaminidase having O-GlcNAcase activity. Nat Struct Mol
Biol. 13:365–371.
Dion M, Fourage L, Hallet JN, Colas B. 1999. Cloning and expression of a beta-
glycosidase gene from Thermus thermophilus. Sequence and biochemical
characterization of the encoded enzyme. Glycoconj J. 16:27–37.
Eijsink V, Hoell I, Vaaje-Kolstada G. 2010. Structure and function of enzymes
acting on chitin and chitosan. Biotechnol Genet Eng Rev. 27:331–366.
Feng HY, Drone J, Hoffmann L, Tran V, Tellier C, Rabiller C, Dion M. 2005.
Converting a β-glycosidase into a β-transglycosidase by directed evolution.
J Biol Chem. 280:37088–37097.
β-N-acetylglucosaminidase activity. Examples of native enzymes
presenting both these activities are quite rare. To our knowledge,
only the glycosidase Nag3 belonging to the family GH3 can cleave
pNPGlc and pNPGlcNAc with a similar efficiency (Mayer et al.
2006). The residual glucose activity with mutants that are thought
to use the substrate-assisted mechanism is surprising. One possible
explanation is that the N163D mutation could replace with low effi-
ciency the nucleophile residue absent with the E338G mutation.
Conclusion
We have demonstrated that a limited number of active site mutations
within a β-glycosidase from Family 1 can change both the specificity of
the enzyme and its mechanism of action so that we were able to induce
a β-N-acetylglucosaminidase activity, which is normally found in
Families 3, 20 and 84. All these glycosidase families are supported
by a similar (β/α)8 fold and these results suggest that an ancestral pro-
miscuous enzyme with both activities could have easily evolved toward
one of these activities with a few occasional mutations.
Hart GW, Slawson C, Ramirez-Correa G, Lagerlof O. 2011. Cross talk between
O-GlcNAcylation and phosphorylation: Roles in signaling, transcription,
and chronic disease. Annu Rev Biochem. 80:825–858.
Koné FM, Le Béchec M, Sine JP, Dion M, Tellier C. 2009. Digital screening
methodology for the directed evolution of transglycosidases. Protein Eng
Des Sel. 22:37–44.
Langley DB, Harty DW, Jacques NA, Hunter N, Guss JM, Collyer CA. 2008.
Structure of N-acetyl-beta-D-glucosaminidase (GcnA) from the endocardi-
tis pathogen Streptococcus gordonii and its complex with the mechanism-
based inhibitor NAG-thiazoline. J Mol Biol. 377:104–116.
Leung DW, Chen E, Goeddel DV. 1989. A method for random mutagenesis of a
defined DNA segment using a modified polymerase chain reaction.
Technique. 1:11–15.
Liu QP, Sulzenbacher G, Yuan H, Bennett EP, Pietz G, Saunders K, Spence J,
Nudelman E, Levery SB, White T, et al. 2007. Bacterial glycosidases
for the production of universal red blood cells. Nat Biotechnol. 25:
454–464.
Regarding the synthetic application of these mutants, we have de-
monstrated that this rational approach to designing β-N-acetylgluco-
saminidases provided enzymes with high transglycosidase activity
with NAG-oxazoline as donor and low hydrolytic activity, which
gave a high yield of transglycosylation product. These trans-β-N-acet-
ylglucosaminidases may find useful applications in the synthesis of
O-GlcNAc derivatives.
Funding
Lombard V, Golaconda Ramulu H, Drula E, Coutinho PM, Henrissat B. 2014.
The carbohydrate-active enzymes database (CAZy) in 2013. Nucleic Acids
Res. 42:D490–D495.
This work was supported in part by the Région Pays de la Loire, France
(AAP Glyconet) and F.M.T.K. from the Ivory Coast Government.
Macauley MS, Whitworth GE, Debowski AW, Chin D, Vocadlo DJ. 2005.
O-GlcNAcase uses substrate-assisted catalysis.
J Biol Chem. 280:
25313–25322.
Acknowledgements
Mark BL, Vocadlo DJ, Knapp S, Triggs-Raine BL, Withers SG, James MN.
2001. Crystallographic evidence for substrate-assisted catalysis in a bacter-
ial beta-hexosaminidase. J Biol Chem. 276:10330–10337.
The authors are grateful to Johann Dion for the mass spectrometry
measurements.
Mayer C, Vocadlo DJ, Mah M, Rupitz K, Stoll D, Warren RA, Withers SG.
2006. Characterization of a beta-N-acetylhexosaminidase and a beta-
N-acetylglucosaminidase/beta-glucosidase from Cellulomonas fimi. FEBS
J. 273:2929–2941.
Meichenin M, Rocher J, Galanina O, Bovin N, Nifant’ev N, Sherman A,
Cassagnau E, Heymann MF, Fraser RH, Le Pendu J. 2000. Tk, a new
colon tumor-associated antigen resulting from altered O-glycosylation.
Cancer Res. 60:5499–5507.
Conflict of interest
None declared.
Abbreviations
Ochiai H, Huang W, Wang LX. 2009. Endo-β-N-acetylglucosaminidase-
catalysed polymerization of β-Glcp-(1→4)-GlcpNAc oxazoline: A revisit
to enzymatic transglycosylation. Carbohydr Res. 344:592–598.
Ogata M, Zeng X, Usui T, Uzawa H. 2007. Substrate specificity of
N-acetylhexosaminidase from Aspergillus oryzae to artificial glycosyl
acceptors having various substituents at the reducing ends. Carbohydr
Res. 342:23–30.
Osanjo G, Dion M, Drone J, Solleux C, Tran V, Rabiller C, Tellier C. 2007. Di-
rected evolution of the α-L-fucosidase from Thermotoga maritima into an
α-L-transfucosidase. Biochemistry. 46:1022–1033.
EPP, error-prone PCR; ESI, electrospray ionization; GH, glycoside hydrolases;
HEPES, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; IPTG, iso-
propylthiogalactoside; LB, Luria-Bertani; NAG-oxazoline: 2-methyl-(1,2-
dideoxy-α-D-glucopyrano) [2,1-d]-oxazoline; NGT: NAG-thiazoline; NMR,
nuclear magnetic resonance; PCR, polymerase chain reaction; PDB, protein
data bank; pNP, product concentration; TAT, Twin Arginine translocation;
TtβGly, Thermus thermophilus; UV, ultra-violet; WT, wild-type; X-GlcNAc,
β-D-N-acetylglucosamine002E; 3D, tridimensional.
Parsons TB, Patel MK, Boraston AB, Vocadlo DJ, Fairbanks AJ. 2010. Strepto-
References
coccus pneumonia endohexosaminidase D. feasibility of using N-glycan ox-
azoline donors for synthetic glycosylation of
a GlcNAc-asparagine
Bode L, Jantscher-Krenn E. 2012. Structure-function relationships of human
milk oligosaccharides. Adv Nutr. 3:383–391.
acceptor. Org Biomol Chem. 8:1861–1869.
Rich JR, Withers SG. 2009. Emerging methods for the production of homoge-
Çetinbas N, Macauley MS, Stubbs KA, Drapala R, Vocadlo DJ. 2006. Identifi-
cation of Asp174 and Asp175 as the key catalytic residues of human
O-GlcNAcase by functional analysis of site-directed mutants.
Biochemistry. 45:3835–3844.
neous human glycoproteins. Nat Chem Biol. 5:206–215.
Shoda SI, Fujita M, Lohavisavapanichi C, Misawa Y, Ushizaki K, Tawata Y,
Kuriyama M, Kohri M, Kuwata H. 2002. Efficient method for the

402
C André-Miral et al.
elongation of the N-acetylglucosamine unit by combined use of chitinase
Vocadlo DJ, Mayer C, Shouming H, Withers SG. 2000. Mechanism of action
and identification of Asp242 as the catalytic nucleophile of Vibrio furnisii
N-acetyl-β-D-glucosaminidase using 2-acetamido-2-deoxy-5-fluoro-α-L-
idopyranosyl fluoride. Biochemistry. 39:117–126.
Vocadlo DJ, Withers SG. 2005. Detailed comparative analysis of the catalytic
mechanism of β-N-acetylglucosaminidases from families 3 and 20 of glyco-
side hydrolases. Biochemistry. 44:12809–12818.
and β-galactosidase. Helv Chim Acta. 85:3919–3936.
Silvestro A, Pommier J, Pascal MC, Giordano G. 1989. The inducible trimethy-
lamine N-oxide reductase of Escherichia coli K12: Its localization and indu-
cers. Biochim Biophys Acta. 999:208–216.
Slàmovà K, Bojarova P, Petraskova L, Kren V. 2010. β-N-acetylhexosaminidase:
What’s in a name?. Biotechnology Adv. 28:682–693.
Teze D, Dion M, Daligault F, Tran V, André-Miral C, Tellier C. 2013. Alkoxyamino
glycoside acceptors for the regioselective synthesis of oligosaccharides using gly-
cosynthases and transglycosidases. Bioorg Med Chem Lett. 23:448–451.
Teze D, Hendrickx J, Czjzek M, Ropartz D, Sanejouand YH, Tran V, Tellier C,
Dion M. 2014. Semi-rational approach for converting a GH1 β-glycosidase
into a β-transglycosidase. Protein Eng Des Sel. 27:13–19.
Vollmer W. 2008. Structural variation in the glycan strands of bacterial peptido-
glycan. FEMS Microbiol Rev. 32:287–306.
Williams SJ, Mark BL, Vocadlo DJ, James MN, Withers SG. 2002. Aspartate
313 in the Streptomyces plicatus hexosaminidase plays a critical role in
substrate-assisted catalysis by orienting the 2-acetamido group and stabiliz-
ing the transition state. J Biol Chem. 277:40055–40065.