Rational design of a thermostable glycoside hydrolase from family 3 introduces β-glycosynthase activity

Article abstract of DOI:10.1093/glycob/cww081

The thermostable β-glucosidase from Thermotoga neapolitana, TnBgl3B, is a monomeric threedomain representative from glycoside hydrolase family 3. By using chemical reactivation with exogenous nucleophiles in previous studies with TnBg13B, the catalytic nucleophile (D242) and corresponding acid/base residue (E458) were determined. Identifying these residues led to the attempt of converting TnBgl3B into a β-glucosynthase, where three nucleophilic variants were created (TnBgl3B-D242G, TnBgl3B-D242A, TnBgl3B-D242S) and all of them failed to exhibit glucosynthase activity. A deeper analysis of the TnBgl3B active site led to the generation of three additional variants, each of which received a single-point mutation. Two of these variants were altered at the -1 subsite (Y210F, W243F) and the third received a substitution near the binding site's aglycone region (N248R). Kinetic evaluation of these three variants revealed that W243F substitution reduced hydrolytic turnover while maintaining KM. This key W243F mutation was then introduced into the original nucleophile variants and the resulting double mutants were successfully converted into β-glucosynthases that were assayed using two separate biosynthetic methods. The first reaction used an α-glucosyl fluoride donor with a 4-nitrophenyl-β-D-glucopyranoside (4NPGlc) acceptor, and the second used 4NPGlc as both the donor and acceptor in the presence of the exogenous nucleophile formate. The primary specificity observed was a β-1,3-linked disaccharide product, while a secondary β-1,4-linked disaccharide product was observed with increased incubation times. Additional analysis revealed that substituting quercetin-3-glycoside for the second reaction's acceptor molecule resulted in the successful production of quercetin-3,4'-diglycosides with yields up to 40%.

Full text of DOI:10.1093/glycob/cww081

Glycobiology Advance Access published August 3, 2016
Rational design of a thermostable glycoside hydrolase from family 3 introduces
β-glycosynthase activity
Tania Pozzo1, Javier Romero-García2, Magda Faijes2 Antoni Planas2, Eva Nordberg
Karlsson1
1Biotechnology, Department of Chemistry, Lund University, P.O. Box 124, SE-221 00 Lund,
2
Sweden, Laboratory of Biochemistry, Institut Quimíc de Sarrià, Universitat Ramon Llull,
Via Augusta 390, E-08017 Barcelona, Spain
1To whom correspondence should be addressed: Tel: + 46 46-222 98 56, Fax: +46 46-222 47
13, e-mail: tania.pozzo@gmail.com
Keywords: α-glucosyl fluoride / β-glycosynthase / formate / quercetin / Thermotoga
neapolitana
Supplementary Data:
 Table TS1 - Primers that were used for site directed mutagenesis of TnBgl3B.
 Fig. S1 - SDS-PAGE image of the purified TnBgl3B protein.
 Fig. S2 - HPLC chromatograms of 4-Nitrophenyl glycoside standards.
 Fig. S3 - HPLC Chromatogram of Glycosynthase reaction products.
 Fig. S4 - Glycosynthetic products formed using TnBgl3B_W243F_D242A.
 Fig. S5 - Glycosynthetic products formed using TnBgl3B_W243F_D242G.
© The Author 2016. Published by Oxford University Press. All rights reserved. For permissions,
please email:journals.permissions@oup.com

Abstract
The thermostable β-glucosidase from Thermotoga neapolitana, TnBgl3B, is a monomeric
three-domain representative from glycoside hydrolase family 3 (GH3). By using chemical
reactivation with exogenous nucleophiles in previous studies with TnBg13B, the catalytic
nucleophile (D242) and corresponding acid/base residue (E458) were determined. Identifying
these residues led to the attempt of converting TnBgl3B into a β-glucosynthase, where three
nucleophilic variants were created (TnBgl3B_D242G, TnBgl3B_D242A, TnBgl3B_D242S)
and all of them failed to exhibit glucosynthase activity. A deeper analysis of the TnBgl3B
active site led to the generation of three additional variants, each of which received a single
point mutation. Two of these variants were altered at the -1 sub-site (Y210F, W243F) and the
third received a substitution near the binding site’s aglycone region (N248R). Kinetic
evaluation of these three variants revealed that W243F substitution reduced hydrolytic
turnover while maintaining KM. This key W243F mutation was then introduced into the
original nucleophile variants and the resulting double mutants were successfully converted
into β-glucosynthases that were assayed using two separate biosynthetic methods. The first
reaction used an α-glucosyl fluoride donor with a 4-nitrophenyl β-D-glucopyranoside
(4NPGlc) acceptor, and the second used 4NPGlc as both the donor and acceptor in the
presence of the exogenous nucleophile formate. The primary specificity observed was a β-
1,3-linked disaccharide product, while a secondary β-1,4-linked disaccharide product was
observed with increased incubation times. Additional analysis revealed that substituting
quercetin-3-glycoside for the second reaction’s acceptor molecule resulted in the successful
production of quercetin-3,4´-diglycosides with yields up to 40%.
2

Introduction
Glycosynthases have been successfully applied to facilitate in vitro biosynthesis of both
artificial and natural analogues of glycopolymers that are commonly used in biomaterials
manufacturing (Faijes et al. 2006; Perez et al. 2011; Zachary and Withers 2013; Cobucci-
Ponzano et al. 2012). These enzymes are engineered Glycoside Hydrolases (GHs) that exhibit
substitution of the key nucleophilic residue, either Glutamate (Glu) or Aspartate (Asp), for a
smaller, uncharged amino acid (aa). By convention, these point mutations are commonly
made to replace the nucleophilic residue either with Alanine (Ala), Glycine (Gly), or Serine
(Ser). The resulting nucleophilic variants often display complete or severely reduced
hydrolytic activity, thus enabling their use as biocatalysts for glycosidic bond formation in
the presence of appropriate donor and acceptor molecules (Williams and Withers 2000). Two
methods for the characterization of glycosynthetic activity have been previously described
and their reaction mechanisms are depicted in figure 1: the first involves the use of α-glucosyl
fluoride donors (Mackenzie et al.1998; Malet et al. 1998), while the second describes the
implementation of exogenous nucleophiles to mimic covalent bond formation between
enzyme and donor molecules (Moracci et al. 1998).
A thorough study of the residues involved in both the catalytic mechanism and active site
morphology allows for a deeper understanding of enzymatic function and specificity. This
concept of rational enzyme design allows for the introduction of new biosynthetic functions,
an example of which is the use of a glycosynthase as a biocatalyst for the in vitro
modification of flavonoids (Plaza et al. 2014; Pozzo et al. 2014). Using glycosynthases for
the modification of flavonoids is a sound application of biotechnology based on the fact that
many of these polyphenolic compounds possess antioxidant activity and are thought to exert
protective effects against different diseases when conjugated with specific sugar residues
3

(Lee et al. 2006). This chemical modification process is referred to as glycosylation, wherein
a flavonoid’s physicochemical properties may be altered to enable its use as an additive over
a wide array of applications in food science technology, pharmaceutical industry and
cosmetic design (Plaza et al. 2014).
A strong candidate for engineering a glycosynthase capable of carrying out these types of
applications in biotechnology was determined to be the β-glucosidase from Thermotoga
neapolitana (TnBgl3B), which was previously reported as the first thermostable, three-
domain representative from GH3 by Pozzo et al. 2010. However, generating glycosynthases
from enzymes belonging to families with scant levels of described activity and structural
morphology can be challenging. In the case of GH3, 25 complete structures have been
reported to date out of 284 enzymes that were characterized from a list of several thousand
sequences deposited in the Carbohydrate-Active enZYmes database (CAZy). This publically
available information, while somewhat limiting in terms of breadth and resourcefulness,
provided a starting point for the conversion of TnBgl3B.
In the first attempts at engineering alternative function into TnBgl3B, stand-alone
construction of three inactive, nucleophile-mutated variants (TnBgl3B_D242G,
TnBgl3B_D242A, TnBgl3B_D242S) proved to be insufficient for the acquisition of
glycosynthase activity. Therefore, a secondary key point mutation was introduced adjacent to
the nucleophile (TnBgl3B_W243F_D242A), and a biosynthetic activity conducive to
oligosaccharide production was achieved. In light of this achievement, the study was
extended to include an evaluation of quercetin type flavonoid efficacy when serving as
acceptor molecules in glycosynthetic reactions. Acquiring this information on catalytic
residues was necessary to establish a viable basis for the further development of TnBgl3B as
the first thermostable glycosynthase in GH family 3.
4

Results
Production of engineered TnBgl3B variants
The gene encoding TnBgl3B from T. neapolitana was cloned into pET-22b(+) by Turner et
al. 2007. This construct was used as a template for the engineering of nine TnBgl3B variants,
three of which include previously reported nucleophilic variants displaying single amino acid
(aa) substitutions (TnBgl3B_D242A/G/S) by Pozzo et al. 2010. Our attempt to engineer
glycosynthetic activity into TnBgl3B resulted in the generation of three doubly mutated
nucleophilic variants (TnBgl3B_W243F_ D242A/G/S), and three active site variants with
intact nucleophilic residues that received single residue mutations at alternative sites
(TnBgl3B_Y210F and TnBgl3B_W243F at the -1 subsite and TnBgl3B_N248R within a loop
near the active site, see Figure 2).
All nine variants and the TnBgl3B wild type (wt) were recombinantly produced using the E.
coli expression strain BL21 (DE3). The use of this system resulted in a soluble protein
production rate of approximately 80% and there was no observable difference in production
among the variants. The enzymes were purified to apparent homogeneity using the two-step
protein purification strategy reported by Turner et al. 2007. The first step involved a standard
heat treatment (80C, 30 min) to denature all non-thermostable proteins. From there, the
targeted proteins were isolated by immobilized metal ion affinity chromatography (IMAC),
where the implemented system utilized a copper ligand to bind C-terminal His×6-tags that
were included in the cloning design (Figure 1 of the supporting material).
Determination of TnBgl3B mutant hydrolytic activity
Enzyme kinetic values were determined for the wt (TnBgl3B) and single active site variants
(TnBgl3B_Y210F, TnBgl3B_W243F and TnBgl3B_N248R) using 4NPGlc as the substrate at
5

80C. All three single mutation variants exhibited a decreased enzymatic turnover rate when
compared to the wt. The smallest turnover rate change was observed with TnBgl3B_N248R
(a mutation introduced with the aim to stabilize a loop located in the vicinity of the active site
region), where turnover was reduced to approximately 33% of the wt. TnBgl3B_Y210F
(interacting with the glycone in the -1 subsite) resulted in a significant increase in KM and a
92% reduction in turnover, which indicated poor binding affinity of the glycone. The KM
value for TnBgl3B_W243F was unaffected by comparison to the wt, but the hydrolytic
turnover showed a 62% decrease (results are summarised in Table I).
Enzyme activity was also assayed using cello-oligosaccharide substrates and the results
indicated that active site variants with unaffected KM values for 4NPGlc were able to
effectively hydrolyse longer substrates (Figure 3). Assay of the three mutants with single aa
substitutions revealed that their initial velocity of hydrolysis was low when compared to the
initial velocities observed with the shorter substrate 4NPGlc. This implied that cello-
oligosaccharide accommodation at the active site was not efficient. The most obvious
indication of this was the similar initial velocities of cello-oligosaccharide hydrolysis that
were observed between the wt and TnBgl3B_N248R variant, which indicated that the N248R
substitution within the loop (residues 242-248 in the first domain, with residue 248 located
approximately 15 Å from the nucleophile) did not improve substrate accommodation for
cellotetraose (C4) and cellopentaose (C5) as originally intended. On the other hand,
TnBgl3B_W243F showed a significantly lower initial velocity using these same cello-oligo
substrates, which indicated that a mutation at the -1 subsite affects substrate accommodation
in a manner that resulted in lowering hydrolytic activity with cello-oligosaccharides. In the
case of the TnBgl3B_Y210F variant, hydrolytic activity on cello-oligosaccharide substrates
was completely absent, which implied a strong importance for establishing contact between
6

the OH1 of the sugar and the -1 subsite to facilitate the formation of a hydrogen bond (Figure
2).
Determination of temperature and pH effects on enzyme activity
Initial velocities were determined for wt, TnBgl3B_W243F, the three single nucleophilic
variants (TnBgl3B_D242A/G/S) and the three doubly mutated nucleophilic variants
(TnBgl3B_W243F_ D242A/G/S) under two conditions. Based on a reported thermostable
range of 60-90 C by Turner et al. 2007, the first set of initial velocities were acquired from
reactions that were carried out at 80C with a pH of 5.6. From there, a second set of initial
velocities were measured using the classical glycosynthase reaction conditions of 35C and a
pH of 7, where these conditions ensure the stability of the α-glucosyl fluoride donor.
Hydrolytic activity under the classical glycosynthase reaction conditions was extremely
reduced for the wt enzyme, and a complete loss of activity was observed for the
TnBgl3B_W243F variant. Additionally, little to no residual hydrolytic activity was detected
for nucleophilic variants displaying both single and double mutations. Interestingly, the
reactions carried out with the single variants under the high temperature conditions displayed
more residual activity than those with the double mutants. This outcome was likely attributed
by poor substrate accommodation at the -1 subsite due to the W243F mutation present in each
of the double mutants (Table II). Additionally, the specific activity of TnBgl3B_W243F was
approximately half of that observed in the wt at high temperature. This observation was in
accordance with the turnover determined at 90C by Pozzo et al. 2010, suggesting that the
mutation located next to the nucleophilic residue (W243F) could be crucial to stabilising the
glycone donor at the -1 subsite.
7

Chemical reactivation of TnBgl3B variants with exogenous nucleophile
Nucleophilic variants were reactivated using the exogenous nucleophile sodium azide
(NaN3). This chemical rescue experiment was carried out for all three doubly mutated
nucleophilic variants (TnBgl3B_W243F_ D242A/G/S) in a pH of 5.6, at 80C. Different
NaN3 concentrations were used to identify the optimal conditions for achieving the maximum
recovery of hydrolytic activity. Enzyme and 4NPGlc concentrations were fixed at 0.05 μM
and 1mM respectively, where NaN3 concentration was varied from 0.2 to 4 mM. The assay
indicated that maximum activity was recovered, only for TnBgl3B_W243F_D242A and
TnBgl3B_W243F_D242S, with 0.5 M NaN3 (Figure 4). The kinetics after chemical rescue at
80C could only be measured for the D242S-variant, which also had a low turnover
compared to the wt (Table III).
Assaying glycosynthase activity
A. “Glycosynthase with α-glycosyl fluoride donor”
The classical glycosynthase reaction was carried out using a 1:5 ratio of α-glucosyl fluoride
and 4NPGlc for all six variants of TnBgl3B. All three of the TnBgl3B variants with single aa
mutations failed to exhibit glycosynthase activity. In the case of the doubly-mutated
nucleophilic variants, the TnBgl3B_W243F_D242G variant failed to generate a biosynthetic
product. However, the TnBgl3B_W243F_D242A and TnBgl3B_W243F_D242S variants
succeeded in carrying out glycosynthase activity, where the same 4-Nitrophenyl-β-1,3-
laminaribiose (4NPLam) product was detected with yields of 13 and 37% respectively (Table
IV).
In an attempt to improve the reaction yield, the protein concentration was increased to 2.5
μM, which resulted in an improved yield for TnBgl3B_W243F_D242A, but not for
8

TnBgl3B_W243F_D242S (data not shown). Increasing the concentration of
TnBgl3B_W243F_D242A also resulted in the production of both β-1,3 (4NPLam) and β-1,4-
(4NPCel) linked disaccharide products. In all instances of observed glycosynthetic activity,
the primary product formed was always observed to contain a β-1,3-linkage, while the slower
formation of the β-1,4-linkage was determined to be secondary. For
TnBgl3B_W243F_D242S (with a 37% yield using 0.5 μM of enzyme concentration), further
increases in enzyme concentration was insufficient for improving the product yield, which
was likely due to residual hydrolytic activity. In this case, a mixture of synthesis and
hydrolysis was observed after performing an HPLC analysis (data not shown), and the
optimal glycosynthetic reaction conditions were determined to be the use of a low enzyme
concentration under prolonged incubation times (Table IV).
B. “Glycosynthase-type reaction with in situ donor generation”
The single- and doubly-mutated nucleophilic variants were also tested in glycosynthetic
reactions using 4NPGlc as both the donor and acceptor molecule, where sodium formate was
the exogenous nucleophile for in situ generation of the activated α-glycosyl formate donor
(Figure 1B). This glycosynthetic approach was also carried out under the same reaction
conditions (80° C, pH 5.6), and used a fixed enzyme concentration of 0.5 μM for all reactions
(Table V and supplementary data Figure S4 and S5). Control reactions that lacked the
addition of an enzyme confirmed that the synthetic reaction was achieved due to the presence
of a glycosynthase. Additionally, spontaneous transglucosylation was not detected in the
control reactions that were carried out without the presence of sodium formate. Optimization
experiments were then conducted by assessing the effect of various sodium formate
concentrations (0.1 to 2 M) over a pH range of 3 to 7. This led to the determination of
maximum initial velocities, which were obtained with 2 M formate, at pH 5.6 (Table V).
9

Among the six TnBgl3B variants, TnBgl3B_W243F_D242S showed the highest
glycosynthetic activity and produced 4-Nitrophenyl-β-1,3-laminaribiose (4NPLam) with
yields of 23%.
Enzymatic glycosylation of flavonoids leads to structural insight
Using the glycosynthase-type reaction and the TnBgl3B-synthase W243F_D242A, flavonoid
glycosylation was observed when using quercetin-3-glucoside, quercetin-3-galactoside and
quercetin as acceptor molecules with yields of 30, 40 and 15% respectively (Table VI). These
results were further investigated by conducting a structural analysis of this β-exo-
glycosynthase from GH3 to acquire a better understanding of donor and acceptor molecule
accommodation at the active site.
The docking of donor and acceptor molecules within the active site was performed using the
structure of TnBgl3B_D242A (PDB code 2X42). In addition, an extended loop (residues 409-
434, between helices K1 and K2 in domain 2), which was too disordered between residues
420-422 to be resolved in the crystal structure of the wt (Pozzo et al. 2010), was modelled.
The tip of this loop (residue 419) was approximately 12Å from the O1 atom of glucose bound
in the -1 subsite and was previously suggested to either interact with the substrate, or to
stabilize the structure.
The Trp243 residue of the TnBgl3B wt has an equivalent position to residue W286 in HvExoI
(Hrmova et al. 2002; Hrmova et al. 2004; Pozzo et al. 2010), but the residue in HvExoI is
located in the +1 sub-site, while the residue in the TnBgl3B wt is in an energetically
unfavourable rotamer (stabilized by several hydrophobic interactions, and by a H-bond from
Ala40) that has swung to the -1 subsite.
Mutagenesis of this residue to Phe will, according to the modelling, likely change the position
of the new aromatic residue into its more energetically favoured position, thus moving the
10

interaction from the -1 to the +1 subsite (Figure 5A and 5B). Mutagenesis of this residue
(W243) appears to be fundamental for the possibility to utilize the enzyme as a
glycosynthase, and it is likely that the donor glucosyl residue in the -1 subsite will receive a
slightly modified accommodation, which is not reflected in the KM value of the mutant
enzyme.
11

Discussion
In an attempt to create a glycosynthase with TnBgl3B, the nucleophilic region of the active
site was targeted for inactivation by introducing single point mutations (TnBgl3B_D242G,
TnBgl3B_D242A, and TnBgl3B_D242S), but these mutants did not exhibit the targeted
activity. Because GH family 3 is characterized as having broad structural variability among
its members in terms of domain arrangements and conformation of the active site, the next
logical step in the targeted evolution of glycosynthase activity in TnBgl3B was to deliver an
additional point mutation to each single nucleophile variant to produce the double
nucleophile
variants
TnBgl3B_D242G_W243F,
TnBgl3B_D242A_W243F
and
TnBgl3B_D242S_W243F. The location of the second mutation was selected based on the
previous observation that the Trp243 residue of the wt forms and influences substrate
accommodation at the -1 subsite, where TnBgl3B_W243F mutants displayed extremely
reduced hydrolytic activity when compared to wt (Pozzo et al. 2010).
Thermostable glycosynthases in oligosaccharide synthesis
TnBgl1A, a thermostable glycoside hydrolase from GH family 1 derived from Thermotoga
neapolitana, was previously reported to exhibit a mutagenesis-induced glycosynthetic
activity that preferentially generated a β-1,3 linked oligosaccharide product in yields of up to
45% when using formate as an exogenous nucleophile, while also producing a β-1,4 linked
oligosaccharide with lower yields of 6% when using the “classical glycosynthase
mechanism” (Pozzo et al. 2014). Both of these glycosynthetic strategies were first applied to
all TnBgl3B single nucleophile variants in order to screen for successful mutagenesis events.
The negative results with these initial variants inspired the generation of the TnBgl3B double
nucleophilic variants, three of which demonstrated glycosynthetic activity using both
aforementioned strategies (TnBgl3B_W243F_D242A/S/G). Under the reaction conditions of
12

the “classical glycosynthase reaction”, 37% product yields were observed for
oligosaccharides with β-1,3 linkages. Similar to the results obtained with TnBgl1A-
glycosynthases, the TnBgl3B-glycosynthases were found to produce lower product yields of
β-1,4 linked oligosaccharides at only 3%.
Flavonoid modification using glycosynthases
Applying modifications to the native structure of flavonoids can alter their physicochemical
properties and, in some cases, lead to the development of novel compounds that may hold the
potential to improve various areas of biotechnology. Such compounds can be produced by
using alternative glycosides as substrate donors, as well as by targeting unusual positions of
the molecular backbone for glycosylation. In this way, natural flavonoid diversity can be
utilized in the laboratory as a tool to promote investigations that aim to enhance the
biological properties of known bioactive molecules based on alterations in bioavailability
(Riva et al. 2002; Bernhardt et al. 2009; is upic va et al. 00 ang et al. 2010; Xiao et
al. 2011). However, while these molecules have the potential to be applied as additives in
pharmaceuticals, cosmetics and food science technologies, the expansion of flavonoid
utilization for industrial applications is limited due to their low solubility in water. If specific
modifications at certain positions could be obtained selectively, the solubility issue could be
reduced by selective glycosylation of the flavonoids by using enzymes that produce
molecular products with a higher solubility (Plaza et al. 2014).
It should be noted that glycosynthases have already been successfully used as tools for the in
vitro modification of flavonoids. An example of this was the work of Yang et al. 2007, where
a glycosynthase from Humicola insolens, Cel7B_E197S, was screened for glycosynthetic
activity with 20 donor and 80 acceptor molecules through the use of a high throughput mass
spectrometry-based method. The work of this group demonstrated that finding the appropriate
13

donor and acceptor molecules is crucial for the success of the glycosynthetic reaction, which
is in large part due to the structural conformation of the substrate binding domain that confers
enzymatic function and substrate specificity. Our alternative strategy to employ rational
modifications to the TnBgl3B active site bolsters both the importance and value in
understanding substrate-binding architecture for the introduction of glycosynthetic activity to
glycosyl hydrolase enzymes from family 3.
A recent study that stresses the significance of determining the proper donor and acceptor
molecules for the targeted enzyme in a glycosynthetic reaction involved the TnBgl1A β-
glycosynthases from T. neapolitana that were used as biocatalysts in the modification of
quercetin and quercetin glycosides. These enzymes produced low synthetic yields (10%) due
to their narrow active site structure that performed poorly in the accommodation of the bulky
flavonoid molecules (Pozzo et al. 2014). By contrast, we have demonstrated here that
TnBgl3B_D242A_W243F successfully synthesised Q3Gal4’Glc in yields of up to 40%,
showing that glycosynthases from GH3 members have the potential to be utilized for the
glycosylation of bulky acceptor molecules such as polyphenolic antioxidants or antibiotics.
Further explanations for the TnBgl3B_D242A_W243F active sites’ ability to successfully
accommodate flavonoid-type molecules could be revealed by carrying out molecular
dynamics simulations similar to those using the ElNémo server that were reported by Suhre et
al. 2004. The simulations showed that the second domain of TnBgl3B presents a long and
highly flexible loop (410-435) that moves close to the active site entrance and possibly
accommodates these bulky substrates. Performing a similar study at this level of detail could
be useful in confirming the feasibility of implementing candidate glycosynthases from GH3
in industrial and pharmaceutical applications. A example of one such GH3-glycosynthase is
that of Jakeman et al. 2011, where the enzyme has been shown to utilize fluorinated sugars
14

and erythromycin to produce a glucosylated erythromycin A product that exhibits β-1,2-
linkages.
Conclusion
In this work, the first thermostable glycosynthase belonging to the GH family 3 was
successfully constructed and characterised in the synthesis of β-1,3 and β-1,4-linked
oligosaccharides. In order to explore the work capacity of this thermostable glycosynthase
further, the new acceptor molecules quercetin and quercetin-3-glycoside were successfully
glycosylated to form the primary quercetin-3,4´-diglycoside product. Gathering insight as to
which targeted substitutions to the active site are capable of bringing about the necessary
conformational changes required for glycosynthase activity is important for future success in
the bioengineering of enzymes that are capable of producing synthetic glycopolymers and
other glycosylated natural products. While the previously reported success of engineering
glycosynthetic activity into TnBgl1A (an enzyme classified under GH1) was presented with
an emphasis on the enzyme’s biosynthetic mechanism (Pozzo et al. 2014), the efforts put
forth to introduce a similar mode of action into TnBgl3B emphasize the importance of
identifying active site mutations that are alternative to the nucleophilic residue, yet required
for the acquisition of glycosynthetic activity.
15

Materials and Methods
Chemicals
4-Nitrophenyl β-D-glucopyranoside (4NPGlc), cellotetraose (β-D-Glc-[1→4])3-D-Glc) and
cellopentaose (β-D-Glc-[1→4])4-D-Glc) and 4-Nitrophenyl-β-1,3-laminaribiose were
purchased from Sigma-Aldrich (Steinheim, Germany). The α-glycosyl fluoride was
chemically synthesized as described previously in Faijes et al. 2006. Briefly, peracetylated
glucose receives fluoride followed by an O-deacetylation by methoxide to produce α-glycosyl
fluoride in high yields (more than 80%). Quercetin 4’-O-β-glucopyranoside, Quercetin 3,4’-
di-O-β-glucopyranoside were purchased from Polyphenols Laboratories AB (Sandnes,
Norway) and Quercetin-3-O-β-galactoside from Extrasynthese (Lyon, France).
Engineering TnBgl3B active site variants
Construction of six thermophilic active site variants was carried out by site directed
mutagenesis using plasmids encoding the β-glucosidase gene from T. neapolitana (TnBgl3B
for single mutation variants and TnBgl3B_W243F for double nucleophile variants) as the
DNA template and mutagenic primers synthesized by Eurofins MWG Operon in Ebersberg,
Germany (primer sequences are described in Supplementary data, Table S1).
PCR reactions were run in a T-Gradient thermocycler (Biometra, Göttingen, Germany) using
iProof high fidelity DNA polymerase (iProofTM High-Fidelity master mix Bio-Rad
laboratories, California, USA), and the thermal cycling protocol was as follows: an initial
denaturing step of 30 s at 95C, followed by 30 cycles of denaturing for 30 s at 95C,
annealing for 60 s at 55C, an extension time of 5 min 30 s at 68C and a final extension of
10 min at 68C.
16

PCR reactions were digested with DpnI (New England, Biolabs) for one hour and
transformed into the electrocompetent E. coli strain NovaBlue (Novagen) using the
GenePulser II (Bio-Rad). The competent cells were immediately incubated for 1 hour at 37C
and spread on LB agar containing 100 µg/mL ampicillin. E. coli colonies were selected from
the plates and an overnight inoculum was prepared for plasmid extraction in 5 ml LB media
with ampicillin (100 µg/mL). Plasmids were purified using the Zyppy plasmid miniprep kit
(Zymo Research, CA, USA) and sent for sequencing using vector specific T7 primers and
pET-RP supplied by the sequencing facility at GATC Biotech (Konstanz, Germany) to obtain
complete coverage of the entire β-glucosidase encoding genes.
Enzyme production and purification
All constructs (active-site variants and TnBgl3B) were used to transform competent E. coli
BL21(DE3) cells (Novagen) for protein production. The cultivation was carried out in a 2 L
shake flask at 37C shaking at 200 rpm. The culture volume was one litre of LB medium with
100 µg/mL ampicillin. Enzyme production was induced using a final concentration of
0.1 mM IPTG when the OD, at λ=600 nm, reached the value of 0.6, and cultivation continued
for up to 4 h. Subsequently, the cultivations were centrifuged (10,000  g, 10 min, 4 C), and
cell pellets were resuspended in binding buffer (20 mM imidazole, 20 mM Tris-HCl, 0.75 M
NaCl, pH 7.5). Cell suspensions were then lysed by sonication over three cycles with two-
minute duration times using a 14 mm titanium probe, 60% sound intensity, and a cycle of 0.5
(UP400S, Dr. Hielscher). Cell lysates were then immediately subjected to an additional round
of centrifugation (30 min, 39,000  g, 4C).
The supernatant was heat treated at 80C for 30 min and centrifuged (10,000  g, 20 min).
The resulting supernatant had the His-tagged enzyme and was used for purification using an
ÄKTA prime system (Amersham Biosciences, Uppsala, Sweden) with a 5 ml HiTrap affinity
17

column (GE health care, Germany) by immobilized metal ion affinity chromatography.
Fractions containing the purified enzyme were pooled and dialyzed against 20 mM citrate
phosphate buffer, pH 5.6, overnight using a dialysis membrane with a molecular weight cut-
off of 3.5 kDa (Spectrum laboratories, Rancho Dominguez, CA, USA). The dialysed protein
fractions were stored at 4C for future use. Purity of all enzymes was confirmed by SDS-
PAGE and protein concentrations were estimated by absorbance measurements at λ=280 nm
using the NanoDrop 1000 (Thermo Fisher Scientific).
Enzyme kinetics using 4NPGlc
Kinetic parameters for TnBgl3B (wt), and its active site variants involved in putative
substrate interactions (TnBgl3B_Y210F, TnBgl3B_W243F and TnBgl3B_N248R), were
determined at 80C using a UV-1650PC spectrometer (Shimadzu, Kyoto, Japan) connected to
the JulaboMB (Labortechnik GMBH, Germany) temperature controlled system. Continuous
spectrophotometric monitoring of 4-nitrophenol (4NP) liberation from 4-nitrophenyl β-D-
glucopyranoside (4NPGlc) was carried out at λ=400 nm for 300 seconds after adding the
enzyme (0.012 to 0.05 µM), with an extinction coefficient (εM) of 1785 M−1cm−1 under the
conditions of pH5.6 and 80C. Under these same operating conditions, different 4NPGlc
concentrations (0.05 to 6 mM) in 50 mM citrate phosphate buffer were used to evaluate the
kinetic parameters using the Michaelis-Menten formula in GraphPad Prism 6 (La Jolla, CA,
USA). A blank was run for each 4NPGlc concentration to neglect spontaneous substrate
hydrolysis.
Enzyme activity on cello-oligosaccharides
Hydrolytic activity was determined at pH 5.6 and 80C using cellotetraose (β-D-Glc-
[1→4])3-D-Glc) and cellopentaose (β-D-Glc-[1→4])4-D-Glc) at different concentrations (0.5
18

to 1.5 mM). The reactions were monitored every 10 min for 2 hours, after adding the enzyme
(1.2 µM), by scanning for the appearance of a glucose peak detected by HPAEC-PAD (high
performance anion-exchange chromatography with amperometric detection) using a Dionex
ICS-5000, Ion Chromatograph (Dionex Corporation, Sunnyvale, CA, USA). A CarboPac PA-
200 anion-exchange column (3 x 150 mm and 3 x 30 mm CarboPac PA-20 guard column)
was used. The oligosaccharides, together with glucose, were eluted by a gradient of eluents:
(A) milliQ water (B) 20 mM NaOH+400 mM NaAc (C) 200 mM NaOH. A Gradient of 50%
eluent A, 25% eluent B and 25% eluent C was constant during for 15-20 min with a flow rate
of 0.5 mL/min. Chromeleon 7 software was used to control the system and analyse the data.
This proportion was kept constant for 30 min. Initial velocities for TnBgl3B wt and its active
site variants where calculated from the glucose peak areas using a standard curve (0.5 to 20
μM glucose).
Hydrolytic activity and chemical reactivation for glycosynthase templates and nucleophilic
variants
A. Hydrolytic activity using 4NPGlc
Initial velocities at the pH and temperature conditions used in the classical glycosynthase
reaction were determined for wt and TnBgl3B_W243F. Residual activity was also measured
on the single and double nucleophilic variants. Experiments were carried out at pH 5.6 and 7,
and at temperatures of 35C and 80C.
The volume of the total reaction was 1 mL with 2 mM 4NPGlc, in 50 mM citrate phosphate
buffer at the respective pH. The reaction started by adding 20 μl of the enzyme (0.01 to 0.5
μM) to a 4NPGlc substrate that was pre-incubated at 35C and 80C, respectively. Extinction
coefficients were measured at λ=400 nm and were calculated to be 1898 M-1cm−1, 3789 M-
19

1cm−1, 14905 M-1cm−1, and 15203 M-1cm−1 under the reaction conditions of pH 5.6 at 35C,
pH 5.6 at 80C, pH 7 at 35C and pH 7 at 80C respectively.
B. Chemical reactivation using sodium azide
Chemical reactivation experiments using sodium azide as an external nucleophile (from 0.5
to
3
M) were carried out for doubly mutated (TnBgl3B_D242G_W243F,
TnBgl3B_D242A_W243F, TnBgl3B_D242_W243F) variants. These double mutants were
first evaluated by 96 well plate assay at 35C for 30 min (Thermo Scientific Multiskan GO
UV/Vis Microplate) using different 4NPGlc concentrations (from 0.2 to 4 mM) in 50 mM
citrate phosphate buffer at pH 5.6. The total assay volume was 250 μl and a fixed enzyme
concentration of 0.05 μM was used. Chemical reactivation reactions were measured at 80C
using 4NPGlc in 50 mM citrate phosphate buffer at pH 5.6, a final sodium azide
concentration of 0.5 M. These reactions for single and double mutants were
spectrophotometrically monitored at λ=400 nm after adding the enzyme (0.025 to 0.05 μM).
Glycosynthase activity of oligosaccharides
A. α-glucosyl fluoride
Glycosynthase activity reactions were run on single and double nucleophilic variants of the
TnBgl3B wt using a donor to acceptor ratio of 1:5 (1 mM α-glucosyl fluoride and 5 mM of
4NPGlc) in 50 mM citrate phosphate buffer, pH 7. The reaction was pre-incubated for 5 min
at 35C before adding the enzyme (enzyme concentration was varied between 0.5 to 2.5 μM).
The final volume of the total reaction was 300 μL. Reaction monitoring began as soon as the
enzyme was added by taking an aliquot of 10 μL, and aliquots continued to be withdrawn
every 15 minutes for a total duration of 16 hours. The samples were diluted with deionized
water (1:10) before being injected into the HPLC.
20

B. Exogenous nucleophile
Glycosynthase activity reactions were run on single and double nucleophile variants from
TnBgl3B, using 6 mM 4NPGlc in 50 mM citrate phosphate over a pH range of 3 to 7, and
different concentrations of the exogenous nucleophile (sodium formate) from 0.1 to 2 M. The
reaction was pre-incubated for 5 min at 80C before adding the enzyme (enzyme
concentration was varied between 0.5 to 4 μM), afterwards monitoring began by withdrawing
aliquots of 20 μL every 15 min for 7 h. The samples were diluted with deionized water (1:10
and 1:100) before HPLC and HPAEC-PAD injections. Blanks were also ran: blank 1,
reaction without enzyme to check spontaneous hydrolysis of the substrate with external
nucleophile and blank 2, reaction without external nucleophile to check possible spontaneous
transglucosylation reactions.
Monitoring enzymatic synthesis of oligosaccharides.
As described previously in Pozzo et al in 2014, HPLC was used to monitor oligosaccharide
formation and identify the type of linkages that were formed by using a sample/standard co-
injection strategy. The mobile phase used was 14% MeOH (v/v) in water, the flow rate was 1
mL/min, UV detection was carried out at λ=300 nm, and a C18 column (Novapak, 4 μm,
Waters) was used.
HPAEC-PAD was used to measure residual hydrolysis in the glucosynthase reaction and was
quantified based on the integration of glucose peak areas to a standard curve (0.5 to 10 μM)
using a Dionex ICS-5000, Ion Chromatograph system (Dionex Corporation, Sunnyvale, CA,
USA). The column was a CarboPac PA-200 anion-exchange column (3 x 150 mm and 3 x 30
mm CarboPac PA-200 guard column). Glucose was eluted by a gradient of eluents: constant
100 mM NaOH 0 to 20 min, Gradient 0-15 min of 0-34mM NaAc from 15 to 20 min constant
21

100 mM NaAc. Chromeleon 7 software was used to control the system and analyse the data
(Pozzo et al. 2014).
Reactions were incubated for 16 hours and were visualised by Thin Layer Chromatography
(TLC). Briefly, μL of each reaction were loaded onto a silica gel (Silica 60 F254, Merck,
Darmstadt, Germany) and the separation was carried out using a mobile phase composed of
ethyl acetate/methanol/water (7:2.5:1). The Plates were developed in sulphuric
acid/methanol/water (10:45:45) followed by baking at 120C.
Enzymatic glycosylation of flavonoids
Enzymatic glycosylation reactions were carried out for TnBgl3B_D242A_W243F (1 μM)
using 4NPGlc as the donor and flavonoid molecules as acceptors at a ratio of 1:3.
Glycosylation reactions occurred using 2 M sodium formate in 50 mM citrate phosphate at
pH 5.6 and 80C. The reaction was pre-incubated for 5 minutes at 80C before adding the
enzyme, and monitoring began by withdrawing aliquots of 10 μL every hour for a total
duration of 16 hours. The sample was diluted with methanol (1:10) before HPLC injection.
The blanks were as follows: 1) reaction without enzyme to check spontaneous hydrolysis of
the substrate with the external nucleophile and 2) reaction without external nucleophile to
check for possible spontaneous transglycosylation reactions.
The HPLC method used for monitoring the glycosidic formation and identification of
linkage-type by coinjection of standards was described previously by Pozzo et al. (2014).
Briefly, The HPLC system used was an UltiMate-3000® (Thermo Fisher, Germering,
Germany) HPLC system from Dionex, consisting of an on-line degasser, a quaternary solvent
pump, an auto-sampler and photo diode array detector (DAD), all controlled by Chromeleon
6.8 software (Thermo Fisher). The detection wavelengths utilized were 280, 300 and 350 nm
and the analytical column used was an Ascentis Express C18 (150 mm × 2.1mm, 2.7 μm)
22

from Supelco (Bellefonte, PA, USA). The mobile phases consisted of (A) water with 0.5 %
(v/v) formic acid (B) methanol with 0.5 % (v/v) formic acid in a gradient elution analysis
programmed as follows: 0 min, 5 % (B); 0-5 min, 5% (B); 5-35 min, 40 % (B); 35-40 min, 40
% (B); at a flow rate of 300 μL/min. The injection volume was 2 μL and the column
temperature was 50°C.
Docking and molecular dynamics simulation
Structural studies of the TnBgl3B improved the understanding of donor and substrate
accommodation at the active site. Docking experiments for the GH3 enzyme were prepared
using the crystal structure of TnBgl3B_D 4 A in complex with α-D-glucose (PDB code
2X42) reported by Pozzo et al. (2010), where the α-D-glucose was substituted for an α-
glucosyl fluoride at the -1 subsite in the active site. In this case α-glucosyl fluoride, 4NPGlc
and flavonoid molecules (quercetin-3-glucoside, quercetin-3,4´-diglucoside) were obtained
using the server PRODRG (Schüttelkopf et al., 2004). The best docking results for the
complex to reach transglucosylation was achieved when the distance between O3 of 4NPCel
or 4´OH from the flavonoid the anomeric carbon of α-glucosyl fluoride was 5 Å.
Molecular dynamics simulations of TnBgl3B_D242A were analysed using ElNémo server
(Suhre et al. 2004) that showed the movement of the second domain close to the entrance of
the active site (domain 1) with its long flexible loop. Therefore, reconstruction of TnBgl3B’s
unresolved loop was done, modelling the crystal structure 2X42 residue 418 to 422, using
Modeller v. 9.8 (Eswar et al. 2006), keeping the backbone and the lateral chains fixed. Five
different loops with five different refinements each where modelled, resulting in a total of 25
structures. Energy minimization using GROMAC (Hess et al. 2008) reduced the initial
energy by a factor of 10 after 2.5 ns. All docking results were visualized and figures were
created using VMD (Humphrey et al. 2006).
23

Supplementary data
Supplementary data is available.
Conflict of interest
None declared
Funding
We wish to acknowledge the financial support from the Swedish research council Formas
(supporting SuReTech, grant 229-2009-1527) and from EU FP7 (supporting AMYLOMICS,
grant 265992) (to ENK), and MINECO, Spain for grant BIO2013-49022-C2-1-R (to AP).
Geneco Research School is highly appreciated for the travel grant that supported Tania Pozzo
as an exchange researcher at the University of Ramon Llull in Barcelona, Spain.
Abbreviations
C4, Cellotetraose; C5, Cellopentaose; GlcF, α-glucosyl fluoride; GH, glycoside hydrolase;
HPAEC-PAD, High-Performance Anion-Exchange Chromatography with Pulsed
Amperometric Detection; IMAC, immobilized metal ion affinity chromatography; 4NPGlc,
4-nitrophenyl β-D-glucosylpyranoside; 4NPCel, 4-nitrophenyl β-D-cellobioside; 4NPLam, 4-
Nitrophenyl-β-1,3-laminaribiose; Q, Quercetin; Q3Glc, Quercetin 3-O-β-glucopyranoside;
Q3Gal, Quercetin-3-O-β-galactoside Q3,4’DiGlc, Quercetin 3,4’-di-O-β-glucopyranoside;
Tn, Thermotoga neapolitana; TnBgl3B, β-glucosidase from GH family 3 and Thermotoga
neapolitana; wt, wild type.
24

References
Bernhardt P, O’Connor SE. 2009. Opportunities for enzyme engineering in natural product
biosynthesis. Curr. Opin. Chem. Biol. 13: 35-42.
Cobucci-Ponzano B, Strazzulli A, Rossi M, Moracci M. 2011. Glycosynthases in
biocatalysis. Adv. Synth. Catal. 353: 2284-2300.
Eswar NMA, Marti-Renom B, Webb MS, Madhusudhan D, Eramian M, Shen U, Pieper A,
Sali. 2006. Comparative Protein Structure Modelling With MODELLER. Curr.
Protocols Bioinformatics. 65: 712-725.
Faijes M, Sauravalls M, Perez X, Conti M, Planas A. 2006. Acceptor-dependent
regioselectivity of glycosynthase reactions by Streptomyces E383A β-glucosidase.
Carbohydr. Res. 341: 2055-2065.
Hess BC, Kutzner DV, VanDerSpoel E, Lindahl J. 2008. "GROMACS 4: Algorithms for
highly efficient, load-balanced, and scalable molecular simulation." J. Chem Theory
Comput. 4: 435-447.
Hrmova M, De Gori R, Smith BJ, Fairweather J.K, Driguez H, Varghese JN, Fincher GB.
00 . Structural basis for broad substrate specificity in higher plant β-D-glucan
glucohydrolases. Plant Cell. 14: 1033-1052.
Hrmova M, De Gori R, Smith BJ, Vasella A, Varghese JN, Fincher GB. 2004. Three-
dimensional structure of the barley β-D-glucan glucohydrolase in complex with a
transition state mimic. J. Biol. Chem. 279: 4980-4980.
Humphrey W, Dalke A. and Schulten K. 1996. "VMD - Visual Molecular Dynamics", J.
Molec. Graphics. 14: 33-38.
25

Jakeman DL. Sadeghi-Khomami A. 2011. A β-(1,2)-Glycosynthase and an Attempted
Selection Method for the Directed Evolution of Glycosynthases. Biochemistry. 50:
10359-10366.
Lee KW, Lee HJ. 2006. The roles of polyphenols in cancer chemoprevention. Biofactors.
26:105-121.
Mackenzie LF, Wang Q, Warren RAJ, Withers SG. 1998. Glycosynthases: Mutant
Glycosidases for Oligosaccharide Synthesis. J. Am. Chem. Soc. 120: 5583-5584.
Malet C, Planas A. 1 8. From β-glucanase to β-glucansynthase: glycosyl transfer to α-
glycosyl fluorides catalyzed by a mutant endoglucanase lacking its catalytic
nucleophile. FEBS Letters. 440: 208-212.
Moracci M. Trincone A, Perugino G, Ciaramella M, Rossi M. 1998. Restoration of the
Activity of Active-Site Mutants of the Hyperthermophilic β-Glycosidase from
Sulfolobus solfataricus: Dependence of the Mechanism on the Action of External
Nucleophiles. Biochemistry. 37: 17262-17280.
Perez X, Faijes M, Planas A. 2011. Artificial mixed-lin ed β-glucans produced by
glycosynthase-catalyzed polymerization: tuning morphology and degree of
polymerization. Biomacromol. 12: 494-501.
Plaza M, Pozzo T, Liu J, Ara KZG, Turner C, Nordberg Karlsson E. 2014. Substituent effects
on in vitro antioxidizing properties, stability and solubility in flavonoids. J.Agr.Food
Chem. 62: 3321-3333.
Pozzo T., Pasten JL, Nordberg Karlsson E, Logan DT. 2010. Structural and Functional
Analyses of β-Glucosidase 3B from Thermotoga neapolitana: A Thermostable Three-
Domain Representative of Glycoside Hydrolase 3. J. Mol. Biol. 397: 724-739.
26

Pozzo T, Plaza M, Romero-García J, Faijes M, Nordberg Karlsson E, Planas A. 2014.
Glycosynthases from Thermotoga neapolitana β-glucosidase 1A: A comparison of α-
glucosyl fluoride reactions with reactions using exogenous nucleophile. J. Mol. Catal.
B: Enzym. 107: 132-139.
Riva S. 2002. Enzymatic modification of the sugar moieties of natural glycosides. J. Mol.
Catal. B: Enzym. 19-20: 43-54.
Schüttelkopf AW, Van Aalten DM. 2004. PRODRG - a tool for high-throughput
crystallography of protein-ligand complexes. Acta Crystallogr. 60: 1355–1363.
Suhre K, Sanejouand YH. 2004. ElNemo: a normal mode web server for protein movement
analysis and the generation of templates for molecular replacement. Nucleic Acids
Research. 32: 610–614.
Turner, P, Svensson D, Adlercreutz, P, and Nordberg Karlsson E. 2007. A novel variant of
Thermotoga neapolitana B-glucosidase B is an efficient catalyst for the synthesis of
alkyl glucoside by transglycosylation. J Biotechnol.130: 67-74.
Viskupicová J, Ondrejovic M, Sturdík E. 2009. The potential and practical applications of
acylated flavonoids. Pharmazie. 64: 355-60.
Wang A, Zhang F, Huang L, Yin X, Li H, Wang Q, Zeng Z, Xie T. 2010. New progress in
biocatalysis and biotransformation of flavonoids. J. Med. Plant Res. 4: 847-856.
Williams SJ, Withers SG. 2000. Glycosyl fluorides in enzymatic reactions. Carbohydr. Res.
327: 27-46.
Xiao M, Shao Y, Yan W, Zhang Z. 2011. Measurement and correlation of solubilities of
apigenin and apigenin 7-O-rhamnosylglucoside in seven solvents at different
temperatures. J. Chem. Thermodyn., 43, 240–243.
27

Yang M, Davies GJ, Davis BG. 2007. A Glycosynthase Catalyst for the Synthesis of
Flavonoid Glycosides. Ange. Chem. Int. Ed. 46: 3885-3888.
Zackari A, Withers SG. 2013. Synthesis of Glycans and Glycopolymers Through Engineered
Enzymes. Bioplymers. 99: 666–674
28

FIGURES
Figure 1. Two mechanisms for achieving the glycosynthase reaction with TnBgl3B
nucleophilic variants.
(A) The “Classical Glycosynthase” reaction involved an α-glucosyl fluoride donor and
4NPGlc as the acceptor for the production of a β-1,3-disaccharide. The experimental
conditions for this reaction were pH 7 and a temperature of 30°C. (B) The “Glycosynthase-
type Reaction” utilized in situ generation of the glycosyl donor by using formate as an
exogenous nucleophile and 4NPGlc as both the donor and acceptor molecule. This reaction
occurred at a pH of 5.6 and a temperature of 80°C.
Figure 2. Close up views of the TnBgl3B active site.
(A) This panel displays modelling that was performed using the α-glucosyl fluoride donor
and the 4NPLam acceptor at the TnBgl3B_D242A active site (PDB code 2X42). D242 is the
catalytic nucleophile that was mutated to A242 in the rational design for the introduction of
glycosynthase activity. The additional residues that were selected for mutagenesis include
W243, Y210 and N248. The catalytic acid/base E458 (pink) has been shown for clarity. (B)
A surface representation that expands upon the modelling depicted in panel 2A was carried
out to emphasize the wt (PDB code 2X41) hinge, where Y210 (green), D242 (red) and W243
(orange) form a “wall” and the α-glucosyl fluoride is both stacked and stabilized by
hydrophobic interactions with the W243 residue.
Figure 3. Hydrolytic Activity of TnBgl3B and TnBgl3B variants on cello-oligosaccharide
substrates.
Cellotetraose (C4) and cellopentaose (C5) were hydrolysed by TnBgl3B and its active site
variants (1 μM). The initial velocities were determined under the reaction conditions of pH
29

5.6 and 80°C. Oligosaccharide substrate accommodation was lost in the W243F mutant, and
initial velocities were observed to be low for both C4 and C5 oligosaccharides. The distal
N248R mutation reduced cello-oligosaccharide activity relative to the wild type, indicating
that the distal mutation also affected long-substrate accommodation to the active site of
TnBgl3B.
Figure 4. Recovery of hydrolytic activity for inactive TnBgl3B nucleophilic variants using an
external nucleophile.
Chemical rescue experiments were performed using 4NPGlc (1mM) as the substrate,
different concentrations of the exogenous nucleophile sodium azide (0.5 to 4 M), a pH of 5.6,
and high heat (80°C). Hydrolytic activity was observed using 0.5M sodium azide for two
glycosynthases (TnBgl3B_D242A_W243F and TnBgl3B_D242S_W243F).
Figure 5. Docking of the quercetin-3,4´-diglucoside at the active site of TnBgl3B_D242A.
(A) Key residues interacting with the flavonoid (yellow molecule) were identified. Different
glycines (G27 and G38 in orange residues) form a confronted hydrophobic patch where the
body of the flavonoid is stabilized. The 3’ Glc (blac residue) is stabilized by the W243F
variant (red). The modelled flexible loop 418-422 (green color) contains the Trp420 (purple)
that could push, with its movement, the flavonoid to the displayed position. (B) Surface
representation of the mutated W420 and the modelled loop 418-422 (yellow) is depicted at
the active site, along with the 4NPLam and GlcF ligands. W420 partially covers the hinge
and stabilizes the glycosynthase reaction. The large size, hydrophobicity and central position
of the W420 residue within the flexible loop forms a “lid” that partially covers the hinge,
stabilizes the acceptor ligand and produces a favourable synthetic reaction.
30