Solvents derived from glycerol modify classical regioselectivity in the enzymatic synthesis of disaccharides with Biolacta β-galactosidase

Article abstract of DOI:10.1039/c1gc15266a

Green solvents made from glycerol change the classical regioselectivity of Biolacta N^o 5 β-galactosidase, from β(1→4) to β(1→6) linkages when a 2 M concentration was used. In order to explain these results, the non-proteic compounds present in

Full text of DOI:10.1039/c1gc15266a

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
Green Chemistry
Cite this: Green Chem., 2011, 13, 2810
www.rsc.org/greenchem
PAPER
Solvents derived from glycerol modify classical regioselectivity in the
enzymatic synthesis of disaccharides with Biolacta b-galactosidase†
a
a
b
c
Mar ´ı a P e´ rez-S a´ nchez, Manuel Sandoval, Alvaro Cort e´ s-Cabrera, H e´ ctor Garc ´ı a-Mar ´ı n,
d
c
a,d
Jos e´ V. Sinisterra, Jos e´ I. Garc ´ı a and Mar ´ı a J. Hernaiz*
Received 10th March 2011, Accepted 25th May 2011
DOI: 10.1039/c1gc15266a
o
Green solvents made from glycerol change the classical regioselectivity of Biolacta N 5
b-galactosidase, from b(1→4) to b(1→6) linkages when a 2 M concentration was used. In order to
explain these results, the non-proteic compounds present in the Biolacta preparation were
separated by precipitation with ammonium sulfate and the remaining protein extract was used to
set reactions with appropriate organic solvents to find that the regioselectivity towards the b(1→6)
isomer is retained. According to proteomic analysis, a 98% homology between Streptococcus
pneumoniae and Biolacta b-galactosidase preparation was found. With these data, molecular
modelling was done which predicts a tridimensional interaction in the enzyme active site with the
donor (GlcNAc) and the water-solvent mixture which explains this phenomenon.
Introduction
and specificity of the enzyme. There are numerous examples
in the literature wherein solvents’ physical properties such as
dielectric constant, dipole moment and hydrophobicity are
related to various effects on enzyme activity, specificity and
In recent years, green chemistry has become an area of great in-
terest for developing more environmentally acceptable processes.
According to the Green Chemistry principles, biocatalysis
and solvents play an important role in green processes, since
1–3
6–9
enantioselectivity.
Nowadays, a new generation of solvents has been developed,
whose search has become an area of interest. Currently, water,
supercritical fluids (SCFs), fluorous solvents, and solvents from
4
they provide a valuable alternative to classic organic chemistry.
First, enzymes offer suitable tools for industrial reactions, which
can be carried out under milder conditions, without using
heavy metals and with a great control over chemo-, regio- and
stereoselectivity using appropriate enzymes. Second, there are
numerous potential advantages in employing enzymes in organic
solvents, such as increased solubility of nonpolar substrates,
shifting of thermodynamic equilibria to favor synthesis over
hydrolysis, decrease of water-dependent side reactions, enhanced
thermal stability of enzymes, and elimination of microbial
renewable sources are considered green solvents and have been
10,11
used in biocatalysis.
In that sense glycerol is a green solvent
that has received attention, especially because it is a by-product
of biodiesel production, and hence it can be obtained from
renewable sources. Glycerol is an environmentally friendly,
non-toxic, biodegradable solvent. It has already been used in
biocatalytic reactions, and its potential for these reactions has
12–14
been demostrated.
5
contamination, among others. On the other hand, the presence
of organic solvents in enzymatic reactions can alter the activity
On the other hand, oligosaccharides are involved in a wide
range of biological processes including bacterial and viral
infection, cancer metastasis, the blood-clotting cascade and
15–18
many other crucial intercellular recognition events.
As the
a
Department of Organic and Pharmaceutical Chemistry, Faculty of
understanding of these biological functions increases, the need
for practical synthetic procedures of oligosaccharides in large
Pharmacy, Complutense University of Madrid, Campus de Moncloa,
2
9
8040, Madrid, Spain. E-mail: mjhernai@farm.ucm.es; Fax: (+) 34
139 41822; Tel: (+) 34 91139 41821
quantities has become a major subject. Organic chemical
b
Unidad de Bioinform a´ tica. Centro de Biolog ´ı a Molecular “Severo
Ochoa” (CBMSO), CSIC, Universidad Aut o´ noma de Madrid (UAM),
C/Nicol a´ s Cabrera 1, 28049, Madrid, Spain
19–21
methods for obtaining them have been developed,
but
they involve several elaborate protection and deprotection
procedures. Glycosyl hydrolases (glycosidases) can be used to
synthesize oligosaccharides in a kinetically controlled reaction,
where a glycosyl donor is used to transfer its glycosyl residue
to a sugar acceptor present in the reaction medium. In spite
of the increased amount of work carried out with glycosyl
hydrolases, their main drawback is the lack of regioselectivity,
c
Instituto de Ciencia de Materiales de Arag o´ n, CSIC-Universidad de
Zaragoza, C/Pedro Cerbuna, 12, E-50009, Zaragoza, Spain
d
Servicio de Interacciones Moleculares, (Pz/Ram o´ n y Cajal s/n. 28040
Madrid) y Servicio de Biotransformaciones Industriales, (C/Santiago
Grisol ´ı a, 28760 Tres Cantos), Parque Cient ´ı fico de Madrid, Spain
†
1
Electronic supplementary information (ESI) available. See DOI:
0.1039/c1gc15266a
2
810 | Green Chem., 2011, 13, 2810–2817
This journal is © The Royal Society of Chemistry 2011

View Article Online
which limits their use for synthetic purposes. In an important
contribution to glycosidase-catalyzed oligosaccharide synthesis,
22
Usui et al. reported for the first time the use of a b-galactosidase
from Bacillus circulans (Biolacta) to synthesize some b-D-
(
1→4) galactosyl disaccharides bearing a GlcNAc residue at
the reducing end. Some b-D-(1→6) linkages were produced
as well, but to the best of our knowledge, it was the first
time that a preparative scale b-D-galactosyltransfer has been
shown to occur preferentially at the O-4 position using a
galactosyl hydrolase. Also it has been shown that the use of the
thioethyl glycoside of N-acetylglucosamine as the acceptor and
p-nitrophenyl b-D-galactoside as donor gave complete selectivity
23–26
for 1→4 transfer.
In this work, we present the first study
of a change of regioselectivity in the enzymatic synthesis of
disaccharides with Biolacta b-galactosidase when green solvents
derived from glycerol are used.
Results and discussion
The benchmark reaction used throughout the work has been
the transglycosilation reaction of N-acetyl glucosamine with
◦
p-nitrophenyl-b-D-galactopyranoside, catalyzed by Biolacta N
5
b-galactosidase (Scheme 1). This reaction can afford Gal-b-
Fig. 1 Structure of different solvents derived from glycerol employed
in the transglycosilation reaction with Biolacta b-galactosidase.
(
1→4)-GlcNAc as mayor product and Gal-b-(1→6)-GlcNAc as
minor product.
solvent set was selected, since hydrophobicity could be a key
factor in solvent-enzyme substrate interactions.
Also, dielectric constants and dipole moments cover a broad
range of values, as well as the Dimroth-Reichardt solva-
N
28
tochromic parameter, E . The latter is mainly a measure of the
T
hydrogen-bond donor ability of the solvent, a feature that could
also be of importance for solvent-substrate and/or solvent-
enzyme interactions. Some of the glycerol-derived solvents used,
namely those bearing fluorinated chains, display simultaneously
both high hydrophobicity and hydrogen-bond donor ability, a
rather unusual combination in conventional organic solvents.
Miscibility with water, another important issue, also depends
on solvent structure. Thus, solvents 1–3 are completely miscible
with water, 5 and 8 are able to dissolve an appreciable amount
of water. On the other hand, triethers (6 and 9) and highly fluo-
rinated alcohols (10 and 11) display a very limited compatibility
27
with water.
Concerning toxicity issues, there are some experimental data
pointing to glycerol-derived solvents as being low hazards. For
instance, toxicities of glycerol based solvents like 1, 8 and the
-
1
closely related 1,3-diethoxy-2-propanol (LD50 > 3000 mg kg ,
32
oral in mice) and 1,2,3-triethoxypropane (LD50 = 660 mg
Scheme 1 General scheme of transglycosilation reaction catalyzed by
Biolacta N 5 b-galactosidase.
-
1
32
◦
kg , intraperitoneal in mice) are very low, and in some cases
-
1
even lower than that of butanol (LD50 = 2680 mg kg , oral
in mice). Fluorinated solvents will probably be more toxic than
aliphatic ones, but lacking in experimental data for them it would
be reasonable to take the toxicities of compounds produced
Synthesis and toxicity issues of glycerol-derived solvents
Glycerol-based solvents were synthesized using the same proce-
in their probable decomposition, like trifluoroethanol (LD50 =
27
-1
33
dures previously described. The complete listing of the solvents
used in this work is given in Fig. 1. Some physico-chemical and
solvatochromic parameters of the solvents used in this work are
gathered in Table 1. As can be seen, the structural diversity of the
glycerol derivatives results in a wide range of hydrophobicities,
covering four orders of magnitude. Bearing this in mind, a
366 mg kg , oral in mice), trifluoroacetic acid (LDlo = 150 mg
-
1
3
3
kg , intraperitoneal in mice) or 2,2,2-trifluoroethyl ethyl ether
-
1
34
(LD50 = 5100 mg kg , intraperitoneal in rats), as referents of
the toxicity of these kinds of solvents.
Moreover using these glycerol derivatives allow to reduce
some other hazardous aspects. For instance, they have low
This journal is © The Royal Society of Chemistry 2011
Green Chem., 2011, 13, 2810–2817 | 2811

View Article Online
Table 1 Transglycosylation yields (%) obtained with commercial Biolacta b-galactosidase in presence of green solvents
Disaccharide
b-(1→4)
Disaccharide
b-(1→6)
a
a
a
Na
Medium
log P
m (D)
e
E
T
Gal
Buffer
Glycerol
TFE
0.00
1.85
—
2.64
—
35
9
84
14
73
15
23
13
—
—
—
17
—
21
—
—
17
9
-1.33
0.71
-0.60
0.14
0.27
1.14
1.42
1.71
1.93
2.07
2.48
2.56
3.77
b
c
d
27.1
13.0
7.0
0.90
0.61
0.48
0.44
0.59
0.70
0.55
0.46
0.45
0.15
0.70
0.07
—
85
77
87
93
100
21
83
92
71
42
—
1
2
3
4
5
6
7
8
9
3.00
2.40
2.36
3.36
4.10
4.20
2.30
2.40
2.20
3.94
3.90
—
—
—
—
—
79
—
—
—
57
100
6.6
12.0
14.6
13.7
5.2
5.6
4.7
1
1
0
1
10.4
8.6
a
b
c
d
Values from ref. 27, unless otherwise indicated. Value taken from ref. 30 Value taken from ref. 29. Value taken from ref. 31.
volatility, so the concentration in the air when using them is
lower than using other conventional solvents. Albeit toxicities
of these solvents are not well known at this moment, it is clear
they are not extremely toxic, meeting one of the key requirements
for green applications.
According to SDS-PAGE results C and S preparation gave
about 11 protein bands in the range of 250–37 kDa, but there are
three important bands at 230, 145 and 86 KDa that agree with
previous a publication reporting three isoforms in the Biolacta
35
b-galactosidase preparation.
First, the regioselectivity showed by this enzyme was in-
vestigated when the reaction was carried out in 50 mM
citrate/phosphate buffer at pH 5.0 (Table 1). Biolacta b-
galactosidase gave a mixture of regioisomers, the major b-(1→4)
linker (84% conversion) and minor product b-(1→6) linker
Enzymatic synthesis in presence of green solvents with
commercial and semipurified Biolacta b-galactosidase
◦
Powder of Biolacta N 5 contains about 11% proteins in its
(
17%).
composition. After the concentration process we achieved 71%
of proteins in the solid sample, which means we reduce non-
proteic components from the crude preparation. Commercial
Second, hydrolytic activity of this b-D-galactosidase was
measured in different concentrations of solvent 1. The best
result was achieved in 2 M concentration of that solvent (Fig. 3).
Therefore, transglycosylation reactions catalyzed by commercial
preparation of the Biolacta b-galactosidase were carried out
in buffer media and green solvent (2 M)-buffer mixture (see
experimental section). Results from these reactions are shown in
Table 1.
(
C) and semipurified (S) enzyme showed a similar specific
-
1
activity (4.0–4.2 U mg respectively). SDS-PAGE (Sodium
Dodecyl Sulfate PolyAcrylamide Gel Electrophoresis) showed
the same gel profile elution for the two preparations mentioned
above (Fig. 2).
Fig. 3 Determination of the optimal concentration of solvent 1 for
◦
the enzyme activity of Biolacta b-galactosidase at 30 C (followed by
hydrolysis of pNP-b-Gal).
The synthetic activity of the b-galactosidase was examined,
and a change in the regioselectivity of the reaction was con-
firmed. In presence of solvents 1, 4, 5, 8 and 10 the regioisomer
b-(1→6) was formed as the sole product, the best yields being
obtained in the presence of 4 (93%), 5 (100%) and 8 (92%),
Fig. 2 Electrophoresis gel (7.5%) with different samples obtained from
Biolacta N 5 (S, semipurified; C, commercial).
◦
2
812 | Green Chem., 2011, 13, 2810–2817
This journal is © The Royal Society of Chemistry 2011

View Article Online
3
9,40
Table 2 Transglycosylation yields (%) obtained with semipurified
Biolacta b-galactosidase in the presence of glycerol-derived solvents
lactose because its role in infection processes.
This feature is
also present in Biolacta crude extract and suggests that both
could indeed be the same enzyme. These data also support
the idea that protein conglomerates are present in Biolacta
preparation and they are cleaved by the denaturing conditions
of SDS-PAGE.
In order to come up with a rational explanation and to confirm
our hypothesis about the influence of the solvent properties in
reaction stereoselectivity in the glycosidic linkage formation,
molecular modeling studies were carried out.
A 3D b-galactosidase S. pneumoniae model was built and
the molecules implicated in the mechanism docked, then, MD
simulations were run in two different solvent conditions and the
results compared.
The active site was located using a combination of sequence
and structure alignment and blind docking. The enzyme model
revealed that the active center could be similar to E. coli lacZ.
A simple superimposition of the structure of a monomer and
the model yielded the identification of the catalytic pair (Glu-
Hydrolysis
(galactose)
Synthesis
b(1→ 4)
Synthesis
b(1→ 6)
Medium
Buffer
—
6
14
9
90
6
11
9
10
88
75
82
4
5
8
while with the rest of the solvents tested the main product
formed was b-(1→6) with yields ranging from 71% to 85%
and b-(1→4) as the minor product (13–23%). This change in
the regioselectivity had also been noticed in previous works by
our group in the reaction catalyzed by Biolacta b-galactosidase,
36
when N,N-dimethylamide-type solvents were employed.
In fact, the change of regioselectivity cannot be easily related
to the available physical parameters (dipole moment, dielectric
constant or log P). In principle, it could be due to either
solvent-enzyme or solvent-substrate interactions, as well as to
the presence of impurities in the enzyme.
In order to explain the changes in the regioselectivity observed
in reactions with some solvents and commercial enzyme, most
of the non-proteic substances present in Biolacta were removed
5
64 and Glu-645) and the chemical behavior of each of them.
Residue Glu-645 corresponds to Glu-537 in lacZ which is
41
accepted as the nucleophile, while Glu-564 was identified as the
acid catalyst. This fact is also consistent with the first docking
results.
It was observed that the compound tended to be located near
the Glu-645 residue, packed over the Trp-708 residue. This kind
of interaction has been deeply investigated and it is identified
as crucial for other galactosidases. Additional interactions with
residues His-450, His-484, Asn-563 and Glu-716 contribute to
stabilize the molecule as seen in Fig. 4.
After modeling the glycosyl-enzyme, GlcNAc was docked
in the active site. The best docking solution was used as a
starting point for MD simulations. Regarding the preferred
stereochemistry of b-galactosidases, distances between anomeric
carbon (C1) and, 4 and 6 hydroxyl groups (O4 and O6) of
the acceptor molecule, were shown to be valid indicators of
the carbon atom in the best position to be attacked and
correlated with the regiospecificity of various enzymes used in
by precipitation with (NH
4
)
2
SO . The semipurified Biolacta b-
4
galactosidase was used with the best green solvents, selected
accordingly with the previous results obtained with the com-
mercial enzyme. Table 2 shows the yields obtained in different
media for these reactions.
Using semipurified enzyme we observed that solvent 4 leads
to the best yields for Galb(1→6)GlcNAc product. This result
represents a very important regioselectivity change in compar-
ison with the natural behavior of this enzyme, which shows a
42
2
4–26
high tendency to give the Galb(1→ 4)GlcNAc product.
Using commercial or semipurified enzyme as the catalyst
in buffer medium usually gives rise to yields of about 80%
or 90% b(1→4) and 17 or 10% b(1→6) product (see Table
1
and 2). Finally, and taking into account that this reaction
was performed after elimination of non-proteic substances
present in Biolacta, one can conclude that those substances
were not related to regioselectivity changes observed during
transglycosylation reactions and could be related to molecular
interactions between enzyme, solvent and substrates.
43
the enzymatic synthesis of carbohydrates.
With this aim, the distances between the C1 and O4 and
O6 groups and between oxygen atoms of Glu-564 and protons
of each hydroxyl group were measured. Special attention was
paid for structural information in the active site, particularly
for differential binding mode between simulations that could
explain the change in the stereoselectivity. Average distances
are presented in Table 3 together with RMSD of alpha
carbons.
Molecular modeling studies
In order to analyze the proteic nature of the commercial Biolacta
b-galactosidase a spot from the SDS-PAGE (approximately
8
6 kDa) was selected according to the molecular mass expected
Although the docked solution showed a clear preference for
O6, not compatible with the experimental product, during the
MD simulation the distances from the anomeric carbon to
O4 and O6 changed at 700 ps, indicating a variation in the
configuration of the acceptor. Indeed, a different binding mode
can be observed, in which the acceptor is found deeper in the
active center, interacting mainly with Tyr-624 and Lys-664.
This change could be a consequence of a pushing effect of the
solvent (water) due to the hydrophobic nature of the acetyl chain,
which can be found interacting with Tyr-624. The distance C1–
O4 was found to be more favorable; nevertheless, O6 is still close
enough to allow both linkages as determined experimentally.
35
for the b-galactosidase for Biolacta crude extract and the mass
spectrometry analysis of the protein spots was done. Results
from the proteomic analysis showed 98% homology with Strep-
tococcus pneumoniae SP9-BS68 b-galactosidase, which posseses
a 247 kDa molecular mass. This information could explain the
synthetic behavior of Biolacta enzymes with high preference
to Gal-b(1→4)GlcNAc synthesis instead of the classical B.
37,38
circulans tendency toward b(1→3) linkages
that does not
occur in Biolacta preparation.
The S. pneumoniae b-galactosidase have proved to be more
selective for catabolic degradation of Gal-b(1→4)GlcNAc than
This journal is © The Royal Society of Chemistry 2011
Green Chem., 2011, 13, 2810–2817 | 2813

View Article Online
Table 3 Average distances (in A˚ ) between anomeric carbon (C1) and hydroxyl groups (O4 and O6) and RMSD of the simulation
-
-
a
RMSD C (A)
˚
C1–O6
C1–O4
COO –HO–C6
COO –HO–C4
a
a
a
a
Water
5.76
5.26
5.73
5.81
8.93
2.27
1.93
5
-Water
4.79
7.89
7.42
a
After change of position.
Fig. 4 Binding modes: Shallow mode in 5-water simulation on the left, deep mode in water simulation on the right.
In the case of solvent 5-water mixture, the initial docked
solution was found to be coincident with the experimental
product. The acceptor exhibited during the entire simulation a
clear preference for O6, supporting the formation of the b(1→6)
linkage. However, a key interaction present in this complex, the
residue Ser-648, is found to be interacting with the O4 hydroxyl
of the acceptor. This interaction is not present in the water
simulation because of the deep binding mode of the molecule
and it could contribute to the stabilization of the complex. In
this case, the acetyl chain is not pushing the acceptor deeper.
Apparently, the interaction with the solvent is not as unfavorable
as in the case of water only, and the molecule adopts a shallow
binding mode with essentially the same interactions of the
deep binding mode (Tyr-624 and Lys-664) with the additional
interaction with Ser-648.
The distances measured during MD simulations are correlated
with the experimental results, in which the formation of b(1→6)
linkage is expected in the 5-water mixture, while the b(1→4)
linkage is the main product in water with some b(1→6) linked
molecules.
In the superimposition of both solutions (water and 5-water
mixture, Fig. 5) can be appreciated that O4wat and O65-wat are
located in the same position, at a good distance from C1 and the
catalytic Glu-564.
Fig. 5 Superimposition of the two binding modes: shallow (in green)
and deep (in yellow). Catalytic residues are shown in light blue.
Superimposition of O6 (shallow) and O4 (deep) can be observed.
The role of the acceptor in the stereochemical changes in
carbohydrate enzymatic synthesis is of great importance as it has
In our system, the interaction between the acetyl chain and the
solvent triggers a binding mode change in which the molecule
may reach a deep position in the active center, improving the
O4/O6 groups accessibility, and therefore, the stereochemistry.
The differences between solvents may be explained by the
different size and characteristics of the derivatives that may be
responsible for the interaction with the acceptor.
38
been previously reported. Particularly, some groups attached
to the acceptor could help to get the substrate molecule closer
to the catalytic residues and induce a change in the sugar
38
conformation. This is the case of the 2-acetamido group, which
greatly influences the stereoselectivity of the overall reaction.
2
814 | Green Chem., 2011, 13, 2810–2817
This journal is © The Royal Society of Chemistry 2011

View Article Online
Conclusions
One enzymatic unit was defined as the amount of protein that
hydrolyses 1 mmol of p-nitrophenyl-b-D-galactopyranoside per
minute under the conditions described before.
In this work we determined a change in classical regioselectivity
of Biolacta b-D-galactosidase when reaction is performed in the
presence of glycerol based solvents. These solvents promote the
b(1→6) linkages between the donor and acceptor instead of
b(1→4). This reactivity change is maintained even after the
Biolacta non protein components are removed, which means
the change is mainly related to enzyme-solvent interactions.
The proteomic analysis showed a high similarity between S.
pneumoniae b-D-galactosidase and one of the Biolacta proteins,
which could explain the tendency toward b(1→4) linkages
present in both enzymes. Using S. pneumoniae protein structure
as a basis for understanding the solvent-enzyme interaction,
we undertook a molecular modeling study and found that the
three-dimensional arrangement between GlcNAc and the water-
solvent (5) mixture in the active site of the enzyme, favors the
b(1→6) linkage. This disposition is also stabilized by the OH
interaction between Ser-648 and O4 from GlcNAc which also
explain the reduced proportion of b(1→4) linkage product.
Semipurification of enzymes
Semipurified Biolacta b-galactosidase was obtained as a pre-
cipitate by centrifugation after the addition of ammonium
◦
sulfate, (55% saturation at 0 C) to a solution of 5% m/v from
commercial enzyme in citrate/phosphate buffer 50 mM, pH 5.0,.
The pellet was resuspended in citrate/phosphate buffer 50 mM,
pH 5.0.
Proteomic Analysis
In order to identify proteic nature of the b-galactosidases
present in the commercial Biolacta preparation, a spot from
46
the SDS-PAGE (7,5%) was selected according to the molec-
ular relative mass for this enzyme described by Vetere and
42
Paoleti. A sample was sent to the Proteomic Unit Services
at Universidad Complutense. The protein was identified by
peptide mass fingerprints and MS/MS sequencing analyses.
Homology search of the sequences was obtained by Blast
Experimental
(
http://www.ncbi.nlm.nih.gov/BLAST).
Materials
p-Nitrophenol (pNP), p-nitrophenyl-b-D-galactopyranoside
Glycerol-derived solvents
(
pNP-b-Gal), N-acetyl-D-glucosamine (GlcNAc), Gal-
D-(+)-galactose,
Glycerol-based solvents were synthesized using the same proce-
b(1→4)GlcNAc,
Gal-b(1→6)GlcNAc,
27
dures previously described. The alcoholysis of either epichloro-
hydrin (symmetrical derivatives) or the appropriate glycidol
ether (unsymmetrical derivatives), gave the corresponding 1,3-
dialkoxy-2-propanols. O-Methylation of some selected targets
leads to the corresponding 1,2,3-trialkoxypropanes. All the
solvents were purified by distillation and used in high pure form.
The complete listing of the solvents used in this work is given in
Fig. 1.
ammonium sulfate and bovine serum albumin (BSA), were
◦
purchased from Sigma Aldrich. Biolacta N 5 b-galactosidase
from B. circulans was received from Daiwa Kasei, Japan. Dye
reagent for protein determination was purchased from Bio Rad.
All other chemicals were from analytical grade.
UV-visible spectra were recorded on a UV-2401 PC Shimadzu.
HPLC Agilent 1100 with UV-vis detector using Mediter-
raneasea18 15 cm ¥ 0.46 5 mm column (Teknokroma) with
water : acetonitrile (75 : 25) as a mobile phase at a flow of 0.7 mL
Effect of solvent molarity on enzyme activity
-
1
min . HPLC Jasco with an evaporative light scattering detector
ELSD) using NH P50-4E amino column (Asahipak, Japan)
with acetonitrile : water (80 : 20) as a mobile phase at a flow
(
2
In order to determine the molarity of green solvent required
for the transglycosylation reactions we measured the hydrolytic
rates of the crude enzyme using solvent 1 in the following
concentrations: 2, 4, 6 and 8 M. The hydrolysis measurements
were performed by quantification of the p-nitrophenyl-b-D-
galactopiranoside by UV-visible detector on HPLC.
-
1
of 1.0 mL min . NMR spectra were recorded on Bruker AV
00 MHz spectrometers. The structures of the enzymatically
5
1
13
synthesized disaccharides were assigned by H, C and 2-D
NMR (see supporting information for details†).
Enzyme assays
Transglycosylation reactions
Hydrolytic activity was determined by spectrophotometri-
cal quantification of pNP liberated from pNP-b-Gal in cit-
rate/phosphate buffer 50 mM, pH 5.0, using both continuous
and discontinuous methods. Continuous method: hydrolysis of
pNP-b-Gal (5 mM) in a 1 mL cell by measuring the increase in
Transglycosylation reactions were carried out using commercial
preparations and semipurified enzymes. To begin with, solvents
from 1 to 11 were employed with a commercial crude extract
of the enzyme. Then, the best solvents were chosen according
to best performance and used to carry out reactions with
semipurified enzymes in order to determine the effect of non-
proteic substances present in Biolacta preparation in reaction
yields.
◦
44
absorbance at 410 nm during 2.5 min at 37 C. Discontinuous
method: hydrolysis of 200 mL of pNP-b-Gal (1.0 mM) for 10 min
◦
at 37 C and then reaction was stopped by adding 1.00 mL of
44,36
Na
2
CO
3
0.20 M.
Each experimental assay was determined
A 0.17 M solution pNP-b-Gal (donor) 0.85 M of GlcNAc
(acceptor) in 1 mL of green solvent (2 M)-buffer mixture
at least three times with a standard deviation under 5% of
the average of samples. Protein concentration was determined
◦
-1
was pre-equilibrated to 30 C. Afterwards, 155 mmol min
45
by Bradford method following the manufacturer’s protocol
Bio-Rad) and with bovine serum albumin as the standard.
(U) of Biolacta b-galactosidase were added to the reaction
mixture. Reaction was monitored by HPLC UV-visible, and final
(
This journal is © The Royal Society of Chemistry 2011
Green Chem., 2011, 13, 2810–2817 | 2815

View Article Online
4
1
products obtained were analyzed by HPLC-ELSD. The reaction
was stopped by heating to 100 C for 5 min.
(nucleophile) in the lacZ of E. coli. The structure of the chain
A of lacZ (PDB ID: 1PX3) was superimposed on the model
using PyMOL. A very good superimposition of the analyzed
residues could be observed depicting a typical active center of b-
galactosidase in which the two glutamic residues were placed at
◦
The reaction mixture was then directly loaded onto a carbon-
celite column. The column was first eluted with water (200 mL)
and then with a linear gradient of 0% to 15% (v/v) of ethanol.
Solvents were eliminated and disaccharides were dissolved in
˚
a distance of 5.5 A, the usual distance for this class of enzymes.
1
13
D
2
O to be characterized by H and C NMR spectroscopy.
Finally, a blind docking using the whole structure as the grid
was performed, and the results confirmed the hypothetic active
site.
Molecular modeling studies
4
3
Upon information available, molecular modeling studies were
carried out to explain interactions between glycerol-based
solvents and substrates as well as solvent-enzyme interactions.
Molecular docking. As in previous work a two dockings
approach was followed. A p-nitrophenol-b-galactopiranose was
initially docked into the active center to determine to the position
of the molecule to simulate the glycosyl-enzyme intermediate.
With this aim, we selected the best pose according the geomet-
rical (close enough to the nucleophile Glu-645) and energetic
criteria (favorable energy) and set up a covalent link between the
Glu-645 residue and the galactose.
Homology modeling. According to the results obtained with
the peptide mass fingerprint, Streptococcus pneumoniae b-
galactosidase was selected as the starting point for the molecular
modeling studies. The 3D model of the domain with catalytic
activity of this b-galactosidase (residues 150–979) was built by
homology modeling based on available crystal structures of
homologous proteins. The complete amino acid sequence of
the target protein was retrieved from NCBI protein sequence
database (Accession No. gi|148992245) in FASTA format. The
Regarding the second docking, the glycosyl-enzyme was used
to perform a docking with the N-acetyl-glucosamine (GlcNAc),
the best solution was selected taking into account energetic
criteria only, and it was taken as a starting structure for the
subsequent molecular dynamics studies.
47
NCBI Basic Local Alignment Search Tool (BLAST), for
sequence similarities was used for searching the crystal structures
of the closest homologues available in the Brookhaven Protein
Data Bank (PDB). The results revealed b-galactosidase from
Bacteroides vulgatus (PDB ID: 3GM8), with a resolution of 2.4
All docking procedures were carried out with Autodock 4
software, using a 60 ¥ 60 ¥ 60 grid points box around Glu-645
residue, to embrace all residues implicated in the mechanism,
55
˚
and with a grid spacing of 0.375 A. A Lamarckian genetic
algorithm with the standard parameters was selected.
˚
A, was a suitable template (81% of homology in the selected
zone and E-value 2e-104).
Solvent parameterization. One of the experimental solvents,
namely 1,3-bis(2,2,2-trifluoroethoxy)propan-2-ol (5) was used
to solvate the protein and simulate the system in order to model
the change in the stereoselectivity induced by the solvent and
get insights into the molecular details of this process. Due to
the lack of parameters for 5 for the GROMOS 96 43a1 force
field, a parameterization of this molecule was carried out. The
structure was optimized within the density functional theory
(DFT) framework, at the B3LYP/6-31G** level, using the
The sequences of b-galactosidases from B. vulgatus, B. fragilis,
B. thetaiotaomicron and E. coli, which were the most similar to
that of S. pneumoniae, were used for a pairwise alignment em-
ploying the Multiple Sequence Comparison by Log-Expectation
48
(
MUSCLE) tool.
Homology modeling of the target was performed by two
49
automated homology modeling servers: SWISS-MODEL and
50
3
D-JIGSAW using as template the protein 3GM8. The first was
56
found to be the best model, and was selected after performing a
the 1000 steps of steepest descent (SD) energy minimization
using the GROMOS 96 force field and GROMACS 4.0.3
Massively Parallel Quantum Chemistry package (MPQC). All
the parameters (bond lengths, angles and dihedrals) and charges
were extracted from the calculations or from the standard force
51
57
package, to regularize the protein structure geometry.
field.
The quality of the model was analyzed with the
To check the derived parameters, a test system was set up
formed by a cubic box of 216 molecules of 5. This box was
equilibrated 300 ps at 298 K and 1 atm. Following, a production
simulation of 5 ns was carried out at the same NpT conditions
recording position and energy values every 1 ps. Finally, density
52
PROCHECK program to calculate the main-chain torsional
angles (Ramachandran plot) and it was verified using the on-
53
54
line software ERRAT and VERIFY_3D at the UCLA-DOE
Structure Analysis and Verification Server (http://www.doe-
mbi.ucla.edu/Services/SV/). Most of the analyzed parameters
had statistical values that were in the range of those expected
for a naturally folded protein and therefore, the model was
considered validated.
58
was calculated and compared with the experimental values,
-
3 29
confirming a good agreement (rexp = 1330 Kg m ;
r
calc
=
-
3
1360 Kg m ).
Molecular dynamics. Molecular dynamics simulations were
set up with the best solution of docking and the glycosyl-
enzyme intermediate. Parameters adopted in GROMOS 96
43a1 were used for the protein, while the parameters re-
quired for the galactose-glutamic residue were derived in a
consistent manner with the force field. GlcNAc parameters
were extracted from Dundee PRODRG 2.5 Server (beta)
(http://davapc1.bioch.dundee.ac.uk/cgi-bin/prodrg_beta) and
the charges were properly checked.
Active site search. The absence of a crystal structure of the
complex forced us to deduce the location of the active center by
means of conservation analysis of residues, blind docking and
structure superimposition.
A comparative analysis of the catalytic residues of resolved
structures (lacZ E. coli) yielded, as a result, a perfect con-
servation of two glutamic acid residues, Glu-564 and Glu-
6
45, corresponding with residues 461 (acid catalyst) and 537
2
816 | Green Chem., 2011, 13, 2810–2817
This journal is © The Royal Society of Chemistry 2011

View Article Online
An initial minimization of the complex was performed with
00 steps of SD followed by 1500 steps of Polak-Ribiere Con-
19 R. R. Schmidt and E. R u¨ cker, Tetrahedron Lett., 1980, 21, 1421–
1
424.
5
2
2
0 R. R. Schmidt, Angew. Chem., Int. Ed. Engl., 1986, 25, 212–235.
jugate Gradients (CG). The minimized complex was solvated
in two different cubic boxes, fulfilling the requirement of a
minimum distance of 1.2 nm between any atom of the complex
and the faces of the box. The first box was filled with SPC
water molecules to reproduce the usual experimental conditions
for the enzyme, while the second one was filled with a mixture
of SPC water molecules and 5 molecules trying to reproduce
the conditions for the stereochemistry change presented in this
work. Then, for each box, a new minimization procedure of 500
steps of SD and 3000 steps of CG was followed. The systems
were equilibrated for 100 ps at 298 K (NVT conditions) and
another 100 ps at 298 K and 1 atm (NPT conditions). Finally, 1
ns production simulations were carried out.
1 H. Paulsen, Angew. Chem., Int. Ed. Engl., 1982, 21, 155–173.
22 T. Usui, S. Kubota and H. Ohi, Carbohydr. Res., 1983, 244, 315–323.
23 M. J. Hernaiz and D. H. G. Crout, J. Mol. Catal. B: Enzym., 2000,
1
0, 403–408.
4 A. Vetere and S. Paoletti, Biochem. Biophys. Res. Commun., 1996,
19, 6–13.
2
2
25 K. Sakai, R. Katsumi, H. Ohi, T. Usui and Y. Ishido, J. Carbohydr.
Chem., 1992, 11, 553–565.
2
2
2
2
3
3
3
6 T. Usui, S. Morimoto, Y. Hayakawa, M. Kawaguchi, T. Murata, Y.
Matahira and Y. Nishida, Carbohydr. Res., 1996, 285, 29–39.
7 J. I. Garc ´ı a, H. Garc ´ı a-Mar ´ı n, J. A. Mayoral and P. P e´ rez, Green
Chem., 2010, 12, 426–434.
8 C. Reichardt, Solvents and Solvent Effects in Organic Chemistry,
Wiley-VCH, Weinheim, 2003.
9 D.-P. Hong, M. Hoshino, R. Kuboi and Y. Goto, J. Am. Chem. Soc.,
1999, 121, 8427–8433.
0 A. M. P. Mainar, J., J. I. Garc ´ı a, F. M. Royo and J. S. Urieta, J. Chem.
Soc., Faraday Trans., 1998, 94, 3595–3599.
Acknowledgements
1 C. Laurence, P. Nicolet and C. Reichardt, Bull. Soc. Chim. Fr., 1987,
1
001–1005.
This work was supported by two research projects of the
Spanish MICINN (Ministerio de Ciencia e Innovaci o´ n de
Espa n˜ a) CTQ2009-11801 and CTQ2008-05138, and one Eu-
ropean project (FP-62003-NMP-SMF-3, proposal 011774-2).
Manuel Sandoval was supported by a Ph.D fellowship granted
from the Universidad Nacional de Costa Rica.
2 J. Wagner and H. Grill, ed. US. Pat. 3888994., U.S.A, 1975.
33 D. A. Blake, H. F. Coscorbi,bR. S. Rozman and F. J. Meyer, Toxicol.
Appl. Pharmacol., 1969, 15, 83.
3
4 M. J. Murphy, D. A. Dunbar and L. S. Kaminsky, Toxicol. Appl.
Pharmacol., 1983, 71, 84–92.
35 J. Bigorra, V. Fabry, M. Perez, J. V. Sinisterra and M. J. Hernaiz.
European Patent EP 08011388, 2008.
3
6 A. Vetere and S. Paoletti, Biochim. Biophys. Acta – General Subjects,
1
998, 1380, 223–231.
37 Y. Ito and T. Sasaki, Biosci., Biotechnol., Biochem., 1997, 61, 1270–
276.
Notes and references
1
1
P. T. W. Anastas, J. C., Green Chemistry: Theory and Practice, Oxford
University Press, New York, 1998.
38 X. X. Zeng, R. Yoshino, T. Murata, K. Ajisaka and T. Usui,
Carbohydr. Res., 2000, 325, 120–131.
39 D. Zahner and R. Hakenbeck, J. Bacteriol., 2000, 182, 5919–5921.
40 R. Zeleny, F. Altmann and W. Praznik, Anal. Biochem., 1997, 246,
96–101.
41 D. H. Juers, T. D. Heightman, A. Vasella, J. D. McCarter, L.
Mackenzie, S. G. Withers and B. W. Matthews, Biochemistry, 2001,
40, 14781–14794.
42 V. Spiwok, P. Lipovov a´ , T. Sk a´ lov a´ , E. Buchtelov a´ , J. Hasek and B.
Kr a´ lov a´ , Carbohydr. Res., 2004, 339, 2275–2280.
43 N. Br a´ s, P. Fernandes and M. Ramos, Theor. Chem. Acc., 2009, 122,
283–296.
2
3
4
P. T. Anastas, Chem. Rev., 2007, 107, 2167–2168.
I. T. Horvath and P. T. Anastas, Chem. Rev., 2007, 107, 2169–2173.
R. W. Kates, W. C. Clark, R. Corell, J. M. Hall, C. C. Jaeger, I.
Lowe, J. J. McCarthy, H. J. Schellnhuber, B. Bolin, N. M. Dickson,
S. Faucheux, G. C. Gallopin, A. Grubler, B. Huntley, J. Jager, N. S.
Jodha, R. E. Kasperson, A. Mabogunje, P. Matson, H. Mooney, B.
Moore, T. O’Riordan and U. Svedin, Science, 2001, 292, 641–642.
J. S. Dordick, Enzyme Microb. Technol., 1989, 11, 194–211.
S. V. Kamat, E. J. Beckman and A. J. Russell, J. Am. Chem. Soc.,
5
6
1
993, 115, 8845–8846.
7
8
K. Ryu and J. S. Dordick, Biochemistry, 1992, 31, 2588–2598.
44 L. Fourage, M. Helbert, P. Nicolet and B. Colas, Anal. Biochem.,
S. Tawaki and A. M. Klibanov, J. Am. Chem. Soc., 1992, 114, 1882–
1999, 270, 184–185.
1
884.
45 M. M. Bradford, Anal. Biochem., 1976, 72, 248–254.
46 U. K. Laemmli, Nature, 1970, 227, 680–685.
47 T. Tatusova and T. Madden, FEMS Microbiol. Lett., 1999, 174, 247–
250.
9
C. R. Wescott and A. M. Klibanov, J. Am. Chem. Soc., 1993, 115,
1
629–1631.
1
1
1
1
1
1
0 M. J. Hernaiz, A. R. Alc a´ ntara, J. I. Garc ´ı a and J. V. Sinisterra,
Chem.–Eur. J., 2010, 16, 9422–9437.
48 R. C. Edgar, Nucleic Acids Res., 2004, 32, 1792–1797.
49 T. Schwede, J. R. Kopp, N. Guex and M. C. Peitsch, Nucleic Acids
Res., 2003, 31, 3381–3385.
50 P. Bates, L. Kelley, R. MacCallum and M. Sternberg, Proteins:
Struct., Funct., Genet., 2001, 45, 39–46.
51 B. Hess, C. Kutzner, D. van der Spoel and E. Lindahl, J. Chem.
Theory Comput., 2008, 4, 435–447.
52 R. Laskowski, M. MacArthur, D. Moss and J. M. Thornton, J. Appl.
Crystallogr., 1993, 26, 283–291.
53 C. Colovos and T. Yeates, Protein Sci., 1993, 2, 1511–1519.
54 D. Eisenberg, R. L u¨ thy and J. Bowie, Methods in enzymology, 1997,
277, 396., 1997, 277–396.
55 G. M. Morris, R. Huey, W. Lindstrom, M. F. Sanner, R. K. Belew, D.
S. Goodsell and A. J. Olson, J. Comput. Chem., 2009, 30, 2785–2791.
56 C. Janssen, I. Nielsen, M. Leininger, E. Valeev and E. Seidl, Autodock
4.2, Scripps Research Institute, 2009.
57 R. Suardiaz, M. Maestre, E. Su a´ rez and C. P e´ rez, THEOCHEM,
2006, 778, 21–25.
58 C. Resende and G. F. L Prado, THEOCHEM, 2007, 847, 93–100.
1 M. S. P e´ rez, J. V. Sinisterra and M. J. Hern a´ iz, Curr. Org. Chem.,
2
010, 14, 2366–2383.
2 A. Wolfson, C. Dlugy and Y. Shotland, Environ. Chem. Lett., 2007,
5
, 67–71.
3 A. Wolfson, C. Dlugy, D. Tavor, J. Blumenfeld and Y. Shotland,
Tetrahedron: Asymmetry, 2006, 17, 2043–2045.
4 L. Andrade, L. Piovan and M. D. Pasquini, Tetrahedron: Asymmetry,
2
009, 20, 1521–1525.
5 M. E. C. Caines, H. Zhu, M. Vuckovic, L. M. Willis, S. G. Withers,
W. W. Wakarchuk and N. C. J. Strynadka, J. Biol. Chem., 2008, 283,
3
1279–31283.
1
6 H. Shirato, S. Ogawa, H. Ito, T. Sato, A. Kameyama, H. Narimatsu,
Z. Xiaofan, T. Miyamura, T. Wakita, K. Ishii and N. Takeda, J.
Virol., 2008, 82, 10756–10767.
1
1
7 G. F. Springer, Science, 1984, 224, 1198–1206.
8 G. F. Springer, P. R. Desai, W. Wise, S. C. Carlstedt, H. Tegt-
meyer, R. Stein and E. F. Scanlon, Immunol Ser, 1990, 53, 587–
6
12.
This journal is © The Royal Society of Chemistry 2011
Green Chem., 2011, 13, 2810–2817 | 2817