See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323671039

QM/MM Approach on the Structural and Stereolectronic Factors governing

Glycosylation by GTF-SI from Streptococcus mutans

Article  in  Organic & Biomolecular Chemistry · March 2018

DOI: 10.1039/C8OB00284C

CITATIONS READS

5 248

7 authors, including:

Gonzalo Jaña Fernanda Mendoza

Universidad Andrés Bello Universidad Andrés Bello
28 PUBLICATIONS   168 CITATIONS    20 PUBLICATIONS   82 CITATIONS   

SEE PROFILE SEE PROFILE

Manuel Osorio Joel Alderete

Universidad Andrés Bello Universidad de Talca
11 PUBLICATIONS   28 CITATIONS    128 PUBLICATIONS   1,626 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

UNVEILING THE FACTORS THAT DETERMINE THE ENZYMATIC C-GLYCOSIDIC BOND FORMATION: A QM/MM APPROACH View project

The structural and thermodynamic factors that determine 3 ', 4' or 6 'transglucosidase activity in the GTF-SI enzyme View project

All content following this page was uploaded by Gonzalo Jaña on 21 March 2018.

The user has requested enhancement of the downloaded file.

 Organic &
Biomolecular Chemistry
View Article Online
PAPER View Journal

A QM/MM approach on the structural and

Published on 09 March 2018. Downloaded by Universidad de Concepcion on 20/03/2018 13:42:30.

Cite this: DOI: 10.1039/c8ob00284c

stereoelectronic factors governing glycosylation
by GTF-SI from Streptococcus mutans†
Gonzalo A. Jaña, *a Fernanda Mendoza, a Manuel I. Osorio, a,b Joel B. Alderete, c

Pedro A. Fernandes, d Maria J. Ramos d and Verónica A. Jiménez a

Received 2nd February 2018,
Accepted 9th March 2018
DOI: 10.1039/c8ob00284c
rsc.li/obc
In this work, QM/MM calculations were employed to examine the catalytic mechanism of the retaining
glucosyltransferase GTF-SI enzyme, which participates in the process of caries formation. Our goal was to
characterize, with atomistic details, the mechanism of sucrose hydrolysis and the catalytic factors that
modulate this reaction. Our results suggest a concerted mechanism for sucrose hydrolysis in which the
first event corresponds to the glycosidic bond breakage assisted by Glu515, followed by the nucleophilic
attack of Asp477, leading to the formation of the Covalent Glycosyl Enzyme (CGE) intermediate. A novel
conformational itinerary of the glucosyl moiety along the reaction mechanism was identified: 2H3 → 2H3–
E3 → 4C1, and the calculated energy barrier is 16.4 kcal mol−1, which is in good agreement with experi-
mental evidence showing a major contribution coming from the glycosidic bond breakage. Our calculations
also revealed that Arg475 and Asp588 play a critical role as TS-stabilizers by electrostatic and charge transfer
mechanisms, respectively. This is the first report dealing with the specific features of the mechanism and
catalytic residues involved in GTF-SI hydrolysis of sucrose, which is a matter of relevance in enzyme catalysis
and could be valuable to aid the design of novel and specific inhibitors targeting GTF-SI.

1. Introduction Glucans are synthesized by 1,3-1,6-α-glucansucrases (also

Dental caries are a major public health problem in most
industrialized countries, affecting 60–90% of school-aged chil-
dren and the vast majority of adults as a consequence of
growing sugar consumption.1–3 This infection relies on the
biosynthesis of an extracellular polymeric film that shelters
microorganisms, and allows their adhesion to biological sur-
faces, thus enabling bacterial growth.4 Glucans are the main
polymeric components of this biofilm that play a key role in
the cariogenesis process, for which they have gained impor-
tance in the search for new treatments to prevent dental
caries.5
a
Departamento de CienciasQuímicas, Facultad de Ciencias Exactas,
Universidad Andres Bello, Sede Concepción, Autopista Concepción-Talcahuano 7100,
Talcahuano, Chile. E-mail: gonzalo.jana@unab.cl
b encoding genes of S. mutans markedly reduced the plaque
Doctorado en Fisicoquímica molecular, Universidad Andres Bello, Republica 275,
Santiago, Chile
c
Departamento de Química Orgánica, Facultad de Ciencias Químicas,
Universidad de Concepción, Casilla 160-C, Concepción, Chile
d
UCIBIO, REQUIMTE, Departamento de Química e Bioquímica, Faculdade de
Ciências, Universidade do Porto, Rua do Campo Alegre, s/n, 4169-007 Porto,
Portugal
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
c8ob00284c
known by the acronyms of GSs or GTFs for glucansucrases and
glucosyltransferases, respectively) which are produced by oral
bacteria streptococci, like Streptococcus mutans or Streptococcus
sobrinus. GTFs are members of the Glycoside Hydrolase (GH)
family 70 and catalyze the formation of glycosidic bonds
between glucose units, producing different α-glucans from
sucrose through transglucosylation reactions.5,6 GTFs are
classified into GTF-I, GTF-SI and GTF-S. GTF-S is a dextransu-
crase that synthesizes predominantly soluble glucans with
α(1 → 6) linkages, whereas GTF-I and GTF-SI are mutansu-
crases that synthesize mainly insoluble glucans with α(1 → 3)
glycosidic linkages (mutans). As the synthesis of mutans aids
Streptococcus mutans (S. mutans) in adhering to the tooth
surface, the inhibition of GTF-SI or GTF-I activities might
impair the cariogenic virulence of tooth plaque.7 Recent
experimental studies have shown that inactivating two GTF
formation on tooth surfaces in vitro, which suggests that mole-
cules that were able to inhibit GTFs could be potential anti-
caries agents. Nowadays, there are natural and synthetic mole-
cules whose inhibitory activity results in poor biofilm for-
mation; however, it is not known how these molecules target
GTFs from S. mutans. Although many of the available inhibi-
tors are highly effective in vitro, it is still unclear whether they

This journal is © The Royal Society of Chemistry 2018 Org. Biomol. Chem.

Paper
are suitable for in vivo use, and their mechanisms remain
largely unknown. Furthermore, the consequences for the
human metabolic GHs (like α-amylases) or the health-associ-
ated microbiota are also unclear.7
Even though the catalytic mechanism is known for several
members from different GH families,8–10 the specific details
about how GTF-SI from Streptococcus mutans catalyzes the for-
mation of α-glucans have never been addressed at a molecular
atomic level due to the lack of a crystalline structure. Thus, the
Published on 09 March 2018. Downloaded by Universidad de Concepcion on 20/03/2018 13:42:30.
catalytic mechanism of glucan synthesis has been proposed by
direct analogy with GH13 (α-amylases) and GH77 (amylomal-
tases) families (all of them belong to GH-H clan).11,12 These
enzymes, and therefore the GTFs,12–15 follow an α-retaining
double displacement mechanism that starts with the attack of
a nucleophilic residue (Asp) to the anomeric carbon of the glu-
cosyl moiety of sucrose concerted with a glutamic acid proto-
nation of the glycosidic oxygen of the fructosyl leaving group.
In this way, the covalent β-glucosyl enzyme intermediate (CGE)
is formed via an oxocarbenium ion-like transition state (TS),
this step being rate-limiting. In the second step, the leaving
fructose is exchanged by a new acceptor sucrose (R-OH, Fig. 1),
which gets deprotonated by a glutamate residue acting as a
base. The R-OH attacks the anomeric carbon of Asp-sucrose
(CGE) while the leaving group Asp departs. Thus, the final
α-glycosidic product is formed in a concerted step via another
oxocarbenium ion-like TS. Pinto et al.16 studied the catalytic
mechanism of human pancreatic α-amylase (HPA) from the
GH13 family by means of QM/MM calculations. These calcu-
lations enabled the analysis of atomic structural details related
to the transfer reaction, which led to the characterization of a
mechanism that turned out to be in good agreement with the
empirically proposed mechanism for α-amylase’s enzymes.
Although the general mechanism is already understood,
homologous catalytic residues within the GH13 (α-amylase)
and GH70 (GTF-SI) families do not necessarily imply identical
catalytic functions17 nor similar mechanisms.18 Additionally,
it is known that different GH-family enzymes catalyzing the
same kind of mechanism can show different conformational
itineraries of the glucosyl ring.19 Thus, the role of several
active site amino acids in the reaction catalyzed by GTFs from
Streptococcus mutans has been speculative and further evidence
must be provided in order to characterize its catalytic mecha-
nism and the conformational itinerary of the glucosyl ring. In
Fig. 1 General double displacement mechanism proposed for the synthesis of α-glucans by GTFs.
Org. Biomol. Chem.
View Article Online
Organic & Biomolecular Chemistry
addition, there are no computational studies considering
sucrose acting as the donor substrate, and therefore a detailed
understanding of the mechanistic and catalytic factors that
rule the mutan biosynthesis performed by either of these
enzymes is a key factor to identify specific inhibitors for GTF-I
or GTF-SI.
The recent elucidation of the crystal structure of the GTF-SI
enzyme (PDB ID: 3AIE) from Streptococcus mutans15 represents
a great opportunity for understanding the catalytic mechanism
involved in mutan synthesis, which could allow the identifi-
cation of structural and electronic changes that are relevant
during the catalysis. The X-ray structure possesses an appropri-
ate resolution (2.10 Å) to conduct the computational modeling
of the catalytic conversion from sucrose to mutan. From this
crystal structure, three protein residues have been proposed to
play a relevant catalytic role, namely Asp477 as a nucleophile,
Glu515 as an acid/base, and Asp588 as a transition state stabil-
izer.15 Homologous catalytic residues have been found in
another GTF crystal structure from Lactobacillus reuteri 180.12
Thus, the 3AIE crystal structure offers a suitable model to
conduct exhaustive computational studies on GTF-SI, aimed at
elucidating the fundamental aspects regarding the mecha-
nism, such as the key amino acid residues stabilizing tran-
sition state structures, and the catalytic steps involved in the
reaction path.
Herein we carried out Quantum Mechanics (QM) and
Quantum Mechanics/Molecular Mechanics (QM/MM) calcu-
lations starting from the structure of sucrose in complex with
the crystalline structure of the GTF-SI enzyme. In a first stage,
we built a reduced cluster model of the active site including
the main catalytic residues Asp477 and Glu515. Then, the
enzymatic environment was incorporated in order to fully
characterize the catalytic strategies of this enzyme by means of
the QM/MM method.
2. Methods
2.1 Models
The crystal structure of GTF-SI (2.10 Å resolution) corresponds
to an octamer composed of equal chains in structure and
sequence, with no sucrose in the active site (apoenzyme).
Thus, to model the GTF-SI·sucrose binary complex the chain C
This journal is © The Royal Society of Chemistry 2018

Organic & Biomolecular Chemistry
Published on 09 March 2018. Downloaded by Universidad de Concepcion on 20/03/2018 13:42:30.
Fig. 2 Active site view of the superimposed crystal structures GTF-SI
(green) and GTF180ΔN (cyan).
was chosen and aligned with the GTF180-ΔN crystal structure,
which is in complex with sucrose (PDB code 3HZ3). The struc-
tural alignment showed a QH20 value of 0.800, which indicates
a high structural and sequence similarity between 3AIE and
3HZ3 structures. Thus, the matching procedure enabled locat-
ing the sucrose unit within the active site of GTF-SI (Fig. 2),
forming this way the GTF-SI·sucrose complex. This model
already features the most relevant interactions for the catalytic
mechanism and was employed to conduct further QM and
QM/MM calculations.
The PROPKA method21 was applied to determine the proto-
nation states of all ionizable residues at pH = 6.0 in which the
GTF-SI enzyme exhibits its maximum activity,22 using the
PDB2PQR2.1.1 server.23 According to PROPKA, Glu515 was
predicted to exist in its protonated state. All other residues
were assigned standard protonation states. This complex was
solvated using a water sphere of a 25 Å radius centered at the
glycosidic oxygen, forming a model of 13 893 atoms.
2.2 Quantum mechanical calculations
The initial model was truncated to a smaller system composed
of sucrose, Asp477, Glu515, and three crystalline water mole-
cules (residue numbers 55, 127 and 128) that interact with the
glucosyl and fructosyl moieties of sucrose. Geometry optimi-
zations in the gas phase were computed at the BP86/6-31+G(d)
level of theory. This functional has been successfully tested for
carbohydrate related enzymes showing a good compromise
between computational efficiency and accuracy.24 All calcu-
lations were carried out using the Q-Chem program
package.25,26 The Freezing String Method27 (FSM) was used to
locate the TS structure. Briefly, this algorithm consists of an
iterative process in which growing pairs of molecular struc-
tures along the reaction path, starting from the reactant and
the product, complete the “string” generating a guess TS struc-
ture. Later, local optimizations of the guess structure were
performed using an exact initial Hessian that was updated
This journal is © The Royal Society of Chemistry 2018
View Article Online
Paper
until obtaining a viable TS structure using the Powell/
Murtagh–Sargent scheme.28,29 The stationary points were
characterized as minima or transition state (TS) structures by
means of frequency calculations. Finally, single point energy cal-
culations were performed with the Hybrid-Generalized Gradient
Approximation (H-GGA)ωB97X-V30 functional with nonlocal cor-
relation, using the 6-311++G(d,p) basis set. This functional has
been proved to describe very well different molecular pro-
perties.31 Natural charges and Second Order Perturbation ana-
lyses were carried out through NBO calculations using the NBO
v5.0 program32 as implemented in Q-Chem.
2.3 QM/MM calculations
The initial GTF-SI·sucrose complex was used as a starting
structure for QM/MM calculations. The system containing
13 893 atoms was partitioned into a Quantum Mechanics part
(QM) defined by 61 atoms (link atoms not included): 45 atoms
of sucrose, the side chain of the nucleophile Asp477 (6 atoms),
and the side chain of the acid Glu515 (10 atoms). Two link
atoms were used, each one of them located along the Cα-QM
of Asp477 and Glu515. The total charge of the QM subsystem
was −1. The geometry of the QM subsystem was described at
the BP86/6-31+G(d) level of theory. The Molecular Mechanics
(MM) part was described using the Charmm27-all33 atoms
force field, and an electronic embedding scheme was used for
the QM/MM treatment. CHARMM34,35 and Q-CHEM packages
were employed to carry out the calculations. Initially, the QM
part was frozen and the rest of the system was optimized using
a gradient criterion of 0.005 kcal mol−1 Å−1. Then, we defined
an active region of 25 Å radius centered at the glycosidic
oxygen. Every residue within the active region was set free,
while the remaining residues were frozen during the optimi-
zations. This partial freezing method was recently found to be
appropriate for the treatment of large systems.36 The pro-
cedure afforded an optimized reactant structure that was used
as a starting point for the initial reaction path calculation.
This initial guess was built using a Reaction Coordinate (RC)
that allowed the conversion of the reactant to the CGE inter-
mediate (Fig. 3).
The initial path was used as a guess to compute the
Minimum Energy Path (MEP), which was calculated using the
Conjugate Peak Refinement (CPR) algorithm37 as
implemented in the TREK module of the CHARMM program.
The CPR algorithm rebuilds the whole path that connects the
desired reactant and product structures, and locates the
correct transition state structure, ensuring that the energy
barrier is not affected by a biasing description of the
process.38 This method allows determination of the mecha-
nism of complex reactions in proteins and has been success-
fully used in conjunction with QM/MM energy functions to
study the hydrolysis of chondroitin disaccharide catalyzed by
chondroitin lyase39 and other complex enzymatic
reactions.40–42 The CPR criterion for deciding whether a tran-
sition state has been found was that a gradient of less than
10−2 kcal mol−1 Å−1 is attained. CPR identifies these points
along the path where the energy is highest (the peaks) and
Org. Biomol. Chem.

Paper
Published on 09 March 2018. Downloaded by Universidad de Concepcion on 20/03/2018 13:42:30.
Fig. 3 QM (black)/MM (gray) partition used in this work. Red lines indi-
cate the location of the link atoms, and the β-face of the glucosyl
moiety is highlighted (opposite face of fructosyl). The labels of several
atoms are shown and will be referenced throughout the results section.
The Reaction Coordinate (RC) has been defined as RC = d(C1–OG) −
d(C1–OD1) − d(HE2–OG).
iteratively moves these points closer to the MEP by controlled
conjugated-gradient minimization. The transition states were
further validated by means of normal mode analyses of the
QM part.
Finally, single point energy calculations were carried out for
the stationary points using the H-GGA ωB97X-V functional
together with the 6-311++G(d,p) basis set. The ωB97X-V/6-
311++G(d,p) method was also used for the NBO and electro-
static calculations.
3. Results and discussion
This work describes the results of QM and QM/MM calcu-
lations dealing with the mechanism of sucrose hydrolysis by
GTF-SI. The characterization and analysis of the mechanism
were carried out considering the structural and stereoelectro-
nic factors that could modulate the nature of the TS and the
conformational itinerary of the glucosyl moiety during the
reaction, which is an undetermined issue for GTF-SI and its
whole GH70 family.
3.1 CGE formation using a QM cluster model
The cluster model employed in the present study includes
sucrose, Asp477, Glu515 and three water molecules. After geo-
metry optimization, a reactant-state system was obtained
showing a series of stabilizing interactions between the
sucrose and the amino acid residues. For example, the hydro-
gen bonds between the glucose that will be transferred and
Asp477: (i) 6H–OD2 = 1.77 Å and (ii) 2H–OD2 = 1.52 Å; and
one interaction between the leaving fructosyl group and
Org. Biomol. Chem.
View Article Online
Organic & Biomolecular Chemistry
Glu515: 3O′–HE2 = 1.75 Å, as displayed in Fig. 4. Besides, a
water molecule is interacting with the 4H and 3O of glucose at
1.84 Å and 1.75 Å, respectively. The conformation of glucose at
the reactant is highly distorted exhibiting ϕ = 155° and θ = 82°,
which is consistent with a skew-boat 2SO type, according to the
Cremer–Pople parameters.43 Additionally, the putative nucleo-
phile Asp477 seems to be well positioned for the nucleophilic
attack, with the OD1 atom located at 3.83 Å from the anomeric
carbon of glucose (C1 in Fig. 4).
The CGE structure was obtained from a flexible scan along
the bond between the glycosidic oxygen and the acid proton of
Glu515 using the optimized reactant structure as a starting
point. Then, the FSM algorithm was applied to localize the TS
structure using the optimized structures of the reactant and
the CGE intermediate. The TS structure has an oxocarbenium
character demonstrated by the change of the C1–O5 bond dis-
tance, which has decreased from 1.43 Å in the reactant to
1.30 Å in the TS, showing a significant double bond character.
The values of ϕ and θ are 205° and 45°, respectively, indicating
a conformation close to the half-chair 4H3, similar to that
observed for the retaining α-galactosidases (GH27 family).44
Normal mode analysis revealed a unique imaginary frequency
of 125.5i cm−1, confirming the nature of a saddle point con-
necting the reactant and the CGE structures. The vibrational
mode was associated with stretching of the C1–OG, C1–
OD1Asp477 and the OG–HE2Glu515 bonds, in agreement with a
concerted mechanism for the CGE formation in which the
proton transfer and the nucleophilic attack occur after the gly-
cosidic bond breakage, as confirmed by IRC calculations.
Additionally, Natural Population Analysis (NPA) revealed that
the TS charge, including the atomic charges of the anomeric
center (O5–C1–H1), was 0.30 a.u. higher than the charge of the
reactant, thus revealing a positive charge development, in
agreement with a oxocarbenium ion-like TS. Natural Bond
Orbital (NBO) analysis confirmed the double bond character of
the C1–O5 bond at the TS and also confirmed that the glycosi-
dic bond is already broken, with a distance of 2.15 Å. The
interaction between Glu515 and the glycosidic oxygen (OG)
is still present in the TS (HE2–OG = 1.65 Å), showing that the
proton transfer has not occurred yet. The nucleophilic oxygen
of Asp477 is located at 2.64 Å from the anomeric carbon,
1.19 Å closer than the distance observed at the reactant, pro-
viding stabilization to the oxocarbenium ion. A new interaction
between the 2-hydroxyl group (2OH) of glucose and the OG
(2H–OG = 1.65 Å) was observed at the TS, which was attributed
to the increase of negative charge on the OG (0.25 a.u.) upon
glycosidic bond breakage. This suggests that the 2H–OG inter-
action has an important TS-stabilizing effect. The nature of the
2H–OG interaction was further analyzed using a Second Order
Perturbation analysis of the Fock matrix in Natural Bond Orbital
(NBO) basis, which detailed important interactions between
donor Lewis-type and acceptor non-Lewis-type NBOs describing
the quantum region (Table 1). The initial geometrical con-
clusion about the relevance of the 2H–OG interaction on the
stabilization of the oxocarbenium was thus confirmed through
this analysis. The results indicate a nOG ! σ*2OH interaction,
This journal is © The Royal Society of Chemistry 2018
 View Article Online

Organic & Biomolecular Chemistry Paper

Published on 09 March 2018. Downloaded by Universidad de Concepcion on 20/03/2018 13:42:30.

Fig. 4 Snapshots of stationary points along the reaction path for the cluster model.

Table 1 Stabilization energy (ΔE) due to the interaction between the ΔΔE (51.5 kcal mol−1) corresponding to nOG ! σ*ðC1O5Þ high-
donor and acceptor orbitals obtained from Second Order Perturbation lights the relevance of stereoelectronic interactions on the
analysis of the QM cluster model. The ΔE values for reactant and TS
oxocarbenium stabilization. This interaction gains relevance at
structures are given in kcal mol−1. The ΔΔE is the difference between
the ΔE at the TS compared to that for the reactant
the TS even though the glycosidic bond (C1–OG) is already
broken (2.15 Å). It is interesting that although the mechanism
Reactant TS ΔΔE was found to be a concerted one, the proton transfer has not
occurred yet (HE2–OG = 1.65 Å). The reason underlying this
n(OE1)–σ*(1OH′) 10.7 29.3 18.6
n(OD1)–σ*(C1–O5) Missing 8.7 8.7 finding could be understood by means of 2nd-order pertur-
n(OD2)–σ*(6OH) 10.5 21.7 11.2 bation analysis, which shows the important stabilizing effect
σ(C2–H2)–σ*(C1–O5) Missing 15.5 15.5 of nOG ! σ*ðHE2OE2Þ (19.8 kcal mol−1), explaining that even
n(O5)–σ*(C1–O2) 3.9 8.9 5.0
n(OG)–σ*(OE2–HE2) 1.6 21.4 19.8 though the proton transfer has not occurred, there is an
n(OG)–σ*(2OH) Missing 16.0 16.0 important favorable effect of this interaction.
n(OG)–σ*(C1–O5) 8.6 60.1 51.5 The CGE intermediate shows a structure in which the
leaving fructosyl group has already been protonated by Glu515,
and Asp477 has attacked the glucosyl moiety. In addition, the
suggesting a strong hydrogen bond. The presumable stabilizing Cremer–Pople parameters (ϕ = 240° and θ = 79°) indicate that
effect of the hydrogen bonds formed by Asp477 and Glu515 glucose reaches an envelope ∼4E conformation with an energy
with sucrose (Fig. 4) is supported by the high calculated ΔΔE, reaction of −7.77 kcal mol−1, which is in agreement with other
for example, for nOD1 ! σ*ð6OHÞ or nOE1 ! σ *ð1OH′ Þ , which provide theoretical predictions.16 Thus both the proton transfer and
some of the strongest stabilizations for the TS of 11.2 and the nucleophilic attack seem to occur without an apparent
18.6 kcal mol−1, respectively. The important effect of the hydro- energy barrier, indicating that the energy barrier is mostly
gen bond between the OD1Asp477 and 6H of glucose is in agree- associated with the glycosidic bond breakage.
ment with what is known about the relevance of this interaction The calculated activation energy for the reaction at the
in other similar enzymes, which usually accounts for a stabiliz- BP86/6-31+G(d) level is 15.6 kcal mol−1, which increases to
ing effect at the TS of ∼5 kcal mol−1.16 Additionally, the highest 26.4 kcal mol−1 when single point energy calculations are

This journal is © The Royal Society of Chemistry 2018 Org. Biomol. Chem.

Paper
carried out at the ωB97X-V/6-311++G(d,p) level. The first
method has been widely used due to its good geometrical per-
formance and computational efficiency; however it has been
found that it underestimates the energy barriers related to
carbohydrate mechanisms.24,45 Thus, all the energy values dis-
cussed from now on are at the ωB97X-V/6-311++G(d,p) level,
which has shown fewer errors regarding both reaction and acti-
vation energies.31 The energy barrier obtained with this func-
tional, 26.4 kcal mol−1, is rather high compared to experi-
Published on 09 March 2018. Downloaded by Universidad de Concepcion on 20/03/2018 13:42:30.
mental activation energy for the related GTF-I enzyme46
(∼16 kcal mol−1). Despite this high discrepancy, the minimal
cluster model is very useful because it separates the pure
chemistry of the process from the electrostatic stabilization of
the enzyme, something that would start mixing if we had used
a larger model of the enzyme. The results of the full enzyme
model will be discussed in the next section.
3.2 CGE formation using a QM/MM model
The minimal cluster model provided a first indication of the
mechanism, revealing that the two catalytic residues that
directly participate in the reaction, Asp477 and Glu515, are
also very important for stabilizing the fructosyl and glucosyl
groups during the reaction. To further analyze the catalytic
strategies of GTF-SI, the reaction was studied using a full
enzyme model. The reaction path obtained at the QM (BP86/6-
31+G(d))/MM(CHARMM27) (Fig. S1†), using the reaction co-
ordinated (RC) previously defined, shows only one reaction
event for the formation of the CGE intermediate, which is
obtained after overcoming an energy barrier of ∼15 kcal mol−1.
Thus, based on this guess reaction path, the Minimum Energy
Path (MEP) was built using the CPR algorithm. The energies
associated with the TS and the CGE structures were corrected
at the ωB97X-V/6-311++(d,p) level (Fig. 5), obtaining an energy
barrier of 17.9 kcal mol−1, which is in very good agreement
with the experimental evidence.46
Although the CPR algorithm guarantees the location of the
true TS,37,38 this was further characterized by means of normal
mode analysis showing one imaginary frequency of 188.9i cm−1
with an intensity of 245 kcal mol−1, which was associated with
Fig. 5 Potential energy profiles (kcal mol−1) that schematically describe
the CGE formation for the cluster and the QM/MM models.
Org. Biomol. Chem.
View Article Online
Organic & Biomolecular Chemistry
an antisymmetric stretching of C1–OG, HE2Glu515–OG and
OD1Asp477–C1 bonds, related to the breakage of the glycosidic
bond along with the proton transfer and the nucleophilic
attack.
The results obtained for the full enzyme model revealed
that the presence of the enzymatic environment decreases the
energy of the TS in about 9 kcal mol−1, compared to the
cluster model (Fig. 5). In addition, QM/MM calculations indi-
cate a concerted mechanism in which the glycosidic bond is
being simultaneously broken as the proton transfer occurs, fol-
lowed by the nucleophilic attack, unlike cluster model results.
The main structural and stereoelectronic factors that might be
relevant for the stabilization of the TS and the conformations
adopted by the glucosyl moiety during the reaction will be dis-
cussed next.
The interactions established by sucrose in the reactant
structure (Fig. S2†) give rise to a conformation of glucose in
which the calculated puckering parameters (ϕ = 150° and θ =
36°) indicate a moderately distorted structure that resembles
the 2H3 conformation. Interestingly, this kind of sugar confor-
mation was previously proposed for the glucose in GTF180ΔN
and therefore, the finding of the 2H3 conformation in GTF-SI
comes to support this proposal.5
The mechanistic calculations show that the reaction
starts with the lengthening of the C1–OG bond while Asp477
and Glu515 are approaching the glucose, which is in part
caused by the growing conformational distortion of the
glucose. In the TS structure the C1–OG is 2.16 Å, the C1–
OD1Asp477 is 2.43 Å and the HE2–OG is 1.11 Å, which demon-
strates that the glycosidic bond breakage and the proton trans-
fer occur in a concerted way (Table 2 and Fig. 6). The confor-
mation of glucose at the TS is closer to both the E3 and the
2
H3 (ϕ = 173° and θ = 43°). The oxocarbenium nature of the TS
is confirmed by the distance C1–O5, which has decreased from
1.38 Å to 1.29 Å (Table 2), evidencing a strong double-bond
character; additionally, the NPA charges on the O5–C1–
H1group of atoms are 0.26 a.u. higher than the calculated
charges for the reactant, which shows a decrease of the elec-
tron density on the anomeric carbon as reactants evolve to the
TS structure.
The formation of a partial positive charge on the anomeric
center was confirmed by NBO analysis, which revealed a
decrease in the p-character of C1 and the formation of a new
Table 2 Selected distances d (Å), conformations, and change of posi-
tive charge at the anomeric center (Δq in a.u.) for the reactant, TS and
CGE intermediate
Reactant TS CGE
d(C1–OG) 1.51 2.16 2.82
d(C1–OD1Asp477) 3.09 2.43 1.49
d (C1–O5) 1.38 1.29 1.40
d(HE2–OG) 1.78 1.11 1.05
d(HE2–OE2) 1.01 1.37 1.55
Conformation ∼2H3 2
H3–E3 ∼4C1
Δq(O5–C1–H1) 0.00 0.26 −0.02
This journal is © The Royal Society of Chemistry 2018
 View Article Online

Organic & Biomolecular Chemistry Paper

Published on 09 March 2018. Downloaded by Universidad de Concepcion on 20/03/2018 13:42:30.

Fig. 6 Snapshots of stationary points along the reaction path for the full enzyme model.

π bonding interaction between C1 and O5, in agreement with Table 3 Stabilization energy (ΔE) due to the interaction between the
the expected Lewis structure of an oxocarbenium ion charac- donor and acceptor orbitals obtained from Second Order Perturbation
analysis of the full enzyme model. The ΔE values for reactant and TS
terized by a rehybridization of C1 from sp3 to sp2.
structures are given in kcal mol−1. The ΔΔE corresponds to the differ-
In order to get deeper insight into the nature of the main ence between the ΔE at the TS compared to that for the reactant
interactions that contribute to the stabilization of the TS from
a molecular orbital point of view, a second-order perturbation Reactant TS ΔΔE
analysis was carried out as displayed in Table 3.
n(OE1)–σ*(1OH′) 7.1 8.9 1.8
Our results revealed that the highest ΔΔE is related to n(OG)–σ*(OE2–HE2) 18.9 Missing −18.9
nOE2 ! σ*ðOG HE2Þ interaction (108 kcal mol−1), revealing that n(OE2)–σ*(OG–HE2) Missing 108.0 108.0
Glu515 acts as a catalytic acid but also as a strong TS stabilizer. n(OD1)–σ*(C1–O5) 0.1 17.6 17.5
n(OD2)–σ*(6OH) 11.2 27.3 16.1
This interaction was not observed in the cluster model. The σ(C5–H5)–σ*(C1–O5) 1.2 14.8 13.6
great contribution of the nOE2 ! σ*ðOG HE2Þ . interaction in the n(O5)–σ*(C1–O2) 4.9 8.9 4.0
QM/MM model can be related to the large positive charge on n(OG)–σ*(C1–O5) 4.6 50.1 45.5
σ(C–OE2)–σ*(1OH′) 4.2 8.2 4.0
the HE2 atom (0.52 a.u.), which is located between two n(OG)–σ*(C1–C2) 8.4 9.8 1.4
oxygens with high electron density: the OE2 (−0.78 a.u.) and n(OG)–σ*(C4–C5) 7.5 10.2 2.7
the OG (−0.82 a.u.) atoms (Fig. 7B). These atoms are separated n(OG)–σ*(C5–C6) 7.8 9.6 1.8
by a distance of 2.48 Å (Fig. 7A), which is characteristic of
special short-strong hydrogen bonds.47 This kind of strong
interaction has not been described for other related GH- transferred proton as compared with GTF-SI and therefore, it
enzymes, such as members of GH13 or GH77 families. The is not able to establish such a short-strong hydrogen bond.16
only QM/MM study carried out with a similar enzyme of the Additionally, the positive charge of the Arg475 residue in
GH13 family, the human pancreatic α-amylase, showed that the active site can assist the proton transfer from Glu515 to
the acid residue homologue to Glu515 is farther from the the fructosyl leaving group by stabilizing the growing negative

This journal is © The Royal Society of Chemistry 2018 Org. Biomol. Chem.

Paper
Published on 09 March 2018. Downloaded by Universidad de Concepcion on 20/03/2018 13:42:30.
Fig. 7 Active site view of the TS indicating: A, the interactions and residues that are important for TS stabilization and B, a short-strong hydrogen
bond depicted as a (red) high and (green) low electron density isosurface for the nOE2 ! σ*ðOG HE2Þ interaction.
charge on the glutamate moiety by means of long-range
electrostatic interactions (Fig. 7). Actually, the distance
between Arg475 and Glu515 decreases from 4.32 to 4.12 Å as
evolving from Reactant to TS. Thus, we propose that Arg475
plays a very important role to properly describe the concerted
nature of this mechanism.
Another important interaction for TS stabilization corres-
ponds to nOG ! σ*ðC1O5Þ (45.5 kcal mol−1), which was also
observed in the cluster model, thus highlighting the impor-
tance of this stereoelectronic interaction in the oxocarbenium
stabilization. The hydrogen bonds formed between sucrose
and Asp477 and Glu515 are also relevant for the stabilization
of the TS, which is supported by the ΔΔE related to the
nOD1Asp477 ! σ*6ðHOÞ interaction (16.1 kcal mol−1), which is
5.0 kcal mol−1 stronger than the interaction observed in the
cluster model. A relevant interaction observed in both models
corresponds to nOD2Asp477 ! σ*C1O5 , which shows a two-fold
increase in the QM/MM model compared to the cluster
system. To further analyze the role of Asp477 in the stabiliza-
tion of oxocarbenium, this amino acid was removed from the
QM part of the QM/MM system, and single point calculations
were carried out for reactant and TS structures. The results
showed an energy barrier increase in about 22 kcal mol−1,
which clearly indicates the key role of Asp477 in the stabiliza-
tion of the oxocarbenium ion-like TS.
Interestingly, the σC5H5 ! σ*C1O5 interaction has a ΔΔE of
13.6 kcal mol−1 and is not present in the cluster model.
However, an analogue interaction between σC2H2 ! σ*C1O5 is
found in the reduced model (15.5 kcal mol−1). In both cases,
the acceptor antibonding orbital is the σ*C1O5 , which suggests
that these two interactions could favor a slightly different dis-
torted conformation of the glucosyl moiety, as is revealed by the
different conformational itineraries obtained from each model.
The interactions established by Asp588 could be crucial for
the catalytic activity of GTF-SI since mutagenesis experiments
in which the homologue of Asp588 (Asp1136) in GTF180-ΔN
was mutated resulted in a drastic decrease in activity.12 Due to
this, both Asp1136 and Asp588 were proposed as a putative TS
Org. Biomol. Chem.
View Article Online
Organic & Biomolecular Chemistry
stabilizer; however this amino acid residue has also been pro-
posed as a CGE stabilizer.4,12,15 Thus, the role of this residue
has not been confirmed yet. As can be seen in Fig. 7, Asp588 is
interacting with sucrose through its glucosyl and fructosyl
moiety along the whole reaction path and thus, it is expected
to exert an influence during the mechanism. It is worth noting
that the interaction between glucose and Asp588 (2OH–
OD2Asp588) is very strong showing an average distance of 1.55 Å
over the reactant, TS and CGE structures. If Asp588 were not
present in the active site, it is likely that the 2OH group will
reorient to form another similar interaction. Actually, in the
reactant of the cluster model the 2OH group was interacting
with Asp477 and then this group was reoriented to the OG
(opposite face of glucose) at the TS and CGE structures. This
suggests that Asp588 is important for maintaining the 2OH
group correctly oriented during the reaction mechanism.
Moreover, the closeness between Asp588 and the Glu515 (∼3 Å)
could be especially important at the reactant state because it
might increase the pKa of Glu515, which is necessary for cataly-
sis. In order to examine whether Asp588 exerts a transfer charge
effect during the mechanism, Asp588 was included into the QM
part of the QM/MM system and the energy barrier and the reac-
tion energy were recalculated.
Thus, the calculated energy barrier and reaction energy are
16.4 and −0.88 kcal mol−1, respectively. This shows a stabiliz-
ing effect of 1.5 and 3.1 kcal mol−1 for the TS and CGE,
respectively, suggesting that transfer charge from Asp588 to
sucrose could be important for both the TS and CGE stabiliz-
ation. Finally, when Asp477 attacks the anomeric carbon the
CGE intermediate is formed showing a distance of 1.49 Å for
the C1–OD1Asp477 (Fig. 6). In the intermediate, the glucosyl
moiety adopts a chair conformation (4C1), similarly to that
observed for several GHs.44
It is worth noting that the conformational itineraries of
glucose along the reaction path are not clear for members of
GH13 neither for GH70 families;19,48 therefore, this is the first
study that reveals conformational features of the glucose during
the catalytic process: ∼2H3 → 2H3–E3 → 4C1. This new knowl-
This journal is © The Royal Society of Chemistry 2018

Organic & Biomolecular Chemistry
edge could be very valuable for the design of specific inhibitors
of GTF-SI that could be employed for dental caries treatment.
4. Conclusion
Despite the number of mechanistic studies regarding the
Glycosyl Hydrolase (GH) family, a number of these enzymes
have not been comprehensively addressed in terms of the
Published on 09 March 2018. Downloaded by Universidad de Concepcion on 20/03/2018 13:42:30.
specific details of their catalytic mechanisms. Among them is
GTF-SI, which is involved in the development of dental caries,
and has been identified as a target for preventive treatments.
Thus, understanding the atomistic details of GTF-SI hydrolysis
of sucrose is relevant for drug design.
In this work QM and QM/MM calculations were carried out
using a cluster and a full enzyme model built from the crystal
structure of GTF-SI in a complex with sucrose. The TS was
found to be strongly stabilized by the hydrogen bond
OG⋯HE2⋯OE2, which according to the distance OG⋯OE2,
charges and 2nd-order perturbation analysis can be considered
as a short-strong hydrogen bond. This strong stabilizing inter-
action seems to be a new insight into the catalytic strategies
used for this enzyme since it has not been observed in other
members of related-GH families.
As for other GH enzymes, the nucleophile (Asp477 in
GTF-SI) is able to provide important stabilization for the TS; in
this case the interaction is established between Asp477 and
the 6OH group of the glucosyl moiety. Besides, Arg475 likely
assists the proton transfer stabilizing the nascent negative
charge on the glutamate moiety. Another important active site
residue is Asp588, which interacts with sucrose through its glu-
cosyl (2OH group) and fructosyl moiety (1OH′) along the whole
reaction path. Asp588 seems to be crucial for maintaining the
2OH group suitably oriented during the reaction mechanism,
exerting a stabilizer role not only in the TS, as has been pro-
posed, but also in the CGE intermediate. Actually, the
inclusion of Asp588 in the QM region results in a decrease of
the activation and reaction energies down to 16.4 and
−0.88 kcal mol−1, respectively.
We consider that this new information provides relevant
new knowledge about specific enzymatic catalysis aspects
with implications not only for GTF-SI but also for the whole
GH70 family. Thus, we propose for the first time a confor-
mational itinerary for the glucosyl moiety of sucrose: ∼2H3 →
2
H3–E3 → 4C1.
We hope that these key structural and stereoelectronic fea-
tures of the TS might be considered in the future design of
specific inhibitors for GTF-SI. Upcoming efforts will be
focused on unraveling the factors that modulate the specificity
(α-1,3 or α-1,6 linkage) of the new glycosidic bond that is
synthesized by GTF-SI.
Conflicts of interest
There are no conflicts to declare.
This journal is © The Royal Society of Chemistry 2018
View Article Online
Paper
Acknowledgements
The authors acknowledge financial support from Grant
FONDECYT no. 1150704.
References
1 P. E. Petersen, S. Estupinan-Day and C. Ndiaye, Bull.
W. H. O., 2005, 83, 642–642.
2 P. E. Petersen, D. Bourgeois, H. Ogawa, S. Estupinan-Day
and C. Ndiaye, Bull. W. H. O., 2005, 83, 661–669.
3 R. A. Giacaman, P. M. Reyes and V. B. Leon, Braz. Oral Res.,
2013, 27, 7–13.
4 X. F. Meng, J. Gangoiti, Y. X. Bai, T. Pijning, S. S. Van
Leeuwen and L. Dijkhuizen, Cell. Mol. Life Sci., 2016, 73,
2681–2706.
5 H. Leemhuis, T. Pijning, J. M. Dobruchowska, S. S. van
Leeuwen, S. Kralj, B. W. Dijkstra and L. Dijkhuizen,
J. Biotechnol., 2013, 163, 250–272.
6 H. Leemhuis, T. Pijning, J. M. Dobruchowska,
B. W. Dijkstra and L. Dijkhuizen, Biocatal. Biotransform.,
2012, 30, 366–376.
7 Z. Ren, L. L. Chen, J. Y. Li and Y. Q. Li, J. Oral Microbiol.,
2016, 8, 31095.
8 L. Petersen, A. Ardevol, C. Rovira and P. J. Reilly, J. Am.
Chem. Soc., 2010, 132, 8291–8300.
9 X. Biarnes, A. Ardevol, J. Iglesias-Fernandez, A. Planas and
C. Rovira, J. Am. Chem. Soc., 2011, 133, 20301–20309.
10 L. Raich, V. Borodkin, W. X. Fang, J. Castro-Lopez,
D. M. F. van Aalten, R. Hurtado-Guerrero and C. Rovira,
J. Am. Chem. Soc., 2016, 138, 3325–3332.
11 B. Henrissat and G. Davies, Curr. Opin. Struct. Biol., 1997,
7, 637–644.
12 A. Vujicic-Zagar, T. Pijning, S. Kralj, C. A. Lopez,
W. Eeuwema, L. Dijkhuizen and B. W. Dijkstra, Proc. Natl.
Acad. Sci. U. S. A., 2010, 107, 21406–21411.
13 V. Monchois, R. M. Willemot and P. Monsan, FEMS
Microbiol. Rev., 1999, 23, 131–151.
14 R. Ohrlein, Biocatalysis - from Discovery to Application, 1999,
vol. 200, pp. 227–254.
15 K. Ito, S. Ito, T. Shimamura, S. Weyand, Y. Kawarasaki,
T. Misaka, K. Abe, T. Kobayashi, A. D. Cameron and
S. Iwata, J. Mol. Biol., 2011, 408, 177–186.
16 G. P. Pinto, N. F. Bras, M. A. S. Perez, P. A. Fernandes,
N. Russo, M. J. Ramos and M. Toscano, J. Chem. Theory
Comput., 2015, 11, 2508–2516.
17 R. I. Sadreyev, B. H. Kim and N. V. Grishin, Curr. Opin.
Struct. Biol., 2009, 19, 321–328.
18 M. Kitamura, M. Okuyama, F. Tanzawa, H. Mori, Y. Kitago,
N. Watanabe, A. Kimura, I. Tanaka and M. Yao, J. Biol.
Chem., 2008, 283, 36328–36337.
19 G. J. Davies, A. Planas and C. Rovira, Acc. Chem. Res., 2012,
45, 308–316.
20 E. Roberts, J. Eargle, D. Wright and Z. Luthey-Schulten,
BMC Bioinf., 2006, 7, 382.
Org. Biomol. Chem.

Paper
21 H. Li, A. D. Robertson and J. H. Jensen, Proteins: Struct.,
Funct., Bioinf., 2005, 61, 704–721.
22 J. S. Chia, C. S. Yang and J. Y. Chen, Infect. Immun., 1998,
66, 4797–4803.
23 T. J. Dolinsky, J. E. Nielsen, J. A. McCammon and
N. A. Baker, Nucleic Acids Res., 2004, 32, W665–W667.
24 A. T. Pereira, A. J. M. Ribeiro, P. A. Fernandes and
M. J. Ramos, Int. J. Quantum Chem., 2017, 117, e25409.
25 Y. Shao, L. F. Molnar, Y. Jung, J. Kussmann, C. Ochsenfeld,
Published on 09 March 2018. Downloaded by Universidad de Concepcion on 20/03/2018 13:42:30.
S. T. Brown, A. T. B. Gilbert, L. V. Slipchenko,
S. V. Levchenko, D. P. O’Neill, R. A. DiStasio Jr.,
R. C. Lochan, T. Wang, G. J. O. Beran, N. A. Besley,
J. M. Herbert, C. Yeh Lin, T. Van Voorhis, S. Hung Chien,
A. Sodt, R. P. Steele, V. A. Rassolov, P. E. Maslen,
P. P. Korambath, R. D. Adamson, B. Austin, J. Baker,
E. F. C. Byrd, H. Dachsel, R. J. Doerksen, A. Dreuw,
B. D. Dunietz, A. D. Dutoi, T. R. Furlani, S. R. Gwaltney,
A. Heyden, S. Hirata, C.-P. Hsu, G. Kedziora,
R. Z. Khalliulin, P. Klunzinger, A. M. Lee, M. S. Lee,
W. Liang, I. Lotan, N. Nair, B. Peters, E. I. Proynov,
P. A. Pieniazek, Y. Min Rhee, J. Ritchie, E. Rosta, C. David
Sherrill, A. C. Simmonett, J. E. Subotnik, H. Lee Woodcock
Iii, W. Zhang, A. T. Bell, A. K. Chakraborty, D. M. Chipman,
F. J. Keil, A. Warshel, W. J. Hehre, H. F. Schaefer Iii,
J. Kong, A. I. Krylov, P. M. W. Gill and M. Head-Gordon,
Phys. Chem. Chem. Phys., 2006, 8, 3172–3191.
26 A. I. Krylov and P. M. W. Gill, Wiley Interdiscip. Rev.:
Comput. Mol. Sci., 2013, 3, 317–326.
27 A. Behn, P. M. Zimmerman, A. T. Bell and M. Head-
Gordon, J. Chem. Phys., 2011, 135, 224108.
28 B. A. Murtagh and R. W. H. Sargent, Comput. J., 1970, 13,
185–194.
29 M. J. D. Powell, Math. Program., 1971, 1, 26–57.
30 N. Mardirossian and M. Head-Gordon, Phys. Chem. Chem.
Phys., 2014, 16, 9904–9924.
31 S. Dasgupta and J. M. Herbert, J. Comput. Chem., 2017, 38,
869–882.
32 E. D. Glendening, J. K. Badenhoop, A. E. Reed,
J. E. Carpenter, J. A. Bohmann, C. M. Morales and
F. Weinhold, NBO 5.0, Theoretical Chemistry Institute,
W. Madison, 2001; http://www.chem.wisc.edu/~nbo5.
33 A. D. MacKerell, D. Bashford, M. Bellott, R. L. Dunbrack,
J. D. Evanseck, M. J. Field, S. Fischer, J. Gao, H. Guo,
Org. Biomol. Chem.
View publication stats
View Article Online
Organic & Biomolecular Chemistry
S. Ha, D. Joseph-McCarthy, L. Kuchnir, K. Kuczera,
F. T. K. Lau, C. Mattos, S. Michnick, T. Ngo, D. T. Nguyen,
B. Prodhom, W. E. Reiher, B. Roux, M. Schlenkrich,
J. C. Smith, R. Stote, J. Straub, M. Watanabe,
J. Wiorkiewicz-Kuczera, D. Yin and M. Karplus, J. Phys.
Chem. B, 1998, 102, 3586–3616.
34 M. Karplus, J. Comput. Chem., 1983, 4, 187–217.
35 B. R. Brooks, C. L. Brooks, A. D. Mackerell, L. Nilsson,
R. J. Petrella, B. Roux, Y. Won, G. Archontis, C. Bartels,
S. Boresch, A. Caflisch, L. Caves, Q. Cui, A. R. Dinner,
M. Feig, S. Fischer, J. Gao, M. Hodoscek, W. Im,
K. Kuczera, T. Lazaridis, J. Ma, V. Ovchinnikov, E. Paci,
R. W. Pastor, C. B. Post, J. Z. Pu, M. Schaefer, B. Tidor,
R. M. Venable, H. L. Woodcock, X. Wu, W. Yang,
D. M. York and M. Karplus, J. Comput. Chem., 2009, 30,
1545–1614.
36 A. R. Calixto, M. J. Ramos and P. A. Fernandes, J. Chem.
Theory Comput., 2017, 13, 5486–5495.
37 S. Fischer and M. Karplus, Chem. Phys. Lett., 1992, 194,
252–261.
38 F. A. Kiani and S. Fischer, Proc. Natl. Acad. Sci. U. S. A.,
2014, 111, E2947–E2956.
39 J. Xiong and D. G. Xu, J. Phys. Chem. B, 2017, 121, 931–941.
40 H. Batebi and P. Imhof, Theor. Chem. Acc., 2016, 135, 262.
41 M. Culka, F. J. Gisdon and G. M. Ullmann, in Structural
and Mechanistic Enzymology, ed. T. KarabenchevaChristova,
Elsevier Academic Press Inc, San Diego, 2017, vol. 109, pp.
77–112.
42 C. Diederich, M. Leypold, M. Culka, H. Weber,
R. Breinbauer, G. M. Ullmann and W. Blankenfeldt, Sci.
Rep., 2017, 7, 6272.
43 D. Cremer and J. A. Pople, J. Am. Chem. Soc., 1975, 97,
1354–1358.
44 A. Ardèvol and C. Rovira, J. Am. Chem. Soc., 2015, 137,
7528–7547.
45 J. Kóňa and I. Tvaroška, Chem. Pap., 2009, 63, 598–607.
46 K. S. Devulpalle, S. D. Goodman, Q. Gao, A. Hemsley and
G. Mooser, Protein Sci., 1997, 6, 2489–2493.
47 E. V. Anslyn and D. A. Dougherty, in Modern Physical
Organic Chemistry, ed. J. Murdzek, 2004, ch. 3.2, binding
forces, p. 177.
48 A. Ardèvol, J. Iglesias-Fernandez, V. Rojas-Cervellera and
C. Rovira, Biochem. Soc. Trans., 2016, 44, 51–60.
This journal is © The Royal Society of Chemistry 2018