Endocyclic cleavage in glycosides with 2,3-trans cyclic protecting groups

Article abstract of DOI:10.1021/ja201024a

An endocyclic pathway is proposed as a reaction mechanism for the anomerization from the β (1,2-trans) to the α (1,2-cis) configuration observed in glycosides carrying 2,3-trans cyclic protecting groups. This reaction occurs in the presence of a weak Lewis or Bronsted acid, while endocyclic cleavage (endocleavage) in typical glycosides was observed only when mediated by protic media or strong Lewis acids. To rationalize the behavior of this class of compounds, the reaction mechanism and the promoting factors of the endocleavage are investigated using quantum-mechanical (QM) calculations and experimental studies. We examine anomerization reactions of thioglycosides carrying 2,3-trans cyclic protecting groups, employing boron trifluoride etherate (BF₃ OEt₂) as a Lewis acid. The estimated theoretical reactivity, based on a simple model to predict transition state (TS) energies from the strain caused by the fused rings, is very close to the TS energies calculated by the TS search along the C1-C2 bond rotation after the endo C-O bond breaking. Excellent agreement is found between the predicted TS energies and the experimental reactivity ranking. The series of calculations and experiments strongly supports the predominance of the endocyclic rather than the exocyclic mechanism. Furthermore, these investigations suggest that the inner strain is the primary factor enhancing the endocleavage reaction. The effect of the cyclic protecting group in restricting the pyranoside ring to a ⁴C₁ conformation, extensively discussed in conjunction with the stereoelectronic effect theory, is shown to be a secondary factor.(Figure Presented)

Full text of DOI:10.1021/ja201024a

ARTICLE
pubs.acs.org/JACS
Endocyclic Cleavage in Glycosides with 2,3-trans Cyclic Protecting
Groups
Hiroko Satoh,*,† Shino Manabe,*,‡ Yukishige Ito,‡ Hans P. L€uthi,§ Teodoro Laino,|| and J€urg Hutter^
†National Institute of Informatics (NII), Tokyo 101-8430, Japan
‡RIKEN Advanced Science Institute, Saitama 351-0198, Japan
§Laboratory of Physical Chemistry, Swiss Federal Institute of Technology (ETH), 8093 Z€urich, Switzerland
IBM Research - Zurich, 8803 R€uschlikon, Switzerland
^Institute of Physical Chemistry, University of Zurich, 8057 Z€urich, Switzerland
S Supporting Information
b
ABSTRACT: An endocyclic pathway is proposed as a reaction
mechanism for the anomerization from the β (1,2-trans) to the
R (1,2-cis) configuration observed in glycosides carrying 2,3-
trans cyclic protecting groups. This reaction occurs in the
presence of a weak Lewis or Brønsted acid, while endocyclic
cleavage (endocleavage) in typical glycosides was observed only
when mediated by protic media or strong Lewis acids. To
rationalize the behavior of this class of compounds, the reaction
mechanism and the promoting factors of the endocleavage are
investigated using quantum-mechanical (QM) calculations and
experimental studies. We examine anomerization reactions of thioglycosides carrying 2,3-trans cyclic protecting groups, employing
boron trifluoride etherate (BF3 OEt2) as a Lewis acid. The estimated theoretical reactivity, based on a simple model to predict
3
transition state (TS) energies from the strain caused by the fused rings, is very close to the TS energies calculated by the TS search
along the C1ꢀC2 bond rotation after the endo CꢀO bond breaking. Excellent agreement is found between the predicted TS
energies and the experimental reactivity ranking. The series of calculations and experiments strongly supports the predominance of
the endocyclic rather than the exocyclic mechanism. Furthermore, these investigations suggest that the inner strain is the primary
4
factor enhancing the endocleavage reaction. The effect of the cyclic protecting group in restricting the pyranoside ring to a C1
conformation, extensively discussed in conjunction with the stereoelectronic effect theory, is shown to be a secondary factor.
1. INTRODUCTION
Carbohydrates, ubiquitously present in living-organisms, play
an important role in biological systems. To unravel the mecha-
nistic details of their reactivity, experimental and theoretical
studies have been performed extensively in the context of
glycoscience and chemical biology.1,2 Although the chemistry
of carbohydrates has been well understood, there is a long-
standing open issue regarding the hydrolytic cleavage sites in
pyranosides, known as the “exocyclic vs endocyclic question”
(Figure 1). Because of their unsymmetrical nature, pyranosides
can undergo hydrolytic cleavage through either an exocyclic or
endocyclic pathway. In the first case, the exocyclic bond, between
the anomeric carbon C1 and the exocyclic oxygen O1 (or sulfur
S1 for thioglycosides), breaks. This pathway, known also as
exocyclic cleavage or exocleavage, leads to a cyclic oxacarbenium
ion, which is the intermediate of glycosylation reactions.3ꢀ9 In
the second one, the endocyclic bond between the anomeric
carbon C1 and the endocyclic oxygen O5 in the pyranose ring
breaks. This pathway leads to an acyclic oxacarbenium ion and is
known as endocyclic cleavage or endocleavage.10ꢀ32 While the
Figure 1. Exocyclic vs endocyclic question.
endocleavage is unanimously considered to be a minor cleavage
mode for pyranosides, the exocleavage plays the role of major
reactivity channel in organic synthesis. Nonetheless, a way to
promote the endocyclic pathway is highly desirable because it will
Received: February 1, 2011
Published: March 21, 2011
r
2011 American Chemical Society
5610
dx.doi.org/10.1021/ja201024a J. Am. Chem. Soc. 2011, 133, 5610–5619
|

Journal of the American Chemical Society
ARTICLE
Figure 2. Proposed pathway of the anomerization reactions of pyranosides with 2,3-trans cyclic protecting groups. Experiments and computations
suggest the endocyclic pathway. The reaction is initiated by BF3 approaching the ring oxygen, activating the endocyclic CꢀO bond cleavage
(endocleavage). As the bond breaking proceeds, the C1ꢀC2 bond consecutively rotates, reaching the R-anomer by recyclization via the transition
state.30 Acyclic intermediates produced by endocleavage were captured in experiments.29.
allow the invention of new strategies for the highly stereoselec-
tive synthesis of several oligosaccharides.
BF3 OEt2. If this anomerization reaction really occurs via an
3
endocyclic pathway, this is the first case of endocleavage of
pyranosides under weak acidic conditions. In fact, it was generally
assumed that protic media or strong Lewis acids (e.g., SnCl4 or
Me2BBr) were necessary to activate pyranosides to undergo
endocleavage.
The general hypothesis for the anomerization starts with the
endocleavage followed by bond rotation and final cyclization
(Figure 2). Experimental and theoretical investigations strength-
en the relevancy of the predominance of endocleavage. Oscarson
and co-workers inferred an endocyclic mechanism from experi-
mental outcomes on anomerization reactions from the β- to the
The first investigation on endocleavage of pyranosides dates
back to the work on the hydrolysis of 3:6-anhydro-methyl
glucosides reported by Haworth in 1941.10 Several decades later,
the first reports on the theoretical investigations of the CꢀO
bond cleavage were published by Gorenstein et al.11 They
discussed the stereoeletronic control in carbonꢀoxygen and
phosphorusꢀoxygen bond breaking processes with ab initio
quantum-mechanical calculations. Subsequently, within an exo-
cyclic vs endocyclic debate on the hydrolysis of oligosaccharides
in a lysozyme complex,11,13,33ꢀ39 Post, Karplus, Phillipes, and co-
workers suggested an endocyclic mechanism using classical
molecular dynamics simulations. Their findings were supported
by the conformational analysis of the crystal structure of a
puckered pyranoside adopting a 4C1 conformation in the active
site of lysozyme.13,14 Afterward, Ram and Franck captured a
cation produced via endocleavage during alkyl β-THP acetal
methanolysis by an intramolecular aza-DielsꢀAlder reaction,15
while Fraser-Reid et al. submitted the experimental evidence of
an endocleavage process, taking place during the acetolysis of the
acyl-protected methyl glycosides in presence of ferric chloride.17
Only in 1994, Anslyn and co-workers examined the exocyclic vs
endocyclic problem using β-alkyl acetal probes of a cis-decalin
type, via deuterium scrambling for solvolysis (in methanol or
aqueous media).18,20 They concluded that the endocyclic path-
way exists as well as the exocyclic pathway at least for the acetal
probes. Several works have been published in subsequent years,
both theoretical and experimental, supporting the existence of an
endocyclic pathway under protic media or strong Lewis acid
conditions. Deslongchamps and Dory investigated reaction path-
ways of enzyme-catalyzed hydrolysis of glycosides based on
quantum-mechanical calculations as well as experimental
study.22 There are systematic analyses on endocleavage of glyco-
sides by Frejd et al.21 and by Murphy and co-workers.26,32
Only recently, an endocyclic pathway was proposed by
Crich,23 Oscarson,24,27 and Manabe,25,29ꢀ31 independently, as
a reaction mechanism for anomerization reactions from the β
(1,2-trans) to the R (1,2-cis) anomer,40 observed in pyranosides
carrying 2N,3O-trans-oxazolidinone groups in the presence of
R-linkage
for
2-N-acetyl-2N,3O-oxazolidinone-protected
glucosamines.27 Experimental evidence of the existence of the
endocleavage pathway in pyranosides carrying 2,3-trans carba-
mate or carbonate was demonstrated observing the adducts of
the acyclic cations produced via endocleavage, captured by
reduction, chloridation, and inter- or intramolecular Friedelꢀ
Crafts reactions.29 In theoretical studies, density functional
theory (DFT) calculations demonstrated that pyranosides, car-
rying the cyclic protecting groups, undergo endocleavage-in-
duced anomerization more easily than typical pyranosides.30
A solvent effect was also observed to play a role in the
anomerization reactions.31 In fact, in comparison with the
reaction in dichloromethane, the reaction is accelerated and
decelerated in acetonitrile and diethyl ether solvents, respec-
tively. Interestingly, it was reported that anomerization does not
occur in pyranosides with 2,3-cis carbonate.29
An open issue is the identification of the predominant factor
enhancing endocleavage in the class of pyranosides with cyclic
protecting groups. The stereoelectronic effect theory is often
used to discuss the feasibility of exocyclic vs endocyclic
cleavage.11ꢀ18,20,22,41 The preferential conformations of a pyr-
anose ring can influence the stereoelectronic contribution.
According to the stereoelectronic effect theory, an exocyclic
CꢀX (X = e.g., O, S) bond in a (pseudo)axial orientation
enhances exocleavage, whereas a CꢀX bond in a (pseudo)
equatorial orientation makes exocleavage energetically unfavor-
able. In the case of pyranosides carrying 2,3-trans cyclic protect-
ing groups, the pyranoside ring will be locked into a
5611
dx.doi.org/10.1021/ja201024a |J. Am. Chem. Soc. 2011, 133, 5610–5619

Journal of the American Chemical Society
ARTICLE
Figure 3. Set of compounds designed for the investigation. The R- and the β-anomers are represented as a and b with the structure no., respectively
(e.g., 1a and 1b).
4C1 conformation by the fused protecting ring. For the β-anomer
in this conformation, the exocyclic CꢀX bond is in an equatorial
orientation. Therefore, the stereoelectronic contribution has a
negative effect on exocleavage and a positive effect on endoclea-
vage. Thus the 2,3-trans fused ring stabilizing a 4C1 conformation
of the pyranoside ring will promote endocleavage. However, the
series of compounds studied here undergo anomerization even
under very mild reaction conditions, in some cases showing a
perfect anomerization in the presence of a weak Lewis acid
at ꢀ30 ꢀC. In contrast, the maximum yield of the R-product
reported for other known pyranoside compounds is even at room
temperature only 30%.15,18,20 The high reactivity of the series of
pyranosides with cyclic protecting groups indicates the existence
of other contributing factors.
Based upon detailed investigations on the geometries and
experimental observations, we present a model in which the inner
strain energy, caused by the two fused rings, is assumed to be the
predominant factor responsible for the promotion of endoclea-
vage. This model allows us to formulate a simple expression to
predict TS energies from the inner strain of the cyclic protecting
group and from the TS anomerization energy of a reference
pyranoside without a 2,3-trans cyclic protecting group. The
predicted values from the inner strain show a good agreement
with the TS energies optimized by quantum mechanical meth-
ods. The trend of the predicted TS energies agrees with the
observed experimental reactivity ranking. These results indicate
that the reactivity can be determined from the inner strain
considered as the predominant factor. In the present article, we
describe the theoretical and experimental studies that identify the
predominant factor and lead to the definition of the reactivity
predictor.
protecting groups (7ꢀ9) were added to identify the influence of
the ring size and the effect of the carbonyl group in the cyclic
protecting groups. The pyranoside rings of 2ꢀ9 are restrained to
a 4C1 conformation by the 2,3-trans fused ring (2ꢀ8) or by the
trans-decalin structure (9).
2. COMPUTATIONAL METHODS
All calculations were performed using density functional theory
(DFT) at the B3LYP/6-31G(d,p) level42,43 with the Gaussian 03
program.44 The model structures 1ꢀ9 were generated via full geometry
optimization in the gas phase for the anomeric isomers (anomers) R
(1aꢀ9a) and β (1bꢀ9b). Optimizations were initiated from an ideal
4C1 conformation (Figure 3). The X-ray structures measured for 3b and
4b are in excellent agreement with the calculated geometries.28,29 This
fact demonstrates the high quality of the geometries from the DFT
calculations.
We searched transition states (TS) in the gas phase for the glycosides
1ꢀ4, 6, and 9, restricting the search along the C1ꢀC2 torsion angle θ
(H1ꢀC1ꢀC2ꢀH2) as a reaction coordinate.30 The final TS structures
were located using the STQN (Synchronous Transit-Guided Quasi-
Newton) method45,46 by using the QST3 option with three molecule
specifications of the β-anomer (reactant), the R-anomer (product), and
an optimized TS candidate. The results from the TS theory were
validated by comparing them to experimental product yields of the R-
and β-anomers. The order of the potential energies of the β-anomer
(reactant), the TS structure, and the R-anomer (product) are preserved
in a dichloromethane (CH2Cl2) solvent as estimated with an IEF-PCM
(Integral-Equation Formation-Polarizable Continuum Model)47 calcu-
lation by using the default parameter of Gaussian 03 for this solvent
(dielectric permittivity ε = 8.93; see the Supporting Information for
details of the results).
The model set of compounds designed for the present
investigations consists of a common moiety of 4-O-acetyl-6-O-
benzyl-thiophenylglycosides (Figure 3). A typical pyranoside,
3,4-diacetyl-N-acetylglucosamine (1) was selected as the refer-
ence. This does not carry any cyclic protecting group and does
not undergo anomerization reaction, either. Out of the pyrano-
sides with cyclic protecting groups that have been thoroughly
experimentally studied, we chose the 2N,3O-trans-N-substituted
oxazolidinone protected glucosamines (2ꢀ5) and a glycoside
with 2,3-trans-carbonate (6). Pyranosides carrying six-membered
Pilot calculations of the energy profile associated with the endoclea-
vage reaction were performed for 1b and 3b in the gas phase. We
performed constraint geometry optimization, starting from the opti-
mized structures and progressively elongating, in steps of 0.01 nm, the
C1ꢀO5 bond from 0.15 nm to 0.27 and to 0.28 nm for 1b and 3b,
respectively. The final bond lengths are close to the bond lengths at TS
(0.269 and 0.278 nm). The amount of energy required to elongate the
C1ꢀO5 bond (131.2 and 91.9 kJ molꢀ1 for 1b and 3b, respectively) is
3
very close to the activation energy from the β-anomer to the TS (159.5
and 83.6 kJ molꢀ1 for 1 and 3, respectively). The resulting energy
3
5612
dx.doi.org/10.1021/ja201024a |J. Am. Chem. Soc. 2011, 133, 5610–5619

Journal of the American Chemical Society
ARTICLE
Figure 4. Model to predict TS energies from inner strain.
profile calculation along the C1ꢀC2 bond rotation after the elongation
of the C1ꢀO5 bond and using a constraint on the C1ꢀO5 distance
(0.27 and 0.28 nm for 1 and 3), reached states close to the TS via a
rather flat potential energy region (from ꢀ156.2 to ꢀ35.6 degrees, and from
ꢀ158.8 to ꢀ28.8 degrees, respectively). The results indicate that C1ꢀC2
bond rotation is almost free after endocleavage. The important factors
enhancing the reactivity should therefore predominantly influence the
initiation of the bond breaking rather than the C1ꢀC2 bond rotation.
On the other hand, the pyranoside rings with a 2,3-trans cyclic
protecting group attached (2ꢀ6, and 8), which undergo anomerization
reaction, were found to be slightly deformed compared with the
reference pyranoside 1. These observations lead us to the assumption
that the strain caused by the fused ring is a primal trigger in activating
endocleavage. We formulated a simple model to predict TS energies
from inner strain (Figure 4). The model assumes that the series of
pyranosides have almost the same TS structures and that the TS relative
to the β-anomer (reactant) appear to be lower in energy because the
inner strain destabilizes the reactants. This model is stimulated by the
so-called extended activation strain model developed by Bickelhaupt
et al.48ꢀ53 In this elegant approach to understanding palladium-induced
bond activation, the activation energy is decomposed into the activation
strain of the reactants and the stabilizing TS interaction between the
reactants. Although their concept for intermolecular reaction systems
cannot be applied to our intramolecular anomerization reaction directly,
we followed the spirit of the model.
energy (Epyranose ), computed by a constrained geometry optimization
(where we constrained all atomic coordinates except the methoxy
fixed
groups), and the potential energy (Epyranose ), obtained by a full
relaxed
relaxation, is assigned to the inner strain (upper scheme of Figure 4):
ΔEpyranose ¼ E
þ E
pyranosefixed
pyranoserelaxed
In the calculation of ΔEfused, C1 and C4 are changed to methyl groups
after eliminating the partial structure including C5, C6, and endo-oxygen
(O5). Similarly to the term described above, the difference between the
potential energy (Efused ), computed by a constrained geometry
optimization (where we ficxoednstrained all atomic coordinates except the
hydrogen atoms of the methyl groups), and the potential energy
(Efused ), obtained by a full relaxation, is assigned to the inner strain
relaxed
(lower scheme of Figure 4):
ΔEfused ¼ E
þ E
fusedfixed
fusedrelaxed
The predicted activation energy (ΔE‡predict) is finally defined by the sum
of those relaxation energies ΔEpyranose and ΔEfused (usually negative)
and the TS energy of the reference system (ΔE‡ref).
ΔE‡
¼ ΔEpyranose þ ΔEfused þ ΔE‡
predict
ref
The inner strain energy ΔEstrain provides a simple relaxation indicator
caused by endocleavage for a static conformation. The definition does
not include the statistical distribution of several conformers of a
pyranose ring. The effect of the inner strain caused by the two fused
rings, independent of the effect of the conformational distribution by the
fused ring restraining the pyranose ring, is an important point to discuss.
Using this model, the TS energies were predicted for pyranosides
With our new method for the pyranoside system, the activation
energy (ΔE‡predict; the potential energy of TS relative to the β-anomer)
is decomposed into the inner strain energy (ΔEstrain) and the activation
energy of a reference pyranoside (ΔE‡ref; the potential energy of TS
relative to the β-anomer).
2ꢀ9. The TS energy relative to the β-anomer for 1 (139.0 kJ molꢀ1
)
3
was used as the reference activation energy ΔE‡
.
ref
ΔE‡
¼ ΔEstrain þ ΔE‡
The validation of the model was performed by comparing the
predicted TS energies with those calculated via TS search along the
C1ꢀC2 torsion angle. The results were further assessed using the
experimentally observed reactivity ranking.
predict
ref
The inner strain energy is decomposed into a contribution from the
pyranose ring (ΔEpyranose) and one from the fused ring (ΔEfused). Both
contributions are calculated as relaxation energies of one ring if the other
ring is removed.
3. RESULTS AND DISCUSSION
ΔEstrain ¼ ΔEpyranose þ ΔEfused
3.1. TS Analysis. Figure 5 shows the potential energies of the TS
as well as of the R-anomer relative to the β-anomer for structures
1ꢀ4, 6, and 9 (Figure 3) obtained by TS search along the C1ꢀC2
In the calculation of ΔEpyranose, the protecting groups at C2 and C3 are
changed to methoxy groups. The difference between the potential
5613
dx.doi.org/10.1021/ja201024a |J. Am. Chem. Soc. 2011, 133, 5610–5619

Journal of the American Chemical Society
ARTICLE
Figure 5. Results of the TS search for 1ꢀ4, 6, and 9 along the C1ꢀC2 torsion angle as a reaction coordinate. For the pyranosides 2ꢀ4, and 6, which
anomerize from the β- to the R-anomer, transition states appear at lower potential energies (solid lines). On the contrary, 1 and 9, which experimentally
do not undergo anomerization reactions, need to overcome higher TS energies (broken lines). The TS geometries are shown with the values of the
C1ꢀO5 bond length (in nm) and the H1ꢀC1ꢀC2ꢀH2 torsion angle (in degrees).
bond rotation coordinate in the gas phase. The yields of the R-andβ-
Table 1. Experimental Yields of the r- and the β-Anomers,
anomers observed in experiments for 2ꢀ9 with BF3 OEt2 (2 equiv)
with BF3 OEt2 (2 equiv) in CH2Cl2 Solution at ꢀ30 ꢀC for
3
3
in a CH2Cl2 solvent at ꢀ30 ꢀC for 12 h are summarized in Table 1.
Transition states are located at around θ (H1ꢀC1ꢀC2ꢀH2)
= ꢀ30 ∼ ꢀ45 degrees (Figure 5). The TS of pyranosides with a
five-membered fused ring at the 2,3-trans position (2ꢀ4 and 6;
solid lines) are found to be lowest in energy (94.3, 83.6, 76.4, and
12 ha
76.9 kJ molꢀ1). In the experiments, anomerization reaction was
3
observed for all of these pyranosides (the R:β ratio in % is 19:69,
63:16, 75:0, and 14:46). On the contrary, the TS of pyranosides 1
and 9, which do not undergo anomerization, are much higher in
energy (139.0 and 119.7 kJ mol; broken lines). Furthermore, the
3
ordering of TS energies is in good agreement with the observed
yields of the R-product. However, in comparison with the
experiments, the TS of the pyranoside with carbonate 6 is
estimated to be lower in energy. The C1ꢀO5 bond lengths of
the transition structures range from 0.269 to 0.284 nm. The energy
calculations in a CH2Cl2 solvent estimated with an IEF-PCM
scheme preserved the ordering of the potential energies, as men-
tioned in Section 2 (see Supporting Information for the details of
the results). These results demonstrate that the TS analysis allows
the qualitative estimation of the reactivity of anomerization.
product yield (ratio)
reactant
entry
(β-anomer)
R-anomer 1aꢀ9a [%]
β-anomer 1bꢀ9b [%]
1
2
3
4
5
6
7
8
9
1b
2b
3b
4b
5b
6b
7b
8b
9b
0
19
63
75
80
14
0
93
69
16
0
4
0
For all compounds studied, the pyranoside rings keep a C1
conformation along the reaction coordinate. According to the
46
92
82
61
stereoelectronic effect theory, the endo CꢀO bond in the β-
4
anomer adopting the C1 conformation is advantageous for
17
endocleavage.12,13,16,19 This means that endocleavage of com-
pounds 1b-9b is favored by the 4C1 conformation. However, they
are split into the anomerizing and nonanomerizing groups
showing a clear difference of the TS in energy. This indicates
0
a Entries 1ꢀ3 (1ꢀ3) and 6 (6) are reported in the literature.29ꢀ31
Entries 4ꢀ5 (4ꢀ5) and 7ꢀ9 (7ꢀ9) were carried out for the present
study (see Supporting Information for the experimental details).
5614
dx.doi.org/10.1021/ja201024a |J. Am. Chem. Soc. 2011, 133, 5610–5619

Journal of the American Chemical Society
ARTICLE
Figure 6. Deformation of pyranoside rings of 2bꢀ9b compared to 1b (4C1).
the existence of an important factor other than the stereoelec-
tronic effect enhanced by the 4C1 conformation.
3.2. Deformation of the Pyranoside Rings in the Reactants
(the β-Anomers). The torsion angles γ1ꢀ6 of the pyranoside ring
of 2bꢀ9b relative to 1b are shown along with their structures in
Figure 6. As can be seen from the geometries, the pyranoside rings
4
all adopt a C1 conformation. However, slight deformation was
found in 2bꢀ6b, and 8b, for which anomerization reactions were
experimentally observed from the β- to the R-anomer. Pyranosides
7b and 9b, which do not anomerize, show almost nodifference. For
2bꢀ6b, and 8b, the largest deformation, by more than 10 degrees,
appeared around the C2ꢀC3 bond, γ3 (C1ꢀC2ꢀC3ꢀC4),
where the cyclic protecting group is attached. In the course of
the TS search along the C1ꢀC2 bond rotation, those fused rings of
2bꢀ6b, and 8b adopt a more planar conformation after endoclea-
vage. Such a transition is not observed for the pyranosides 7b and
9b. These observations led us to assume that the strain from the
fused ring preferring to be in a more planar conformation is a key
factor for the enhancement of the reactivity of endocleavage.
3.3. Prediction of TS Energies from Inner Strain. Using the
inner strain energy caused by the two fused rings (the pyranoside
ring and the cyclic protecting group) defined in Figure 4, we derived
TS energies ΔE‡predict. Figure 7 shows the comparison between the
predicted TS energies ΔE‡predict and the energies of the transition
states located by the TS search ΔE‡ (Figure 5) for 2ꢀ4, 6, and 9.
The prediction model based on inner strain successfully estimates
TS energies (ΔE‡predict) very close to those from the TS search
(ΔE‡). The difference between the ΔE‡predict and ΔE‡ values ranges
Figure 7. Validation of the predicted TS energies from inner strain
values are compared with the energies
predict
calculated by the TS search ΔE‡.
ΔE‡predict. The ΔE‡
4ꢀ5ꢀ6ꢀ3ꢀ2ꢀ8ꢀ9ꢀ7, shown in the left chart in Figure 8.
The product ratios (β:R in %) in experiments with BF3 OEt2
3
(2 equiv.) in CH2Cl2 solution at ꢀ30 ꢀC for 12 h (Table 1) are
compared with the theoretically predicted reactivity, shown in
the right chart in Figure 8, in which the white and black bars
represent the ratios of the β- and R-anomers, respectively.
The difference in the predicted TS energy splits the model set
of compounds into two groups. One is the group showing
ΔE‡predict values lower than 90 kJ molꢀ1 (2, 3, 4, 5, 6, and 8).
3
The other group consists of 7 and 9, showing ΔE‡predict values
from 0.3 to 5.4 kJ molꢀ1 (for9and 4, respectively). The ordering of
3
higher than 120 kJ molꢀ1. We can see that, this grouping is in
the TS energies stays the same: 4ꢀ6ꢀ3ꢀ2ꢀ9.
3
The predicted TS energies ΔE‡
for 2ꢀ9 are validated
good agreement with the experiments: anomerization reaction
occurs in the former group while it does not in the latter group.
Good agreement is also found in the ordering between the
ΔE‡predict values and the experimental reactivity estimated based
on the product ratios. Again, there is only one exception for the
predict
using the experimental data in Table 1. Figure 8 graphically
shows the comparison between the prediction and the experi-
predict
ments. The results are sorted in ascending order of the ΔE‡
values, that is, the ordering is from higher to lower reactivity:
5615
dx.doi.org/10.1021/ja201024a |J. Am. Chem. Soc. 2011, 133, 5610–5619

Journal of the American Chemical Society
ARTICLE
Figure 8. Validation of the predicted TS energies from inner strain. The ordering of TS energies predicted from the inner strain (left side) is compared
with the experimental product ratios in Table 1 (right side).
Table 2. Experimental Yields of the r- and the β-Anomers for 4 and 5, with BF3 OEt2 (2 equiv) in CH2Cl2 Solution at ꢀ50 ꢀCa
3
product yield (ratio)
entry
reaction time [hours]
reactant (β-anomer)
R-anomer 4aꢀ5a [%]
β-anomer 4bꢀ5b [%]
1
2
3
4
2
4
2
4
4b
4b
5b
5b
67
96
0
19
0
97
95
0
a See Supporting Information for the experimental details.
pyranoside with the carbonate protecting group 6. The prediction
overestimates the reactivity of 6 compared to the experiment.
For 4 and 5, the lowest ΔE‡predict values were obtained, suggesting
cyclic protecting group attached, only 8b, carrying the N-acetyl
substituted cyclic protecting group, underwent anomerization to
the R-anomer. The difference in the reactivity observed for 8
compared with 7 and 9 is also reflected by a large difference in the
high reactivity (71.1 (4) and77.0(5) kJ molꢀ1). In the experiments,
3
these pyranosides anomerize perfectly, giving only the R-product in
high yield at ꢀ30 ꢀC for 12 h (75% and 80% for 4and 5, respectively,
shown in Table 1). Under these reaction conditions, the yield of the
R-product of 4 is 5% lower than 5 because of the lower solubility of
the R-product of 4, but the experiments at lower temperature
ꢀ50 ꢀC confirmed that 4 is more reactive than 5 (Table 2).
Therefore, we conclude that the predicted theoretical reactivity for
these pyranosides is in excellent agreement with the experiments.
Experiments demonstrate that the N-acetyl substituted cyclic
protecting groups act as a strong promoter of the reaction (e.g., 4
and 8). Out of the pyranosides 7bꢀ9b, with a six-membered
ΔE‡predict values (89.9 (8), 124.9 (7) and 123.0 (9) kJ molꢀ1).
3
We want to further discuss the possibility of stereoelectronic
effects related to the preferential conformation of pyranoside
rings. The DFT calculations predict that the pyranoside ring is
kept in a 4C1 conformation along the reaction coordinate and the
TS prediction was performed based only on the 4C1 conforma-
tion for all pyranosides. The preference of a 4C1 conformation is
reasonable from a chemical point of view: in structures 2ꢀ8, the
pyranoside rings are restrained by the 2,3-trans cyclic protecting
groups, while in structure 9 the 4C1 conformation is stabilized by
the double anomeric effects in the trans-decalin. The X-ray
5616
dx.doi.org/10.1021/ja201024a |J. Am. Chem. Soc. 2011, 133, 5610–5619

Journal of the American Chemical Society
ARTICLE
Figure 9. Energy profile associated with the exo vs endo cleavage. The amount of energy required to elongate the CꢀO bond from the ground state to
the length at TS on the endocyclic pathway is calculated. The units of the length and the energy are presented in nm and kJ molꢀ1, respectively.
3
structures available for 3b and 4b are well reproduced by the
calculations. If the stereoelectronic effect caused by the C1
exo CꢀO bond from 0.140 to 0.280 nm is 135.0 kJ molꢀ1, ∼25
3
4
kJ molꢀ1 larger than that for the endo CꢀO bond elongation.
3
conformation was the predominant factor enhancing endoclea-
vage, then a similarly high preference for endocleavage would be
observed for all compounds. However, the TS prediction shows,
in good agreement with the experiments, a clear variation in
reactivity. The reactivity ordering is: protecting groups with an
N-acetyl substituted 5-membered ring (4, 5) > with a 5-mem-
bered ring (2, 3, 6) > with an N-acetyl substituted 6-membered
ring (8) . with a 6-membered ring (7, 9) > without a cyclic
protecting group (1). This ordering correlates with the stiffness of
This means that endo cleavage is energetically favored in the
pyranoside with the 2,3-trans cyclic protecting group. In contrast,
higher energies are required for both of the elongations in the
typical pyranoside 11. For this pyranoside, the amount of energy
required to elongate the endo CꢀO bond from 0.144 to 0.270 nm
is 137.5 kJ molꢀ1, while the amount of energy required to
3
elongate the exo CꢀO bond from 0.140 to 0.270 nm is 123.1
kJ molꢀ1, ∼15 kJ molꢀ1 smaller than that for the endo CꢀO
3
3
bond elongation. This means that endo cleavage is energetically
less favorable in the pyranoside in the absence of a cyclic
protecting group attached. These energy profiles also indicate
that, for the pyranoside 11, both cleavages would not be observed
in mild conditions. As compared with 11, in the cyclically
protected 10, the endo cleavage should be activated in mild
conditions; however, it would be difficult for the exo cleavage to
take place. The torsion angle γ3 (C1ꢀC2ꢀC3ꢀC4) of the
optimized structures of 10 and 11 were ꢀ52.3 and ꢀ67.2
degrees, respectively. These results can be interpreted that exo
cleavage is energetically favorable in the pyranoside with more
planar γ3 torsion angle. This trend is consistent with the findings
by Crich and co-workers: the oH5 half-chair conformation,
forming planar γ3, reduces the energy barrier to cyclic oxacarbe-
nium ion formation.8,9,54
4
the C1 pyranoside ring conformation, but calculations and
experiments show that this is not enough to strongly enhance
endocleavage. These results lead to the conclusion that the inner
strain is the predominant factor enhancing the endocleavage-
induced anomerization reactions. The stereoelectronic effect due
to the restrained 4C1 conformation is only a secondary factor.
3.4. Energy Profile of Exocyclic vs Endocyclic Cleavage.
The experimental and theoretical investigations suggest that the
anomerization reaction observed for this class of pyranosides
occurs predominantly via the endocyclic pathway rather than the
exocyclic pathway. In order to corroborate this assumption, we
calculated the referential energy profile in the gas phase asso-
ciated with the exo or endo cleavage for O-glycosides 10 and 11
(Figure 9).
Although mechanistic details of the exocyclic pathway as well
as analyses in solution would allow the investigations of more
precise behavior of the pyranosides, these simple pilot calcula-
tions support the relevancy of the predominance of the endo-
cleavage in the anomerization reactions of pyranosides carrying
2,3-trans cyclic protecting groups. The inner strain caused by the
fused rings is found to be the predominant factor to enhance
the endo cleavage. However, this does not mean to rule out
the possibility of the existence of other contributing factors. The
fused rings can have an influence on the geometrical and
electronic structures of pyranosides to diminish the effects
enhancing exo cleavage,55 for example, the ease of flattering of
γ3 (C1ꢀC2ꢀC3ꢀC4), remote participation from a protecting
group (e.g., at O3)56, and stereoelectronic contributions.
Our computations suggest that, for pyranosides carrying 2,3-
trans cyclic protecting groups, the barrier to endo cleavage ΔEendo
The reference structures were generated via full geometry
optimization, initiated from an ideal C1 conformation. The exo
4
CꢀO bond length is 0.140 nm for the optimized structures of 10
and11. Theendo CꢀO bond lengths are 0.145 and 0.144 nm for 10
and 11, respectively. The full geometry optimization preserved the
4C1 conformation of the pyranoside rings. The energy profiles were
then calculated by constrained geometry optimization, progres-
sively elongating the exo or endo CꢀO bond (the C1ꢀO1 or
C1ꢀO5 bond), in steps of 0.02 nm, from 0.15 to 0.28 nm for 10
and to 0.27 nm for 11. The final bond lengths are close to the endo
CꢀO bond lengths at TS (0.279 and 0.263 nm), found in the
endocyclic pathway along the C1ꢀC2 bond rotation.28
These energy profiles confirm the predominance of the endo
cleavage of the pyranoside carrying a cyclic protecting group at
the 2,3-trans position (Figure 9). For the pyranoside with the
cyclic protecting group 10, the amount of energy required to
elongate the endo CꢀO bond from 0.145 to 0.280 nm is 106.1
is approximately 20ꢀ40 kJ molꢀ1 smaller than the barrier to exo
3
kJ molꢀ1, while the amount of energy required to elongate the
cleavage ΔEexo. In contrast, for normal pyranosides, ΔEendo is
3
5617
dx.doi.org/10.1021/ja201024a |J. Am. Chem. Soc. 2011, 133, 5610–5619

Journal of the American Chemical Society
ARTICLE
estimated to be larger than ΔEexo with about a similar difference
in energy. The theoretical model is in good agreement with
experiments, which show, that under the selected mild reaction
NMR and 13C NMR spectral characteristics. This material is
available free of charge via the Internet at http://pubs.acs.org.
conditions using the week Lewis acid BF3 OEt2, the pyranosides
’ AUTHOR INFORMATION
3
carrying 2,3-trans cyclic protecting groups undergo an endoclea-
vage-induced anomerization reaction. If stronger conditions are
employed, e.g., a strong acid and/or higher temperature, these
substrates can undergo anomerization also via the exocyclic
pathway. Under these conditions, the normal pyranosides will
undergo exocleavage-induced anomerization reaction. Employ-
ing a glycosylation promoter specifically attacking the sulfur at
the anomeric site (S1) will also reduce the barrier to exo cleavage.
This results in fast anomerization reactions via the exocyclic
pathway, as reported by Boons and Stauch employing IDCP
(iodonium dicollidine perchlorate) as a promoter.57 Finally, our
findings suggest that a careful selection of substrates, promoters,
as well as reaction conditions will allow the control of the
reactivity channel to either the exocyclic or endocyclic pathway.
These findings will contribute to create new strategies for highly
stereoselective synthesis to diversify oligosaccharides by control-
ling anomerization reactions.
Corresponding Author
hsatoh@nii.ac.jp; smanabe@riken.jp
’ ACKNOWLEDGMENT
The present calculations were performed on Obelix, a com-
puter cluster of the Competence Center for Computational
Chemistry (C4). H.S. and J.H. gratefully thank the Swiss National
Science Foundation for funding this research with the Interna-
tional Short Visits Program. S.M. was supported by a Grant-
in-Aid for Scientific Research (C) (Grant No. 21590036) from
the Japan Society for Promotion of Science. S.M. thanks
Ms. Akemi Takahashi for her technical assistance. We thank
Professor David M. Birney (Texas Tech University, Lubbock,
TX) and Professor Andrea Vasella (ETH Zurich, Switzerland)
for valuable discussions.
’ REFERENCES
(1) Stallforth, P.; Lepenies, B.; Adibekian, A.; Seeberger, P. H.
J. Med. Chem. 2009, 52, 5561–5577.
(2) Seeberger, P. H. Nat. Chem. Biol. 2009, 5, 368–372.
(3) Crich, D.; Chandrasekara, N. S. Angew. Chem., Int. Ed. 2004,
43, 5386–5389.
(4) Galoniꢀc, D. P.; Gin, D. Y. Nature 2007, 446, 1000–1007.
(5) Boltje, T. J.; Buskas, T.; Boons, G.-J. Nature Chem. 2009,
1, 611–622.
(6) Walvoort, M. T. C.; Dinkelaar, J.; van den Bos, L. J.; Lodder, G.;
Overkleeft, H. S.; Codꢀee, J. D. C.; Van der Marel, G. A. Carbohydr. Res.
2010, 345, 1252–1263.
4. CONCLUSIONS
Regarding anomerization reaction of the series of 4-O-acetyl-
6-O-benzyl-thiophenylglycosides carrying 2,3-trans cyclic pro-
tecting groups, we formulated a simple model to predict TS
energies from inner strain caused by the fused rings. Our model is
an adaptation of Bickelhaupt’s extended activation strain model
for bimolecular reactions to unimolecular reaction systems. The
predicted TS energy serves as a good reactivity predictor of
the anomerization reaction from the β- to the R-anomer. The
predicted reactivity is in a good agreement with the reactivity
estimated from the TS search along the C1ꢀC2 bond rotation
coordinate, and also with the experiments. The energy profile
calculations to estimate the feasibility of the elongation of the
endo- or exo CꢀO bond indicate that endocleavage is energeti-
cally preferred in the pyranosides with 2,3-trans cyclic protecting
groups. These series of investigations strongly support the
endocyclic pathway as a reaction mechanism of the anomeriza-
tion. The results lead to an important conclusion: the inner strain
caused by the fused rings (the pyranoside ring and the cyclic
protecting group) is the predominant factor for the enhancement
of the endocleavage-induced anomerization. The stereoelectro-
nic contribution to the endocleavage induced by the restrained
conformation of pyranoside rings is a secondary factor. This
inner strain model acts as a qualitative reactivity predictor, which
can be calculated ∼30 times faster than the TS search. This
predictor is expected to be useful for further investigations, such
as the virtual screening of compounds, possibly even at a
molecular mechanics level. Furthermore, it will be an important
concept in the discussion on the exocyclic vs endocyclic debate.
(7) Mydock, L. K.; Demchenko, A. V. Org. Biomol. Chem. 2010,
8, 497–510.
(8) Crich, D. Acc. Chem. Res. 2010, 43, 1144–1153.
(9) Bohꢀe, L.; Crich, D. C. R. Chim. 2011, 14, 3–16.
(10) Haworth, W. N.; Owen, L. N.; Smith, F. J. Chem. Soc.
1941, 88–102.
(11) Gorenstein, D. G.; Findlay, J. B.; Luxon, B. A.; Kar, D. J. Am.
Chem. Soc. 1977, 99, 3473–3479.
(12) Kirby, A. J. Acc. Chem. Res. 1984, 17, 305–311.
(13) Post, C. B.; Karplus, M.
J. Am. Chem. Soc. 1986,
108, 1317–1319.
(14) Post, C. B.; Brooks, B. R.; Karplus, M.; Dobson, C. M.;
Artymiuk, P. J.; Cheetham, J. C.; Phillips, D. C. J. Mol. Biol. 1986,
190 (3), 455–479.
(15) Gupta, R. B.; Franck, R. W. J. Am. Chem. Soc. 1987,
109, 6554–6556.
(16) Guindon, Y.; Anderson, P. C. Tetrahedron Lett. 1987,
28, 2485–2488.
(17) McPhail, D. R.; Lee, J. R.; Fraser-Reid, B. J. Am. Chem. Soc.
1992, 114, 1905–1906.
(18) Liras, J. L.; Anslyn, E. V. J. Am. Chem. Soc. 1994,
116, 2645–2646.
(19) Deslongchamps, P.; Dory, Y. L.; Li, S. Can. J. Chem. 1994,
72, 2021–2027.
’ ASSOCIATED CONTENT
(20) Liras, J. L.; Lynch, V. M.; Anslyn, E. V. J. Am. Chem. Soc. 1997,
119, 8191–8200.
S
Supporting Information. Complete reference descrip-
b
tions of ref 44, the number of imaginary frequencies for the
transition states, geometries of the optimized R-and β-anomers
and transition structures, transition state energies in a CH2Cl2
solvent using an IEF-PCM, the strain energies ΔEpyranose
ΔEfused, and ΔEstrain, the experimental procedures, and the H
(21) Olsson, R.; Berg, U.; Frejd, T.
Tetrahedron 1998,
54, 3935–3954.
(22) Deslongchamps, P.; Li, S.; Dory, Y. L. Org. Lett. 2004,
6, 505–508.
(23) Crich, D.; Vinod, A. U. J. Org. Chem. 2005, 70, 1291–1296.
,
1
5618
dx.doi.org/10.1021/ja201024a |J. Am. Chem. Soc. 2011, 133, 5610–5619

Journal of the American Chemical Society
ARTICLE
(24) Boysen, M.; Gemma, E.; Lahmann, M.; Oscarson, S. Chem.
Commun. 2005, 3044–3046.
(25) Manabe, S.; Ishii, K.; Ito, Y. J. Am. Chem. Soc. 2006,
128, 10666–10667.
(26) O’Brien, C.; Polꢀakovꢀa, M.; Pitt, N.; Tosin, M.; Murphy, P. V.
Chem.—Eur. J. 2007, 13, 902–909.
(27) Olsson, J. D. M.; Eriksson, L.; Lahmann, M.; Oscarson, S.
J. Org. Chem. 2008, 73, 7181–7188.
(28) Manabe, S.; Ishii, K.; Hashizume, D.; Ito, Y. Acta Crystallogr.
2008, E64, o1868.
(29) Manabe, S.; Ishii, K.; Hashizume, D.; Koshino, H.; Ito, Y. Chem.—
Eur. J. 2009, 15, 6894–6901.
(30) Satoh, H.; Hutter, J.; L€uthi, H. P.; Manabe, S.; Ishii, K.; Ito, Y.
Eur. J. Org. Chem. 2009, 1127–1131.
(31) Manabe, S.; Ito, Y. Tetrahedron Lett. 2009, 50, 4827–4829.
(32) Pilgrim, W.; Murphy, P. V. J. Org. Chem. 2010, 75, 6747–6755.
(33) Chipman, D. M.; Sharon, N. Science 1969, 165, 454–465.
(34) Ford, L. O.; Johnson, L. N.; Machin, P. A.; Phillips, D. C.; Tjian,
R. J. Mol. Biol. 1974, 88, 349–360.
(35) Sinnott, M. L. Chem. Rev. 1990, 90, 1171–1202.
(36) Strynadka, N. C. J.; James, M. N. G. J. Mol. Biol. 1991,
220, 401–424.
(37) Hadfield, A. T.; Harvey, D. J.; Archer, D. B.; Mackenzie, D. A.;
Jeenes, D. J.; Radford, S. E.; Lowe, G.; Dobson, C. M.; Johnson, L. N.
J. Mol. Biol. 1994, 243, 856–872.
(38) McCarter, J.D.;Withers,S.G. J. Am. Chem. Soc. 1996, 118, 241–242.
(39) Bols, M. Acc. Chem. Res. 1998, 31, 1–8.
(40) The stereo-descriptors R and β are adopted to notate the
stereochemistry at the anomeric site as defined by IUPAC conventions.
The stereoselectivity is often described using relative stereo-descriptors
to describe the configuration at the anomeric carbon, R and β, instead of
the notation 1,2-cis and 1,2-trans. The R descriptor is used for the case
where the OH at the anomeric carbon is on the same side as the OH at
the reference atom in the Fischer projection, and the β descriptor is used
for the opposite side. In the case of glucopranosides, the 1,2-cis and 1,2-
trans positions are defined as the R and β configurations for the
anomeric carbon, respectively. For more details, see the IUPAC
definition at http://www.chem.qmul.ac.uk/iupac/2carb/06n07.html
(accessed September 1, 2010).
(41) Kirby, A. J. Stereoelectronic Effects; Oxford University Press:
New York, 1996.
(42) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(43) Stephenes, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J.
J. Phys. Chem. 1994, 98, 11623–11627.
(44) Frisch, M. J.; et al. Gaussian 03, Revision C.02; Gaussian, Inc.:
Wallingford, CT, 2004.
(45) Peng, C.; Schlegel, H. B. Israel J. Chem. 1993, 33, 449–454.
(46) Peng, C.; Ayala, P. Y.; Schlegel, H. B.; Frisch, M. J. J. Comput.
Chem. 1996, 17, 49–56.
(47) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005,
105, 2999–3093.
(48) Bickelhaupt, F. M. J. Comput. Chem. 1999, 20, 114–128.
(49) Diefenbach, A.; Bickelhaupt, F. M. J. Chem. Phys. 2001,
115, 4030–4040.
(50) Diefenbach, A.; de Jong, G. T.; Bickelhaupt, F. M. Mol. Phys.
2005, 103, 995–998.
(51) Diefenbach, A.; de Jong, G. T.; Bickelhaupt, F. M. J. Chem.
Theory Comput. 2005, 1, 286–298.
(52) de Jong, G. T.; Bickelhaupt, F. M. Chem. Phys. Chem. 2007,
8, 1170–1181.
(53) van Zeist, W.-J.; Visser, R.; Bickelhaupt, F. M. Chem.—Eur. J.
2009, 15, 6112–6115.
(54) Crich, D.; Vinod, A. U.; Picione, J. J. Org. Chem. 2003,
68, 8453–8458.
(55) Demchenko, A. V. Synlett 2003, 1225–1240.
(56) Baek, J. Y.; Lee, B.-Y.; Jo, M. G.; Kim, K. S. J. Am. Chem. Soc.
2009, 131, 17705–17713.
(57) Boons, G.-J.; Stauch, T. Synlett 1996, 906–908.
5619
dx.doi.org/10.1021/ja201024a |J. Am. Chem. Soc. 2011, 133, 5610–5619