Thermodynamic origin of selective binding of β-cyclodextrin derivatives with chiral chromophoric substituents toward steroids

Article abstract of DOI:10.1021/jp105821s

Two β-cyclodextrin derivatives with chiral chromophoric substituents, that is, l- (1) and d-tyrosine-modified β-cyclodextrin (2), were synthesized and fully characterized. Their inclusion modes, binding abilities, and molecular selectivities with four ste

Full text of DOI:10.1021/jp105821s

J. Phys. Chem. B 2010, 114, 16147–16155
16147
Thermodynamic Origin of Selective Binding of ꢀ-Cyclodextrin Derivatives with Chiral
Chromophoric Substituents toward Steroids†
Yong Chen, Fang Li, Bo-Wen Liu, Bang-Ping Jiang, Heng-Yi Zhang, Li-Hua Wang, and
Yu Liu*
Department of Chemistry, State Key Laboratory of Elemento-Organic Chemistry, Nankai UniVersity,
Tianjin 300071, People’s Republic of China
ReceiVed: June 23, 2010
Two ꢀ-cyclodextrin derivatives with chiral chromophoric substituents, that is, L- (1) and D-tyrosine-modified
ꢀ-cyclodextrin (2), were synthesized and fully characterized. Their inclusion modes, binding abilities, and
molecular selectivities with four steroid guests, that is, cholic acid sodium salt (CA), deoxycholic acid sodium
salt (DCA), glycochoic acid sodium salt (GCA), and taurocholic acid sodium salt (TCA), were investigated
by the circular dichroism, 2D NMR, and isothermal titration microcalorimetry (ITC). The results obtained
from the circular dichroism and 2D NMR showed that two hosts adopted the different binding geometry, and
these differences subsequently resulted in the significant differences of molecular binding abilities and
selectivities. As compared with native ꢀ-cyclodextrin and tryptophan-modified ꢀ-cyclodextrin, host 2 showed
the enhanced binding abilities for CA and DCA but the decreased binding abilities for GCA and TCA; however,
host 1 showed the decreased binding abilities for all four bile salts. The best guest selectivity and the best
2-TCA
1-CA
host selectivity were KS2-DCA/KS
) 12.6 and KS2-CA/KS
) 10, respectively, both exhibiting great
enhancement as compared with the corresponding values of the previously reported L- and D-tryptophan-
modified ꢀ-cyclodextrins. Thermodynamically, it was the favorable enthalpic gain that led to the high guest
selectivity and host selectivity.
Introduction
CHART 1: Structures of Host Molecules
In recent years, molecular recognition has received much
attention in chemistry, biology, environmental science, and many
other fields.1-3 Cyclodextrins (CDs), a class of macrocyclic
oligosaccharides consisting of six, seven, or eight glucose units
linked by R-1,4-glucose bonds, have been widely used as
receptors in molecular recognition.4 It has been demonstrated
that the formation of inclusion complexes between CDs and
guest molecules is cooperatively governed by several weak
forces, such as van der Waals, hydrophobic, electrostatic,
dipole-dipole, and hydrogen-bonding interactions, and every
weak force does its contribution to the complexation. The
introduction of a suitable substituent may change the original
balance of the weak forces between CDs and guests and thus
influence the binding ability and selectivity. Therefore, a great
deal of endeavor has been devoted to the design and synthesis
of novel CD derivatives as well as their interactions with various
guest molecules.5 Among the numerous guests researched, bile
salts, a group of physiologically important steroids in vivo, play
a crucial role in lipid digestion, transportation, and absorption.6
They are composed of a steroid skeleton and a hydrophilic tail,
and the size/shape of bile salts is well fit for that of ꢀ-CD cavity.
Vazquez Tato and coworkers7 studied the complex geometry
of ꢀ-CD and its derivatives in D2O by ROESY experiments,
observing different binding modes upon inclusion complexation
with bile salts. Ueno et al.8 employed the dansyl moiety as a
fluorescence probe to investigate the complexation of CD-amino
acid-chromophore triad systems with bile salts and alcohols,
demonstrating that the molecular recognition behavior of these
sensors was delicately affected by the chirality of the amino
acid. Reinhoudt et al.9 reported a microcalorimetric study on
the cooperative binding of bile salts by CD dimers in a basic
environment, showing the enhanced binding abilities toward CA
and DCA. Recently, we reported the molecular recognition
behavior of L- and D-tryptophan-ꢀ-CDs10 (L- and D-Trp-ꢀ-CD,
3 and 4 in Chart 1) toward bile salts, and a satisfactory CA/
TCA selectivity was obtained as 4.6 by D-Trp-ꢀ-CD. More
recently, we also reported the molecular recognition of several
permethylated ꢀ-CD derivatives11 toward bile salts, showing the
enhanced binding abilities. In the present work, we synthesized
a pair of ꢀ-CD derivatives (1 and 2) with L- and D-tyrosine
substituents (Chart 1). Their molecular recognition and binding
behavior toward four bile salts (CA, DCA, TCA, and GCA)
were investigated by the circular dichroism, 2D NMR, and ITC
experiments and compared with those of L- and D-Trp-ꢀ-CDs.
Significantly, both a good guest selectivity as KS2-DCA/KS
2-TCA
) 12.6 and a good host selectivity as KS2-CA/KS1-CA ) 10 were
obtained. The factors that resulted in the great enhancement of
selectivity were discussed from the viewpoint of the binding
† Part of the “Robert A. Alberty Festschrift”.
* Corresponding author. E-mail: yuliu@nankai.edu.cn.
10.1021/jp105821s  2010 American Chemical Society
Published on Web 08/09/2010

16148 J. Phys. Chem. B, Vol. 114, No. 49, 2010
Chen et al.
heat measured in the titration experiments to give the net
reaction heat. The net reaction heat in each run was analyzed
by using “one set of binding sites” model for all ꢀ-CD
derivatives. The ORIGIN software (Microcal) allowed us to
determine the binding constant (KS) and reaction enthalpy (∆H°)
simultaneously with the standard derivation based on the scatter
of data points from a single titration experiment. The binding
stoichiometry was also given as a parameter when fitting the
binding isotherm (panel b in Figure 2). Knowledge of the
binding constant (KS) and molar reaction enthalpy (∆H°) enabled
the calculation of the standard Gibbs free energy of binding
(∆G°) and entropy change (∆S°) according to equation
∆G° ) -RT ln KS ) ∆H◦ - T∆S°
where R is the gas constant and T is the absolute temperature.
To check the accuracy of observed thermodynamic param-
eters, we carried out at least two independent titration experi-
ments to afford self-consistency of the thermodynamic param-
eters, and their average values are listed in Table 1. For a
comparison purpose, the thermodynamic quantities reported for
L- and D-Trp-ꢀ-CDs10 are also included in Table 1.
Figure 1. Calorimetric titration curve of 2 with DCA in pH 7.2
phosphate buffer solution (I ) 0.1 M) at 25 °C. (a) Raw data for
sequential 10 µL injections of solution 2 (3.90 mM) into solution DCA
(0.16 mM) (b) Heats of reactions, as obtained from the integration of
the calorimetric traces.
Results and Discussion
Circular Dichroism Spectra. It is widely documented that
the inclusion of a chromophoric group in a chiral cavity, such
as CD cavity, will produce either positive- or negative-induced
circular dichroism (ICD) signals according to the location and
orientation of the transition dipole moment of the chromophore
with respect to the CD axis.14 Therefore, the circular dichroism
spectrometry becomes a convenient and widely employed
method to elucidate the conformation of the CD-containing
systems.15 As can be seen from Figure 3, the circular dichroism
spectrum of host 1 displayed moderate positive Cotton effect
geometry and the binding thermodynamics. It is our special
interest to compare the binding abilities and molecular selectivi-
ties of ꢀ-CD derivatives with the aromatic amino acid substit-
uents in different size, which will help us deeply understand
the factors governing the molecular recognition of amino-acid-
modified CDs as well as their thermodynamic origin.
Microcalorimetric Titration. We performed ITC experi-
ments at atmospheric pressure in aqueous phosphate buffer
solution at 25 °C by using a Microcal VP-ITC titration
microcalorimeter, which allows us to determine the enthalpy
and equilibrium constant from a single titration curve simulta-
neously. We calibrated the instrument chemically by performing
the complexation reaction of ꢀ-CD with cyclohexanol; the
thermodynamic parameters obtained were in excellent agreement
with the literature data.12 All solutions were degassed and
thermostatted using a Thermo Vac accessory before each
titration. In each run, a phosphate buffer solution of the host
molecule in a 250 µL syringe was sequentially injected into
the calorimeter sample cell containing a buffer solution of bile
salt guests with a stirring speed at 300 rpm. The sample cell
volume was 1.4227 mL in all experiments. Each titration
experiment was composed of 25 successive injections (10 µL
per injection). The bile salt solutions that applied here were at
a concentration range of 0.16 to 0.53 mM, which was below
their critical micelle concentration (CMC). (The CMCs of the
steroids employed here are >1 mM13.) Each addition of CDs to
the sample cell gave rise to a heat of reaction, caused by the
formation of inclusion complexes between bile salt molecules
and CDs. The heats of reaction decrease after each injection of
host CD because fewer and fewer bile salt molecules are
available to form inclusion complexes. Typical titration curves
are shown in Figure 1.
1
peak at 225 nm (∆ε ) 2.79 M-1 cm-1) assigned to the La
transitions of L-tyrosine substituent. Moreover, the circular
dichroism signals of free L-tyrosine were relatively weak (∆ε
< 0.8 M-1 cm-1), whereas ꢀ-CD itself showed almost no
appreciable signals in the wavelength range of 210-300 nm.
In addition, the circular dichroism spectrum of the equimolar
mixture of L-tyrosine with ꢀ-CD resembled the corresponding
circular dichroism spectrum of free tyrosine. (See the Supporting
Information.) In the control experiment, the corresponding
circular dichroism signals of 1 resulting from displacement of
the tyrosine substituent from the CD cavity by measuring the
circular dichroism spectra in a water/alcohol mixture were also
examined, which showed that the circular dichroism signals of
1 in a water/alcohol mixture were weaker than those of 1 in
water. Therefore, we could deduce that the moderate circular
dichroism signals of 1 should be partially attributed to the self-
inclusion of the L-tyrosine substituent into the ꢀ-CD cavity with
1
its La transition band nearly parallel to the C7 axis of ꢀ-CD,
although the L-tyrosine substituent was optically active and
possessed the intrinsic circular dichroism signals.16 Upon
inclusion complexation with DCA, the circular dichroism signal
of 1 decreased to ∆ε ) 1.77 M-1 cm-1 at 225 nm. This
phenomenon indicated that the L-tyrosine moiety slightly moved
to the external of ꢀ-CD cavity after complexation with DCA.
The circular dichroism signal of host 2 was different from that
of 1, showing a moderate negative Cotton effect peak at 219
nm (∆ε ) -2.27 M-1 cm-1) for the 1La transition of D-tyrosine
substituent. In the control experiment, the circular dichroism
signals of free D-tyrosine were also very weak (|∆ε| < 0.8 M-1
We performed a control experiment to determine the heat of
dilution by injecting a host solution into a pure buffer solution
containing no guest molecules. The dilution heat determined in
the control experiment was subtracted from the apparent reaction

Selective Binding of ꢀ-Cyclodextrin Derivatives
J. Phys. Chem. B, Vol. 114, No. 49, 2010 16149
Figure 2. (a) Heat effects of dilution (I) and complexation (II) of 2 with DCA for each injection during microcalorimetric titration experiment. (b)
“Net” heat effect obtained by subtracting the heat of dilution from the heat of reaction, which was analyzed by computer simulation using the “one
set of binding sites” model.
TABLE 1: Complex Stability Constants (KS), Enthalpic Changes (∆H°), Entropic Changes (T∆S°), and Gibbs Free Energy
Changes (∆G°) for 1:1 Inclusion Complexation of Bile Salts with Host CDs in Phosphate Buffer Solution at 298.15 K
guests
CA
hosts
ꢀ-CDa
L-Tyr-ꢀ-CD (1)
D-Tyr-ꢀ-CD (2)
L-Trp-ꢀ-CD (3)b
D-Trp-ꢀ-CD (4)b
ꢀ-CDa
L-Tyr-ꢀ-CD (1)
D-Tyr-ꢀ-CD (2)
L-Trp-ꢀ-CD (3)b
D-Trp-ꢀ-CD (4)b
ꢀ-CDa
L-Tyr-ꢀ-CD (1)
D-Tyr-ꢀ-CD (2)
L-Trp-ꢀ-CD (3)b
D-Trp-ꢀ-CD (4)b
ꢀ-CDa
KS (M-1
)
∆G° (kJ mol-1
)
∆H° (kJ mol-1
)
T∆S° (kJ mol-1
)
4070 ( 80
871 ( 10
-20.6
-16.8
-22.5
-18.9
-23.4
-21.0
-17.3
-22.8
-19.2
-21.9
-19.3
-15.0
-17.4
-17.4
-18.5
-19.2
-14.8
-16.6
-17.3
-18.1
-23.0 ( 0.5
-26.7 ( 0.5
-41.7 ( 0.6
-23.2 ( 0.1
-37.9 ( 0.3
-25.8 ( 0.0
-33.1 ( 2.5
-50.5 ( 0.0
-32.1 ( 0.2
-46.0 ( 0.3
-23.0 ( 0.1
-28.3 ( 0.3
-30.5 ( 0.1
-23.4 ( 0.3
-24.9 ( 0.3
-23.8 ( 0.1
-25.7 ( 0.2
-26.7 ( 0.5
-23.1 ( 0.2
-24.3 ( 0.1
-2.4
-9.9
8689 ( 183
2020 ( 20
6680 ( 130
4840 ( 20
1087 ( 45
9962 ( 358
2310 ( 30
6770 ( 190
2350 ( 70
428 ( 1.4
1105 ( 3
-19.2
-4.3
-14.5
-4.8
DCA
GCA
TCA
-15.8
-27.9
-12.9
-24.1
-3.7
-13.3
-13.1
-6.0
-6.4
-4.6
1110 ( 20
1760 ( 40
2290 ( 10
391 ( 3
809 ( 9
1060 ( 20
1470 ( 10
L-Tyr-ꢀ-CD (1)
D-Tyr-ꢀ-CD (2)
L-Trp-ꢀ-CD (3)b
D-Trp-ꢀ-CD (4)b
-10.9
-10.1
-5.8
-6.2
a Ref 19. b Ref 10.
cm-1), whereas the circular dichroism spectrum of the equimolar
mixture of D-tyrosine with ꢀ-CD resembled the corresponding
circular dichroism spectra of free D-tyrosine. (See the Supporting
Information.) This phenomenon indicated that the D-tyrosine
substituent of 2 might be self-included in the ꢀ-CD cavity with
CHART 2: Structures of Guest Molecules Employed
1
the La transition band nearly perpendicular to the C7 axis of
ꢀ-CD. The circular dichroism signal of host 2 became more
negative after complexation DCA, indicating that there may be
1
a change of the angle between the La transition band and the
C7 axis of ꢀ-CD. It should be noted that the circular dichroism
spectra could not provide us with the enough information about
the relative position of substituent or guest molecule in the CD
cavity, so the conformations of hosts 1-2 and their complexes
with bile salts were further investigated by the 2D NMR
experiments.
2D NMR. 2D NMR spectroscopy has become a powerful
method to study the conformation of CD derivatives and their
complexes because one can conclude that two protons are
closely located in space (0.4 nm apart at most) if an NOE
correlation is detected between the relevant proton signals in
the NOESY or ROESY spectrum.17,18 Therefore, it is possible
to estimate the orientation of substituent or guest molecule in

16150 J. Phys. Chem. B, Vol. 114, No. 49, 2010
Chen et al.
Figure 3. Circular dichroism spectra of 1 and 2 (1.0 × 10-4 M) in the
presence and absence of DCA (1.0 × 10-3 M) in pH 7.20 phosphate
buffer solution (I ) 0.1 M).
Figure 4. ROESY spectrum of host 1 with a mixing time of 250 ms
at 298.1 K.
CD cavity with the aid of the assigned NOE correlations. On
this basis, we performed ROESY experiments to obtain further
information about the conformation of host CDs and their
complexes. The ROESY spectrum of host 1 was shown in
Figure 4. It is well known that only the H3, H5, and H6 protons
of CDs could give the NOE cross-peaks for analyzing the
conformation of host-guest complexes because H2 and H4 were
facing the outer cavity. (Hn was used for host protons.) As seen
from Figure 4, the cross-peak A was assigned to the NOE
correlations between the H3/H5 protons of ꢀ-CD and the Ha/
Hb protons of L-tyrosine substituent. A further comparison
showed that Ha exhibited the stronger NOE correlations with
the H5 protons than with the H3 protons of ꢀ-CD. Moreover,
the cross-peaks B and C were assigned to the NOE correlations
of the H6 protons of ꢀ-CD with the Ha/Hc protons of L-tyrosine
substituent, respectively. Because the H5/H6 protons were
located at the narrow opening, whereas the H3 protons were at
the wide opening, of the ꢀ-CD cavity, these results unambigu-
ously showed that the L-tyrosine moiety was self-included in
Figure 5. ROESY spectra of 1/DCA (2.8 and 5.3 mM, respectively)
with a mixing time of 250 ms at 298.1 K.
the ꢀ-CD cavity from the narrow opening. Moreover, the
MALDI-MS experiments ruled out the possibility of the
formation of oligomeric species. Furthermore, the ROESY
experiment of the 1/DCA system was also performed to
investigate the binding geometry between L-Tyr-ꢀ-CD and bile
salts. In Figure 5, the NOE correlations between the H3/H5
protons of ꢀ-CD and the Ha/Hb protons of L-tyrosine substituent
(peak A) were also observed after the addition of DCA. This
phenomenon indicated that the L-tyrosine moiety was not driven

Selective Binding of ꢀ-Cyclodextrin Derivatives
J. Phys. Chem. B, Vol. 114, No. 49, 2010 16151
Figure 6. ROESY spectrum of host 2 with a mixing time of 250 ms
at 298.1 K.
Figure 7. ROESY spectrum of 2/DCA (2.6 and 5 mM, respectively)
with a mixing time of 250 ms at 298.1 K.
out of the ꢀ-CD cavity, even after the guest inclusion. In a
general way, the bile salt molecule (Gn was used for DCA
protons) was marked with A, B, C, and D rings according to
the previous report.7 As seen in Figure 5, the peak B cor-
responded to the NOE correlations between the DCA’s G18
protons and the ꢀ-CD’s H3 protons, the peak C corresponded
to the NOE correlations between the DCA’s G21 protons and
the ꢀ-CD’s H3/H5 protons, and the peaks D, E, and F were
assigned to the NOE correlations of the protons of the DCA’s
D ring and the hydrophobic tail with the ꢀ-CD’s H3 or H5
protons. From the above information, we deduced that the DCA
entered the ꢀ-CD cavity from the wide opening with the tail
and the D ring and coexisted with the L-tyrosine substituent in
the ꢀ-CD cavity to form a cooperative inclusion manner.
In the case of D-Tyr-ꢀ-CD (Figure 6), similarly, the cross-
peak A was assigned to the NOE correlations between the H3/
H5 protons of ꢀ-CD and the Ha/Hb protons of D-tyrosine
substituent, and the correlation intensity of Ha with H3/H5 was
basically the same as that of Hb. This implied that the distances
between Ha/Hb and H3/H5 might be equal. At the same time,
the cross-peak B corresponding to the NOE correlations of the
D-tyrosine’s Hc protons with the ꢀ-CD’s H3/H5 protons and
the cross-peak C corresponding to the NOE correlations of the
D-tyrosine’s Hd protons with the ꢀ-CD’s H3 protons were also
observed. Combining this information with the circular dichro-
ism spectra results, we deduced that the D-tyrosine substituent
was deeply self-included in the ꢀ-CD cavity and might be
located in the center of the ꢀ-CD cavity. The ROESY spectra
of 2/DCA are shown in Figure 7. The NOE correlations between
the ꢀ-CD’s H3/H5 protons and the D-tyrosine’s Ha/Hb protons
still existed, but the NOE correlations of the D-tyrosine’s Hc
protons with the ꢀ-CD’s H3/H5 protons and the NOE correla-
tions of the D-tyrosine’s Hd protons with the ꢀ-CD’s H3 protons
(peak B and C in Figure 6) disappeared, indicating that the
D-tyrosine substituent of 2 partially moved out of the ꢀ-CD
cavity. Similarly, the cross-peak B corresponded to the NOE
correlations of the carboxylate tail and D ring protons of DCA
with the ꢀ-CD’s H3 or H5 protons. A close comparison of the
NOE signals of 1/DCA and 2/DCA systems showed that for
2/DCA the DCA’s G15/G14 protons exhibited the NOE
correlations with the ꢀ-CD’s H3/H5 protons (peak D and E),
but for 1/DCA, the DCA’s G15/G14 protons only exhibited the
NOE correlations with the ꢀ-CD’s H3 protons, and the NOE
correlation intensity of 2/DCA was stronger than that of 1/DCA.
Meanwhile, the peak C in Figure 7 should be assigned to the
NOE correlation of the Hd proton of D-tyrosine side arm in 2
with the DCA’s G18 protons. From this comparison, it is
reasonable to come to the conclusion that DCA penetrated into
the ꢀ-CD cavity of 2 more deeply than into that of 1. These
subtle differences in conformation between 1/DCA and 2/DCA
systems may subsequently result in the significant differences
of binding ability and molecular selectivity between hosts and
guests, which would be discussed below.
Binding Stoichiometry. The ITC experiments of hosts 1 and
2 with bile salts (CA, DCA, GCA, and TCA) showed the typical
titration curves of the 1:1 complex formation. The stoichiometric
ratio “N” observed from the curve-fitting results was within the
range 0.9 to 1.1, which clearly indicated that the majority of
the inclusion complexes had a 1:1 binding mode. Moreover,
the molecular modeling study demonstrated that hosts 1 and 2
could only accommodate one bile molecule in their hydrophobic
cavity. Therefore, the possible conformations of hosts 1 and 2
and their complexes with DCA are illustrated in Figure 8, along
with the reported conformations of hosts 3 and 4 and their DCA
complexes. In addition, their computational structures are in the
Supporting Information.
Binding Behavior. ITC is becoming a widely accepted
method to determine the thermodynamic parameters associated
with the noncovalent interactions of two (or more) molecules.20
According to the data presented in Table 1, thermodynamically,
the binding of all CDs with the bile salt guests was entirely
driven by the favorable enthalpic changes accompanied by the
unfavorable entropic changes (∆H° < 0; T∆S° < 0). The different
host-guest binding ability directly led to the huge differences

16152 J. Phys. Chem. B, Vol. 114, No. 49, 2010
Chen et al.
Figure 10. Enthalpic (∆H°) and entropic changes (T∆S°) for the
inclusion complexation of bile salts with hosts 1-4 in phosphate
aqueous buffer solutions at 298 K.
Figure 8. Conformation of hosts 1-4 and their possible binding modes
with DCA.
system and thus led to the favorable entropic changes. With
the tyrosine side arm located at the narrow opening of the CD
cavity, the shape of hosts 1 and 2 was like a barrel with a
bottom, and the close location of “bottom” to the bile salt guest
made the accommodated bile salt more immovable in the CD
cavity. This large loss of conformational freedom upon host-guest
binding consequently resulted in the big entropic loss, which
overwhelmed the entropic gain arising from the solvent reor-
ganization. This was consistent with the widely reported
enthalpy/entropy compensation effect that the large enthalpic
changes always accompanied by the large entropic changes.23
Complexation of Bile Salts by D-Tyr-ꢀ-CD (2). As can be
seen from Table 1, host 2 displayed the distinct binding abilities
toward four bile guests. Although ꢀ-CD, D-Trp-ꢀ-CD, and 2
all formed the 1:1 inclusion complexes with bile guests, 2 gave
the higher bind ability toward CA and DCA than ꢀ-CD and
D-Trp-ꢀ-CD (4). This might be attributed to the introduction of
D-tyrosine substituent and the conformational difference between
2 and 4. According to the binding geometry deduced from the
circular dichroism spectra and the 2D NMR experiments, after
complexation with guest molecules, the D-tyrosine moiety of
host 2 was not expelled out of the ꢀ-CD cavity. This conforma-
tion decreased the effective volume of the ꢀ-CD cavity to some
extent, and thus led to the higher size-fit efficiency between
host 2 and guest molecules. For D-Trp-ꢀ-CD (4), the D-
tryptophan substituent was expelled out of CD cavity10 after
complexation with the bile salt, which would lead to the
relatively weak size-fit efficiency as compared with that of the
2/bile salt system. Thermodynamically, the large entropic losses
of the 2/bile salt system (T∆S° ) -19.2 kJ mol-1 for CA and
-27.9 kJ mol-1 for DCA) could be compensated by the larger
negative enthalpic gain (∆H° ) -41.7 kJ mol-1 for CA and
-50.5 kJ mol-1 for DCA), which directly resulted in the high
binding constant (KS2-CA ) 8689 M-1 and KS2-DCA ) 9962 M-1).
In addition, the bind constant of 2 for DCA was slightly bigger
than that for CA. Possessing a more hydrophobic structure due
to the absence of the C-7 hydroxyl group as compared with
CA, DCA was easier to bind to the ꢀ-CD cavity than CA, which
consequently led to the more favorable hydrophobic interactions
between hosts and guests.
Figure 9. Complex stability constants (KS) of bile salts with hosts
1-4 in phosphate aqueous buffer solution at 298 K.
of the host-guest selectivity. To visualize and easily discuss
the binding behavior and molecular selectivity between hosts
1-4 and bile salt molecules, we plotted the changing profiles
of the complex stability constants (KS) as well as the enthalpic
changes (∆H°) and the entropic changes (T∆S°) in Figures 9
and 10, respectively. Thermodynamically, the large negative
enthalpic changes should mainly be attributed to the strong van
der Waals interactions between host and guest. It was reported
that aminated ꢀ-CDs would be positively charged at pH 7.2.19
Therefore, the electrostatic interactions between the positively
charged side arm and the negatively charged bile salt would
lead to the negative enthalpic changes. Moreover, the release
of the high-energy water in the ꢀ-CD cavity,21 the C-H· · ·O
hydrogen bond between the tyrosine group and the carboxyl
group of bile salt,22 and the O-H· · ·O hydrogen bond between
the -OH groups of CD cavity and the carboxyl group of bile
salt also contributed to the favorable enthalpic gain. Before the
host-guest binding, both the host CD and the guest molecule
were highly solvated, and the solvent molecules around the host
and the guest were highly ordered. During the host-guest
binding, the solvation shells of both the host and the guest
underwent reorganization. This process created disorder in the
Host 2 exhibited the obviously smaller binding abilities for
GCA and TCA guests than ꢀ-CD and D-Trp-ꢀ-CD (4). This
unusual phenomenon prompted us to explore the reasons that
caused the decrease in binding constants. Thermodynamically,
the values in Table 1 indicated that the decreased binding
affinities of host 2 toward GCA and TCA arose from the entropy

Selective Binding of ꢀ-Cyclodextrin Derivatives
J. Phys. Chem. B, Vol. 114, No. 49, 2010 16153
rather than the enthalpy (-∆∆H° < -∆T∆S°). The large
entropic loss might be attributed to at least two parts. First, GCA
(or TCA) possessed a cholic acid skeleton and a glycine (or
taurine) tail, and the side chain at C23 for GCA (or TCA)
became more polar and more hydrophilic than that for CA (or
DCA), which weakened the hydrophobic interactions between
hosts and guests and thus affected their binding thermodynamics
and binding constants. In addition, GCA (or TCA) possessed a
longer side chain than CA (or DCA), and it might not exist in
a serrated form but twist into a ring. This phenomenon would
increase the volume of the side chain to some extent and thus
lead to the relatively poor size-fit between host and guest
because 2 had a smaller effective volume of ꢀ-CD cavity, as
described above. Possessing a free ꢀ-CD cavity, ꢀ-CD and
D-Trp-ꢀ-CD (whose substituent was expelled out of the ꢀ-CD
cavity) could better bind the bigger GCA (or TCA) guest and
thus gave higher binding constants than 2.
the electrostatic repulsion with the anionic carboxylate or
sulfonate tail of the bile salt guest. These factors jointly led to
the weaker interactions between host and bile salts and thus
the lower binding ability. According to our previously report,10
DCA complexed L-Trp-ꢀ-CD (3) with the bigger A ring and B
ring penetrating into the ꢀ-CD cavity from the narrow opening
of the ꢀ-CD cavity, which consequently led to the better size-
fit between ꢀ-CD cavity and the bile salt guest. Moreover, the
L-tryptophan substituent of 3 also showed the NOE correlation
with the B ring of the accommodated bile salt guest. This NOE
correlation indicated that a possible C-H· · ·π interaction
resulted from the close location of these two groups. As a result,
3 gave stronger binding abilities toward the bile salt guest than
1.
Diastereomeric Host Recognition. The diastereomeric host
selectivity for the modified CDs with the enantiomeric side arms
has rarely been discussed. We have reported the molecular
recognition behaviors of L- and D-Trp-ꢀ-CDs toward bile salts,
but the results showed only the moderate host selectivity. As
Molecular Selectivity of 2 toward Bile Salts. The sharp
decrease in stability constants of 2 toward GCA and TCA
seemed disappointing, but the high complex stability constants
did not always refer to the high molecular selectivity. In general,
both the increase and the decrease in the stability constants were
important to the improvement of the molecular selectivity. As
seen in Table 1, the selectivity of 2 toward four bile guests was
DCA > CA > GCA > TCA, and KS2-DCA/KS2-TCA was up to 12.6,
the best result, CA gave a binding constant by D-Trp-ꢀ-CD (4)
4-CA
that was 3.3 times higher than that by L-Trp-ꢀ-CD (3) (KS
/
KS3-CA ) 3.3). After the tryptophan substituent was changed to
the smaller tyrosine substituent, the host selectivity got the
satisfactory enhancement. Also, for CA, the binding constants
for the hosts in Table 1 were in an order of 2 > 4 > ꢀ-CD > 3
a sharp comparison with KS4-DCA/KS4-TCA ) 4.6 for D-Trp-ꢀ-CD
> 1, and the host selectivity of the 2/1 couple was up to 10
ꢀ-CD-TCA
and KSꢀ-CD-DCA/KS
) 2.1 for native ꢀ-CD. Thermody-
(KS2-CA/KS
) 10). Thermodynamically, the higher ∆∆H°
1-CA
namically, the difference of the Gibbs free energy changes for
value (∆∆H°2-1 ) ∆H°2 - ∆H°1 ) -15 kJ mol-1, ∆T∆S°2-1
) T∆S°2 - T∆S°1 ) -9.3 kJ mol-1 for CA) indicated that the
favorable enthalpic gains, probably arising from the van der
Waals forces, hydrogen bond, and the release of the high-energy
water in the ꢀ-CD cavity, apparently contributed to the high
selectivity.
the complexation of 2 with DCA and TCA (∆∆G° ) ∆G°DCA
- ∆G°TCA) was up to -6.2 kJ mol-1, and ∆∆H° ) ∆H°DCA
∆H°TCA ) -24.8 kJ mol-1, ∆T∆S° ) T∆S°DCA - T∆S°TCA
-
)
-17 kJ mol-1. In contrast, for the complexation of 4 with DCA
and TCA, the enthalpic gain (∆∆H° ) -21.7 kJ mol-1) was
less, but the entropic loss (∆T∆S° ) -17.9 kJ mol-1) was larger,
than that of 2. That means, the higher enthalpic gain and the
less entropic loss together contributed to the great improvement
of the guest selectivity.
Conclusions
In summary, we have successfully synthesized a pair of ꢀ-CD
derivatives with L- and D-tyrosine as chiral side arms, and their
binding behaviors and molecular selectivities toward four typical
bile salts (CA, DCA, GCA, and TCA) were carefully studied.
The results obtained from the circular dichroism and 2D NMR
showed that the D-tyrosine moiety of 2 was more deeply self-
included into the ꢀ-CD cavity than the L-tyrosine moiety of 1.
Upon inclusion complexation, bile salt guests inserted into the
ꢀ-CD cavity of host 2 more deeply than into that of host 1.
The different binding geometry subsequently affected their
binding abilities and selectivities. Thermodynamically, the
binding behaviors of two hosts toward four bile salts were
entirely driven by favorable enthalpic changes, accompanied
by unfavorable entropic changes (∆H° < 0; T∆S° < 0). Host 2
displayed the enhanced binding ability for CA and DCA and
decreased binding abilities for GCA and TCA; however, host
1 showed clearly decreased binding ability toward all four bile
salts. The increase and the decrease in the stability constants
cooperatively contributed to the high selectivities between hosts
Complexation of Bile Salts by L-Tyr-ꢀ-CD (1). We could
also observe from Table 1 that compared with ꢀ-CD and L-Trp-
ꢀ-CD (3), 1 showed clearly decreased binding abilities toward
all four of the bile salts. In particular, for GCA and TCA, the
binding constants were only 427 and 391 M-1, respectively,
and even its strongest binding was only up to 1087 M-1 toward
DCA. These lower binding abilities of 1 might imply a change
of the binding thermodynamics. Thermodynamically, the inclu-
sion complexation of 1 with four bile salts exhibited the
favorable negative ∆H° and unfavorable T∆S°. The favorable
enthalpic gain of 1 was slightly higher than those of ꢀ-CD and
L-Trp-ꢀ-CD (3), but the entropic loss of 1 was much more than
those of ꢀ-CD and L-Trp-ꢀ-CD (3) toward corresponding guests.
Taking DCA as an example, the ∆∆H° and ∆T∆S° (∆H°1 -
H°3 ) -1.0 kJ mol-1, T∆S°1 - T∆S°3 ) -2.9 kJ mol-1) data
showed that the entropic changes mainly contributed to the
decrease in the binding abilities. The similar results were also
observed in the case of the three other bile salt guests. From
the viewpoint of binding geometry, 1 showed a conclusion mode
upon complexation with the bile salt guest. The bile salt guest
entered ꢀ-CD cavity of 1 with its D-ring and tail from the wide
opening of the ꢀ-CD cavity. Therefore, the relatively small
volume of the D-ring and the tail gave the relatively poor size-
fit with the ꢀ-CD cavity. In addition, the anionic carboxylate
of L-tyrosine substituent that was self-included in the ꢀ-CD
cavity of 1 would not only decrease the microenvironment
hydrophobicity of ꢀ-CD cavity to some extent but also give
and guests. As a result, host 2 presented the best recognition
ability for DCA/TCA pair (KS2-DCA/KS
gave the best host selectivity (KS2-CA/KS
) 12.6), and CA
) 10). Compared
2-TCA
1-CA
with their analogues L- and D-Trp-ꢀ-CD, L- and D-Tyr-ꢀ-CD
with the smaller substituents adopted the different inclusion
modes, which resulted in the better host-guest size-fit relation-
ship and the higher molecular selectivities between hosts and
guests.

16154 J. Phys. Chem. B, Vol. 114, No. 49, 2010
Chen et al.
Experimental Section
References and Notes
(1) (a) Nguyen, B.; Neidle, S.; Wilson, W. D. Acc. Chem. Res. 2009,
42, 11–21. (b) Kim, S. K.; Lee, D. H.; Hong, J. I.; Yoon, J. Acc. Chem.
Res. 2009, 42, 23–31. (c) Tomizawa, M.; Casida, J. E. Acc. Chem. Res.
2009, 42, 260–269.
(2) (a) Nishino, T.; Umezawa, Y. Anal. Chem. 2008, 80, 6968–6973.
(b) Ludden, M. J. W.; Li, X.; Greve, J.; van Amerongen, A.; Escalante,
M.; Subramaniam, V.; Reinhoudt, D. N.; Huskens, J. J. Am. Chem. Soc.
2008, 130, 6964–6973. (c) Choi, H. S.; Takahashi, A.; Ooya, T.; Yui, N.
ChemPhysChem. 2006, 7, 1668–1670.
Materials. ꢀ-CD of reagent grade was recrystallized twice
from water and dried in vacuo at 100 °C for 24 h prior to use.
L- and D-tyrosine were purchased from Peptide Institute. All
bile salt guests (CA, DCA, GCA, and TCA) were purchased
from Sigma and used as received. Mono[6-O-(p-toluenesulfo-
nyl)]-ꢀ-CD (6-OTs-ꢀ-CD) was prepared by the reaction of ꢀ-CD
with p-toluenesulfonyl chloride in aqueous alkaline solution.24
Phosphate buffer solution of pH 7.20 (I ) 0.1 M) was used for
circular dichroism spectral measurements and ITC experiments.
Instruments. Circular dichroism spectral measurements were
performed in a conventional quartz cell (light path 1 cm) on a
JASCO J-720W spectropolarimeter equipped with a temperature
controller, and the temperature of the cell was kept constant at
(3) (a) Lehn, J. M. Angew. Chem., Int. Ed. Engl. 1990, 29, 1304–1319.
(b) Liu, Y.; Chen, Y. Acc. Chem. Res. 2006, 39, 681–691. (c) Riela, S.;
D’Anna, F.; Lo Meo, P.; Gruttadauria, M.; Giacalone, R.; Noto, R.
Tetrahedron 2006, 62, 4323–4330. (d) Liu, Y.; Chen, G. S.; Chen, Y.; Ding,
F.; Chen, J. Org. Biomol. Chem. 2005, 3, 2519–2523. (e) Liu, Y.; Yang,
E. C.; Yang, Y. W.; Zhang, H. Y.; Fan, Z.; Ding, F.; Cao, R. J. Org. Chem.
2004, 69, 173–180. (f) Liu, Y.; Zhao, Y. L.; Zhang, H. Y.; Fan, Z.; Wen,
G. D.; Ding, F. J. Phys. Chem. B 2004, 108, 8836–8843. (g) Rekharsky,
M. V.; Inoue, Y. J. Am. Chem. Soc. 2002, 124, 813–826. (h) Schmidtchen,
F. P. Chem.sEur. J. 2002, 8, 3522–3529. (i) Bom, A.; Bradley, M.;
Cameron, K.; Clark, J. K.; van Egmond, J.; Feilden, H.; MacLean, E. J.;
Muir, A. W.; Palin, R.; Rees, D. C.; Zhang, M. Q. Angew. Chem., Int. Ed.
2002, 41, 265–270. (j) Kano, K.; Hasegawa, H. J. Am. Chem. Soc. 2001,
123, 10616–10627.
(4) (a) Szejtli, J. Chem. ReV. 1998, 98, 1743–1753. (b) Connors, K. A.
Chem. ReV. 1997, 97, 1325–1357. (c) Wallimann, P.; Marti, T.; Furer, A.;
Diederich, F. Chem. ReV. 1997, 97, 1567–1608. (d) Wenz, G.; Strassnig,
C.; Thiele, C.; Engelke, A.; Morgenstern, B.; Hegetschweiler, K.
Chem.sEur. J. 2008, 14, 7202–7211. (e) Han, C.; Li, H. small 2008, 4,
1344–1350. (f) Clark, J. L.; Peinado, J.; Stezowski, J. J.; Vold, R. L.; Huang,
Y.; Hoatson, G. L. J. Phys. Chem. B 2006, 110, 26375–26387. (g) Liu, Y.;
Song, Y.; Chen, Y.; Li, X. Q.; Ding, F.; Zhong, R. Q. Chem.sEur. J. 2004,
10, 3685–3696. (h) Ravoo, B. J.; Jacquier, J. C.; Wenz, G. Angew. Chem.,
Int. Ed. 2003, 42, 2066–2070.
1
25.0 °C. H NMR experiments were recorded on a Bruker
AV400 spectrometer, and 2D ROESY (rotating frame Over-
hauser effect spectroscopy) spectra were recorded on a Varian
Mercury VX300 instrument. All NMR experiments were carried
out in D2O.
Synthesis of L-Tyr-ꢀ-CD (1). L-Tyrosine (0.7 g) and 6-OTs-
ꢀ-CD (1.3 g) were dissolved in water (30 mL) containing
triethanolamine (20 mL), and the resulting mixture was heated
to reflux for 24 h with stirring under an argon atmosphere. After
the evaporation of most of the solvent under a reduced pressure,
the resulting solution was poured in anhydrous acetone (200
mL) with vigorous stirring to produce a pale-yellow precipitate.
The crude solid was collected by filtration and then purified by
Sephadex C-25 column chromatography with aqueous ammonia
(1.0 M) as an eluent, followed by chromatography on a
Sephadex G-25 column with deionizated water as eluent to give
a pale-yellow product (0.57 g, 44%). ESI-MS: m/z (relative
intensity), 1298.1 ([M + H]+, 100%), 1320.1 ([M + Na]+, 7%).
1H NMR (D2O, δ): 2.54-2.64 (m, 1H), 2.72-2.84 (m, 1H),
2.99-3.12 (d, 2H), 3.21-3.98 (m, 41H), 4.84-5.01(m, 7H),
6.71-6.85 (d, 2H), 6.98-7.09 (d, 2H). 13C NMR (100 MHz,
D2O, δ): 154.9, 129.6, 115.2, 101.8, 100.1, 82.9, 80.9, 73.1,
71.9, 60.0, 46.0, 35.5. Anal. Calcd for L-Tyr-ꢀ-CD:
C51H79NO37 ·6H2O, C, 43.56%; H, 6.52%; N, 1.00%. Found:
C, 43.43%; H, 6.59%; N, 0.85%. FT-IR (KBr) ν(cm-1): 3404,
2928, 1635, 1516, 1246, 849. UV λ (ε): 224 nm (6.23 × 103
dm3 mol-1 cm-1), 277 nm (1.36 × 103 dm3 mol-1 cm-1).
Synthesis of D-Tyr-ꢀ-CD (2). D-Tyr-ꢀ-CD was synthesized
in 41% yield using a similar method to the synthesis of 1. ESI-
MS: m/z (relative intensity), 1298.3 ([M + H]+, 100%). 1H NMR
(D2O, δ): 2.78-2.84 (m, 2H), 2.90-3.08 (m, 1H), 3.18-3.90
(m, 42H), 4.86-5.02 (m, 7H), 6.70-6.85 (d, 2H), 6.96-7.10
(d, 2H). 13C NMR (150 MHz, D2O, δ): 154.9, 130.6, 129.1,
115.4, 102.0, 101.2, 83.35, 81.1, 73.3, 72.2, 65.8, 60.4, 48.7,
37.8. Anal.. Calcd for D-Tyr-ꢀ-CD: C51H79NO37 ·8H2O, C,
42.47%; H, 6.64%; N, 0.97%. Found C, 42.62%; H, 6.54%; N,
0.95%. FT-IR (KBr) ν(cm-1): 3388, 2927, 1635, 1547, 1517,
1241, 836. UV λ (ε): 224 nm (8.04 × 103 dm3 mol-1 cm-1),
275 nm (1.65 × 103 dm3 mol-1 cm-1).
(5) Khan, A. R.; Forgo, P.; Stine, K. J.; D’Souza, V. T. Chem. ReV.
1998, 98, 1977–1996.
(6) (a) Nakai, K.; Tazuma, S.; Nishioka, T.; Chayama, K. Biochim.
Biophys. Acta 2003, 1632, 48–54. (b) Venneman, N. G.; van Kammen,
M.; Renooij, W.; VanBerge-Henegouwen, G. P.; van Erpecum, K. J.
Biochim. Biophys. Acta 2005, 1686, 209–219.
(7) (a) Cabrer, P. R.; Alvarez-Parrilla, E.; Meijide, F.; Seijas, J. A.;
Nunez, E. R.; Vazquez Tato, J. Langmuir 1985, 1, 5489–5495. (b) Singh,
A. P.; Cabrer, P. R.; Alvarez-Parrilla, E.; Meijide, F.; Vazquez Tato, J.
J. Inclusion Phenom. Macrocyclic Chem. 1999, 35, 335–348. (c) Cabrer,
P. R.; Alvarez-Parrilla, E.; Al-Soufi, W.; Meijide, F.; Rodriguez, Nunez,
E.; Vazquez Tato, J. Supramol. Chem. 2003, 15, 33–43.
(8) (a) Hamasaki, K.; Ikeda, H.; Nakamura, A.; Ueno, A.; Toda, F.;
Suzuki, I.; Osa, T. J. Am. Chem. Soc. 1993, 115, 5035–5040. (b) Wang,
Y.; Ikeda, T.; Ikeda, H.; Ueno, A.; Toda, F. Bull. Chem. Soc. Jpn. 1994,
67, 1598–1607. (c) Ikeda, H.; Nakamura, M.; Ise, N.; Oguma, N.; Nakamura,
A.; Ikeda, T.; Toda, F.; Ueno, A. J. Am. Chem. Soc. 1996, 118, 10980–
10988. (d) Matsushita, A.; Kuwabara, T.; Nakamura, A.; Ikeda, H.; Ueno,
A. J. Chem. Soc., Perkin Trans. 2 1997, 1705–1710.
(9) De Jong, M. R.; Engbersen, J. F. J.; Huskens, J.; Reinhoudt, D. N.
Chem.sEur. J. 2000, 6, 4034–4040.
(10) Wang, H.; Cao, R.; Ke, C. F.; Liu, Y.; Wada, T.; Inoue, Y. J. Org.
Chem. 2005, 70, 8703–8711.
(11) Liu, Y.; Shi, J.; Guo, D.-S. J. Org. Chem. 2007, 72, 8227–8234.
(12) Rekharsky, M. V.; Schwarz, F. P.; Tewari, Y. B.; Goldberg, R. N.;
Tanaka, M.; Yamashoji, Y. J. Phys. Chem. 1994, 98, 4098–4103.
(13) (a) Collo, A.; Meijide, F.; Rodriguez Nunez, E.; Vazquez Tato, J.
J. Pharm. Sci. 1996, 85, 9–15. (b) Cooper, A.; Nutley, M. A.; Camilleri, P.
Anal. Chem. 1998, 70, 5024–5028. (c) Liu, Y.; Zhang, N.; Chen, Y.; Chen,
G. S. Bioorg. Med. Chem. 2006, 14, 6615–6620.
(14) (a) Shimizu, H.; Kaito, A.; Hatano, M. Bull. Chem. Soc. Jpn. 1979,
52, 2678–2684. (b) Shimizu, H.; Kaito, A.; Hatano, M. Bull. Chem. Soc.
Jpn. 1981, 54, 513–519. (c) Kajta´r, M.; Horvath-Toro, C.; Kuthi, E.; Szejtli,
J. Acta Chim. Acad. Sci. Hung. 1982, 110, 327–355. (d) Kodaka, M. J. Am.
Chem. Soc. 1993, 115, 3702–3705.
(15) (a) Yang, S. Y.; Green, M. M.; Schultz, G.; Jha, S. K.; Muller,
A. H. E. J. Am. Chem. Soc. 1997, 119, 12404–12405. (b) McAlpine, S. R.;
Garcia-Garibay, M. A. J. Am. Chem. Soc. 1998, 120, 4269–4275. (c) Inoue,
Y.; Wada, T.; Sugahara, N.; Yamamoto, K.; Kimura, K.; Tong, L.-H.; Gao,
X.-M.; Hou, Z.-J.; Liu, Y. J. Org. Chem. 2000, 65, 8041–8050. (d) Balan,
B.; Gopidas, K. R. Chem.sEur. J. 2007, 13, 5173–5185.
(16) Sreerama, N.; Manning, M. C.; Powers, M. E.; Zhang, J. X.;
Goldenberg, D. P.; Woody, R. W. Biochemistry 1999, 38, 10814–10822.
(17) Bothner-By, A. A.; Stephens, R. L.; Lee, J.-M.; Warren, C. D.;
Jeanloz, R. W. J. Am. Chem. Soc. 1984, 106, 811–813.
Acknowledgment. We thank the 973 Program (2006CB-
932900), NNSFC (nos. 20932004, 20772062, and 20721062)
and Tianjin Natural Science Foundation (07QTPTJC29600) for
financial support.
Supporting Information Available: UV-vis spectra, cir-
cular dichroism spectra, and computational structures. This
material is available free of charge via the Internet at http://
pubs.acs.org.
(18) Neuhaus, D.; Williamson, M. The Nuclear OVerhauser Effect in
Structural and Conformational Analysis; VCH Publishers: New York, 1989;
pp 123-135.

Selective Binding of ꢀ-Cyclodextrin Derivatives
J. Phys. Chem. B, Vol. 114, No. 49, 2010 16155
(19) Liu, Y.; Yang, Y.-W.; Cao, R.; Song, S.-H.; Zhang, H.-Y.; Wang,
L.-H. J. Phys. Chem. B 2003, 107, 14130–14139.
(23) (a) Rekharsky, M. V.; Inoue, Y. Chem. ReV. 1998, 98, 1875–1917.
(b) Schneider, H. J. Angew. Chem., Int. Ed. 2009, 48, 3924–3977. (c) de
Bokx, P. K.; Boots, H. M. J. J. Phys. Chem. 1989, 93, 8240–8243.
(24) Petter, R. C.; Salek, J. S.; Sikorski, C. T.; Kumaravel, G.; Lin,
F.-T. J. Am. Chem. Soc. 1990, 112, 3860–3868.
(20) Ababou, A.; Ladbury, J. E. J. Mol. Recognit. 2006, 19, 79–89.
(21) Bender, M. L.; Komiyama, M. Cyclodextrin Chemistry; Springer-
Verlag: Berlin, 1978. (b) Komiyama, M.; Bender, M. L. J. Chem. Phys.
1978, 100, 2259.
(22) Hobza, P.; Havlas, Z. Chem. ReV. 2000, 100, 4253–4264.
JP105821S