A boronic acid-diol interaction is useful for chiroselective transcription of the sugar structure to the Δ- Versus Λ-[CoIII(bpy)3]3+ ratio

Article abstract of DOI:10.1039/a803382j

In order to apply boronic acid-saccharide interactions to the chiroselective synthesis of Δ- and Λ-[Co^III(bpy)₃]³⁺ saccharide-binding ligands, 2,2′-bipyridine-4-boronic acid (bpymb) and 2,2′-bipyridine-4,4′-diboronic acid (bpydb) were newly synthesized. It was shown that most D-saccharides form cyclic 1:1 complexes with bpydb to afford the CD-active species. The positive exciton coupling band implies that two pyridine rings are twisted in a clockwise direction ((R)-chirality). In contrast, such a CD-active species was not yielded from bpymb. The treatment of the bpydb-D-saccharide complexes with Co(OAc)₂ gave the substitution-active [Co^II(bpyba)₃]^4--saccharide complexes, which were oxidized to the substitution-inactive [Co^III(bpyba)₃]^3--saccharide complexes. In this stage, the Δ vs. Λ ratio was fixed. The complexes were converted to [Co^III(bpy)]³⁺ by treatment with AgNO₃ and the e.e. was determined by comparison with authentic Δ- or Λ-[Co^III(bpy)]³⁺. The Δ-isomer was obtained in excess from most D-saccharides but the Λ-isomer was also obtained from D-fructose and D-fucose. At 4°C, the largest e.e. for bpydb was attained with D-glucose (47% e.e.; Δ excess). Under the same reaction conditions the bpymb + D-glucose system gave 16% e.e. (Δ excess). The e.e. of the bpydb + D-glucose system increased with lowering the reaction temperature and at -25°C it reached 79% e.e. The foregoing results clearly establish that the saccharide-templated synthesis is useful as a new concept for the preparation of chiral tris(2,2′-bipyridine)-metal complexes. Furthermore, the Δ vs. Λ equilibrium can be shifted in either direction by the selection of saccharide enantiomers.

Full text of DOI:10.1039/a803382j

View Article Online / Journal Homepage / Table of Contents for this issue
A boronic acid–diol interaction is useful for chiroselective
transcription of the sugar structure to the Ä- versus
Ë-[CoIII(bpy)3]3؉ ratio
Toshihisa Mizuno,a Masayuki Takeuchi,a Itaru Hamachi,a Kazuaki Nakashimab and
Seiji Shinkai*ab
a Department of Chemical Science & Technology, Faculty of Engineering, Kyushu University,
Fukuoka 812, Japan
b Chemotransfiguration Project-JST, 2432 Aikawa, Kurume, Fukuoka 839, Japan
Received (in Cambridge) 5th May 1998, Accepted 20th July 1998
In order to apply boronic acid–saccharide interactions to the chiroselective synthesis of ∆- and Λ-[CoIII(bpy)3]3ϩ
saccharide-binding ligands, 2,2Ј-bipyridine-4-boronic acid (bpymb) and 2,2Ј-bipyridine-4,4Ј-diboronic acid (bpydb)
were newly synthesized. It was shown that most -saccharides form cyclic 1:1 complexes with bpydb to afford the
CD-active species. The positive exciton coupling band implies that two pyridine rings are twisted in a clockwise
direction ((R)-chirality). In contrast, such a CD-active species was not yielded from bpymb. The treatment of the
bpydb–-saccharide complexes with Co(OAc)2 gave the substitution-active [CoII(bpyba)3]4Ϫ–saccharide complexes,
which were oxidized to the substitution-inactive [CoIII(bpyba)3]3Ϫ–saccharide complexes. In this stage, the ∆ vs. Λ
ratio was fixed. The complexes were converted to [CoIII(bpy)]3ϩ by treatment with AgNO3 and the e.e. was determined
by comparison with authentic ∆- or Λ-[CoIII(bpy)]3ϩ. The ∆-isomer was obtained in excess from most -saccharides
but the Λ-isomer was also obtained from -fructose and -fucose. At 4 ЊC, the largest e.e. for bpydb was attained
with -glucose (47% e.e.; ∆ excess). Under the same reaction conditions the bpymb ϩ -glucose system gave 16% e.e.
(∆ excess). The e.e. of the bpydb ϩ -glucose system increased with lowering the reaction temperature and at Ϫ25 ЊC
it reached 79% e.e. The foregoing results clearly establish that the saccharide-templated synthesis is useful as a new
concept for the preparation of chiral tris(2,2Ј-bipyridine)–metal complexes. Furthermore, the ∆ vs. Λ equilibrium can
be shifted in either direction by the selection of saccharide enantiomers.
gives us a variety of saccharides with chiral structures. The
object of the present study is to exploit a new cascade-type
transmission system for (i) reading-out of the ‘structural
information’ stored in the saccharide configuration, (ii) chiro-
selective transcription into other vehicles and (iii) immobiliz-
ation as a new information source. With this object in mind
we chose 2,2Ј-bipyridine-4-monoboronic acid (bpymb) and
Introduction
Saccharides are abundant chiral resources that Nature presents
us. However, the chemistry utilizing saccharides as chiral
building-blocks or chiral auxiliaries has not yet been sufficiently
exploited. This is due to the difficulty or the lack of method-
ology to transcribe the chirality information stored in sac-
charides into desired molecules or molecular assemblies. Since
the chemistry of saccharides plays a very important role in
(HO)2B
B(OH)2
B(OH)2
the metabolic pathways of living organisms, the detection of
biologically important saccharides is necessary in a variety of
medical, diagnostic and industrial contexts. Then, how can we
‘touch’ saccharides which have been ‘untouchable’ with an arti-
ficial receptor, particularly, when they are dissolved in aqueous
solution? Although hydrogen-bonding interactions play a cen-
tral role in most reported artificial receptors,1 it has recently
been shown that boronic acid–diol covalent interactions which
readily and reversibly form in aqueous media represent an
important alternative binding force in the recognition of sac-
charides and related molecular species.2–7 Stable boronic acid-
based saccharide receptors offer the possibility of designing
saccharide sensors which are selective and sensitive for a
specific saccharide.6 Of particular interest are fluorescent sac-
charide sensors in which the photoinduced electron-transfer
(PET) mechanism is integrated within a molecule.8 Through
these studies we have noticed that boronic acid–saccharide
interactions create a variety of supramolecular structures which
are founded on the absolute configuration of each sacchar-
ide.9–11 This implies that saccharides are very useful as chiral
building-blocks to create oriented supramolecular assemblies
or as chiral auxiliaries for asymmetric syntheses. This viewpoint
has escaped attention for a long time, even though Nature
N
N
N
N
bpydb
bpymb
2,2Ј-bipyridine-4,4Ј-diboronic acid (bpydb) as an ‘interface’:
since tris(2,2Ј-bipyridine: bpy)–metal complexes have inherent
∆ vs. Λ chirality, two pyridine rings in bpydb may be asym-
metrically oriented by the formation of a cyclic 1:1 complex
with saccharides.12 This saccharide-induced chiral orientation
of this ligand should eventually lead to the enantiomeric excess
of either ∆- or Λ-isomer in the resultant metal complexes. If
this idea really works, it follows that one can achieve the process
(i)→(ii)→(iii) using bpydb as an ‘interface’ to transmit the
saccharide structure to the metal complex chirality.13,14
Results and discussion
Determination of the pKa values and saccharide-induced CD
spectra of bpymb and bpydb
To find out the pH range useful for saccharide recognition, pKa
values for bpymb and bpydb were determined. Fig. 1 shows a
J. Chem. Soc., Perkin Trans. 2, 1998, 2281–2288
2281

View Article Online
Fig. 2 Plot of A320 vs. pH for bpydb.
Fig. 1 Phototitration of bpydb (6.00 × 10Ϫ5 mol dmϪ3) at 25 ЊC: buffer
concentration 50 mmol dmϪ3 (for details see Experimental section).
pH-dependent absorption spectral change in bpydb which is a
more important ligand than bpymb for the present purpose.
For the pH range 1.00–6.00 a tight isosbestic point appears at
289.5 nm whereas for the pH range 6.50–12.00 another tight
isosbestic point appears at 286.0 nm. The results indicate that
three species which can be differentiated by phototitration exist
in bpydb. It is known that pyridine-4-boronic acid (pymb) has
pKa1 = 3.6 and pKa2 = 7.9 which are assigned to OHϪ-adduct
formation with pymbHϩ to yield a zwitterionic species and
deprotonation to yield an anionic species, respectively. Hence,
the pH-dependent acid dissociation for bpydb can be expressed
as in Scheme 1. A plot of A320 versus pH is shown in Fig. 2.
From the analysis of this plot the pKa1 and pKa2 for bpydb can
be estimated to be 3.6 and 7.9, respectively.
Fig. 3 Plot of A304 vs. pH for bpymb.
–
B(OH)2
B(OH)3
H+
H2O
N
N
N
N
+
H+
H2O
H
H
pK1 = 3.2
H+
–
–
(HO)2B
B(OH)2
H2O
(HO)3B
B(OH)3
pK2 = 7.0
H+
H+
–
B(OH)3
N
N
N
N
+
+
H+
pK1 = 3.9
H2O
H
H
N
N
H+
pK2 = 7.9
Scheme 2
to the OHϪ-adducts at neutral pH and, therefore, can form the
stable complexes with saccharides. This is a large advantage of
bpymb and bpydb as an interface for saccharides.
H+
–
–
(HO)3B
B(OH)3
Chiral orientation of ligands by the saccharide-binding
N
N
Previously, we designed diphenyl-3,3Ј-diboronic acid for the
detection of disaccharides.15 The distance between the two
boronic acids (ca. 7.4 Å) is comparable with or a little
shorter than that between 1,2-cis-diol and 4Ј-OHؒ5Ј-CH2OH in
disaccharides with an extended conformation (ca. 8.7 Å).
As expected, the two boronic acids were bridged by certain
disaccharides and the resultant disaccharide-containing macro-
cycles became CD-active because of a chiral twist of the two
phenyl rings. From the sign and the intensity of the CD spectra
it is possible to distinguish these disaccharides by CD spectro-
scopy. Strangely, some monosaccharides (e.g., -glucose),
the molecular size of which is apparently smaller than the
diboronic acid in diphenyl-3,3Ј-diboronic acid, could make
Scheme 1
Bpymb also showed the similar pH-dependent spectral
change with a tight isosbestic point at 295.5 nm for pH 1.53–
6.00 and another tight isosbestic point at 286.0 nm for pH 6.50–
11.00. A plot of A304 versus pH is shown in Fig. 3. From analysis
of this plot the pKa1 and pKa2 can be estimated to be 3.8 and
7.2, respectively (Scheme 2).
It is known that the boronic acid group forms stable sacchar-
ide complexes at pH regions where the group can exist as its
OHϪ-adduct. The foregoing pKa1 values indicate that the
boronic acid groups in bpymb and bpydb are already converted
2282
J. Chem. Soc., Perkin Trans. 2, 1998, 2281–2288

View Article Online
Fig. 6 Job plot: the sum of [-glucose] and [bpydb] is kept constant
(1.69 × 10Ϫ2 mol dmϪ3). Other conditions are recorded in the Fig. 4
caption.
active species is a cyclic 1:1 bpydb–-glucose complex. Based
on the foregoing findings, one can propose that -glucose is
bound to bpydb according to Scheme 3. This view was further
Fig. 4 CD spectra of bpydb (1.69 × 10Ϫ2 mol dmϪ3) in the presence of
-glucose (0–3.38 × 10Ϫ1 mol dmϪ3): water (pH 7.5 with 0.050 mol dmϪ3
MOPS)–methanol = 1:1 v/v, 4 ЊC, 0.2 mm cell width.
O
O
O
–
O
–
–
–
B
B
(HO)3B
B(OH)3
OH
HO
N
N
N
N
+
+
H
H
CD-active
1 : 1 complex
OH
OH
OH
OH
Fig. 5 Plot of [θ]295 vs. the -glucose concentration: the bpydb concen-
tration was kept constant (1.69 × 10Ϫ2 mol dmϪ3). Other conditions are
recorded in the Fig. 4 caption.
HO
HO
OH
OH
: Saccharide
O
O
O
O
–
the complexes CD-active.15 Examination by Corey–Pauling–
Koltun (CPK) molecular models and MM3 computational
studies suggest that when 1,2-cis-diol forms a cyclic ester with
one boronic acid group, 6-OH can reach another boronic acid
group:15 in particular, 6-OH in the furanose form can complex
with another boronic acid group.16 Hence, it can be postulated
that the origin of the CD activity induced by monosaccharides
also stems from the formation of saccharide-containing
macrocycles.
–
B
B
OH
HO
N
N
+
H
CD-silent
1 : 2 complex
Scheme 3
As shown in Fig. 4, bpydb yielded a CD-active species in the
presence of -glucose. Since the wavelength at θ = 0 (283 nm) is
comparable with the λmax in the absorption spectrum at the
same pH (see Fig. 1), one can observe that the CD spectra
feature the exciton coupling band consisting of the first positive
Cotton effect and the second negative Cotton effect. This CD
sign implies that two pyridine rings are twisted in a clockwise
direction ((R)-chirality) by -glucose. In Fig. 5, the CD inten-
sity ([θ]295) is plotted against the -glucose concentration. It is
seen from Fig. 5 that the [θ]295 first increases at [-glucose]/
[bpydb] < 2 and then decreases with increasing -glucose con-
centration. This kind of biphasic dependence has frequently
been observed for the saccharide-binding to diboronic acid
receptors and can be rationalized in terms of a stoichiometrical
change from a cyclic 1:1 to a noncyclic 1:2 diboronic acid–
saccharide complex. To obtain explicit evidence for the stoichio-
metry we applied a Job plot to this system (Fig. 6). Fig. 6 clearly
shows that a maximum [θ]295 value appears at [-glucose]/
([-glucose] ϩ [bpydb] ) = 0.5, supporting the idea that the CD-
supported by the fact that bpymb does not give the CD-active
species in the presence of any saccharides.
The CD spectra of bpydb complexes with several -mono-
saccharides and one disaccharide (-cellobiose) are shown in
Fig. 7. As shown in Fig. 7(a), most -monosaccharides gave a
positive exciton coupling band as -glucose did. The result
implies that in general, -monosaccharides forming cyclic 1:1
complexes with bpydb tend to twist the two pyridine rings in
a clockwise direction ((R)-chirality). -monosaccharides which
gave CD spectra different from the typical positive exciton
coupling band are summarized in Fig. 7(b). For the bpydb–
-galactose complex the first positive Cotton effect at around
310 nm could not be observed clearly and the second negative
Cotton effect at around 270 nm was weaker than those for other
bpydb–-monosaccharide complexes. The bpydb–-fucose
complex gave more complicated CD bands. Conceivably, the
complexity is due to the formation of a few different complex
J. Chem. Soc., Perkin Trans. 2, 1998, 2281–2288
2283

View Article Online
O
O
O
O
–
–
B
B
(HO)2B
B(OH)2
OH
HO
saccharide
N
N
N
+
N
H
bpydb
bpydb-saccharide
complex
Co(OAc)2
O
O
B
O
O
–
–
B
HO
OH
N
N
HO
OH
B
II
B
Co
N
N
–
–
O
O
O
O
N
N
–
O
O
–
O
O
B
B
HO
OH
[Co(bpydb)3]4- - saccharide
complex
Fig. 7 CD spectra of bpydb–monosaccharide complexes with a
(R)-chirality exciton coupling band (a) and those with unusual exciton
coupling bands and a bpydb–-cellobiose (disaccharide) complex (b):
[bpydb] = 1.69 × 10Ϫ2 mol dmϪ3, [saccharide] = 3.38 × 10Ϫ2 mol dmϪ3
.
oxidation
Other conditions are recorded in the Fig. 4 caption.
species. On the other hand, -cellobiose (disaccharide) also
gave a positive exciton coupling band with bpydb.
The foregoing results indicate that the bpydb ligand can be
converted to a chiral atropisomer by an intramolecularly-
bridged saccharide prior to the metal complex formation. Next,
we estimated the possibility of this saccharide-induced chirality
being transcripted to the ∆ vs. Λ chirality ratio of the bpydb
metal complexes.
O
O
O
–
O
–
B
B
HO
OH
N
N
HO
O
OH
III
B
Co
N
N
B
Evaluation of the transcription methods
–
–
O
O
O
Previously, we tested the above-mentioned working hypothesis
with the FeII complexes.12 The reasons why we chose FeII
were (i) FeII tends to selectively afford the tris-bpy-coordinated
[FeII(bpy)3]2ϩ species with bpy in aqueous solution, (ii) there
is a precedent for successful optical resolution of ∆- and
Λ-[FeII(bpy)3]2ϩ complexes and (iii) the FeII complexes usually
give a relatively strong absorbance in the MLCT region.17
Although strong CD bands were observed in the MLCT region
for [FeII(bpydb)3]4Ϫ in the presence of several saccharides, the
CD sign could be easily reversed by the addition of the
enantiomeric saccharides even at 4 ЊC.12 The result implies that
in these FeII complexes racemization between the ∆- and the
Λ-isomer can easily occur. To transmit the saccharide chirality
one should adopt some substitution-inactive metal complex.
RuII can be such a candidate metal ion, but the RuII complexes
can be prepared only at high temperature where the chirality
control is energetically disadvantageous. To overcome this
dilemma we contrived the reaction scheme depicted in Scheme
4. Firstly, bpydb (or bpymb: hereafter, the method for bpydb is
mainly described but that for bpymb is basically the same) and
saccharide were dissolved in a water (pH 7.4 with 1.7 × 10Ϫ2
mol dmϪ3 MOPS buffer)–methanol (1:1 v/v) solution (1.0 ml),
and then the solution was mixed with a water–methanol (1:1
v/v) solution (1.0 ml) containing Co(OAc)2 at the appropriate
temperature. The reaction mixture was stirred at this temper-
ature for 2 days under a nitrogen atmosphere. As [CoII(bpy)3]2ϩ
N
N
–
O
–
O
O
B
B
O
HO
OH
[Co(bpydb)3]3- - saccharide
complex
HO
HO
OH
: Saccharide
OH
Scheme 4
is known to isomerize between ∆- and Λ-isomer (i.e., substitu-
tion-active),17 the [CoII(bpydb)3]4Ϫ-saccharide complex should
reach some ∆ vs. Λ equilibrium under the influence of the
added saccharide. Then, the [CoII(bpydb)3]4Ϫ-saccharide com-
plex was oxidized to the [CoIII(bpydb)3]3Ϫ-saccharide complex
by bubbling O2 into this solution for 48 h. As [CoIII(bpydb)3]3Ϫ
is substitution-inactive, the ∆ vs. Λ ratio has been fixed at this
step. By this method one can prepare the substitution-inactive
CoIII complexes in the presence of saccharides.
It is known that a redox couple of [CoII(bpy)3]2ϩ/[CoIII-
(bpy)3]3ϩ has E1/2 0.34 V and the [CoII(bpy)3]2ϩ complex is slowly
2284
J. Chem. Soc., Perkin Trans. 2, 1998, 2281–2288

View Article Online
Fig. 8 Absorption spectral change for the dioxygen-oxidation of
[CoII(bpymb)3]Ϫ (6.00 × 10Ϫ5 mol dmϪ3) to [CoIII(bpymb)3]: 25 ЊC, pH
7.0 with 5.0 × 10Ϫ2 mol dmϪ3 phosphate buffer.
Fig. 9 Absorption spectra of authentic [CoIII(bpydb)3]3Ϫ (dotted line)
and [CoIII(bpydb)3]3Ϫ prepared by dioxygen-oxidation of [CoII-
(bpydb)3]4Ϫ (solid line). The measurement conditions are recorded in
the Fig. 8 caption.
oxidized by dioxygen.18 This oxidation is characterized by a red
shift of the absorption maximum from 301 nm to 317 nm.18 We
measured an absorption spectral change for the oxidation of
[CoII(bpymb)3]Ϫ and found a similar red shift from 307 nm (for
the CoII complex) to 317 nm (for the CoIII complex) under the
present measurement conditions. The CV measurements of
[CoII(bpymb)3]Ϫ resulted in a reversible redox peak at 0.011 V
(25 ЊC, pH 7.0 with 5.0 × 10Ϫ2 mol dmϪ3 phosphate buffer,
[CoII(bpymb)3]Ϫ = 7.50 × 10Ϫ4 mol dmϪ3, sweep speed 20 mV
sϪ1). Thus, the [CoII(bpymb)3]Ϫ complex should be also oxidized
by dioxygen. Fig. 8 shows an absorption spectral change of
[CoII(bpymb)3]Ϫ in an aerobic solution. It is seen from Fig. 8
that the λmax peak at 294 nm disappears with the appearance of
a new λmax peak at 318 nm and that the spectral change holds a
few tight isosbestic points. The final spectrum after 12 h was
ascribable to [CoIII(bpymb)3]. In the subsequent experiments,
therefore, we analyzed the CD spectra of the solutions
dioxygen-oxidized for 48 h. On the other hand, the dioxygen-
oxidation of [CoII(bpydb)3]4Ϫ was so fast that the spectral
change could not be followed by the conventional spectro-
scopic method. In the CV measurements, the sharp redox peak
did not appear in the range of Ϫ0.4–0.4 V. Judging from the
fact that the E1/2 value for [CoII(bpymb)3]Ϫ is more negative
than that for [CoII(bpy)3]2ϩ, [CoII(bpydb)3]4Ϫ should possess an
E1/2 value more negative than [CoII(bpymb)3]Ϫ (i.e., more
negative than 0.011 V). Because of this negative E1/2 shift,
[CoII(bpydb)3]4Ϫ has become sensitive to dioxygen-oxidation.
Thus, completion of this oxidation process was confirmed
by the following lines of indirect evidence: (i) the absorption
spectrum of the resultant complex is the same as that of the
authentic [CoIII(bpydb)3]3Ϫ complex prepared from [CoIII-
Cl(NH3)5]2ϩ and bpydb (Fig. 9), (ii) the absorption spectrum is
unaffected by further oxidation with H2O2, (iii) the complex is
ESR-silent (because CoII is paramagnetic whereas CoIII is
diamagnetic) and (iv) the 1H NMR spectrum does not show any
indication that the complex is paramagnetic (e.g., line-
broadening, unusual chemical shift, etc.). We mixed the
bpydb–saccharide complex and CoII(OAc)2 under the anaero-
bic conditions and introduced dioxygen after 10 min so that
CoII could form the [CoII(bpydb)3]3Ϫ–saccharide complex.
As [CoIII(bpymb)3]Ϫ and [CoIII(bpydb)3]3Ϫ are substitution-
inactive,17 the ∆ vs. Λ ratio has been fixed at these oxidation
steps.
Fig. 10 CD spectra of [CoIII(bpydb)3]3Ϫ (5.56 × 10Ϫ4 mol dmϪ3
)
and the influence of added cis-CPD at 25 ЊC: (a) Sample A, [-
glucose] = 3.38 mmol dmϪ3, [cis-CPD] = 0, 10, 30, 50, 100, 200, 500, 800
mmol dmϪ3 (from bottom to top at 460 nm); (b) Sample B, [-
glucose] = 3.38 mmol dmϪ3, [cis-CPD] = 0, 10, 30, 50, 100, 200, 500, 800
mmol dmϪ3 (from bottom to top at 460 nm).
CD-active, one must consider that the CD band arising from
∆ vs. Λ chirality overlaps with the saccharide-induced CD
band. In fact, a mixture of racemic [CoIII(bpydb)3]3Ϫ and
-glucose resulted in a CD spectrum similar to Fig. 10(b). To
estimate whether the CD spectra truly contain the CD band
arising from ∆ vs. Λ chirality we added an excess amount of
optically-inactive cis-cyclopentane-1,2-diol (cis-CPD): this
compound shows a high affinity with boronic acids and can
competitively substitute -glucose bound to the boronic acid
groups in [CoIII(bpydb)3]3Ϫ.19 As shown in Fig. 10(b), the CD
CD spectral measurements
The CD spectrum of the [CoIII(bpydb)3]3Ϫ–saccharide complex
obtained at 4 ЊC in the presence of -glucose (Sample A) is
shown in Fig. 10(a). We also prepared a complex (Sample B) in
which -glucose was added after [CoII(bpydb)3]4Ϫ was oxidized
to [CoIII(bpydb)3]3Ϫ (Fig. 10(b)). Although both samples are
J. Chem. Soc., Perkin Trans. 2, 1998, 2281–2288
2285

View Article Online
Table 1 Configuration and enantiomeric excess of [CoIII(bpy)3]3ϩ
obtained from the reaction of [CoIII(bpydb)3]3Ϫ and AgNO3
band of Sample B decreased with increasing cis-CPD concen-
tration and almost disappeared at [cis-CPD]/[-glucose] = 25.
In contrast, the CD band of Sample A also decreased with
increasing cis-CPD concentration but remained still CD-active
with [θ]460 = 1100 deg cm2 dmolϪ1 even in the presence of excess
cis-CPD. The difference clearly indicates that Sample B consists
of a 1:1 racemic mixture of ∆- and Λ- isomers whereas in
Sample A one isomer exists in excess over the other.
Saccharide
Configuration
e.e. (%)
47
-Glucose
-Glucose
-Galactose
-Mannose
-Talose
∆
Λ
∆
∆
∆
—
Λ
∆
Λ
∆
∆
47
5
10
2
0
Similar CD spectra were also observed for the [CoIII-
(bpymb)3]Ϫ–-glucose complexes. When excess cis-CPD was
added, the CD band of the complex prepared in a manner
similar to Sample B disappeared whereas the CD band of the
complex prepared in a manner similar to Sample A remained
although the intensity ([θ]460 = 180 deg cm2 dmolϪ1) was weaker
than that for Sample A in the [CoIII(bpydb)3]3Ϫ–-glucose
complex. To estimate the enantiomeric excess (e.e.) of the
[CoIII(bpydb)3]3Ϫ–saccharide complexes we applied a HPLC
-Allose
-Fructose
-Xylose
-Fucose
-Ribose
5
6
13
2
-Cellobiose
2
1
method with chiral-packed columns or a H NMR spectro-
scopic method with chiral shift reagents but unfortunately these
methods were not successful. These methods are not suitable to
optical resolution of the metal complex containing six boronic
acid groups. It is known that a boronic acid group in phenyl-
boronic acid can be eliminated by treatment with AgNO3.20
We applied this method to the present metal complexes and
tried to determine the e.e. by comparison with authentic ∆- and
Λ-[CoIII(bpy)3]3ϩ.21 Thus, AgNO3 (500 mg, 2.9 mmol) was
added to the Sample A solution and the mixture was stirred at
room temperature. The progress of the reaction was monitored
by HPLC (Zorbax ODS, water–methanol = 1:4 v/v). It showed
that both [CoIII(bpymb)3] and [CoIII(bpydb)3]3Ϫ are quanti-
tatively converted to [CoIII(bpy)3]3ϩ after one day (Scheme
5).18,21 After membrane filtration (Millipore LCR 13-LH), the
filtrate was purified using gel filtration (Toyopearl HW-40,
water–methanol = 1:1 v/v). The yields of recovered [CoIII-
Fig. 11 CD spectrum of [CoIII(bpy)3]3ϩ (1.00 × 10Ϫ3 mol dmϪ3
)
obtained from Sample A ([-glucose] = 3.38 mmol dmϪ3 in feed)
after elimination of boronic acid groups by AgNO3: 25 ЊC, water–
methanol = 1:1 v/v.
(bpy)3]3ϩ after gel filtration were higher than 80%. The CD
spectrum of [CoIII(bpy)3]3ϩ obtained from the [CoIII(bpydb)3]3Ϫ–
-glucose complex is shown in Fig. 11. By comparison with the
CD spectrum of optically-pure ∆-[CoIII(bpy)3]3ϩ 21 the optical
purity of this sample was determined to be 47% e.e. (∆ excess).
By a similar operation, the optical purity of [CoIII(bpy)3]3ϩ
obtained from the [CoIII(bpymb)3]–-glucose complex was
estimated to be 16% e.e. (∆ excess). The result indicates that
bpydb is a better ligand than bpymb for chiroselective tran-
scription. We repeated the sample preparation from [CoIII-
(bpydb)3]3Ϫ in the presence of various saccharides and deter-
mined the optical purities of the resultant [CoIII(bpy)3]3ϩ by CD
spectroscopy. The results are summarized in Table 1. Examin-
ation of Table 1 reveals that the highest optical yield is obtained
from glucose: - and -glucose yield the ∆- and Λ-enantiomer
in excess, respectively both in 47% e.e. Hence, the optimization
was carried out with [CoII(bpydb)3]4Ϫ and -glucose which gave
the highest e.e. among the saccharides tested.
O
O
O
–
O
–
B
B
HO
OH
N
N
HO
OH
III
B
Co
N
N
B
–
–
O
O
O
O
N
N
O
O
O
O
–
–
B
B
HO
OH
Fig. 12(a) shows the influence of the -glucose concentration
(in feed) on the e.e. of ∆-[CoIII(bpy)3]3ϩ. The maximal e.e. (47%
e.e.) appeared at [-glucose] = 5.66 × 10Ϫ3 mol dmϪ3, which
corresponds to [-glucose]/[bpydb] = 2.0. This ratio is the same
as that observed for the plot of the CD intensity of bpydb vs.
[-glucose] (see Fig. 5). Since bpydb and -glucose form a cyclic
1:1 complex (see Fig. 6), the result supports the view that the
stoichiometry of bpydb/-glucose in the [CoIII(bpydb)3]3Ϫ–
-glucose complex should be also 1:1 at the maximal CD
intensity. Fig. 12(b) shows the influence of the reaction temper-
ature on the e.e. of ∆-[CoIII(bpy)3]3ϩ. It is seen from Fig. 12(b)
that the e.e. increases with lowering the reaction temperature and
at Ϫ25 ЊC it even reaches 79% e.e. Under similar conditions
in the presence of -glucose we could obtain Λ-[CoIII(bpy)3]3ϩ
in 79% e.e.
HO
HO
OH
OH
: Saccharide
AgNO3
elimination of
boronic acid groups
N
N
N
CoIII
N
N
N
Comments on the mechanism of chiroselective transcription
The data collected in Table 1 indicate that in general,
Scheme 5
2286
J. Chem. Soc., Perkin Trans. 2, 1998, 2281–2288

View Article Online
Experimental
Materials
2,2Ј-Bipyridine-4-boronic acid (bpymb) was synthesized from
2,2Ј-bipyridine via its N-oxide intermediate. Since the prepar-
ation of 4-nitro-2,2Ј-bipyridine-N-oxide has already been
reported,25 the synthetic procedures starting from this com-
pound are described below. 2,2Ј-Bipyridine-4,4Ј-diboronic acid
(bpydb) was also synthesized from 2,2Ј-bipyridine. In the syn-
thetic route the preparation of 4,4Ј-dibromo-2,2Ј-bipyridine
was already reported, so that the synthetic procedures starting
from this compound are described below.
4-Bromo-2,2Ј-bipyridine-N-oxide. 4-Nitro-2,2Ј-bipyridine-N-
oxide (5.2 g, 23.9 mmol) was dissolved in acetic acid (50 ml)
containing acetyl bromide (26.0 ml, 35.2 mmol) and the solu-
tion was refluxed for 9 h under a nitrogen atmosphere. The
progress of the reaction was monitored by a TLC method (sil-
ica gel, Rf = 0.1, chloroform). After cooling, the solution was
poured into an ice–water mixture (100 g) and the pH of the
aqueous phase was adjusted to 11 with aqueous KOH solution.
The product was extracted from the aqueous phase with chloro-
form. The organic solution was dried over MgSO4 and evapor-
ated to dryness. The residue was subjected to flush column
chromatography (silica gel, chloroform) for the purification:
(5.3 g, 89%), mp 65.8–66.2 ЊC; δH [250 MHz; CDCl3] 7.37–7.44
(2H, m, 5-H, 4Ј-H), 7.85 (1H, t, J 7.1, 8.3, 5Ј-H), 8.26 (1H, d,
J 7.0, 3Ј-H), 8.40 (1H, s, 3-H), 8.75 (1H, d, J 8.7, 6-H), 8.90
(1H, d, J 8.1, 6Ј-H). This product was used for the next reaction
without further purificaton.
4-Bromo-2,2Ј-bipyridine. 4-Bromo-2,2Ј-bipyridine-N-oxide
(5.3 g, 21.1 mmol) and PBr3 (7.0 ml, 73.9 mmol) were dissolved
in chloroform (50 ml) and the solution was refluxed for 4 h
under a nitrogen atmosphere. The progress of the reaction was
monitored by a TLC method (silica gel, Rf = 0.2, chloroform).
After cooling, the solution was poured into an ice–water mix-
ture and the pH of the aqueous phase was adjusted to 11 with
aqueous KOH solution. The product was extracted with chloro-
form, the organic layer being dried over MgSO4. The solution
was evaporated to dryness and the residue was subjected to
flush column chromatography (silica gel, chloroform) for the
purification: (2.5 g, 50%), mp 52.9–53.6 ЊC (Found: C, 51.26;
H, 3.01; N, 11.96. C10H7N2Br requires C, 51.26; H, 3.06; N,
11.95%); δH [250 MHz; CDCl3] 7.34 (1H, dd, J 4.8, 7.7, 5Ј-H),
7.48 (1H, d, J 5.5, 5-H), 7.83 (1H, t, J 8.1, 7.6, 4Ј-H), 8.39 (1H,
d, J 8.1, 3Ј-H), 8.49 (1H, d, J 5.2, 6-H), 8.63 (1H, s, 3-H), 8.69
(1H, d, J 4.8, 6Ј-H).
2,2Ј-Bipyridine-4-boronic acid (bpymb). 4-Bromo-2,2Ј-
bipyridine (1.0 g, 4.25 mmol) was dissolved in anhydrous
diethyl ether (80 ml) in a nitrogen-flushed flask and the solution
was cooled to Ϫ100 to Ϫ110 ЊC with a N2 (liq.)–methanol bath.
An n-hexane solution (3.12 ml) containing n-BuLi (1.5 mol
dmϪ3, 4.68 mmol) was added dropwise and the reaction mixture
was stirred at this temperature for 30 min. Trimethyl borate
(0.49 ml, 4.68 mmol) was added: the mixture was stirred for 1 h
at this temperature and then for one day while gradually raising
the reaction temperature to room temperature. The reaction
mixture was concentrated to dryness, the residual solid being
dissolved in methanol. The methanol solution was poured into
acetone, the precipitate being recovered by filtration: (450 mg,
52%), mp (decomp.) >250 ЊC (Found: C, 59.87; H, 4.60; N,
13.89. C10H9BN2O2 ϩ 0.05 CH3OH requires C, 60.67; H, 4.21;
N, 13.45%); δH [250 MHz; CD3OD] 7.38 (1H, dd, J 4.6, 8.3,
5Ј-H), 7.58 (1H, d, J 4.5, 5-H), 7.90 (1H, t, J 7.6, 8.5, 4Ј-H), 8.08
(1H, d, J 7.8, 3Ј-H), 8.32 (1H, s, 3-H), 8.38 (1H, d, J 4.5, 6Ј-H),
8.62 (1H, d, J 4.4, 6-H).
Fig. 12 Influence of the -glucose concentration (a) and the reaction
temperature (b) on the e.e. of [CoIII(bpy)3]3ϩ
.
-saccharides tend to orientate bpydb to form the ∆ enantio-
mers in preference to the Λ enantiomers. The reason for this is
not clear yet. However, the comparison of the CD spectral data
of the bpydb–-saccharide complexes with the ∆ vs. Λ prefer-
ence supports the view that the chiral twist of the two pyridine
rings in bpydb by -saccharides is the origin of the chirality
creation in the final [CoIII(bpy)3]3ϩ complex, because (i) the
bpydb–-fucose complex which has a CD spectrum different
from the others (Fig. 7(b)) results in Λ enantiomers with 13%
e.e., (ii) -allose and -ribose which only give weak CD inten-
sity with bpydb (Fig. 7(a)) result in a very low e.e. (0–2% e.e.)
and (iii) the highest e.e. value for [CoIII(bpydb)3]3Ϫ is obtained at
the -glucose concentration (i.e., [-glucose]/[bpydb] = 2.0)
where the highest CD intensity for the bpydb–-glucose
complex is observed.
It is also true, however, that the chirality creation is not only
due to the chiral twist of the two pyridine rings, because (i) the
magnitude of the e.e. values is not necessarily correlated with
the CD intensity of the bpydb–-saccharide complexes and (ii)
bpymb which cannot form the -saccharide-containing cyclic
complex also yields ∆-excess [CoIII(bpy)3]3ϩ in the presence
of -glucose although the e.e. is lower than that for bpydb.
These findings support the view that steric crowding among
saccharide-carrying bpymb or bpydb and the inter-ligand
bridging by saccharides in the CoII complexes also affect the
∆ vs. Λ preference. We now consider that the chiral twist of
the two pyridine rings by intra-ligand bridging is the primary
factor and the other effects in the CoII complex form the
secondary factor.
Conclusions
The foregoing results clearly establish that the saccharide-
templated synthesis is useful as a new concept for the pre-
paration of chiral tris(2,2Ј-bipyridine)–metal complexes.
Furthermore, the ∆ vs. Λ equilibrium can be shifted in either
direction by the selection of saccharide enantiomers. The
chemistry of chiral tris(2,2Ј-bipyridine)–metal complexes has
become a very active area of research in relation to asym-
metric synthesis, asymmetric memory transduction, nonlinear
optics, photo-induced electron or energy transfer, etc.22–24 How-
ever, these fields have always been hampered by difficulty in the
preparation of the chiral metal complexes. We believe that the
present method provides an important breakthrough for these
advanced fields of science.
2.2Ј-Bipyridine-4,4Ј-diboronic acid (bpydb). This compound
was synthesized from 4,4Ј-dibromo-2,2Ј-bipyridine in a manner
similar to that described for bpymb except for the reaction
solvent (anhydrous THF was used instead of anhydrous diethyl
ether). The reaction was stopped by the addition of water (1.0
J. Chem. Soc., Perkin Trans. 2, 1998, 2281–2288
2287

View Article Online
4 Y. Nagai, K. Kobayashi, H. Toi and Y. Aoyama, Bull. Chem. Soc.
Jpn., 1993, 66, 2965.
5 G. Wulff, S. Krieger, B. Kubneweg and A. Steigel, J. Am. Chem.
Soc., 1994, 116, 409 and references cited therein.
6 K. Tsukagoshi and S. Shinkai, J. Org. Chem., 1991, 56, 4089;
Y. Shiomi, M. Saisho, K. Tsukagoshi and S. Shinkai, J. Chem. Soc.,
Perkin Trans. 1, 1993, 2111; J. C. Norrild and H. Eggert, J. Am.
Chem. Soc., 1995, 117, 1479.
7 For comprehensive reviews on boronic acid-based saccharide
receptors, see: T. D. James, K. R. A. S. Sandanayake and S. Shinkai,
Supramol. Chem., 1995, 6, 141; T. D. James, P. Linnane and
S. Shinkai, Chem. Commun., 1996, 281; T. D. James, K. R. A. S.
Sandanayake and S. Shinkai, Angew. Chem., Int. Ed. Engl., 1996,
35, 1911.
8 T. D. James, K. R. A. S. Sandanayake and S. Shinkai, Angew. Chem.,
1994, 106, 2287; Angew. Chem., Int. Ed. Engl., 1994, 22, 2207;
J. Chem. Soc., Chem. Commun., 1994, 477.
9 M. Mikami and S. Shinkai, J. Chem. Soc., Chem. Commun., 1995,
153; Chem. Lett., 1995, 603; M. Takeuchi, T. Imada and S. Shinkai,
J. Am. Chem. Soc., 1996, 118, 10 658.
10 S. Arimori, M. Takeuchi and S. Shinkai, J. Am. Chem. Soc., 1996,
118, 245; L. D. Sarson, K. Ueda, M. Takeuchi and S. Shinkai,
Chem. Commun., 1996, 619; M. Takeuchi, Y. Chin, T. Imada and
S. Shinkai, ibid., 1996, 1867.
11 H. Yamashita, K. Amano, S. Shimada and K. Narasaka, Chem.
Lett., 1996, 537; H. Yamashita and S. Narasaka, ibid., 1996, 539.
12 K. Nakashima and S. Shinkai, Chem. Lett., 1994, 1267.
13 Preliminary communication: T. Mizuno, M. Takeuchi, I. Hamachi,
K. Nakashima and S. Shinkai, Chem. Commun., 1997, 1793.
14 In 1964 Liu et al. reported that the asymmetric synthesis of
[RuII(bpy)3]2ϩ is achieved (although in low e.e.) by the reaction
of Ru2ϩ and 2,2Ј-bipyridine in the presence of sodium -tartrate or
sucrose: C. F. Liu, N. C. Liu and X. Bailar, Jr., Inorg. Chem., 1964,
3, 1085. This chirality is induced when -tartrate or sucrose
coordinated to Ru2ϩ is substituted with 2,2Ј-bipyridine and the
concept is entirely different from the present system.
15 This prospect is supported by our previous finding that the
two planes of biphenyl-3,3Ј-boronic acid can be asymmetrically
immobilized by added saccharides: K. Kondo, Y. Shiomi,
M. Saisho, T. Harada and S. Shinkai, Tetrahedron, 1992, 48, 8239.
16 In the present research, the furanose vs. pyranose structural problem
is not specified in detail. The appearance of a clear exciton coupling
band in bpydb supports the view that the two boronic acid groups
are intramolecularly bridged by the monosaccharide. Judging
from the distance between the two boronic acid groups (ca. 7.4 Å),
the distance between the 1,2-diol and the 4,6-diol which are useful
as building sites in -glucopyranoside is too short whereas that
between the 1,2-diol and the 5,6-diol which are useful as building
sites in -glucofuranoside is long enough to interact with the two
boronic acids. We now believe that the furanose form is used for the
intramolecular bridging of the two boronic acids.
ml). The gelatinous reaction mixture thus obtained was concen-
trated in vacuo to dryness, the solid residue being dissolved in
methanol. The methanol solution was poured into acetone, the
precipitate being recovered by filtration. Finally, the product
was washed with a small amount of methanol and then with
diethyl ether: (700 mg, 80%), mp (decomp.) >250 ЊC (Found: C,
51.11; H, 6.44; N, 7.45. C16H24B2Li2N2O6 requires C, 51.03; H,
6.03; N, 7.35%); δH [250 MHz; CD3OD] 7.54 (2H, d, J 4.6, 5,5Ј-
H), 8.16 (2H, s, 3,3Ј-H), 8.36 (2H, d, J 4.6, 6,6Ј-H). The elem-
ental analysis data indicates that the boronic acid groups are
isolated as ϪBϪ(OMe)3 Liϩ. Although these methoxy protons
could not be found by 1H NMR spectroscopy in CD3OD, they
were easily found as CH3OD (3.34 ppm, integral intensity 18H
as a hydrolysis product of ϪBϪ(OMe)3 Liϩ) in D2O.
General procedure for the preparation of [CoIII(bpy)3]3؉ from
[CoIII(bpymb)3] or [CoIII(bpydb)3]3؊. AgNO3 (500 mg, 2.9 mmol)
was added to the sample solution (2.0 ml: pH 7.4 with
5.0 × 10Ϫ5 mol dmϪ3 MOPS buffer–methanol (1:1 v/v)) con-
taining a [CoIII(bpymb)3]–saccharide complex or a [CoIII-
(bpydb)3]3Ϫ–saccharide complex and the mixture was stirred at
room temperature for one day. The progress of the reaction was
monitored by a HPLC method (Zorbax ODS, water–methanol
(1:4 v/v)). It showed that these complexes are quantitatively
converted to [CoIII(bpy)3]3ϩ after one day. After membrane fil-
tration (Millipore LCR 13-LH), the filtrate was purified using
gel filtration (Toyopearl HW-40, water–methanol (1:1 v/v)).
The yields of recovered [CoIII(bpy)3]3ϩ were higher than
80%. The products thus obtained were subjected to the CD
measurement.
Miscellaneous. For the phototitration of bpymb and bpydb,
the following buffer solutions were used: 50 mmol dmϪ3 HCl for
pH 1.0–2.0, 50 mmol dmϪ3 HCOOH for pH 2.5–3.0, 50 mmol
dmϪ3 acetate for pH 3.5–5.0, 50 mmol dmϪ3 phosphate for pH
6.5–9.0, 50 mmol dmϪ3 carbonate for pH 9.5–10.5, 50 mmol
dmϪ3 NaOH for pH 11.0–12.0. Absorption spectra, CD spectra
1
and H NMR spectra were measured with Hitachi V-3000,
Jasco J-720 and Bruker AC-250P spectrometers. HPLC
analyses were performed on a Jasco PU-980 pump equipped
with a Zorbax ODS column.
Acknowledgements
We thank Dr Tony D. James (University of Birmingham) for
helpful discussions. This work was supported by a Grant-in-
Aid for COE Research ‘Design and Control of Advanced
Molecular Assembly Systems’ from the Ministry of Education,
Science and Culture, Japan (#08CE2005).
17 D. F. Shriver, P. W. Atkins and C. H. Langfold, Inorganic Chemistry,
2nd edn., Oxford, 1994, p. 266.
18 F. H. Burstall and R. S. Nyholm, J. Chem. Soc., 1952, 3570.
19 T. Kimura, S. Arimori, M. Takeuchi, K. Nagasaki and S. Shinkai,
J. Chem. Soc., Perkin Trans. 2, 1995, 1889; M.Takeuchi, T. Mizuno,
M. Shinmori and S. Shinkai, Tetrahedron, 1996, 52, 1195.
20 K. Torssell, Ark. Kemi, 1957, 10, 473.
References
21 S. F. Mason and B. J. Deart, J. Chem. Soc., Dalton Trans., 1973, 949.
22 K. Ohkubo, H. Ishida, T. Hamada and T. Inaoka, Chem. Lett.,
1989, 1545; H. Ishida, T. Hamada, Y. Fujishima, Y. Saito and
K. Ohkubo, Bull. Chem. Soc. Jpn., 1933, 66, 714.
1 For comprehensive reviews on hydrogen-bond-based synthetic
receptors, see: J. Rebek, Jr., Angew. Chem., Int. Ed. Engl., 1990,
29, 245; A. D. Hamilton, Bioorg. Chem. Front., 1991, 2, 115;
J.-H. Fuhrhop and J. Köning, Membranes and Molecular
Assemblies: The Synkinetic Approach, ed. J. F. Stoddart, Royal
Society of Chemistry, Cambridge, 1994; G. M. Whitesides, E. E.
Simanek, J. P. Mathias, T. S. Christopher, N. C. Donovan,
M. Mathai and M. G. Dana, Acc. Chem. Res., 1995, 28, 37.
2 J. Yoon and A. W. Czarnik, J. Am. Chem. Soc., 1992, 114, 5874;
L. K. Mohler and A. W. Czarnik, ibid., 1993, 115, 2998.
23 A. Tatehara and A. Muraida, Chem. Lett., 1996, 461.
24 C. Dhenaut, I. Ledoux, I. D. W. Samuel, J. Zyss, M. Bourgault and
H. L. Bozec, Nature, 1995, 374, 339.
25 D. Wenkert and R. B. Woodward, J. Org. Chem., 1983, 48, 283.
3 P. R. Westmark and B. D. Smith, J. Am. Chem. Soc., 1994, 116, 9343
and references cited therein.
Paper 8/03382J
2288
J. Chem. Soc., Perkin Trans. 2, 1998, 2281–2288