Glycosylated tris-bipyridine ferrous complexes as molecular mimics of densely packed glycoclusters on cell surfaces: spatial carbohydrate packing of glycoclusters changes on additions of salts

Article abstract of DOI:10.1080/07328303.2019.1615500

Tris-bipyridine ferrous complexes having β-lactosides, β-maltosides or α-mannosides with serinol spacers were prepared as molecular mimics of densely packed carbohydrate clusters on cell surfaces. Conformational analysis on these glycosylated complexes we

Full text of DOI:10.1080/07328303.2019.1615500

Journal of Carbohydrate Chemistry
ISSN: 0732-8303 (Print) 1532-2327 (Online) Journal homepage: https://www.tandfonline.com/loi/lcar20
Glycosylated tris-bipyridine ferrous complexes as
molecular mimics of densely packed glycoclusters
on cell surfaces: spatial carbohydrate packing of
glycoclusters changes on additions of salts
Naoto Chigira, Nao Maeda, Kanako Tachikawa, Maki Sekiguchi, Yoshitsugu
Amano, Mayu Inokuchi, Qintong Li & Teruaki Hasegawa
To cite this article: Naoto Chigira, Nao Maeda, Kanako Tachikawa, Maki Sekiguchi, Yoshitsugu
Amano, Mayu Inokuchi, Qintong Li & Teruaki Hasegawa (2019): Glycosylated tris-bipyridine
ferrous complexes as molecular mimics of densely packed glycoclusters on cell surfaces: spatial
carbohydrate packing of glycoclusters changes on additions of salts, Journal of Carbohydrate
Chemistry, DOI: 10.1080/07328303.2019.1615500
To link to this article: https://doi.org/10.1080/07328303.2019.1615500
View supplementary material
Published online: 10 Jun 2019.
Submit your article to this journal
View Crossmark data
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=lcar20

JOURNAL OF CARBOHYDRATE CHEMISTRY
https://doi.org/10.1080/07328303.2019.1615500
Glycosylated tris-bipyridine ferrous complexes as
molecular mimics of densely packed glycoclusters on
cell surfaces: spatial carbohydrate packing of
glycoclusters changes on additions of salts
Naoto Chigiraa, Nao Maedab, Kanako Tachikawab, Maki Sekiguchib,
Yoshitsugu Amanoa, Mayu Inokuchib, Qintong Lib, and Teruaki Hasegawab,c
b
aGraduate School of Life Sciences, Toyo University, Ora-gun, Japan; Faculty of Life Sciences,
c
Toyo University, Ora-gun, Japan; Bio-Nano Electronics Research Centre, Toyo University,
Kawagoe, Japan
ABSTRACT
ARTICLE HISTORY
Received 23 March 2019
Accepted 30 April 2019
Tris-bipyridine ferrous complexes having b-lactosides,
b-maltosides or a-mannosides with serinol spacers were pre-
pared as molecular mimics of densely packed carbohydrate
clusters on cell surfaces. Conformational analysis on these gly-
cosylated complexes were conducted by UV–vis and circular
dichroism spectroscopy measurements, which disclosed that
the chloride, nitrate and sulfate salts induced the conform-
ational changes of the glycosylated complexes in the anion-
and carbohydrate-dependent manners.
KEYWORDS
Carbohydrate–carbohydrate
interaction; carbohydrate
cluster; circular dichroism;
dynamic combinatorial
library; spatial carbohydrate
packing; tris-bipyridine
ferrous complex
GRAPHICAL ABSTRACT
Introduction
Glycosphingolipids (GSLs) in cell membranes aggregate laterally to form
GSL-enriched microdomains with densely packed carbohydrate clusters
(glycoclusters) on cell surfaces (Fig. 1a).[1] In the last three decades,
increasing effort has been paid to their functional analysis and, as a result,
CONTACT Teruaki Hasegawa
t-hasegawa@toyo.jp
Faculty of Life Sciences, Toyo University, 1-1-1 Izumino,
Itakura-machi, Ora-gun, Gumma 374-0193, Japan.
Supplemental data for this article can be accessed on the publisher’s website.
ß 2019 Taylor & Francis Group, LLC

2
N. CHIGIRA ET AL.
Figure 1. (a) Illustration of glycoclusters on a cell surface and (b) structural similarity between
LacCer and bpySerbLac as representatives of GSLs and glycobpys in this work, respectively.
it has been found that cell-surface glycoclusters play substantial roles in
cell–cell adhesion; multiple carbohydrate–carbohydrate interactions (CCIs)
take place in interfacial manners between two adjacent cells and the result-
ant cell–cell adhesion is then reinforced by strong carbohydrate–protein
and protein–protein interactions.[2] Thus far, some combinations of GSLs
were reported for inducing the CCIs; specific interactions between LeX and
[4]
LeX [3] and that between GM3 and Gg3
were reported to induce compac-
tion of blastomeres and migration of cancer cells, respectively (see Chart S1
for structures of LeX, GM3, and Gg3). These two examples of CCIs clearly
indicate that a clearer understanding on a mechanism behind CCIs should

JOURNAL OF CARBOHYDRATE CHEMISTRY
3
provide essential clues about human embryogenesis, and facilitate the
design of new drugs for preventing unfavorable cell–cell adhesions.
However, structural diversity of GSLs, fluidic nature of cell membranes,
and fluctuations of their GSL levels, all combined together, has strongly
hindered quantitative and molecular-level investigation on CCIs. In fact,
very few information has been obtained on CCIs thus far. To circumvent
these difficulties and to obtain the molecular-level information on CCIs,
some research groups developed the artificial glycoclusters that mimicked
cell-surface glycoclusters.[5–12] For example, Penades et al. developed the
ꢀ
gold nanoparticles (GNPs) carrying multiple LeX-trisaccharides (LeX ,
3
Chart S1) for probing selective LeX-LeX interactions.[5] In their work, the
transmission electron microscopic (TEM) imagings were conducted for
monitoring the Ca2þ-mediated aggregations of the LeX -immobilized
3
GNPs, thereby indicating Ca2þ as the essential cation to mediate LeX-LeX
interactions. Reynolds et al. also developed the similar GNPs-based systems
having multiple b-lactosides (bLac, Chart S1) and demonstrated that Ca2þ
induced the aggregation of the bLac-immobilized GNPs, thereby indicating
that Ca2þ also induced selective bLac-bLac interactions.[6] Note that bLac
is the carbohydrate moiety of lactosylceramide (LacCer), which is one of
major GSLs in cell membranes as well as a common structural motif of
various GSLs (e.g., GM3 and Gg3). The specific CCIs were probed also by
using the other artificial glycoclusters, such as the glycosylated linear poly-
mers,[7] dendrimers,[8] cyclodextrins,[9] and liposomes.[10] It should be
important to herein note that researchers in this field commonly believe
the followings; (1) specific metal cations (e.g., Ca2þ) function as molecular
glue joining two carbohydrate units together; and (2) formation of multiple
carbohydrate–cation–carbohydrate complexes is the main driving force for
inducing CCIs. Based on these ideas, most of the previous works focused
the cation- and the carbohydrate-selectivity of the CCIs.
In 2000, one of the authors (TH) reported the synthesis and the func-
tional analysis of the artificial glycoclusters along poly(phenylisocyanide)
(PPI) mainchains.[13] Since the PPI mainchains had the rigid and helical
structures, spatial carbohydrate packing (hereafter referred to as glycopack-
ing) of the PPI-based glycoclusters was fixed into the helical ones. Of quite
interest, these PPI-based glycoclusters adsorbed onto the hydrophilic surfa-
ces in a multi-layer manner via the inter-molecular CCIs between their
densely packed carbohydrates. Conversely, the polyacrylamide-type glyco-
clusters having flexible glycopacking did not have the adsorption property,
thereby indicating that glycopacking is an important factor to induce CCIs.
However, only few works have been conducted thus far to reveal the rela-
tionship between glycopackings and CCIs.[14,15] A pioneering work was the
conformational analysis of the LeX -dimers reported by Schmidt et al.[14]
3

4
N. CHIGIRA ET AL.
They developed the LeX -dimers in which two LeX units were linked
3
3
together with the flexible spacers, and conducted the 2 D nuclear magnetic
resonance (NMR) spectroscopy measurements on the LeX -dimers. They
3
thereby revealed that Ca2þ induced the cross-shaped glycopackings of the
two LeX units.
3
We also developed the molecular mimics for investigating the effects of
the ions on glycopacking.[16,17] In a series of our works, we recently devel-
oped the tris-bipyridine (tris-bpy) ferrous complexes presenting hexavalent
bLac, b-maltosides (bMal), and b-cellobiosides (bCel) and, found that the
glycopacking of these [Febpy3]2þ-based artificial glycoclusters (metallogly-
coclusters) changed on additions of the salts, mainly in the anion-depend-
ent manners.[16] These data suggest that the glycopacking of cell-surface
glycoclusters may change in the presence of the anions. In addition, rhe-
ology of the metalloglycocluster solution also changed on additions of the
specific salts, possibly owing to the intermolecular CCIs. This result indi-
cated that the salt-induced glycopacking changes and the CCIs were linked
to each other. However, the metalloglycoclusters in this previous work had
a serious structural problem; their carbohydrate appendages and [Febpy3]2þ
cores were tethered with the phenyl spacers. By contrast, the carbohydrate
heads and the hydrocarbon tails are connected with the serinol spacers in
natural GSLs (Fig. 1b). This structural difference deteriorated validity of the
metalloglycoclusters as the molecular mimics, and reliability of our new
finding on the function of the specific anions for changing the glycopack-
ing. To obtain more reliable evidences on the anion-induced glycopacking
change of cell-surface glycoclusters, the phenyl spacers in the metallogly-
coclusters should be changed into the serinol ones to mimic reasonably the
glycoclusters on cell surfaces. In this respect, we designed the new bis-
glycosylated bpy derivatives (glycobpys) in which the carbohydrate appen-
dages were tethered to the bpy units through the serinol spacers. We herein
report a synthetic route to access the glycobpys having bLac, bMal, and
a-mannosides (aMan), the complexation of the glycobpys with Fe2þ to pre-
pare the corresponding metalloglycoclusters, and their glycopacking
changes induced by the alkaline/alkaline-earth metal salts.
Results and discussion
Synthesis of glycobpys
To obtain the glycobpys, we prepared the L-serine derivative having a Cbz-
protected amino group and a methyl carboxylate via a two-step synthetic
scheme (Scheme 1). Briefly, commercially available L-serine (1) was con-
verted into the corresponding methyl ester (SerMe, 2) by SOCl2 in MeOH.
SerMe was then treated with benzyl chloroformate (CbzCl) in the presence

JOURNAL OF CARBOHYDRATE CHEMISTRY
5
Scheme 1. Reaction conditions: (i) MeOH, SOCl2, reflux, overnight, 32%, (ii) CbzCl, NaHCO3 aq,
CH2Cl2, room temperature (rt), overnight, 92%.
Scheme 2. Reaction conditions: (i) CbzSerMe, BF3OEt2, CH2Cl2, rt, 2–12 h, (ii) H2, Pd/C, THF, rt,
overnight, 19–28% (2 steps), (iii) bpyCOCl, Et3N, THF, rt, 20 min, 80–97%, iv) NaBH4, THF, water,
rt, 1–4 day, 69–97%.
of Et3N for the selective protection of its amino group. The resultant serine
derivative (CbzSerMe, 3) was then glycosylated with per-acetylated lactose
(LacAc), maltose (MalAc), and mannose (ManAc) in the presence of
BF3OEt2 to give the corresponding L-serine derivatives having carbohydrate
units (CbzSerMeXAc: X ¼ bLac, bMal, and aMan, respectively, Scheme 2).
After removal of the Cbz groups, the glycosylated L-serine derivatives hav-
ing free amino groups were coupled with 2,2-bipyridine-5,5-dicarboxylyl
chloride (bpyCOCl) to give the bpy derivatives having two carbohydrate
units (bpySerMeXAc: X ¼ bLac, bMal, and aMan). The conversion of the
methyl carboxylates into the hydroxymethyl groups and the removal of the

6
N. CHIGIRA ET AL.
Scheme 3. Reaction conditions: FeCl2 (3 eq.), H2O, rt, quant.
acetyl groups were simultaneously achieved by using NaBH4 in the mixed
solvent system composed of H2O and tetrahydrofurane (THF). The glyco-
bpys having two carbohydrates with the serinol spacers (bpySerX: X ¼
bLac, bMal, and aMan) were thereby obtained. Chemical constitution of
1
the glycobpys was confirmed by their H and 13C NMR, infrared (IR), and
high-resolution electrospray ionization time-of-flight mass (HR ESI-TOF
MS) spectra.
Preparation of tris-bpy ferrous complexes having six carbohydrate
appendages
The glycobpys were highly water-soluble and thus, smoothly coordinated
onto Fe2þ in water (Scheme 3). For example, aqueous bpySerbLac solution
was colorless and transparent before the complexation. However, it imme-
diately turned to red purple on a mixing with aqueous FeCl2 solution (3
eq.). The resultant solution showed a UV-vis spectrum having a clear
metal-to-ligand charge transfer (MLCT) band at 562 nm, thereby proving
the successful formation of the corresponding tris-bpy ferrous complexes
([Fe(bpySerbLac)3]2þ). The complexation of the other glycobpys with Fe2þ
and the preparation of the corresponding tris-bpy ferrous complexes were
also proved by the appearance of the MLCT bands on the complexation.
Tris-bpy ferrous complexes are known to have unique 3 D structures in
which their bpy units coordinate onto Fe2þ with right-handed (D) or left-
handed (K) screw-like spatial arrangements. The metalloglycoclusters in
this work showed CD spectra having negative-to-positive (weak) and

JOURNAL OF CARBOHYDRATE CHEMISTRY
7
Figure 2. CD spectra of [Fe(bpySerbLac)3]2þ (black plane line), [Fe(bpySerbMal)3]2þ (gray thin
line), and [Fe(bpySeraMan)3]2þ (gray bold line); [glycobpys] ¼ 2.44 mM, [FeCl2] ¼ 7.33 mM,
d ¼ 0.2 mm, rt, in water.
positive-to-negative (strong) circular dichroism (CD) signals at around 550
and 310 nm, respectively, on going from longer to shorter wavelength
ꢀ
(Fig. 2). The strong CD signals at 310 nm were assignable to d-d transi-
tion bands of the [Febpy3]2þ cores, and the positive-to-negative CD pat-
terns in this wavelength region indicated that the metalloglycoclusters
existed predominantly in their K-isomers.[18] However, the CD signal
intensities were much weaker in comparison to those of enantio-pure
K-[Febpy3]2þ, thereby indicating the low diastereomeric excesses of the
metalloglycoclusters. It was quite difficult to evaluate their diastereomeric
excesses in quantitative manners owing to the magnetic nature of Fe2þ and
the reversible Fe-N bonds that strongly hindered the NMR and high-per-
formance liquid chromatography (HPLC) analyses, respectively.
We assumed that the (R)-configuration of the asymmetric carbons in the
serinol spacers should play a predominant role for inducing the K-isomers.
This assumption was reasonably supported by the following facts; (1) the
asymmetric carbons in the serinol spacers had closer proximity to the com-
plex cores than that between the carbohydrate units and the complex cores;
and (2) the different carbohydrate units had negligible impact on the
CD spectra.

8
N. CHIGIRA ET AL.
Salt-induced UV-vis and CD spectral changes of the metalloglycoclusters
Salt-induced conformational changes of [Fe(bpySerbLac)3]2þ
We conducted the UV-vis and CD spectroscopy titration assays on the met-
alloglycoclusters for investigating the effect of salts on glycopacking, and
found that the UV-vis and CD spectra of [Fe(bpySerbLac)3]2þ changed on
additions of the salts (Fig. 3). For example, additions of the chloride salts
(NaCl, KCl, and CaCl2, gray symbols in Fig. 4a, top) slightly reduced the
MLCT intensity of [Fe(bpySerbLac)3]2þ. Conversely, additions of the sulfate
salts (Na2SO4 and K2SO4, white symbols) slightly enhanced the MLCT inten-
sity. Note that owing to limited water solubility of the sulfate salts, we could
not assess the effects of Na2SO4 and K2SO4 on the MLCT intensity with their
concentrations higher than 1.0 and 0.5 M, respectively. In contrast to these
UV-vis spectral changes by the chloride and the sulfate salts, little UV-vis
spectral change was observed on additions of the nitrate salts (NaNO3,
KNO3, and Ca(NO3)2, black symbols). It should be noted that although the
different salts with the same cation (e.g., NaCl and NaNO3) induced the dif-
ferent UV-vis spectral changes of [Fe(bpySerbLac)3]2þ, the different salts
with the same anion (e.g., NaCl and KCl) induced the almost identical UV-
vis spectral changes. These facts suggested that not the cations but the anions
were responsible for the UV-vis spectral changes.
[Fe(bpySerbLac)3]2þ also had different CD spectral responses to the dif-
ferent salts. For example, additions of the alkaline-earth metal salts (e.g.,
CaCl2 and Ca(NO3)2) strongly reduced the CD signals of
[Fe(bpySerbLac)3]2þ (Fig. 4a, bottom). The reduced CD signals indicated
that D-[Fe(bpySerbLac)3]2þ was stabilized by these salts and, as a result,
the K-isomer excess was reduced. Although the CD signals were also
reduced by the alkaline metal salts (e.g., NaCl, KCl, NaNO3, and KNO3),
the effects of the alkaline metal salts to reduce the CD signals were smaller
than that of the alkaline-earth metal salts. This difference between the alka-
line-earth metal salts and the alkaline metal salts strongly supported our
assumption that the anions were responsible for the CD spectral changes.
Note that, for example, CaCl2 released two Cl- ions and thus the
Cl-concentration in the aqueous CaCl2 was two times as high as that in the
aqueous NaCl with the same salt concentrations. The sulfate salts did not
affect the CD spectrum of [Fe(bpySerbLac)3]2þ although the binding
between the sulfate salts and [Fe(bpySerbLac)3]2þ was clearly proved by the
above-mentioned UV-vis spectral changes.
Salt-induced conformational changes of [Fe(bpySerbmal)3]2þ
We conducted the titration assays on [Fe(bpySerbMal)3]2þ to reveal a
mechanism behind the salt-induced UV-vis and CD spectral changes of the

JOURNAL OF CARBOHYDRATE CHEMISTRY
9
Figure 3. (a) UV-vis and (b) CD spectral changes of [Fe(bpySerbLac)3]2þ in the presence of
varying concentrations of CaCl2; [bpySerbLac] ¼ 2.44 mM, [FeCl2] ¼ 7.33 mM, d ¼ 0.2 mm, rt,
in water.
metalloglycoclusters. We thereby found that [Fe(bpySerbMal)3]2þ showed
the salt-induced UV-vis and CD spectral changes that were almost identical
to those of [Fe(bpySerbLac)3]2þ. For example, additions of the chloride
salts (NaCl, KCl, and CaCl2) slightly reduced the MLCT intensity of
[Fe(bpySerbMal)3]2þ (Fig. 4b, top, gray symbols). Among these chloride

10
N. CHIGIRA ET AL.
Figure 4. UV-vis and CD spectral changes of (a) [Fe(bpySerbLac)3]2þ, (b) [Fe(bpySerbMal)3]2þ
,
and (c) [Fe(bpySeraMan)3]2þ, in the presence of varying concentrations of the salts;
[glycobpys] ¼ 2.44 mM, [FeCl2] ¼ 7.33 mM, d ¼ 0.2mm, rt, in water.
salts, CaCl2 reduced the MLCT intensity more significantly than NaCl or
KCl. This difference between the alkaline-earth metal salt and the alkaline
metal salts could be also rationalized by the higher Cl- concentrations in
the aqueous CaCl2 solution than that in the aqueous NaCl or KCl solutions
with the same salt concentrations. In the cases of the sulfate and the nitrate
salts, the sulfate salts enhanced the MLCT intensity (Fig. 4b, top, white
symbols) and the nitrate salts did not change the MLCT intensity of
[Fe(bpySerbMal)3]2þ (Fig. 4b, top, black symbols).
The CD spectrum of [Fe(bpySerbMal)3]2þ also changed on additions of
the salts. For example, additions of the chloride salts clearly reduced the
CD signals of [Fe(bpySerbMal)3]2þ; CaCl2 reduced the CD signals of
[Fe(bpySerbMal)3]2þ more significantly than NaCl or KCl (Fig. 4b, bot-
tom). The nitrate salts also reduced CD signals of [Fe(bpySerbMal)3]2þ in
similar manners to the corresponding chlorides with the same cation (e.g.,
NaNO3 and NaCl). By contrast, the sulfate salts induced negligible CD
spectral change of [Fe(bpySerbMal)3]2þ. It should be again emphasized that
the salt-induced UV-vis and CD spectral change of [Fe(bpySerbMal)3]2þ
was almost identical to that of [Fe(bpySerbLac)3]2þ.
Of the UV-vis and CD spectral changes, the latter directly correlated to
the conformational changes of the [Febpy3]2þ cores although a mechanism
behind the former was still unclear. The salt-induced conformational

JOURNAL OF CARBOHYDRATE CHEMISTRY
11
Table 1. CD spectral changes of the metalloglycoclusters by the anions.
CIꢁ
NO3
SO4
ꢁ
2ꢁ
metalloglycoclusters
carbohydrates
[Fe(bpySerbLac)3]2þ
[Fe(bpySerbMal)3]2þ
[Fe(bpySeraMan)3]2þ
bGal-1,4-bGlc-
aGlc-1,4-bGlc-
aMan-
#
#
#
–
–
–
a
#
#
b
##
aReduced CD intensity and bhighly reduced CD intensity.
change of [Fe(bpySerbLac)3]2þ and [Fe(bpySerbMal)3]2þ were characterized
as follows; (1) not the cations but the anions were responsible for the con-
formational changes; and (2) little difference was found between the salt-
responses of [Fe(bpySerbLac)3]2þ and that of [Fe(bpySerbMal)3]2þ
(Table 1). These facts clearly indicated that the serinol spacers and/or the
reducing terminal carbohydrates—the common structure of the two metal-
loglycoclusters—participated in the binding with the anions and the result-
ant UV-vis and CD spectral changes of the metalloglycoclusters. To
support this assumption, we conducted the titration assays on
[Fe(bpySeraMan)3]2þ, as follows.
Salt-induced conformational changes of [Fe(bpySerbman)3]2þ
By the titration assays on [Fe(bpySeraMan)3]2þ, we found that only small
changes in its MLCT intensity were induced by the salts (Fig. 4c, top). This
is clear contrast to the significant UV-vis spectral changes of
[Fe(bpySerbLac)3]2þ and [Fe(bpySerbMal)3]2þ induced by the chloride and
the sulfate salts. On the other hand, [Fe(bpySeraMan)3]2þ showed salt-
induced CD spectral changes that were similar to those of
[Fe(bpySerbLac)3]2þ and [Fe(bpySerbMal)3]2þ; the nitrate and the chloride
salts reduced the CD signal of [Fe(bpySeraMan)3]2þ, thereby suggesting
that the serinol spacer—the common structure of the three metallogly-
coclusters—play major roles in the salt-induced glycopacking changes. The
hydrogen bonding between the anions and the amide units (i.e.,
–NH–CO–) would be mainly responsible for the affinity of the anions with
the serinol spacers. The similar binding between the anions and the amides
was also observed by using the metalloglycoclusters with the phenyl
spacers, as reported in our previous work.
Although these three metalloglycoclusters had the similar salt-induced
ꢁ
CD spectral responses, [Fe(bpySeraMan)3]2þ had a difꢁferent NO3 -
response compared to the other metalloglycoclusters; NO3 reduced the
CD signal of [Fe(bpySeraMan)3]2þ more significantly than those of the
others (Fig. 4c, bottom). For example, CD signal of [Fe(bpySeraMan)3]2þ
at 320 nm became 15 mdeg when the Ca(NO3)2 concentration was 2.0 M,
and this CD intensity was 0.17-fold compared to the CD intensity (CD0)
without the salt. The relative CD intensity (CD/CD0 ¼ 0.17) under this
Ca(NO3)2 concentration was significantly smaller than those of

12
N. CHIGIRA ET AL.
[Fe(bpySerbLac)3]2þ or [Fe(bpySerbMal)3]2þ; the relative CD intensities
were 0.32 forꢁ[Fe(bpySerbLac)3]2þ and 0.39 for [Fe(bpySerbMal)3]2þ. The
sensitive NO3 -response of [Fe(bpySeraMan)3]2þ compared to the others
indicated that the non-reducing terminal carbohydrates also played consid-
erable roles in the anion-induced glycopacking change of the metallogly-
coclusters. Given the design concept of the metalloglycoclusters, these
findings suggest the possibility that the anions bind to the serinol spacers
and the reducing terminal carbohydrates of GSLs, and affect the glycopack-
ing of the cell-surface glycoclusters.
Molecular dynamics simulation of the glycosylated tris-bipyridine ferrous
complexes with/without the salts
To visualize a binding between the anions and the metalloglycoclusters, we
conducted molecular dynamics (MD) simulations (10 ns equilibration, 90 ns
dynamics, 300 K, e ¼ 80, CHARMM) on a three-components molecular
-
assembly composed of K-[Fe(bpySerbLac)3]2þ, Naþ, and NO3 . As a result,
-
we found a close proximity of NO3 to K-[Fe(bpySerbLac)3]2þ (Fig. 5a).
Conversely, no such close proximity was observed between Naþ and
K-[Fe(bpySerbLac)3]2þ.
To reveal the mechanism behind the interaction between the anions and
ꢁ
the metalloglycoclusters, we carefully monitored the pꢁositions of NO3 dur-
ing the MD simulation. We thereby found that NO3 moved freely round
K-[Fe(bpySerbLac)3]2þ and its positiꢁon was not fixed. These data indicated
the weak interactions between NO3 and K-[Fe(bpySerbLac)3]2þ, and the
MD simulation did not supply any decisive information on their interactions.
For example, as shown above, the UV-vis and the CD titration assays clearly
indicated the binding of the anions onto the serinol and/or the reducing ter-
ꢁ
minal carbohydrate units of the metalloglycoclusters. However, NO3 in the
most stable conformation of the three-components molecular assembly had
closer proximity not to the serinol and/or the reducing terminal carbohy-
drate units (i.e., the Glc units) but to the Gal units (Fig. 5b). This contradict-
ory result obtained by the wet chemistry (i.e., the UV-vis and CD
spectroscopy titration assays) and the dry chemistry (i.e., the MD simulation)
should arise from the weak interactions between the anions and the metallo-
glycoclusters, although we strongly believe that the data obtained by the for-
mer should be more reliable than that obtained by the latter.
Conclusion
We developed the [Febpy3]2þ-based artificial glycoclusters having bLac,
bMal, or bMan appendages. The carbohydrate appendages were tethered to

JOURNAL OF CARBOHYDRATE CHEMISTRY
13
Figure 5. The most stable conformation of the three-components molecular assembly com-
ꢁ
posed of K-[Fe(bpySerbLac)3]2þ, Naþ, and NO3 during the MD simulation (10 ns equilibration,
90 ns dynamics, 300 K, e ¼ 80)ꢁ(a) and an enlarged view around the nitrate anion (b). The
[Febpy3]2þ core, Naþ, and NO3 were depicted in CPK models for a clear presentation. The Gal
and Gal’ represent the Gal units of the different Lac units.
the [Febpy3]2þ cores with the serinol spacers and the similar glycosyl seri-
nol structures were almost invariably observed as parts of the hydrophilic
moieties of the natural GSLs. Therefore, we believe that the metallogly-
coclusters in this work reasonably function as molecular mimics of the cell-
surface glycoclusters. In the absence of the salts, the metalloglycoclusters
showed the almost identical CD spectra, thereby indicating the same glyco-
packing of the glycoclusters irrespectively of the carbohydrate structures.
However, their CD spectra changed on additions of the salts in the anion-
dependent manners. In addition, our experiments indicated that the serinol
spacers and the reducing terminal carbohydrates were important structural
motifs for the anion-induced glycopacking changes. Given the design con-
cept of the metalloglycoclusters, these findings suggest that the anions
would affect the glycopacking of the cell-surface glycoclusters. However,
the salt concentrations required for the glycopacking changes of the

14
N. CHIGIRA ET AL.
metalloglycoclusters were much higher than their physiological concentra-
tions. Further investigation is thus required to reach the final conclusion of
the relationship between the salts and the glycopacking of the cell-surface
glycoclusters.
Experimental section
General experimental methods
1H and 13C NMR spectra were acquired on a JEOL AL300 spectrometer
(JEOL Resonance, Ltd., Japan) using CDCl3 or D2O as solvents. Chemical
shifts were reported in ppm (d) relative to TMS (for CDCl3) or HOD (for
D2O). Matrix-assisted laser desorption ionization time-of-flight mass
(MALDI-TOF MS) spectra were recorded on a SHIMAZU AXIMA-
Confidence (SHIMAZU, Ltd., Japan). HR ESI-TOF MS spectra were
recorded on a SYNAPT G2 HDMS (Waters, Co., USA). IR spectra were
recorded on a JASCO FT/IR-4100 Fourier transform infrared spectrometer
(JASCO Co., Ltd., Japan). UV-vis spectra were recorded on a V-630 UV-
visible spectrophotometer (JASCO Co., Ltd., Japan) equipped with a one-
drop accessory (SAH-769). CD spectra were acquired on a JASCO J-820
spectropolarimeter (JASCO Co., Ltd., Japan) using a micro sampling disk
(MSD-462). Thin layer chromatography (TLC) was carried out using
R
Whatman TLC glass plates (PartisilV K6F) pre-coated with silica gel (60 Å)
with a fluorescent indicator (k ¼ 254 nm). Silica gel 60 N (spherical, neutral,
particle size 63–210 lm) for column chromatography, dry MeOH, dry
CH2Cl2, SOCl2, MgSO4, Et3N, Na2SO4, NaCl, NaNO3, FeCl2, and CHCl3
were purchased from Kanto Chemical Co. Inc. (Japan). CDCl3, D2O,
CbzCl, 35% HCl aq, BF3OEt2, K2SO4, KNO3, and THF were purchased
from FUJIFILM Wako Pure Chemical Co. (Japan). NaBH4, KCl, and
Ca(NO3)2 were purchased from Kishida Chemical Co. Ltd. (Japan). L-
Serine was purchased from Tokyo Chemical Industry Co. Ltd. (Japan).
Toluene, hexane, ethyl acetate (AcOEt), CH2Cl2, and MeOH were pur-
chased from GODO Co. Ltd. (Japan). All these chemicals and materials
were used without further purification.
Methyl-L-serine (SerMe, 2)
To L-serine (1) (4.02 g) in MeOH (50 mL), SOCl2 (2.7 mL) was added drop-
wise and the resultant mixture was refluxed with magnetic stirring for
overnight. The mixture was then evaporated to dryness and the residue was
re-dissolved into MeOH (25 mL). Toluene (25 mL) was then added to the
mixture, and the mixture was stirred by shaking the vessel slowly with
hands at every tens minutes until the precipitate formed. The precipitate

JOURNAL OF CARBOHYDRATE CHEMISTRY
15
was retrieved by filtration and dried in vacuo to give the title compound
1
(SerMe, 2) as a white powder in 32% yield. The product showed a H
NMR spectrum that was identical to that in the literature.[19]
N-(benzyloxycarbonyl)-serine methyl ester (CbzSerMe, 3)
To SerMe (2) (0.35 g) and NaHCO3 (8.00 g) in a biphasic mixture com-
posed of water (43.5 mL) and CH2Cl2 (60.0 mL), CbzCl (6.1 mL) was added
dropwise at 0 ꢂC, and the resultant mixture was vigorously stirred for over-
night at ambient temperature. The mixture was then poured into ice water
and diluted with AcOEt. The organic layer was washed with 0.5 N HCl aq
and saturated NaCl aq repeatedly. The organic layer was dried over anhyd-
rous MgSO4, filtered, evaporated, and dried in vacuo. The residue was then
subjected to purification on silica gel (hexane/AcOEt ¼ 1/0–1/2, v/v, gradi-
ent) to give the title compound (CbzSerMe, 3) in 92% yield as sirup. The
1
product showed a H NMR spectrum that was identical to that in the
literature.[20]
General synthetic protocol to access N-Cbz-protected serine methyl ester
having O-linked carbohydrate per-acetates (4)
To each of the per-acetylated carbohydrates (XAc; X ¼ Lac, Mal, or
Man),[21] CbzSerMe (3), CH2Cl2, and BF3OEt2 were added and the mixture
was magnetically stirred under N2 at ambient temperature. After complete
consumption of CbzSerMe was confirmed based on TLC analysis (hexane/
acetone ¼ 3/2 v/v), the reaction mixture was evaporated to give brown
sirup. Although the resultant residue was subjected to purification on silica
gel using varieties of solvent systems (e.g., CHCl3/MeOH ¼ 50/1, toluene/
acetone ¼ 9/1, and hexane/acetone ¼ 3/2, v/v), the unreacted XAc could
not be removed and the title compound (CbzSerMeXAc: X ¼ bLac, bMal,
or aMan) was not obtained in its pure form. The crude product was thus
used in the following section without any further purification and detailed
characterization. The detailed reaction conditions were summarized in
Section 4.4.1–4.4.3.
N-Benzyloxycarbonyl-O-(2,3,6,2’,3’,4’,6’-hepta-O-acetyl-lactosyl)-serine methyl
ester (CbzSerMebLacAc)
LacAc: 12 g; CbzSerMe: 2.3 g; CH2Cl2: 60 ml; BF3OEt2: 12 ml; reaction time:
12 h; MALDI-TOF MS: [M þ K]þ ¼ 910.64 (calc. 910.24).

16
N. CHIGIRA ET AL.
N-Benzyloxycarbonyl-O-(2,3,6,2’,3’,4’,6’-hepta-O-acetyl-maltosyl)-serine methyl
ester (CbzSerMebMalAc)
MalAc: 10.56 g; CbzSerMe: 3.28 g; CH2Cl2: 60 mL; BF3OEt2: 12 mL; reaction
time: 2 h; MALDI-TOF MS: [M þ Na]þ ¼ 894.26 (calc. 894.39).
N-Benzyloxycarbonyl-O-(2,3,4,6-tetra-O-acetyl-mannosyl)-serine methyl ester
(CbzSerMeaManAc)
ManAc: 3.4 g; CbzSerMe: 0.9 g; CH2Cl2: 10 mL; BF3OEt2: 4.0 mL; reaction
time: 5 h; MALDI-TOF MS: [M þ H]þ ¼ 606.05 (calc. 606.18).
General synthetic protocol to access serine methyl ester having O-linked
carbohydrate per-acetates (5)
To each of the N-Cbz-protected L-serine methyl esters having O-linked
carbohydrate per-acetates (CbzSerMebLacAc, CbzSerMebMalAc, or
CbzSerMeaManAc) obtained at Section 4.4 in THF, Pd/C was added and
the resultant mixture was vigorously stirred under H2 at ambient tempera-
ture. The mixture was then filtrated through a celite pad, and the filtrate
was evaporated and dried in vacuo. The residue was subjected to purifica-
tion on silica gel (CHCl3/MeOH ¼ 40/1, v/v) to give the title compound
(5) as a white powder. The product was relatively unstable and thus, it was
used in Section 4.7 without detailed characterization. The detailed reaction
conditions were summarized in Section 4.5.1–4.5.3.
O-(2,3,6,2’,3’,4’,6’-Hepta-O-acetyl-lactosyl)-L-serine methyl ester
(SerMebLacAc, 5Lac
)
THF: 70 mL; Pd/C: 0.38 g; reaction time: overnight; yield: 22% (two steps
from CbzSerMe).
O-(2,3,6,2’,3’,4’,6’-Hepta-O-acetyl-maltosyl)-L-serine methyl ester
(SerMebMalAc, 5Mal
)
THF: 50 mL; Pd/C: 0.30 g; reaction time: overnight; yield: 19% (two steps
from CbzSerMe).
O-(2,3,4,6-Tetra-O-acetyl-mannosyl)-L-serine methyl ester (SerMeaManAc, 5Man
)
THF: 50 mL; Pd/C: 0.38 g; reaction time: overnight; yield: 28% (two steps
from CbzSerMe).
2,2-Bipyridine-5,5-dicarboxylyl chloride (bpyCOCl)
To 2,2-bipyridine-5,50-dicarboxylic acid (bpyCOOH), SOCl2 was added and
then, the resultant heterogeneous mixture was refluxed with a magnetic

JOURNAL OF CARBOHYDRATE CHEMISTRY
17
stirring until it became homogeneous (ca. 5 h). The resultant mixture was
evaporated and dried in vacuo to afford the title compound as a pale yel-
low powder. The product was then used in Section 4.7. The reaction condi-
tions were summarized in Section 4.7.1–4.7.3.
General synthetic protocol to access bpys having two carbohydrates with
their hydroxy and carboxy groups protected as the acetates and the methyl
esters, respectively (6)
To each of the L-serine methyl esters having O-linked per-acetylated carbo-
hydrates (SerMebLacAc, SerMebMalAc, or SerMeaManAc) and Et3N in
THF (40 mL), the THF solution (20 mL) containing bpyCOCl prepared
from bpyCOOH was added, and the resultant mixture was stirred at ambi-
ent temperature. The mixture was then diluted with AcOEt, and the
organic layer was washed with 0.5 N HCl aq and saturated NaHCO3 aq
repeatedly. The organic layer was dried over anhydrous Na2SO4, filtered,
and evaporated to give the title compound (6) as a pale yellow powder.
5,5-Bis-(2’-(2’’,3’’,6’’,2’’’,3’’’,4’’’,6’’’-hepta-O-acetyl-b-lactosyl)oxy-1’-(R)-methox-
ycarbonyl-ethyl-aminocarbonyl)-2,2-bipyridine (bpySerMebLacAc, 6Lac
)
bpyCOOH: 163 mg; SOCl2: 15 mL; SerMebLacAc (5Lac): 1.23 g; Et3N: 1 mL;
1
reaction time: 20 min; yield: 97% (2 steps from bpyCOOH); H NMR
(300 MHz, CDCl3) d 9.15 (s, 2 H), 8.51 (d, J 8.4 Hz, 2 H), 8.32 (d, J 8.1 Hz,
2 H), 7.2 (d, J 7.5 Hz, 2 H), 5.35 (d, J 3.3 Hz, 2 H), 5.19 (t, J 9.3 Hz, 2 H),
5.10 (dd, J 8.1, 10.2 Hz, 2 H), 4.97 (dd, J 3.3, 7.5 Hz, 2 H), 4.89 (t, J 8.7 Hz,
2 H), 4.57–4.48 (m, 6 H), 4.32 (dd, J 2.7, 10.5 Hz, 2 H), 4.19–4.02 (m, 8 H),
3.94–3.75 (m, 6 H), 3.82 (s, 6 H), 3.63–3.59 (m, 2 H), 2.15 (s, 6 H), 2.13 (s,
6 H), 2.07 (s, 6 H), 2.05 (s, 12 H), 2.00 (s, 6 H), 1.97 (s, 6 H); 13C NMR
(75 MHz, CDCl3) d 170.4, 170.3, 170.2, 170.1, 170.0, 169.7, 169.7, 169.1,
165.0, 157.7, 148.2, 136.0, 129.3, 121.4, 101.1, 100.6, 75.9, 72.9, 72.4, 71.5,
70.9, 70.7, 69.0, 68.8, 68.0, 66.5, 61.7, 60.8, 53.0, 20.9, 20.8, 20.7, 20.7, 20.5;
IR (KBr, cmꢁ1): 2012, 1752, 1659, 1586, 1430, 1371, 1224, 1046; MALDI-
TOF MS: [M þ H]þ ¼ 1683.27 (calc. 1683.50).
5,5-Bis-(2’-(2’’,3’’,6’’,2’’’,3’’’,4’’’,6’’’-hepta-O-acetyl-b-maltosyl)oxy-1’-(R)-
methoxycarbonyl-ethyl-aminocarbonyl)-2,2-bipyridine (bpySerMebMalAc, 6Mal
)
bpyCOOH: 247 mg; SOCl2: 20 mL; SerMebMalAc (5Mal): 1.85 g; Et3N: 2 mL;
1
reaction time: 20 min; yield: 85% (two steps from bpyCOOH); H NMR
(300 MHz, CDCl3) d 9.15 (d, J 3.5 Hz, 2 H), 8.61 (d, J 8.4 Hz, 2 H), 8.31
(dd, J 2.4, 8.4 Hz, 2 H), 7.12 (d, J 7.5 Hz, 2 H), 5.40 (d, J 3.9 Hz, 2 H), 5.35
(t, J 9.9 Hz, 2 H), 5.25 (t, J 9.0 Hz, 2 H), 5.06 (t, J 9.9 Hz, 2 H), 4.98 (quint. J

18
N. CHIGIRA ET AL.
3.3 Hz, 2 H), 4.85 (dd, J 4.2, 10.2 Hz, 2 H), 4.81 (dd, J 8.1, 9.6 Hz, 2 H), 4.61
(d, J 7.5 Hz, 2 H), 4.52 (dd, J 2.1, 12.0 Hz, 2 H), 4.35 (dd, J 3.0, 10.8 Hz,
2 H), 4.28–4.19 (m, 4 H), 4.07–3.93 (m, 8 H), 3.83 (s, 6 H), 3.69–3.67 (m,
2 H), 2.15 (s, 6 H), 2.10 (s, 6 H), 2.04 (s, 6 H), 2.03 (s, 6 H), 2.01 (s, 6 H),
2.00 (s, 6 H), 1.98 (s, 6 H); 13C NMR (75 MHz, CDCl3) d 170.6, 170.5,
170.1, 170.0, 169.7, 169.4, 100.3, 95.7, 74.9, 72.4, 72.0, 70.0, 69.2, 68.9, 68.5,
67.9, 62.5, 61.4, 53.0, 20.9, 20.8, 20.7, 20.7, 20.6; IR (KBr, cmꢁ1): 2959,
1753, 1668, 1595, 1530, 1463, 1438, 1370, 1234, 1042; MALDI-TOF MS:
[M þ H]þ ¼ 1682.92 (calc. 1683.50).
5,5-Bis-(2’-(2’’,3’’,4’’,6’’-tetra-O-acetyl-b-mannosyl)oxy-1’-(R)-methoxycarbonyl-
ethyl-aminocarbonyl)-2,2-bipyridine (bpySerMeaManAc, 6Man
)
bpyCOOH: 107 mg; SOCl2: 7 mL; SerMeaManAc (5Man): 0.4 g; Et3N:
1
0.56 mL; reaction time: 20 min; yield: 80% (two steps from bpyCOOH); H
NMR (300 MHz, CDCl3) d 9.20 (d, J 1.8 Hz, 2 H), 8.61 (d, J 8.1 Hz, 2 H),
8.33 (dd, J 2.4, 8.1 Hz, 2 H), 7.40 (d, J 7.5 Hz, 2 H), 5.31 (dd, J 3.0, 8.1 Hz,
2 H), 5.29–5.19 (m, 4 H), 5.07–5.05 (m, 2 H), 4.86 (d, J 1.8 Hz, 2 H),
4.26–4.08 (m, 6 H), 3.98–3.94 (m, 2 H), 3.88 (s, 6 H); 13C NMR (75 MHz,
CDCl3) d 170.4, 170.2, 169.7, 169.7, 169.6, 169.4, 165.0, 157.1, 148.1, 135.8,
129.1, 120.8, 98.2, 69.0, 68.4, 65.6, 62.1, 52.8, 20.6, 20.4; IR (KBr, cmꢁ1):
2891, 1750, 1660, 1590, 1368, 1229, 1045; MALDI-TOF MS: [M þ H]þ
¼
1107.43 (calc. 1107.33).
General synthetic protocol to access bipyridine derivatives having two
carbohydrates (7)
To each of the bpys having two per-acetylated carbohydrates
(bpySerMebLacAc, bpySerMebMalAc, or bpySerMeaManAc) in a mixture
of H2O and THF, NaBH4 was added and the resultant mixture was stirred
at ambient temperature for the given periods of time. After confirming that
the reaction was completed, the mixture was evaporated and dried in
vacuo. The residue was purified through a glass column packed with HP-
20 (Mitsubishi Chemical Co.) using water with increasing concentrations of
MeOH as eluent (MeOH contents (v/v) ¼ 0, 10, 20, 30, 40, 50, 60, 70, 80,
90, and 100%). The fractions eluted out from the column with aqueous
MeOH ([MeOH] ¼ 10 and 20%) were combined, evaporated, and dried in
vacuo to give the title compound (7) as a white powder.
5,5-Bis-(2’-b-lactosyloxy-1’-(R)-hydroxymethyl-ethylaminocarbonyl)-2,2-bipyri-
dine (bpySerbLac, 7Lac
)
bpySerMebLacAc (6Lac): 280 mg; NaBH4: 150 mg; H2O: 5 mL; THF: 10 mL;
1
reaction time: 4 days; yield: 97%; H NMR (300 MHz, D2O): 9.04 (br, 2 H),

JOURNAL OF CARBOHYDRATE CHEMISTRY
19
8.53 (br, 2 H), 8.41 (br, 2 H), 4.38 (d, J 7.5 Hz, 2 H), 4.30 (t, J 5.4 Hz, 2 H),
4.27 (d, J 7.5 Hz, 2 H), 3.97 (dd, J 6.6, 10.5 Hz, 2 H), 3.83–3.74 (m, 6 H),
3.69–3.53 (m, 12 H), 3.50–3.47 (m, 8 H), 3.37 (t, J 8.7 Hz, 2 H), 3.19 (t, J
7.8 Hz, 2 H); 13C NMR (75 MHz, CDCl3) d 169.0, 157.5, 148.7, 137.9, 131.0,
123.0, 103.5, 103.0, 78.9, 75.9, 75.4, 74.8, 73.3, 73.0, 71.5, 69.3, 69.1, 61.6,
61.1, 60.6, 52.4; IR (KBr): 3297, 2952, 2882, 1645, 1599, 1556, 1464, 1344,
1307, 1043 cmꢁ1; HR ESI-TOF MS: [M þ Na]þ
¼
1061.4534
(calc. 1039.3545).
5,5-Bis-(2’-b-maltosyloxy-1’-(R)-hydroxymethyl-ethylaminocarbonyl)-2,2-bipyri-
dine (bpySerMal, 7Mal
)
bpySerMebMalAc (6Mal): 307 mg; NaBH4: 257 mg; H2O: 7 mL; THF: 10 mL;
1
reaction time: 3 days; yield: 76%; H NMR (300 MHz, D2O): 8.81 (s, 2 H),
8.15 (d, J 8.4 Hz, 2 H), 8.02 (d, J 8.4 Hz, 2 H), 5.21 (d, J 3.9 Hz, 2 H), 4.35
(d, J 8.1 Hz, 2 H), 4.26 (t, J 5.7 Hz, 2 H), 3.94 (dd, J 6.6, 10.5 Hz, 2 H),
3.76–3.36 (m, 26 H), 3.25–3.14 (m, 4 H); 13C NMR (75 MHz, D2O) d 169.3,
157.7, 149.0, 138.2, 131.3, 123.3, 103.4, 100.5, 77.6, 77.0, 75.5, 73.9, 73.7,
73.6, 72.5, 70.2, 69.7, 61.6, 61.5, 61.4, 52.7; IR (KBr) 3335, 2924, 1647,
1596, 1542, 1466, 1362, 1033 cmꢁ1; HR ESI-TOF MS: [M þ Na]þ
¼
1061.4054 (calc. 1061.3545).
5,5-Bis-(2’-b-mannoloxy-1’-(R)-hydroxymethyl-ethylaminocarbonyl)-2,2-bipyri-
dine (bpySerMan, 7Man
)
bpySerMeManAc (6Man): 340 mg; NaBH4: 406 mg; H2O: 8 mL; THF: 10 mL;
1
reaction time: 1 day; yield: 69%; H NMR (300 MHz, D2O): 8.78 (br s, 2 H),
8.12 (d, J 8.4 Hz, 2 H), 7.99 (d, J 8.1 Hz, 2 H), 4.75 (br s, 2 H), 4.24 (t, J
5.4 Hz, 2 H), 3.84–3.80 (m, 4 H), 3.72–3.48 (m, 16 H); 13C NMR (75 MHz,
D2O) d 168.7, 157.1, 148.6, 137.7, 130.8, 122.8, 100.8, 73.5, 71.0, 70.5, 67.2,
66.8, 61.4, 61.1, 52.5; IR (KBr) 3324, 2932, 1647, 1588, 1545, 1466, 1348,
1051 cmꢁ1; HR ESI-TOF MS: [M þ H]þ ¼ 715.1749 (calc. 715.2674).
UV-vis and CD spectroscopy titration experiments
We prepared the aqueous metalloglycocluster solutions and the aqueous salt
solutions as the stock solutions. In the case of the former, we mixed the
glycobpys and FeCl2 in water to prepare the aqueous metalloglycocluster
solutions ([bpySerbLac], [bpySerbMal], or [bpySeraMan]¼ 24.4 mM,
[FeCl2] ¼ 73.3 mM). We then mixed the metalloglycocluster solutions (3 lL),
the aqueous salt solutions (0–27 lL), and pure water (27–0 lL) to prepare the
sample solutions (30 lL) with varying concentrations of the salts. After the
incubation of these sample solutions for 1 day, their UV-vis and CD spectra

20
N. CHIGIRA ET AL.
were measured using a V-630 UV-visible spectrophotometer and a JASCO J-
820 spectropolarimeter equipped with SAH-769 and MSD-462, respectively.
MD Simulations
Before the MD simulation of the three-components moꢁlecular assembly
composed of K-[Fe(bpySerbLac)3]2þ, Naþ, and NO3 , we built a
K-[Fe(bpySerbLac)3]2þ with its stable conformation following a molecular
mechanics (MM) calculation. Briefly, we constructed a K-[Febpy3]2þ with-
out the b-lactosyl serinol unit and then, conducted a MM calculation to
obtain the stable K-[Febpy3]2þ. We then fixed the spatial coordinates of all
atoms of the stable K-[Febpy3]2þ. This spatial fixing was essential to suc-
cessfully complete the MD simulation on the three-components molecular
assembly, since CHARMM force field was not suitable for the molecules
having Fe-N bonds. We then attached the b-lactosyl serinol units onto all
C5 positions of pyridine units of the stable K-[Febpy3]2þ and thereby con-
-
structed K-[Fe(bpySerbLac)3]2þ. We then added Naþ and NO3 ions nearby
K-[Fe(bpySerbLac)3]2þ to construct the three-components molecular
ꢁ
assembly composed of K-[Fe(bpySerbLac)3]2þ, Naþ, and NO3 . Before
starting the MD simulation, we applied the distance restrictions (the dis-
-
tance between Fe2þ and Naþ, and that between Fe2þ and NO3 were fixed
to be shorter than 15 Å) to this three-components molecular assembly so
-
that the ions (Naþ and NO3 ) did not leave far away from
K-[Fe(bpySerbLac)3]2þ. The subsequent MD simulation (10 ps equilibra-
tion, 90 ps dynamics, 300 K, e ¼ 80, CHARMM) gave the most stable con-
formation of the three-components molecular assembly.
Data related to this article containing structures of LeX, GM3, Gg3, and
LacCer (Chart S1), a colored version of Fig. 2 (Fig. S1), a colored version
1
of Fig. 5 (Fig. S2), and H/13C NMR spectra of the newly synthesized com-
pounds are available in Supplementary material.
Funding
This work was supported by Grant-in-Aid for Young Scientists (B) [No. 25810105] from
Japan Society for the Promotion of Science (JSPS) and Special Research Fund of
Toyo University.
References
[1] Hakomori, S. Structure, Organization, and Function of Glycosphingolipids in
Membrane. Curr. Opin. Hematol. 2003, 10, 16–24.
[2] (a) Hakomori, S. Carbohydrate-Carbohydrate Interaction as an Initial Step in Cell
Recognition. Pure Appl. Chem 1991, 63, 473–482. (b) Yu, S.; Kojima, N.; Hakomori,
S.; Kudo, S.; Inoue, S.; Inoue, Y. Binding of Rainbow Trout Sperm to Egg Is

JOURNAL OF CARBOHYDRATE CHEMISTRY
21
Mediated by Strong Carbohydrate-to-Carbohydrate Interaction between (KDN)GM3
(Deaminated Neuraminyl Ganglioside) and Gg3-like Epitope. PNAS. 2002, 99,
2854–2859.
[3] Handa, K.; Hakomori, S. Changes of Glycoconjugate Expression Profiles
during Early Development. Glycoconj. J. 2017, 34, 693–699.
[4] (a) Iwabuchi, K.; Yamamura, S.; Prinetti, A.; Handa, K.; Hakomori, S. GM3-
Enriched Microdomain Involved in Cell Adhesion and Signal Transduction
Through Carbohydrate-Carbohydrate Interaction in Mouse Melanoma B16 Cells. J.
Biol. Chem. 1998, 273, 9130–9138. (b) Handa, K.; Hakomori, S. Carbohydrate to
Carbohydrate Interaction in Development Process and Cancer Progression.
Glycoconj. J. 2012, 29, 627–637.
ꢀ
ꢀ
[5] de la Fuente, J. M.; Eaton, P.; Barrientos, A. G.; Menendez, M.; Penades, S.
Thermodynamic Evidence for Ca2þ-Mediated Self-Aggregation of Lewis X Gold
Glyconanoparticles. A Model for Cell Adhesion via Carbohydrate ꢁ Carbohydrate
Interaction. J. Am. Chem. Soc. 2005, 127, 6192–6197.
[6] Reynolds, A. J.; Haines, A. H.; Russell, D. A. Gold Glyconanoparticles for Mimics
and Measurement of Metal Ion-Mediated Carbohydrate-Carbohydrate Interactions.
Langmuir 2006, 22, 1156–1163.
[7] (a) Matsuura, K.; Kobayashi, K. Analysis of GM3-Gg3 Interaction Using Clustered
Gycoconjugate Models Constructed from Glycolipid Monolayers and Artificial
Glycoconjugate Polymers. Glycoconj. J. 2004, 21, 139–148. (b) Matsuura, K.; Oda,
R.; Kitakouji, H.; Kiso, M.; Kitajima, K.; Kobayashi, K. Surface Plasmon Resonance
Study of Carbohydrate-Carbohydrate Interaction Between Various Gangliosides and
Gg3-Carrying Polystyrene. Biomacromolecules 2004, 5, 937–941. (c) Chen, Q.; Cui,
Y.; Zhang, T. L.; Cao, J.; Han, B. H. Fluorescent Conjugated Polyfluorene with
Pendant Lactopyranosyl Ligands for Studies of Ca2þ-Mediated Carbohydrate-
Carbohydrate Interaction. Biomacromolecules 2010, 11, 13–19.
[8] Seah, N.; Santacroce, P. V.; Basu, A. Probing the LactoseꢃGM3
Carbohydrate#Carbohydrate Interaction with Glycodendrimers. Org. Lett. 2009, 11,
559–562.
[9] Bavireddi, H.; Bharate, P.; Kikkeri, R. Probing Carbohydrate–Carbohydrate
Interactions by Photoswitchable Supramolecular Glycoclusters. Chem. Commun.
2013, 49, 3988–3990.
[10] (a) Pincet, F.; Le Bouar, T.; Zhang, Y.; Esnault, J.; Mallet, J.-M.; Perez, E.; Sina€y, P.
Ultraweak Sugar-Sugar Interactions for Transient Cell Adhesion. Biophys. J. 2001,
€€
80, 1354–1358. (b) Kunze, A.; Bally, M.; Hook, F.; Larson, G. Equilibrium-
Fluctuation-Analysis of Single Liposome Binding Events Reveals How Cholesterol
and Ca2þ Modulate Glycosphingolipid Trans-Interactions. Sci. Rep 2013, 3,
ꢀ
1452–1459. (c) Lorenz, B.; de Cienfuegos, L. A.; Oelkers, M.; Kriemen, E.; Brand, C.;
Stephan, M.; Sunnick, E.; Yu€ksel, D.; Kalsani, V.; Kumar, K.; et al. Model System for
Cell Adhesion Mediated by Weak Carbohydrate ꢁ Carbohydrate Interactions. J. Am.
Chem. Soc. 2012, 134, 3326–3329.
[11] (a) Lai, C.-H.; Hu€tter, J.; Hsu, C.-W.; Tanaka, H.; Varela-Aramburu, S.; De Cola, L.;
Lepenies, B.; Seeberger, P. H. Analysis of Carbohydrate-Carbohydrate Interactions
Using Sugar-Functionalized Silicon Nanoparticles for Cell Imaging. Nano Lett. 2016,
16, 807–811. (b) Zhao, J.; Liu, Y.; Park, H.-J.; Boggs, J. M.; Basu, A. Carbohydrate-
Coated Fluorescent Silica Nanoparticles as Probes for the Galactose/
3 ꢁ Sulfogalactose Carbohydrate ꢁ Carbohydrate Interaction Using Model Systems
and Cellular Binding Studies. Bioconjugate Chem. 2012, 23, 1166–1173.

22
N. CHIGIRA ET AL.
[12] Chen, L.; Sun, P.; Chen, G. Fluorous-Based Carbohydrate Quartz Crystal
Microbalance. Carbohydr. Res. 2015, 405, 66–69.
[13] Hasegawa, T.; Matsuura, K.; Ariga, K.; Kobayashi, K. Multilayer Adsorption and
Molecular Organization of Rigid Cylindrical Glycoconjugate Poly(Phenylisocyanide)
on Hydrophilic Surfaces. Macromolecules 2000, 33, 2772–2775.
[14] Gege, C.; Geyer, A.; Schmidt, R. R. Carbohydrate_Carbohydrate Recognition
Between Lewis X Blood Group Antigens, Mediated by Calcium Ions. Eur. J. Org.
Chem. 2002, 2002, 2475–2485.
[15] Hasegawa, T.; Sasaki, T. Glyco-Helix: Parallel Lactose-Lactose Interactions Stabilize
an a-Helical Structure of Multi-Glycosylated Peptide. Chem. Commun. 2003,
978–979.
[16] Otsuka, A.; Sakurai, K.; Hasegawa, T. Ferrocenes with Two Carbohydrate
Appendages at the Upper and Lower Rings Are Useful for Investigating
Carbohydrate–Carbohydrate Interactions. Chem. Commun. 2009, 5442–5444.
[17] Chigira, N.; Dai, F.; Nonaka, Y.; Sato, K.; Amano, Y.; Sekiguchi, M.; Inokuchi, M.;
Hagio, M.; Hasegawa, T. Glycosylated Tris-Bipyridine Ferrous Complexes for
Probing a Mechanism behind Carbohydrate-Carbohydrate Interactions: Spatial
Carbohydrate Packing of Glycoclusters Changes on Additions of Salts in
Carbohydrate- and Anion-Dependent Manners. Tetrahedron 2018, 74, 5898–5907.
[18] Milder, S. J.; Gold, J. S.; Kliger, D. S. Circular Dichroism of a Subnanosecond State:
(.Delta.)-Fe(Bpy)32þ. J. Am. Chem. Soc. 1986, 108, 8295–8296.
[19] Foss, F. W.; Snyder, A. H.; Davis, M. D.; Rouse, M.; Okusa, M. D.; Lynch, K. R.;
Macdonald, T. L. Synthesis and Biological Evaluation of c-Aminophosphonates as
Potent, Subtype-Selective Sphingosine 1-Phosphate Receptor Agonists and
Antagonists. Bioorg. Med. Chem 2007, 15, 663–677.
[20] Hernandez, J. N.; Martin, V. S. First Practical Protection of a-Amino Acids as N,N-
Benzyloxycarbamoyl Derivatives. J. Org. Chem. 2004, 69, 3590–3592.
[21] Arya, P.; Barkley, A.; Randell, K. D. Automated High-Throughput Synthesis of
Artificial Glycopeptides. Small-Molecule Probes for Chemical Glycobiology. J. Comb.
Chem. 2002, 3, 193–198.