A Self-Assembled Spherical Complex Displaying a Gangliosidic Glycan Cluster Capable of Interacting with Amyloidogenic Proteins

Article abstract of DOI:10.1002/anie.201501981

Physiological and pathological functions of glycans are promoted through their clustering effects as exemplified by a series of gangliosides, sialylated glycosphingolipids, which serve as acceptors for bacterial toxins and viruses. Furthermore, gangliosid

Full text of DOI:10.1002/anie.201501981

Angewandte
Chemie
International Edition: DOI: 10.1002/anie.201501981
German Edition: DOI: 10.1002/ange.201501981
Self-Assembly
A Self-Assembled Spherical Complex Displaying a Gangliosidic
Glycan Cluster Capable of Interacting with Amyloidogenic Proteins**
Sota Sato,* Yutaka Yoshimasa, Daishi Fujita, Maho Yagi-Utsumi, Takumi Yamaguchi,
Koichi Kato,* and Makoto Fujita*
Abstract: Physiological and pathological functions of glycans
are promoted through their clustering effects as exemplified by
a series of gangliosides, sialylated glycosphingolipids, which
serve as acceptors for bacterial toxins and viruses. Further-
more, ganglioside GM1 clusters on neuronal cell membranes
specifically interact with amyloidogenic proteins, triggering
their conformational transitions and leading to neurodegener-
ation. Here we develop a self-assembled spherical complex that
displays a cluster of the GM1 pentasaccharide, and successfully
demonstrate its ability to interact with amyloid b and a-
synuclein. Due to the lack of hydrophobic lipid moieties, which
would stably trap these cohesive proteins or give rise to toxic
aggregates, this artificial cluster enabled NMR spectroscopic
characterization of the early encounter stage of protein
interactions with its outer carbohydrate moieties, which were
not observable with previous glycan clusters.
a receptor for cholera toxin, polyomavirus, and autoanti-
bodies associated with the Guillain–BarrØ syndrome.[3–5]
These interactions are characterized by their multivalent
binding to GM1 clusters. Therefore, artificial clusters of the
GM1 pentasaccharide moiety could be designed and created
for detecting, capturing, and adsorbing toxic entities.
Recently, accumulating evidence has indicated that GM1
clusters on neuronal cell membranes specifically interact with
amyloid b-protein (Ab), triggering its conformational tran-
sition into toxic aggregates responsible for the onset and
development of Alzheimer’s disease.[6] Pathological interac-
tions of ganglioside clusters have also been proposed for other
amyloidogenic proteins, including a-synuclein and prion
protein, which are associated with Parkinson’s disease and
Creutzfeldt-Jakob syndrome, respectively.[7–9] The ganglioside
binding of these proteins is not described as lectin-like
stoichiometric binding, although they exhibit certain specific-
ities for the outer carbohydrate moieties. Rather, ganglioside
clusters offer catalytic environments for conformational
alteration of the proteins coupled with their transient, multi-
step binding.[2,6b] To characterize the underlying mechanisms
of these pathological molecular processes, artificial ganglio-
side clusters have been prepared based on micelles and
bicelles, primarily for solution-phase NMR spectroscopic
analyses.[10,11] These studies indicated that the size and
curvature of the ganglioside clusters are determining factors
in the interactions and subsequent aggregate formation,
highlighting the need for development of gangliosidic carbo-
hydrate assemblies that are systematically controllable in
terms of these factors. These previous studies also revealed
that the hydrophobic environment provided by the lipid
hydrocarbon chains is responsible for the formation and
accommodation of a-helical segments of Ab and a-synucle-
in.[10a,c,12] Furthermore, the hydrophilic/hydrophobic interface
facilitates amyloid formation of these pathogenic proteins.
These findings suggest that creation of oligosaccharide
clusters lacking the ceramide moiety will open up new
possibilities for characterizing the early stages of interactions
of ganglioside clusters with amyloidogenic proteins.
O
ligosaccharides play pivotal roles in physiological and
pathological contexts. Oligosaccharide functions are often
promoted when they form clusters on cellular membranes; as
exemplified by gangliosides, a group of glycosphingolipids
containing sialic acid, which mediate cell adhesion and signal
transduction and provide acceptor sites for microbial toxins
and viruses.[1] Hence, ganglioside clusters are potential
medicinal and therapeutic targets.[2] Particular attention has
been paid to ganglioside GM1, which is known to serve as
[*] Dr. S. Sato,[+] Y. Yoshimasa, Dr. D. Fujita, Prof. Dr. M. Fujita
Department of Applied Chemistry, School of Engineering
The University of Tokyo
7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656 (Japan)
E-mail: satosota@m.tohoku.ac.jp
mfujita@appchem.t.u-tokyo.ac.jp
Dr. M. Yagi-Utsumi, Dr. T. Yamaguchi, Prof. Dr. K. Kato
Department of Life and Coordination-Complex Molecular Science
Institute for Molecular Science and Department of Bioorganization
Research, Okazaki Institute for Integrative Bioscience
National Institutes of Natural Sciences
5-1 Higashiyama, Myodaiji, Okazaki, Aichi 444-8787 (Japan)
E-mail: kkato@phar.nagoya-cu.ac.jp
[+] Current address: WPI-AIMR, Tohoku University (Japan)
Here, we created well-defined GM1 clusters on the
periphery of an M12L24 spherical complex,[13] a large scaffold
assembled from 12 Pd2+ ions (M) and 24 bent ligands (L) that
define a stable discrete interface.[13c] An NMR spectroscopic
study revealed that the initial encounter complex formation
between the GM1 pentasaccharide cluster and Ab occurs at
the N-terminal segment of Ab. This approach was also
successfully applied to the observation of the interaction
between the GM1 pentasaccharide cluster and a-synuclein.
[**] This research was supported by JST, ACCEL, and KAKENHI for
Young Scientists (A) (24685010), for Scientific Research on
Innovative Areas (25102007, 25102001, and 25102008), for Young
Scientists (B) (15K21680), and for Challenging Exploratory Research
(26560451), the Research Funding for Longevity Sciences (25-19)
from National Center for Geriatrics and Gerontology, the Okazaki
ORION project, and the Nanotechnology Network Project and
Nanotechnology Platform Program at IMS.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201501981.
Angew. Chem. Int. Ed. 2015, 54, 8435 –8439
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
8435

.
Angewandte
Communications
GM1 pentasaccharide clusters were constructed by self-
assembly of Pd2+ ions with bidentate ligand 1, which bear the
sugar moiety of ganglioside GM1 at the apex of the ligand.
Since the commercially available GM1 source is expensive
and limited, an efficient and short synthetic route to introduce
the GM1 sugar onto the ligand was explored (Figure 1). The
reductive amination reaction was performed by addition of
NaBH3CN (1.0 equiv) to a solution of lyso-GM1 2 (1–5 mg
scale) and bidentate ligand 3 (5.0 equiv), which bears an
aldehyde group at the apex, in hexafluoro-2-propanol, to
afford GM1-anchored ligand 4. The lipid chain was removed
by treatment of crude product 4 in a mixed solvent system of
MeOH/CH2Cl2 (1:1) with ozone (4.0 equiv), followed by
addition of NaBH4 (10 equiv), affording ligand 1 in 67% yield
from lyso-GM1 2 (Figure 1B). Structural determination of
1 was performed by spectroscopic analyses and high-resolu-
tion mass spectrometry.
GM1 sugar cluster 5, which accumulates just 24 GM1
sugar moieties on an M12L24 spherical complex, was self-
assembled from the ligand-bearing GM1 sugar 1 (1.91 mmol)
in DMSO (89.3 mL). This was treated with Ca(NO3)2
(46.7 mmol) in D2O (6.34 mL) at 508C for 10 min, followed
by addition of Pd(NO3)2 (1.15 mmol) in DMSO at 508C for
12 h (Figure 1A). The initial addition of Ca(NO3)2 was
essential for forming the spherical complex, presumably
because the weak interaction between the Ca2+ ions and
sugar moieties prevented trapping of the Pd2+ ions at the
sugar moieties. Dialysis of the solution of sphere 5 in DMSO
against H2O removed the DMSO and Ca(NO3)2, giving
a solution of sphere 5 in H2O. The 1H NMR spectra of ligand 4
and sphere 5 show a downfield shift for the 1H NMR signals at
the pyridyl a-positions, which indicates the formation of
coordination bonds between the pyridyl groups and Pd2+ ions
1
(Figure 2). The H signals of sphere 5 are broad due to the
large size of the structure. The number of signals did not
change between ligand 4 and sphere 5, suggesting the
formation of a highly symmetric structure. The diffusion
coefficient (D) of sphere 5 was determined by DOSY
measurements; the obtained value, D = 6.3 ” 10À11 m2 sÀ1, is
consistent with a diameter of 5.7 nm, which corresponds to
the value obtained for the structure simulated through
molecular mechanics (MM) calculations.[14]
The possible interaction between sugar cluster 5, bearing
an artificial GM1 cluster without hydrophobic ceramide
moieties, and Ab was examined to characterize their encoun-
tering state. A solution of uniformly 15N-labeled Ab-40 with
40 amino acid residues (240 mL, 0.11 mm, 25 nmol) in aqueous
phosphate buffer (30 mm) was added to a solution of sphere 5
(360 mL, 0.070 mm, 25 nmol) in aqueous phosphate buffer
(30 mm, pH 6.8) at 278C. Under these conditions, 1.0 equiv of
sphere 5 bearing 24 pentasaccharides was used for every
Figure 1. A) Synthesis of sugar cluster 5 from Pd2+ ions and bidentate
ligand 1 bearing GM1 sugar. The molecular structure of sphere 5 was
optimized by MM calculations (Pd: yellow, C: light blue, N: blue, O:
red, H: gray). B) Synthesis of ligand 1 by the ligand-tethering reaction i
[conditions: 1) 2,3-bis(4-pyridyl)benzaldehyde 3, HFIP, 508C, 1 h;
2) NaBH3CN, HFIP/CH3OH (15:1), 508C, 2 h] and the lipid chain
cleavage reaction ii [conditions: 1) O3, CH2Cl2/CH3OH (1:1), À808C,
3–6 h; 2) NaBH4, CH2Cl2/CH3OH (1:1), À808C to RT, 2 h]. The overall
yield from GM1 2 was 67%.
1
1.0 equiv of Ab-40. The solution was examined by H–15N
HSQC NMR spectroscopy to observe the amide signals of
1
À
Ab-40. When we compared the signal intensities in the H
15N HSQC spectra for Ab-40 with and without sugar cluster 5,
all signals were attenuated. Intriguingly, the signals originat-
ing from the N-terminal segment became undetectable due to
extreme line broadening (Figure 3A, dark blue). Although
the NH signal intensities were recovered upon reduction of
the amount of 5 used (2.5 nmol, 0.10 equiv), the N-terminal
signals of Ab-40 remained significantly attenuated (Fig-
8436 www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 8435 –8439

Angewandte
Chemie
Figure 2. 1H NMR spectra (600 MHz, 278C, D2O) of A) ligand 1 and
B) complex 5.
ure 3A, light blue). Effective interactions were observed in
the range of 0.10 to 1.0 equiv of sugar cluster 5, with signal
decays for all amino acid residues. The specific attenuation of
the N-terminal signals was presumably due to selective
involvement of the N-terminal segment of Ab-40 in the
interaction with the GM1 pentasaccharide cluster on sphere 5.
This is characterized as a dynamic process in the intermediate
regime, whereas the C-terminal segment is scarcely involved
in the interaction.
Figure 3. A) Relative intensities of backbone NH signals of 15N-labeled
For Ab-28,[15] which has 28 amino acid residues and lacks
the C-terminal residues 29–40, 1.0 equiv of sugar cluster 5 was
added in the same manner (Figure 3B). The truncated Ab-28
showed a larger magnitude of signal decay than Ab-40 over all
amino acid residues, which again underscored the selective
involvement of the N-terminal segment, confirming the
results obtained for Ab-40. This was presumably due to the
decrease in size of the noninteracting part of the protein. As
Ab-40 protein plotted against the amino acid sequence. The applied
amount of sugar cluster 5 was either 1.0 equiv (dark blue) or
0.10 equiv (light blue). The signals were observed from 1H–15N
HSQC NMR spectra (600 MHz, 278C, 30 mm aqueous phosphate
buffer, pH 6.8, H2O/D2O (90:10)). Intensity ratio was obtained by
dividing each peak height value of Ab-40 protein in the presence of
sugar cluster 5 by that of this protein alone. Asterisks indicate amino
acid residues that did not exhibit observable peaks in the spectrum
due to severe broadening. B) Relative intensities of backbone NH
signals of 15N-labeled Ab-28 protein plotted against the amino acid
sequence. The applied amount of sugar cluster 5 was 1.0 equiv. The
signals were observed from 1H–15N HSQC NMR spectra (600 MHz,
278C, 30 mm aqueous phosphate buffer, pH 6.8, H2O/D2O (90:10)).
Intensity ratio was obtained by dividing each peak height value of Ab-
28 protein in the presence of sugar cluster 5 by that of this protein
alone. Asterisks indicate amino acid residues that did not exhibit
observable peaks in the spectrum due to severe broadening.
1
control experiments, H–15N HSQC NMR spectra of Ab-40
(25 nmol) with monomeric GM1 pentasaccharide (25 nmol)
or Pd(NO3)2 (2.5 nmol) were measured. In both cases, signal
decay was negligible, demonstrating that cluster formation of
the GM1 pentasaccharide is a prerequisite for its specific
interaction with Ab-40 (see Figures S9 and S10 in the
Supporting Information). Inspection of all data suggests that
Ab could form an encounter complex specifically with the
sugar clusters of GM1 gangliosides through its N-terminal
region. This observation is in marked contrast to the
topological mode of Ab-40 lying on the hydrophilic/hydro-
phobic interface of GM1 micelles, in which the two a-helices
(His14-Val24 and Ile31-Val36) and the C-terminal dipeptide
segment (Val39-Val40) are in contact with the hydrophobic
interior of the ganglioside clusters.[10a]
Brain deposition of hereditary Ab variants with single
amino acid substitutions crucially depends on the presence of
favorable gangliosides.[16] For example, Dutch-, Italian-, and
Iowa-type Ab variants require GM3 ganglioside for their
assembly, whereas GD3 is upregulated at the site of Flemish-
type Ab deposition. These data suggest that the outer
carbohydrate structures displayed on neural cells are critical
determining factors for specificities of the interactions
between the hereditary Ab variants and the various gangli-
osides. Our self-assembling cluster will be applicable for
displaying other gangliosidic oligosaccharides, opening the
door to studies on the structural mechanisms underlying such
pathological processes.
To examine the applicability of artificial gangliosidic sugar
cluster 5, without a hydrophobic lipid chain for identifying the
carbohydrate interacting sites in other amyloidogenic pro-
teins, a-synuclein was subjected to NMR spectroscopic
analysis under the same conditions using 1.0 equiv of sugar
cluster 5. Because a-synuclein is a large, intrinsically unstruc-
tured protein composed of 140 amino acid residues, we
performed ultrahigh-resolution NMR spectroscopic experi-
1
ments with a 920 MHz spectrometer.[17] Analysis of the H–
15N HSQC NMR spectra revealed reduction in peak intensity
over all amino acid residues of a-synuclein, with marked
Angew. Chem. Int. Ed. 2015, 54, 8435 –8439
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8437

.
Angewandte
Communications
Figure 4. Relative intensities of backbone NH signals of 15N-labeled a-synuclein protein plotted against the amino acid sequence. The applied
amount of sugar cluster 5 was 1.0 equiv. The signals were obtained from 1H–15N HSQC NMR spectra (920 MHz, 108C, 15 mm aqueous
phosphate buffer, pH 6.8, H2O/D2O (90:10)). Intensity ratio was obtained by dividing each peak height value of a-synuclein in the presence of
sugar cluster 5 by that of this protein alone. Asterisks indicate proline residues or amino acid residues that did not exhibit observable peaks in
the spectrum due to severe broadening.
attenuation in its N- and C-terminal segments, suggesting
these regions participate in the interaction with the GM1
pentasaccharide cluster (Figure 4). Our previous NMR spec-
troscopic analysis using GM1-embedding small bicelles
suggested that the N-terminal segment of a-synuclein is
involved, as the most “ganglioside-philic” region, in the
membrane-landing process on ganglioside clusters.[11c] This is
most likely followed by membrane-anchoring a-helix forma-
tion, as observed with anionic vesicles.[12] Artificial sugar
cluster 5 presumably exhibits a large continuous surface
covered by GM1 pentasaccharides, as compared with the
GM1-embedding small bicelles, and lacks the hydrophobic
interior required for accommodation of the a-helical seg-
ments of a-synuclein. This enables us to identify the
previously unidentified secondary ganglioside binding site,
that is, the C-terminal segment, thereby offering a new tool to
provide us with information on the interaction mode com-
plementarily with micelle- and bicelle-based systems.
In summary, a well-defined sugar cluster was constructed
through the modification of a self-assembled M12L24 sphere
with the sugar moiety of natural bioactive ganglioside GM1.
The spherical complex showed recognition abilities toward
amyloidogenic proteins Ab and a-synuclein at their specific
sites, which were promoted by oligosaccharide clustering
without the use of hydrophobic lipid chains. These chains
would otherwise stably trap the cohesive proteins, precluding
the observation of early encounter interactions or give rise to
toxic aggregates. Further variation of ganglioside clusters with
different size and curvature would be accessible for system-
atic studies using a series of self-assembled spherical com-
plexes as the scaffolds.[18] Our novel artificial ganglioside–
glycan cluster will be useful not only as a novel analytical tool,
but also as a prototype of self-assembled sorbent targeting the
ganglioside-interacting hazardous substances used in thera-
peutic and diagnostic applications, offering new possibilities
for developing bioprobes, biosensors, and drug delivery
systems.
Keywords: amyloid b · NMR spectroscopy · self-assembly ·
sugar cluster · a-synuclein
How to cite: Angew. Chem. Int. Ed. 2015, 54, 8435–8439
Angew. Chem. 2015, 127, 8555–8559
[1] a) M. E. Taylor, K. Drickamer, Introduction to Glycobiology;
Oxford University Press, New York, 2003; b) R. A. Dwek,
Chem. Rev. 1996, 96, 683 – 720.
[2] Y. Kamiya, M. Yagi-Utsumi, H. Yagi, K. Kato, Curr. Pharm. Des.
2011, 17, 1672 – 1684.
[3] E. A. Merritt, P. Kuhn, S. Sarfaty, J. L. Erbe, R. K. Holmes, W. G.
Hol, J. Mol. Biol. 1998, 282, 1043 – 1059.
[4] U. Neu, K. Woellner, G. Gauglitz, T. Stehle, Proc. Natl. Acad.
Sci. USA 2008, 105, 5219 – 5224.
[5] N. Yuki, Curr. Opin. Immunol. 2005, 17, 577 – 582.
[6] a) K. Yanagisawa, A. Odaka, N. Suzuki, Y. Ihara, Nat. Med.
1995, 1, 1062 – 1066; b) K. Matsuzaki, K. Kato, K. Yanagisawa,
Biochim. Biophys. Acta Mol. Cell Biol. Lipids 2010, 1801, 868 –
877; c) L.-P. Choo-Smith, W. K. Surewicz, FEBS Lett. 1997, 402,
95 – 98.
[7] J. Fantini, N. Yahi, Adv. Exp. Med. Biol. 2013, 991, 15 – 26.
[8] G. P. Gorbenko, P. K. Kinnunen, Chem. Phys. Lipids 2006, 141,
72 – 82.
[9] D. R. Taylor, N. M. Hooper, Mol. Membr. Biol. 2006, 23, 89 – 99.
[10] a) M. Utsumi, Y. Yamaguchi, H. Sasakawa, N. Yamamoto, K.
Yanagisawa, K. Kato, Glycoconjugate J. 2009, 26, 999 – 1006;
b) M. Yagi-Utsumi, T. Kameda, Y. Yamaguchi, K. Kato, FEBS
Lett. 2010, 584, 831 – 836; c) M. Yagi-Utsumi, K. Matsuo, K.
Yanagisawa, K. Gekko, K. Kato, Int. J. Alzheimers Dis. 2010,
2011, 925073.
[11] a) J. Fantini, N. Yahi, J. Mol. Biol. 2011, 408, 654 – 669; b) Z.
Martinez, M. Zhu, S. Han, A. L. Fink, Biochemistry 2007, 46,
1868 – 1877; c) T. Yamaguchi, T. Uno, Y. Uesaka, M. Yagi-
Utsumi, K. Kato, Chem. Commun. 2013, 49, 1235 – 1237.
8438
www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 8435 –8439

Angewandte
Chemie
[12] a) T. S. Ulmer, A. Bax, N. B. Cole, R. L. Nussbaum, J. Biol.
Chem. 2005, 280, 9595 – 9603; b) C. R. Bodner, C. M. Dobson, A.
Bax, J. Mol. Biol. 2009, 390, 775 – 790; c) C. R. Bodner, A. S.
Maltsev, C. M. Dobson, A. Bax, Biochemistry 2010, 49, 862 – 871;
d) T. Bartels, L. S. Ahlstrom, A. Leftin, F. Kamp, C. Haass, M. F.
Brown, K. Beyer, Biophys. J. 2010, 99, 2116 – 2124.
[13] a) M. Tominaga, K. Suzuki, M. Kawano, T. Kusukawa, T. Ozeki,
S. Sakamoto, K. Yamaguchi, M. Fujita, Angew. Chem. Int. Ed.
2004, 43, 5621 – 5625; Angew. Chem. 2004, 116, 5739 – 5743; b) N.
Kamiya, M. Tominaga, S. Sato, M. Fujita, J. Am. Chem. Soc.
2007, 129, 3816 – 3817; c) S. Sato, Y. Ishido, M. Fujita, J. Am.
Chem. Soc. 2009, 131, 6064 – 6065; d) T. Kikuchi, S. Sato, M.
Fujita, J. Am. Chem. Soc. 2010, 132, 15930 – 15932; e) D. Fujita,
K. Suzuki, S. Sato, M. Yagi-Utsumi, Y. Yamaguchi, N. Mizuno, T.
Kumasaka, M. Takata, M. Noda, S. Uchiyama, K. Kato, M.
Fujita, Nat. Commun. 2012, 3, 1093.
[16] a) N. Yamamoto, Y. Hirabayashi, M. Amari, H. Yamaguchi, G.
Romanov, W. E. Van Nostrand, K. Yanagisawa, FEBS Lett.
2005, 579, 2185 – 2190; b) N. Yamamoto, W. E. Van Nostrand, K.
Yanagisawa, Neuroreport 2006, 17, 1735 – 1737.
[17] H. Sasakawa, E. Sakata, Y. Yamaguchi, M. Masuda, T. Mori, E.
Kurimoto, T. Iguchi, S. Hisanaga, T. Iwatsubo, M. Hasegawa, K.
Kato, Biochem. Biophys. Res. Commun. 2007, 363, 795 – 799.
[18] a) K. Suzuki, S. Sato, M. Fujita, Nat. Chem. 2010, 2, 25 – 29;
b) Q.-F. Sun, J. Iwasa, D. Ogawa, Y. Ishido, S. Sato, T. Ozeki, Y.
Sei, K. Yamaguchi, M. Fujita, Science 2010, 328, 1144 – 1147;
c) Q.-F. Sun, T. Murase, S. Sato, M. Fujita, Angew. Chem. Int. Ed.
2011, 50, 10318 – 10321; Angew. Chem. 2011, 123, 10502 – 10505;
d) Q.-F. Sun, S. Sato, M. Fujita, Nat. Chem. 2012, 4, 330 – 333;
e) Q.-F. Sun, S. Sato, M. Fujita, Angew. Chem. Int. Ed. 2014, 53,
13510 – 13513; Angew. Chem. 2014, 126, 13728 – 13731; f) D.
Fujita, H. Yokoyama, Y. Ueda, S. Sato, M. Fujita, Angew. Chem.
Int. Ed. 2015, 54, 155—158; Angew. Chem. 2015, 127, 157 – 160.
[14] A satisfactory mass spectrum for sphere 5 was not obtained on
ESI mass spectrometers despite the optimization in sample
preparation and measurement conditions presumably due to the
large molecular weight and fragile, multi-charged structures.
[15] D. A. Kirschner, H. Inouye, L. K. Duffy, A. Sinclair, M. Lind,
D. J. Selkoe, Proc. Natl. Acad. Sci. USA 1987, 84, 6953 – 6957.
Received: March 3, 2015
Revised: March 31, 2015
Published online: May 26, 2015
Angew. Chem. Int. Ed. 2015, 54, 8435 –8439
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 8439