Rational design of β-sheet ligands against Aβ42- induced toxicity

Article abstract of DOI:10.1021/ja107675n

A β-sheet-binding scaffold was equipped with long-range chemical groups for tertiary contacts toward specific regions of the Alzheimer's Aβ fibril. The new constructs contain a trimeric aminopyrazole carboxylic acid, elongated with a C-terminal binding si

Full text of DOI:10.1021/ja107675n

ARTICLE
pubs.acs.org/JACS
Rational Design of β-Sheet Ligands Against Aβ42-Induced Toxicity
Katrin Hochd€orffer,† Julia M€arz-Berberich,† Luitgard Nagel-Steger,‡ Matthias Epple,†
Wolfgang Meyer-Zaika,† Anselm H.C. Horn,§ Heinrich Sticht,§ Sharmistha Sinha,|| Gal Bitan,||,^,# and
Thomas Schrader†,*
†Universit€at Duisburg-Essen, Fachbereich Chemie, Universit€atstrasse 5, 45117 Essen, Germany
‡Universit€at D€usseldorf, Fachbereich Biologie, Universit€atsstrasse 1, 40225 D€usseldorf, Germany
§Universit€at Erlangen-N€urnberg, Institut f€ur Biochemie, Fahrstrasse 17, 91054 Erlangen, Germany
Department of Neurology, David Geffen School of Medicine, ^Brain Research Institute, and #Molecular Biology Institute, University of
California at Los Angeles, 635 Charles East Young Drive South, Los Angeles, California 90095, United States
S Supporting Information
b
ABSTRACT: A β-sheet-binding scaffold was equipped with long-range chemical groups for
tertiary contacts toward specific regions of the Alzheimer’s Aβ fibril. The new constructs
contain a trimeric aminopyrazole carboxylic acid, elongated with a C-terminal binding site,
whose influence on the aggregation behavior of the Aβ42 peptide was studied. MD
simulations after trimer docking to the anchor point (F19/F20) suggest distinct groups
of complex structures each of which featured additional specific interactions with char-
acteristic Aβ regions. Members of each group also displayed a characteristic pattern in their
antiaggregational behavior toward Aβ. Specifically, remote lipophilic moieties such as a
dodecyl, cyclohexyl, or LPFFD fragment can form dispersive interactions with the nonpolar
cluster of amino acids between I31 and V36. They were shown to strongly reduce
Thioflavine T (ThT) fluorescence and protect cells from Aβ lesions (MTT viability assays). Surprisingly, very thick fibrils and a
high β-sheet content were detected in transmission electron microscopy (TEM) and CD spectroscopic experiments. On the other
hand, distant single or multiple lysines which interact with the ladder of stacked E22 residues found in Aβ fibrils completely dissolve
existing β-sheets (ThT, CD) and lead to unstructured, nontoxic material (TEM, MTT). Finally, the triethyleneglycol spacer
between heterocyclic β-sheet ligand and appendix was found to play an active role in destabilizing the turn of the U-shaped
protofilament. Fluorescence correlation spectroscopy (FCS) and sedimentation velocity analysis (SVA) provided experimental
evidence for some smaller benign aggregates of very thin, delicate structure (TEM, MTT). A detailed investigation by dynamic light
scattering (DLS) and other methods proved that none of the new ligands acts as a colloid. The evolving picture for the
disaggregation mechanism by these new hybrid ligands implies transformation of well-ordered fibrils into less structured aggregates
with a high molecular weight. In the few cases where fibrillar components remain, these display a significantly altered morphology
and have lost their acute cellular toxicity.
’ INTRODUCTION
the most promise for a breakthrough; however, after a halt in
phase III due to several cases of severe encephalopathy, currently
running trials have yielded beneficial results only in ApoE4
carriers and still have adverse effects in quite a large number of
patients. As a consequence, the call for small molecules was
reinitiated.
A plethora of small molecules has been screened in the past 3
decades for their antiaggregation potential against Aβ.2 Among
these are colored heterocyclic compounds (Congo Red,
cucurmin),3 Zn and Cu chelators,4 or peptides, in some cases
taken directly from putative nucleation sites within the Aβ
molecule;5 these were often modified derivatives of the LVFFA
sequence, expected to bind in a self-complementary fashion to
the central hydrophobic cluster (CHC) of Aβ. Soto presented
the β-sheet breaker LPFFD (iAbeta5),6 which retained the high
Alzheimer’s disease (AD) is a steadily increasing threat
especially for industrialized countries with a growing percentage
of old individuals; today an estimated 5.3 million U.S. citizens are
suffering from AD, and the number is predicted to double within
the next 50 years. Research on potential therapies has been going
on for several decades now, without producing one single drug
which is able to cure the disease. Since AD is accompanied by
many diverse pathologic mechanisms, numerous avenues have
been exploited in the search for a therapy. Antiinflammatory,
antihypertensive, as well as hypolipidemic agents, passive and
active immunization, cholinergic therapies, neuroprotective
agents, glutamate receptor antagonists, β- and γ-secretase in-
hibitors, amyloid β-protein and tau aggregation inhibitors, metal
chelating agents, monoamine oxidase inhibitors, and medicinal
plants are only a subset of the most prominent classes.1 In recent
years passive immunization with Aβ-specific antibodies has held
Received: August 25, 2010
Published: March 07, 2011
r
2011 American Chemical Society
4348
dx.doi.org/10.1021/ja107675n J. Am. Chem. Soc. 2011, 133, 4348–4358
|

Journal of the American Chemical Society
ARTICLE
Figure 1. Structural properties of the Aβ42 fibril. (a) Pentameric stack from the experimentally determined Aβ42 fibril15 indicating four candidate
regions for the interaction of C-terminally extended aminopyrazoles. These regions are color coded in orange, green, magenta, and blue. The
aminopyrazole moiety and the interacting phenylalanines are shown as red and yellow sticks, respectively. (b) Top view on one layer of the U-shaped
fibril showing the location of the individual amino acids (color coding according to their biophysical properties). (c) Detailed view of the hydrogen-
bonding interactions of the aminopyrazole with residues L17-A21. (d) Trimer moiety (in sticks and cpk coloring) docked to Aβ42. F19/20 of chain A
are depicted in yellow sticks and E22 residues of all chains (A-E) in red sticks. The dashed line indicates the orientation of ligand substituents after
docking.
affinity toward the self-complementary LVFFA region (Aβ(17-
21)) but impaired its β-sheet forming propensity by introducing
a proline kink.
The monomeric subunits form a longitudinal stack, thereby
creating two parallel in-register β-sheets. Amyloid growing can
proceed longitudinally, i.e., adsorption of new monomers along
the fibril axis, and laterally, i.e., association of another stack of
monomers at the lateral surface of the growing aggregate.16
Aminopyrazoles are rationally designed β-sheet ligands with a
specific DAD sequence of hydrogen-bond donors and acceptors,
perfectly complementary to that of a β-sheet.17 In numerous
experiments, they have been shown to bind selectively to the
backbone of misfolded peptides in which the predominant
structural element is a cross-β-sheet conformation. Several
successive generations were synthesized with more than one
heterocyclic binding unit for enhanced β-sheet affinity and
additional charged amino acids for enhanced water solubility. A
combination of two consecutive proteinogenic amino acids
flanked by external aminopyrazolecarboxylates was shown to
be exactly complementary to an extended β-sheet. Such deriva-
tives were synthesized and also evaluated on the solid phase.18
Direct interaction of dimeric and trimeric aminopyrazole deri-
vatives with Aβ42 was shown and characterized by FCS, AUC,
density gradient centrifugation, as well as HRMS. β-Sheet
recognition as well as the individual strength of all hydrogen
bonds involved were studied in great detail by R2PI spectroscopy
on a cooled argon jet stream.19 Recently, it was discovered that
the trimeric aminopyrazole carboxylate could also disassemble
preformed Aβ fibrils in a dose- and time-dependent manner.20
Concept. On the basis of the information about the Aβ-fibril
topology and the mode of aminopyrazole action we decided to
design a number of Trimer derivatives, which contain a variety of
C-terminal extensions of the original aminopyrazole trimer. Our
work includes a verification of the aminopyrazole binding site
itself and application of molecular dynamics simulations to
identify the binding regions of the C-terminal extensions.
Another prominent class are peptides containing alternating
N-methylated and nonmethylated peptide amides or esters,
presented by Meredith, Hughes, and Kapurniotu.7 These mole-
cules are able to cap growing β-sheets, because their back is
blocked for hydrogen bonding due to the sterically demanding
N-methyl groups or ester oxygens. They have recently been
optimized with respect to their antiaggregatory capacity by
introduction of three cyclohexylglycine units and reached nano-
molar IC50 values. The most potent compound, however, is
thought to accelerate Aβ self-assembly and thereby deplete the
level of neurotoxic Aβ oligomers.8 Other more recent examples
comprise the small molecule homotaurin (3-amino-1-propane-
sulfonic acid),9 scyllo-inositol,10 and (-)-epigallocatechin-3-gal-
late (EGCG).11 Surprisingly, little knowledge/information,
however, is available on the exact mechanism of action for most
Aβ complexing agents, even less on experimentally verified
structural details.12 To the best of our knowledge, the only case
is Sato’s concept of β-sheet packing: peptide inhibitors based on
a GxFxGxF framework disrupt sheet-to-sheet packing and inhibit
the formation of mature Aβ fibrils.13 This strategy was developed
from inspection of solid-state NMR structures of amyloid fibrils
and confirmed by 13C NMR spectroscopy for the best peptide
candidate in its direct complex with the 30-35 region of Aβ40.
The study above emphasizes the usefulness of structural
information on the Aβ topology for inhibitor design. In this
14
15
context, the structures of Aβ40 and Aβ42 in the fibril state,
which were determined by NMR spectroscopy, provide valuable
insight into Aβ aggregation. With respect to the monomeric
subunits, both structures consistently reveal two β-strands con-
nected by a turn, thus forming a U-shaped topology (Figure 1a).
4349
dx.doi.org/10.1021/ja107675n |J. Am. Chem. Soc. 2011, 133, 4348–4358

Journal of the American Chemical Society
ARTICLE
Figure 2. Final complex structures resulting from 20 ns molecular dynamics simulations of different C-terminally extended aminopyrazole ligands (cpk
coloring) in complex with a pentameric Aβ42-protofilament. Interacting residues are shown in stick representation: Phe19/20 (yellow), Glu22 (red), and
Met35 (orange). (a) Trimer-OMe, (b) Trimer-TEG-Dd, (c) Trimer-K-Che, (d) Trimer-TEG-NHMe, (e) Trimer-TEG-K, and (f) Trimer-TEG-
KKKKKG-OH.
On the basis of their binding site and the effect on the Aβ42
structure, the ligands can be divided in at least three distinct
classes. A large variety of ligands was subsequently synthesized,
and their effect on Aβ42 aggregation was systematically investi-
gated with a broad range of biophysical methods. These studies
revealed a close correlation between the modeled mode of
binding and the experimental properties observed. To probe
the relevance of these results in the environment of living cells,
we tested the toxicity of the new aminopyrazole derivatives and
their ability to inhibit Aβ42-induced toxicity in cell culture.21
In the present study, near-UV CD measurements of water-
soluble trimeric ligands in 1:1 mixtures with Aβ42 reveal that in
most cases a new CD band evolves at 260-320 nm with a
positive maximum at 280 nm, typical for complexes of aromatic
moieties (Phe 260-270 nm; Tyr 270-280 nm; Trp 290-
300 nm). Since at this wavelength aminopyrazole ligands are
CD silent, we attribute the new band to tight complex formation
between aromatic units in Aβ and ligand, most likely between the
two consecutive phenylalanines (F19/F20) and the pyrazole
nuclei. The importance of the π-π-stacking interactions is also
confirmed from molecular dynamics simulations, which prove
that the respective binding site exhibits a considerably higher
conformational stability compared to adjacent binding sites
(Supporting Information, Figure S1). These biophysical and
computational observations provide the basis for a structural
model of the aminopyrazole trimer in complex with the fibrillar
form of Aβ42. In this model, the trimer is hydrogen bonded to the
solvent-exposed top face of the KLVFF backbone and simulta-
neously stacks its pyrazole nuclei with both phenylalanines in the
form of a hydrophobic cleft (Figure 1).
Design of the C-Terminally Extended Ligand. We at-
tempted to match the main binding regions identified above
with complementary recognition sites on the complexing ami-
nopyrazole trimers and to vary sizes and distances from their
attachment point. (a) For the polar region (E22-K28) carbox-
ylate and ammonium groups were placed close to the aminopyr-
azoles. (b) Pentacationic appendices are probes for a potential
complexation of the glutamate-22 ladder formed by adjacent
layers of parallel β-sheets in the aggregated Aβ (Figures 1a and
1d). (c) The central cleft between the polar and the unpolar
region was addressed by a TEG (triethyleneglycol) unit, which
also serves as a water-soluble linker for remote recognition
events. (d) Unpolar residues (I31-V36) are matched by flexible
’ RESULTS AND DISCUSSION
Identification of Candidate Interaction Regions. The struc-
ture of the Aβ42 protofilament, which was experimentally
determined by L€uhrs et al.,15 gives important information about
those candidate regions, which might be targeted by C-terminally
extended aminopyrazoles (Figures 1a and 1b). There are at least
four regions with distinct structural properties: (a) a polar region,
formed by the stretch from D23-K28, (b) the E22 ladder, formed
by the E22 side chains of adjacent layers of the cross-β-sheet, (c)
the central cleft formed in the interior of the U-shaped turn, and
(d) the unpolar region, formed by residues A30-V36 of the
C-terminal β-strand.
Anchor Point for Molecular Modeling Studies. In order to
design suitable C-terminal extensions, the binding site of the
aminopyrazole trimer must first be determined with sufficient
accuracy. In the past, detailed conformational NMR analyses
between aminopyrazole ligands and the model peptide
(K)KLVFF, a putative nucleation site in Aβ, furnished strong
hints for direct hydrogen bonds between ligand and peptidic
backbone as well as π-stacking interactions with both phenylalanines.22
4350
dx.doi.org/10.1021/ja107675n |J. Am. Chem. Soc. 2011, 133, 4348–4358

Journal of the American Chemical Society
ARTICLE
Scheme 1. Lewis Structures of Unprotected Aminopyrazole Trimer Derivatives with Appendices Xa
a (Left) Small neutral, anionic, and cationic moieties as well as unpolar and TEG-spacer groups. (Right) Peptidic attachments.
Figure 3. Synthetic access to the new C-terminally modified aminopyrazole trimers via peptide coupling of various amines onto the PMB-protected
trimer and final total deprotection.
branched hydrocarbons as found in cyclohexylglycine, already
introduced by the Stott group.8 In addition, small peptide
fragments from the central hydrophobic core (CHC) were
employed.
In order to investigate whether these extensions are actually
capable of interacting with the Aβ42 fibril, molecular dynamics
simulations were performed for several representative ligands. All
computational studies started from the Aβ42-fibril structure
obtained from NMR spectroscopic data by L€uhrs et al.15 Starting
from the aminopyrazole binding site, C-terminally modified
ligands were modeled in complex with Aβ42 and the mode of
interaction was subsequently refined using molecular dynamics
simulations (see Supporting Information for details).
that both the backbone hydrogen bonds as well as the π-stacking
interactions of the aminopyrazole moiety remained stable
(Supporting Information Figures S1 and S2). Intriguingly,
several additional favorable interactions were found for the
C-terminally extended trimers, which were typical for each major
class of trimer extensions (Figure 2; see Supporting Information
Table S2 for a detailed list of the intermolecular interactions
formed).
Extended or cyclic unpolar groups align with nonpolar side
chains and undergo hydrophobic as well as dispersive interac-
tions (Trimer-Chg-Che, Trimer-K-Che, Trimer-TEG-Dd). Ac-
cording to a conformational search, unpolar binding sites on the
ligand indeed prefer the cluster of hydrophobic residues from I31
to V36. A TEG spacer in Trimer-TEG-Dd allows the attached
dodecyl tail to explore the entire M35 ladder (formed by adjacent
Modeling Results. Molecular dynamics simulations of the
Aβ42 ligand complexes in an explicit solvent environment proved
4351
dx.doi.org/10.1021/ja107675n |J. Am. Chem. Soc. 2011, 133, 4348–4358

Journal of the American Chemical Society
ARTICLE
Figure 4. Continued
Figure 4. Equilibrium of 33 μM Aβ42 aggregation vs disaggregation in the absence or presence of trimeric aminopyrazole inhibitors (each at 198 μM):
(top) inhibition of aggregation and (middle) disaggregation (10 mM PBS, pH 7.3). The color code for each aminopyrazole trimer derivative is
maintained throughout the entire article. For each bar 6 measurements were averaged. (Bottom) FCS measurements at 5 nm Aβ with Trimer-OH in
comparison to Trim-TEG-OH/OEt (each at 100 nM).
layers of parallel β-sheets in the aggregated Aβ) during MD
simulations for extended van der Waals interactions on the back
of the pentameric Aβ42 fibril (Figure 2b). For other nonpolar
ligands carrying a cyclohexyl moiety or a TEG-NMe extension,
penetration into the central cleft is observed (Figures 2c and 2d).
The latter ligands also exhibit the strongest destabilizing effect of
the Aβ42 fibril (Supporting Information Figure S3).
Trimer-TEG-K-OH) are capable of interacting with the E22
ladder of Aβ42 (Figures 2e and 2f). However, only long tethers
allowed simultaneous complexation of more than one carbox-
ylate group of the glutamate ladder by pentalysine attachments.
Two polar interactions are observed on average over the entire
simulation time for Trimer-TEG-KKKKKG-OH (Supporting
Information Tables S2 and S3).
Aminopyrazole derivatives, which carry one or multiple lysine
residues (Trimer-KKKKKG-OH, Trimer-TEG-KKKKKG-OH,
The simulations above suggest that three of the four candidate
interaction regions defined in Figure 1a can indeed be targeted by
4352
dx.doi.org/10.1021/ja107675n |J. Am. Chem. Soc. 2011, 133, 4348–4358

Journal of the American Chemical Society
ARTICLE
Figure 5. TEM pictures: (A) Aβ(1-42) fibrils; (B-H) Aβ(1-42) mixtures (20 μM) with aminopyrazole trimer derivatives (200 μM) ((B) Trimer-
OMe, (C) Trimer-OH, (D) Trimer-Che, (E) Trimer-K-OMe, (F) Trimer-TEG-K-OMe, (G) Trimer-TEG-OEt, (H) Trimer-TEG-OH.)
Table 1. Classification of CD Spectra to Three Main Types
the C-terminal extensions. Only for the polar region (D23-
K28) we observed very few interactions with the ligands. This
might be explained by the fact that D23 and K28 form a stable salt
bridge, and putative hydrogen bonds with S26/N27 have only a
small energetic contribution in the aqueous environment.
Another interesting finding is the fact that the TEG spacer
itself forms numerous interactions with the Aβ42 fibril in all
simulations of TEG-containing derivatives (Supporting Informa-
tion Table S2). These interactions are formed either with the
stretch F20-V24 of the N-terminal β-strand or with I32-L34 of
the C-terminal β-strand, suggesting that the TEG spacer itself
might play a beneficial role for the anti-Aβ activity of the ligands.
In summary, the MD simulations indicate that different types
of extension of the trimer core are capable of targeting different
sites on the Aβ42 fibril and undergo distinct interactions. To
investigate whether these different modes of binding also result
in different effects on Aβ aggregation, a large number of new
Trimer derivatives was synthesized and experimentally charac-
terized. Scheme 1 shows an overview of all synthesized amino-
pyrazole derivatives and their chemical classification.
Synthesis. The trimeric aminopyrazole’s core structure was
elongated with the above-discussed additional binding sites. To
this end, its C-terminal carboxylic acid was connected to the
respective amines by way of an amide bond. Conventional
coupling reagents comprised EDC/HOBt, HCTU/Cl-HOBt,
and Mukaiyama’s reagent, which produced the hybrid com-
pounds in high yields. In an economic fashion, all protecting
groups of the tether were finally cleaved by TFA together with all
PMB moieties on the aminopyrazole nuclei. Peptidic tethers
were first synthesized by manual solid-phase peptide synthesis
(SPPS) on a Wang resin, followed by covalent attachment at the
aminopyrazole trimer (HBTU, DIEA). Final deblocking of all
acid-labile protecting groups at 70 ꢀC for ∼3 h furnished, after
precipitation and recrystallization from ether, analytically pure
final products (Figure 3). All these new trimeric aminopyrazoles
are soluble in DMSO, most of them also in water.
type A (polar)
type B (very polar)
type C (unpolar)
Trim-K-OMe
Trim-KKKKKG-OH
Trim-K-Che
Trim-TEG-K-OMe Trim-TEG-KKKKKG-OH Trim-TEG-KLVFF-OH
Trim-TEG-OH Trim-K-TEGDA-K-Trim Trim-TEG-Dd
thermodynamically using the thioflavine T (ThT) fluorescence
assay.23 Care was taken to eliminate fluorescence changes
induced by any other events than the aggregation process; thus,
each ligand was separately shown to be nonfluorescent and not to
alter ThT fluorescence in mixtures. Controls with pure Aβ (1-
42) were identical in peptide concentration as well as buffer and
solvent composition.
Inhibition. After 72 h, the aggregation process has reached a
maximum in the absence or presence of any trimeric ligand. The
respective equilibrium concentration then was indicated by the
relative fluorescene intensity reached at the predefined end point
after 3 days. The full series of new aminopyrazole trimer
derivatives was subjected to ThT inhibition experiments, and
the final equilibrium was analyzed by comparison of the respec-
tive fluorescence intensities of intercalated dye at 482 nm (exc. at
442 nm). Figure 4 (top) reveals a significant structure-activity
relation for the ligands with respect to aggregation prevention of
Aβ(1-42): structurally related compounds in general display
comparable inhibition properties. Two classes of the modified
derivatives significantly surpass the original trimer activity:
Trimer-Che/Trimer-K-Che/Trimer-LPFFD-OH with attached
extended lipophilic groups and Trimer-TEG-K-OMe/Trimer-
KKKKKG-OH with a distant or multiple lysine residues. Their
proposed mechanism of action is discussed within the next
sections together with all the other biophysical and biochemical data.
The strong influence of the spacer unit becomes particularly
evident for compounds with a single lysine residue. A single
C-terminal lysine directly attached to the trimer (Trim-K-OMe)
exhibits very special properties: Contrary to almost all other
trimer derivatives, this aminopyrazole accelerates Aβ aggregation
(data not shown). On the other hand, Trim-TEG-K-OMe, with a
Aggregation Studies with ThT. The influence of the new β-
sheet ligands on the Aβ self-assembly process was first studied
4353
dx.doi.org/10.1021/ja107675n |J. Am. Chem. Soc. 2011, 133, 4348–4358

Journal of the American Chemical Society
ARTICLE
Figure 6. Typical CD spectra recorded for aggregated Aβ(1-42) alone (black) and after addition of aminopyrazole ligand (type A, Trimer-K-OMe:
red; type B, Trimer-KKKKKG-OH blue; type C, Trimer-K-Che: brown). Time progresses in the direction of the embedded arrows, indicated by lighter
colors (0 h, 1 h, 1d, 2d). Test solutions contained 10 μM Aβ(1-42), 5 μM potassium phosphate buffer (pH = 7.3), 2% HFIP, and 10 μM of the
respective aminopyrazole trimer derivative.
Table 2. Cell Viability and IC50 Values of the Most Potent
aminopyrazole ligands operate in a fully reversible fashion and
reach an open equilibrium when the thermodynamically most
favorable complex is formed. The total reduction in Aβ aggregate
concentration by the trimer parent compound corresponds to an
estimated dissociation constant of the respective complex in the
low micromolar range (2 μM Kd in PBS buffer).
Colloidal Mechanism? It might be argued that the observed
effect could in principle also stem from colloidal aggregates,
which have been shown to inhibit amyloid polymerization in
prion proteins in an unspecific way.24 To clarify this point,
extensive DLS measurements were carried out in combination
with AFM and DOSY experiments as well as cmc determinations
(all experiments in the Supporting Information). With one
exception (Trimer-TEG-K-OMe) none of our new ligands forms
colloids, which is further corroborated by inspection of TEM
images of the pure compounds. Even this ligand did not show a
colloidal mechanism, when we transferred the established tests
for colloidal inhibition from prion protein to seeded Aβ
aggregation.
The structural influence of differently modified trimer deriva-
tives on Aβ aggregation was now further studied with fluores-
cence correlation spectroscopy (FCS), sedimentation velocity
analysis (SVA, based on analytical ultracentrifugation), and
transmission electron microscopy (TEM).
Fluorescence Correlation Spectroscopy. In the FCS experi-
ment a 5 nM Oregon-Green-labeled Aβ42 solution was prepared
in PBS buffer with ∼3% DMSO.25 These Aβ concentrations are
also found in the human brain.26-28 Trimer-OH was added at
100 nM and reduced the peaks ꢀ height value to ∼60% of
the control. Since the number of peaks remained unchanged,
their molecular weight was significantly reduced.29 Much more
pronounced is the influence of Trimer-TEG derivatives: FCS
here witnesses the uniform transition to very small oligomers
within all Aβ aggregates (Figure 4, bottom). This observation
inhibitors of Aβ-Induced Toxicitya
inhibitor
viability [%]
IC50 [μM]
Trim-LPFFD-OH
Trim-TEG-LPFFD-OH
Trim-GABA-OH
97
3.1
113
103
97
10.1
18.5
20.3
21.5
35.3
52.7
81.0
>100
>100
Trim-GABA-OMe
Trim-K-TEGDA-K-Trim
Trim-TEG-OH
116
111
117
97
Trim-TEG-K-OMe
Trim-Chg-Che
Trim-TEG-KLVFF-OH
Trim-TEG-Dd
100
108
a Aβ lesion control (10 μM) was set at ∼75% viability.
single lysine separated from the trimeric aminopyrazole core unit
by the TEG spacer, leads to the most efficient suppression of Aβ
fibril formation of all tested derivatives (20% ThT fluorescence).
This structurally closely related pair of aminopyrazole trimers is a
striking example of how strongly the exact placement of binding
moieties influences the degree and path of Aβ aggregation.
Disaggregation. Starting from preformed fibrils, the disas-
sembly process was also monitored with the full series of
modified trimeric aminopyrazoles (Figure 4). Remarkably, the
inhibition pattern (top) looks quite similar to the disaggregation
pattern (bottom); in several cases, the total percentage of
remaining ThT fluorescence is identical. In other words, little
difference is observed between the end points of aggregation
assays, in which the ligand inhibits de novo aggregation starting
from monomeric Aβ molecules, and those experiments which
require disaggregation of preformed fibrils by externally added
ligand. Experimental evidence is thus provided for the fact that
4354
dx.doi.org/10.1021/ja107675n |J. Am. Chem. Soc. 2011, 133, 4348–4358

Journal of the American Chemical Society
ARTICLE
Figure 7. Viability assays of PC-12 cells with trimeric aminopyrazoles (100 μM). Aβ lesion control (10 μM) was set at ∼75% viability. Most effective
drugs are marked with asterisks.
various kinds of aminopyrazole trimers and Aβ (Figure 5). TEM
pictures were obtained from mature Aβ fibrils (>600 nm long) as
well as globular particles (3-30 nm diameter), grown in the
absence of aminopyrazole ligands.33
While twisted fibrils from pure Aβ had a diameter of 10 nm,
thin filaments were produced in the presence of the Trimer-OH
(5 nm diameter) and the number of mature fibrils was greatly
reduced. We tentatively conclude that the trimer seems to break
the mature fibrils into protofilaments by a combination of
backbone recognition and hydrophobic interactions. The un-
polar Trimer-Che, on the other hand, produced very thick
twisted fibrils (up to 70 nm), with a length of at least 600 nm.
It thus seems that here fibrillogenesis is not prevented but rather
shifted to a much more compact form, which does not accom-
modate well-ordered fluorescent ThT molecules. TEM pictures
of mixtures of aggregation accelerator Trimer-K-OMe with Aβ
show very thin fibrils (5-10 nm) of 800 nm length, while
Trimer-TEG-K-OMe redirects the aggregation process of Aβ
to unstructured material and thin bent filaments, distinct from
Aβ fibrils. Finally, from Aβ solutions in the presence of Trim-
TEG-OH or -OMe, very thin, delicate structures evolve, which
show no helical twist. In total, similar appendices again produced
similar TEM morphologies of aggregated material, which must
originate from their similar complexation behavior.
CD Spectroscopy. CD spectra could only be measured from
Aβ complexes with the most soluble aminopyrazole trimers.
None of the examined ligands had its own CD spectrum above
210 nm, so that spectral changes could be directly attributed to
interactions between Aβ and the respective ligand. They can be
divided into 3 groups (Table 1, Figure 6): Type A is found for
derivatives with a single lysine or TEG spacer only; it is
characterized by an induced CD in the aromatic region around
280 nm, without a significant loss of β-sheet conformation. Type
B is obtained with the most polar derivatives which contain
multiple lysine residues; these are the most potent of all
synthesized aminopyrazole derivatives and deplete the ThT
fluorescence level down to 20% in inhibition and disaggregation
experiments (Figure 4). The corresponding CD spectrum fea-
tures again the additional CD band at ∼280 nm, typical for
aromatic amino acids; however, in this case, the β-sheet band
almost completely disappears with time, indicating dissolution of
the secondary peptide structure.
Figure 8. Dose-response curve for inhibition of Aβ-induced toxicity in
PC-12 cells by Trimer-LPFFD (IC50 = 3.1 μM).
underlines the overall beneficial effect of the TEG spacer on the
antiaggregation potency of the ligands. In addition, the differences
between Trimer-TEG-OHand Trimer-TEG-OEt indicate that the
polarity of the extension also plays an important role for the
interaction with the Aβ42 fibril. This is in line with the numerous
interactions observed for Trimer-TEG-NMe in the modeling
study (Figure 2d and Supporting Information Table S2).
Sedimentation Velocity Analysis. Markedly different behav-
ior was also observed for various classes of modified aminopyr-
azole trimers in ultracentrifugation experiments with Oregon-
Green-labeled Aβ:30 weight-averaged sedimentation coefficients
for pure Aβ were about 50 S, whereas those for its complex with
Trimer-OH continuously decreased in a dose-dependent man-
ner down to 25S, corresponding to a 50% molecular weight
reduction (Supporting Information Figures S4 and S5).31 By
contrast, the TEG-elongated aminopyrazole trimer does not
significantly change the medium aggregate size of aggregated
Aβ but instead produces a significant amount of smaller oligo-
mers (monomers-18mers). On the other extreme, the excep-
tionally high aggregation propensity of Trim-K-OMe is also well
documented in sedimentations, featuring the total loss of soluble
Aβ-peptide signal during rotor acceleration. For all these trimer
derivatives, sedimentation assays prove direct complex formation
with Aβ.32
Transmission Electron Microscopy. A broad spectrum of
different morphologies were produced in complexes between
4355
dx.doi.org/10.1021/ja107675n |J. Am. Chem. Soc. 2011, 133, 4348–4358

Journal of the American Chemical Society
ARTICLE
Table 3. Synopsis of All Experimental Results Ordered by Ligand Classes
E22 ladder
polar region
unpolar region
central pore
(TEG derivatives)
experiment
(polylysines)
(single charged groups)
(extended alkyl residues)
ThT (inhibition)a
ThT (disaggregation)a
FCSb
sedimentationc
TEMd
24%
25%
21%
27-50%
35-62%
48-54%
28%
42-68%
3-21%
50Sþ
amorphous material
50Sþ
50Sþ
thick twisted fibrils
(70 nm ꢀ 600 nm)
increased β-sheet band
5-10S þ 50S
very thin, delicate structures
unstructured material þ
thin bent filaments
ICD (280 nm),
less β-sheet (218 nm)
IC50 g 22 μM
CDe
ICD (280 nm);
no β-sheet (218 nm)
IC50 > 80 μM
ICD (280 nm),
less β-sheet (218 nm)
MTTf
IC50 g 3 μM
IC50 g 35 μM
a Percent ThT fluorescence of a 6:1 mixture relative to Aβ42 control (10 μM). b Percent OregonGreen fluorescence of a 20:1 mixture relative to Aβ42
control (5 nM). c 10:1 mixture; Aβ42 control (20 μM) shows only 50S particles. d Ten-fold diluted sedimentation samples. e 1:1 mixture with Aβ42 (10
μM each). f PC-12 cells with Aβ42 lesion (10 μM).
CD spectra of Trimer-K-Che, Trimer-TEG-KLVFF-OH, and
Trimer-TEG-DD all feature/display an almost doubled β-sheet
band intensity and the total loss of the aromatic signal at 280 nm
(type C). A straightforward interpretation suggests a remarkable
stabilization of the cross-β-sheet (in line with TEM results) with
concomitant withdrawal of the aminopyrazole from its Phe-
Phe cleft.
Although it is tempting to correlate CD spectra with modeling
structures, it should be kept in mind that the time regimes are
drastically different: while MD simulations last only for a few
nanoseconds and would rather represent structures formed in
the early stages of fibril disassembly, the CD spectra as well as
ThT assays cover several days, including potential major con-
formational rearrangements.
Inhibition of Aβ-Induced Toxicity. Initially, all compounds
were screened for any toxic effect they might have by adding each
compound at 100 μM and measuring the effect on cell viability using
the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide
(MTT) reduction assay34 in differentiated rat pheochromocytoma
(PC-12) cells.35 Encouragingly, none of the compounds was toxic to
the cells at this concentration. Next, to examine whether the
aminopyrazole trimers could protect from Aβ42-induced neurotoxi-
city, cells were treated for 24 h with 10 μMAβ(1-42) in the absence
or presence of 100 μM of each compound. The viability of untreated
cells decreased by 30-40% (MTT). Inhibition of Aβ(1-42)-
induced toxicity by Trimer-OH/OMe was very moderate (<10%
viability increase); however, many of the new derivatives achieved a
complete rescue of cell viability (Table 2, Figure 7).
Intriguingly, the most efficient inhibition of Aβ toxicity was
achieved with 3 lipophilic extensions and Trimer-TEG-K-OMe,
which were also superior in ThT and related assays. The two
GABA derivatives are a surprise; they might potentially interact
with GABA receptors and not with the Aβ peptide itself. The
above-delineated findings demonstrate that trimeric aminopyr-
azoles are indeed active against Aβ-induced toxicity in living cells;
they also provide experimental evidence for their low toxicity at
relatively high doses of 0.1 mM, inspite of, e.g., the presence of an
N-terminal nitro group.
increasing concentrations of β-sheet ligand (0.3, 1, 3, 10, 30, and
100 μM). IC50 in this respect is defined as the concentration of
the β-sheet ligand (aminopyrazole trimer derivative), at which
the inhibition of Aβ toxicity just reaches 50% (Figure 8).
With respect to IC50 values, the two nonpolar LPFFD
derivatives were now found to be the most effective. Since the
Aβ42 concentration was always kept constant at 10 μM, it should
be emphasized that substoichiometric IC50 values such as that of
Trim-LPFFD-OH indicate very high affinity toward the target
peptide, even if a 1:1 complex is assumed. A 3 μM IC50 value is a
remarkable number by itself and testifies to an efficient disag-
gregation capability in the cell environment. It should be kept in
mind that in the brain Aβ concentrations are in the low
nanomolar range, similar to the situation in enzyme assays. Only
in such a scenario, IC50 values can be expected to drop to
nanomolar concentrations.
’ CONCLUSIONS AND OUTLOOK
This investigation provides experimental evidence for the fact
that small structural changes in β-sheet ligands can have a
profound influence on the aggregation behavior of misfolding
proteins. Only one of all examined aminopyrazole ligands forms
colloids, and none displayed the characteristics of a colloidal
mechanism, so that direct interaction of amyloid peptide with
monomeric ligand species can be assumed. Moreover, a common
anchor point (F19-F20) could be identified, allowing one to
perform docking experiments and subsequent MD simulations.
Intriguingly, various types of computationally designed binding
sites on the ligands allow tight interactions with specific Aβ
regions (cf. MD simulations) and correlate well with different
kinds of modulated Aβ aggregation behavior, throughout the
entire series of analytical experiments, Table 3.
It should however be emphasized that MD simulations treat
only a very short time span in the beginning of the aminopyr-
azoles’ interactions with Aβ. CD measurements as well as ThT
aggregation assays on the other hand display pronounced con-
tinuous changes within more than 2 h, which may involve con-
formational changes that are not yet visible even during extended
molecular dynamics simulations. A preliminary structure-
activity relation can therefore be suggested only with great
care, evolving from the combination of different biophysical,
biochemical, and “in silico” experiments. Two major binding
motifs were discovered, which greatly improve the β-sheet
On the basis of this initial screen, we decided to evaluate the
IC50 value of compounds that increased the viability of PC-12
cells to g90%. The data are summarized in Table 2. To
determine the IC50 value of each of the compounds, dose-
dependence MTT experiments were conducted with the ami-
nopyrazole trimers at a fixed Aβ42 concentration of 10 μM and
4356
dx.doi.org/10.1021/ja107675n |J. Am. Chem. Soc. 2011, 133, 4348–4358

Journal of the American Chemical Society
ARTICLE
breaker ability of the aminopyrazole trimer: remote lipophilic
moieties for dispersive interactions with the nonpolar cluster of
amino acids between I31 and V3636 and distant polyLys stretches
which interact with the E22 ladder. Only the latter, however,
effectively destroys the cross-β-sheet (CD, TEM). In addition,
the TEG spacer was found to play an active role in destabilizing
the turn of the U-shaped protofilament (Trim-TEG-OMe; Trim-
TEG-K-OMe). Importantly, only the synopsis of ThT assay,
sedimentation experiment, and TEM picture allows one to draw
conclusions about the aggregate species formed by direct inter-
action between Aβ and the aminopyrazole trimer derivatives: In
most cases the content of well-ordered soluble fibrils is greatly
diminished (ThT), because these are transformed into less
structured aggregates with a high molecular weight (UC). The
few remaining fibrillar components display a significantly altered
morphology (TEM) and have lost their acute cellular toxicity
(MTT). We will in the future try to confirm these structure
motifs postulated from modeling and aggregation experiments.
Direct evidence should be gained from cocrystals of these
complexes as well as from 2D solid-state NMR experiments.
K. E.; Ali, S. A. E.; Qureshi, M. M.; Allsop, D.; El-Agnaf, O. M. A.
Biochemistry 2008, 47, 1984–1992. (c) KLVFF aggregation and gelation:
Krysmann, M. J.; Castelletto, V.; Kelarakis, A.; Hamley, I. W.; Hule,
R. A.; Pochan, D. J. Biochemistry 2008, 47, 4597–4605.
(6) Soto, C.; Sigurdsson, E. M.; Morelli, L.; Kumar, R. A.; Casta~no,
E. M.; Frangione, B. Nat. Med. 1998, 4, 822–826.
(7) Gordon, D. J.; Meredith, S. C. Biochemistry 2003, 42, 475–485.
Hughes, E.; Burke, R. M.; Doig, A. J. J. Biol. Chem. 2000, 275,
25109–25115. Kapurniotu, A.; Schmauder, A.; Tenidis, K. J. Mol. Biol.
2002, 315, 339–350.
(8) Kokkoni, N.; Stott, K.; Amijee, H.; Mason, J. M.; Doig, A. J.
Biochemistry 2006, 45, 9906–9918.
(9) Aisen, P. S; Saumier, D.; Briand, R.; Laurin, J.; Gervais, F.;
Tremblay, P.; Garceau, D. A. Neurology 2006, 67, 1757–63. Wright,
T. M. Drugs Today 2006, 42, 291–298 (Homotaurin dropped out of
phase III clinical trial).
(10) (a) McLaurin, J.; Goloumb, R.; Jurewicz, A.; Antel, J. P.; Fraser,
P. E. J. Biol. Chem. 2000, 275, 18495–18502. (b) McLaurin, J.; Kierstead,
M. E.; Brown, M. E.; Hawkes, C. A.; Lambermon, M. H. L.; Phinney,
A. L.; Darabie, A. A.; Cousins, J. E.; French, J. E.; Lan, M. F.; Chen, F.;
Wong, S. S. N.; Mount, H. T. J.; Fraser, P. E.; Westaway, D.; St. George-
Hyslop, P. Nat. Med. 2006, 12, 801–808.
(11) Ehrnhoefer, D. E.; Bieschke, J.; Boeddrich, A.; Herbst, M.;
Masino, L.; Lurz, R.; Engemann, S.; Pastore, A.; Wanker, E. E. Nat.
Struct. Mol. Biol. 2008, 15, 558–566.
’ ASSOCIATED CONTENT
(12) Structure of an affibody/Aβ40 complex: Hoyer, W.; Gronwall,
C.; Jonsson, A.; Stahl, S.; Hard, T. Proc. Natl. Acad. Sci. U.S.A. 2008,
105, 5099–5104. Hoyer, W.; Hard, T. J. Mol. Biol. 2008, 378, 398–411.
(13) Sato, T.; Kienlen-Campard, P.; Ahmed, M.; Liu, W.; Li, H. L.;
Elliott, J. I.; Aimoto, S.; Constantinescu, S. N.; Octave, J. N.; Smith, S. O.
Biochemistry 2006, 45, 5503–5516.
(14) Petkova, A. T.; Ishii, Y.; Balbach, J. J.; Antzutkin, O. N.;
Leapman, R. D.; Delaglio, F.; Tycko, R. Proc. Natl. Acad. Sci. U.S.A.
2002, 99, 16742–16747. Petkova, A. T.; Yau, W. M.; Tycko, R.
Biochemistry 2006, 45, 498–512.
S
Supporting Information. Detailed synthetic experimen-
b
tal procedures, computational studies, thioflavine T-assays, de-
termination of the equilibrium from inhibition and disaggrega-
tion experiments, colloid experiments, UV-vis extinctions, DLS,
AFM, DOSY, cmc determination, seeded aggregations, CD-
spectroscopic measurements, FCS, sedimentation analysis, trans-
mission electron microscopy, and MTT viability assays. This
material is available free of charge via the Internet at http://pubs.
acs.org.
(15) L€uhrs, T.; Ritter, C.; Adrian, M.; Riek-Loher, D.; Bohrmann, B.;
Dobeli, H; Schubert, D.; Riek, R. Proc. Natl. Acad. Sci. U.S.A. 2005,
102, 17342–17347.
’ AUTHOR INFORMATION
(16) Ma, B.; Nussinov, R. Curr. Opin. Chem. Biol. 2006, 10, 445–452.
Sato, T.; Kienlen-Campard, P.; Ahmed, M.; Liu, W.; Li, H.; Elliott, J. I.;
Aimoto, S.; Constantinescu, S. N.; Octave, J. N.; Smith, S. O. Biochem-
istry 2006, 45, 5503–5516.
Corresponding Author
Thomas.Schrader@uni-due.de.
(17) Schrader, T.; Kirsten, C. J. Chem. Soc., Chem. Commun.
1996, 2089. Schrader, T.; Kirsten, C. N. J. Am. Chem. Soc. 1997,
119, 12061–12068.
’ ACKNOWLEDGMENT
Financial support from the American Health Assistance
Foundation (grant A2008-350) and the UCLA Jim Easton
Consortium for Alzheimer’s Drug Discovery and Biomarker
Development is gratefully acknowledged. This work was also
supported by the Deutsche Forschungsgemeinschaft and the
Volkswagen foundation.
(18) (a) Rzepecki, P.; Wehner, M.; Molt, O.; Zadmard, R.; Schrader, T.
ꢀ
Synthesis 2003, 1815–1826. (b) Cernovskꢁa, K.; Kemter, M.; Gallmeier, H.-C.;
Rzepecki, P.; Schrader, T.; K€onig, B. Org. Biomol. Chem. 2004, 2, 1603–1611.
(c) Rzepecki, P.; Gallmeier, H.; Geib, N.; Cernovska, K.; K€onig, B.; Schrader,
T. J. Org. Chem. 2004, 69, 5168–5178. (d) Rzepecki, P.; Geib, N.; Peifer, M.;
Biesemeier, F.; Schrader, T. J. Org. Chem. 2007, 72, 3614–3624.
(19) (a) Rzepecki, P.; Nagel-Steger, L.; Feuerstein, S.; Linne, U.;
Molt, O.; Zadmard, R.; Aschermann, K.; Wehner, M.; Schrader, T.;
Riesner, D. J. Biol. Chem. 2004, 279, 47479–47505. (b) Fricke, H.; Funk,
A.; Schrader, T.; Gerhards, M. J. Am. Chem. Soc. 2008, 130, 4692–4698.
(c) Fricke, H.; Gerlach, A.; Unterberg, C.; Wehner, M.; Schrader, T.;
Gerhards, M. Angew. Chem. 2009, 48, 900–904.
(20) Biesemeier, F.; K. Hochd€orffer, L. Nagel-Steger, Riesner, D.;
Schrader, T.; . DE 10 2006 015 140.2, Patent application March 28, 2006.
(21) This study was conducted with Aβ42, which is far more aggressive
than Aβ40; however, since both peptides contain the same well-oriented central
peptide fragment, we expect the inhibitors to work just as well for Aβ40.
(22) Rzepecki, P.; Schrader, T. J. Am. Chem. Soc. 2005, 127,
3016–3025.
’ REFERENCES
(1) Review: Jakob-Roetne, R.; Jacobsen, H. Angew. Chem., Int. Ed.
2009, 48, 3030–3059.
(2) Bisstyrylbenzenes: Flaherty, D. P.; Walsh, S. M.; Kiyota, T.;
Dong, Y.; Ikezu, T.; Vennerstrom, J. L. J. Med. Chem. 2007,
50, 4986–4992. (b) Porat, Y; Abramowitz, A.; Gazit, E. Y. Chem. Biol.
Drug Des. 2006, 67, 27–37.
(3) Yang, F.; Lim, G. P.; Begum, A. N.; Ubeda, O. J.; Simmons, M. R.;
Ambegaokar, S. S.; Chen, P. P.; Kayed, R.; Glabe, C. G.; Frautschy, S. A.;
Cole, G. M. J. Biol. Chem. 2005, 280, 5892–5901.
(4) Zn and Cu chelators:Cherny, R. A.; Atwood, C. S.; Xilinas, M. E.;
Gray, D. N.; Jones, W. D.; McLean, C. A.; Barnham, K. J.; Volitakis, I.;
Fraser, F. W.; Kim, Y. Neuron 2001, 30, 665–676.
(23) LeVine, H., III Methods Enzymol. 1999, 309, 274–284.
(24) Feng, B. Y.; Toyama, B. H.; Wille, H; Colby, D. W; Collins,
S. R.; May, B. C. H.; Prusiner, S. B.; Weissman, J.; Shoichet, B. K. Nat.
Chem. Biol. 2008, 4, 197–199.
(5) KLVFF derivatives: (a) Tjernberg, L. O.; Naslund, J.; Lindqvist,
L.; Johansson, J.; Karlstrom, A. R.; Thyberg, J.; Terenius, L.; Nordstedt,
C. J. Biol. Chem. 1996, 271, 8545–8548. (b) Austen, B. M.; Paleologou,
4357
dx.doi.org/10.1021/ja107675n |J. Am. Chem. Soc. 2011, 133, 4348–4358

Journal of the American Chemical Society
ARTICLE
(25) Chang, L.; Bakhos, L.; Wang, Z.; Venton, D. L.; Klein, W. L.
J. Mol. Neurosci. 2003, 20, 305–313.
(26) Haass, C.; Schlossmacher, M. G.; Hung, A. Y.; Vigo-Pelfrey, C.;
Mellon, A.; Ostaszewski, B. L. Nature 1992, 359, 322–325.
(27) Shoji, M.; Golde, T. E.; Ghiso, J.; Cheung, T. T.; Estus, S.;
Shaffer, L. M. Science 1992, 258, 126–129.
(28) Seubert, P.; Vigo-Pelfrey, C.; Esch, F.; Lee, M.; Dovey, H.;
Davis, D. Nature 1992, 359, 325–327.
(29) The Aβ concentration is well below Aβ’s cmc:Klajnert, B.;
Cortijo-Arellano, M.; Bryszewska, M; Cladera, J. Biochem. Biophys. Res.
Commun. 2006, 339, 577–582.
(30) Molecular Probes; Invitrogen: Karlsruhe, Germany.
(31) Demeler, B.; van Holde, K. E. Anal. Biochem. 2004, 335,
279–288.
(32) Demeler, B.; Brookes, E.; Nagel-Steger, L. Methods Enzymol.
2009, 454, 87–113.
(33) Nagel-Steger, L.; Demeler, B.; Meyer-Zaika, W.; Hochd€orffer,
K.; Schrader, T.; Willbold, D. Eur. Biophys. J. 2009, Feb, 24.
(34) Datki, Z.; Juhasz, A.; Galfi, M.; Soos, K.; Papp, R.; Zadori, D.;
Penke, B. Brain Res. Bull. 2003, 62, 223–229.
(35) Shearman, M. S. Methods Enzymol. 1999, 309, 716–723.
(36) The most powerful inhibitor of Aβ-induced toxicity Trimer-
TEG-LPFFD-OH uses the well-known β-sheet-breaker motif intro-
duced by Soto et al.
4358
dx.doi.org/10.1021/ja107675n |J. Am. Chem. Soc. 2011, 133, 4348–4358