Design of β-amyloid aggregation inhibitors from a predicted structural motif

Article abstract of DOI:10.1021/jm201332p

Drug design studies targeting one of the primary toxic agents in Alzheimer's disease, soluble oligomers of amyloid β-protein (Aβ), have been complicated by the rapid, heterogeneous aggregation of Aβ and the resulting difficulty to structurally characterize the peptide. To address this, we have developed [Nle³⁵, d-Pro³⁷]Aβ₄₂, a substituted peptide inspired from molecular dynamics simulations which forms structures stable enough to be analyzed by NMR. We report herein that [Nle ³⁵, d-Pro³⁷]Aβ₄₂ stabilizes the trimer and prevents mature fibril and β-sheet formation. Further, [Nle³⁵, d-Pro³⁷]Aβ₄₂ interacts with WT Aβ₄₂ and reduces aggregation levels and fibril formation in mixtures. Using ligand-based drug design based on [Nle³⁵, d-Pro³⁷] Aβ₄₂, a lead compound was identified with effects on inhibition similar to the peptide. The ability of [Nle³⁵, d-Pro ³⁷]Aβ₄₂ and the compound to inhibit the aggregation of Aβ₄₂ provides a novel tool to study the structure of Aβ oligomers. More broadly, our data demonstrate how molecular dynamics simulation can guide experiment for further research into AD.

Full text of DOI:10.1021/jm201332p

Article
pubs.acs.org/jmc
Design of β-Amyloid Aggregation Inhibitors from a Predicted
Structural Motif
Paul A. Novick,*,#,† Dahabada H. Lopes,#,‡ Kim M. Branson,#,† Alexandra Esteras-Chopo,§
Isabella A. Graef,§ Gal Bitan,‡,∥,⊥ and Vijay S. Pande†
†Department of Chemistry, Stanford University, Stanford, California 94305, United States
‡Department of Neurology, UCLA, Los Angeles, California 90095, United States
§Department of Pathology, Stanford University, Stanford, California 94305, United States
∥Brain Research Institute, UCLA, Los Angeles, California 90095, United States
⊥Molecular Biology Institute, UCLA, Los Angeles, California 90095, United States
S
* Supporting Information
ABSTRACT: Drug design studies targeting one of the
primary toxic agents in Alzheimer’s disease, soluble oligomers
of amyloid β-protein (Aβ), have been complicated by the
rapid, heterogeneous aggregation of Aβ and the resulting
difficulty to structurally characterize the peptide. To address
this, we have developed [Nle35, D-Pro37]Aβ42, a substituted
peptide inspired from molecular dynamics simulations which
forms structures stable enough to be analyzed by NMR. We
report herein that [Nle35, D-Pro37]Aβ42 stabilizes the trimer
and prevents mature fibril and β-sheet formation. Further,
[Nle35, D-Pro37]Aβ42 interacts with WT Aβ42 and reduces aggregation levels and fibril formation in mixtures. Using ligand-based
drug design based on [Nle35, D-Pro37]Aβ42, a lead compound was identified with effects on inhibition similar to the peptide. The
ability of [Nle35, D-Pro37]Aβ42 and the compound to inhibit the aggregation of Aβ42 provides a novel tool to study the structure of
Aβ oligomers. More broadly, our data demonstrate how molecular dynamics simulation can guide experiment for further research
into AD.
early cognitive deficits to soluble Aβ oligomers has emerged.9,17,18
Still, because of the rapid aggregation kinetics of Aβ and the
existence of Aβ oligomers in dynamically changing mixtures, it
INTRODUCTION
Alzheimer’s disease (AD) is a degenerative brain disorder and a
major cause of dementia for which no cure is known. Currently,
■
has been difficult to elucidate the exact mechanism of assembly
>35 million people worldwide are believed to suffer from the
and the relationship between specific oligomer structures and
disease and with the aging of the population this number is
expected to increase by a factor of 3−4 over the next 40 years.1
AD is characterized by two pathological hallmarks, extracellular
deposits composed primarily of amyloid β-protein (Aβ) fibrils
and intracellular, hyperphosphorylated tau fibrils. While the
their toxic activity. Bitan et al. have used photo-induced cross-
linking of unmodified proteins (PICUP) to demonstrate the
rapid equilibration of monomers and oligomers in freshly
prepared Aβ, with distinct oligomer size distributions for Aβ40
and Aβ42.19 Recently, the relationship between oligomer order
identity of the molecular species responsible for the initiation of
and toxicity was elucidated for Aβ40 monomer through
AD pathology has been debated historically, genetic evidence
tetramer using PICUP-stabilized oligomers.20 Nonetheless,
testing the biological activity of single oligomeric Aβ species
has been highly challenging.
from early onset familial AD,2−5 and experimentally observed
neurotoxicity of Aβ,6−11 have indicated that Aβ assembly into
β-sheet rich aggregates is a crucial, early stage, neurotoxic
element in AD.
An additional consequence of the heterogeneous composi-
tion of Aβ assemblies is the difficulty determining the structure
of soluble Aβ oligomers under physiological conditions.
Traditional high-resolution methods, such as X-ray crystallog-
raphy and solution-state NMR spectroscopy, have been un-
successful at determining the structure of the pathologically and
physiologically relevant soluble Aβ oligomers. Efforts have been
Although Aβ is an important neurotoxin in the etiology of
AD, the heterogeneous nature of Aβ assemblies has made it
difficult to study particular toxic Aβ oligomers individually. An
early hypothesis of Aβ toxicity stated that the energetically
stable, insoluble Aβ fibrils were the primary pathologic
species.12 However, a large body of evidence demonstrated
poor correlation between fibril content and disease severity,
thus potentially contradicting the original amyloid fibril hypo-
thesis.13−16 Instead, evidence connecting synaptotoxicity and
Received: September 2, 2011
Published: March 15, 2012
© 2012 American Chemical Society
3002
dx.doi.org/10.1021/jm201332p | J. Med. Chem. 2012, 55, 3002−3010

Journal of Medicinal Chemistry
Article
made to stabilize low-molecular-weight oligomers, but to date,
high-resolution structures of Aβ oligomers are not available. To
gain insight into the structural features of Aβ oligomers and the
mechanisms by which they cause neurotoxicity, more
information is needed.
Toward this end, we have previously reported the results of a
Markov state model molecular dynamics (MSM-MD) simu-
lation of four C-terminal fragments of Aβ42 at physiological
concentrations.21 Because of the strong dependence of
aggregation kinetics on the exact amino acid sequence of the
C-terminus, C-terminal fragments of Aβ21−43 were used in the
simulation. The study revealed that Aβ trimers were a semi-
stable state in the aggregation pathway, in which monomers
contained a β-turn in the C-terminus. Similar structural motifs
containing a C-terminal β-turn had been previously predicted in
coarse-grained simulations.22 On the basis of these results, a
Gly37→D-Pro substitution was predicted to stabilize the β-turn.
Given that this turn is not seen in any proposed structure of Aβ
fibrils, it was hypothesized that the Gly37→D-Pro substitution
would stabilize soluble forms of Aβ42.
Recently, we have reported experimental studies of the
structure and behavior of an Aβ analogue containing both
Gly37→D-Pro and Met35→Nle (norleucine) substitutions
([Nle35,D-Pro37]Aβ42.23 The Nle substitution was included to
preclude oxidation of Met35, which can impact assembly
kinetics and toxicity.24 Solution-state NMR studies of [Nle35,
D-Pro37]Aβ42 showed that this analogue indeed formed a
C-terminus β-turn centered around residues 37−39. Additional
evidence from atomic force microscopy (AFM) and circular
dichroism (CD) spectroscopy suggested that [Nle35,D-Pro37]-
Aβ42 did not form fibrils and instead was stable in soluble
oligomeric forms comprising 3−7 monomers. Here, we report
further characterization of the behavior of [Nle35,D-Pro37]Aβ42
and its effect on the aggregation of WT Aβ42. In addition,
intrigued by the results of these studies, we used ligand-based
drug design to identify a small molecule capable of recap-
itulating the behavior of [Nle35,D-Pro37]Aβ42. In doing so, we
have identified and characterized novel modulators of the Aβ42
aggregation pathway capable of stabilizing Aβ42 oligomers that
may be used for detailed structural and mechanistic
investigations of their interactions with their cellular targets.
Our studies demonstrate the utilization of molecular simula-
tions to guide experiment in developing small-molecule inhibi-
tors targeting toxic amyloidogenic protein oligomers.
Figure 1. ThT fluorescence experiments of mixtures incubated for four
days containing [Nle35,D-Pro37]Aβ42 and Aβ42. Mixtures contained
25 μM Aβ42 with [Nle35,D-Pro37]Aβ42 at 6.25 μM in the 4:1 mixture
and 25 μM in the 1:1 mixture. Samples were incubated at 37 °C for
the duration of the experiment.
To further evaluate the effect of the Gly37→D-Pro sub-
stitution on fibril formation, peptide preparations were imaged
using transmission electron microscopy (TEM) at the same
time point as the ThT fluorescence measurement. Aβ42 showed
extensive fibril formation with codeposition of amorphous
aggregates (Figure 2A,B). In contrast, in identically prepared
solutions of [Nle35,D-Pro37]Aβ42, we observed only amorphous
aggregates and no fibrils (Figure 2C,D). Additionally, [Nle35,
D-Pro37]Aβ42 samples aggregated for up to 6 months showed no
fibril formation (data not shown).
Our previous simulations had indicated that, apart from the
C-terminal turn, [Nle35,D-Pro37]Aβ42 is largely disordered in
solution and should not form stable secondary structure
elements. To assess this, [Nle35,D-Pro37]Aβ42 was analyzed by
circular dichroism (CD). At time zero, both WT Aβ42 and
[Nle35,D-Pro37]Aβ42 exhibited spectra characteristic of a
statistical coil, with a minimum at 197 nm (Figure 3). The
spectrum of WT Aβ42 changed gradually over the course of the
five-day experiment to develop a mimimum at 217 nm and a
maximum at 196 nm, indicative of β-sheet. In contrast, the
spectrum of [Nle35,D-Pro37]Aβ42 exhibited almost no change
over five days.
Finally, to investigate the effect of [Nle35,D-Pro37]Aβ42 on
equilibrium mixtures of freshly prepared Aβ42, we conducted
PICUP experiments, fractionated the cross-linked oligomers
using SDS-PAGE, and visualized them by silver staining as
previously reported (Figure 4A).25 Quantitative analysis of the
gels with ImageJ indicated a 70% increase in the relative
abundance of the trimer state for [Nle35,D-Pro37]Aβ42 compared
with WT Aβ42 (Figure 4B). Concomitant with the increase in
trimer abundance, we observed a decrease in monomer and
tetramer in [Nle35,D-Pro37]Aβ42 relative to WT Aβ42, consistent
the prediction that the Gly37→D-Pro substitution stabilizes
Aβ42 trimer.
As a control, we examined an analogue with an identical
amino acid composition to [Nle35,D-Pro37]Aβ42 but with seque-
nce positions randomly assigned (Aβscramble). In both turbidity
and Thioflavin T experiments, Aβscramble aggregated substan-
tially more rapidly than WT Aβ42 (Supporting Information
Figure 3). This result demonstrates that any D-proline
RESULTS
■
[Nle35,D-Pro37]Aβ42 Forms Stable Oligomers, Not
Fibrils, In Solution. In a previous computational study,
Kelley et al. hypothesized that a turn-promoting substitution
such as Gly37→D-Pro would stabilize trimeric Aβ oligomers.21
To test this hypothesis, an Aβ42 analogue with the Gly37→D-Pro
substitution was synthesized. An additional Met35→Nle
substitution was introduced to avoid oxidation of the native
methionine at position 35 during assay. Thioflavin T
experiments, in which fluorescence of the dye is proportional
to the extent of β-sheet character of the protein aggregates,
were performed to assess the extent of Aβ aggregation (Figure 1).
Following four days of incubation, the [Nle35,D-Pro37]Aβ42
samples showed negligible fluorescence whereas substantial
fluorescence was measured for WT Aβ42, indicating negligible
β-sheet character in the substituted peptide. This result is
consistent with our predictions that a Gly37→D-Pro substitution
stabilizes oligomers.
3003
dx.doi.org/10.1021/jm201332p | J. Med. Chem. 2012, 55, 3002−3010

Journal of Medicinal Chemistry
Article
Figure 2. Morphological analysis of mixtures of [Nle35,D-Pro37]Aβ42 and Aβ42 after a four-day incubation by TEM. Solutions containing 25 μM Aβ42,
(A) and (B), 25 μM [Nle35,D-Pro37]Aβ42, (C) and (D), or a 1:1 mixture of 25 μM [Nle35,D-Pro37]Aβ42 and 25 μM Aβ42, (E) and (F), were incubated
at 37 °C until immediately prior to imaging. After incubation, samples were transferred to carbon-coated Formvar grids and stained with uranyl
acetate.
Information Figure 4). These results agree with other PICUP
studies which demonstrate that a Met35→Nle substitution
slightly destabilizes the trimer state and is otherwise dissimilar
from the distribution of [Nle35,D-Pro37]Aβ42.26
[Nle35,D-Pro37]Aβ42 Reduces Aggregation in Mixtures
with WT Aβ42 in a Dose-Dependent Manner. Next, we
asked whether [Nle35,D-Pro37]Aβ42 could interact with and
affect the aggregation of WT Aβ42. To answer this question, we
measured ThT fluorescence in solutions of 25 μM WT Aβ42
mixed with 25 or 6.25 μM [Nle35,D-Pro37]Aβ42 (1:1 or 4:1
Aβ42/[Nle35,D-Pro37]Aβ42 concentration ratio, respectively).
Following four-day incubation experiments, the fluorescence
of the mixtures was intermediate between the values observed
for each peptide alone and inversely proportional to [Nle35,
D-Pro37]Aβ42 concentration (Figure 1). The fluorescence signal
Figure 3. Secondary structure analysis of [Nle35,D-Pro37]Aβ42 and Aβ42
by CD spectroscopy. CD spectra of [Nle35,D-Pro37]Aβ42 and Aβ42
taken immediately after dissolution of the monomer (light-blue and
pink, respectively), and after five-day incubation (dark-blue and red,
respectively). Samples of either [Nle35,D-Pro37]Aβ42 or Aβ42 (25 μM)
were incubated at 37 °C with continuous shaking for the duration of
the experiment.
of the 1:1 and 4:1 mixtures were 31.4 12.5% and 62.2 3.3%
that of WT Aβ42 alone, respectively.
Morphological examination by TEM of the [Nle35,D-Pro37]-
Aβ42:Aβ42 mixtures supported the ThT data (Figure 2E,F). At a
1:1 molar ratio, the extent of mature fibril formation was re-
duced and an increase in amorphous aggregated was observed.
Taken together with ThT data, the results suggest that
[Nle35,D-Pro37]Aβ42 interacts with WT Aβ42, decreases
formation of β-sheet-rich Aβ42 fibrils, and increases the relative
abundance of nonfibrillar assemblies.
substitution alone is not sufficient to replicate the effects of the
Gly37→D-Pro substitution on aggregation. Additionally, a seleno-
methionine substituted peptide [Sem35,D-Pro37]Aβ42 was
created for attempted X-ray studies. This peptide gave similar
ThT and PICUP results as [Nle35,D-Pro37]Aβ42, thus demon-
strating that the Met35→Nle substitution is not required to
reproduce the effects of [Nle35,D-Pro37]Aβ42(Supporting
Identification of a Small-Molecule Aβ42 Aggregation
Inhibitor. Intrigued by the effects of [Nle35,D-Pro37]Aβ42 on
aggregation, we conducted a ligand-based screen to identify
small molecules capable of recapitulating structural and chemi-
cal motifs of the substituted peptide. The previously reported
3004
dx.doi.org/10.1021/jm201332p | J. Med. Chem. 2012, 55, 3002−3010

Journal of Medicinal Chemistry
Article
Figure 4. PICUP reveals distinct oligomer size distributions for [Nle35,D-Pro37]Aβ42 and Aβ42 following cross-linking immediately after preparation
of fresh monomer. (A) SDS-PAGE of cross-linked samples of 25 μM Aβ42 and [Nle35,D-Pro37]Aβ42 (1 and 2, respectively), following silver staining.
Arrows to the left of the gel images indicate the identity of oligomer bands, as deduced from the molecular weight marker. (B) Densiometric analysis
of [Nle35,D-Pro37]Aβ42 and Aβ42 created using ImageJ. The data are the result of three independent experiments, each of which is well represented by
the images in (A).
NMR structure of [Nle35,D-Pro37]Aβ42 was stripped of side
chain atoms, except for the D-Pro, and used as the query for the
screen (Figure 5A). In doing so, molecules mimicking the
β-turn structure and recapitulating hydrogen-bonding interac-
tions were selected. The top 100 scoring hits from the Maybridge
Screening Database were purchased and investigated for their abil-
ity to reduce Aβ42 aggregation.
The compounds were screened in a single-dose ThT
aggregation assay (data not shown). Compounds identified as
potential hits were further evaluated in a dose−response ThT
aggregation assay. Of these compounds, 4-chloro-N-[2-[(E)-
(3,5-dichloro-2-hydroxyphenyl)methylideneamino]-4-methyl-
phenyl] benzenesulfonamide (1) demonstrated reproducible,
dose-dependent decreases in fluorescence (Figure 5B,C). 1 was
found to inhibit WT Aβ42 aggregation as read by ThT fluore-
scence with an IC50 of 13 μM.
Characterization of the Effects of 1 on WT Aβ42
Aggregation. We then evaluated the effect of 1 on the
secondary structure of aggregated Aβ42 by CD. In initial scans,
identically prepared solutions of WT Aβ42 and a 1:1 ratio of
drug/Aβ42 showed a minimum absorbance around 197 nm and
were otherwise characteristic of a statistical coil conformation
(Figure 6A). Over the course of the five-day aggregation, the
WT Aβ42 spectra shifted to have an absorbance maximum
around 196 nm and minimum at 217 nm, indicating the
transition of the WT peptide from a statistical coil to β-sheet
rich conformation. In comparison, the CD spectrum of the 1:1
solution did not change over the course of the experiment and
remained consistent with a random coil structure.
To supplement our data on the effect of 1 on Aβ42 fibril
formation, we investigated the formation of fibrils by TEM in
solutions of the compound at a 1:1 ratio to Aβ42. As reported
Figure 5. (A) Query used in the ligand-based drug design screen to
above, WT Aβ42 forms mature fibrils over the course of a four-
identify 1, as visualized in PyMol. Residues 35−40 of the NMR
day aggregation experiment (Figure 4A,B). However, in the 1:1
structure of [Nle35,D-Pro37]Aβ42 were stripped of side chain atoms
mixture, no fibrils were observed and only amorphous aggre-
except for the D-Pro residue. Carbons are shown in purple, oxygen in
gates were seen (Figure 6B). In comparison with [Nle35,
red, and nitrogen in blue. (B) Dose−response effect of 1 on Aβ42
D-Pro37]Aβ42, which did not completely block fibril formation
aggregation as measured by ThT fluorescence. Aβ42 was incubated at a
at this concentration ratio, 1 was more effective than [Nle35,
concentration of 25 μM in the presence of the indicated
D-Pro37]Aβ42 at inhibiting fibril growth.
concentrations of 1. (C) Structure of 1.
3005
dx.doi.org/10.1021/jm201332p | J. Med. Chem. 2012, 55, 3002−3010

Journal of Medicinal Chemistry
Article
Figure 6. Analysis of the effect of 1 on the fibril and secondary structure formation of Aβ42. (A) TEM image of a 1:1 mixture of 25 μM 1 and 25 μM
Aβ42. Samples were incubated for four days at 37 °C. After incubation, samples were transferred to carbon-coated Formvar grids and stained with
uranyl acetate. (B) CD spectra of 25 μM Aβ42 and a 1:1 mixture of 25 μM 1 and 25 μM Aβ42 incubated for five days at 37 °C. Spectra for Aβ42 alone
and the 1:1 mixture were obtained at t = 0d (light-blue and pink, respectively), and after a five-day incubation (dark-blue and red, respectively).
Neurotoxicity of [Nle35,D-Pro37]Aβ42 and 1. We then
examined the effect of [Nle35,D-Pro37]Aβ42 and 1 on cell
viability using the (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenylte-
trazolium bromide (MTT) reduction assay in differentiated
PC-12 cells. Cells exposed to 5, 10, or 20 μM [Nle35,D-Pro37]-
Aβ42 demonstrated a dose-dependent decrease in cell viability
(Figure 7). The toxicity was weaker than that of Aβ42, which led
identified as 1−20 (mass spectra and NMR available in the
Supporting Information). Either 1 or 1−20 undergoes hydro-
lysis of the imine to yield the degredation products 2 or 2−20,
respectively, and 3. Unexpectedly, mass chromatograms and
UV spectra of samples incubated up to 24 h were identical
within error, indicating that the equilibrium does not trend
toward any product on time scales relevant to these experi-
ments. When compound 2 was analyzed alone with the same
protocol, the product 2−20 was not identified, suggesting
that the dechlorination/epoxidation occurs only in the parent
compound, 1.
To assess the activity of the fragments relative to compound 1,
we obtained 2 and 3 to evaluate in ThT assays. While
compound 3 was found to be inactive (data not shown),
compound 2 exhibited a dose-dependent inhibition of Aβ42
aggregation (Figure 9). Interestingly, the inhibition of Aβ42 by 2
exhibited notably different kinetics in comparison to 1. At t = 0 d
(Figure 9A), the IC50 of 1 as measured by percent of
flourescence of an untreated control was 62 μM and the IC50
of 2 was 30 μM. Over the course of the four-day aggregation,
1 increased in efficacy in comparison to control with an IC50 of
13 μM at t = 4 d (Figure 9B). Conversely, compound 2 showed
similar efficacy after the aggregation period, with an IC50 of
23 μM at t = 4 d.
Figure 7. Results of MTT reduction assay to assess the effect of
[Nle35,D-Pro37]Aβ42 or 1 on the viability of PC-12 cells in the presence
and absence of 10 μM Aβ42. Assay was performed after maintaining
cells with Aβ42 mixtures for 48 h.
DISCUSSION
■
In our effort to design a stable oligomer of Aβ42, we have
predicted21 that the substitution Gly37→D-Pro would result in
enhanced β-hairpin formation and increased stability of low-
order oligomers, particularly trimers. Here, our PICUP results
confirmed these computational predictions. demonstrating a
70.7 5.5% increase in trimer abundance of [Nle35,D-Pro37]-
Aβ42 compared to Aβ42 (Figure 4). The CD, Thioflavin T, and
TEM experiments demonstrated that the Gly37→D-Pro
substitution prevented formation of β-sheet-rich fibrils and
promoted formation of amorphous assemblies devoid of regular
secondary structure elements (Figures 1−3). The data suggest
that the β-turn stabilized by the Gly37→D-Pro mutation either
(a) stabilizes the small, soluble oligomers to such a degree as
to prohibit further aggregation into fibrils and/or (b) the
to a 45% decrease in viability at 10 μM, while [Nle35,D-Pro37]-
Aβ42 showed only 14% decrease in viability at this concen-
tration. 1 also showed a dose-dependent increase in toxicity,
reducing cell viability by 40% at 50 μM. (Figure 6)
Stability Assessment of 1 and Identification of Active
Fragment. Given the intriguing effects of 1 on Aβ42, we sought
to verify its stability under experimental conditions by LC-MS.
We found that upon entering an aqueous environment 1 re-
ached an equilibrium with several degredation products (Figure 8).
Dechlorination of 1 at the para chlroine, followed by epoxida-
tion, yields a product with a mass 20 g/mol lighter than 1,
3006
dx.doi.org/10.1021/jm201332p | J. Med. Chem. 2012, 55, 3002−3010

Journal of Medicinal Chemistry
Article
Figure 8. Equilibrium reaction diagram of the products created upon dissolution of 1 in aqueous buffer, pH = 7.4. Dechlorination of 1 at the para
position yields product 1−20. Hydrolysis of the imine bond of 1 yields 2, and analogous hydrolysis of 1−20 yields 2−20, in addition to the aldehyde 3.
Figure 9. Results from ThT assays measuring the dose dependence of ThT flouresence in solutions of Aβ42 aggregated with the indicated
concentrations of 1 and 2. (A) Flouresence measured immediately after resuspending Aβ42 and combining with either 1 or 2. (B) Flouresence of
identical samples, measured after a four-day aggregation period.
conformation assumed by [Nle35,D-Pro37]Aβ42 represents a
conformation that is off pathway of fibril formation all together.
Interestingly, although other parts of Aβ are unaltered, most
notably, the central hydrophobic cluster (residues 17−21),
which is known to be important for β-sheet formation, the
Gly37→D-Pro substitution is sufficient for preventing formation
of β-sheets.
Importantly, for the development of our research on
[Nle35,D-Pro37]Aβ42 toward a lead compound for Alzheimer’s
disease, the change in aggregation induced by the Gly37→D-Pro
substitution is not limited to solutions containing only [Nle35,
D-Pro37]Aβ42 but is observed in mixtures of [Nle35,D-Pro37]Aβ42
and WT Aβ42 as well. The results from ThT and TEM analysis
of mixtures of [Nle35,D-Pro37]Aβ42 and WT Aβ42 clearly
3007
dx.doi.org/10.1021/jm201332p | J. Med. Chem. 2012, 55, 3002−3010

Journal of Medicinal Chemistry
Article
indicate that [Nle35,D-Pro37]Aβ42 interacts with WT Aβ42 to
reduce the extent of aggregation and fibril formation when
compared to Aβ42 alone (Figures 1,3). At a 1:1 ratio, aggre-
gation as read by ThT was reduced by 60% when compared to
WT peptide alone. In TEM images, fibrils were reduced in the
solution containing a mixture of [Nle35,D-Pro37]Aβ42 and Aβ42.
We further attempted to recapitulate the effects of [Nle35,
D-Pro37]Aβ42 on Aβ42 aggregation with a small molecule.
Structural and chemical motifs from the NMR structure of
[Nle35,D-Pro37]Aβ42 were used in a ligand-based drug discovery
effort. Using this approach, we discovered 1 and found that it
inhibited Aβ42 β-sheet formation with an IC50 of 13 μM over a
four-day aggregation (Figure 5). Results from CD and TEM
experiments were consistent with the ThT data and indicated
that 1 completely inhibited Aβ42 fibril and β-sheet formation at
a 1:1 molar ratio (Figure 6). Considering the results of the LC-
MS stability assessment in combination with the ThT studies
with compound 2, the substructure represented by 2 clearly is
the primary fragment of 1 responsible for inhibition of Aβ42.
Although 3 does not inhibit aggregation on its own, the
addition of fragment 3 to fragment 2 (yielding 1) significantly
alters the kinetics of the aggregation in comparison to 2 alone.
The data suggest that 2 has a faster rate of association with Aβ42
in comparison to 1.
MATERIALS AND METHODS
■
Sample Preparation. WT Aβ42 was purchased from a single
production lot from Bachem (Bachem H-6466 lot 1013710). Synthetic
peptide [Nle35,D-Pro37]Aβ42 (DAEFRHDSGY10EVHHQKLVFF20-
AEDVGSNKGA30IIGLNLVDPGVV40IA) was synthesized by Anaspec
(San Jose, CA). Because the work was carried out in two different
laboratories, with slightly different protocols, two sample preparations
exist. Peptides were prepared for all experiments in Eppendorf Protein
Lo-Bind tubes (Eppendorf no. 022431081). For Thioflavin T studies,
the entire contents of the 1 mg Aβ42 peptide vial was dissolved in
1,1,1,3,3,3-hexafluoro-2-propanol (HFIP) (Sigma no. 325244) to a
concentration of 1 mg/mL and incubated for 2 h at room temperature.
After aliquoting the solution, samples were covered with Parafilm
in which holes were poked. The solvent was evaporated using a
SpeedVac for 1 h and the dry peptide films stored at −80 °C.
Immediately before experiments, samples were dissolved in DMSO
(Sigma no. D2438), Biotechnology Performance grade) at 2 mM and
sonicated for 5 min. The peptide solutions were diluted with PBS, pH
7.4 (GIBCO no. 10010) to the desired concentration. Final DMSO
concentration was kept below 3%. For all other experiments, peptides
were prepared as described previously.28 Dried peptide films were
dissolved by adding sequentially 10% of the final volume 60 mM
NaOH, 45% water, and 45% 20 mM NaH2PO4 buffer. After dissolu-
tion in 60 mM NaOH, peptides were sonicated for 1 min in a bath
sonicator. Peptide concentration was measured by Bradford protein
assay (Bio-Rad) or quantitative amino acid analysis and was generally
70−80% of the nominal concentration. Peptide mixtures were pre-
pared immediately prior to the beginning of experiments.
Compounds were purchased from the Maybridge Screening
Collection (Maybridge Chemical Company, Cornwall, UK) and Sigma.
Compounds were dissolved in DMSO to 10 mM and then diluted
with PBS to the desired concentration. Purity of the compounds was
assessed by NMR, and the degredation of the parent compound was
investigated by LC-MS. For MTT and CD experiments, which are
sensitive to the presence of DMSO, 1 was dissolved to 10 mM in
NaOH and then diluted further with PBS.
PICUP/SDS-PAGE. Photo-induced cross-linking of unmodified
proteins (PICUP) was performed as described previously.25 Briefly,
1 μL of 40 mM ammonium persulfate (APS, Aldrich) and 1 μL of
2 mM tris(2,2′-bipyridyl)dichlororuthenium(II) (Ru(Bpy)) (Aldrich)
were added to 18 μL of 25 μM WT Aβ42 sample in a clear PCR tube.
Irradiation was carried out for 1 s. The cross-linking reaction is
quenched immediately by adding 1 μL of 1 M DTT and 10 μL of
Tricine sample buffer (Invitrogen). Samples are then analyzed using
1 mm thick, 10−20% Tris-Tricine gradient gels (Invitrogen) and silver
stained (SilverXpress, Invitrogen). Following air drying, gels were
analyzed using ImageJ. Reported values are from averaging of three
separate gels.
Both [Nle35,D-Pro37]Aβ42 and 1 were found to cause a dose-
dependent decrease in cell viability. Notably, [Nle35,D-Pro37]-
Aβ42 is significantly less toxic than WT Aβ42 (Figure 7).
Recently, O’Nuallain et al. found that freshly prepared dimers
of [Cys26]Aβ42 did not block long-term potentiation (LTP) but
that protofibrils built from these dimers did reduce LTP.27
Similarly, our results suggest that the trimer stabilized by
[Nle35,D-Pro37]Aβ42 is a minimally toxic species, in comparison
to WT Aβ42. Given the results, it is hypothesized that [Nle35,
D-Pro37]Aβ42 and 1 reduce Aβ42 aggregation by either stabilizing
soluble oligomers to such an extent that further aggregation is
prohibited and/or by stabilizing a conformation of Aβ42 that is
off pathway altogether of protofibril and fibril formation.
CONCLUSIONS
■
We designed [Nle35,D-Pro37]Aβ42 based on a structural motif
observed in the structure of the Aβ42 trimer predicted from
molecular dynamics simulations. Recently, we have reported
solution NMR structures confirming the existence of the
predicted β-hairpin, and the data herein demonstrate a 70%
increase in the relative population of the trimer in freshly
prepared solutions. The use of computation to predict the
structure of a stable species in a heterogeneous aggregation
pathway and guide the development of an experimental investi-
gation is notable and suggests a methodology for developing
aggregation inhibitors for other amyloid-related diseases, which
are complicated due to a lack of structural information. Both
[Nle35,D-Pro37]Aβ42 and the compounds identified based on its
structure, 1 and 2, are able to interact in solution with Aβ42 to
stabilize soluble oligomers and reduce fibril formation. The shift
in oligomer distribution induced by [Nle35,D-Pro37]Aβ42
correlated to a reduction in toxicity compared to Aβ42. Given
the evidence herein that [Nle35,D-Pro37]Aβ42, 1, and 2 inhibit
Aβ42 aggregation and β-sheet formation, we hope to further
exploit their structural motifs to develop more efficacious and
less toxic lead compounds.
Circular Dichroism. Samples of 25 μM WT peptide, in the
absence or presence of [Nle35,D-Pro37]Aβ42 or 1 were incubated at
37 °C with continuous agitation using an orbital shaker at 200 rpm.
Spectra were recorded every 24 h during 5 days using a J-810
spectropolarimeter (Jasco, Easton, MD) equipped with a thermostable
sample cell at 37 °C using 1 mm path-length cuvettes. Spectra were
collected from 190 to 260 nm with 1 s response time, 100 nm/min
scan speed, 0.2 nm resolution, and 1 nm bandwidth. Spectra were
averaged after background subtraction.
Thioflavin T Fluoresence. Samples were incubated at 37 °C for
four days in triplicates with shaking. Then 10 μL samples were added
in black, flat-, and clear-bottom 96-well plates (Corning no. 3631) to
190 μL of 25 μM Thioflavin T (Sigma no. T3516) in 50 mM glycine,
pH 8 (Fisher no. G48−212). After a 5 min incubation with the dye,
the fluorescence of ThT was measured using a Spectromax M5
fluorometer (Molecular Devices) at an excitation wavelength of 446
nm and emission at 490 nm.
Transmission Electron Microscopy. Samples of Aβ42 were
incubated in the absence or presence of [Nle35,D-Pro37]Aβ42 or 1 for
four days at 37 °C. Aliquots of 7 μL were spotted onto 400 mesh,
glow-discharged, carbon-coated Formvar grids (Electron Microscopy
3008
dx.doi.org/10.1021/jm201332p | J. Med. Chem. 2012, 55, 3002−3010

Journal of Medicinal Chemistry
Article
Science, Hatfield, PA) for 2 min and stained with 7 μL of 1% uranyl
acetate. The samples were analyzed using a transmission electron
microscope JEM1200-EX (JEOL).
Transport Physics with Uncertainty Quantification) for com-
puting resources.
MTT Assay. MTT assays were conducted as previously reported.26
Rat pheochromocytoma (PC-12) cells were maintained in F-12
nutrient mixture with Kaighn’s modification (F-12K) with 15% heat-
inactivated horse serum and 2.5% FBS at 37 °C in an atmosphere of
5% CO2. For cell viability assays, cells were plated in 96-well plates at a
density of 30000 cells per well in differentiation media (F-12K, 0.5%
FBS, 100 μM nerve growth factor) and maintained for 48 h.
To assess the biological activity of Aβ42, [Nle35,D-Pro37]Aβ42, and 1,
solutions were prepared as indicated above and diluted into RPMI
media to yield concentrations 10 times higher than experimental
concentrations. Aliquots of 10 μL were added to PC-12 cells to yield
final concentrations of 10 μM WT Aβ42 and incubated for 15 h. Cell
viability was assessed qualitatively by visual observation and
quantitatively by the CellTiter 96 nonradioactive cell proliferation
assay (Promega). Briefly, 15 μL of dye solution was incubated with the
cells for 3 h. Then 100 μL of solubilization/stop solution was added
and the plates were incubated overnight in the dark to ensure
complete solubilization. Plates were read by using a Synergy HT
microplate reader (BioTek), and the absorbance at 570 nm (formazan
product) minus the absorbance at 630 nm (background) was recorded.
At least three independent experiments with six replicates (n = 18)
were carried out, and the results were averaged.
Virtual Screening. A representative structure from the previously
reported NMR studies into the [Nle35,D-Pro37]Aβ42 peptide was used
as a basis for the ligand-based screen. Residues 35−40 were extracted
from the full structure, and side chain atoms were deleted except for
the proline atoms. The Maybridge Screening Collection was used as a
screening database. Molecules were stripped of salts, and conformers
of the remaining molecules were created using Omega (version 2.3,
OpenEye Scientific Software). The screening database was then com-
pared to the modified peptide structure using the program ROCS
(version 2.3, OpenEye Scientific Software). The Explicit Mills Dean
force field was used, with the exception that ring interactions were
turned off. The top 100 compounds ranked by combo Tanimoto after
optimization by shape and color and were selected for evaluation by
ThT. Compound 1 was ranked 48th out of 56897 compounds scree-
ned (details available as Supporting Information).
ABBREVIATIONS USED
■
Aβ, amyloid β-protein; AD, Alzheimer’s disease; Nle,
norleucine; D-Pro, D-proline; ThT, Thioflavin T; PICUP,
photo-induced cross-linking of unmodified peptides
REFERENCES
■
(1) World Alzheimer Report 2010; Alzheimer’s Disease International:
London, 2010.
(2) Goate, A.; Chartier-Harlin, M. C.; Mullan, M.; Brown, J.;
Crawford, F.; Fidani, L.; Giuffra, L.; Haynes, A.; Irving, N.; James, L.
Segregation of a missense mutation in the amyloid precursor protein
gene with familial Alzheimer’s disease. Nature 1991, 349, 704−706.
(3) Van Broeckhoven, C.; Haan, J.; Bakker, E.; Hardy, J. A.; Van Hul,
W.; Wehnert, A.; Vegter-Van der Vlis, M.; Roos, R. A. Amyloid beta
protein precursor gene and hereditary cerebral hemorrhage with
amyloidosis (Dutch). Science 1990, 248, 1120−1122.
(4) Tomiyama, T.; Nagata, T.; Shimada, H.; Teraoka, R.; Fukushima,
A.; Kanemitsu, H.; Takuma, H.; Kuwano, R.; Imagawa, M.; Ataka, S.;
Wada, Y.; Yoshioka, E.; Nishizaki, T.; Watanabe, Y.; Mori, H. A new
amyloid β variant favoring oligomerization in Alzheimer’s-type
dementia. Ann. Neurol. 2008, 63, 377−387.
(5) Chartier-Harlin, M. C.; Crawford, F.; Houlden, H.; Warren, A.;
Hughes, D.; Fidani, L.; Goate, A.; Rossor, M.; Roques, P.; Hardy, J.
Early-onset Alzheimer’s disease caused by mutations at codon 717 of
the beta-amyloid precursor protein gene. Nature 1991, 353, 844−846.
(6) Pike, C. J.; Walencewicz, A. J.; Glabe, C. G.; Cotman, C. W. In
vitro aging of beta-amyloid protein causes peptide aggregation and
neurotoxicity. Brain Res. 1991, 563, 311−314.
(7) Lambert, M. P.; Barlow, A. K.; Chromy, B. A.; Edwards, C.;
Freed, R.; Liosatos, M.; Morgan, T. E.; Rozovsky, I.; Trommer, B.;
Viola, K. L.; Wals, P.; Zhang, C.; Finch, C. E.; Krafft, G. A.; Klein, W. L.
Diffusible, nonfibrillar ligands derived from Abeta1−42 are potent
central nervous system neurotoxins. Proc. Natl. Acad. Sci. U.S.A. 1998,
95, 6448−6453.
(8) Haass, C.; Selkoe, D. J. Soluble protein oligomers in
neurodegeneration: lessons from the Alzheimer’s amyloid beta-
peptide. Nature Rev. Mol. Cell Biol. 2007, 8, 101−112.
(9) Klein, W. L.; Krafft, G. A.; Finch, C. E. Targeting small Abeta
oligomers: the solution to an Alzheimer’s disease conundrum? Trends
Neurosci. 2001, 24, 219−224.
ASSOCIATED CONTENT
■
S
* Supporting Information
Results from experiments conducted with Aβscramble and
[Sem35,D-Pro37]Aβ42. LC-MS and H NMR of 1, 2, and 3.
1
(10) Georganopoulou, D. G.; Chang, L.; Nam, J. M.; Thaxton, C. S.;
Mufson, E. J.; Klein, W. L.; Mirkin, C. A. Nanoparticle-based detection
in cerebral spinal fluid of a soluble pathogenic biomarker for
Alzheimer’s disease. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 2273−
2276.
Rankings and structures of top scoring compounds from the
virtual screen. This material is available free of charge via the
Internet at http://pubs.acs.org.
(11) Fukumoto, H.; Tokuda, T.; Kasai, T.; Ishigami, N.; Hidaka, H.;
Kondo, M.; Allsop, D.; Nakagawa, M. High-molecular-weight {beta}-
amyloid oligomers are elevated in cerebrospinal fluid of Alzheimer
patients. FASEB J. 2010, 24, 2716−2726.
AUTHOR INFORMATION
Corresponding Author
*Phone: (650)723-3660. E-mail: pnovick@stanford.edu.
Address: Department of Chemistry, Stanford University,
318 West Campus Drive, Stanford, California 94305.
■
(12) Hardy, J. A.; Higgins, G. A. Alzheimer’s disease: the amyloid
cascade hypothesis. Science 1992, 256, 184−185.
(13) Neve, R. L.; Robakis, N. K. Alzheimer’s disease: a re-
examination of the amyloid hypothesis. Trends Neurosci. 1998, 21,
15−19.
Notes
The authors declare no competing financial interest.
#Joint first authors.
(14) Giannakopoulos, P.; Herrmann, F. R.; Bussier
̀
e, T.; Bouras, C.;
Kovari, E.; Perl, D. P.; Morrison, J. H.; Gold, G.; Hof, P. R. Tangle and
̈
ACKNOWLEDGMENTS
neuron numbers, but not amyloid load, predict cognitive status in
Alzheimer’s disease. Neurology 2003, 60, 1495−1500.
■
We acknowledge the UCLA Jim Easton Consortium
Alzheimer’s Drug Discovery and Biomarker Development, the
NIH Nanomedicine Center (PN1 EY016525), and the
Stanford Center for Molecular Analysis and Design (CMAD)
for funding. We also acknowledge NSF award CNS-0619926
and no. 0960306 (MRI-R2: Acquisition of a Hybrid CPU/
GPU and Visualization Cluster for Multidisciplinary Studies in
(15) Chui, D. H.; Tanahashi, H.; Ozawa, K.; Ikeda, S.; Checler, F.;
Ueda, O.; Suzuki, H.; Araki, W.; Inoue, H.; Shirotani, K.; Takahashi,
K.; Gallyas, F.; Tabira, T. Transgenic mice with Alzheimer presenilin 1
mutations show accelerated neurodegeneration without amyloid
plaque formation. Nature Med. 1999, 5, 560−564.
(16) Mucke, L.; Masliah, E.; Yu, G. Q.; Mallory, M.; Rockenstein, E.
M.; Tatsuno, G.; Hu, K.; Kholodenko, D.; Johnson-Wood, K.;
3009
dx.doi.org/10.1021/jm201332p | J. Med. Chem. 2012, 55, 3002−3010

Journal of Medicinal Chemistry
Article
McConlogue, L. High-level neuronal expression of abeta 1−42 in wild-
type human amyloid protein precursor transgenic mice: synaptotox-
icity without plaque formation. J. Neurosci. 2000, 20, 4050−4058.
(17) Watson, D.; Castano, E.; Kokjohn, T. A.; Kuo, Y. M.;
̃
Lyubchenko, Y.; Pinsky, D.; Connolly, E. S.; Esh, C.; Luehrs, D. C.;
Stine, W. B.; Rowse, L. M.; Emmerling, M. R.; Roher, A. E.
Physicochemical characteristics of soluble oligomeric Abeta and their
pathologic role in Alzheimer’s disease. Neurol. Res. 2005, 27, 869−881.
(18) De Felice, F. G.; Wu, D.; Lambert, M. P.; Fernandez, S. J.;
Velasco, P. T.; Lacor, P. N.; Bigio, E. H.; Jerecic, J.; Acton, P. J.;
Shughrue, P. J.; Chen-Dodson, E.; Kinney, G. G.; Klein, W. L.
Alzheimer’s disease-type neuronal tau hyperphosphorylation induced
by A beta oligomers. Neurobiol. Aging 2008, 29, 1334−1347.
(19) Bitan, G.; Lomakin, A.; Teplow, D. B. Amyloid beta-protein
oligomerization: prenucleation interactions revealed by photo-induced
cross-linking of unmodified proteins. J. Biol. Chem. 2001, 276, 35176−
35184.
(20) Ono, K.; Condron, M. M.; Teplow, D. Structure−neurotoxicity
relationships of amyloid beta-protein oligomers. Proc. Natl. Acad. Sci.
U.S.A. 2009, 106, 14745−14750.
(21) Kelley, N.; Vishal, V.; Krafft, G.; Pande, V. Simulating
oligomerization at experimental concentrations and long timescales:
A Markov state model approach. J. Chem. Phys. 2008, 129, 214707.
(22) Urbanc, B.; Cruz, L.; Yun, S.; Buldyrev, S. V.; Bitan, G.; Teplow,
D. B.; Stanley, H. E. In silico study of amyloid beta-protein folding and
oligomerization. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 17345−17350.
(23) Rajadas, J.; Liu, C. W.; Kelley, N. W.; Novick, P. A.;
Inayathullah, M.; LeMieux, M. C.; Pande, V. S. Rationally designed
turn promoting mutation in the amyloid-β peptide sequence stabilizes
oligomers in solution. PLOS One 2011, 6, e21776 DOI: 10.1371/
journal.pone.0021776.
(24) Maiti, P.; Piacentini, R.; Ripoli, C.; Grassi, C.; Bitan, G.
Surprising toxicity and assembly behaviour of amyloid β-protein
oxidized to sulfone. Biochem. J. 2011, 433, 323−332.
(25) Bitan, G.; Teplow, D. Rapid photochemical cross-linkinga
new tool for studies of metastable, amyloidogenic protein assemblies.
Acc. Chem. Res. 2004, 37, 357−364.
(26) Maiti, P.; Lomakin, A.; Benedek, G. B.; Bitan, G. Despite its role
in assembly, methionine 35 is not necessary for amyloid β-protein
toxicity. J. Neurochem. 2010, 11.
(27) O’nuallain, B.; Freir, D.; Nicoll, A.; Risse, E.; Ferguson, N.;
Herron, C.; Collinge, J.; Walsh, D. Amyloid β-Protein Dimers Rapidly
Form Stable Synaptotoxic Protofibrils. J. Neurosci. 2010, 30, 14411−
14419.
(28) Rahimi, F.; Maiti, P.; Bitan, G. Photo-induced cross-linking of
unmodified proteins (PICUP) applied to amyloidogenic peptides.
J. Vis. Exp. 2009, No. 23, 1071 DOI: 10.3791/1071.
3010
dx.doi.org/10.1021/jm201332p | J. Med. Chem. 2012, 55, 3002−3010