Cholesterol seco-sterol-induced aggregation of methylated amyloid-β peptides - Insights into aldehyde-initiated fibrillization of amyloid-β

Article abstract of DOI:10.1002/anie.200705922

(Chemical Equation Presented) Hot spot on amyloid-β? Atheronal-B-induced aggregation of amyloid-β (Aβ) involves a site-specific adduction of the aldehyde to the ε-amino group of Lys 16, suggesting that Lys 16 is a hot spot for atheronal-induced fibrilliza

Full text of DOI:10.1002/anie.200705922

Angewandte
Chemie
DOI: 10.1002/anie.200705922
Alzheimer’s Disease
Cholesterol seco-Sterol-Induced Aggregation of Methylated
Amyloid-b Peptides—Insights into Aldehyde-Initiated Fibrillization of
Amyloid-b**
Johanna C. Scheinost, Hong Wang, Grant E. Boldt, John Offer, and Paul Wentworth, Jr.*
The nucleation and aggregation of amyloid-b (Ab) peptides,
Ab(1–40) and Ab(1–42), into neurotoxic oligomers is consid-
ered a primary event in Alzheimerꢀs disease (AD) patho-
genesis.[1,2] The vast majority of this disease is sporadic in
origin (> 85%),[3] involving the oligomerization of native Ab
peptide oligomers, therefore, research is ongoing to classify
the in vivo environmental triggers of AD onset that facilitate
intracellular and extracellular nucleation, aggregation and
deposition of native Ab peptide.
As part of our ongoing research in this area,[4] we
discovered cholesterol seco-sterol aldehyde 1 (termed athe-
ronal-B) in vivo.[5] Aldehyde 1 is quantifiable in all human
plasma and central nervous system (CNS), but is significantly
elevated in the plasma and inflamed arteries of patients with
advanced atherosclerosis, and in the CNS of patients with
inflammatory neurological disease.[5,6] We have further shown
that adduction of 1 to apoB-100, a protein component of low-
density lipoprotein (LDL), causes this protein to misfold in
vitro,[5] a misfolding process that renders LDL particles
susceptible for uptake into macrophages.[7] In addition, we
have shown that 1 accelerates the aggregation of Ab(1–40)
and Ab(1–42) in vitro[8] hinting that 1 could be a plausible
Figure 1. Atheronal-B(1)-induced aggregation of Ab(1–40) peptide
(2a). a) Schiff-base equilibrium between Lys16 of Ab(1–40) (2a) and
aldehyde 1. Similar equilibria exist for Schiff-base formation at the e-
amino group of Lys28 and the a-amino group of Asp1. b) Amino acid
chemical factor linking the known AD association with
atherosclerosis.[9] An important and largely unanswered
question regarding lipid aldehyde-induced protein misfolding
is the nature of the interaction between the aldehyde and the
sequences of Ab(1–40) (2a) and N,N-dimethylamino-containing pep-
tide sequences (2b–f) synthesized for these studies. The central
protein and how this facilitates protein aggregation.[4] Herein
we show, by kinetic analyses of the atheronal-B (1)-induced
oligomerization and fibrillization of a panel of synthetic
mono-, bis- and tris-N,N-dimethylamine-containing Ab(1–40)
protein sequences 2b–f (Figure 1b), that the aggregation of
hydrophobic cluster (CHC) of Ab is underlined. K*=Ne,Ne-dimethyl
Lys.
Ab(1–40) peptide 2a is accelerated by 1 only when the
aldehyde adducts to the e-amino group of Lys16; no initiation
in oligomerization of 2a is observed when aldehyde 1 adducts
to either the e-amino group of Lys28 or the a-amino group
amine of Asp1. In addition, the atheronal-B-induced aggre-
gation of peptide Ab(1–40) 2a is inhibited by cholesterol.
Both data combine to suggest that the atheronal-B-induced
aggregation of Ab(1–40) involves a high degree of structural
recognition between the lipid and the peptide that involves, in
part, binding of 1 into the putative cholesterol-binding
domain of 2a.
[*] J. C. Scheinost, Dr. G. E. Boldt, Dr. J. Offer, Prof. P. Wentworth, Jr.
The Scripps-Oxford Laboratory
Dept. of Biochemistry, University of Oxford
South Parks Rd., Oxford OX13QU (UK)
Fax: (+44)1865-285-329
E-mail: paul.wentworth@bioch.ox.ac.uk
Homepage: http://www2.bioch.ox.ac.uk/wentworthlab
Dr. H. Wang, Prof. P. Wentworth, Jr.
Department of Chemistry and The Skaggs Institute for Chemical
Biology, The Scripps Research Institute
Atheronal-B (1) was synthesized as outlined previously.[5]
Peptides 2a–f were synthesized by Boc/benzyl solid phase
peptide synthesis (SPPS) using in situ neutralization.[10] Ne,Ne-
Dimethyl Lys (K*) substitutions were incorporated using
Boc-Lys(Me)2-CO2H. The Na,Na-dimethyl Asp group
required for peptides 2d and 2 f was incorporated using
Me2N-Asp(OcHx)-CO2H, prepared by reductive amination
10550 N. Torrey Pines Rd., La Jolla CA 92037 (USA)
[**] This work was supported by a grant from The Scripps Research
Institute. The authors thank R. P. Troseth (TSRI) for synthesis of 1,
Dr. M. Wood for EM images, and Dr. P. R. Antrobus (UO) for high
resolution ES-MS of peptides 2a–f.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2008, 47, 3919 –3922
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3919

Communications
of H2N-Asp(OcHx)-CO2H (for full synthetic details see the
Supporting Information).
A comparison of the kinetic profile of aggregation of the
methylated peptides 2b–f relative to native 2a was deter-
mined from a timecourse of thioflavin T (ThT) fluorescence
of a deseeded peptide solution (100 mm) in sodium acetate
buffer (20 mm, pH 5.0) and NaCl (100 mm) by incubating with
standing at 378C (see Supporting Information for experimen-
tal details).[12] This analysis revealed that the methylated
peptides 2b–f have a classical sigmoidal aggregation profile,
indicative of a nucleation-dependent polymerization pro-
cess,[13,14] with all the peptides having a lag phase similar in
duration to that observed with native 2a, about 1 h (Figure S3
in the Supporting Information).
Timecourses of ThT binding and fluorescence of peptides
2a–f (100 mm) in the presence or absence of atheronal-B
(100 mm) were then measured (Figure 3a–f) using a modifi-
cation of our previous method.[8] In brief, peptides 2a–f were
deseeded in hexafluoroisopropanol (HFIP),[15] dissolved in
dimethylsulfoxide (DMSO) and added into buffer (PBS,
pH 7.4). The aggregation was initiated by addition of
aldehyde 1 in isopropyl alcohol (IPA) and then proceeded
quiescently at 378C. Aliquots were removed periodically,
added to a ThT solution and the fluorescence of the ThT
solution was measured. The concentration of peptide 2a–f
(100 mm) was selected such that no measurable oligomeriza-
tion of protein in the absence of 1 would occur (Figure 3a–f).
Incubation of atheronal-B (1) with peptide 2a leads to
rapid formation of ThT-positive aggregates (time to half
maximal fluorescence, t50 = 14 h, Figure 3a) in a process that
is thermodynamically favoured from the outset,[8,16] a so-
called “downhill polymerization”, that has no measurable lag
phase (Figure 3a).[17] Aldehyde 1 also generates ThT-positive
aggregates when incubated with the N,N-dimethylated pep-
tides 2c (K*28, t50 = 25 h, Figure 3c) and 2d (Me2N-D1, t50 =
13 h, Figure 3d) in a manner similar to that of 2a; thus, the
ThT profile in each case has no lag phase, with oligomeriza-
tion proceeding immediately. In contrast, aldehyde 1 does not
initiate the formation of ThT-positive aggregates when
incubated with N,N-dimethylated peptides 2b (K*16, Fig-
ure 3b), 2e (K*16, K*28, Figure 3e) and 2 f (Me2N-D1, K*16,
K*28, Figure 3 f).
The Ab(1–40) peptide analogs (2b–f, Figure 1b) incorpo-
rate tertiary amines that replace the primary amines present
in native Ab(1–40) (2a) at specific loci along the peptide
chain. Tertiary amines are unable to form Schiff-base adducts
with aldehydes, and are isoelectronic with primary amines at
physiological pH (7.4), rendering them ideal for a study to
elucidate putative site-specific modification of peptide pri-
mary amines by aldehydes. However, N,N-dimethylation of
primary amine-containing peptides may lead to increased
basicity and hydrophobicity of the resulting peptide,[11]
coupled with an impaired ability to form H-bonds. Therefore,
the impact of the non-isostructural N,N-dimethylamino
modifications on oligomerization and fibrillization propensity
of peptides 2b–f was determined using a combination of TEM
(Figure 2 and Figure S2 in the Supporting Information) and
Figure 2. Representative TEM images of fibrils of N,N-dimethylated
amyloid-b peptides: a) 2b (K*16), b) 2c (K*28), and c) 2 f (Me2N-D1,
K*16, K*28).
thioflavin-T fluorescence (Figure S3). Fibrillization of the
methylated Ab peptide analogs 2b–f (100 mm) was induced by
incubation with shaking in phosphate buffered saline (PBS)
(pH 7.4) containing NaN3 (0.02% w/v) at 378C. TEM analysis
of the protein aggregates revealed that the methylated Ab
analogues (2b–f) form a classical network of non-branched
fibrillar aggregates, several micrometer in length, indistin-
guishable from native Ab(1–40) peptide 2a (Figure 2 and
Figure S2 in the Supporting Information).
Figure 3. Kinetics of atheronal-B-induced aggregation of amyloid-b peptides 2a–f. a–f) ThT analyses, ex: 440 nm and em: 485 nm reported as
mean ꢀSD. g–l) Far-UV CD analyses; mean residue ellipticity [V] at 217 nm reported as average of three scans of peptides 2a (c, wild-type);
2b (c, K*16); 2c (c, K*28); 2d (c, Me2N-D1); 2e (c, K*16, K*28); 2 f (c, Me2N-D1, K*16, K*28). In each case, the peptide
&
(100 mm) is incubated quiescently in the presence ( ) or absence (&) of aldehdye 1 (100 mm) at 378C.
3920 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 3919 –3922

Angewandte
Chemie
We next investigated the secondary structure character-
istics during atheronal-B-induced fibrillization of peptides
2a–f spectroscopically using far-UV circular dichroism (CD).
For the CD analyses, the deseeding protocol was modified
from that used in the ThT assay to remove the DMSO co-
solvent. Thus, a typical aggregation involved deseeding in
urea (8m) and glycine/NaOH (10 mm, pH 10). This stock
solution was then diluted into PBS (final pH 7.4) containing
NaN3 (0.02% w/v) in the presence or absence of 1 (100 mm) in
IPA (0.5% v/v) and the aggregation mixtures were then
incubated quiescently at 378C. Fibrillization was followed by
the time-dependent change in the mean residue ellipticity [V]
at 217 nm (the wavelength minimum for b-strand) (Figure 3).
All the peptides 2a–f were random coil (RC) at t = 0
(Figure 3g–l) and in the absence of 1, all the peptides
remained in this form until ca. day 6 when the RC!b-
strand transition started to occur; the classic conformational
change that occurs during fibrillization of Ab peptides (for
full CD spectra see Figure S5).[18]
toxic species in AD,[1] is sensitive not only to the physical
environment (quiescent or shaking) but also the chemical
environment in which the aggregation is occurring.
The data presented here adds further support to our
original hypothesis, that covalent modification of Ab-peptides
by lipid aldehyde 1 is a key event that facilitates fibrillization.
This is emphasized with peptide 2 f, that has no free primary
amines with which to form a Schiff base with aldehyde 1, and
whose aggregation is not initiated with atheronal-B (Fig-
ure 3l).
However, the most significant result to come from this
work is the clear observation that the fibrillization of Ab-
peptides 2a–f is only accelerated by atheronal-B (1), if the e-
amino group of Lys16 is available for covalent modification.
This observation is the most dramatic in the case of peptide
2b (K*16) in which the e-amino group of Lys28 and the a-
amino group of Asp1 are both available for adduction, but
there is still no initiation in fibrillization by atheronal-B
(Figure 3b and h).
For all peptides 2a–f, the amount of b-strand increases
from day 6 until the plateau phase is reached by ca. day 12.
The time to half maximal [V] at 217 nm, t50V, for all the
peptides was very similar, ranging from 8–10 d.
In line with current thinking on how hydrophobic peptide
mutations contribute to peptide fibrillization through burial
of hydrophobic surface,[19,20] we had initially speculated that
covalent modification of the e-amino group of Lys16, Lys28
and the a-amino group of Asp1 with the hydrophobic
aldehyde 1 would be sufficient to trigger amyloidogenesis.
The hydrophobic effect may still impact amyloidogenesis
once aldehyde 1 is adducted to Ab, but what is clear is that
this process is specific to Lys16. Increasing local hydro-
phobicity at Lys28 and Asp1 by adduction of atheronal-B is
not sufficient to trigger fibrillization.
Lys16 sits at the N-terminus of the central hydrophobic
cluster (CHC) which has been suggested to be a cholesterol
binding domain of Ab (Figure 1b).[21] The binding of choles-
terol by Ab has been linked to a role of membrane
stabilization.[22,23] Given that atheronal-B (1) and cholesterol
share structural simile, it seems plausible that upon adduction
to Lys16 binding of the seco-sterol 1 in the CHC may occur.
We investigated such a hypothesis by studying the effect of
cholesterol on atheronal-B(1)-induced fibrillization of Ab(1–
40) (2a) (Figure 4). Cholesterol exhibits a concentration-
dependent reduction of the ability of atheronal 1 to induce
fibrillization of Ab(1–40) 2a, with an EC50 (effective concen-
tration that reduces the maximum ThT-positive aggregates to
50% of the untreated) of ca. 30 mm. This is the first example of
The kinetics of the far-UV CD of peptides 2a–f in the
presence of 1 supported the ThT fluorescence data (Fig-
ure 3g–l). Specifically, atheronal-B accelerates the onset of b-
strand formation when incubated with peptides 2a [t50V=
5.5 d, (8.0 d in the absence of 1), Figure 3 g], 2c [K*28, t50V=
6.0 d (9.5 d), Figure 3i] and 2d [Me2N-D1, t50V= 4.0 d (9.0 d),
Figure 3j]. In contrast, atheronal-B has no effect on the onset
of b-strand formation of peptides, 2b [K*16, t50V= 9 d (9 d),
Figure 3h], 2e [K*16, K*28, t50V= 9 d (9 d), Figure 3k] and
2 f [Me2N-D1, K*16, K*28, t50V= 7.5 d (8 d), Figure 3l].
During the CD analyses, the atheronal-induced aggrega-
tion of 2a, 2c, and 2d proceeds with a measurable lag phase,
indicative of a nucleated polymerization process (Figure 3g, i,
and j). This observation is in contrast to the ThT fluorescence
kinetic data, where ThT-positive aggregates start to form
immediately (Figure 3a, c, and d). In our previous reports of
lipid aldehyde 1 induced Ab(1–40) fibrillization, quiescent
aggregation protocols, such as those used throughout this
study, lead to a downhill polymerization with no lag phase.[8,16]
AFM analysis has revealed that quiescent aggregation Ab(1–
40) with 1 yields spherical aggregates, not fibrils.[8] However, if
agitation is applied during the incubation of atheronal 1 with
Ab(1–40) (2a), fibrils are formed in a two stage process that
does proceed through a nucleation process that has a
measurable lag phase.[16] To help answer the quandary as to
why with quiescent aggregation conditions we observe a lag
phase in the atheronal-induced aggregation of 2a, we
repeated the ThT fluorescence analysis of the aggregation
of 2a (100 mm) in the presence and absence of 1 (100 mm), but
using the deseeding method and buffer conditions employed
for the CD assay. This approach led to a profile of aldehyde-
induced ThT-positive aggregate generation indicative of a
nucleation polymerization with a measurable lag phase (see
Figure S4 in the Supporting Information). Clearly, the process
by which atheronal-B (1) is able to obviate Ab nucleation and
form spherical aggregates, considered to be the main neuro-
Figure 4. Cholesterol inhibits aldehyde 1-induced Ab(1–40) fibrilliza-
&
tion. ThT fluorescence of Ab(1–40) (2a) (100 mm) in the presence ( )
or absence (&) of 1 (100 mm) and cholesterol (0–50 mm) (PBS, pH 7.4
and 378C). Each point is the mean ꢀSD of duplicate measurements.
Angew. Chem. Int. Ed. 2008, 47, 3919 –3922
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3921

Communications
[2] W. L. Klein, W. B. Stine, Jr., D. B. Teplow, Neurobiol. Aging
2004, 25, 569.
inhibition of atheronal-induced Ab aggregation by a molecule
that does not trap the aldehyde of atheronal-B with a
nucleophile.[4,8,16] Cholesterol could be impacting aldehyde-
induced fibrillization of Ab by competing with atheronal-B
(1) for the CHC domain or alternatively, by binding in the
CHC domain cholesterol may block adduction of atheronal-B
to Lys16. Further studies are ongoing to unravel this effect,
but clearly there are molecular recognition events occurring
between atheronal-B and Ab that go beyond Schiff-base
formation with Lys16 and could well form the basis for the
development of inhibitors as potential therapeutics for
Alzheimerꢀs disease.
In summary, we have discovered that cholesterol seco-
sterol-1-induced aggregation of Ab involves a site-specific
adduction of the aldehyde to the e-amino group of Lys16,
suggesting that Lys16 is a hot spot for atheronal-induced
fibrillization of Ab. This process can be inhibited by
molecules that compete for the CHC binding domain.
Although in this report we have focused on the interaction
between 1 and 2a, the implications are more wide reaching
for protein misfolding and disease. We have previously shown
that lipid aldehydes cause the misfolding of other disease-
related proteins such as apoB-100[5] and a-synuclein.[6] This
study suggests that there may be specific hot spots on
misfolding prone proteins at which molecular recognition
events with specific lipid aldehydes may bind, and in such
cases this process may be amenable to competition and
inhibition by small molecules that may ultimately lead to new
therapeutic targets.
[3] F. Chiti, C. M. Dobson, Annu. Rev. Biochem. 2006, 75, 333.
[4] J. Bieschke, Q. Zhang, D. A. Bosco, R. A. Lerner, E. T. Powers,
P. Wentworth, J. W. Kelly, Acc. Chem. Res. 2006, 39, 611.
[5] P. Wentworth, Jr., J. Nieva, C. Takeuchi, R. Galve, A. D.
Wentworth, R. B. Dilley, G. A. DeLaria, A. Saven, B. M.
Babior, K. D. Janda, A. Eschenmoser, R. A. Lerner, Science
2003, 302, 1053.
[6] D. A. Bosco, D. M. Fowler, Q. Zhang, J. Nieva, E. T. Powers, P.
Wentworth, Jr., R. A. Lerner, J. W. Kelly, Nat. Chem. Biol. 2006,
2, 249.
[7] C. Takeuchi, R. Galve, J. Nieva, D. P. Witter, A. D. Wentworth,
R. P. Troseth, R. A. Lerner, P. Wentworth, Biochemistry 2006,
45, 7162.
[8] Q. Zhang, E. T. Powers, J. Nieva, M. E. Huff, M. A. Dendle, J.
Bieschke, C. G. Glabe, A. Eschenmoser, P. Wentworth, Jr., R. A.
Lerner, J. W. Kelly, Proc. Natl. Acad. Sci. USA 2004, 101, 4752.
[9] I. Casserly, E. J. Topol, Lancet 2004, 363, 1139.
[10] M. Schnolzer, P. Alewood, A. Jones, D. Alewood, S. B. Kent, Int.
J. Pept. Protein Res. 1992, 40, 180.
[11] S. Hyun, H. J. Kim, J. N. Lee, K. H. Lee, Y. Lee, D. R. Ahn, K.
Kim, S. Jeong, J. Yu, J. Am. Chem. Soc. 2007, 129, 4514.
[12] H. Levine III, Protein Sci. 1993, 2, 404.
[13] J. D. Harper, P. T. Lansbury, Jr., Annu. Rev. Biochem. 1997, 66,
385.
[14] D. B. Teplow, Amyloid 1998, 5, 121.
[15] W. B. Stine, Jr., K. N. Dahlgren, G. A. Krafft, M. J. LaDu, J. Biol.
Chem. 2003, 278, 11612.
[16] J. Bieschke, Q. Zhang, E. T. Powers, R. A. Lerner, J. W. Kelly,
Biochemistry 2005, 44, 4977.
[17] A. R. Hurshman, J. T. White, E. T. Powers, J. W. Kelly, Bio-
chemistry 2004, 43, 7365.
[18] H. Y. Shao, S. C. Jao, K. Ma, M. G. Zagorski, J. Mol. Biol. 1999,
285, 755.
[19] G. Bitan, S. S. Vollers, D. B. Teplow, J. Biol. Chem. 2003, 278,
34882.
[20] F. Chiti, M. Stefani, N. Taddei, G. Ramponi, C. M. Dobson,
Nature 2003, 424, 805.
[21] T. J. Nelson, D. L. Alkon, J. Biol. Chem. 2005, 280, 7377.
[22] N. A. Avdulov, S. V. Chochina, U. Igbavboa, C. S. Warden, A. V.
Vassiliev, W. G. Wood, J. Neurochem. 1997, 69, 1746.
[23] Z.-X. Yao, V. Papadopoulos, FASEB J. 2002, 16, 1677.
Received: December 22, 2007
Revised: January 26, 2008
Published online: April 11, 2008
Keywords: alzheimer’s disease · amyloids · lipid peroxidation ·
.
methylation · protein misfolding
[1] C. Haass, H. Steiner, Nat. Neurosci. 2001, 4, 859.
3922
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 3919 –3922