Novel multi-target compounds in the quest for new chemotherapies against Alzheimer's disease: An experimental and theoretical study

Article abstract of DOI:10.1016/j.bmc.2018.08.019

The lack of any effective therapy along with the aging world population anticipates a growth of the worldwide incidence of Alzheimer's disease (AD) to more than 100 million cases by 2050. Accumulation of extracellular amyloid-β (Aβ) plaques, intracellular tangles in the brain, and formation of reactive oxygen species (ROS) are the major hallmarks of the disease. In the amyloidogenic process, a β-secretase, known as BACE 1, plays a fundamental role in the production of Aβ fragments, and therefore, inhibition of such enzymes represents a major strategy for the rational design of anti-AD drugs. In this work, a series of four multi-target compounds (1–4), inspired by previously described ionophoric polyphenols, have been synthesized and studied. These compounds have been designed to target important aspects of AD, including BACE 1 enzymatic activity, Aβ aggregation, toxic concentrations of Cu²⁺ metal ions and/or ROS production. Two other compounds (5 and 6), previously reported by some of us as antimalarial agents, have also been studied because of their potential as multi-target species against AD. Interestingly, compounds 3 and 5 showed moderate to good ability to inhibit BACE 1 enzymatic activity in a FRET assay, with IC₅₀′s in the low micromolar range (4.4 ± 0.3 and 1.7 ± 0.3 μM, respectively), comparable to other multi-target species, and showing that the observed activity was in part due to a competitive binding of the compounds at the active site of the enzyme. Theoretical docking calculations overall agreed with FRET assay results, displaying the strongest binding affinities for 3 and 5 at the active site of the enzyme. In addition, all compounds selectively interacted with Cu²⁺ metal ions forming 2:1 complexes, inhibited the production of Aβ-Cu²⁺ catalyzed hydroxyl radicals up to a ～100% extent, and scavenged AAPH-induced peroxyl radical species comparably to resveratrol, a compound used as reference in this work. Our results also show good anti-amyloidogenic ability: compounds 1–6 inhibited both the Cu²⁺-induced and self-induced Aβ(1–40) fibril aggregation to an extent that ranged from 31% to 77%, while they disaggregated pre-formed Aβ(1–40) mature fibrils up to a 37% and a 69% extent in absence and presence of Cu²⁺, respectively. Cytotoxicity was additionally studied in Tetrahymena thermophila and HEK293 cells, and compared to that of resveratrol, showing that compounds 1–6 display lower toxicity than that of resveratrol, a well-known non-toxic polyphenol.

Full text of DOI:10.1016/j.bmc.2018.08.019

Accepted Manuscript
Novel Multi-target Compounds in the Quest for New Chemotherapies against
Alzheimer’s Disease: An Experimental and Theoretical Study
Alberto Martínez, Mai Zahran, Miguel Gomez, Coreen Cooper, Johnny
Guevara, Erik Ekengard, Ebbe Nordlander, Ralph Alcendor, Sarah Hambleton
PII:
S0968-0896(18)31385-3
https://doi.org/10.1016/j.bmc.2018.08.019
BMC 14506
DOI:
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Accepted Date:
2 August 2018
14 August 2018
Please cite this article as: Martínez, A., Zahran, M., Gomez, M., Cooper, C., Guevara, J., Ekengard, E., Nordlander,
E., Alcendor, R., Hambleton, S., Novel Multi-target Compounds in the Quest for New Chemotherapies against
Alzheimer’s Disease: An Experimental and Theoretical Study, Bioorganic & Medicinal Chemistry (2018), doi:
https://doi.org/10.1016/j.bmc.2018.08.019
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Novel Multi-target Compounds in the Quest for New
Chemotherapies against Alzheimer’s Disease: An Experimental and
Theoretical Study
a,
b
a
b
b
Alberto Martínez *, Mai Zahran , Miguel Gomez , Coreen Cooper , Johnny Guevara , Erik
c
c
b
a
Ekengard , Ebbe Nordlander , Ralph Alcendor, and Sarah Hambleton
a
Department of Chemistry, New York City College of Technology, The City University of New
York (CUNY), Brooklyn, New York, 11201, USA.
b
Department of Biology, New York City College of Technology, The City University of New York
(CUNY), Brooklyn, New York, 11201, USA.
c
Inorganic Chemistry Research Group, Chemical Physics, Department of Chemistry, Lund
University, Box 124, SE-221 00 Lund, Sweden.
*
Corresponding author: Dr. Alberto Martínez
Tel: 1-718-260-5855
E-mail: amartinez@citytech.cuny.edu

ABSTRACT
The lack of any effective therapy along with the aging world population anticipates a growth of
the worldwide incidence of Alzheimer’s disease (AD) to more than 100 million cases by 2050.
Accumulation of extracellular amyloid-β (Aβ) plaques, intracellular tangles in the brain, and
formation of reactive oxygen species (ROS) are the major hallmarks of the disease. In the
amyloidogenic process, a β-secretase, known as BACE 1, plays a fundamental role in the
production of Aβ fragments, and therefore, inhibition of such enzymes represents a major strategy
for the rational design of anti-AD drugs. In this work, a series of four multi-target compounds (1-
4
), inspired by previously described ionophoric polyphenols, have been synthesized and studied.
These compounds have been designed to target important aspects of AD, including BACE 1
enzymatic activity, Aβ aggregation, toxic concentrations of Cu2+ metal ions and/or ROS
production. Two other compounds (5 and 6), previously reported by some of us as anti-malarial
agents, have also been studied because of their potential as multi-target species against AD.
Interestingly, compounds 3 and 5 showed moderate to good ability to inhibit BACE 1 enzymatic
activity in a FRET assay, with IC50’s in the low micromolar range (4.4 ± 0.3 and 1.7 ± 0.3µM,
respectively), comparable to other multi-target species, and showing that the observed activity was
in part due to a competitive binding of the compounds at the active site of the enzyme. Theoretical
docking calculations overall agreed with FRET assay results, displaying the strongest binding
affinities for 3 and 5 at the active site of the enzyme. In addition, all compounds selectively
interacted with Cu2+ metal ions forming 2:1 complexes, inhibited the production of Aβ-Cu2+
catalyzed hydroxyl radicals up to a ~100% extent, and scavenged AAPH-induced peroxyl radical
species comparably to resveratrol, a compound used as reference in this work. Our results also
2
+
show good anti-amyloidogenic ability: compounds 1-6 inhibited both the Cu -induced and self-

induced Aβ(1-40) fibril aggregation to an extent that ranges from 31% to 77%, while they
disaggregated pre-formed Aβ(1-40) mature fibrils up to a 37% and a 69% extent in absence and
2
+
presence of Cu , respectively. Cytotoxicity was additionally studied in Tetrahymena thermophila
and HEK293 cells, and compared to that of resveratrol, showing that compounds 1-6 display lower
toxicity than that of resveratrol, a well-known non-toxic polyphenol.
KEYWORDS: Alzheimer’s disease, polyphenol, ionophoric compound, amyloid-beta
aggregation, multi-target compound, reactive oxygen species, antioxidant, BACE1, Tetrahymena,
β-secretase inhibitor, molecular design, docking
1
. Introduction
Alzheimer’s disease (AD) is a chronic neurodegenerative disease that causes 60-70% cases of
dementia, and affects more than 30 million people worldwide. This number is expected to grow to
up to more than 100 million by 2050 due to the rise in life expectancy and the aging population
[
1]. There is no effective therapy currently available, although some acetylcholinesterase (AChE)
inhibitors and N-methyl-D-aspartate (NMDA) receptor antagonists are in clinical use to
temporarily relieve the symptoms. However, the benefits from their application are very limited
and none of them have been shown to clearly delay or halt the progression of AD. In addition, the
use of anticholinergic drugs has been recently linked to higher risk of dementia [2] and, thus, the
search for novel chemotherapies for Alzheimer’s disease clearly requires urgent attention.
A large body of research shows that the accumulation of extracellular amyloid−β (Aβ) plaques,
intracellular neurofibrillary tangles in the brain, and the production of reactive oxygen species
(
ROS) are the major hallmarks of AD [3-7]. The amyloid cascade hypothesis, which has dominated
AD research for the past 25 years, proposes that the amyloid precursor protein (APP), a

transmembrane protein that is concentrated in the synapses of neurons, is enzymatically cleaved
to produce various Aβ peptides [4,5]. Cleavages taking place after residues 40 and 42 of APP
produce Aβ(1-40) and Aβ(1-42), respectively, the two typical components that finally assemble
into toxic Aβ aggregates resulting in oxidative and inflammatory damage [6-9]. Aβ is produced
by the sequential enzymatic cleavage of APP by two proteases, the β-secretase (BACE – beta site
APP cleaving protein enzyme) and γ-secretase [10]. BACE is the rate limiting step and, therefore,
inhibition of β-secretase decreases the production of all forms of Aβ, including the most
pathogenic Aβ(1-40) and Aβ(1-42). BACE is a transmembrane aspartyl protease that exists in two
isoforms [11]: BACE 1 located in the central nervous system (CNS) and BACE 2 expressed mainly
in the periphery [12]. Studies show that BACE 1 protein levels and activity are elevated in sporadic
AD brains [13], and increasing evidence reveals that BACE 1 levels are upregulated under stress
conditions related to AD incidence [14-16]. Since its discovery, BACE 1 has been heavily pursued
as a small-molecule drug target. The first generation of BACE 1 inhibitors included large
peptidometic molecules which could not penetrate the blood brain barrier (BBB), and thus a
substantial amount of efforts is currently being devoted to the design of anti-AD small BACE1
inhibitors with BBB permeability. In this context, compounds from a second generation consisting
of smaller, orally bioavailable agents have already entered clinical trials [17-21].
In particular, LY2811376 (Figure 1), developed by Eli Lilly Company, reached phase I clinical
trials in 2009 [19], and it is used in this work as a BACE 1 inhibitor control. This drug is able to
strongly inhibit BACE 1 activity by binding in the catalytic site of the enzyme. The BACE 1
catalytic domain is composed of two aspartyl protease motifs located on two loops from residues
3
2-35 and 228-231, respectively. The active-site aspartate residues (Asp32 and Asp228) have
become a promising target for BACE 1 inhibiting drugs. These residues have been reported to

form a hydrogen bond network with a single water molecule in the apoprotein, and to form
hydrogen bond interactions in the presence of LY2811376, thus preventing BACE 1 from
performing its function. The BACE 1 flap region (from residue 67-77) is reported to maintain the
ligands in the catalytic pocket, thus contributing to the BACE 1 – ligand binding stability.
Evidence also points to the contribution of metal ions to the progression of AD. In particular, Cu2+
has been detected in elevated µΜ concentrations in Aβ plaques [22-24]. Essential copper (Cu, at
different oxidation states) plays an important role in electron and oxygen transport proteins, and it
can also be found in synaptic vesicles loosely bound to low affinity ions, for instance glutamate
[
25]. Relevant to AD, there are signs indicating that Cu homeostasis is altered in AD brains,
resulting in anomalous concentrations of this metal ion in the synaptic cleft [9,26]. The binding of
aberrant Cu to Aβ seems to promote Aβ aggregation into toxic and neurodegenerative forms of
the protein [22,27], and to catalyze the production of reactive oxygen species (ROS) and oxidative
stress leading to neuronal death [28-30]. This mounting evidence supporting the role of metal ions,
2
+
specifically Cu , in the pathophysiology of AD has rendered metal ionophoric therapy a promising
treatment strategy, and some drugs aimed at chelating toxic concentrations of metal ions have
reached clinical trials [31-34].
Due to the complexity of AD and the identification of many potential targets, the multi-target
directed ligand (MTDL) approach, in which a single drug is designed to contain two or more
pharmacophores, has recently attracted the attention of researchers in the pursuit of improved
drugs [35,36]. The MTDL strategy has the extra benefit of synergy, i.e. the activity of the
combination is superior to the added activity of the separate parts. Several multi-target compounds
have been investigated, although the great majority are still based on the old and ineffective
cholinergic hypothesis, and only a few of them have reached clinical trials without further relevant

therapeutic success [37,38]. A more promising approach consists of designing MTDL drugs that
contain disease-modifying pharmacophores, that target the Aβ production-aggregation-removal,
ROS generation or aberrant concentration of undesired metal ions.
1
3
2
LY2811376
Resveratrol
4
5
6
Ionophoric Polyphenol
Figure 1. Multi-target compounds 1-6 evaluated in this work as ionophoric, antioxidant and
antiamyloidogenic agents, and BACE 1 inhibitors. Resveratrol, LY2811376 and general structure
of previous generation of ionophoric polyphenols (see ref. 39) are also represented.
In previous work we have demonstrated that a family of multi-target ionophoric polyphenols
inspired by the structure of the natural product resveratrol and incorporating a motif in the

molecular design that provides the ability to complexate with AD-related Cu2+ ions show
2
+
promising potential as anti-AD agents [39]. These compounds selectively chelate Cu over other
2
+
biologically relevant metal ions, inhibit up to a 99% of the Cu -catalyzed ROS production and
2
+
prevent up to a 90% extent the Cu -induced and self-induced Aβ aggregation, while being non-
toxic towards healthy eukaryotic cells.
Because of the enormous potential of i) BACE 1 inhibitors, ii) metal ionophoric agents and iii)
our previous family of selective ionophoric polyphenols, in the present work we have synthesized
and evaluated a family of novel multi-target and disease-modifying anti-AD compounds (1-4,
Figure 1) where a BACE 1 inhibitor fragment is incorporated into the structure of a ionophoric
polyphenol. We, therefore, aim at providing a detailed study of a small family of compounds that
exploit the multi-target approach in the rational design of anti-AD agents. We thus report: 1) their
ability to inhibit BACE 1 enzymatic activity with a comparison to clinically tested LY2811376, 2)
a detailed theoretical study of docking for a better understanding of the observed BACE 1
2
+
inhibition properties and mechanisms, 3) their Cu binding properties and ion selectivity, 4) their
2
+
ability to both inhibit the Cu -Aβ(1-40) induced ROS formation and AAPH (2,2’-azobis(2-
amindino-propane) dihydrochloride)-induced free radicals, 5) their potential to inhibit the
aggregation of Aβ(1-40) into pathological forms of the peptide, and 6) their cytotoxicity toward
healthy eukaryotic cells of Tetrahymena thermophila and human embryonic kidney cells
(
HEK293). Compounds 5 and 6 (Figure 1), previously synthesized by some of us and tested for
their antimalarial properties [40,41], were also studied and included because of their potential as
multi-target anti-AD compounds (see section 2.1 for more details on the rational design of
compounds).

2
2
. Results and discussion
.1. Drug design and drug likeliness
The design of multi-target compounds 1-4 (Figure 1 and Schemes 1 and 2) was inspired by the
previous family of ionophoric polyphenols [39], because of their promising potential and lack of
toxicity. The polyphenol moiety was maintained due to the well-known antioxidant and anti-
amyloidogenic properties of this kind of compounds, and it is inspired by the structure of
resveratrol (3,5,4’-trans-trihydroxystilbene), a natural polyphenol that has currently completed
phase II clinical trials in 120 patients with possible AD diagnosis [42]. Several health-promoting
properties have been reported for resveratrol due to its intrinsic antioxidant capacity. Interestingly,
this natural compound is able to penetrate the blood-brain barrier (BBB) and exert strong
neuroprotective effects [43-45], even at low concentrations, by acting as an anti-amylodiogenic
and fibril-destabilizing polyphenol [46,47].
2
+
Importantly, the multi-target structures were provided with both Cu selective ionophoric and
BACE 1 motifs for increased anti-AD potential. Thus, compounds 1-4 were expected to maintain
and/or enhance all Cu2+ binding, antioxidant and anti-amyloidogenic properties of compounds
previously synthesized, and display the added benefit of BACE 1 inhibition (Scheme 1). The
molecular design of compounds 1-4 provides a low-medium affinity and selective chelating motif
(
bidentate N-O) into the structure of a polyphenol-like compound (see section 2.5 below for
expanded discussion and results on metal ion selectivity) and adds a benzimidazole, indole or
adenine BACE 1 inhibitor moiety. These fragments have been chosen as BACE 1 inhibition
moieties because some analogues, such 2-aminobenzimidazole, have been identified in fragment-
based discovery as effective BACE 1 inhibitor fragment, based upon screening Evotec’s library of
compounds with more than 20,000 drugs through high concentration enzyme assays [48,49].

Importantly, compounds 5 and 6, previously reported by some of us [40,41], were also included in
this study because of their potential as BACE 1 inhibitors and multi-target agents. We note that
compounds 5 and 6 are structurally related to 1-4 by containing a selective ionophoric motif
(
bidentate N-O) and a BACE 1 inhibitor moiety, a chloroquinoline in their case. Aminoquinolines
and aminoheterocycles have also been identified as promising BACE 1 inhibitors through XRD
screening and enzyme assays [50]. We, therefore, expected that the presence of different BACE 1
inhibitor moieties providing opportunities for a variety of hydrogen bonding and hydrophobic
interactions, and also lengths and chemical differences in the linkers/spacers and polyphenol
moieties would allow us to further explore the multi-target approach in the search for new
chemotherapeutic agents against AD.
BACE 1 Inhibitor
Chelating Motif
Moiety
BACE 1 Inhibitor
Moiety
+
Ionophoric Polyphenol
moiety
Chelating Motif
Multi-target compound
Ionophoric Polyphenol
Moiety
Scheme 1. Design of a multi-target compound incorporating a BACE 1 inhibitor fragment to the
structure of an ionophoric polyphenol.

Table 1. Physicochemical properties of the multi-target compounds 1-6, resveratrol and BACE 1
inhibitor LY2811376.
Lipinski’s
1
2
3
4
5
6
LY2811376 Resveratrol
Rules
3
60.84 377.83 280.33 281.32 327.77 325.80
320.36
2.894
4
228.25
2.833
3
ꢀ450
MW
clogP
HBA
2
.399
6
0.419
9
3.382
4
2.088
5
3.031
6
4.492
4
ꢀ5.0
ꢀ10
3
4
2
2
2
2
1
3
ꢀ5
HBD
8
5.03
135.77
56.06
65.42
78.21
56.989
63.1
60.69
-0.27
ꢀ90
PSA
-
0.772 -1.821
-0.186
-0.521 -0.567 -0.0305
-0.364
*
logBB
*
Calculated according to logBB = -0.0148xPSA + 0.152xclogP + 0.130. Drugs with logBB below -1 are expected to be poorly
distributed in the brain [52,53].
MW, molecular weight; clogP, calculated logarithm of the octanol-water partition coefficient; HBA, hydrogen bond acceptor;
HBD, hydrogen bond donor; PSA, polar surface area.
Importantly, blood brain barrier (BBB) permeability for compounds 1-6 was predicted from
Lipinski rules’s as previously stated [52,53]. The calculation of logBB for predicted BBB
penetration ability (see equation in footnote of Table 1) is comparable in all parameters to those
for the clinical BACE 1 inhibitor LY2811376 and confirms probable brain distribution for the
multi-target compounds.
2
.2. Chemistry
Compounds 1 and 2 were synthesized in a multi-step process. In the first step, benzimidazole or
adenine were deprotonated with NaH and reacted with 1-bromo-2-chloroethane by means of a
nucleophilic substitution to produce 1-(2-chloroethyl)-1H-benzimidazole or N9-(2-
chloroethyl)adenine, respectively [54]. These substances were further reacted with pure
ethylenediamine (also used as a solvent for the reaction) to yield intermediates ethylenediamine-

N,1-ethylbenzimidazole hydrochloride (I1) or ethylenediamine-N,9-ethyladenine hydrochloride
I2), respectively, which were finally condensed in both cases with 2,4-dihydroxybenzaldehyde
(
through a Schiff base reaction in dry DMF to produce compounds 1 and 2 as orange powders in
relatively good yields (Scheme 2 and Scheme S1 in Supplementary Information).
Compounds 3 and 4 were also synthesized by a multi-step process inspired by a previously
described procedure [55]. Indole or 7-azaindole were deprotonated with NaH, reacted with
bromoacetonitrile by means of a nucleophilic substitution and reduced to the amino derivative with
LiAlH
azaindole-1-ethanamine (I4), respectively. These amino intermediates were later condensed with
,4-dihydroxybenzaldehyde through a Schiff base reaction in dry DMF to produce compounds 3
4
in dry diethyl ether, yielding intermediates 1-H-indole-1-ethanamine (I3) or 1-H-7-
2
and 4 as orange powders in good yields (Scheme 2 and Scheme S1 in Supplementary Information).
It is interesting to note that compounds 1 and 2 were synthesized as salts, while compounds 3 and
4
are neutral compounds, and therefore 1 and 2 were expected to have a better aqueous solubility.
All compounds remained stable in hydrated DMSO solution for more than 48 hours at room
temperature, with less than 5% decomposition only in the case of compounds 1 and 2 (see figure
S17 in Supporting Information).

a
1
-(2-Chloroethyl)-1H-
Benzimidazole
Ethylenediamine-N,1-
Ethylbenzimidazole Hydrochloride
(I1)
I1
1
CN
NH2
H
i.
ii.
N
N
N
iii.
b
1
-H-Indole-1-Ethan-Amine
(
1-(Cyanomethyl)-1H-Indole
I3)
I3
3
Scheme 2. Synthesis of compounds 1 and 3 as representative examples. a) i. NaH, rt, 20 h,
BrCH CH Cl; ii. NH CH CH NH , rt, 4 days; iii. 2,4-dihydroxybenzaldehyde, 60ºC, 4 h then rt
for 20 h, dry DMF; b) i. NaH, rt, 30 min, dry DMF; ii. BrCH CN, 65ºC, 30 min then rt, 20 h; iii.
LiAlH , reflux, 4 h, diethyl ether; iv. 2,4-dihydroxybenzaldehyde, 60ºC, 4 h, dry DMF. See
Scheme S1 in Supplementary Information for synthesis of compounds 2 and 4.
2
2
2
2
2
2
2
4

2
.3. BACE 1 inhibitory activity
Strategies employed to design BACE 1 inhibitors include substrate-based design, high throughput
screening and fragment-based lead generation. Imidazolic and aminoquinolinic compounds have
been found to be good candidates for BACE 1 inhibition [56-59], which provided the rationale for
the molecular design of 1-6 and the testing in their ability to inhibit BACE 1 enzymatic activity.
Results of BACE 1 inhibition measured through a fluorescence resonance energy transfer (FRET)
are shown in Table 2 and supplementary information (Figures S1 and S3). In this assay,
fluorescence of a substrate is measured upon cleavage of a quencher by the BACE 1 enzyme (see
Experimental Section for more details). Inhibition was measured for all compounds at a single
concentration of 50 µM and also IC50 values were calculated when possible. Results were then
compared to the well-known BACE 1 inhibitor LY2811376, a compound that has reached phase I
clinical trials in humans [18,60], used as positive control to validate our experiments. Resveratrol
was also used as a control for all our experiments. Results for LY2811376 and resveratrol generally
agree with previously published results [61,62], and minor discrepancies can be attributed to
differences in the FRET assay and experimental design in use.
Results in table 2 allow to speculate that changes in the structure of the compounds may alter their
ability to interact with the peptide through the active site by forming hydrogen bonds or
hydrophobic interactions, which might explain the different activity. Results for compound 1, and
especially compounds 3 and 5 are in the moderate/good activity range, and comparable to
previously reported imidazolic compounds [56,58]. In general, compounds 1 and 2 with longer
linkers between the polyphenol and the BACE 1 fragment seem to be less active. Curiously
enough, compound 4, which is structurally almost identical to compound 3 but displaying an extra
nitrogen atom in the indole fragment shows no ability to inhibit BACE 1, which underscores the

the importance of even small changes in the BACE 1 inhibitory fragment. Differences between
compounds 5 and 6 suggest that the presence of the amide bond in 5 might play an important role
in the observed activity. Resveratrol and 5 ionophoric polyphenols synthesized in previous work
[
39] were also studied in their ability to inhibit BACE 1 activity (see Supplementary Information
Figure S2). None of the ionophoric polyphenols could reach 50% inhibition in any case when
tested at 10 µM and 50 µM, which validates the incorporation of indolic or aminoquinolinic
fragments into the molecular design as BACE 1 inhibitor fragment.
Table 2. BACE 1 inhibition activity of compounds 1-6, LY2811376 and resveratrol through a
FRET assay.
%
BACE 1 Inhibition at
ʌ
Compound
BACE 1 IC50
*
BACE 1 IC50
5
0 µM
LY2811376
100
0.173 ± 0.008
--
--
--
Resv
41.1 ± 2.3
62.5 ± 3.6
59.2 ± 8.3
100
1
2
3
4
5
6
15.6 ± 2.7
41.4 ± 1.2
4.4 ± 0.3
106.7 ± 6.1
1.7 ± 0.3
>160
18.5 ± 0.8
42.0 ± 1.5
5.7 ± 0.3
--
31.6 ± 9.9
91.0 ± 5.2
29.6 ± 8.7
9.9 ± 0.1
--
*
All concentrations in µM. Substrate concentration of 250 nM
ꢁ
Experiments performed at double concentration of substrate (500 nM)
To gain further insights into the mechanism of action for the observed activity, BACE 1 inhibition
of compounds 1-3 and 5 was measured at double substrate concentration to ascertain whether the

binding mode to the enzyme was competitive or noncompetitive. Compounds 4 and 6 were not
studied any further given the poor activity observed. Typically, competitive inhibitors prevent
binding of substrate by blocking the active site of the enzyme, while noncompetitive do it by
interacting with allosteric locations of the protein, which in turns induces conformational changes
that can lead to loss of enzymatic activity. Results are also shown in Table 2 and reveal that IC50
values seem to be slightly higher than in experiments at initial substrate concentration. Higher
IC50’s (lower activity) might be an indication that binding of the multi-target compounds to the
enzyme is to some extent competitive, and that the observed activity is due, at least in part, to
binding in the active site, as per the molecular design of the compounds. It is interesting to note
that doubling the substrate concentration had the strongest effect on compound 5, which could
suggest that this compound and the substrate might be interacting through common residues in the
active site of BACE 1. Docking calculations shown below (section 2.4) were aimed at obtaining a
clearer picture of the mode of action of compounds 1-6.
One last observation concerns with the strength of the observed inhibition. It has been reported
that strong inhibition of BACE 1 might induce disruptive effects on synaptic functions and
plasticity, indicating the need for either careful dosage of drugs targeting this protease or to achieve
partial inhibition of the enzyme [18,63]. In this context, the moderate inhibition range displayed
by some of our compounds might become an interesting property, especially when considering
that these compounds are also designed to target other important aspects of AD.
2
.4. Molecular docking of compounds 1-6 with BACE 1
We have experimentally shown the potential of compounds 1-6 to inhibit the β-secretase BACE
. Despite the structural similarities between some of the compounds, their ability to inhibit the
enzymatic activity was substantially different in most of the cases. In addition, results at double
1

concentration of substrate suggested a competitive mode of action for compounds 1-3 and 5. Thus,
molecular modeling of the compounds and docking were performed in order to gain further
insights into the binding mode, binding affinity and the underlying interactions between
compounds 1-6 and BACE 1. Molecular docking is a computational method that predicts a
preferred position, conformation, and orientation a ligand can take to effectively bind to a receptor.
Targeted docking studies of compounds 1-6 to the BACE 1 active site were performed using
AutoDock Vina [64], and noncovalent intermolecular interactions were calculated using Arpeggio
[
65] (see Experimental Section for more details). LY2811376 and resveratrol were also docked to
BACE 1 and used as controls.
The structure of the ligands was first modeled using Avogadro 1.2.0 [66], and energy was
minimized in the obtained structures to an optimized conformation. Structures of the modeled
compounds 1-6, as well as the structures of LY2811376 and resveratrol can be found in
Supplementary Information, Figure S5. Table 3 shows a summary of the docking calculation
results and interactions formed between the best compound pose and BACE 1. The docking results
show that LY2811376, compounds 3, 5 and 1 retained higher binding affinities (-8.6 kcal/mol, -
8
.2 kcal/mol, -7.9 kcal/mol, and -7.8 kcal/mol, respectively) than compounds 2, 4, 6, and
resveratrol, indicating a tighter binding capacity to the active site.
Analysis of the interactions at the binding sites shows that compounds 2, 4, 6, and resveratrol
participate in different hydrogen bonds and hydrophobic interactions than LY2811376, 3, 5, and
1
, supporting their lower binding affinity to the active site. Moreover, LY2811376, 3, 5, and 1
form hydrogen bonds with various residues, including the catalytic Asp32 and Asp228 residues
Figure 2). Compounds 2, 4, 6, and resveratrol all lack hydrogen bonds to the Asp dyad.
(

Figure 2: Representation of the four compounds forming hydrogen bonds with one or both
aspartyl acids (Asp32 and Asp228) of BACE 1. Compounds 1, 3, 5 and LY2811386 are shown in
A, B, C and D respectively. Residues in green are the aspartyl acids, and in yellow are residues in
the flap region of the enzyme.
Interestingly, compound 1 interacts with more residues than compounds 3 and 5, but has a lower
binding affinity. The larger number of interactions may be due to the fact that compound 1 is
bigger in size than 3 and 5, hence it is able to interact with more residues to form hydrogen bonds,
hydrophobic interactions and π-π interactions, but not necessarily to form a tighter binding. The
combination of the docking score and the analysis of the noncovalent interactions show that among
compounds 1-6, compounds 3 and 5 show the best potential to inhibit BACE 1. Figure 3 shows
the three main noncovalent interactions (hydrogen bonds, hydrophobic and π-π interactions)
analyzed here for compounds 3 and 5. Both compounds exhibit relatively similar number of

hydrogen bond interactions, the difference being that compound 3 interacts with both aspartyl
acids, Asp32 and Asp228, like the control LY2811376, while compound 5 interacts only with
Asp228. Additionally, both compounds 3 and 5 are forming hydrogen bonds to other active site
residues: Thr231 for compound 3, while compound 5 interacts with Gly34 from the catalytic loop
and Thr72 from the flap region. Furthermore, hydrophobic as well as aromatic interactions with
residues Tyr71, Gln73 and Ile110 seem to be a common feature of 3 and 5, as well as LY2811376.
All these observations taken together suggest that 3 and 5 exhibit good potential to inhibit BACE
1
based on their docking binding energy and the interactions they are forming with the catalytic
aspartyl dyad. Interestingly, these results are in agreement with the FRET experimental results,
showing that 3 and 5 have the highest BACE 1 inhibitory activity (Table 2, section 2.3). The 3D
structures of best poses of each compound 1-6, LY2811376, and resveratrol can be seen in Figure
S5 of the Supplementary Information.
Table 3. Summary of docking scores and noncovalent interactions for compounds 1-6,
LY2811376 and resveratrol with BACE 1 active site.
Binding Affinity
kcal/mol)
H-bond interacting
residues
Hydrophobic interacting
residues
Aromatic interacting
residues
Compound
(
Ile126, Arg128,
Tyr198, Asp228,
Thr231
Leu30, Val69, Tyr71, Phe108,
Ile118, Trp115, Tyr198
1
-7.8
Tyr71, Phe108
Thr72, Lys107,
Phe108, Gly230
Asp32, Asp228,
Thr231
Tyr71, Phe108, Trp115
Tyr198
2
-7.7
-8.2
-7.2
-7.9
-6.9
-8.6
-7.2
Tyr71
3
Tyr71, Gln73, Ile110
Tyr71, Phe108
Tyr71, Phe108
Tyr71, Phe108, Tyr198
Tyr71, Phe108, Tyr198
Tyr71, Phe108
4
Gly230
Gly34, Thr72,
Asp228
5
Tyr71, Gln73, Phe108, Ile118
Tyr71, Phe108, Ile110, Ile118
6
Gln73
Asp32, Asp228,
Leu30, Tyr71, Gln73, Phe108,
Ile110, Trp115
Leu30, Tyr71, Ile118, Tyr198,
Ile226
LY2811376
Resveratrol
Tyr71
Thr231
Gly34
Tyr71, Phe108

Figure 3: Representation of noncovalent interactions between compounds 3 and 5 and BACE 1.
A, B and C show the hydrogen bonds, hydrophobic and π-π interactions between compound 3 and
BACE 1 respectively. D, E and F show the hydrogen bonds, hydrophobic and π-π interactions
between compound 5 and BACE 1, respectively. The amino acids in green are the two aspartyl
acids of the catalytic site. The yellow loop is the flap region. A-D, B-E and C-F are shown in the
same view to ease the comparison between the compounds. The views differ for each type of
interaction. The yellow dashed lines in A and D refer to hydrogen bonds, the cyan dotted lines in
B and E refer to hydrophobic interactions, and the magenta lines in C and F refer to π-
π interactions.

In order to better understand the mechanism of action of the multi-target compounds, we also
focused our attention to the atoms or molecular fragments of compounds 1-6 responsible for
interacting with the active site of BACE 1. In section 2.3 we speculated that: 1) longer linkers
between the BACE 1 inhibitory fragment (indole, imidazole and aminoquinoline) and the
polyphenol moiety, 2) changes in the BACE 1 inhibitory fragment, and 3) the presence of the
amide bond in compound 5 could have an impact on BACE 1 inhibitory activity. Docking results
agree with our previous observations and reveal, for instance, that longer linkers in compounds 1
and 2, extra nitrogen atoms in the six-membered ring of the BACE 1 inhibitory fragment
(
compound 2 compared to compound 1, and compound 4 compared to compound 3), or the absence
of the amide bond in compound 6 with respect to compound 5, in general, lower the binding
affinity for the catalytic site of the enzyme, and as a result, the inhibitory potency of the compound.
In detail: Compound 3 establishes hydrogen bonds with the catalytic Asp32 and Asp228 residues
through the OH groups of the polyphenol moiety, while compound 5 uses the amide nitrogen to
interact with Asp228 (Figures 2 and 3). Interestingly, none of the other multi-target compounds
binds the catalytic Asp dyad through the polyphenol moiety (Figure S5 in Supplementary
Information). Furthermore, although the indolic or the aminoquinolinic fragments of 3 and 5 do
not seem directly involved in the interaction with Asp32 and Asp228, we theorize that their
hydrophobic and aromatic interactions with other residues of BACE 1 active site (Tyr71, Gln73
and Ile110 seem to be common features, see Table 3 and Figure 3) are important for the compounds
to end in the appropriate orientation. Some evidence supporting that observation could come from
the complete lack of activity of compound 4 (Table 2, section 2.3), structurally almost identical to
compound 3 but replacing the indole by an azaindole moiety. The presence of the extra nitrogen
in the six-membered ring of the BACE 1 inhibitory fragment prevents the azaindole group of 4

from establishing hydrophobic interactions with the BACE 1 catalytic site, resulting in an
orientation that does not allow hydrogen bond interactions with the catalytic Asp32 and Asp228
(
see Figure S5 in Supplementary Information). A similar observation can be made between
compounds 1 and 2. The extra nitrogen groups in the adenine moiety of compound 2 do not allow
such fragment to form hydrophobic interactions with the catalytic site (see Figure S5 in
Supplementary Information), resulting also in the loss of hydrogen bonding with the Asp residues
and a lower activity.
In summary, it seems reasonable to hypothesize that compounds 1, 3 and 5 owe the BACE 1
inhibitory activity to their ability to interact with the catalytic Asp32 and Asp228 residues (as it
happens for LY2811376), and that any molecular feature of the compounds favoring that
interaction induces an enhancement of the activity. These observations could have important
implications in the rational design of compounds with BACE 1 inhibitory properties.
2
.5. Metal ion binding properties and selectivity
Due to the active role that several metal ions play in the progression of AD (i.e. promoting Aβ
aggregation and ROS production), metal chelation represents a promising strategy for the
treatment of this disease. Ideal therapeutic ionophoric agents must be selective and display a
moderate binding affinity for the target ions, which would allow them to compete with Aβ for
2
+
metal binding (Aβ forms 1:1 complexes with Cu with logK values in the range 8-10, depending
a
on experimental conditions [67,68]), but not with metalloenzymes. Compounds 1-6 were studied
spectrophotometrically to determine 1) their ability to selectively chelate Cu2+ versus other
2
+
2+
2+
2+
2+
2+
biologically relevant metal ions, such as Zn , Fe , Ca , Mn and Mg , and 2) their Cu binding
2
+
properties, including stoichiometry and association constants of the compound-Cu adducts. The
interaction of 1-6 with metal ions was spectrophotometrically followed by changes in the

absorption spectra upon interaction with metal ions. Results for 1-6 (see Figure S6 in
Supplementary Information) show that, with the only exception of compound 5, these multi-target
2
+
compounds bind Cu with a remarkable selectivity, which could have implications in preventing
side effects.
2
+
Binding stoichiometry and apparent association constants for the complex Cu :compound were
2
+
calculated from spectrophotometric titrations of 1-6 upon stepwise additions of a Cu solution
see Figure S7 in Supplementary Information). These experiments revealed that multi target
(
2
+
2+
compounds 1-4 and 6 form 2:1 complexes with Cu (compound: Cu ; binding through the
chelating motif N-O is proposed). Interestingly, compounds 1-4 and 6 displayed moderate values
for logK (see Table 4), as proposed in the molecular design (Section 2.1), adequate to have a
a
therapeutic value as ionophoric agents while preventing competition with cuproenzymes. These
results are comparable to the previous family of ionophoric polyphenols reported by us [39],
validating our molecular design aimed at maintaining the properties observed in that first
generation of compounds.
Interestingly, changes in the spectra of buffered solutions of free compound 5 were not observed
upon additions of Cu2+ under the experimental conditions in use, which did not allow us to
determine the copper binding properties for this compound. This might be indicative of a weaker
interaction with the studied metal ion, possibly due to the difference in chelating motif and the
poorer ability of carbonyl groups to form coordination complexes through the oxygen atom.
The experiments described in the following sections were aimed at correlating Cu2+ binding
properties with the ability of the compounds to inhibit ROS production and Aβ(1-40) aggregation.

2
.6. Antioxidant ability: Inhibition of Cu -Aβ(1-40) hydroxyl radical production and
+
2
scavenging of AAPH-induced peroxyl radicals
There seems to be consensus in the implication of oxidative stress in AD [76-79]. A relationship
seems to exist between signs of oxidative stress, Aβ cortical deposits and homeostatic imbalance
of metal ions like Cu (copper in its different oxidation states). On one hand, variation of oxidation
biomarkers suggests that Aβ amyloid plaques represent environments of elevated oxidative stress
[
[
70]. On the other hand, the brain levels of Cu and its binding to proteins are dysregulated in AD
2
+
71-73], and enrichment in Cu may not only serve to assemble Aβ deposits. The complex Cu -
.
-
.
Aβ is known to catalyze the generation of O
2
, H
2
O and OH through Fenton reactions in the
2
presence of molecular oxygen and a reducing agent [74]. The brain environment is well supplied
in oxygen and several reducing agents are found at moderately high concentrations (for instance,
brain concentration of ascorbate ranges between 200 µM and 10 mM).
2
+
In this work we have measured the ability of compounds 1-6 to inhibit the Cu -Aβ(1-40)-
.
catalyzed formation of hydroxyl radical (OH ), one of the most toxic radicals in biological systems.
The effect of resveratrol, well-known for its antioxidant properties, was also analyzed for
2
+
comparison purposes. Hydroxyl radical production was promoted through Fenton reaction of Cu -
Aβ(1-40) with dissolved O in presence of ascorbate and 2 eq. of the compounds under study (see
2
Experimental Section for more details). Time course plots (Figure S9 in Supplementary
Information) were used to calculate the apparent rate of ROS formation by determining the slope
of the linear segment of the graphs. All results are displayed in Figure S8 of Supplementary
Information and Table 4, and show that, overall, compounds 1-6 both inhibit and slow down the
2
+
Cu -Aβ(1-40) catalyzed generation of hydroxyl radical better than natural resveratrol. In general,

2
+
.
we observed that compounds with higher Cu affinity proved to be better inhibitors of the OH
+
2
radical production, which allowed us to hypothesize that trapped Cu ions could not participate
in the Fenton reaction that leads to hydroxyl radical formation.
In addition, the total antioxidant capacity of compounds 1-6 was also determined in the absence of
2
+
Cu ions by a modification of the ORAC (oxygen radical absorbance capacity) assay [75,76].
.
Assays based on the scavenging of peroxyl radicals (ROO ) induced by 2,2’-azobis(2-amindino-
propane) dihydrochloride (AAPH), like the ORAC assay, are considered relevant because they use
a biologically significant ROS source. This method determines the ability of antioxidants to protect
fluorescein (FS) from damage by free radicals. Thus, the protective effect on FS of the multi-target
compounds was then compared to that of resveratrol and ascorbic acid used as controls. Results
are shown in Table 4 and Figure S10 in Supplementary Information (dose vs. response curves).
Overall, compounds 1-4 and 6 display a comparable activity to that of natural resveratrol and the
previous family of ionophoric polyphenols studied by us [39], while ascorbic acid proved to have
a low peroxyl radical protecting ability when compared to the polyphenols, as previously reported
by us and others [39,76,77].
The poor total antioxidant ability obtained for compound 5 seems to correlate well with the low
activity against hydroxyl radical observed before, suggesting that the antioxidant capacity of 1-4
2
+
and 6 against hydroxyl radicals might not only be related to their ability to complexate Cu ions.
We hypothesize then that the antioxidant behavior could be partly due to radical scavenging by
hydrogen transfer from amino or hydroxo groups. Taking both antioxidant assays, we observed
that the cationic compounds (1 and 2) were in general better antioxidant agents, probably due to
2
+
1
) their higher Cu binding affinity, and 2) the larger number of groups in their structures to

provide hydrogen transfer opportunities. These observations might be of importance in the rational
design of future generations of anti-AD agents with antioxidant properties.
2
+
Table 4. Antioxidant properties of compounds 1-6. Ability to inhibit Cu -Aβ(1-40) catalyzed
hydroxyl radical production and to scavenge AAPH-induced free radicals. Apparent association
2
+
constants ligand:Cu (as logK ).
a
2
inhibition of Cu -
+
%
Rate of radical
formation
Apparent association
IC50 for AAPH-
2
+
Compound
Aβ(1-40) catalyzed
ROS production
0
constants for Cu
induced radicals**
(a.u./min)*
5.11 ± 0.42
(logKa)
2
Cu -Aβ(1-40)
+
--
--
Resv
9.0 ± 3.3
--
4.46 ± 0.35
--
1.70 ± 0.04
62.2 ± 0.2
1.42 ± 0.25
2.39 ± 0.35
5.22 ± 0.06
8.13 ± 0.62
NA***
6.70 ± 0.25
--
Ascorbic Acid
1
2
3
4
5
6
93.7 ± 0.5
97.5 ± 0.2
68.5 ± 0.5
71.9 ± 4.7
36.2 ± 2.7
57.9 ± 9.5
0.22 ± 0.02
0.03 ± 0.03
1.68 ± 0.03
1.68 ± 0.37
3.08 ± 0.24
2.05 ± 0.49
14.85 ± 0.62
14.76 ± 0.47
13.36 ± 0.27
12.65 ± 0.34
NA****
2.33 ± 0.18
15.34 ± 0.56
*Calculated as the slope of the lineal part of the time course plot fluorescence intensity vs. time. Figure S8 in Supplementary
Information.
*
*
*
*All values in µM concentration.
** No radical inhibition detected when measured up to 64 µM concentration.
***No spectroscopic changes observed upon addition of Cu solution.
2+
2
.7. Inhibition of Cu -induced and self-induced Aβ(1-40) aggregation
+
2
Amyloid-β aggregation is widely accepted as one of the major hallmarks of AD. According to the
amyloid cascade hypothesis, the imbalance between production and clearance of Aβ peptides first

leads to the formation of oligomeric aggregates that end up organizing in fibril aggregates, these
being the main constituent of Aβ plaques [6,7]. In addition, some evidence suggests that pH affects
2
+
the Cu -induced Aβ aggregation into neurodegenerative forms of the protein. It has been reported
2
+
that Cu -catalyzed aggregation is more significant at slightly acidic pH (6.6-6.8) than at 7.4 [78].
This fact is important given that cerebral acidosis (altered proton homeostasis) seems to be
2
+
recurrent in AD [79]. Thus, the ability of 1-6 to inhibit both the Cu -induced and self-assembly
of Aβ(1-40) aggregation was tested and compared with that of resveratrol at pH=6.60. Solutions
of Aβ(1-40) were incubated up to 10 days with or without 2 eq. of compounds 1-6 in presence or
2
+
absence of 1 eq. of Cu . The amount of aggregates was then measured after 3, 7 and 10 days by
means of the thioflavin-T (ThT) and turbidity assays. The ThT assay is used as a key diagnostic
for the formation of mature Aβ fibril aggregates. On the other hand, the turbidity assay measures
light scattering due to the presence of protein aggregates in solution and, therefore, it can determine
the total amount of aggregates, including both the initial amorphous oligomeric form and mature
fibril aggregates that form Aβ plaques. However, as we have previously described [39], and in
agreement with reports by others [80], turbidity results can be taken as evidence of mainly
oligomeric aggregates. Whether the oligomeric or the fibril is the most toxic form of Aβ aggregates
still remains controversial [81]. Figure 4 shows results when ThT fluorescence and turbidity were
monitored up to 10 days. Several interesting observations can be made when turbidity and ThT
assays results are taken together: 1) As previously reported by us [39] and others [82-84] the
2
+
presence of Cu catalyzes the formation of amorphous aggregates (when compared to Aβ self-
assembly) and inhibits the fibrilization of Aβ. This phenomenon translated into an increase in
turbidity and decrease of ThT fluorescence when Aβ(1-40) was treated with Cu2+ in our
experiments (see Figure 4), (see Table S1 in Supplementary Information for all values of

fluorescence and percent inhibition). 2) ThT and turbidity assay results for resveratrol are
comparable to what us [39] and other researchers have published [80,85]. 3) ThT results showed
2
+
that although all compounds seemed to initially promote the fibril aggregation of Cu -Aβ(1-40)
2+
(
increase of fluorescence at 3 days of incubation with respect to Cu -Aβ(1-40) control), overall
2
+
inhibition of Cu -catalyzed aggregation in its fibril form was observed after 10 days for all
compounds between a ~52% and a ~77%. The initial increase was attributed to the complexation
2
+
of Cu and partial restoration of Aβ(1-40) aggregation profile, while the latter decrease can be
explained by the ability of compounds 1-6 to destabilize fibril forms of Aβ(1-40). 4) Turbidity
results revealed that, with the exception of compound 6, compounds 1-5 also lowered the amount
of amorphous aggregates between 18% to 52% after 10 days, a fact that was differential with
respect to resveratrol and our previous family of ionophoric polyphenols. Thus, comparison of 1-
5
with resveratrol and with the previous family of ionophoric polyphenols leads us to postulate
that our multi-target compounds prevent Aβ(1-40) aggregation by mechanisms different from
2
+
both, and that the observed anti-amyloidogenic effect is probably a consequence of Cu removal
from Aβ(1-40), as well as fibril destabilizing interactions of 1-5 (or their respective Cu2+
complexes) with the peptide. These effects resulted in a decrease of all forms of Aβ(1-40)
aggregates. In general, we also observed that compounds 1-4 displayed a better ability to inhibit
2
+
Cu -catalyzed aggregation than compounds 5 and 6, suggesting that the presence of the
polyphenol moiety in 1-4 might play an important role.

a
500
400
300
200
100
0
3 days incubation
7 days incubation
10 days incubation
Compound
b
0
0
0
0
0
.08
.06
.04
.02
.00
3
7
1
days incubation
days incubation
0 days incubation
Compound
2
+
Figure 4. Inhibition of Cu -induced Aβ(1-40) aggregation by compounds 1-6 and resveratrol
studied by ThT (a) and turbidity (b) assays. All values of fluorescence and percent inhibition of
fibril aggregation can be found in Table S1 in Supplementary Information.

2
+
Our results seem to suggest mechanism(s) additional to ion chelation to explain the Cu -induced
aggregation inhibition of compounds 1-5, thus providing a rationale to further investigate the
ability of the multi-target compounds to inhibit Aβ self-induced aggregation in the absence of
metal ions. The ability of 1-6 and resveratrol to inhibit self-induced Aβ(1-40) aggregation was also
studied by the ThT and turbidity assays. A solution of Aβ(1-40) was incubated for 3-10 days in
the absence or presence of 2 eq. of compounds 1-6 or resveratrol at pH=6.60 for the purpose of
2
+
comparison with the Cu -induced experiment. Results in Figure 5 reveal that compounds 1-4 and
display a moderate inhibition ability of the Aβ(1-40) self-assembly into fibril aggregates ranging
6
between ~6% and a ~31% of the Aβ(1-40) after 10 days of incubation (see Table S2 in
Supplementary Information for all values of fluorescence and percent inhibition). On the other
hand, compound 5 seems to promote the fibril aggregation after 7 and 10 days of incubation,
although we cannot explain this different behavior with the data at hand. Turbidity results show a
reverse trend to that of ThT fluorescence. Compounds 1-4 and 6 seem to slightly promote
amorphous aggregation between a ~6% and a ~33% extent, while compound 5 displays a minor
2
+
inhibition. Resveratrol follows a parallel pattern to that observed in presence of Cu , suggesting
that its effect on Aβ(1-40) aggregation is metal ion independent, in good agreement with its low
metal binding affinity.