Donepezil-like multifunctional agents: Design, synthesis, molecular modeling and biological evaluation

Article abstract of DOI:10.1016/j.ejmech.2015.10.001

Currently available drugs against Alzheimer's disease (AD) are only able to ameliorate the disease symptoms resulting in a moderate improvement in memory and cognitive function without any efficacy in preventing and inhibiting the progression of the pathology. In an effort to obtain disease-modifying anti-Alzheimer's drugs (DMAADs) following the multifactorial nature of AD, we have recently developed multifunctional compounds. We herein describe the design, synthesis, molecular modeling and biological evaluation of a new series of donepezil-related compounds possessing metal chelating properties, and being capable of targeting different enzymatic systems related to AD (cholinesterases, ChEs, and monoamine oxidase A, MAO-A). Among this set of analogues compound 5f showed excellent ChEs inhibition potency and a selective MAO-A inhibition (vs MAO-B) coupled to strong complexing properties for zinc and copper ions, both known to be involved in the progression of AD. Moreover, 5f?exhibited moderate antioxidant properties as found by in?vitro assessment. This compound represents a novel donepezil–hydroxyquinoline hybrid with DMAAD profile paving the way to the development of a novel class of drugs potentially able to treat AD.

Full text of DOI:10.1016/j.ejmech.2015.10.001

Accepted Manuscript
Donepezil-like Multifunctional Agents: Design, Synthesis, Molecular Modeling and
Biological Evaluation
Ming-Yu Wu, Gerard Esteban, Simone Brogi, Masahi Shionoya, Li Wang, Giuseppe
Campiani, Mercedes Unzeta, Tsutomu Inokuchi, Stefania Butini, Jose Marco-
Contelles
PII:
S0223-5234(15)30283-X
10.1016/j.ejmech.2015.10.001
EJMECH 8136
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 6 May 2015
Revised Date: 9 September 2015
Accepted Date: 1 October 2015
Please cite this article as: M.-Y. Wu, G. Esteban, S. Brogi, M. Shionoya, L. Wang, G. Campiani, M.
Unzeta, T. Inokuchi, S. Butini, J. Marco-Contelles, Donepezil-like Multifunctional Agents: Design,
Synthesis, Molecular Modeling and Biological Evaluation, European Journal of Medicinal Chemistry
(2015), doi: 10.1016/j.ejmech.2015.10.001.
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.

ACCEPTED MANUSCRIPT
Donepezil-like Multifunctional Agents: Design, Synthesis,
Molecular Modeling and Biological Evaluation
a,§
b,§
c,§
a
a
c
Ming-Yu Wu, Gerard Esteban, Simone Brogi, Masahi Shionoya, Li Wang, Giuseppe Campiani,
b,*
a,*
c,∗
d,∗
Mercedes Unzeta, Tsutomu Inokuchi, Stefania Butini, and Jose Marco-Contelles
a
Division of Chemistry and Biotechnology, Graduate School of Natural Science and Technology,
Okayama University, 3.1.1 Tsushima-Naka, Kita-ku, Okayama 700-8530, Japan.
b
Departament de Bioquímica i Biologia Molecular, Facultat de Medicina, Institut de Neurociències,
Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain.
c
European Research Centre for Drug Discovery and Development (NatSynDrugs) and Department of
Biotechnology, Chemistry and Pharmacy, Università degli Studi di Siena Via A. Moro, 53100 Siena,
Italy.
d
Laboratory of Medicinal Chemistry (IQOG, CSIC), C/Juan de la Cierva 3, 28006-Madrid, Spain.
*
Corresponding author. Tel. +34 93 5811523; fax +34 935811573; e-mail: mercedes.unzeta@uab.cat
*
Corresponding author. Tel. +81862945045; fax +81862518021; e-mail: inokuchi@cc.okayama-u.ac.jp
∗
Corresponding author. Tel.: +39 0577234161; fax: +39 0577 234254; e-mail: butini3@unisi.it
∗
Corresponding author. Tel.: + 34 915622900; fax: +39 915644853; e-mail: iqoc21@iqog.csic.es
§
Authors equally contributed to this work
1

ACCEPTED MANUSCRIPT
Abstract
Currently available drugs against Alzheimer’s disease (AD) are only able to ameliorate the disease
symptoms resulting in a moderate improvement in memory and cognitive function without any efficacy
in preventing and inhibiting the progression of the pathology. In an effort to obtain disease-modifying
anti-Alzheimer’s drugs (DMAADs) following the multifactorial nature of AD, we have recently
developed multifunctional compounds. We herein describe the design, synthesis, molecular modeling
and biological evaluation of a new series of donepezil-related compounds possessing metal chelating
properties, and being capable of targeting different enzymatic systems related to AD (cholinesterases,
ChEs, and monoamine oxidase A, MAO-A). Among this set of analogues compound 5f showed
excellent ChEs inhibition potency and a selective MAO-A inhibition (vs MAO-B) coupled to strong
complexing properties for zinc and copper ions, both known to be involved in the progression of AD.
Moreover, 5f exhibited moderate antioxidant properties as found by in vitro assessment. This
compound represents a novel donepezil-hydroxyquinoline hybrid with DMAAD profile paving the way
to the development of a novel class of drugs potentially able to treat AD.
Keywords: MAO-A, MAO-B, Human Acetylcholinesterase, Human Butyrylcholinesterase, Molecular
Modeling, Molecular Docking, Alzheimer’s Disease, Metal Chelating Properties, Antioxidant
properties, ADME+T properties.
2

ACCEPTED MANUSCRIPT
1
. Introduction
Alzheimer’s disease (AD) represents the most prevalent pathology associated to neurodegenerative
senile dementia, characterized by selective loss of cholinergic neurons and reduced level of
acetylcholine (ACh) neurotransmitter, driving to memory deficit and progressive impairment of
cognitive functions up to debilitating dementia before death [1, 2]. This pathology has been found to be
related to many factors including ACh levels, β-amyloid plaques deposition, oxidative damage, free
radicals formation and metal ions [3-6]. AD currently affects millions of aging people (WHO report
2
012, Dementia: a public health priority) [7], a number which, with the greying of society, will lead to
a dramatic increase expecting a double of patients by 2030 and triple by 2050 [8]. To date, healthcare
or financial systems in developed countries are not prepared to manage this worrisome scenario [7].
Indeed the only available therapeutics for AD, focused on increasing cholinergic neurotransmission in
the brain (by inhibiting cholinesterases, ChEs), have resulted in a modest improvement in memory and
cognitive function and are devoid of any therapeutic potential concerning prevention and progression
of neurodegeneration [9]. All the currently registered drugs for the treatment of AD are ChEIs
(
donepezil, 1, rivastigmine, 2, and galantamine, 3, Figure 1). The N-methyl-D-aspartate (NMDA)
antagonist memantine was approved for the treatment of moderate to severe AD [10, 11], while tacrine,
(Figure 1) the first drug approved for the treatment of AD, was withdrawn from the market [12] due
4
to its hepatotoxicity [13, 14]. The narrowness of therapies for AD has become as the current biggest
unmet medical need in neurology unless an effective disease-modifying anti-Alzheimer’s drug
(
DMAAD) is found [15].
3

ACCEPTED MANUSCRIPT
Figure 1 Current cholinesterase inhibitors approved for the treatment of AD.
The multifactorial nature of AD has led to “the failure of the so-called one-drug-one-target paradigm”
[
16, 17]. Besides the well-recognized symptomatic efficacy of ChEs inhibition for the treatment of AD
a therapeutic potential was proposed for monoamine oxidase inhibitors (MAOIs). Indeed, among the
factors that may contribute to cell death in neurodegenerative disorders such as AD, oxidative stress
plays a relevant role. MAOIs can reduce the formation of neurotoxic products such as H O and
2
2
aldehydes, which in turn, promote the generation of reactive oxygen species (ROS) [18], endorsing
MAOIs with strong neuroprotective properties [19].
Amyloid beta (Aβ) aggregation and Aβ  plaques formation represent the hallmarks of AD. The
amyloid cascade hypothesis of AD postulates that excessive Aβ formation is one of the causes of
neurodegeneration. Furthermore, the metallostasis evidence points to iron, copper and zinc as relevant
metals in the pathogenesis of AD, either by direct interaction with Aβ, thus increasing its aggregation,
or by enhancing the production of H O and ROS. Consequently, the modulation of the homeostasis of
2
2
these biometals is a potential therapeutic strategy for AD therapy [20]. In healthy people amyloid
precursor protein and Aβ interact with the mentioned metal species in a subtle equilibrium. Disruption
4

ACCEPTED MANUSCRIPT
of this balance drives to an abnormal Aβ metabolism as well as an alteration in copper and zinc uptake,
leading to the formation of Aβ fibrils. The formation of fibrils may induce an inflammatory reaction
with consequent decrease of pH, also modifying the zinc homeostasis [21]. Zinc is thus released into
the cellular environment triggering the events that lead to oxidative stress-induced cell death. Copper
ions are then free to compete for zinc binding site on Aβ, and these ions can catalyse H O production
2
2
by Aβ. As a consequence of the inflammatory response, a highly reactive free radical, nitric oxide is
released to regulate the process [22].
Accordingly, in the last years the multifactorial etiology of AD has driven the research to develop
novel multitarget anti-AD agents [23-28]. In this frame, we have previously explored: i)
multifunctional compounds able to inhibit both ChEs [29-31], ii) compounds with multifunctional
profile coupling the inhibition of ChEs with Aβ aggregation as well as with the disruption of preformed
fibrils [32, 33], iii) compounds with acetylcholinesterase (AChE) and MAOs inhibition capacity [34,
3
5], and iv) compounds with AChE and MAOs inhibition capacity showing additional antioxidant
properties [36, 37]. Based on our knowledge and by a scaffold hopping approach, we herein report the
development of novel chemical entities (compounds 5a-f, Figure 2) which represent an improved series
of our previously developed analogues typified by DPH6 (Table 1) [38]. Compounds 5a-f combine the
N-benzylpiperidine moiety of 1 with a metal chelating portion represented by a 8-hydroxyquinoline
system functionalized at position 2. Since 1 is a weak MAO-B inhibitor and is devoid of MAO-A
affinity, for improving the MAO inhibition properties of the compounds, we also introduced the
propargylamine moiety.
5

ACCEPTED MANUSCRIPT
Figure 2. Design of the title compounds 5a-f.
Pharmacological analysis revealed the developed compounds as potent AChE and
butyrylcholinesterase (BuChE) inhibitors, with IC50 values in the nanomolar range combined with
micromolar MAO-A selective inhibition (vs MAO-B), metal-chelating capability and moderate
antioxidant properties. In-depth bioinformatics analysis of the analogues with AChE BuChE and
MAO-A and -B helped rationalizing the observed enzyme inhibition data. Among the analogues
described, compound 5f showed the most interesting profile as multifuntional tool, thus representing an
innovative scaffold for the development of improved donepezil-hydroxyquinoline hybrids as potential
DMAADs.
2
2
. Results and Discussion
.1. Chemistry
To obtain the compounds presented in this paper the following synthetic schemes 1-3 were adopted.
6

ACCEPTED MANUSCRIPT
ω-(1-Benzylpiperidin-4-yl)alkanals 10a,b were prepared in our previous methods [38]. Thus, two-
carbon homologation of the commercial ketone 6a and aldehyde 6b with Horner-Wadsworth-Emmons
reagent 7 (derived from Arbuzov reaction between triethyl phosphite and 1-chloroacetonitrile) gave
8
a,b. On the latter olefins double bond reduction with Mg-I in MeOH followed by DIBAL-H
2
reduction of the resulting alkanenitriles 9a,b gave the correspondent aldehydes 10a,b (Scheme 1).
The secondary amine, N-(2-propynyl)-N-(quinolin-2-yl)methylamine, 14 was prepared by reductive
amination of the quinoline-2-carbaldehyde 13 with propargylamine, using NaBH under acidic
4
conditions. For this process, 13 was derived from commercially available 2-methylquinolin-8-ol (11)
2
by SeO oxidation followed by silylation of 12 with TBSCl (Scheme 2).
Reductive amination of 6b and ω-(1-benzylpiperidin-4-yl)alkanals 10a,b with the secondary amine 14,
using NaBH under acidic conditions, afforded two component coupled tertiary amines 15a-c, which
4
were desilylated by treatment with Bu NF to give the desired donepezil + propargylamine + 8-
4
hydroxyquinoline hybrid compounds (DPHs) 5a-c (Scheme 3). DPHs 5d-f were prepared in short
synthetic sequences through Lewis acid-catalyzed Strecker reaction as the key operation [39]. Thus, the
2
aldehydes 6a, 10a and 10b were treated with TMSCN in the presence of ZnI , forming the
corresponding cyanohydrin TMS ethers 16a-c, which was immediately followed by addition of the
secondary amine 14 to give the corresponding α-aminonitriles 17a-c, respectively. Finally, the TBS-
protective group of 17a-c was removed by treatment with Bu NF in THF, affording the desired DPHs
4
5
d-f (Scheme 3). All new molecules showed analytical and spectroscopic data in good agreement with
their structures, and then submitted to biological evaluation as racemic mixtures.
7

ACCEPTED MANUSCRIPT
Scheme 1. Synthesis of aldehydes 10a,b. Reagents and
conditions: (a) K CO , THF, rt–reflux; (b) Mg/I ,
2 3 2
MeOH, 0 °C, rt; (c) DIBAL-H/toluene, -78 °C.
Scheme 2. Synthesis of intermediate
amine 14. Reagents and conditions: (a)
SeO /1,4-dioxane, 60 °C–reflux; (b)
2
propargylamine,
TBSCl/DCM,
imidazole,
°C–rt; (c)
0
CF COOH/MeOH, 0 °C–rt then NaBH ,
3
4
0
°C–rt.
8

ACCEPTED MANUSCRIPT
Scheme 3. Synthesis of title compounds 5a-f. Reagents and conditions: (a)
compound 14, CF CO H/MeOH, 0 °C–rt then NaBH , 0 °C–rt; (b) TBAF/THF, rt;
3
2
4
(
c) TMSCN, ZnI /DCM, rt; (d) compound 14, MeOH, 50 °C, 3 h.
2
2
2
.2. Pharmacological evaluation, structure-activity relationship, and molecular modeling studies
.2.1. Enzyme assays
®
Donepezil (Aricept ) is an AChE inhibitor able to increase concentrations of ACh in the brain and
improving the symptoms of disease without delaying the progression of AD. Donepezil was approved
®
by the FDA in 1996 with the trade name of Aricept and in 2006 the drug was also approved for
treating severe dementia. As shown in Table 1, donepezil displayed an IC value in the nanomolar
50
range regarding AChE activity inhibition with weak selectivity for other enzymatic systems (BuChE
and MAOs). In an effort to design multifunctional compounds as DMAADs the benzyl-piperidine
moiety of donepezil and 8-hydroxyquinoline nucleus, were combined for maintaining both the ChEs
inhibitory activity and metal-chelating properties. A propargylamine moiety for inhibiting MAO was
also inserted between these two moieties by a carbon chain of variable length. This approach led to the
9

ACCEPTED MANUSCRIPT
design of compounds 5a-f (Figure 2 and Table 1). Further in compounds 5d-f the spacer was decorated
with a cyano group as a further interacting group for potentiating enzyme inhibition. The obtained
compounds were evaluated in vitro for assessing their inhibition properties at the human enzymes of
interest (MAO-A, MAO-B, AChE, and BuChE). As shown in Table 1 the compound 5f exhibited the
most interesting profile of the series. In fact, 5f inhibited AChE activity in the nanomolar range with an
IC value of 29 nM, similar to that found for 1. Furthermore, our prototypic multifunctional compound
5
0
5
f showed an excellent inhibitory potency (IC = 39 nM) against the BuChE being more potent than
50
reference compound 1 (IC = 7.5 ꢀM). In addition 5f was found to be a selective MAO-A (vs MAO-B)
50
inhibitor in the micromolar range (IC50 = 10.1 ꢀM), whereas 1 did not show appreciable MAO-A
inhibitory activity and displayed MAO-B inhibition potency in the micromolar range (IC50 = 15.2 ꢀM).
DPH6 (Table 1) [38], the progenitor molecule of all these compounds, when compared with 5f, it
showed similar IC values versus MAO-A, but higher affinity versus MAO-B and lower inhibition
5
0
potency towards AChE and BuChE. Taken into account these data, we can assess that compound 5f
nicely challenged the in vitro profile of DPH6.
1
0

ACCEPTED MANUSCRIPT
Table 1. Inhibitory activities (IC50) and selectivity ratios of derivatives inhibiting human
recombinant MAO (hrMAO) isoforms and human recombinant ChEs.
a
a
Compound
IC50 (µM)
IC50 (µM)
b
c
hrMAO-A
hrMAO-B
SI
hrAChE
hrBuChE
SI
4
6.5 ± 4.4
>100
>2.2
9.7 ± 1.0
0.064 ± 0.005
152
5a
3
3
5.6 ± 3.7
7.0 ± 3.8
>100
>100
>2.8
>2.7
4.1 ± 0.6
1.5 ± 0.2
0.079 ± 0.013
0.023 ± 0.003
52
65
5
b
5c
1
1
1
0.2 ± 1.2
2.2 ± 1.3
0.1 ± 1.1
>100
>100
>100
>9.8
>8.2
>9.9
64.2 ± 7.1
0.030 ± 0.004
0.002 ± 0.0003
2140
1650
5
d
3.3 ± 0.6
5
e
0
.029 ± 0.003
0.039 ± 0.003
0.052 ± 0.002
(Ki)
0.7
2.9
0.015 ± 0.004
Ki)
5
f
(
Clorgyline
l-Deprenyl
(Donepezil)
0.016 ± 0.003
55.4 ± 6.5
>1000
65.5 ± 7.1
0.02 ± 0.003
15.2 ± 1.7
4094
0.0004
<0.0152
>500
>500
>500
-
-
>500
1
0.009 ± 0.001
7.5 ± 0.8
0.0012
6
.7 ± 0.8
11.8 ± 1.5
1.8
0.35 ± 0.11
2.1± 0.2
0.17
DPH6
a
Values obtained after 30 min pre-incubation expressed as the mean ± S.E.M. of at least three independent
b
c
experiments. MAO-A selectivity index: [IC (hrMAO-B)]/[IC (hrMAO-A)]. hrBuChE selectivity index:
50
50
[
IC50(hrBuChE)]/[IC50(hrAChE)].
11

ACCEPTED MANUSCRIPT
In order to provide some relevant information about the mechanism of action of 5f, an assessment to
establish the reversibility and time-dependent inhibition of hrMAO-A and hrChEs was performed. In
Figure 3 the output of these in vitro studies concerning MAO-A is shown.
Figure 3. Assessment for reversibility and time-dependent inhibition of hrMAO-A by derivative 5f.
A) hrMAO-A activity was measured after a 30 min pre-incubation with either still water as a control,
1
00 ꢀM 5f or 200 nM clorgyline (10-fold of the IC50 values) and followed by a large dilution of the
enzyme-inhibitor complex; B) dose-response curves of hrMAO-A inhibited by a 5f performed at
different pre-incubation times (0-240 min); C) IC50 values determined from dose-response curves of
hrMAO-A inhibition by 5f after different pre-incubations.
Unequivocally, the lack of recovery of activity after enzyme inactivation by 5f (Figure 3A) confirms an
irreversible MAO-A inhibition similarly to that observed with the reference inhibitor clorgyline. This
result is consistent with the covalent interaction between the propargylamine moiety and the flavin
adenine dinucleotide (FAD) co-factor present in the MAO-A enzyme. This type of inhibition is also in
agreement with molecular modelling studies discussed below (see paragraph 2.2.2 for further details).
Moreover, a time-dependent inhibition mechanism of hrMAO-A (Figure 3B and 3C) was evidenced
from the dose-response curves (IC ) when different pre-incubations times were employed.
5
0
1
2

ACCEPTED MANUSCRIPT
Figure 4. Kinetic assessment of 5f as hrChEs inhibitor. Lines were derived from a
weighted least-squares analysis of data. Acetylthiocholine iodide (ASCh) and S-
butyrylthiocholine iodide (BuSCh) were used as substrates.
For better characterizing compound 5f inhibition properties at ChEs, kinetic assessments were
performed as reported in Materials and Methods section. The results of this study are summarized in
Figure 4. The assessment was performed by means of steady-state inhibition (Lineweaver-Burk plots of
initial velocity) of hrAChE (Figure 4A) and hrBuChE (Figure 4B) by 5f (0-200 nM) using
acetylthiocholine iodide and S-butyrylthiocholine iodide (0.1-2 mM) as substrates, respectively. The
data obtained by kinetic studies clearly showed a mixed-type inhibition for both enzymes since
m
decreasing K values and increasing Vmax values were determined as higher 5f concentrations were
used (Figure 4C). Moreover, non-linear analyses allowed to calculate inhibition constants (Ki) for 5f
inhibiting hrChEs, and values of 14.9 ± 4.4 nM and 52.5 ± 16.6 nM were determined for hrAChE and
hrBuChE, respectively.
1
3

ACCEPTED MANUSCRIPT
2
.2.2. Structure-activity relationship and molecular modeling studies
Computational studies allowed us to rationally explain the experimentally determined different
selectivity degrees of compounds 5a-f on the hrMAO-A and -B enzyme isoforms. The structural
requirements governing the inhibitory potency of the title compounds on hrAChE were also
investigated exploiting our experience in the field [29, 30, 40]. The potential binding modes of a subset
of the analyzed compounds were found by applying the Induced Fit Docking (IFD) protocol [41]
available in Schrödinger Suite [42] as already reported by us [31, 32]. The mentioned protocol involves
various steps such as ligands and proteins preparation and molecular docking. In particular, in order to
improve the reliability of our calculations, the ligands were prepared taking into account their
protonation state at physiological pH (7.4) [33, 43-46]. In agreement with the literature data and
software output, we have considered the 8-hydroxyquinoline moiety in its zwitterionic form, and the
piperidine nitrogen as protonated (Table S1). Analysis of the data suggested that the presence of a CN
group is relevant for improving MAO-A activity and selectivity (5d vs 5a, 5e vs 5b, and 5f vs 5c, Table
1
). For verifying this issue we have performed docking calculations for 5f and 5c in complex with
hMAO enzymes. Results are depicted in Figure 5 for 5f S-enantiomer (docking calculations of R-
enantiomer were provided in Supporting Information in Figure S1) and in Figure 6 for 5c. As shown in
Figure 5A, compound (S)-5f interacts in the active site of MAO-A with a combination of polar contacts
and hydrophobic interactions, targeting the key residues of the protein [47]. In particular, (S)-5f
establishes a series of H-bonds with Lys305, Tyr444 and Phe208 (backbone), and a π-π stacking with
Phe208 and with Phe352 (Figure 5A). In line with the experimentally determined irreversible mode of
inhibition of 5f (Figure 3), the propargylamine moiety of (S)-5f was found to lay in close proximity to
the FAD molecule (3.3 Å). Accordingly, (S)-5f showed satisfactory docking score and free binding
energy values (GlideXP score -10.82 kcal/mol, and ꢁGbind = -180.03 kcal/mol, respectively) which is
consistent with good affinity for the isoform A of MAO enzyme. Notably, we observed a partial
1
4

ACCEPTED MANUSCRIPT
stereoselective interaction with the enzyme. The R-enantiomer (Figure S1A) showed indeed a different
pattern of interaction with respect to those found for the S-enantiomer. Although free binding energy
values (GlideXP score and ꢁGbind of -8.54 kcal/mol and -176.16 kcal/mol) were found less favorable
with respect to those obtained for the S-enantiomer, the different binding mode of (R)-5f with MAO-A,
could prevent strong interactions with the enzyme. In fact, (R)-5f can establish only a π-π stacking with
Phe173 and polar contacts with Val210 and Thr336. However, the propargylamine moiety lies far from
the FAD molecule (14.1 Å) thus precluding a potential covalent interaction with the enzyme. These
findings are in agreement with the micromolar inhibition potency found for the racemic mixture (the S-
enantiomer showed an excellent predicted affinity with a binding mode that could allow a correct
interaction within the enzyme, while the R-enantiomer appeared to possess a reduced predicted
affinity).
1
5

ACCEPTED MANUSCRIPT
Figure 5. Docked pose of (S)-5f into hMAO-A (PDB ID: 2Z5X) (panel A) binding site and into
hMAO-B (PDB ID: 2V5Z) (panel B). In the bottom part of the picture is represented a schematic
ligand-protein interaction. The docked poses are visualized by means of PyMOL (The PyMOL
Molecular Graphics System, v1.6-alpha; Schrodinger LLC, New York, NY, 2013), while the ligand-
protein interaction diagram was generated by means of Maestro.
On the contrary, the IFD results obtained for (S)-5f revealed a poor affinity for MAO-B as depicted in
Figure 5B. Due to the different arrangement of the enzyme active sites, (S)-5f could not engage a
suitable pattern of interaction with the MAO-B binding site. In particular, the replacement of Ile335
1
6

ACCEPTED MANUSCRIPT
and Ile180 (MAO-A) with Tyr326 and Leu171 (MAO-B) respectively appeared to limit cavity
accessibility for our inhibitors. In addition, the replacement of Phe208 (MAO-A) with Ile199 (MAO-B)
did not assure a correct accommodation of (S)-5f into MAO-B active site (Figure 5B). Consequently,
the compound produced only a stacking with Trp119 that is located at the boundary of the binding
pocket and no other contacts along the cleft were evident (Figure 5B). Moreover, the 8-
hydroxyquinoline moiety appeared to be solvent exposed and the distance between the propargylamine
and FAD is very large (8.74 Å) and not consistent with any potential covalent interaction. All these
data and free binding energy calculation (GlideXP score -7.62 kcal/mol, and ꢁGbind is -127.49
kcal/mol) are in line with the observed MAO-A and -B selectivity. For, the (R)-5f (Figure S1B) we
obtained similar results and no stereoselective interactions with MAO-B were detected (GlideXP score
-
8.04 kcal/mol and the ꢁGbind -130.81 kcal/mol). Similarly to the S-enantiomer into MAO-B with the
R-enantiomer the distance of the propargylamine group from FAD (6.7 Å) did not assure any chance of
covalent interaction thus well explaining the selectivity profile of 5f for the two MAO isoforms.
For evaluating the influence of structural differences in MAO-A inhibition we have performed IFD
studies also for 5c for comparison with data obtained for 5f. Compound 5c, lacking the CN group,
showed a decreased MAO-A inhibition activity when compared to 5f, and the molecular docking
output (Figure 6A) evidenced a slight different pattern of interaction with respect to 5f. Indeed, 5c
could reproduce contacts with Lys305 and Phe208 (H-bonds), and could establish a π-π stacking with
Phe352, but the absence of the CN group did not allow interaction with Tyr444 and the π-π stacking
with Phe208 as for 5f (Figure 6A). Moreover, the propargylamine group was found at 4.4 Å distance
from the FAD molecule a larger distance than that found for 5f (3.3 Å). All these evidences clearly
demonstrate that the presence of the CN group is relevant for a better interaction with MAO-A binding
site (5c indeed showed a slight decrease of docking score value and estimated free binding energy with
respect to 5f) and for higher inhibition potency (Table 1).
1
7

ACCEPTED MANUSCRIPT
Figure 6. Docked pose of 5c into hMAO-A (PDB ID: 2Z5X) (panel A; Docking score -9.96 kcal/mol;
ꢁ
Gbind = -174.29 kcal/mol) binding site and into hMAO-B (PDB ID: 2V5Z) (panel B; Docking score -
7
.56 kcal/mol; ꢁGbind = -131.68 kcal/mol). In the bottom part of the picture is represented a schematic
ligand-protein interaction.
Similarly to 5f, also 5c showed a poor MAO-B inhibition potency. Our docking results (Figure 6B)
showed that 5c is able to form only a H-bond with Tyr435 and a π-π stacking with Phe168 in a similar
fashion to 5f. Since the stacking with aromatic residues located at the boundary of the MAO-B binding
site prevented a deep accommodation of the compound into the cleft, precluding strong interactions of
1
8

ACCEPTED MANUSCRIPT
5
c within the active site, the propargylamine group was found at 5.8 Å from FAD. All together these
features led to a decrease of docking score values and of the estimated binding affinity in line with the
poor MAO-B inhibition potency of 5c (Table 1).
Overall this series of compounds (5a-f) was found to inhibit the hrChEs from the micromolar to the
nanomolar range and the compounds were in general more potent inhibitors of hrBuChE than hrAChE.
Analysis of the data reported in Table 1 evidenced that hrAChE inhibition potency of the compounds is
more sensible to small variations of the structure with respect to hrBuChE inhibition potency and this is
in line with the wider gorge dimensions of the BuChE enzyme.
In order to better explain this behavior we performed IFD calculations for all the compounds within
hAChE (PDB ID:1B41). The results (Figures 7 and 8) highlighted the importance of the length of the
molecule for correct interaction with the enzyme since potency increased with the length of the
compounds in both the 5a-c and the CN containing 5d-f sub-series. The lengthiest compounds 5c and
5
f were indeed found to be the most potent hrAChE inhibitors of the series since they possess the
correct distance between the aromatic moieties and an appropriately positioned basic center for
interacting with the peripheral anionic site (PAS), mid-gorge, and catalytic site (CAS). In addition, the
length of 5f (S- and R-enantiomer) also assured a better accommodation of the CN group into the gorge
of the enzyme with respect to the 5d and 5e (Figure 8 and Figure S2).
1
9

ACCEPTED MANUSCRIPT
Figure 7. Docked poses of selected compounds (panels A-C; A: 5a; B: 5b; C: 5c) into hAChE enzyme
(PDB ID: 1B41). The pictures were generated by means of PyMOL, while the ligand-interaction
diagrams were generated by means of Maestro.
In particular, the obtained binding mode of compound 5a (Figure 7A) clearly revealed its impossibility
to reach the CAS of hAChE. In fact, 5a established a series of limited hydrophobic and polar contacts
only with the PAS (Trp286 and Tyr72). Accordingly, the docking score and the ꢁGbind of 5a are -7.91
kcal/mol and -84.26 kcal/mol, respectively. A moderate improvement in the inhibitory activity was
observed with the homologation of linker between the propargylamine moiety and piperidine group. In
fact, compound 5b displayed a better accommodation into the binding site of hAChE (Figure 7B),
interacting by a π-π stacking with Trp286 the PAS region with its 8-hydroxyquinoline moiety, and can
engage a polar contact with the mid-gorge Asp74. These interactions account for the docking score and
2
0

ACCEPTED MANUSCRIPT
the ꢁGbind of -8.91 kcal/mol and -102.22 kcal/mol, respectively. A further homologation was beneficial
for activity and 5c strongly interacted with PAS, mid-gorge and CAS recognition sites (Figure 7C). In
particular, 5c established a double π-π stacking with Trp286 and Tyr72 in the PAS and a polar contact
with the mid-gorge region targeting the key residue Asp74. Notably, 5c was also able to interact with
the CAS by a π-π stacking with Trp86 by its benzyl moiety (Figure 7C). Also the docking score (-9.95
kcal/mol) and the ꢁGbind (-112.08 kcal/mol) supported the observed improvement of affinity of 5c vs
5
a for hrAChE.
The same trend was observed with the introduction of CN group on the molecules 5d-f (Figure 8). In
fact, 5f displayed the correct length and spatial arrangements of its features for appropriate interaction
with hAChE. In fact, the CN group is perfectly tolerated in 5e and 5f establishing a series of relevant
polar interactions with Phe295 or/and Arg296 (backbone) as reported in Figure 8B,C and in Figure S2,
accounting for the favorable docking score and ꢁGbind. Notably, and in line with the IC values (Table
50
1
), our computational analysis also revealed that the mid-gorge Asp74 is targeted by both enantiomers
of 5f and by only one enantiomer of 5e.
2
1

ACCEPTED MANUSCRIPT
Figure 7. Docked poses of S-enantiomer of selected compounds (A-C; A: 5d; B: 5e; C: 5f) into hAChE
enzyme (PDB ID: 1B41). The docking output for the R-enantiomers is provided in Figure S2. The
pictures were generated by means of PyMOL, while the ligand-interaction diagrams were generated by
means of Maestro.
Our calculations also explained the worst hrAChE inhibition properties of 5d (S- and R-enantiomer)
which did not establish interactions at the mid-gorge and CAS level as observed for the analogue 5a
devoid of the CN group. 5d only interacts with PAS by a H-bond with Tyr72 by its 8-hydroxyquinoline
moiety (Figure 8A and Figure S2A). Notably, a non-classical conformation of Trp286 was detected
which supported the poorest affinity of 5d for hrAChE. Also the docking score values (S-enantiomer -
7
7
.58 kcal/mol; R-enantiomer -7.33 kcal/mol) and ꢁGbind (S-enantiomer -79.42 kcal/mol; R-enantiomer -
8.15 kcal/mol) are in agreement with a 6-fold lower potency with respect to 5a. As above reported for
compound 5b, the elongation of the spacer in this scaffold improved hrAChE inhibition potency. In
2
2

ACCEPTED MANUSCRIPT
fact, 5e (S- and R-enantiomer, Figure 8B and Figure S2B, respectively) inhibiting hrAChE in the
micromolar range revealed a better pattern of interaction than 5d. In fact, 5e established a series of
contacts with PAS and mid-gorge, but could not reach the CAS due to the lack of appropriate length of
the linker. In particular (S)-5e established a π-π stacking by its aromatic moieties with Trp286 and with
Tyr337 (Figure 8B). In addition, the CN group interacts with the backbone of Phe295 by a H-bond. The
improvement of affinity by one order of magnitude was confirmed by the presence of a stronger pattern
of interaction with hAChE and a satisfactory docking score (-9.46 kcal/mol) coupled to a significant
ꢁ
Gbind of -109.81 kcal/mol. In addition, the R-enantiomer lacks the interaction with Tyr337 but it forms
a H-bond with the mid-gorge site Asp74 (Figure S2B). A comparable docking score (-10.34 kcal/mol)
and ꢁGbind (-107.60 kcal/mol) were found for the R-enantiomer with respect to the S-enantiomer.
Finally, the IFD calculation performed on 5f, possessing the appropriate length for optimally targeting
all the key interaction sites of hAChE gorge (Figures 8C and S2C), revealed a good fitting within the
enzyme. In particular, (S)-5f was able to establish a double π-π stacking by its 8-hydroxyquinoline and
benzyl groups with Trp286 and Trp86/Tyr337, respectively. Moreover, a series of H-bonds were found
with Phe295 by the CN group, and with Asp74 by the piperidine moiety. This enantiomer showed a
very high docking score (-12.93 kcal/mol) as well as the estimated free-binding energy (ꢁGbind -129.67
kcal/mol). Similarly, the R-enantiomer (Figure S2C) was able to reproduce all the contacts found for
the S-enantiomer (Figure 8C) with few additional contacts (H-bonds with the backbone of Arg296 by
its CN group, with Tyr72 sidechain by its 8-hydroxyquinoline moiety, and a π-π stacking with His447
by the benzyl moiety, Figure S2C). These additional contacts led to a significant improvement in the
docking score (-14.87 kcal/mol) as well as in the ꢁGbind (-146.31 kcal/mol). This strong pattern of
interaction found for both 5f enantiomers perfectly parallels and explains the in vitro data (Table 1).
2
3

ACCEPTED MANUSCRIPT
Last, but not least, all the compounds showed an excellent drug-like profile (ADME+T
properties prediction) as calculated by means of QikProp [48] (Table 2).
Table 2. Calculated physicochemical properties for compounds presented in this
study by means of QikProp application.
a
b
c
d
e
Cpd
SASA QPlogP QPlogS QPlogHERG
%HOA
5
a
-8.21
-8.90
-8.89
-8.23
-8.87
-8.97
88.0
743.9
3
.78
-3.37
5b
95.4
84.7
75.2
84.6
84.2
800.9
4
.84
-4.59
-4.91
5
c
827.7
757.1
818.1
851.7
5.19
3.03
4.23
4.50
5d
-3.80
-5.18
-5.61
5
e
5
f
The calculated properties and their recommended range according to QikProp user manual are follows:
a
b
c
d
e
SASA 300-1000; QPlogP -2.00 – 6.5; QPlogS -6.5 – 0.5; QPlogHERG <-5; %HOA: %human oral
absorption >80% high, <25% poor.
2
.2.3. Assessment of metal-chelating properties of 5f
Our investigation of the relevant properties of compound 5f as a potential DMAAD was implemented
with the in vitro assay for establishing its metal-chelating properties. Chelating tests reported in Figure
9
allowed to understand the metal-chelating properties of 5f evaluating the formation of metal-
complexes. Indeed, the ability of 5f to interact with metals relevant for Aβ misfolding such as zinc,
copper, and iron was assessed [49].
2
4

ACCEPTED MANUSCRIPT
Figure 9. Assessment of metal-chelating properties of compound 5f. UV-VIS spectra of solutions
containing 10 ꢀM 5f shortly incubated with increasing concentrations (0-5 ꢀM) of metals ZnSO
4
(A),
CuSO (B) and Fe (SO (C). Determination of the stoichiometry of complexes 5f-Zn(II) (D) and 5f-
4
2
4 3
)
Cu(II) (E) by using the Job’s method at 262 nm. All solutions were prepared in still water and experiments
were performed at room temperature.
2+
2+
In particular, the influence of abnormally high levels of redox-active metal ions such as Cu , Fe was
characterized as well as the compromised regulation of redox-active metals in the promotion and
formation of cytotoxic ROS and neuronal damage [50]. Moreover, the role of mentioned metals in
formation and progression of aberrant Aβ species causing the misfolding of proteins and their
subsequent aggregation was also established. Numerous studies have demonstrated that a surplus of
2
+
2+
2+
3+
metal ions such as Cu , Fe , Zn and Al were present in Aβ plaques of AD patients’ brain [49].
2
+
2+
Furthermore an excess of Cu and Zn promotes Aβ aggregation [51, 52]. For these reasons, the
research with multitarget compounds additionally decorated for obtaining metal-chelating effect might
provide an additional and therapeutic, rather than only symptomatological, strategy for the treatment of
AD. Our in vitro analyses (Figure 9) reveal that 5f is able to interact with metals. Additionally, an
-1
-1
extinction coefficient (ε) of 20800 M ·cm was determined with 5f, while a maximum complexation
2
5

ACCEPTED MANUSCRIPT
peak was detected at 262 nm. As showed in Figure 9A, 5f displayed a spectral changes at 10 ꢀM
2
+
2+
forming a complex with Zn after incubations of 15 h. Remarkably, 5f formed a complex with Cu
3
+
(
Figure 9B) after 5 minutes of incubation. On the contrary, with Fe (Figure 9C) no significant spectral
changes were observed at 262 nm after longer pre-incubations. In summary, 5f was determined as
2
+
metal-chelating agent, a stoichiometry complexation of 1:2 metal-compound with both Cu (Figure
2
+
9
D) and Zn (Figure 9E) revealed a metal-compound ration of 1:2.
2
.2.4. Evaluation of antioxidant properties of 5f
The assessment of antioxidant properties of compound 5f was carried out by different in vitro assays.
Figure 10A shows the Cu(II)-chelating activity of 5f at blocking the production of H O in vitro by
2
2
Cu(II) and ascorbic acid using the Amplex Red reagent. This effect was in agreement with the metal-
chelating properties previously observed as a stoichiometry of 1:2 Cu(II)/compound was also found.
Compound 5f also displayed oxygen radical absorbance capacity (ORAC) values equivalent to those
observed with the antioxidant Trolox (Figure 10B). Lipid peroxidation induced in vitro by linoleic acid
and 2,2'-azobis(2-amidinopropane) dihydrochloride (AAPH) (Figure 10C) was significantly reduced by
5
f though this effect was prolonged similarly (Tinh) over time as that observed in the control. Finally,
compound 5f was shown to induce weak radical scavenger activity (RSA) compared to trolox by using
,2-difenil-1-picrilidrazile (DPPH, Figure 10D). Although a dose-dependent effect was found with 5f,
2
being significant from concentrations higher than 50 ꢀM, no similar antioxidant response was reached
compared to that of 50 ꢀM trolox at any of the concentrations assayed with 5f.
2
6

ACCEPTED MANUSCRIPT
A
3000
500
000
500
000
00
B 3.0
2
.51 ± 0.15
2
2
.5
.0
1.5
.0
.5
.0
2
2
1
.12 ± 0.43
1
***
1
1
5
0
0
0
5
f
Melatonin
trolox
[
5f](µM)
C
D
1
20
Tinh=40-60
120
100
**
1
00
***
***
80
Tinh=100-120
8
0
Tinh=50-60
60
60
40
40
***
20
20
0
0
control
5f
trolox
[5f](µM)
Figure 10. Evaluation of in vitro antioxidant properties of compound 5f.
A) Effect of compound 5f on the in vitro production of H in presence of Cu(II) and ascorbic acid
2
O
2
expressed in fluorescence units (FU); B) ORAC-FL values of 5f and melatonin as trolox
equivalents; C) inhibition of lipid peroxidation (LPO) by compound 5f and trolox (inhibition times
(Tinh) are expressed in min); D) radical scavenger activity (RSA) induced by DPPH with compound
5
f and 50 ꢀM trolox. Statistical analysis was performed using one-way ANOVA test followed by
Bonferroni post-test. **, p<0.01, ***, p<0.001.
3
. Conclusion
In summary, in this study, the development of novel donepezil-related compounds (5a-f) characterized
by a multifunctional profile is reported. Among the series of the novel chemical entities assayed,
compound 5f exhibited the most interesting profile to be further characterized as AD potential
therapeutic. In fact, 5f was defined as a mixed-type hrAChE and hrBuChE inhibitor of nanomolar
2
7

ACCEPTED MANUSCRIPT
potency, and was also found a selective irreversible MAO-A inhibitor in the micromolar range.
Molecular modeling studies were performed to rationalize the mode of action and the selectivity profile
of the developed compounds. These studies evidenced the structural requirements of these donepezil-
related analogues to optimally interact with the enzymes of interest. Additionally, 5f, conceived for
interacting with metals targeting the cytotoxicity derived from A
β
misfolding leading to oligomers
2
+
2+
formation, was demonstrated to be able to form complexes with Zn and Cu . These data confirm 5f
metal-chelating properties which might be relevant for modulating Aβ misfolding. Furthermore, 5f is
also endowed with some antioxidant properties. The calculated physico-chemical properties enrich the
profile of this promising hit as it shows a drug-likeness in line with other marketed drugs. Taken
together these findings will pave the way to the development of novel anti-AD based on 5f that
deserves to be further investigated as DMAAD.
4
4
4
. Experimental section
.1. Chemistry
.1.1. General procedures
Unless otherwise specified, materials were purchased from commercial suppliers and used without
1
further purification. Silica gel 60 (0.040 - 0.063 mm) were used for column chromatography. H NMR,
1
3
and C NMR, spectra were recorded on a Varian INOVA-600, Varian INOVA-400 or Varian Mercury
00MHz, spectrometers by using the residual signal of the deuterated solvent as internal standard.
Splitting patterns are described as singlet (s), doublet (d), triplet (t), quartet (q), quintet (p), and broad
br); the value of chemical shifts (d) are given in ppm and coupling constants (J) in Hertz (Hz). IR
spectra were obtained with a Shimazu, Model FT-IR 8400. Purity of the compounds was assessed by
3
(
2
8

ACCEPTED MANUSCRIPT
HRMS, spectra were recorded on a Bruker Daltonics Bio TOF mass spectrometer. Yields refer to
purified products and are not optimized. All moisture-sensitive reactions were performed under argon
atmosphere using oven-dried glassware and anhydrous solvents.
4
.1.2. 2-(1-Benzylpiperidin-4-ylidene)acetonitrile (8a
)
Diethyl (cyanomethyl)phosphonate (
7
). Chloroacetonitrile (3 g, 40 mmol) and triethyl phosphite (6.65
o
g, 40 mmol) were heated at 150 C for 3.5 h. And the crude product was directly used in the next
reactions.
A solution of diethyl (cyanomethyl)phosphonate 7 (2.13 g, 12 mmol) and K CO (1.66 g, 12 mmol) in
2
3
1
0 mL of dry THF was stirred for 15 min at rt. Then the mixture was heated to reflux for 20 min. After
cooling down to rt, ketone 6a (10 mmol of in 5 mL of dry THF) was added dropwise to this mixture
and then resulting mixture was heated at reflux for 12 h. After cooling down to rt, 10% K CO aqueous
2
3
solution was added. The reaction mixture was extracted with AcOEt three times, the organic phase
were collected, dried with anhydrous Na SO and concentrated in vacuum to obtain the crude product
2
4
which was directly used for the following reactions without further purification. Title compound was
1
obtained as yellow oil, 92.0% yield; H NMR (300 MHz, CDCl ) δ 7.26–7.34 (m, 5 H), 5.10 (s, 1 H),
3
3
.55 (s, 2 H), 2.51–2.63 (m, 6 H), 2.40 (t, J = 5.1 Hz, 2 H).
.1.3. 3-(1-Benzylpiperidin-4-yl)acrylonitrile (8b).
4
Following the procedure described for compound 8a the title compound was obtained as yellow oil,
1
9
1
1
5.0% yield; H NMR (300 MHz, CDCl ) δ7.19–7.38 (m, 5 H), 6.31 (t, J = 10.4 Hz, 1 H), 5.23 (d, J =
3
0.9 Hz, 1 H), 3.50 (s, 2 H), 2.89 (d, J = 11.7 Hz, 2 H), 2.51–2.71 (m, 1 H), 1.93–2.14 (m, 2 H), 1.62–
.77 (m, 2 H), 1.40–1.60 (m, 2 H).
4
.1.4. 2-(1-Benzylpiperidin-4-yl)acetonitrile (9a).
2
9