Design, biological evaluation and X-ray crystallography of nanomolar multifunctional ligands targeting simultaneously acetylcholinesterase and glycogen synthase kinase-3

Article abstract of DOI:10.1016/j.ejmech.2018.12.063

Both cholinesterases (AChE and BChE) and kinases, such as GSK-3α/β, are associated with the pathology of Alzheimer's disease. Two scaffolds, targeting AChE (tacrine) and GSK-3α/β (valmerin) simultaneously, were assembled, using copper(I)-catalysed azide alkyne cycloaddition (CuAAC), to generate a new series of multifunctional ligands. A series of eight multi-target directed ligands (MTDLs) was synthesized and evaluated in vitro and in cell cultures. Molecular docking studies, together with the crystal structures of three MTDL/TcAChE complexes, with three tacrine-valmerin hybrids allowed designing an appropriate linker containing a 1,2,3-triazole moiety whose incorporation preserved, and even increased, the original inhibitory potencies of the two selected pharmacophores toward the two targets. Most of the new derivatives exhibited nanomolar affinity for both targets, and the most potent compound of the series displayed inhibitory potencies of 9.5 nM for human acetylcholinesterase (hAChE) and 7 nM for GSK-3α/β. These novel dual MTDLs may serve as suitable leads for further development, since, in the micromolar range, they exhibited low cytotoxicity on a panel of representative human cell lines including the human neuroblastoma cell line SH-SY5Y. Moreover, these tacrine-valmerin hybrids displayed a good ability to penetrate the blood-brain barrier (BBB) without interacting with efflux pumps such as P-gp.

Full text of DOI:10.1016/j.ejmech.2018.12.063

Accepted Manuscript
Design, biological evaluation and X-ray crystallography of nanomolar multifunctional
ligands targeting simultaneously acetylcholinesterase and glycogen synthase
kinase-3
Killian Oukoloff, Nicolas Coquelle, Manuela Bartolini, Marina Naldi, Rémy Le Guevel,
Stéphane Bach, Béatrice Josselin, Sandrine Ruchaud, Marco Catto, Leonardo Pisani,
Nunzio Denora, Rosa Maria Iacobazzi, Israel Silman, Joel L. Sussman, Frédéric
Buron, Jacques-Philippe Colletier, Ludovic Jean, Sylvain Routier, Pierre-Yves Renard
PII:
S0223-5234(18)31109-7
DOI:
https://doi.org/10.1016/j.ejmech.2018.12.063
EJMECH 10997
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 16 October 2018
Revised Date: 21 December 2018
Accepted Date: 24 December 2018
Please cite this article as: K. Oukoloff, N. Coquelle, M. Bartolini, M. Naldi, Ré. Le Guevel, Sté. Bach,
Bé. Josselin, S. Ruchaud, M. Catto, L. Pisani, N. Denora, R.M. Iacobazzi, I. Silman, J.L. Sussman,
Fréé. Buron, J.-P. Colletier, L. Jean, S. Routier, P.-Y. Renard, Design, biological evaluation and X-
ray crystallography of nanomolar multifunctional ligands targeting simultaneously acetylcholinesterase
and glycogen synthase kinase-3, European Journal of Medicinal Chemistry (2019), doi: https://
doi.org/10.1016/j.ejmech.2018.12.063.
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

ACCEPTED MANUSCRIPT
Design, Biological Evaluation and X-ray Crystallography of Nanomolar
Multifunctional Ligands Targeting Simultaneously Acetylcholinesterase
and Glycogen Synthase Kinase-3
a
b,c
d
d
e
Killian Oukoloff, Nicolas Coquelle, Manuela Bartolini, Marina Naldi, Rémy Le Guevel,
f
f
f
g
g
Stéphane Bach, Béatrice Josselin Sandrine Ruchaud, Marco Catto, Leonardo Pisani, Nunzio
g
h
i
j
k
Denora, Rosa Maria Iacobazzi, Israel Silman, Joel L. Sussman, Frédéric Buron, Jacques-
b
*,a
k
*,a
Philippe Colletier, Ludovic Jean, Sylvain Routier, and Pierre-Yves Renard
a
Normandie Univ, UNIROUEN, INSA Rouen, CNRS, COBRA (UMR 6014), 76000 Rouen,
France
b
Institut de Biologie Structurale, Université Grenoble Alpes, Centre National de la Recherche
Scientifique (CNRS)-Commissariat à l’Énergie Atomique (CEA) (UMR 5075), F-38054
Grenoble, France
c
Institut Laue Langevin, 71, avenue des Martyrs - CS 20156, 38042 Grenoble cedex 9, France
d
Department of Pharmacy and Biotechnology, Alma Mater Studiorum University of Bologna,
Via Belmeloro 6, I-40126, Bologna, Italy
e
Plateforme ImPACcell-SFR BIOSIT UMS-CNRS3480 UMS-INSERM018, Université de
Rennes1, 35043, Rennes Cedex, France
f
Sorbonne Université, CNRS USR 3151, Protein Phosphorylation & Human Diseases, Station
Biologique de Roscoff, CS 90074, Roscoff Cedex F-29688, France
g
Dipartimento di Farmacia-Scienze del Farmaco, Università degli Studi di Bari "Aldo Moro",
via Edoardo Orabona 4, 70125 Bari, Italy
h
Istituto Tumori IRCCS Giovanni Paolo II, via le Orazio Flacco 65, 70124 Bari, Italy
i
Department of Neurobiology, Weizmann Institute of Science, 6100 Rehovot, Israel
j
Department of Structural Biology, Weizmann Institute of Science, 76100 Rehovot, Israel
k
Institut de Chimie Organique et Analytique, Université d’Orléans, UMR CNRS 7311, rue de
Chartres, BP 6759, 45067 Orléans Cedex 2, France
1

ACCEPTED MANUSCRIPT
Corresponding Author
*
*
For L. J.: phone: +33 2 35 52 24 51; E-mail: ludovic.jean@univ-rouen.fr
For P.-Y. R.: phone: +33 2 35 52 24 76; E-mail: pierre-yves.renard@univ-rouen.fr.
Keywords
Alzheimer’s disease, acetylcholinesterase, butyrylcholinesterase, kinases, glycogen synthase
kinase-3, cyclin-dependent kinase 5, synthesis, SAR, crystallography.
Abstract
Both cholinesterases (AChE and BChE) and kinases, such as GSK-3α/β, are associated with the
pathology of Alzheimer’s disease. Two scaffolds, targeting AChE (tacrine) and GSK-3α/β
(valmerin) simultaneously, were assembled, using copper(I)-catalysed azide alkyne
cycloaddition (CuAAC), to generate a new series of multifunctional ligands. A series of eight
multi-target directed ligands (MTDLs) was synthesized and evaluated in vitro and in cell
cultures. Molecular docking studies, together with the crystal structures of three MTDL/TcAChE
complexes, with three tacrine-valmerin hybrids allowed designing an appropriate linker
containing a 1,2,3-triazole moiety whose incorporation preserved, and even increased, the
original inhibitory potencies of the two selected pharmacophores toward the two targets. Most of
the new derivatives exhibited nanomolar affinity for both targets, and the most potent compound
of the series displayed inhibitory potencies of 9.5 nM for human acetylcholinesterase (hAChE)
and 7 nM for GSK-3α/β. These novel dual MTDLs may serve as suitable leads for further
development, since, in the micromolar range, they exhibited low cytotoxicity on a panel of
representative human cell lines including the human neuroblastoma cell line SH-SY5Y.
2

ACCEPTED MANUSCRIPT
Moreover, these tacrine-valmerin hybrids displayed a good ability to penetrate the blood-brain
barrier (BBB) without interacting with efflux pumps such as P-gp.
1. Introduction
Alzheimer’s disease (AD), the most common cause of senile dementia, is a major public health
concern with devastating social impact. In 2016, Alzheimer’s Disease International, the
worldwide federation of Alzheimer associations, estimated that 47 million people were affected
by this type of dementia, and predicted that 131 million people will be affected in 2050. Thus,
the economic impact of the disease on future long-term care costs will be enormous. Already in
2
018 the cost of treatment of dementia is estimated to reach US$ 1 trillion, rising to US$ 2
trillion by 2030.[1]
AD results from a neurodegenerative process occurring in the central nervous system (CNS). It
is clinically characterized by loss of memory and cognitive impairment, which parallel the
deterioration of cholinergic neurons in the basal-forebrain with concurrent reduced levels of the
neurotransmitter acetylcholine (ACh). The disease is histologically characterized by extracellular
deposits of β-amyloid peptide (Aβ) and intracellular formation of neurofibrillary tangles
(NFTs).[2, 3] Other hallmarks include oxidative stress, dys-homeostasis of biometals and
calcium, inflammation and loss of synaptic connections.[4] Current therapeutic options for
treatment of AD mostly aim at restoring physiological ACh levels by inhibiting the enzyme
acetylcholinesterase (AChE), which is responsible for the hydrolysis of ACh. Therefore, three
drugs approved by the Food and Drug administration (FDA) for the management of AD are
AChE inhibitors (AChEIs), namely donepezil, rivastigmine and galantamine.[5] However,
treatment with AChEIs is mostly symptomatic, and effective only for AD patients with mild-to-
moderate symptoms, enabling in some cases the progression of cognitive and functional
3

ACCEPTED MANUSCRIPT
impairments to be retarded. Memantine, which is an N-methyl-D-aspartate (NMDA) receptor
antagonist, is the latest drug approved for the treatment of the disease.[6] The efficacy of
memantine is similar to that of the AChEIs among patients with mild-to-moderate symptoms, but
has been reported to be more useful for treating patients at more advanced stages of the
disease.[7-9] Yet, all four drugs only provide temporary symptomatic relief, but do not provide a
cure. Despite efforts over the last few decades of both academic research teams and
pharmaceutical companies, no curative treatment has emerged yet.[10-12]
An attrition factor in the development of curative drugs is the fact that the aetiology of AD is
not fully understood. It is recognized, however, that the processes resulting in its development
and progression are complex and multifactorial.[13] Thus, it is conceivable to hypothesize that a
combination of synergetic strategies targeting the various players involved in the onset and
development of the disease should be more beneficial, and should present a higher rate of
success. The “multifactorial hypothesis” has fuelled the so-called multi-target-directed ligand
(MTDL) strategy, which foresees the development of a single molecule able to affect several key
targets/pathways.[14-22] Besides the advantage of acting simultaneously on more than one
biological target, MTDLs have the advantages of a higher patient compliance (a parameter of
paramount importance for such neurodegenerative disease), lower risk of drug-drug interactions,
and lower costs to perform ADMET studies.[23]
Kinases have been shown to be associated with the progression of AD, and have received
increased attention during the past decade.[24, 25] Some members of this family are involved in
the hyperphosphorylation of the tau protein, which results in the production of intracellular
NFTs. GSK-3β [26] and CDK5 [27] are the two main kinases involved in this process, whereas
GSK-3α appears to regulate the production of Aβ.[28, 29] Thus, GSK inhibitors may be useful
4

ACCEPTED MANUSCRIPT
for treatment of AD, as well as other neurodegenerative diseases.[30, 31] A study identified
novel mechanisms linking GSK-3 with the Aβ pathology. The inhibition of GSK-3 reversed AD
pathogenesis via lysosomal acidification and reactivation/restoration of the mammalian target of
rapamycin (mTOR) in a mouse model of AD.[32] Furthermore, evidence has been presented that
GSK-3 inhibitors may reduce Aβ-oligomer-induced neuronal toxicity and may promote
neurogenesis in vitro and in vivo.[33, 34] These various studies indicate that GSK-3 inhibitors
should be further studied as candidates for treatment of AD.[35]
A renewed interest in AChEIs for AD treatment arose when evidence was presented that, in
addition to AChE inhibition, certain AChEIs may block the involvement of AChE in mediating
the aggregation and deposition of Aβ peptides.[36-38] Thus, AChE has been reported to
accelerate the formation of neurotoxic Aβ aggregates via a mechanism that involves its
peripheral anionic site (PAS).[39] This resulted in numerous studies aimed at developing dual
binding site AChEIs interacting with both the catalytic site and the PAS.[21, 40, 41] Moreover,
recent studies have suggested a possible role for AChE in hyperphosphorylated tau (P-Tau)
dysregulation.[42] There is thus also still a great interest in developing MTDLs endowed with an
anti-ChE activity.
To combine the beneficial effects of ChE inhibition and GSK-3 inhibition within a single
molecule, a tacrine scaffold was selected to provide activity toward ChEs (Figure 1). This
selection was also based on the fact that the tacrine pharmacophore can easily be substituted on
its primary amino group while retaining a satisfactory anticholinesterase activity. Consequently,
many tacrine-based MTDLs exhibiting high affinity for AChE and/or BChE have been reported
in the past decade.[41, 43-45] However, only one study describing GSK-3/AChE bifunctional
5

ACCEPTED MANUSCRIPT
inhibitors has appeared.[46] These MDTLs presented good inhibitory activities towards both
enzymes, and alleviated cognitive impairment in the mouse model treated with scopolamine.
Amongst the GSK-3 kinase inhibitors developed for application to the CNS,[47] we selected
the valmerins (Figure 1), which contain a tetrahydropyrido[1,2-a]isoindolone core[48, 49] linked
to an heteroaryl moiety by a ureido group. The pharmacophore that interacts with the kinase
active site consists of the isoindolone-urea moiety linked to the pyridine ring at its C-2’ position.
NH
2
N
Tacrine
Br
N
O
O
N
N
N
O
N
N
O
H
H
H
H
N
N
I
IC50 GSK-3α/β = 170 nM
II IC50 GSK-3α/β = 22 nM
IC50 CDK5/p25 = 80 nM
IC50 CDK5/p25 = 25 nM
Figure 1. Structures of tacrine, a ChE inhibitor and valmerins I and II, which display potent
CDK5 and GSK-3α/β inhibition.[49]
Docking studies and SAR studies suggested that substitutions at the pyridine C4’, C5’ and C6’
positions should not adversely affect affinity for either GSK-3α/β or CDK5. Indeed, as seen in
Fig. 1, the presence of a bromine atom on C4’ actually enhances inhibitory capacity several
fold.[49] Thus, effective valmerin-tacrine based MTDLs could be obtained using either of these
positions to link the two functional groups. Linkage of the two pharmacophores could be
advantageously performed using the copper(I)-catalysed azide alkyne cycloaddition (CuAAC)
(
Figure 2) as an interesting convergent synthetic strategy for MTDLs. Furthermore, the
introduction of a 1,2,3-triazole ring within the linker connecting the two pharmacophores would
allow formation of favourable hydrogen bond interactions within the active-site gorge of
6

ACCEPTED MANUSCRIPT
AChE.[50-52] The crystal structure of the complex of mouse AChE (mAChE) with one of the
most active bifunctional AChEIs, anti-TZ2-PA6 (PDB 1Q84), which contains a triazole ring
attached to a tacrine moiety through a two-carbon linker, revealed additional hydrogen bonds
between the triazole moiety and well identified residues.[50] Taking these different factors into
consideration, we designed a series of MTDLs based on the skeleton depicted in Figure 2. These
hybrids present a single chiral centre on the tetrahydroisoindolone moiety. Since this moiety
could potentially interact with both of the targeted enzymes, for the most promising candidate
MTDLs, each enantiomer was synthesized and evaluated independently.
O
N
*
kinases inhibitor scaffold
HN
AChE peripheral site (PAS)
interactions
ATP binding site interactions
Val-135; Lys-85)
(
HN
O
(Trp-286)
N
4'
6'
5'
AChE hydrophobic aromatic gorge
interactions
interaction with the kinase
tunnel toward solvent
exposure
(Tyr-124, Tyr-341)
N
N
N
(Lys-135; Tyr-140)
NH
N
Cholinesterases inhibitor scaffold
AChE catalytic site interactions
(Trp-86; Tyr-337; His-447)
Figure 2. Design strategy of proposed MTDLs targeting both ChEs and GSK-3 kinases.
2. Results and discussion
2.1 Design and docking studies
7

ACCEPTED MANUSCRIPT
Before investing synthetic efforts in the preparation of a large number of MTDLs, we
performed modelling studies to determine the optimal length and chemical characteristics of the
linker connecting the two pharmacophores. These modelling studies took as a starting point the
crystal structure of the anti-TZ2-PA6 complex with mAChE that was referred to above (PDB
1
Q84) [50]. The active-site gorge of AChE is known to display some conformational flexibility.
Binding of bifunctional ligands can induce some side-chain rotations, e.g., of Trp286, and
movement of the protein backbone at the PAS and within the region of the gorge between the
PAS and the catalytic site.[50, 53-55] Taking into account this flexibility, we modelled eight
MTDLs containing different linkers between the triazole and the valmerin pyridine ring by
varying the length (0-3 methylene units) and the chemical nature of the linker (a simple alkyl
chain or a chain containing an oxygen atom or a tertiary amine). The docking studies were
performed with both enantiomers of the tetrahydropyridoisoindolone moiety, and the linker was
attached to the pyridine C-4’ atom since it had been earlier reported that this position allows
substitution without altering the binding activity for the targeted kinases.[56] The binding
energies for AChE of these MTDLs, as calculated by use of Autodock Vina,[57] are listed in
Table 1. Low binding energies, below -15.0 kcal/mol, were obtained for both enantiomers of all
the tested compounds, except for the S enantiomer of 1, which displayed a higher value, -12.8
kcal/mol. All docking poses showed a productive conformation of the tetrahydroacridine scaffold
within the active site of AChE (Figure 3), with an optimal double π-π stacking interaction with
Trp86 and Tyr337, and formation of a hydrogen bond with the backbone carbonyl group of
His447. Furthermore, it superimposed very well with that of anti-TZ2-PA6 (see additional
docking poses in supporting information). Moreover, the lower binding energy predicted for 2,
8

ACCEPTED MANUSCRIPT
relative to 3, could be explained by an additional π-π stacking interaction observed between
Trp286 and the valmerin pyridine ring.
4'
O
O
5
'
H
N
N
N
N
n
X
N
n'
H
6'
N
H
N
N
N
Table 1. Stabilization energies calculated for MTDLs 1-8.
Linker
Binding energy (kcal/mol)
position
Entry
Compound
X
n
n’
on
R enantiomer S enantiomer
pyridine
1
2
3
4
5
6
7
8
9
1
4’
4’
4’
4’
4’
4’
5’
6’
CH2
O
1
1
1
1
0
0
0
0
-
1
1
1
0
0
0
0
0
-
-14.9
-15.2
-14.3
-14.6
-15.8
-16.6
-16.0
-16.3
-12.8
-15.2
-14.7
-14.9
-15.6
-16.8
-16.2
-16.6
2
3
N-Me
CH2
CH2
Ø
4
5
6
7
Ø
8
Ø
anti-TZ2-PA6
-
-14.6
Figure 3. Molecular docking into hAChE of (R)-2 (left panel), (R)-3 (centre panel) and (R)-6
right panel).
(
9

ACCEPTED MANUSCRIPT
In addition, as reported for the crystal structure of the anti-TZ2PA6/mAChE complex,
molecular docking revealed potential hydrogen bonds between the 1,2,3-triazole moiety and
Tyr124 and Tyr337. The lowest stabilization energies were obtained for the two enantiomers of
6
. This could be explained by an additional π-π stacking interaction of the triazole with Tyr341.
These low stabilization energies could also be explained by hydrogen bonds formed between the
C=O of the urea moiety of the valmerins and Tyr286, and between the backbone carbonyl group
of Tyr341 and the NH of the urea. In addition, we modelled MTDLs 7 and 8 substituted
respectively at positions C5’ and C6’ on the pyridine ring of the valmerins. This docking study
also showed promising binding affinity for AChE.
As a further step, MTLDs 1-8 were docked into the crystal structure of GSK-3
K5N).[58] The lowest binding energy, -10.6 kcal/mol, was obtained for the S enantiomer of 6
Figure 4). In the docked structure, the tetrahydropyridoisoindole scaffold points toward the
hinge region, forming a hydrogen bond between Val135 and the C=O of the isoindolone.[56] In
addition, a hydrogen bond between Lys183 and triazole N-3, and a cation- interaction between
β (PDB code
5
(
π
Lys183 and the pyridine ring, were observed. As reported in our earlier SAR studies,[48, 56, 59]
better binding was predicted for compounds whose linker is attached to C4’ of the pyridine ring.
Indeed, docking poses showed that the tacrine scaffold is oriented towards the solvent, thus
minimizing interaction with the kinase. Moreover, the principal docking pose seen for 6 revealed
a T-shaped interaction between the tacrine scaffold and Tyr140.
10

ACCEPTED MANUSCRIPT
Figure 4. Docking of (S)-6 into GSK-3β.
To validate these molecular modelling studies, we decided to synthesize the MTDLs 2, 3, 4,
, 6, 7 and 8 as racemic mixtures, as well as both enantiomers of 6, in order to determine
5
experimentally their inhibitory activities on ChEs and selected kinases. Since docking of
compound 1 presenting higher binding energies, especially for the S enantiomer, we focused our
synthetic efforts on molecules presenting binding energies lower than -14 kcal/mol for both
enantiomers.
2.2 Chemistry
First, the tetrahydroacridine scaffold 14, bearing a clickable azide moiety, was prepared from
the commercially available 2-aminobenzoic acid 9 and cyclohexanone 10 (Scheme 1).
Compound 14 was thus obtained in four steps, with an overall yield of 18%. The detailed
procedures for synthesis of 14 are descried in Supporting Information.
11

ACCEPTED MANUSCRIPT
Scheme 1. Reagents and conditions. (i) POCl (10.0 equiv), 0 °C, then 12 h reflux, 52%; (ii)
3
ethanolamine (3.0 equiv), 18 h reflux, 76%; (iii) SOCl (21.0 equiv), 45 min reflux, 75%; (iv)
2
NaN (4.0 equiv), DMF, 80 °C, 24 h, 61%.
3
The synthesis of the pyridine moieties 19 and 23 (Scheme 2) started with esterification of the
commercially available 4-(hydroxymethyl)picolinic acid 15 to give the corresponding methyl
ester 16 in 89% yield. Preparation of compound 17 involved treatment of 16 with sodium
hydride at 0 °C in DMF followed by the addition of propargyl bromide, giving 17 in 72% yield.
Treatment with an excess of hydrazine in methanol gave the acyl hydrazine 18 in near
quantitative yield. Finally, the acyl azide 19 was synthesized using sodium nitrite in aqueous 12
N HCl at 0 °C in 88% yield. In conclusion, acyl azide 19 was obtained in four steps from 4-
(hydroxymethyl)picolinic acid 15 in a 55% overall yield.
The first step in the preparation of compound 23 involved the conversion of alcohol 16 to the
corresponding chloride 20, followed by the nucleophilic substitution with N-methylpropargyl
amine in the presence of potassium carbonate in refluxing acetonitrile, to give compound 21 in
7
4% yield. The following step involved conversion of methyl ester 21 to the corresponding acyl
azide 23. In conclusion, derivative 23 was obtained in five steps from 4-
hydroxymethyl)picolinic acid 15, in a 50% overall yield.
(
12

ACCEPTED MANUSCRIPT
Scheme 2. Reagents and conditions. (i) SOCl (4.0 equiv), MeOH, reflux, 48 h, 89%; (ii) NaH
2
(2.0 equiv), propargyl bromide (80% in toluene) (2.0 equiv), DMF, 0 °C to rt, 18 h, 72%; (iii)
hydrazine monohydrate (7.0 equiv), MeOH, 45 min rt, 99% for 18 and 22; (iv) NaNO (2.0
2
equiv), 12 N aqueous HCl, 0 °C, 2 h, 88% (for 19) and 83% (for 23); (v) Mesyl chloride (1.5
equiv), Et
3 2 2
N (2.0 equiv), CH Cl , 0 °C, then 24 h reflux, 92%; (vi) N-methylpropargylamine (1.5
equiv), K CO (2.0 equiv), CH CN, 24 h reflux, 74%.
2
3
3
Curtius rearrangement between acyl azides 19 and 23 and tetrahydropyrido[2,1-a]isoindolone
±)-24 [60] in refluxing 1,4-dioxane led to the desired ureas (±)-25 and (±)-26 in 74% and 91%
(
yield, respectively. Finally, CuAAC reaction with 14 using copper sulfate (0.3 equiv) and
sodium ascorbate (0.6 equiv) in DMF for 48 h produced (±)-2 and (±)-3 in 65% and 72% yield,
respectively (Scheme 3).
13

ACCEPTED MANUSCRIPT
O
N
NH
NH
O
N
H
2
N
O
N
X
X
N
X
N
N
(
±)-24
ii
O
N3
i
HN
N
N
N
N
N
O
H
H
O
N
1
9 (X = O)
(±)-25 (X = O)
(±)-26 (X = NMe)
(±)-2 (X = O)
(±)-3 (X = NMe)
23 (X = NMe)
Scheme 3. Reagents and conditions. (i) (±)-24 (1.0 equiv), 1,4-dioxane, reflux, 24 h, 74% (for
±)-25) and 91% (for (±)-26); (ii) 14 (1 equiv), CuSO .5H O (0.3 equiv), sodium ascorbate (0.6
(
4
2
equiv), DMF, 48 h rt, 65% (for (±)-2), 72% (for (±)-3).
We subsequently decided to synthesize several fragments of 2 in order to determine their
structural contributions (especially those of the tacrine and triazole scaffolds) to the inhibition of
the kinases (Figure 5). Thus, 27, 28A-C and 29 were also prepared as described below.
14

ACCEPTED MANUSCRIPT
Figure 5. Structures of isoindolone and tacrine-based fragments of 2 that were synthesized.
The first step in the synthesis of (±)-27 involved the CuAAC reaction between
trimethylsilylmethyl azide and compound 17 to furnish 30 in 74% yield (Scheme 4). Treatment
with TBAF yielded the N-methyl triazole 31 in 76% yield. The following step was formation of
the acyl azide 33 from the methyl ester 31 via a similar route to that previously described for 2,
giving 33 in 24% overall yield. Then, Curtius rearrangement between 33 and
tetrahydropyrido[2,1-a]isoindolone, (±)-24, resulted in the desired product (±)-27. Thus, (±)-27
was synthesized in five steps from 17 in 3% overall yield.
15

ACCEPTED MANUSCRIPT
Scheme 4. Reagents and conditions. (i) trimethylsilylmethyl azide (1.1 equiv), CuSO .5H O
4
2
(0.3 equiv), sodium ascorbate (0.6 equiv), DMF, 24 h, rt, 74%; (ii) TBAF (1.0 M in THF) (2.0
equiv), THF, 0 °C, then 24 h rt, 76%; (iii) hydrazine monohydrate (7.0 equiv), MeOH, 45 min,
rt, 99%; (iv) NaNO (2.0 equiv), 12 N aqueous HCl, 2 h, 0 °C, 25%; (v) (±)-24 (1.0 equiv), 1,4-
2
dioxane, 24 h, reflux, 25%.
The syntheses of isoindolone-based fragments 28A-C involved the CuAAC reaction between
alkyne (±)-25 and azides 34A,[61] 34B and 34C. The latter were prepared from 3-phenylpropan-
1
-ol, 4-aminopyridine and 4,7-dichloroquinoline, respectively (see supporting information).
Using 0.3 equiv of CuSO and 0.6 equiv of sodium ascorbate, 28A-C were obtained in 99, 41
and 47% yields, respectively (Scheme 5).
4
16

ACCEPTED MANUSCRIPT
Scheme 5. Reagents and conditions. (i) 34A-C (1.0 equiv), CuSO .5H O (0.3 equiv), sodium
4
2
ascorbate (0.6 equiv), DMF, 48 h, rt, 99% (for (±)-28A), 41% (for (±)-28B), 47% (for (±)-28C).
Finally, the tacrine-based fragment 29 was synthesized from 14 as shown in Scheme 6. The
CuAAC reaction between 14 and trimethylsilylacetylene, followed by treatment with tetra-n-
butylammonium fluoride, gave 29 in two steps in 50% overall yield.
Scheme 6. Reagents and conditions. (i) trimethylsilylacetylene (1.5 equiv), CuSO .5H O (0.3
4
2
equiv), sodium ascorbate (0.6 equiv), DMF, 3 h, rt, 99%; (ii) TBAF (1.0 M in THF) (2.0 equiv),
THF, 3 h, reflux, 50%.
We then prepared the precursor alkynes (±)-47 and (±)-48 for synthesis of the MTDLs (±)-4
and (±)-6, respectively (Scheme 7). Oxidation of the primary alcohols 16 and 37 to the
17

ACCEPTED MANUSCRIPT
corresponding aldehydes 38 and 40 was performed using 2-iodoxybenzoic acid (IBX) in refluxed
EtOAc in a near quantitative yield. Unfortunately, all attempts to prepare aldehyde 39 by
oxidation of alcohol 36 failed. Oxidation using IBX, Dess-Martin periodinane, PCC or Swern
conditions led only to degradation, probably due to the low stability of 39. We were thus unable
to synthesis the bifunctional ligand 5. Alcohols 36 and 37 were prepared in five steps from
commercially available 4-pyridineethanol and 4-pyridinepropanol, respectively (see Supporting
Information). Then, Seyferth-Gilbert homologation was carried out using 1.5 equivalent of the
Ohira-Bestmann reagent to give the desired alkynes 41 and 42 in 36% and 48% yield,
respectively. Conversion of 41 and 42 to the corresponding acyl azides 45 and 46 was performed
in two steps in 63% and 82% yields, respectively, as described earlier for 19 and 21. Finally,
Curtius rearrangement between acyl azides 45 and 46 and (±)-24 gave (±)-47 and (±)-48 in 92%
and 40% yields, respectively.
Scheme 7. Reagents and conditions. (i) IBX (3.0 equiv), EtOAc, 3 h reflux for 38, and 12 h
for 40, 99%; (ii) Ohira-Bestmann reagent (1.5 equiv), K
1), 48% (for 42); (iii) hydrazine monohydrate (7.0 equiv), MeOH, 45 min, rt, 99%; (iv) NaNO
2.0 equiv), 12 N aqueous HCl, 2 h, 0 °C, 64% (for 45), 83% (for 46); (v) (±)-24 (1.0 equiv), 1,4-
dioxane, 24 h, reflux, 92% (for (±)-47), 40% (for (±)-48).
2 3
CO (2.0 equiv), MeOH, rt, 36% (for
4
2
(
18

ACCEPTED MANUSCRIPT
The final step in accessing the MTDLs (±)-4 and (±)-6 consisted of a CuAAC reaction
between azide 14 and alkynes (±)-47 and (±)-48 (Scheme 8). Following this procedure, (±)-4 and
(±)-6 were obtained in 19% and 18% yields, respectively.
4 2
Scheme 8. Reagents and conditions. (i) 14 (1.0 equiv), CuSO .5H O (0.3 equiv), sodium
ascorbate (0.6 equiv), DMF, 48 h, rt, 18% for (±)-6, and 19% for (±)-4.
The syntheses of single enantiomers of 6 started with the enantioseparation of both
enantiomers of (±)-24 by supercritical fluid chromatography (SFC) (scheme 9) (see Supporting
Information). The absolute stereochemistry of (R)-24 was determined by single crystal X-ray
diffraction after crystallization by vapour diffusion (CH Cl /pentane) (see Supporting
2
2
Information). Then, Curtius rearrangement between each enantiomer of 24 and acyl azide 45
gave the corresponding optically pure ureas (R)-47 and (S)-47 in 73% and 76% yield,
respectively. The final step consisted of a CuAAC reaction between azide 14 and alkyne (R)-47
and (S)-47 to furnish (R)-6 and (S)-6 in 56% and 45% yield, respectively. The higher yields in
the CuAAC reaction achieved in the synthesis of the pure enantiomers with respect to the
racemic mixtures suggests that the poor yields obtained for synthesis of (±)-4 and (±)-6 could be
improved.
19

ACCEPTED MANUSCRIPT
Scheme 9. Reagents and conditions. (i) Separation of enantiomers by SFC; (ii) 45 (1.0 equiv),
,4-dioxane, 24 h reflux, 73% for (-)-(R)-47, 76% for (+)-(S)-47; (iii) 14 (1.0 equiv),
1
CuSO .5H O (0.3 equiv), sodium ascorbate (0.6 equiv), DMF, 48 h, rt, 56% for (-)-(R)-6, 45%
4
2
for (+)-(S)-6.
Subsequently, we prepared both regioisomers of MTDL 6 substituted at position C-5’
(Schemes 10 and 11). The syntheses started with a Sonogashira cross-coupling reaction between
the commercially available bromomethyl picolinates, 49 and 50, and ethynyltrimethylsilane, to
furnish alkynes 51 and 52 in 97% and 96% yield, respectively. Deprotection of the alkynes was
performed with potassium fluoride to give compounds 53 and 54 in 77% and 82% yield,
respectively. The conversion of esters 53 and 54 to the corresponding acyl azides, 57 and 58, was
performed as previously described, in two steps, in 95% and 90% overall yields, respectively.
Finally, Curtius rearrangement between acyl azides 57 and 58 and (±)-24 gave (±)-59 and (±)-60
in 65% and 50% yield, respectively. To summarize, ureas (±)-59 and (±)-60 were prepared in
five steps from methyl picolinates 49 and 50 in 46% and 35% overall yield, respectively. The last
step consisted of the CuAAC reaction between azide 14 and alkynes (±)-59 and (±)-60 to give
the MTDLs (±)-7 and (±)-8 in 81% and 94% yield, respectively (Scheme 11).
20

ACCEPTED MANUSCRIPT
Scheme 10. Reagents and conditions. (i) Ethynyltrimethylsilane (3.0 equiv), CuI (0.1 equiv),
PdCl (PPh ) (0.05 equiv), Et N/THF (1:1, v/v), 5 h, 60 °C, 97% for 51, 96% for 52; (ii) KF (3.0
2
3 2
3
equiv), MeOH/CH Cl (1:1, v/v), 12 h, rt, 77% for 53, 82% for 54; (iii) hydrazine monohydrate
2
2
(
7.0 equiv), MeOH, 45 min, rt, 99%; (iv) NaNO (2.0 equiv), 12 N aqueous HCl, 2 h 0 °C, 96%
2
for 57, 91% for 58; (v) (±)-24 (1.0 equiv), 1,4-dioxane, reflux for 24 h, 65% for (±)-59, 50% for
±)-60.
(
4 2
Scheme 11. (i) 14 (1.0 equiv), CuSO .5H O (0.3 equiv), sodium ascorbate (0.6 equiv), DMF,
4
8 h rt, 81% for (±)-7, 94% for (±)-8.
2.3 In vitro Assays
The newly synthesized MTDLs (±)-2-4, (±)-6-8, (R)-6 and (S)-6, were evaluated for their
inhibitory potency on kinases GSK-3α/β and CKD5/p25, and on human AChE (hAChE) and
human BChE (hBChE) (Table 2).
21

ACCEPTED MANUSCRIPT
Table 2. Inhibition of ChEs and kinases by MTDLs (±)-2-4, (±)-6-8, (R)-6 and (S)-6.
hBChE
GSK-3α/β
(nM)
CDK5/p25
(nM)
a
b
Entry Compound
hAChE (nM)
a
SI
c
c
(
nM)
1
2
3
4
5
6
7
8
9
1
1
(±)-2
20.8 ± 0.9
58.6 ± 3.1
23.6 ± 2.3
11.4 ± 1.7
9.5 ± 0.4
13.7 ± 1.0
0.8 ± 0.1
133 ± 6
424 ± 21
-
169 ± 6
206 ± 7
65.7 ± 3.5
301 ± 14
395 ± 27
254 ± 18
185 ± 47
21.1 ± 1.6
33.5 ± 1.0
-
8.1
3.5
2.8
26
10
300
(±)-3
10
310
(±)-4
21
800
(±)-6
16
800
(R)-6
41.5
18.5
227
0.16
0.08
-
7
500
(S)-6
19
1100
>10 000
3100
>10 000
80
(±)-7
>10 000
4200
>10 000
170
22
(±)-8
Tacrine
Valmerin I
Valmerin II
0
1
-
-
-
25
a
Recombinant human AChE and BChE from human serum were used. Values are expressed as
b
the mean
± SEM of two independent experiments each performed in triplicate. Selectivity Index
c
SI: IC50 (hBChE)/IC50 (hAChE). All data points for construction of dose-response curves were
recorded in triplicate. Typically, the standard deviation of single data points was below 10%.
In addition, the inhibitory activity, expressed as IC50 values, toward GSK-3α/β and CKD5/p25
of all isoindolone-based fragments, i.e., (±)-25-27, (±)-28A-C, (±)-47, (
±)-59 and (±)-60 and of the tacrine-based fragment 29 were determined and are displayed in
Table 3.
R)-47, (S)-47, (±)-48,
(
22

ACCEPTED MANUSCRIPT
Table 3. Kinase inhibition by isoindolone-based and tacrine-based fragments.
a
IC50 (nM)
Entry
Compound
GSK-3α/β
CDK5/p25
1
2
3
4
5
6
7
8
9
(±)-25
(±)-26
60
11
150
90
(±)-27
25
250
260
200
240
>10 000
68
(±)-28A
(±)-28B
(±)-28C
29
8
30
>10 000
>10 000
41
(±)-47
(R)-47
(S)-47
33
18
1
1
1
1
1
1
0
1
2
3
4
5
90
50
(±)-48
15
70
(±)-59
>10 000
300
170
22
>10 000
80
(±)-60
Valmerin I
Valmerin II
80
25
a
All data points for construction of dose-response curves
were recorded in triplicate. Typically, the standard deviation
of single data points was below 10%.
23

ACCEPTED MANUSCRIPT
The most active inhibitor (±)-7 showed very high inhibitory potency in the subnanomolar
range (IC50 = 815 pM). However, all the other assayed hybrids were active against hAChE in the
nanomolar range (Table 2). In detail, the hybrid (±)-2 was 2.8-fold more potent than (±)-3,
probably due to better interaction of the pyridine ring with Trp286 at the PAS of AChE. In
addition, as predicted by docking studies, the shorter the linker between triazole and pyridine
ring is, the better the inhibitory potency. For instance, compound (±)-6 was 2-fold more potent
than compound (±)-4, 1.8-fold more potent than compound (±)-2 and 5-fold more potent than
compound (±)-3. Regarding the enantiomers of hybrid 6, the enantiomer (R)-6 was only 1.5-fold
more potent than the enantiomer (S)-6. All MTDLs, except (±)-8, showed selectivity of one order
of magnitude toward hAChE relative to hBChE, except for compound (±)-7, which was 227-fold
more potent on hAChE. Conversely. (±)-8 acted as a BChE selective inhibitor. Among all the
hybrids, compound (±)-8 showed the highest potency and selectivity toward hBChE with an IC50
value of 21.1 nM and a selectivity index relative to hAChE of 6.3. As has been reported, BChE
inhibition could also be of benefit for AD patients.[62]
Interestingly, compounds (±)-2, (±)-3, (±)-4 and (±)-6 showed inhibitory activity in the
nanomolar range toward GSK-3
of the linker on pyridine position C-4’ proved to be the most suitable to preserve a strong
inhibitory activity toward GSK-3 . In detail, compound (±)-6 was 625-fold and 262-fold
more potent than compounds (±)-7 and (±)-8, whose linkers are attached on pyridine positions C-
’ and 6’, respectively. Conversely to the parent valmerins (Figure 1), all new hybrids showed
low inhibitory potency toward CDK5 kinases, and thus a selectivity for GSK-3 (Table 2). As
far as the influence of the absolute configuration of the stereocentre of the isoindolone scaffold is
α/β (Table 2). As reported in previous studies, the substitution
α/β
5
α/β
24

ACCEPTED MANUSCRIPT
concerned, (R)-6 exhibited a GSK-3 kinase inhibition potency ~3-fold higher than the (S)-
enantiomer. This relatively small difference indicates that the shape differences of the
isoindolone scaffold related to its stereocentre absolute configuration have a limited effect.
Determination of the inhibitory activity of the isoindolone-based and tacrine-based fragments
(
Table 3) provided additional information for the SAR study, especially regarding the
contribution of the tacrine and triazole moieties to kinase inhibition. Indeed, the racemic hybrid
±)-2 was 6-fold more potent than its precursor (±)-25, indicating that the triazole or/and tacrine
(
scaffolds contributed to the kinase inhibition. This difference of potency was not observed for
compound (±)-3, or for its precursor (±)-26, probably due to an additional interaction of the N-
Me group within the GSK-3 active site. Addition of the triazole ring allowed a good GSK-3
binding mode. For instance, hybrid (±)-27, containing a triazole ring, was 2.4-fold better
inhibitor than (±)-25. In contrast, depletion of the isoindolone scaffold, such as in compound 29,
led to a complete loss of activity toward GSK-3 (IC > 10 µM). These results suggested an
5
0
unexpected synergistic effect of the triazole and isoindolone scaffolds in GSK-3 inhibition. The
addition of a hydrophobic moiety led to an increase in the inhibitory potency of the isoindolone.
In comparison with compound (±)-27, the isoindolone-based fragment (±)-28A was a 3-fold
more potent GSK-3 inhibitor. In contrast, the addition of more polar substituents (e.g., 4-
aminopyridine and 4-aminoquinoline) decreased the binding affinity toward GSK-3 kinase (up to
1
250-fold lower potency). in vitro results confirmed the importance of the triazole in kinase
inhibition. Indeed, (S)-6 and (R)-6 were up to 4.7-fold more potent than their precursors, (S)-47
and (R)-47, respectively. Finally, as expected, substitution at position 5 or 6 on the pyridine ring
had a deleterious effect for the inhibition of GSK-3 kinase. Only the isoindolone-based fragment
(R)-47 showed inhibition potency toward CDK5 in the nanomolar range (IC50 = 18 nM). In
25

ACCEPTED MANUSCRIPT
conclusion, the hybrid (R)-6 showed the most promising MTDL profile for the inhibition of
hAChE and GSK-3
nM for hAChE and 7 nM for GSK-3
the triazole and isoindolone scaffolds in the inhibition of GSK-3α/β.
α/β with potencies in the nanomolar range for both enzymes (e.g., IC = 9.5
50
α/β). These results also highlighted the synergistic effect of
The selectivity of (±)-6 (10 µM) was also evaluated on a panel of 468 kinases (Discoverx,
TM
Kinomescan ). The results showed that (±)-6 displayed poor selectivity, with several kinases
are targeted (see supporting information). At this concentration (10 µM), (±)-6 displayed weak
inhibition of GSK-3β (17%), whereas total inhibition of GSK-3α was observed.
2.4 Cell-based assays: Cytotoxicity and Brain Penetration
The cytotoxicity of these new MTDL hybrids was evaluated against a panel of representative
human cell lines including HuH7 (liver), Caco2 (colon), MDA-MB231 (breast), HCT-116
(colon), PC3 (prostate), NCIH727 (lung), HaCaT (skin). The concentrations that produce 50%
inhibition of cell growth are reported in Table 4. Interestingly, the new hybrids were significantly
less cytotoxic, whereas the tacrine fragment alone exhibited IC values in the micromolar range,
50
and the isoindolone fragments alone in the nanomolar range, thus both being quite toxic. In
general, (±)-2-4, (±)-6-8, (R)-6 and (S)-6 assembling the two fragments were, on the contrary
weakly cytotoxic, with IC values in the micromolar range regardless of the cell line. Hybrids
50
(±)-7 and (±)-8 presented the lowest cytotoxicity, with a double-digit micromolar IC50 range
(higher than 25 µM for (±)-7), while compound (±)-4 showed slightly higher cytotoxicity, with
IC values of 1.1
± 0.4 µM for colon HCT 116 cell lines.
5
0
26

ACCEPTED MANUSCRIPT
Table 4: Cytotoxicity of MTDLs on human cell lines
a
IC (µM) on Human cell lines
5
0
MDA-
MB231
NCI-
H727
Entry
MTDL
HuH7
Caco2
HCT-116
PC3
HaCaT
1
2
3
4
5
6
7
8
(±)-2
(±)-3
(±)-4
(±)-6
(R)-6
(S)-6
(±)-7
(±)-8
4.0 ± 0.4
2.0 ± 0.3
2.9 ± 0.4
3.0 ± 0.9
3.5 ± 0.3
2.6 ± 0.5
>25
20 ± 3
9.0 ± 0.3
4.8 ± 0.6
3.0 ± 0.6
8 ± 1
4.0 ± 0.4
2.0 ± 0.1
2.0 ± 0.1
2.0 ± 0.1
2.0 ± 0.2
4.0 ± 0.3
>25
4.0 ± 0.4
2.0 ± 0.1
1.1 ± 0.4
3.0 ± 0.5
2.2 ± 0.1
2.7 ± 0.1
>25
7.0 ± 0.1
3.0 ± 0.3
1.6 ± 0.1
5 ± 1
9 ± 2
4.0 ± 0.4
2.0 ± 0.1
1.9 ± 0.2
5.0 ± 0.2
4.0 ± 0.6
6.3 ± 1.5
2.0 ± 0.3
5.0 ± 0.9
7.0 ± 0.8
>25
b
3.0 ± 0.3
6.0 ± 0.9
>25
-
b
10 ± 1
>25
-
b
-
b
18 ± 7
24 ± 3
17 ± 1
17 ± 2
27 ± 3
>25
-
a
b
Values are expressed as the mean of triplicate ± SD. Reference compound used (not shown) was roscovitine. not determined.
27

ACCEPTED MANUSCRIPT
Then, the ability of compounds (±)-2, (±)-3 and (±)-6 to decrease the viability of neuroblastoma
cells SHSY5Y after incubation for 24 h and 72 h was determined as a function of drug
concentrations of 0.1-100 µM. After 24 h, reduction in cell viability of SHSY5Y was observed
for all selected compounds. At the highest tested concentration (100 µM) cell viability was
decreased about 50% relative to untreated cells. After 72 h, SHSY5Y cells showed a higher
sensitivity to the compounds (Figure 6) and it was possible to determine the concentrations that
induced 50% cell death (IC , µM). The IC values of compounds (±)-2, (±)-3 and (±)-6 were
5
0
50
2
2±2, 25±2 and 13±2 µM, respectively.
In order to exclude any cytotoxic effect of the compounds on MDCK-MDR1 cells during the
permeability experiments, we performed the MTT assay on MDCK-MDR1 cells after incubation
with compounds for 2 h at 75 µM. None of the compounds affected the cell viability of the cells
under the conditions described above (data not shown).
1
00
(
(
(
±)-3
±)-2
±)-6
8
6
4
2
0
0
0
0
0
2
4 h
72 h
[
Conc. 100 µM of Drug]
Figure 6. Percentage of cell viability of SHSY5Y neuroblastoma cells after 24 h and 72 h of co-
incubation with compounds at a concentration of 100 µM.
28