From Kinase Inhibitors to Multitarget Ligands as Powerful Drug Leads for Alzheimer's Disease using Protein-Templated Synthesis

Article abstract of DOI:10.1002/anie.202106295

Multitarget directed ligands (MTDLs) are arising as promising tools to tackle complex diseases. The main goal of this work is to create powerful modulating agents for neurodegenerative disorders. To achieve this aim, we have combined fragments that inhibi

Full text of DOI:10.1002/anie.202106295

Angewandte
Research Articles
Chemie
How to cite:
International Edition: doi.org/10.1002/anie.202106295
German Edition: doi.org/10.1002/ange.202106295
Drug Discovery
From Kinase Inhibitors to Multitarget Ligands as Powerful Drug Leads
for Alzheimerꢀs Disease using Protein-Templated Synthesis
Vanesa Nozal+, Alfonso Garcꢀa-Rubia+, Eva P. Cuevas, Concepciꢁn Pꢂrez, Carlota Tosat-
Bitriꢃn, Fernando Bartolomꢂ, Eva Carro, David Ramꢀrez, Valle Palomo,* and Ana Martꢀnez*
Abstract: Multitarget directed ligands (MTDLs) are arising as
promising tools to tackle complex diseases. The main goal of
this work is to create powerful modulating agents for neuro-
degenerative disorders. To achieve this aim, we have combined
fragments that inhibit key protein kinases involved in the main
pathomolecular pathways of Alzheimerꢄs disease (AD) such as
tau aggregation, neuroinflammation and decreased neurogen-
esis, whilst looking for a third action in beta-secretase
(BACE1), responsible of b-amyloid production. We obtained
well-balanced MTDLs with in vitro activity in three different
relevant targets and efficacy in two cellular models of AD.
Furthermore, computational studies confirmed how these
compounds accommodate adequately into the long and rather
narrow BACE1 catalytic site. Finally, we employed in situ click
chemistry using BACE1 as protein template as a versatile
synthetic tool that allowed us to obtain further MTDLs.
single-target molecule strategy has failed in the treatment of
several pathologies including infectious diseases, cancer and
neurodegenerative disorders and seems no longer appropri-
ate.[3] In order to address these pathologies, polypharmacy has
been used in the clinical settings for numerous years targeting
several biological systems at the same time.[4] For example, it
has been used in the treatment of cardiovascular diseases,
cancer or HIV.[5] However, the combination of different drugs
for pharmacological treatment entails several risks and raises
many difficulties such as unpredicted drug-drug interactions
or multiplied side effects in addition to lower patient
compliance. With the specific aims of tackling several
pathomolecular pathways simultaneously while reducing risks
associated to polypharmacy, MTDLs are arising as an ideal
strategy for the treatment of complex diseases.
Neurodegenerative diseases, and specifically Alzheimerꢀs
disease (AD), would greatly benefit from a MTDL ap-
proach.[6] These complex diseases are characterized by their
unknown etiology, their intricate molecular pathology and
their multifactorial nature. AD is highly prevalent, and
despite the great efforts to find an effective drug, it still
presents high clinical failure. Furthermore, the complexity of
the molecular pathology suggests that traditional drugs will
not be able to produce a therapeutic effect.[7] Due to this
evidence, the association of multitarget compounds and
neurodegenerative diseases has been growing in recent years
and therefore some multitarget compounds have reached
clinical trials.[8] The specific design of these compounds needs
to be carefully studied, being the main optimal characteristics
the following: a) having a similar potency in all the modulated
targets in order to enable a proper dose to tackle every
pathobiological pathway efficiently, b) targeting proteins that
Introduction
The lack of effective treatments for severe diseases
characterized by a multifactorial etiology or therapeutic
resistances, such as neurodegenerative disorders, cancer or
bacterial infections, has led to the emergence of a new drug
design paradigm: the multitarget approach strategy.[1,2] Multi-
target directed ligands (MTDLs) are characterized for show-
ing activity in more than one molecular target taking
advantage of multiple additive or synergic pharmacodynamic
activities in a single molecule. This novel approach, that
carefully selects polypharmacological drugs, contrasts with
the previous classical design where drugs were designed to
modulate a specific target with high selectivity, considering
the multiple modulations of several targets undesirable. The
[*] V. Nozal,[+] Dr. A. Garcꢀa-Rubia,[+] Dr. E. P. Cuevas, C. Tosat-Bitriꢁn,
Dr. V. Palomo, Prof. A. Martꢀnez
Dr. F. Bartolomꢃ, Dr. E. Carro
Hospital Universitario 12 de Octubre Research Institute (imas12),
Group of Neurodegenerative Diseases
Structural and Chemical Biology Department, Centro de Investiga-
ciones Biolꢂgicas-CSIC
Juan de la Cierva 3, 28006 Madrid (Spain)
Ramiro de Maeztu 9, 28040 Madrid (Spain)
E-mail: vpalomo@cib.csic.es
Dr. D. Ramꢀrez
Instituto de Ciencias Biomꢃdicas, Universidad Autꢂnoma de Chile
Llano Subercaseaux 2801—piso 6, Santiago (Chile)
ana.martinez@csic.es
Dr. A. Garcꢀa-Rubia,[+] Dr. V. Palomo, Prof. A. Martꢀnez
Centro de Investigaciꢂn Biomꢃdica en Red de Enfermedades
Neurodegenerativas (CIBERNED), Instituto de Salud Carlos III
28031 Madrid (Spain)
[+] These authors contributed equally to this work.
Supporting information and the ORCID identification number(s) for
the author(s) of this article can be found under:
https://doi.org/10.1002/anie.202106295.
Dr. C. Pꢃrez
Instituto de Quꢀmica Mꢃdica-CSIC)
Juan de la Cierva 3, 28006 Madrid (Spain)
ꢄ 2021 The Authors. Angewandte Chemie International Edition
published by Wiley-VCH GmbH. This is an open access article under
the terms of the Creative Commons Attribution License, which
permits use, distribution and reproduction in any medium, provided
the original work is properly cited.
&&&&
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
These are not the final page numbers!

Angewandte
Research Articles
Chemie
present synergistic or additive effects reached with moderate
activity that provides the opportunity to encounter a better
safety profile, and c) preserving drug-like properties.[9]
Specifically, the main pathological processes in AD are
neurodegeneration, protein aggregation and brain inflamma-
tion. The major pathological aggregate hallmarks are b-
amyloid plaques and neurofibrillary tangles. b-Amyloid
plaques are extracellular neurotoxic deposits of b-amyloid
peptide caused, among other factors, by an abnormal
processing by the beta-secretase (BACE1).[10] Neurofibrillary
tangles are intracellular deposits composed of hyperphos-
phorylated protein tau. Multiple kinases including GSK3b,
LRRK2 and CK1d, among others, are involved in this
pathological aggregation.[11,12] Currently, different BACE1
and protein kinase inhibitors have reached clinical trials but
none has been approved by the regulatory agencies so far.[13,14]
Our goal is to design multitarget compounds able to interfere
simultaneously with three different targets involved in AD:
BACE1 and, at least, two different protein kinases. In that
sense, the new compounds could interfere with both protei-
nopathies present in AD. Moreover, since kinase inhibitors
are also involved in the reduction of neuroinflammation and
beta-amyloid is emerging as a risk factor for many other
diseases associated with aging,[15,16] the combination of all the
biological activities in a single molecule could create powerful
drugs not only for AD but also for other dementias associated
to frailty and aging. Several bifunctional molecules have
already been discovered for the potential treatment of AD,
however molecules able to combine 3 biological functions are
Figure 1. Different strategies used to generate multitarget compounds.
more scarce. The first reported trifunctional molecules
exhibited acetylcholinesterase (AChE) inhibitory activity
together with the reduction of AChE-induced amyloid-b
aggregation and metal chelating properties.[17] Other discov-
eries have found molecules that tackle relevant AD targets
such as AChE and its induced b-amyloid aggregation,
butyrylcholinesterase, monoamine oxidase or metal chela-
tion.[18–20]
The rational design of MTDs is based in three main
strategies: linkage, fusion and incorporation (Figure 1).
Although the increase in molecular size and weight
usually present in the design of multitarget compounds is
one of the major disadvantages,[21] the long and rather narrow
size of the BACE1 catalytic site appears to be suitable for the
linkage strategy in a MTDL design. The volume gained by
linking two different protein kinase inhibitors could adjust
into BACE1 catalytic site, which is able to accommodate
eleven amino acid substrates (Figure 2a).[22] Inhibitors for this
protease are normally large molecules able to maximize
Figure 2. A) Representation of the catalytic site in BACE1. Flap and Tyr71 are shown in blue. B) Example of BACE1 inhibitors with heteroaromatic
rings and amide functional groups.
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
www.angewandte.org
&&&&
These are not the final page numbers!

Angewandte
Research Articles
Chemie
interactions in this cavity stabilizing the flexible loop on its N-
terminal domain known as “the flap” through its Tyr71 and
locking BACE1 in an inactive conformation.[23] The central
binding pockets on the active site bind hydrophobic groups
such as the ones found in peptidomimetic compounds based
on substrate transition-state analogues[24] or other scaffolds
like those showed in Figure 2b, including verubecestat,[25]
OM-003,[26] 5I3Y ligand[27] and 5V0N ligand.[28,29] The BACE1
pharmacophore of amidine compounds, the most advanced
BACE1 inhibitors in clinical trials, highlights the importance
of heteroaromatic rings linked through an amide bond to
optimally occupy S1 and S3 pockets.[30] Therefore our working
hypothesis consisted in the use of the linkage design to build
BACE1 inhibitors starting from kinase inhibitor fragments
and thus, obtaining MTDLs with both activities on the
different AD hallmarks. Additionally, the linkage strategy
also offers the possibility of connecting original scaffolds in
a manner that can retain their primitive activity and
interactions with the initial target.
In this work, we describe the design, synthesis, biological
evaluation and suggest a binding mechanism of novel multi-
target compounds with BACE1 and protein kinase inhibition
activities. Furthermore, click chemistry and protein-templat-
ed synthesis using BACE1 as scaffold has been used as
versatile synthetic tools. Finally, different cellular assays have
also been used to confirm the polypharmacology and applic-
ability of our MTDLs.
Scheme 1. Synthesis of compounds 3 and 8. i) SOCl2, 808C; ii) 2-
amino-benzothiazole, 1008C MW, THF; iii) K2CO3, propane-1,3-dithiol,
THF, 808C; iv) EDCI, DMAP, DMF, r.t., 5 min. then 2-amino-benzothia-
zole, r.t.; v) 3-bromopropyne, K2CO3, MeCN, 908C; vi) NaOH,
MeOH:THF (1:1 v/v) r.t.; vii) EDCI, DMAP, DMF, r.t. 5 min. then 2-
amino-benzothiazole, r.t.; viii) CuSO4·5H2O, sodium ascorbate, DMF,
r.t.
Results and Discussion
Design and Evaluation of a Multitarget hit as Proof of Concept
In order to test our working hypothesis, we firstly
synthesized two multitarget compounds, 3 and 8, to evaluate
whether linking two derivatives of a kinase inhibitor would be
sufficient to gain BACE1 inhibitory activity by accommodat-
ing them into the catalytic site. Specifically, we selected the
benzothiazole-based LRRK2 inhibitors as starting frag-
ments.[31] To connect the two protein kinase fragments,
different linkers were assayed: an aliphatic thioether chain
and the heterocyclic 1,2,3-triazole. Compounds 3 and 8 would
serve as a proof of concept of activity preservation for
LRRK2 inhibition and acquired BACE1 inhibition with
molecules containing heteroaromatic rings linked through
amide bonds.
with IC50 values of 7.14 and 3.72 mM respectively, without
losing LRRK2 inhibition (IC50 values of 1.88 and 1.83 mM,
respectively). These compounds served as a first proof of
concept for our working hypothesis. Additionally, kinetic
experiments were done using human recombinant BACE1
and different concentrations of substrate and compound 8.
Lineweaver–Burk double reciprocal plot (Figure S1) showed
a substrate-competitive behavior in the inhibition of BACE1
by the multitarget compound 8.
Docking and binding free energy calculations showed how
both inhibitors adopted similar poses within the BACE1
catalytic site (Figure 3). These large molecules established
multiple hydrophobic contacts within pockets S1, S2, S3,
S1’and S2’ of the protease and also stabilized the flap through
a p-p interaction with Tyr71. In fact, compared to the original
BACE1 crystal structure, inhibitors 3 and 8 induced a closed
conformation of the flap at the catalytic site, moving the
Tryr71 2.7 ꢁ and 3.6 ꢁ, respectively.
Synthesis of the thioether derivative 3 was performed
coupling 2-aminobenzothiazole 2 with propane-1,3-dithiol in
basic conditions (Scheme 1).
Derivative 8 was obtained by click-chemistry through
a convergent synthesis. Azide fragment 4 was easily synthe-
tized from 4-(azidomethyl)-benzoic acid and 2-aminobenzo-
thiazole coupling with EDCI and DMAP to form the amide
bond. Alkyne 7 was obtained after three synthetic steps.
Finally, the triazole 8 was synthesized using classic copper(I)-
catalyzed alkyne-azide cycloaddition (CuAAC) (Scheme 1).
After in vitro biological evaluation using human recombi-
nant enzymes, compounds 3 and 8 showed a well-balanced
inhibitory activity for the two targets. BACE1 was inhibited
Considering that both linkers presented similar inhibitory
activity against the protease adopting a comparable binding
mode, we selected the triazole 8 as our lead compound to
obtain further multitarget molecules. Moreover, 1,2,3-tria-
zoles may be considered as amide bioisosteres[32] and are
readily obtainable through click chemistry. Therefore, we
considered CuAAC as the best methodology to connect
&&&&
www.angewandte.org
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
These are not the final page numbers!

Angewandte
Research Articles
Chemie
range with kinase selectivity, to design new multitarget
compounds.
As the synthetic pathway chosen to create the MTDL
involves the 1,3-dipolar addition between azides and alkynes,
we firstly examined the best positions in each heterocyclic
family to introduce the alkyne and azide fragments without
affecting the original biological activity of these kinase
inhibitors. The most relevant leads for each family of kinase
inhibitors are showed in Figure 5, indicating the potential
chemical modifications tolerated based on previous SAR
analysis.[31,34–36] We selected one GSK3b inhibitor, one CK1
inhibitor, and two chemically diverse LRRK2 inhibitors as
starting fragments for the MTDLs synthesis.
Figure 3. Suggested binding mode of compounds 3 and 8 within
BACE1 catalytic site. Both the flap and residue Tyr71 from BACE1
crystal structure (PDB: 5I3Y) are represented in white. Compounds 3
and 8 as well as the flap and Tyr71 after docking are represented in
green. p-p interactions are showed as pink dotted lines.
diverse heterocyclic fragments due to its versatility, accessi-
bility and functional group tolerance.
To explore the cellular activity of this MTDL regarding
BACE1 inhibition, we tested the ability of compound 8 to
decrease the production of toxic Ab40 and Ab42, in an amyloid
protein precursor over-expressing human neuroblastoma SH-
SY5Y cell line (SK-APP cells). Levels of Ab40 and Ab42, the
two toxic species present in AD brains, were measured both in
cellular lysates and in extracellular medium after 48 h of
compound treatment. As shown in Figure 4, compound 8
exerts biological action at the cellular level probably due to
BACE1 inhibition, reducing the Ab42 levels in the extracel-
lular fluid and Ab40 levels in extracellular fluid as well as in
cellular lysates.
Figure 5. Different chemical modifications tolerated by the heterocyclic
kinase inhibitors available at the MBC chemical library.
After the selection of the optimal sites for modification,
different alkynes and azides were synthesized taking into
account commercial availability of necessary reagents and
synthetic efficiency.
Overall, five different fragments were obtained; two
azides derived from LRRK2 and CK1d inhibitors and three
alkynes from LRRK2, GSK3b, and CK1d inhibitors (deriv-
atives 4, 7, 11–13). In all cases, the synthesized fragments
followed the original design incorporating the alkyne or azide
at the optimal site for not interfering with the original protein
kinase inhibitory activity. However, for the synthesis of the
GSK3b inhibitor derivative, we encountered some synthetic
difficulties since the needed reagents for the synthesis of the
thiadiazolinone ring, isocyanates and isothiocyanates, are
incompatible with a fair amount of chemical reactions. In this
case, the strategy was subtly changed in order to obtain the
fragment in an efficient and rapid manner. The aromatic ring
needed for GSK3b activity was removed from the starting
fragment, hypothesizing that the triazole formed after the
click chemistry reaction would be positioned in a similar
fashion in the final compound. In this way the linker was
incorporated in the original GSK3b inhibitor structure
(compound 12).
Figure 4. Reduced levels of Ab42 and Ab40 in SK-APP cells after 48 h
treatment with compound 8 at 10 mM. A) and C) extracellular fluid; B)
and D) cellular lysates. Data are presented as mean Æ SEM of at least
three independent experiments performed in triplicate. *p<0.05;
**p<0.01.
With these excellent preliminary data showing that the
new BACE1 in vitro activity found in the multitarget
molecule is confirmed at the cellular level, we worked in
exploring further compounds.
Design, Synthesis and In Vitro Evaluation of New MTDLs:
BACE1/Protein Kinases Inhibitors
The other fragments (alkynes and/or azides) were synthe-
sized without further inconvenient (Scheme 2).
The multitarget compounds to be synthesized were
designed in order to pair different kinase activities, building
four different compounds that would yield MTDLs ideally
showing inhibition of BACE1 and LRRK2-GSK3b (14),
CK1d-GSK3b (15), or CK1d-LRRK2 (16, 17), respectively
(Scheme 3).
Starting from molecules present in our in-house chemical
library,[33] we selected the most interesting heterocyclic
families targeting different protein kinases. Concretely, we
chose GSK3b, LRRK2, and CK1d inhibitors, that have shown
inhibitory activity in the micromolar and submicromolar
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
www.angewandte.org
&&&&
These are not the final page numbers!

Angewandte
Research Articles
Chemie
well-balanced biological activities in three different targets
and pointing to good candidates to modulate the AD
pathobiological pathways efficiently at the same dose.
The BACE1 inhibitory activity of the original protein
kinase inhibitors used to build the multitarget compounds was
also tested, showing that the initial fragments alone did not
have the capacity to bind to this enzyme (Table S1). These
results suggest that the starting protein kinase inhibitors lack
of BACE1 activity, thus BACE1 inhibition found in com-
pounds 14–17 was gained by the molecular volume acquired
after the linkage of the two protein kinase inhibitors,
confirming our initial hypothesis.
Scheme 2. i) SOCl2, THF, 708C; ii) 2-amino-benzothiazole, 1008C, MW;
iii) NaN3, DMSO, r.t.; iv) SO2Cl2, 08C; v) 4-(prop-2-yn-1-yloxy)aniline,
EtOH, reflux.
The binding mode of these new MTDLs was explored
with docking and free binding energy molecular dynamics
studies, showing similar binding poses to compounds 3 and 8
with minor differences (Figure 6). For example, compounds
14 and 15 are smaller molecules and they preferentially
occupy pockets S1, S2’ and partially S1’. Compound 15 and 16
showed a shifted pose leaving pocket S3 unoccupied.
Cellular Activity of MTDLs
Scheme 3. i) CuSO4·5H2O, sodium ascorbate, DMF, r.t.
Cellular activity of the four new compounds (14–17) was
evaluated in two different models in order to prove the
beneficial effects of inhibiting BACE1 and the two different
protein kinases.
Firstly, the human neuroblastoma cell line transfected
with the precursor amyloid protein (SK-APP cells) was used
in order to confirm the ability of the MTDLs treatment to
modulate Ab40 and Ab42 levels. A preliminary cellular toxicity
study of these compounds at a fixed concentration of 10 mM
and after 24 and 48 h of treatment, discarded compounds 16
and 17 from the experiment as they showed a significant
decrease of cell viability at 48 hours (Figure S2).
The inhibitory activity of the synthesized triazoles 14–17
was evaluated against human recombinant BACE1 and their
two respective kinases, in order to establish whether these
new chemical entities conserved the original inhibitory
activities from their fragments and gained the new one as
protease inhibitor (Table 1). Given that multitarget com-
pounds come from selective kinase inhibitors serves as strong
indicative of their selectivity against their kinase.[27,30–32]
Results showed that most of the compounds inhibited
BACE1 at the low micromolar range. Moreover, the original
activity of the fragments was generally maintained, obtaining
The remaining compounds, 14 and 15 continued with the
cellular characterization. Derivatives 14 and 15 exhibited
a significant decrease in Ab42 levels
in extracellular fluids (Figure 7A),
while compound 15 was also able
to decrease levels of Ab40 both in
cell lysates and extracellular me-
dia, showing their potential to
interfere with toxic Ab pathway
(Figure 7C,D).
These decreased Ab levels
could be the consequence of a re-
duction in Ab secretion from APP
processing. Although intracellular
Ab42 expression was unchanged,
Ab40 levels were reduced in cellular
lysates mainly after compound 15
administration. This later result
might suggest that these com-
pounds were modulating Ab pro-
duction, and the subsequent re-
lease to medium. In view of our
data, we suggest that compounds
14 and 15 provoked an important
Table 1: Structure and activity of the 1,2,3-triazoles in their respective enzymes.
No
Structure
LRRK2
CK1d
GSK3b
BACE1
IC50 (mM)
IC50 (mM)
IC50 (mM)
IC50 (mM)
14
0.82Æ0.14
–
2.36Æ0.27
3.85Æ0.32
3.31Æ0.27
15
–
1.25Æ0.21
30%[a]
16
17
13.46Æ5.23
7.09Æ2.22
6.47Æ0.68
–
–
7.24Æ0.38
2.51Æ0.43
26%[a]
[a] Enzyme inhibition at 10 mM. IC50 was not calculated due to poor solubility of the compound.
&&&&
www.angewandte.org
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
These are not the final page numbers!

Angewandte
Research Articles
Chemie
Figure 6. Binding mode of compounds 3 and 8 as well as synthesized triazoles within BACE1 catalytic site. Left panel show the crystallized
structure of BACE1 in complex with 5I3Y ligand (PDB id 5I3Y).
Figure 7. Reduction in the levels of Ab40 and Ab42 in SK-APP cell line
after 48 h treatment with compounds 14 and 15 at 10 mM. A) and C)
extracellular fluid; B) and D) cellular lysates. Data are presented as
mean Æ SEM of at least three independent experiments performed in
triplicate. *p<0.05.
reduction of Ab-induced neurotoxicity, associated with di-
minished Ab presence in the cell medium. Further dose
dependent studies may be designed for more optimized lead
Figure 8. Effects of multitarget compounds and their precursors
protein kinase inhibitors on the viability of SH-SY5Y cells treated with
OA.(In blue) Effects of GSK3b and LRRK2 inhibition. Cells were treated
at the indicated doses with the combination of the individual kinase
compounds.
The protein kinase inhibition of the multitarget com-
pounds, and specifically GSK3b, CK1d, and LRRK2 inhib-
itory activities present in MTDLs 14 and 15 may interfere
with tau phosphorylation, other of the hallmarks of AD. To
evaluate the beneficial effects of our MTDLs in tau phos-
phorylation, the well-known model of okadaic acid (OA) in
human neuroblastome cell line SH-SY5Y was used. OA is
a serine/threonine phosphatase inhibitor able to induce tau
hyperphosphorylation in cell culture with the subsequent cell
death.[33] Thus, neuroprotection of triazoles 14 and 15 was
evaluated after 24 h of treatment in this model and compared
to the biological effect produced by the simultaneous treat-
ment with their individual protein kinase precursors, deriv-
atives MBC-2138 and MBC-2137 for 14; and MBC-2138 and
MBC-858 for 15. In Figure 8, the rescue from the toxic effect
of OA can be observed after the treatment with new MTDLs
inhibitors (MBC-2138 and MBC-2137) or the multitarget triazole 14.
(In green) Effects of GSK3b and CK1d. Cells were treated at the
indicated doses with the combination of the individual kinase inhib-
itors (MBC-2138 and MBC-858) or the multitarget triazole 15. Results
expressed as % of control (DMSO) are the mean (Æ SEM) of at least
three independent experiments performed in triplicate. (*p<0.05,
**p<0.005, ***p<0.001 vs. treatment with OA only).
14 and 15. The MTDLs were able to maintain or improve the
neuroprotection showed by the combination of the two
precursor fragments at cellular level, which highlights the
advantage of MTDL versus polypharmacology approaches.
Altogether, these data combined with the reduction of toxic
b-amyloid deposits in the SK-APP cells demonstrates the
potential of these compounds to treat complex neurodege-
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
www.angewandte.org
&&&&
These are not the final page numbers!

Angewandte
Research Articles
Chemie
nerative diseases such as AD and other dementias associated
to aging.
On Target BACE1 Directed Synthesis
In order to identify multitarget compounds with BACE1
and kinase inhibitory activity easily and effectively, we
implemented the technology of in situ click chemistry.[37] On
template synthesis enables the formation of a product when
reagents bound to the macromolecule acquire an optimal
orientation for the chemical reaction to occur. In this manner,
Huisgen cyclization can be performed in the absence of
catalyst and at room temperature in a short time frame. We
envisioned that by leveraging this methodology we could
quickly identify BACE1 binders starting from kinase inhibitor
fragments prior to synthesizing them in a bigger scale and thus
saving resources.
Azide 4 and alkyne 7 were chosen to test the in situ click
reaction with BACE1. Derivatives 4 and 7 were mixed in
presence or absence of BACE1 in aqueous buffer and the
mixtures were analyzed by LC/MS-SIM. Concentration of the
enzyme (0.25, 0.5, and 1 mM) and reagents (10, 50, 100, and
200 mM), temperature (20 and 378C), time (24 and 48 h) and
buffer (phosphate-buffered saline (PBS) pH 7.0 or sodium
acetate pH 4.5) were screened in order to find the optimal
conditions to test multiple reactions. Finally, selected experi-
mental conditions were 24 h of incubation at room tempera-
ture using 50 mM sodium acetate buffer at pH 4.5, BACE1 at
0.5 mM and reagents at 100 mM. Only in the case of enzyme
presence, a peak with the molecular weight of the expected
triazole 8 was observed (Figure 9). Similarly, a control
reaction in the presence of BSA was performed and no
triazole MTDL was observed.
Scheme 4. i) 3-bromopropyne, K2CO3, THF, 808C; ii) NaOH, H2O:THF
(1:4 v/v) r.t.; iii) SOCl2, CH2Cl2, 708C; then iv) 6-R-2-amino-benzothia-
zole, THF, 1108C, MW; v) SOCl2, THF, 708C; then vi) ethyl 2-amino-4-
methylthiazole-5-carboxylate, THF, 1208C, MW; vii) 3-bromopropyne,
NaH, DMF, 08C; viii) Diethylmalonate, NaH, DMF, 858C; ix) dodeca-
nohydrazide, DMF, 1608C; x) NaN3, DMF, 808C; xi) EDCI, DMAP,
DMF,r.t., 5 min then 2-amino-benzothiazole, r.t.
following previous described experimental conditions
(Scheme 4).
In total, three different azides (4, 11, and 27) and eight
diverse alkynes (7, 12, 13, 20–23, and 25), with inhibitory
activities in their respective protein kinases LRRK2, CK1d
and GSK3b, were combined in order to further select
multitarget compounds (Table 2).
To implement this methodology, other building blocks
starting from different known protein kinase inhibitors were
designed, inspired by kinase inhibitors in our chemical library.
Thus, five new alkynes derived from inhibitors of CK1d and
GSK3b together with a new LRRK2 azide were synthesized
In situ click chemistry experiments were set incubating
both fragments in the presence or absence of BACE1 and the
reaction was analyzed by SIM LC-MS. A total of nineteen
independent reactions were set, from which we obtained
eleven positive hits, observing that without the presence of
Table 2: In situ click chemistry building blocks with their inhibitory
kinase properties.
Figure 9. LC/MS-SIM chromatograms of the incubation of azide 4 and
alkyne 7: A) without BACE1 and B) with 0.5 mM of BACE1.
&&&&
www.angewandte.org
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
These are not the final page numbers!

Angewandte
Research Articles
Chemie
residue, or cation- p interactions with Arg128) that their
corresponding starting fragments. Finally, a molecular dy-
namics study performed on MTDL compounds 14 and 32,
active and inactive on BACE1 respectively, together with
their starting fragment MBC-2137 show clearly how the active
compound 14 present lower DGBind binding energy than the
inactive 32 as well as the fragment MBC-2137 (Figure 11).
This analysis indicates that MTDL14 presents more favorable
interactions making its affinity for BACE1 higher, which is
directly related to the inhibitory activity of BACE1 (Fig-
ures S8-S13).
Figure 10. Hits obtained by in situ click chemistry reaction with BACE1
after incubation with the corresponding building blocks for 24 hours.
Compounds synthetized to confirm the hypothesis are numbered
above the bars: 28–31 positive hits and 32–33 negative hits. Data
presented here is a result of two independent runs.
BACE1 the subsequent triazole MTD was not being formed
(Figure 10 and S3–S6).
To fully confirm that hits obtained in this protein template
synthesis correspond to BACE1 inhibitors, we synthesized
four different MTDLs based in the availability of starting
building blocks (compounds 28–31). Furthermore, to control
the reaction selectivity, two derivatives more (compounds 32
and 33) were also prepared as negative controls (Scheme 5).
In all cases, 1,2,3-disubstituted triazoles were obtained using
classical CuAAC.
Figure 11. DGBind binding energy calculated using MMGBSA along the
simulations. The DGBind average Æ standard deviation (kcalmolÀ1) is
displayed in the legend.
Determination of Blood–Brain Barrier Penetration of MTDLs
As stated in the introduction, one of the main disadvan-
tages of the MTDLs is their increase in size and subsequent
decline in pharmacokinetic properties. For the treatment of
neurodegenerative diseases, a key pharmacokinetic feature is
the permeation through the blood–brain barrier (BBB) that
determines which compounds will be suitable candidates for
the treatment of these diseases. Although there are several
strategies to increase the BBB permeability of drug candi-
dates using nanoparticles,[38] in an attempt to characterize the
pharmacokinetic profile of our multitarget compounds (8, 14–
17, 28–31) we employed the parallel artificial membrane
permeability assay (PAMPA) to predict the in vitro perme-
ability (Pe) by passive diffusion (Table S3 and Figure S14).
While some compounds were insoluble in the conditions of
the experiment (8, 17 and 31), brain permeation prediction
was calculated for MTDLs derivatives 14–16 and 28–30.
Compounds 15 and 30 showed apparent permeability values
compatible with a positive brain penetration while compound
16 remains in the uncertainty region of prediction (CNS + /
CNS-). Compounds 14 and 29 were not detected in the
acceptor well by HPLC-MS and therefore experimental Pe
was assumed as 0 (Figure 12). These data confirm the
difficulty of MTDLs to maintain good pharmacokinetic
profiles but show two of the new MTDLs as potential brain
penetrant compounds and therefore suitable agents to treat
Scheme 5. i) CuSO4·5H2O, sodium ascorbate, DMF, r.t.
Inhibitory activity of compounds 28–33 was tested on
BACE1 and the corresponding protein kinases (Table 3).
Results confirmed how hits selected by BACE1-templated
synthesis show inhibitory activity on this protease while
negative controls lack this activity. In general, protein kinase
inhibition profile was maintained in the multitarget com-
pounds in a well-balanced manner.
The in silico binding mode of the active triazoles on
BACE1 was also studied, and it resembled those of com-
pounds 3 and 8 (Figure S7). A detailed study of predicted
binding energies (MMGSBA values) and interactions with
key BACE1 residues have been done with MTDLs and the
initial fragments (Table S2), showing lower energy values and
different interactions for MTDLs BACE1 inhibitors (such as
hydrophobic, hydrogen bonds and p-p contacts with Try71
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
www.angewandte.org
&&&&
These are not the final page numbers!

Angewandte
Research Articles
Chemie
Table 3: Structure and activity of the new 1,2,3-triazoles in their respective enzymes.
treatments for multifactorial se-
vere diseases has been noted. We
have here designed and synthe-
tized new multitarget compounds
that maintained their original kin-
ase inhibitory activity adding a new
BACE1 inhibitory effect. There-
fore, we describe new MTDLs with
great potential to treat complex
diseases such as AD where Ab
pathology and aberrant phosphor-
ylation events on tau protein are
key events. The suggested binding
mode of these compounds to
BACE1 was simulated by molec-
ular dynamic calculations showing
how the compounds fit in the
catalytic cavity closing it by direct
interaction with the protein flap.
The new BACE1 inhibitory
Structure
LRRK2
CK1d
GSK3b
BACE1
IC50 (mM)
IC50 (mM)
IC50 (mM)
IC50 (mM)
28
0.71Æ0.26
0.40Æ0.06
–
7.21Æ0.35
29
0.44Æ0.10
12.58Æ0.70
–
11.00Æ1.36
activity was further confirmed in
a
cellular model where these
MTDLs were able to reduce Ab40
and Ab42 levels. Additional cellular
assays were performed to test the
potential of these multitarget com-
pounds for protecting neural cells
in the okadaic acid-induced cell
death model. MTDLs showed neu-
roprotective potential against the
tau phosphorylation toxic effect
with similar or improved potency
than the simultaneous treatment
with an equimolar mixture of their
protein kinase inhibitor precursors.
Besides we have demonstrated for
the first time, how BACE1 can be
used as template to select the frag-
ments that may be exploited to
obtain these and others interesting
MTDLs. Additionally, we experi-
mentally predicted that some of
these MTDLs may penetrate into
the brain by passive diffusion using
PAMPA methodology.
30
3.19Æ1.01
30%[a]
–
8.57Æ0.45
31
–
5.40Æ0.91
26%[a]
6.76Æ0.61
32
33
54%[a]
–
–
–
–
<20%
<20%
98%
Our strategy based on the in-
teraction with the long catalytic
site of BACE1 starting from indi-
vidual protein kinase inhibitors to
[a] Enzyme inhibition at 10 mM. IC50 was not calculated due to poor solubility of the compound.
AD and other dementias. Further studies aiming to improve
brain penetration of these compounds using PLGA nano-
encapsulation are ongoing.[39,40]
obtain new MTDLs has been confirmed. The cellular potency
of the compounds has demonstrated how exploring this
strategy could lead to promising compounds for the treatment
of neurodegenerative diseases and specially Alzheimerꢀs
disease. In vivo evaluation of the brain penetrating multi-
target compounds or the nanoencapsulated ones in a AD
model will finally confirm the translational potential of these
compounds.
Conclusion and Future Perspectives
In the recent decade, an unprecedented increase of
polypharmacology and multitarget directed ligands (MTDLs)
design strategies as the new chance to develop effective
&&&&
www.angewandte.org
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
These are not the final page numbers!

Angewandte
Research Articles
Chemie
[11] L. Martin, X. Latypova, C. M. Wilson, A. Magnaudeix, M. L.
Perrin, C. Yardin, F. Terro, Ageing Res. Rev. 2013, 12, 289 – 309.
[12] R. M. Bailey, J. P. Covy, H. L. Melrose, L. Rousseau, R.
Watkinson, J. Knight, S. Miles, M. J. Farrer, D. W. Dickson,
B. I. Giasson, J. Lewis, Acta Neuropathol. 2013, 126, 809 – 827.
[13] G. Novak, J. R. Streffer, M. Timmers, D. Henley, H. R. Brashear,
J. Bogert, A. Russu, L. Janssens, I. Tesseur, L. Tritsmans, L.
Van Nueten, S. Engelborghs, Alzheimer’s Res. Ther. 2020, 12, 58.
[14] S. Lovestone, M. Boada, B. Dubois, M. Hull, J. O. Rinne, H. J.
Huppertz, M. Calero, M. V. Andres, B. Gomez-Carrillo, T. Leon,
T. del Ser, J. Alzheimer’s Dis. 2015, 45, 75 – 88.
[15] A. Gabelle, S. Schraen, L. A. Gutierrez, C. Pays, O. Rouaud, L.
Buee, J. Touchon, C. Helmer, J. C. Lambert, C. Berr, Alzheimer’s
Dementia 2015, 11, 672 – 680.
[16] J. Krell-Roesch, M. Vassilaki, M. M. Mielke, W. K. Kremers, V. J.
Lowe, P. Vemuri, M. M. Machulda, T. J. Christianson, J. A.
Syrjanen, G. B. Stokin, L. M. Butler, M. Traber, C. R. Jack, Jr.,
D. S. Knopman, R. O. Roberts, R. C. Petersen, Y. E. Geda,
Transl. Psychiatry 2019, 9, 123.
[17] E. Nepovimova, E. Uliassi, J. Korabecny, L. E. Pena-Altamira, S.
Samez, A. Pesaresi, G. E. Garcia, M. Bartolini, V. Andrisano, C.
Bergamini, R. Fato, D. Lamba, M. Roberti, K. Kuca, B. Monti,
M. L. Bolognesi, J. Med. Chem. 2014, 57, 8576 – 8589.
[18] J. M. Miezan Ezoulin, B. Y. Shao, Z. Xia, Q. Xie, J. Li, Y. Y. Cui,
H. Wang, C. Z. Dong, Y. X. Zhao, F. Massicot, Z. B. Qiu, F.
Heymans, H. Z. Chen, Int. J. Neuropsychopharmacol. 2009, 12,
1409 – 1419.
[19] D. Silva, E. Mendes, E. J. Summers, A. Neca, A. C. Jacinto, T.
Reis, P. Agostinho, I. Bolea, M. L. Jimeno, M. L. Mateus,
A. M. F. Oliveira-Campos, M. Unzeta, J. Marco-Contelles, M.
Majekova, R. R. Ramsay, M. C. Carreiras, Drug Dev. Res. 2020,
81, 215 – 231.
[20] A. Sharma, V. Pachauri, S. J. S. Flora, Front. Pharmacol. 2018, 9,
1247.
[21] M. G. Savelieff, G. Nam, J. Kang, H. J. Lee, M. Lee, M. H. Lim,
Chem. Rev. 2019, 119, 1221 – 1322.
Figure 12. BBB prediction of multitarget compounds assayed by
PAMPA methodology. CNS+ (green) for compounds with Pe
>3.20ꢅ10À6 cmsÀ1, CNS- (red) for compounds with Pe
<1.88ꢅ10À6 cmsÀ1 and CNS+/CNS- (yellow) for compounds with Pe
between both values. Data presented here is a result of two independ-
ent runs.
Acknowledgements
This work was supported by MINECO (grant SAF2016-
76693-R and PID2019-105600RB-I00 to A.M.), ISCiii_CI-
BERNED (CB18/05/00040 to A.M and A.G.-R., and CB07/
502 to E.C. and F.B.), ISCiii_FIS (grant PI18/00118 to E.C.),
MECD (FPU16-04466 to V.N., and FPU18-06310 to C.T.B.),
CONICYT-PCI (grant REDES190074 to D.R. and A.M.) and
FONDECYT (grant no. 11180604 to D.R.). V.P. has received
financial support through the Postdoctoral Junior Leader
Fellowship Program (LCF/BQ/PR18/11640007) from “la
Caixa” Banking Foundation.
[22] N. M. Moussa-Pacha, S. M. Abdin, H. A. Omar, H. Alniss, T. H.
Al-Tel, Med. Res. Rev. 2020, 40, 339 – 384.
[23] S. A. Spronk, H. A. Carlson, Proteins Struct. Funct. Bioinf. 2011,
79, 2247 – 2259.
[24] “Discovery of BACE1 Inhibitors for the Treatment of Alzheim-
erꢀs Disease”: Y. K. Hamada, in Quantitative Structure-activity
Relationship (Ed.: F. Kandermili), IntechOpen 2017, pp. 27 – 43.
[25] G. Evin, BioDrugs 2016, 30, 173 – 194.
[26] L. Hong, R. T. Turner 3rd, G. Koelsch, D. Shin, A. K. Ghosh, J.
Tang, Biochemistry 2002, 41, 10963 – 10967.
Conflict of Interest
The authors declare no conflict of interest.
Keywords: Alzheimer’s disease · BACE1 ·
in situ click chemistry · multitarget directed ligands ·
protein kinase inhibitors
[27] J. B. Jordan, D. A. Whittington, M. D. Bartberger, E. A. Sickm-
ier, K. Chen, Y. Cheng, T. Judd, J. Med. Chem. 2016, 59, 3732 –
3749.
[28] A. K. Ghosh, M. Brindisi, Y. C. Yen, E. L. Cardenas, J. R. Ella-
Menye, N. Kumaragurubaran, X. Huang, J. Tang, A. D. Mesecar,
Bioorg. Med. Chem. Lett. 2017, 27, 2432 – 2438.
[29] J. R. M. Coimbra, D. F. F. Marques, S. J. Baptista, C. M. F.
Pereira, P. I. Moreira, T. C. P. Dinis, A. E. Santos, J. A. R.
Salvador, Front. Chem. 2018, 6, 178.
[30] C. C. Hsiao, F. Rombouts, H. J. M. Gijsen, Bioorg. Med. Chem.
Lett. 2019, 29, 761 – 777.
[31] J. Zaldivar-Diez, L. Li, A. M. Garcia, W. N. Zhao, C. Medina-
Menendez, S. J. Haggarty, C. Gil, A. V. Morales, A. Martinez, J.
Med. Chem. 2020, 63, 2638 – 2655.
[32] E. Bonandi, M. S. Christodoulou, G. Fumagalli, D. Perdicchia, G.
Rastelli, D. Passarella, Drug Discovery Today 2017, 22, 1572 –
1581.
[1] M. L. Bolognesi, Curr. Med. Chem. 2013, 20, 1639 – 1645.
[2] M. M. Ibrahim, M. T. Gabr, Neural Regener. Res. 2019, 14, 437 –
440.
[3] R. R. Ramsay, M. R. Popovic-Nikolic, K. Nikolic, E. Uliassi,
M. L. Bolognesi, Clin. Transl. Med. 2018, 7, 1639 – 1645.
[4] S. L. Cheong, S. Federico, G. Spalluto, K. N. Klotz, G. Pastorin,
Drug Discovery Today 2019, 24, 1769 – 1783.
[5] G. W. Rutherford, P. R. Sangani, G. E. Kennedy, Cochrane
Database Syst. Rev. 2003, CD002037.
[6] A. Cavalli, M. L. Bolognesi, A. Minarini, M. Rosini, V. Tumiatti,
M. Recanatini, C. Melchiorre, J. Med. Chem. 2008, 51, 347 – 372.
[7] C. J. Van der Schyf, Expert Rev. Clin. Pharmacol. 2011, 4, 293 –
298.
[8] S. O. Bachurin, E. V. Bovina, A. A. Ustyugov, Med. Res. Rev.
2017, 37, 1186 – 1225.
[33] V. Sebastiꢂn-Pꢃrez, C. Roca, M. Awale, J. L. Reymond, A.
Martinez, C. Gil, N. E. Campillo, J. Chem. Inf. Model. 2017, 57,
2143 – 2151.
[9] G. Bottegoni, A. D. Favia, M. Recanatini, A. Cavalli, Drug
Discovery Today 2012, 17, 23 – 34.
[10] D. J. Selkoe, J. Hardy, EMBO Mol. Med. 2016, 8, 595 – 608.
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
www.angewandte.org
&&&&
These are not the final page numbers!

Angewandte
Research Articles
Chemie
[34] A. Martinez, M. Alonso, A. Castro, C. Perez, F. J. Moreno, J.
Med. Chem. 2002, 45, 1292 – 1299.
[38] S. M. Lombardo, M. Schneider, A. E. Tureli, N. Gunday Tureli,
Beilstein J. Nanotechnol. 2020, 11, 866 – 883.
[39] E. Rojas-Prats, C. Tosat-Bitrian, L. Martinez-Gonzalez, V.
Nozal, D. I. Perez, A. Martinez, Pharmaceutics 2021, 13, 180.
[40] V. Nozal, E. Rojas-Prats, I. Maestro, C. Gil, D. I. Perez, A.
Martinez, Int. J. Mol. Sci. 2021, 22, 3206.
[35] I. G. Salado, J. Zaldivar-Diez, V. Sebastian-Perez, L. Li, L.
Geiger, S. Gonzalez, N. E. Campillo, C. Gil, A. V. Morales, D. I.
Perez, A. Martinez, Eur. J. Med. Chem. 2017, 138, 328 – 342.
[36] I. G. Salado, M. Redondo, M. L. Bello, C. Perez, N. F. Liachko,
B. C. Kraemer, L. Miguel, M. Lecourtois, C. Gil, A. Martinez,
D. I. Perez, J. Med. Chem. 2014, 57, 2755 – 2772.
Manuscript received: May 10, 2021
[37] R. Manetsch, A. Krasinski, Z. Radic, J. Raushel, P. Taylor, K. B.
Sharpless, H. C. Kolb, J. Am. Chem. Soc. 2004, 126, 12809 –
12818.
Revised manuscript received: June 24, 2021
Accepted manuscript online: June 24, 2021
Version of record online: && &&
,
&&&&
&&&&
www.angewandte.org
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
These are not the final page numbers!

Angewandte
Research Articles
Chemie
Research Articles
Drug Discovery
Using different small molecules targeting
protein kinases and BACE1 templated
synthesis, we have obtained well-bal-
anced multitarget compounds. They have
three different biological activities and
may be a promising therapeutic approach
for the treatment of complex diseases
such as AD.
V. Nozal, A. Garcꢀa-Rubia, E. P. Cuevas,
C. Pꢃrez, C. Tosat-Bitriꢁn, F. Bartolomꢃ,
E. Carro, D. Ramꢀrez, V. Palomo,*
A. Martꢀnez*
&&&& — &&&&
From Kinase Inhibitors to Multitarget
Ligands as Powerful Drug Leads for
Alzheimer’s Disease using Protein-
Templated Synthesis
Angew. Chem. Int. Ed. 2021, 60, 2 – 13
ꢄ 2021 The Authors. Angewandte Chemie International Edition published by Wiley-VCH GmbH
www.angewandte.org
&&&&
These are not the final page numbers!