Substituent effect of N-benzylated gramine derivatives that prevent the PP2A inhibition and dissipate the neuronal Ca2+ overload, as a multitarget strategy for the treatment of Alzheimer's disease

Article abstract of DOI:10.1016/j.bmc.2018.04.019

Following the premises of the multitarget-directed ligands approach for the drug R&D against neurodegenerative diseases, where Alzheimer's disease (AD) outstands, we have synthesized and evaluated analogues of the gramine derivative ITH12657 (1-benzyl-5-m

Full text of DOI:10.1016/j.bmc.2018.04.019

Accepted Manuscript
Substituent effect of N-benzylated gramine derivatives that prevent the PP2A
inhibition and dissipate the neuronal Ca2+ overload, as a multitarget strategy for
the treatment of Alzheimer’s disease
Dorleta Gonzalez, Raquel L. Arribas, Lucia Viejo, Rocio Lajarin-Cuesta,
Cristóbal de los Ríos
PII:
S0968-0896(18)30196-2
https://doi.org/10.1016/j.bmc.2018.04.019
BMC 14302
DOI:
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
29 January 2018
6 April 2018
7 April 2018
Please cite this article as: Gonzalez, D., Arribas, R.L., Viejo, L., Lajarin-Cuesta, R., de los Ríos, C., Substituent
effect of N-benzylated gramine derivatives that prevent the PP2A inhibition and dissipate the neuronal Ca2+ overload,
as a multitarget strategy for the treatment of Alzheimer’s disease, Bioorganic & Medicinal Chemistry (2018), doi:
https://doi.org/10.1016/j.bmc.2018.04.019
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Substituent effect of N-benzylated gramine derivatives that prevent the PP2A
inhibition and dissipate the neuronal Ca2+ overload, as a multitarget strategy
for the treatment of Alzheimer’s disease
Dorleta Gonzaleza, Raquel L. Arribasa, Lucia Viejoa,b, Rocio Lajarin-Cuestaa,
Cristóbal de los Ríosa,b,*
aInstituto Teófilo Hernando and Departmento de Farmacología y Terapéutica,
Universidad Autónoma de Madrid, C/ Arzobispo Morcillo, 4, 28029, Madrid, Spain;
bInstituto de Investigación Sanitaria, Servicio de Farmacología Clínica, Hospital
Universitario de la Princesa, 28006 Madrid, Spain
Corresponding author: Cristóbal de los Ríos, e-mail address:
cristobal.delosrios@inv.uam.es
Abbreviations: ACh, acetylcholine; AD, Alzheimer’s disease; Aβ, amyloid β
peptide; [Ca2+]c, cytosolic Ca2+; CALHM1, Ca2+ homeostasis modulator-1 channel;
FBS, fetal bovine serum; Mem, memantine; NCLX, mitochondrial Na+/Ca2+
exchanger; NFT, neurofibrillary tangles; NMDA, N-methyl-D-aspartate; OA,
okadaic acid; PAD, PP2A-activating drug; PP2A, phosphoprotein phosphatase 2A;
PP2Ac, PP2A catalytic subunit C; pNPP, para-nitrophenol phosphate; PPP,
Ser/Thr phosphoprotein phosphatases; VGCC, voltage-gated Ca2+ channels.
1

Abstract
Following the premises of the multitarget-directed ligands approach for the drug
R&D against neurodegenerative diseases, where Alzheimer’s disease (AD)
outstands, we have synthesized and evaluated analogues of the gramine
derivative ITH12657 (1-benzyl-5-methyl-3-(piperidin-1-ylmethyl-1H-indole, 2),
which had shown important neuroprotective properties, such as blocking effect of
voltage-gated Ca2+ channels (VGCC), and prevention of phosphoprotein
phosphatase 2A (PP2A) inhibition. The new analogues present different
substitutions at the pending phenyl ring, what slightly modified their
pharmacological characteristics. The VGCC blockade was enhanced in derivatives
possessing nitro groups, while the pro-PP2A feature was ameliorated by the
presence of fluorine. Chlorine atoms supplied good activities over the two
biological targets aimed; nevertheless that substitution provoked loss of viability at
100-fold higher concentrations (10 µM), what discards them for a deeper
pharmacological study. Overall, the para-fluorine derivative of ITH12657 was the
most promising candidate for further preclinical assays.
Keywords: gramine analogues, multitarget strategy, neuroprotection, Ca2+
overload, Tau hyperphosphorylation, phosphoprotein phosphatase 2A,
Alzheimer’s disease
2

1. Introduction
The improvement of life span in developed countries leads to the aging of
population and the prevalence of associated pathologies, such as
neurodegenerative diseases, where Alzheimer’s disease (AD) highlights. AD has a
remarkable impact on economy and society, enhanced by the lack of efficient
medicines.1 Hence, the search of both innovative and alternative therapeutic
strategies keeps being a priority objective.
Brains of AD patients are mainly characterized by two morphological hallmarks,
which are thought to contribute to the neuronal damage in the brain: the senile
plaques formed by aggregates of the amyloid β peptide (Aβ),2 and neurofibrillary
tangles (NFT),3 whose generation has its origin in the altered phosphorylation of
the microtubule-stabilizing protein Tau. Every year, a plethora of scientific
contributions report possible causes for the rise of these aberrant structures in
brain, as well as the pathological consequences that trigger. In the end,
therapeutic targets studied for AD are not conveniently validated, what leads to a
scientific “tower of Babel” that hinders rational drug discovery programs. Among all
of the pharmacological approaches discussed, three of them have prevailed since
the 90s, 1) the recovery of the cholinergic transmission, generally by inhibiting the
acetylcholine (ACh)-metabolizing enzymes, i.e. the cholinesterases,4 2) the
removal of the Aβ-formed senile plaques, mainly through the inhibition of enzymes
that catalyze the synthesis of Aβ monomers, β-secretase and γ-secretase,5 and, in
a lesser extension, 3) the disruption of the NFT formation in neurons, principally by
inhibiting Tau kinases, such as GSK-3β, cdk-5 or PKA.3 Despite all the scientific
efforts spent up to date, none of these strategies have afforded promising crops,6
as the current marketed drugs [three anticholinesterasics and a N-methyl-D-
3

aspartate sensitive (NMDA) receptor blocker]7 only attack the symptoms with
limited efficacy.
Given the multifactorial nature of AD, it is widely accepted the hypothesis stating
that several physiopathological mechanisms exacerbating neuronal impairment
converge to result in an unstoppable neurodegeneration scenario.8 For this
reason, a great part of the scientific community devoted to drug discovery for
neurodegenerative diseases has realized that a potential optimal drug should
interact with two or more biological targets implicated in AD. Thus, the
development of multitarget-directed ligands has given rise to interesting
compounds probed for AD.9-10 Nonetheless, these have commonly focused on the
same biological targets, being mostly cholinesterase inhibitors that feature an
additional pharmacological property. In this sense, other alternative strategies
keep waiting for an opportunity to demonstrate their therapeutic relevance for AD.
Taking into account all the previous evidences, our research group quests
innovative approaches to face neurodegenerative diseases that, despite having
been proven to be involved in the AD progression, have not been conveniently
aimed with the goal of designing new efficient drugs. For instance, we reported the
first selective blocker of the recently discovered Ca2+ homeostasis modulator-1
channel (CALHM1);11 the polymorphism P86L of this channel has been implicated
in the AD onset, or a new generation of mitochondrial Na+/Ca2+ exchanger (NCLX)
blockers,12 on the basis that NCLX plays a chief role in the neurodegeneration
triggered by neuronal Ca2+ overload. Moreover, we have recently reported the
interesting outcomes of some gramine (1) derivatives (Fig. 1) that exerted a
multitarget profile over voltage-gated Ca2+ channels (VGCC) and Ser/Thr
phosphoprotein phosphatases (PPP).13
4

The neuronal Ca2+ overload, highly contributed by altered VGCC opening,14 is a
common event in many neurodegenerative diseases,15 so several Ca2+ antagonists
have been positioned as investigational drugs. As far as the Ser/Thr
phosphatases, they comply an essential function as dephosphorylating enzymes
of the Tau protein, thus counteracting the NFT synthesis, being phosphatase 2A
(PP2A) the principal enzyme.16 Indeed, PP2A is found inactivated or inhibited in
both in vivo and in vitro models of AD,17-18 as well as in post-mortem AD patient
brains.19 Hence, there is currently an increasing interest in developing drugs able
to reduce the endogenous inhibition of PP2A, thus named PP2A-activating drugs
(PADs).20-21 Indeed, some of them are being clinically tested for diseases where
PP2A activity appears depressed.22-23 Consequently, the multitarget-directed
ligands that we have reported, by blocking VGCC and avoid the PP2A inhibition,
elicited neuroprotection in various in vitro models of neurodegeneration, standing
out the derivative ITH12657 (1-benzyl-5-methyl-3-(piperidin-1-ylmethyl-1H-indole,
2; Fig. 1).13 Attempting to optimize its therapeutic strengths, we herein present the
evaluation of how aromatic substitution at the pending phenyl ring could modulate
the affinity with both VGCC and PP2A, and their consequences over the
neuroprotection observed in an experimental model of neuronal damage related to
AD.
5

Fig. 1. Chemical structures of gramine and the lead compound ITH12657 (2),13
together with the chemical substitution herein proposed.
2. Results and discussion
2.1. Chemistry
We have selected various substituents possessing different steroelectronic
behavior that would affect the phenyl ring electronegativity, but also eligible to
make hydrogen bonds or to pose into hydrophobic binding pockets within the
PP2A. Thus, halogens, methyl group or the highly electron-demanding and polar
nitro group, were selected.
Scheme 1. Synthesis of N-benzylated 3-piperidinylmethylgramine derivatives
Functionalization of the indole nitrogen was carried out by deprotonation with NaH
in DMSO, followed by addition of the corresponding benzyl bromide bearing the
corresponding substitution, furnishing intermediates 3–8 in good yields, except for
the reaction to obtain the o-nitro derivative (32%), presumably due to the stereo-
electronic repulsions with the negative-charged indole amide. Otherwise, the
presence of nitro groups provoked reaction crudes with high presence of by-
6

products, likely by alkylation on C3 or C7, which made chromatographic
purifications necessary.
Next, a Mannich-type reaction was carried out to prepare the gramine derivatives
9–14 analogues to the head compound 2. The order of reagents addition was
essential to get the best chemical yields, so piperidine and formaldehyde were
firstly mixed in glacial acetic acid and stirred for 10 min, to subsequently be added
to intermediates 3–8. It is worthwhile mentioning that 3-aminomethylindoles are
sensitive to low pH in aqueous media, undergoing a retro-Mannich transformation
that generates 3-methylidenindoles. To avoid such by-reaction, both quenching
and extractions were carried out with basic (pH = 14) aqueous phases. Oily
products 9–14 were obtained with medium to excellent yields (50–92%). As keys
to confirm the full conversion of intermediates into products, the aromatic doublet
assigned to H3 disappeared from the crude 1H NMR, along with the rise of a two
proton-integrated singlet that corresponded to the new CH2 bonded to C3.
Although nicely dissolved in DMSO, compounds 9–14 presented partial solubility
in water (below 0.1 mM). With the goal of enhancing both aqueous solubility and
manipulation for the biological assays, they were salinized to their oxalate salts by
treatment with oxalic acid 1M in dry ethyl acetate (Supplementary data), obtaining
white and powder solids with high yields (72–91%), and improved aqueous
solubility (10 mM). The foreseeable protonation at the piperidine nitrogen was
confirmed by the low-field shifting of the neighboring C3CH2 protons (from 3.58 to
4.33 ppm) and the piperidine signals, together with a broadening of the H2” signal.
2.2. Effect of compounds 9–14 on the neuronal Ca2+ overload
7

We assessed the ability of compounds to minimize altered Ca2+ elevations into
neurons by two ways. Firstly, warned by the documented pharmacological results
of 2 and its analogues,13 we subjected SH-SY5Y neuroblastoma cells with high K+
(70 mM), which evokes cell depolarization and VGCC opening.24 The consequent
cytosolic Ca2+ ([Ca2+]c) increase was dramatically prevented by compounds 9–14,
pre-incubated at the concentration of 10 µM (Table 1). This huge blockade
prompted us to analyze their IC50 values, defined as the concentration of
compound capable of reducing by 50% the 70 mM K+-triggered Ca2+ entry into SH-
SY5Y cells (Fig. 2). Compounds incubated at several concentrations depicted nice
dose-response curves, with similar IC50 values, between 1.8 and 4.8 µM, though
substitutions at meta position slightly offered better IC50 data. Iwata et al reported
closely related gramine derivatives, possessing benzyl substituent at the indole
NH, which blocked VGCC in a similar fashion.25 By electrophysiological
experiments, under the patch-clamp whole-cell configuration, we have proposed
that this family of gramine derivatives, leaded by 2, would directly interact to the
VGCC ionic channel.13
Secondly, keeping our attention on the participation of neuronal Ca2+ overload in
physiopathological processes driving to neurodegeneration, no one would dispute
the participation of NMDA receptors in the progression of AD.26 Indeed, one of the
sole four prescribed drugs for AD is the NMDA receptor blocker memantine.7 For
this reason, we were interested in assessing the blocking effect of compounds 2
and 9–14 on NMDA receptors. To achieve this purpose, we selected a neuronal
type that possesses functional NMDA-sensitive glutamate receptors, i.e. the rat
cortical embryonic neurons,27 inducing [Ca2+]c increase through these channels with
NMDA, at the concentration of 10 µM (Table 1).
8

Table 1
Effect of compounds 2 and 9–14, tested at 10 µM, on the neuronal Ca2+ increases
induced by either depolarization (70 mM K+) or NMDA receptors recruitment
(NMDA 10 µM),a measured with the Ca2+-sensitive fluorescent probe Fluo-4.
c
Compounds
R
70 K+b
NMDA
-
Nifedipine
-
32
-
6
Memantine
-
41 14
2
H
87
94
97
95
79
90
90
8
1
1
3
2
6
4
42
32
3
3
9
p-CH3
m-Cl
p-Cl
p-F
o-NO2
m-NO2
10
11
12
13
14
37 14
52
34
1
6
24 16
22
6
aData are expressed as % blockade compared with the control situation [neurons
only incubated with vehicle (DMSO 0.1%)], and are means SEM of at least four
independent experiments in triplicates. bSH-SY5Y neuroblastoma cells exposed to
high extracellular concentration of KCl (70 mM). cRat cortical embryonic neurons,
9

from two different pregnant rats, stimulated with NMDA 10 µM. Nifedipine28 and
memantine7 at concentrations of 10 µM were used as standard VGCC and NMDA
receptors blockers, respectively. Compounds were tested at the concentration of
10 µM.
The consequent elevation of [Ca2+]c by the contest of NMDA receptors was slightly
reduced by compounds 9–14 preincubated at 10 µM (Table 1), averaging a 30%
blockade. Compound 11, which presents a chlorine atom at para position, was the
best NMDA receptors blocker, as halved the Ca2+ entry at 10 µM (Table 1).
Preliminary structure-activity relationships point out that substituents well
accommodated into hydrophobic pockets, like alkyl groups or halogens, would
favor the blockade of the NMDA-evoked Ca2+ entry. Comparing with blockade data
from the experiments of depolarization-induced Ca2+ entry, compounds 9–14 were
much better blockers of VGCC than of NMDA receptors, as 2 and its derivatives
halved the 70 K+-evoked Ca2+ entry in SH-SY5Y cells at concentrations around 3
µM (Fig. 2). This differential blockade is consistent with data from literature, where
gramine and its derivatives have exerted substantial blockade of VGCC,25 unlike
that reported over NMDA receptors, where gramine did not show blocking
activities at higher concentrations.29
10

Fig. 2. Dose-response curves of the blockade exerted by compounds 2 (R = H,
black line) and 9–14 [R = CH3 (yellow line), Cl (blue lines), F (green line), NO2
(reddish lines)] on the cytosolic Ca2+ ([Ca2+]c) increases originated by 70 mM K+-
evoked VGCC opening in SH-SY5Y cells. Data points are presented as means
SEM of three experiments in triplicates. Curves were built under non-linear
regression fit with variable slope, of log (compound) vs normalized responses
respect to the ∆[Ca2+], obtained from cultures only preincubated with vehicle
(DMSO 0.1%). (COLOR SHOULD BE USED IN PRINT, PLEASE)
The little blocking effect of compounds 9–14 over NMDA receptors could be due to
an indirect interaction, as it has been reported that active PP2A generates a stable
complex with NMDA receptors, and thus catalyzing the dephosphorylation of two
serines at the NMDA-NR1 subunit. This dephosphorylation desensitizes NMDA

receptors, producing a lesser Ca2+ influx.30 Therefore, should compounds 9–14
behaved as PAD, a possibility exists that this activity was responsible of the slight
NMDA receptors blockade observed in the Fluo-4-monitored Ca2+ entry
measurements under NMDA stimulation. To demonstrate the pro-PP2A activity of
compounds 9–14, we conducted phosphatase activity assays in SH-SY5Y cells
compromised by the PP2A inhibitor okadaic acid (OA).
2.3. Effect of compounds 9–14 on the PP2A-performed phosphatase activity
The inhibition of PP2A performed by OA is a widely used experimental model of
AD, due to this polyether toxin causes PP2A dysfunction and thereby causing Tau
hyperphosphorylation and further NFT generation.31 The administration of OA in
animal models elicits manifestations of cognitive impairment compatible with AD,
such as both short and long-term memory formation,32 and spatial memory
deficits.33 In addition, people intoxicated with OA, from the intake of dinoflagellates-
contaminated seafood, experimented memory losses that were more pronounced
in elderly.34 Hence, the scope of the inhibition of PP2A by OA allows us to analyze
the ability of our compounds to recover the phosphatase activity in situations were
PP2A is depressed, as those observed in AD. SH-SY5Y cells mainly express the
PP2A isoform with the ubiquitously found PR55α B regulatory subunit.35 Thus, this
cell line can be considered a reliable model for appraising the PP2A impairment
occurred in AD brains. Applied to SH-SY5Y cells at 15 nM for 18 h (Fig. 3, top),
OA did not compromise cell viability, but provoked a noticeably decay of the
phosphatase activity by 25 2% (Fig. 3), recorded by the method of p-nitrophenol
phosphate (pNPP), which is a colorimetric dye hydrolyzed by phosphatase
12

enzymes that produce chromogenic (λ = 405 nm) p-nitrophenolate in basic
environment.. The PP2A activity fall was prevented by the presence of most of the
compounds, pre- and co-incubated at 0.1 µM (Fig. 3), standing out the p-
fluorinated compound 12, which kept the phosphatase activity of SH-SY5Y cells in
a 93 3% comparing with control cells (cells not exposed to OA), which implies a
73% recovery of the PP2A-dependent phosphatase activity comparing with the OA
group. Therefore, compounds 9–12, acting as PADs,20 improved the effect of the
head 2, confirming that, for this family of 1 derivatives, some substitutions at
pending phenyl ring could favor a better fitting into the PP2A binding site.
Memantine 10 nM, was used as standard in these experiments, because of its
reversal effect in situations of PP2A inhibition.36
Fig. 3. Effect of compounds 9–14 on the PP2A-leaded phosphatase activity
compromised by okadaic acid (OA) in SH-SY5Y cells, measured by the method of

the para-nitrophenol phosphate (pNPP). Data are means SEM of at least three
independent experiments, normalized over the maximum phosphatase activity
(control), obtained from SH-SY5Y cells not exposed to OA. Memantine (Mem) at
10 nM was used as standard in these experiments. ###p < 0.001, compared with
control group (white bar), ***p < 0.001, **p < 0.01, *p < 0.05 compared with cells
stimulated with OA in absence of compounds (black bar).
2.4 Molecular docking
We hypothesize that compounds 9–14 exert their PAD profile by a direct
interaction with PP2A. Computational docking experiments were executed to
support this proposal, using the constitutive PP2A AC dimer, deposited in the
Protein Data Bank (2IE4, www.rcsb.org).37 Thus, we predicted the best pose of
compound 12 that would explain why it is the best PAD (Fig. 4). The molecular
docking revealed several contacts between 12 and various amino acids into the
PP2A catalytic subunit C (PP2Ac) unable to hinder the catalysis played by PP2A.
It has been documented that the selective inhibitor OA binds PP2Ac in a vast area,
where its C1–C12 fragment blocks the hydrolytic site of PP2Ac, as shown in Fig.
4A, and the two spare C19–C27 and C29–C38 fragments only confer affinity and
selectivity over PP1, respectively.38-39 Interestingly, compound 12, in a slightly
different manner to that performed by 2,13 bound preferentially to amino acids that
also bind fragments the C19–C27 and C–29–C38 fragments of OA (Fig. 4A),
highlighting Trp200, together with both Arg214 and Gln122 skeletons, which form
a hydrophobic pocket where both indole and the pending phenyl ring of 2 pose
(Fig. 4B). His118 illustrates a strong H-bond with the charged nitrogen of the
14

piperidine moiety and, in addition, a weak H-Bond can be appreciated between the
skeleton NH of Ile123 and the fluorine atom of 12 (Fig. 4C). Hydrogen bonding
with fluorine, despite its weakness, has been detected in several crystal PDB
structures, and explains improvements in pharmacodynamic outcomes,40 as what
is found out herein.
(A)
15

(B)

(C)

Fig. 4. (A) Molecular docking of compound 12 (yellow) within PP2A catalytic
subunit (PP2Ac) binding site (main interacting residues and Mn2+ ions in gray
color), including the pose shaped by OA (green). (B) Pose of 12 into the
hydrophobic pocket built by Arg214, Gln122, and Trp200. (C) Hydrogen bonds
between 12 and His118 and Ile123 (energy scores provided in Supplementary
Material, Table S1)
2.5. Cell viability experiments
The positive outcomes on the two therapeutic targets aimed, prevention of the
neuronal Ca2+ overload and reactivation of PP2A, prompted us to assess the
neuroprotective potential of compounds 9–14 with a in vitro model of PP2A
inhibition. The administration of OA depicts a plethora of physiopathological events
beyond the expected Tau hyperphosphorylation and subsequent NFT formation,
which collectively contribute to the AD onset. For instance, the PP2A inhibition

exacerbated by OA drew Aβ aggregation,41 activation of several kinases, such as
ERK1/2 and PKC,42 decrease of ACh levels,33 activation of NMDA receptors,43 free
radical generation,44 mitochondrial dysfunction,45 oxidative DNA damage,46 and
sustained activation of VGCC,47 among others. All of these observations make OA
a reliable tool to study neuroprotective drugs addressed to AD. When exposed to
SH-SY5Y cells at the concentration of 20 nM for 20 h, OA substantially diminished
the cell viability by 50.1 0.2 % respect to that of untreated cells. In this scenario,
the pre- and co-administration of compounds 9–14 (0.1 µM) mitigated this
decrease in a very similar manner, augmenting the SH-SY5Y cells viability by
about 67%. The o-nitro-derivate 13 the best neuroprotectant of the family, as it
elevated cell viability to 70 3%, what supposes a recovery of 40%. Otherwise,
none of the compounds overcame the cell viability of OA-challenged cells
incubated with compound 2 (73 3%). Memantine was used as reference
compound due to its confirmed neuroprotective efficacy in situations of PP2A
depression.36
19

Fig. 5. Effect of compounds 2 and 9–14 on the cell viability challenged by OA at
20 nM in SH-SY5Y cells, measured by the method of the MTT reduction. Data are
means SEM of four independent experiments, and expressed as % cell viability,
normalized respect to the cell viability of SH-SY5Y cells non-incubated with OA
nor compounds. Memantine (Mem) at 10 nM was used as standard in these
experiments. ###p < 0.001, compared with basal group (white bar), ***p < 0.001, **p
< 0.01, compared with cells exposed to OA in absence of compounds (black bar).
Likewise, an additional feature to take into account is the possible toxicity by
themselves. Thus, compounds 9–14 per se exposed to SH-SY5Y cells at 3 µM, a
30-fold higher concentration than the used in neuroprotection assays, did not
diminished the cell viability (Fig. S1, supplementary data), which was only
compromised from 10 µM, underlining chlorinated derivatives 10 and 11.

Conversely, the fluorinated derivative 12 scarcely reduced cell viability at 10 µM, in
a non-statistically significant fashion.
3. Conclusions
In summary, we have studied the influence of the chemical substitution at the
pending phenyl ring over the pharmacological properties of N-benzylgramine
derivatives studied as neuroprotectants, regarding to their modulating effect on
both PP2A and cell Ca2+ overload, demonstrating that these activities can be
modified depending on the substituent selected. Nitro groups at the phenyl
achieved neuroprotection in this series by enhancing VGCC blockade, while
fluorine improved the recovery of the PP2A-conducted phosphatase activity.
Chlorine atoms provided derivatives with good pharmacological activities in all the
assays, but dramatically damaged SH-SY5Y cells at 10 µM (Fig. S1) and
therefore, though presenting a high therapeutic index, their further preclinical
appraisal, as well as the preparation of new chlorinated 2 derivatives, are not
advisable. On the other hand, regarding to the neuroprotective assays, despite
there are not significant differences among all the analyzed compounds, the facts
that compound 12, the best PAD of the family, was not the best neuroprotectant,
and the good neuroprotective actions of 13, which was unable to prevent the
inhibitory actions of OA PP2A, should indicate us that neuroprotection against a
model of PP2A impairment can be achieved downstream, where other biological
targets, such as VGCC or NMDA receptors, are recruited. This can be considered
an accolade to the multitarget-directed ligands strategy, pointing out that
21

pharmacological actions by neuroprotectants over different biological targets
should bear in mind to accurately describe drug mechanisms of action.
The derivative 12, which possesses a hydrogen-isosteric fluorine atom at para
position of the phenyl ring, is presumably the most stable against the CYP-
performed para-oxidation. Moreover, it showed the less toxicity of the family,
measured as cell viability of SH-SY5Y cells exposed to compounds for 48 h, being
compound 12 non-toxic at 100-fold concentrations than those where it protected
against OA neurotoxicity and PP2A inhibition. Once evaluated, compound 12
seems to be the most optimal candidate for further preclinical studies by its good
down-regulating effect over cell Ca2+ overload, excellent recovery of the OA-
inhibited PP2A activity, and lack of toxicity at high concentrations, together with a
promising neuroprotective profile.
4. Materials and methods
4.1. General
5-methylindole, NaH, selected arylmethyl bromides, piperidine, 37% aqueous
formaldehyde, OA sodium salt, memantine, KCl, nifedipine, the colorimetric dye
MTT, melatonin, glutamate, and reagents for SH-SY5Y cells culture, were
purchased from Merck (Madrid, Spain). Acetic acid was purchased from Alfa-
Aesar (Karlsruhe, Germany). Sodium sulfate was purchased from LabKhem
(Mataró, Spain). Kit for the determination of phosphatase activity (pNPP assay)
and solvents used for syntheses, isolation, and chromatographic separations, with
analytical grade, were purchased from VWR (Barcelona, Spain) and, when used
for the preparation of final compounds, neutralized from acid traces with basic
22

alumina (Merck, Madrid, Spain). Fluorescent dye Fluo-4/AM and reagents for rat
cortical neurons culture were purchased from Life Technologies (Alcobendas,
Spain). Reactions were monitored with TLC at 254 nm UV light, using silica gel (60
Å) as stationary phase for the N-alkylation, and basic alumina (60 Å) for the
Mannich-type reaction, both purchased from Merck. Reactions were carried out in
glassware systems purged from air with vacuum/argon cycles and run under Ar
atmosphere. When necessary, chromatographic purifications were carried out with
automatized flash chromatographic station (Biotage, Uppsala, Sweden), using pre-
charged columns. Melting points were obtained with a Stuart SMP-10 apparatus
(Staffordshire, UK), without correction. IR spectrum was obtained in a Bruker
IFS60v FTIR spectrophotometer. MS spectra were acquired in an ABSciex
QSTAR spectrometer under the high-resolution configuration with electrospray as
ionization source. 1H and 13C NMR spectra were acquired in a Bruker AVANCE
300 MHz station (corresponding to a 75.4 MHz for the 13C detection), expressed as
ppm, and calibrated upon the residual proton signal from deuterated solvents.
Purity of final compounds was confirmed by elemental analysis in a LECO CHNS-
932 station.
4.2. General procedure for the N-alkylation of 5-methyl-1H-indole
Following an experimental procedure recently described,13 NaH (1.2 equiv, 60% in
mineral oil) was added to a solution of 5-methyl-1H-indole (1 equiv) in DMSO (5
mL/mmol) and stirred at rt for 2 h. Afterwards, the corresponding arylmethyl
bromide (1.2 equiv) was injected to the reaction mixture, which was stirred at rt for
several hours up to no significant evolution of the reaction or the complete
disappearance of substrate. Reaction was quenched by addition of water (10
mL/mmlol) and extracted with CH2Cl2 (3 ⨯ 30 mL/mmol), dried (Na2SO4), filtered,
23

and evaporated. When necessary, the crude was purified with automatized flash
chromatography, yielding pure intermediates with spectral data according to their
structure.
4.3. 5-methyl-1-(4-methylbenzyl)-1H-indole (3)
Following the general procedure 4.2, 5-methyl-1H-indole (200 mg, 1.52 mmol)
[NaH 60% in mineral oil (74 mg, 1.83 mmol), 4-methylbenzyl bromide (338 mg,
1.83 mmol)] yielded 3 (299 mg, 84%). 1H NMR (CDCl3, 300 MHz) δ 7.46 (bs, 1H,
H4), 7.20 (d, 1H, J6-7 = 8.4 Hz, H7), 7.13–7.01 (m, 6H, Ar), 6.48 (d, 1H, J2-3 = 3.3
Hz, H3), 5.27 (s, 2H, CH2), 2.47 (s, 3H, C5CH3), 2,33 (s, 3H, C4’CH3).
4.4. 1-(3-chlorobenzyl)-5-methyl-1H-indole (4)
Following the general procedure 4.2, 5-methyl-1H-indole (200 mg, 1.52 mmol)
[NaH 60% in mineral oil (74 mg, 1.83 mmol), 3-chlorobenzyl bromide (388 mg,
1.83 mmol)] yielded 4 (318 mg, 82%). 1H NMR (CDCl3, 300 MHz) δ 7.51 (bd, 1H,
H4), 7.32–6.99 (m, 7H, Ar), 6.55 (bd, 1H, H3), 5.32 (s, 2H, CH2), 2.52 (s, 3H, CH3).
4.5. 1-(4-chlorobenzyl)-5-methyl-1H-indole (5)
Following the general procedure 4.2, 5-methyl-1H-indole (200 mg, 1.52 mmol)
[NaH 60% in mineral oil (74 mg, 1.83 mmol), 4-chlorobenzyl bromide (388 mg,
1.83 mmol)] yielded 5 (258 mg, 66%). 1H NMR (CDCl3, 300 MHz) δ 7.55 (bd, 1H,
H4), 7.32 (d, 2H, J2’-3’ = 8.4, H3’, H5’), 7.20 (d, 1H, J6-7 = 8.1, H7), 7.14–7.10 (m,
24

2H, H6, H2), 7.06 (d, 2H, J2’-3’ = 8.4, H2’, H6’), 6.58 (d, 1H, J2-3 = 3.0 Hz, H3), 5.28
(s, 2H, CH2), 2,55 (s, 3H, CH3).
4.6. 1-(4-fluorobenzyl)-5-methyl-1H-indole (6)
Following the general procedure 4.2, 5-methyl-1H-indole (200 mg, 1.52 mmol)
[NaH 60% in mineral oil (74 mg, 1.83 mmol), 4-fluorobenzyl bromide (346 mg, 1.83
mmol)] yielded 6 (256 mg, 70%). 1H NMR (CDCl3, 300 MHz) δ 7.44 (bs, 1H, H4),
7.15 (d, 1H, J6-7 = 8.4, H7), 7.09–6.94 (m, 6H, Ar), 6.48 (bd, 1H, J2-3 = 3.0 Hz, H3),
5.25 (s, 2H, CH2), 2.45 (s, 3H, CH3).
4.7. 5-methyl-1-(2-nitrobenzyl)-1H-indole (7)
Following the general procedure 4.2, 5-methyl-1H-indole (200 mg, 1.52 mmol)
[NaH 60% in mineral oil (74 mg, 1.83 mmol), 2-nitrobenzyl bromide (395 mg, 1.83
mmol)] yielded 7 (131 mg, 32%). 1H NMR (CDCl3, 300 MHz) δ 8.17 (m, 1H, H3’),
7.48 (s, 1H, H4), 7.40 (m, 2H, Ar), 7.11 (d, 1H, J2-3 = 3.0 Hz, H2), 7.02 (m, 2H, Ar),
6.56 (d, 1H, J2-3 = 3.0 Hz, H3), 6.47 (m, 1H, H5’),5.74 (s, 2H, CH2), 2.46 (s, 3H,
CH3).
4.8. 5-methyl-1-(3-nitrobenzyl)-1H-indole (8)
Following the general procedure 4.2, 5-methyl-1H-indole (200 mg, 1.52 mmol)
[NaH 60% in mineral oil (74 mg, 1.83 mmol), 3-nitrobenzyl bromide (395 mg, 1.83
mmol)] yielded 8 (264 mg, 65%). 1H NMR (CDCl3, 300 MHz) δ 8.11 (d, 1H, J5’-6’
=
25

6.3 Hz, H4’), 8.04 (s, 1H, H2’), 7.49–7.42 (m, 2H, H4, H5’), 7.31 (bd, 1H, J6’-5’ = 7.8
Hz, H6’), 7,12–7,08 (m, 2H, H2, H7), 7.01 (bd, 1H, J6-7 = 8.7, H6), 6.53 (d, 1H, J2-3 =
3,3 Hz, H3), 5.39 (s, 2H, CH2), 2,45 (s, 3H, CH3).
4.9. General procedure for the synthesis of 1-arylmethyl-5-methyl-3-(piperidin-1-
methyl)-1H-indoles 9–14
Following an experimental procedure recently described,13 piperidine (1 equiv) and
formaldehyde (37%aq, 1 equiv) were stirred in glacial acetic acid (3 mL/mmol) for
10 min. Then, this mixture was added to an ice-cold, round-bottomed flask with the
corresponding intermediate 3–8 (1 equiv). The reaction mixture was stirred at 0 º C
for 5 min, then at rt for several hours until the reaction was complete (TLC).
Reaction was quenched by addition of NaOH (30%aq) up to pH 14, extracted with
neutralized CH2Cl2 (3 × 30 mL/mmol), dried (anh. Na2SO4), filtered, and evaporated
to get compounds 9–14 as yellowish oils, which did not require further
chromatographic purification in most of the cases. Compounds 9–14 were
dissolved in dry ethyl acetate (10 mL/ mmol) under inert atmosphere, and a
solution of oxalic acid (1 equiv, 1M in dry ethyl acetate) was dropwisely injected.
Salinization was allowed to proceed for 2 h and the fine white precipitate formed
was isolated by centrifugation (10 min, 750 rpm) accompanied by dry ethyl acetate
wash-ups. Traces of ethyl acetate were removed by vacuum. Oxalate salts of
compounds 9–14 showed spectral and analytical data according to their structure.
4.10. 5-methyl-1-(4-methylbenzyl)-3-(piperidin-1-ylmethyl)-1H-indole (9)
26

Following the general procedure 4.9, reaction of 3 (150 mg, 0.64 mmol) [piperidine
(54 mg, 0.64 mmol), formaldehyde (30%aq, 47 µL, 0.64 mmol)] yielded 9 (195 mg,
92%) as a yellow oil that did not need further chromatographic purification. 1H
NMR (acetone-d6, 300 MHz) δ 7.49 (s, 1H, H4), 7.22 (d, 1H, J6-7 = 8.4 Hz, H7), 7.17
(s, 1H, H2), 7.11–7.05 (AA’BB’, 4H, H2’, H3’), 6.91 (dd, 1H, J6-7 = 8.4 Hz, J4-6 = 1.5
Hz, H6), 5.30 (s, 2H, N1CH2C1’), 3.58 (s, 2H, C3CH2), 2.38 (m, 7H, H2’’, C5CH3),
2.26 (s, 3H, C4’CH3), 1.50 (m, 4H, H3’’), 1.40 (m, 2H, H4’’). Its oxalate salt was
prepared following the procedure described in 4.9. Compound 9 (178 mg, 0.54
mmol) [oxalic acid 1M (49 mg, 0.54 mmol)] yielded its oxalate salt as a white solid
(142 mg, 73%). Mp 178–180 ºC; 1H NMR (DMSO-d6, 300 MHz) δ 7.60 (s, 1H, H4),
7.54 (s, 1H, H2), 7.36 (d, 1H, J6-7 = 8.4 Hz, H7), 7.10 (s, 4H, H2’, H3’), 6.99 (d, 1H,
J6-7 = 8.4 Hz, H6), 5.34 (s, 2H, N1CH2C1’), 4.33 (s, 2H, C3CH2), 3.08 (m, 4H, H2’’),
2.38 (s, 3H, C5CH3), 2.23 (s, 3H, C4’CH3), 1.70 (m, 4H, H3’’), 1.48 (m, 2H, H4’’);
13C NMR (DMSO-d6, 75.4 MHz) δ 164.7, 136.8, 134.8, 134.3, 132.0, 129.2, 128.8,
128.5, 127.2, 123.6, 118.5, 110.5, 102.3, 51.3, 50.6, 49.2, 22.7, 21.6, 21.3, 20.7;
IR (KBr) ν 2920, 1630, 1380, 1176, 730 cm-1; HRMS (API-ES) m/z (M+) calcd. for
C23H29N2 333.2325, found, 333.2340 ( 4.5 ppm). Anal. C23H28N2·C2O4H2·1/2 H2O (C,
H, N).
4.11. 1-(3-chlorobenzyl)-5-methyl-3-(piperidin-1-ylmethyl)-1H-indole (10)
Following the general procedure 4.9, reaction of 4 (150 mg, 0.59 mmol) [piperidine
(50 mg, 0.59 mmol), formaldehyde (30%aq, 44 µL, 0.59 mmol)] yielded 10, which
was purified by basic alumina flash chromatography with ethyl acetate/hexane
mixtures as eluent, obtaining a yellow oil (195 mg, 76%). 1H NMR (acetone-d6, 300
27

MHz) δ 7.52 (bs, 1H, H4), 7.34–7.27 (m, 2H, Ar), 7.23–7.21 (m, 2H, Ar), 7.12 (m,
2H, Ar), 6.94 (dd, 1H, J6-7 = 8.4, J4-6 = 1.2 Hz, H6), 5.40 (s, 2H, N1CH2C1’), 3.60 (s,
2H, C3CH2), 2.39 (m, 7H, H2’’, CH3), 1.52 (m, 4H, H3’’), 1.41 (m, 2H, H4’’). Its
oxalate salt was prepared following the procedure described in 4.9. Compound 10
(40 mg, 0.11 mmol) [oxalic acid 1M (10 mg, 0.11 mmol)] yielded its oxalate salt as
a white solid (42 mg, 84%). Mp 175–177 ºC; 1H NMR (DMSO-d6, 300 MHz) δ 7.65
(s, 1H, H4), 7.57 (s, 1H, H2), 7.39 (m, 1H, H5’), 7.34 (m, 2H, H7, H4’), 7.22 (s, 1H,
H2’), 7.14 (bd, 1H, J6’-7’ = 6.5 Hz, H6’), 7.02 (bd, 1H, J6-7 = 8.4 Hz, H6), 5.45 (s, 2H,
N1CH2C1’), 4.35 (s, 2H, C3CH2), 3.09 (bs, 4H, H2’’), 2.39 (s, 3H, CH3), 1.71 (m,
4H, H3’’), 1.49 (m, 2H, H4’’); 13C NMR (DMSO-d6, 75.4 MHz) δ 164.3, 140.5, 140.5,
134.3, 133.2, 131.9, 130.5, 128.9, 128.4, 127.4, 126.8, 125.7, 123.7, 118.6, 110.3,
51.4, 50.5, 48.6, 22.7, 21.5, 21.2; HRMS (API-ES) m/z (M+) calcd. for C22H26N2Cl
353.1779, found, 353.1787 ( 2.3 ppm). Anal. C22H25N2Cl·C2O4H2 (C, H, N).
4.12. 1-(4-chlorobenzyl)-5-methyl-3-(piperidin-1-ylmethyl)-1H-indole (11)
Following the general procedure 4.9, reaction of 5 (150 mg, 0.59 mmol) [piperidine
(50 mg, 0.59 mmol), formaldehyde (30%aq, 44 µL, 0.59 mmol)] yielded 11, which
was purified by basic alumina flash chromatography with ethyl acetate/hexane
mixtures as eluent, obtaining a yellow oil (162 mg, 78%). 1H NMR (acetone-d6, 300
MHz) δ 7.52 (s, 1H, H4), 7.31 (d, 2H, H2’), 7.21–7.14 (m, 4H, H2, H7, H3’), 6.93
(bd, 1H, J6-7 = 8.4 Hz, H6, 5.34 (s, 2H, N1CH2C1’), 3.59 (s, 2H, C3CH2), 2.39 (m,
7H, H2’’, CH3), 1.52 (m, 4H, H3’’), 1.41 (m, 2H, H4’’). Its oxalate salt was prepared
following the procedure described in 4.9. Compound 11 (114 mg, 0.32 mmol)
[oxalic acid 1M (29 mg, 0.32 mmol)] yielded its oxalate salt as a white solid (115
28

mg, 81%). Mp 174–176 ºC; 1H NMR (DMSO-d6, 300 MHz) δ 7.63 (s, 1H, H4), 7.55
(s, 1H, H2), 7.36 (m, 3H, H2’, H7), 7.20 (d, 2H, J2’-3' = 8.4 Hz, H3’), 6.99 (bd, 1H, J6-7
= 8.6 Hz, H6), 5.41 (s, 2H, N1CH2C1’), 4.34 (s, 2H, C3CH2), 3.09 (bs, 4H, H2’’),
2.38 (s, 3H, CH3), 1.70 (m, 4H, H3’’), 1.49 (m, 2H, H4’’); 13C NMR (DMSO-d6, 75.4
MHz) δ 164.7, 136.9, 134.3, 132.1, 132.0, 129.0, 128.9, 128.6, 128.5, 123.7,
118.6, 110.4, 102.5, 51.3, 50.5, 48.6, 22.6, 21.5, 21.2; HRMS (API-ES) m/z (M+)
calcd. for C22H26N2Cl 353.1779, found, 353.1788 ( 2.5 ppm). Anal.
C22H25N2Cl·C2O4H2·1/2H2O (C, H, N).
4.13. 1-(4-fluorobenzyl)-5-methyl-3-(piperidin-1-ylmethyl)-1H-indole (12)
Following the general procedure 4.9, reaction of 6 (177 mg, 0.74 mmol) [piperidine
(63 mg, 0.74 mmol), formaldehyde (30%aq, 55 µL, 0.74 mmol)] yielded 12, which
was purified by basic alumina flash chromatography with ethyl acetate/hexane
mixtures as eluent, obtaining a yellow oil (150 mg, 60%). 1H NMR (acetone-d6, 300
MHz) δ 7.51 (bs, 1H, H4), 7.22–7.18 (m, 4H, H2’, H7, H2), 7.07–7.01 (m, 2H, H3’),
6.93 (bd, 1H, J6-7 = 8.4 Hz, H6), 5.33 (s, 2H, N1CH2C1’), 3.58 (s, 2H, C3CH2), 2.38
(m, 7H, H2’’, CH3), 1.51 (m, 4H, H3’’), 1.41 (m, 2H, H4’’). Its oxalate salt was
prepared following the procedure described in 4.9. Compound 12 (119 mg, 0.34
mmol) [oxalic acid 1M (30 mg, 0.34 mmol)] yielded its oxalate salt as a white solid
(107 mg, 72%). Mp 107–109 ºC; 1H NMR (DMSO-d6, 300 MHz) δ 7.63 (s, 1H, H4),
7.56 (s, 1H, H2), 7.38 (d, 1H, J6-7 = 8.4 Hz, H7), 7.28–7.23 (dd, 2H, J2’-F = 5.6 Hz,
J2’-3’ = 8.6 Hz, H2’), 7.14 (m, 2H, H3’), 7.00 (bd, 1H, J6-7 = 8.4 Hz, H6), 5.40 (s, 2H,
N1CH2C1’), 4.35 (s, 2H, C3CH2), 3.10 (bs, 4H, H2’’), 2.39 (s, 3H, CH3), 1.71 (m,
4H, H3’’), 1.46 (m, 2H, H4’’); 13C NMR (DMSO-d6, 75.4 MHz) δ 164.4, 134.2, 134.0,
29