Carboxamides vs. methanimines: Crystal structures, binding interactions, photophysical studies, and biological evaluation of (indazole-5-yl)methanimines as monoamine oxidase B and acetylcholinesterase inhibitors

Article abstract of DOI:10.1016/j.ejmech.2019.06.041

A comprehensive study was performed for the first time to compare two structurally related substance classes, namely indazole-5-carboxamides (11–16) and (indazole-5-yl)methanimines (17–22). Both chemical entities are potent, selective and reversible MAO-B inhibitors and, therefore, may serve as promising lead structures for the development of drug candidates against Parkinson's disease (PD) and other neurological disorders. Compounds 15 (K_i = 170 pM, SI = 25907) and 17 (K_i = 270 pM, SI = 16340) were the most potent and selective MAO-B inhibitors in both series. To investigate the multi-target inhibitory activity, all compounds were further screened for their potency against human AChE and BuChE enzymes. Compound 15 was found to be the most potent and selective AChE inhibitor in all series (hAChE IC₅₀ = 78.3 ± 1.7 μM). Moreover, compounds 11 and 17 showed no risk of drug-induced hepatotoxicity and a wider safety window, as determined in preliminary cytotoxicity screening. Molecular modeling studies into the human MAO-B enzyme-binding site supported by a HYDE analysis suggested that the imine linker similarly contributes to the total binding energy in methanimines 17–22 as the amide spacer in their carboxamide analogs 11–16. Amplified photophysical evaluation of compounds 17 and 20, including single X-ray analysis, photochemical experiments, and quantum-chemical calculations, provided insights into their more favourable isomeric forms and structural features, which contribute to their biologically active form and promising drug-like properties.

Full text of DOI:10.1016/j.ejmech.2019.06.041

Accepted Manuscript
Carboxamides vs. methanimines: Crystal structures, binding interactions,
photophysical studies, and biological evaluation of (indazole-5-yl)methanimines as
monoamine oxidase B and acetylcholinesterase inhibitors
Nikolay T. Tzvetkov, Hans-Georg Stammler, Maya G. Georgieva, Daniela Russo,
Immacolata Faraone, Aneliya A. Balacheva, Silvia Hristova, Atanas G. Atanasov,
Luigi Milella, Liudmil Antonov, Marcus Gastreich
PII:
S0223-5234(19)30568-9
DOI:
https://doi.org/10.1016/j.ejmech.2019.06.041
EJMECH 11444
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 1 June 2019
Revised Date: 13 June 2019
Accepted Date: 14 June 2019
Please cite this article as: N.T. Tzvetkov, H.-G. Stammler, M.G. Georgieva, D. Russo, I. Faraone,
A.A. Balacheva, S. Hristova, A.G. Atanasov, L. Milella, L. Antonov, M. Gastreich, Carboxamides vs.
methanimines: Crystal structures, binding interactions, photophysical studies, and biological evaluation
of (indazole-5-yl)methanimines as monoamine oxidase B and acetylcholinesterase inhibitors, European
Journal of Medicinal Chemistry (2019), doi: https://doi.org/10.1016/j.ejmech.2019.06.041.
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
Carboxamides vs. methanimines: Crystal structures, binding interactions,
photophysical studies, and biological evaluation of (indazole-5-yl)methanimines as
monoamine oxidase B and acetylcholinesterase inhibitors
a,
b
c
d,e
Nikolay T. Tzvetkov ∗ , Hans-Georg Stammler , Maya G. Georgieva , Daniela Russo , Immacolata
d,e
c
f
g,h,i
d,e
Faraone , Aneliya A. Balacheva , Silvia Hristova , Atanas G. Atanasov , Luigi Milella , Liudmil
f
j
Antonov , Marcus Gastreich
a
b
c
NTZ Lab Ltd., Krasno selo 198, Sofia 1618, Bulgaria
Department of Chemistry, Bielefeld University, Universitätsstr. 25, 33615 Bielefeld, Germany
Bulgarian Academy of Sciences, Institute of Molecular Biology “Roumen Tsanev”, Department of
Biochemical Pharmacology and Drug Design, Acad. G. Bonchev Str., Bl. 21, 1113 Sofia, Bulgaria
d
Department of Science, Università degli Studi della Basilicata, V.le Ateneo Lucano 10, 85100 Potenza,
Italy
e
Spinoff BioActiPlant S.R.L. Department of Science, University of Basilicata, V.le dell’Ateneo lucano,
1
0, 85100 Potenza, Italy
f
Bulgarian Academy of Sciences, Institute of Organic Chemistry with Centre of Phytochemistry, Acad.
G. Bonchev Str., Bl. 9, Sofia 1113, Bulgaria
g
Institute of Genetics and Animal Breeding of the Polish Academy of Sciences, 05-552 Jastrzebiec,
Poland
h
Department of Pharmacognosy, University of Vienna, Althanstrasse 14, A-1090 Vienna, Austria
I
Institute of Neurobiology, Bulgarian Academy of Sciences, 23 Acad. G. Bonchev str., 1113 Sofia,
Bulgaria
j
BioSolveIT GmbH, An der Ziegelei 79, 53757 St. Augustin, Germany
KEYWORDS:
MAO-B inhibitors; Molecular modelling; (Indazole-5-yl)methanimines; Parkinson’s disease; X-ray
∗
Corresponding author.
E-mail address: ntzvetkov@ntzlab.com (N.T. Tzvetkov). Phone: +49-179-528-4358
ORCID iD: 0000-0002-8482-0481 (Dr. Nikolay T. Tzvetkov)
1

ACCEPTED MANUSCRIPT
Abstract
A comprehensive study was performed for the first time to compare two structurally related substance
classes, namely indazole-5-carboxamides (11–16) and (indazole-5-yl)methanimines (17–22). Both
chemical entities are potent, selective and reversible MAO-B inhibitors and, therefore, may serve as
promising lead structures for the development of drug candidates against Parkinson’s disease (PD) and
other neurological disorders. Compounds 15 (K
i i
= 170 pM, SI = 25907) and 17 (K = 270 pM, SI =
1
6340) were the most potent and selective MAO-B inhibitors in both series. To investigate the multi-
target inhibitory activity, all compounds were further screened for their potency against human AChE
and BuChE enzymes. Compound 15 was found to be the most potent and selective AChE inhibitor in all
series (hAChE IC = 78.3 ± 1.7 µM). Moreover, compounds 11 and 17 showed no risk of drug-induced
5
0
hepatotoxicity and a wider safety window, as determined in preliminary cytotoxicity screening.
Molecular modeling studies into the human MAO-B enzyme-binding site supported by a HYDE
analysis suggested that the imine linker similarly contributes to the total binding energy in
methanimines 17–22 as the amide spacer in their carboxamide analogs 11–16. Amplified photophysical
evaluation of compounds 17 and 20, including single X-ray analysis, photochemical experiments, and
quantum-chemical calculations, provided insights into their more favourable isomeric forms and
structural features, which contribute to their biologically active form and promising drug-like properties.
2

ACCEPTED MANUSCRIPT
1
. Introduction
Neurodegenerative diseases (NDs) are characterized by slow progressive death of neurons in the
brain, leading to a loss of structure and function. Among NDs, Alzheimer’s (AD) and Parkinson’s
disease (PD) are the two most prevalent age-related disorders of the central nervous system (CNS). As a
consequence of the aging population, both NDs represent a substantial socioeconomic burden on
society, currently affecting 7–8% of the population over age 65 [1–4]. Although there are common
features in the cellular events that develop as a result of neurodegeneration in specific brain regions of
affected patients, AD and PD display differences in pathogenesis and symptoms [3–7]. While an
irreversible cognitive decline, such as severe memory, attention, and learning deficits associated with an
early-onset of AD, a progressive impairment of the core motor functions is often an early indicator of
PD [8–11]. AD, for example, is principally characterized by a degradation of cholinergic neurons and
synapses in the neocortex and hippocampus, resulting in a decrease of acetylcholine neurotransmitter
levels in these brain regions involved in higher cognitive functions [12]. The loss of dopaminergic
neurons in the substantia nigra pars compacta (SNc) of midbrain, associated with an abnormal α-
synuclein aggregation as well as formation of Lewy bodies and Lewy neurites, lead to the typical
symptoms of PD like resting tremor, rigidity, bradykinesia (slowness), and postural instability [9,10]. In
the advanced stages, PD is accompanied by various non-motor symptoms including depression,
behavioral and cognitive complications [12–14]. Thus, AD and PD are multifactorial disorders, in which
complex pathophysiological processes trigger the neuronal cell impairment and death [11,15].
Therefore, disease-specific mechanisms necessitate disease-tailored therapeutic strategies.
Since an excessive loss of dopamine (DA) neurons is considered as a neuropathological hallmark of
PD, the majority of the therapies introduced so far in the PD treatment have been focused on increasing
the DA levels in the brain [2,16,17]. In particular, the DA replacement therapy combining the prodrug
levodopa (L-DOPA) with DA agonists, DOPA-decarboxylase inhibitors (DDIs), monoamine oxidase B
(MAO-B), and/or catechol-O-methyltransferase (COMT) inhibitors is still commonly used for PD
3

ACCEPTED MANUSCRIPT
treatment [2,13,17–19]. However, the long-term administration of L-DOPA has been linked to major
adverse effects including dyskinesia, dose failure (drug resistance), freezing during movement, but also
other side effects such as DA dysregulation, hypotension, nausea, and others [17]. To date, there are
only few approved drugs for AD treatment; the most of them are acetylcholinesterase (AChE) inhibitors
[12,18,20]. Currently used medicines have an impact on several symptoms in different AD or PD stages
by affecting mainly the brain’s cholinergic or dopaminergic system, respectively, but do not change or
stop the disease progression [12,18–20]. Despite enormous research efforts, including multitarget
strategies rooted on the principle that a simultaneously action on two or more pharmacological targets
may be beneficial for the treatment of multifactorial diseases [12,18], the medical need for new AD and
PD modifying therapeutics still exists.
Monoamine oxidases (MAOs, EC 1.4.3.4) are flavoenzymes localized on the mitochondrial outer
membrane that catalyze the oxidative deamination of xenobiotic and endogenous monoamines.
Therefore, MAOs are important for the modulation the levels of monoamine neurotransmitters in the
central and peripheral nervous system (CNS and PNS). Two isoforms, namely MAO-A and MAO-B,
are present in most mammalian tissues. Both MAOs share ∼ 73% identity of protein sequence, but
display regional differences in enzyme activity, substrate specificity, and distribution in the brain and
periphery [21–24]. The expression levels and activity of MAO-B, but not of MAO-A isoform, in the
human brain increase ∼ 4-fold with aging [25]. The increased activity and overexpression of MAO-B
lead to an overproduction of reactive oxygen species (ROS) and, therefore, is associated with oxidative
stress and loss of neuronal function [25]. In fact, increased MAO-B levels and activity is observed in the
brain of PD and AD patients [26,27]. Thus, inhibition of the MAO-B isoenzyme by selective MAO-B
inhibitors (IMAO-B) is an established therapeutic approach for PD treatment [28]. For example, the
irreversible MAO-B inhibitors selegiline (Zelapar) and resagiline (Azilect), as well as the reversible
inhibitor safinamide (Xadago) are currently in clinical use alone or as add-on therapy to L-DOPA in
late-stage PD (for structures, see Fig. 1) [16,27–31].
4

ACCEPTED MANUSCRIPT
Fig. 1. Structures, brand names, and mechanism of action of approved drugs for PD and AD.
Acetylcholinesterase (AChE; EC 3.1.1.7) and butyrylcholinesterase (BuChE, EC 3.1.1.8) are
cholinesterase (ChE) enzymes that are responsible for the hydrolysis (deacetylation) of choline-based
esters, including specific neurotransmitters in both CNS and PNS [32,33]. Both ChE isoforms are ∼ 65%
identical at protein level, but differ in their substrate preference, enzyme activity, and distribution in the
brain and periphery [34]. Acetylcholine (ACh) is degraded more quickly by AChE, while BuChE
hydrolyses preferably butyrylcholine. In CNS, AChE is located mainly in neurons, accounting ∼ 90% of
ChE activity in the temporal cortex of the normal human brain, whereas BuChE is primary associated
with glial cells and accounts ∼ 10% of ChE activity [34]. Therefore, inhibition of AChE enzyme activity
is an established therapeutic approach for AD treatment. Galantamine, for example, is a reversible
AChE inhibitor of plant origin that is widely prescribed for the treatment of mild-to-moderate AD and
AD-related dementia [35]. Inhibitors of BuChE may also be useful in the AD therapy due to its
increased activity in advanced forms of the disease [36].
Considering their pharmacological potential and more preferable safety profile, we have been
particularly focused on developing selective, reversible IMAO-B with potential multitarget activity on
other CNS relevant biological targets [37–40]. Within the new chemical entities discovered by us so far,
the C5-substituted indazole-carboxamide and pyrrolo-pyridine derivatives were outlined as the most
favorable series for further investigation [39,40]. Consequently, we continued with exploration of
compounds comprising a privileged indazole-5-carboxamide scaffold, which has provided a beneficial
therapeutic potential, acting either as selective IMAO-B or dual MAO-A/B inhibitors [39] (Fig. 2).
5

ACCEPTED MANUSCRIPT
Indazole-5-carboxamide scaffold
(Indazole-5-yl)methanimine scaffold
N-unsubstituted
N1-alkylated
Introduce
methanimine
Indazole
Linkage
A
F/Cl
Cl/F
A
F/Cl
Cl/F
O
Linker
replacement
N
H
N
5
5
R
N 1
R
N 1
N
N
Phenyl
Pyridine
A = -N or -CH
R = alkyl, O-alkyl
3,4-di-substituted
Aromatic Ring
keep constant
Fig. 2. Generic structure of previously developed selective MAO-B inhibitors 11–16 [39] (left) and
intended chemical modifications towards (indazole-5-y)methanimine scaffold (compounds 17–22)
subjected in the present work (right).
Based on previously described drug design strategy [39], facile structural modification of the key
indazole-5-carboxamide scaffold (Fig. 2, left) was performed to replace the carboxamide spacer by an
imine linker, obtaining a series of (indazole-5-yl)methanimine derivatives (Fig. 2, right).
In the present work, we performed a comprehensive study to compare two structurally related
substance classes, namely indazole-5-carboxamides and (indazole-5-yl)methanimines. Both series of
compounds were tested at human MAO-A and B, and further screened for their inhibitory activity
against human AChE and BuChE enzymes. Furthermore, we evaluated kinetics, mechanism of MAO-B
inhibition, drug-like and physicochemical properties, as well as the cytotoxic effects of representative
IMAO-B. To understand the most important inhibitor–enzyme interactions within the binding pocket of
different pharmacological targets, docking experiments were conducted using the single X-ray
structures of selected compounds under study. Photophysical experiments under different conditions
supported by quantum-chemical calculations were performed to investigate the more stable isomeric
forms (e.g., E/Z-isomers and tautomers), as well as the photochemical stability of selected (indazole-5-
yl)methanimine derivatives. Finally, Pan assay interference properties (PAINS) of the studied
compounds were investigated and discussed.
6

ACCEPTED MANUSCRIPT
2
. Results and discussion
2
.1. Chemistry
Synthesis and chemical structures of compounds under study are shown in Scheme 1. The
preparation of indazole-5-carboxamide derivatives 11–16 was performed following optimized reaction
procedures to obtain the compounds in higher yields (>81% yield in average) [39]. Accordingly,
compounds 11–14 can be prepared via amide coupling reaction of 1H-indazole-5-carboxylic acid (1)
with the appropriate 3,4-disubstituted anilines 3–6 using EDC hydrochloride as a coupling reagent
(Scheme 1, left). The N1-alkylated indazole-5-carboxamide derivatives 15 and 16 can be synthesized
following a four-step regioselective approach [39].
Scheme 1. Synthetic routes to indazole-5-carboxamides 11–16 and (indazole-5-yl)methanimines 17–22.
Reagents and conditions: (i) EDC–HCl, methanol, r.t., 3–16 h; (ii) conc. acetic acid (10 mol-%),
ethanol, reflux, 1–24 h; (iii) 1) methanol, c. H
2 4
SO (10 mol-%), 60-70 °C, 1–3 h, 2) methyl
methanesulfonate (MMS, for 15) or 1-bromo-2-methoxyethane (for 16), K CO , DMF, reflux, 9–16 h,
2
3
3
2
) 2M NaOH, THF–H O (1:1), 30–35 °C, 1–3 h; (iv) MMS (for 9) or 1-bromo-2-methoxyethane (for
1
0), K CO , DMF, reflux, 68–72 h.
2
3
7

ACCEPTED MANUSCRIPT
For the preparation of methanimines 17–20, 1H-indazole-5-carbaldehyde (2) was reacted with the
respective anilines 3–6 in the presence of a catalytic amount of acetic acid in ethanol under reflux
(
Scheme 1, right). The N1-alkylated aldehydes 9 and 10 can be prepared via one-step synthetic
procedure (for 10, see Supporting Information). Subsequently, aldehydes 9 and 10 were condensed with
,4-dichloroaniline (3) to obtain the N1-alkylated methanimines 21 and 22, respectively. To investigate
3
the chemical stability of methanimines 17–22 (i.e., Schiff bases), extensive stability studies for selected
compounds were performed (see Section 2.5.1 and Supporting Information). All final products were
purified by column chromatography following by recrystallization and their structures confirmed by
NMR spectroscopy and mass spectrometry. The described procedures allow the introduction of further
structural modifications, to access small compound libraries for rapid evaluation of structure-activity
relationships (SARs). In addition, we obtained the single X-ray structures of 17 and 20, confirming not
only the preferable formation of the 1H-indazole tautomeric form, but also the trans-isomerism of
(indazole-5-yl)methanimine derivatives 17–22 (see Section 2.5.2 and Supporting Information).
2
.2. Biological evaluation
Biological evaluation of all compounds toward human (h) MAOs and human ChEs (hAChE and
hBuChE) was performed according to established protocols [39–42].
2
.2.1. Monoamine oxidase studies
The enzyme inhibition of the studied compounds against MAOs was performed using a
fluorescence-based assay with p-tyramine as a substrate and the commercial kit Amplex Red [43,44].
Determined in vitro MAOs inhibitory activities (IC values) and selectivity (expressed as selectivity
5
0
index, SI) toward hMAO-B for all compounds and reference inhibitors are shown in Table 1.
Determination of the kinetic parameters (K and Vmax) of the hMAO-B enzyme was performed at
m
different concentrations of p-tyramine and the K values of the tested compounds were obtained using
i
the Cheng-Prusoff equation (see also Experimental Section) [45].
8

ACCEPTED MANUSCRIPT
Table 1 Monoamine oxidase activity and selectivity of the tested compounds.
a
IC50 ± SEM (nM)
1
b
a,c
Compd.
R
R
hMAO-A
≥10000
hMAO-B
SI
i
K ± SEM (nM)
d
1
1
3,4-Cl
H
0.586 ± 0.087
17065
0.26 ± 0.04
(
NTZ-1006)
e
1
1
2
3
3-Cl, 4-F
3-F, 4-Cl
H
H
>10000
>10000
0.679 ± 0.044
14727
14970
0.30 ± 0.02
0.29 ± 0.02
d
0.668 ± 0.053
d
1
1
4
5
3,4-Cl
3,4-Cl
H
>10000
>10000
5.42 ± 0.20
>1845
25907
2.39 ± 0.18
0.17 ± 0.02
e
Me
0.386 ± 0.052
(
NTZ-1091)
e
e
1
1
1
1
2
2
2
6
7
8
9
0
1
2
3,4-Cl
3,4-Cl
3-Cl, 4-F
3-F, 4-Cl
3,4-Cl
3,4-Cl
3,4-Cl
–
EtOMe
2870 ± 218
>10000
1.08 ± 0.08
>2657
16340
4762
5236
>7812
9709
3157
259
0.48 ± 0.04
0.27 ± 0.03
0.93 ± 0.13
0.84 ± 0.12
0.57 ± 0.02
0.46 ± 0.08
0.57 ± 0.07
na
d
H
0.612 ± 0.065
2.10 ± 0.30
1.91 ± 0.27
1.28 ± 0.02
H
>10000
H
>10000
H
>10000
d
Me
>10000
1.03 ± 0.09
EtOMe
4072 ± 272
1.29 ± 0.08
f
f
Selegiline
Rasagiline
–
–
1424 ± 69.1
5.50 ± 0.26
f
f
–
680 ± 129
13.0 ± 0.90
>52
na
g
g
g
g
Safinamide
–
–
>25900
5.18 ± 0.04
5000
2.29 ± 0.02
a
b
c
n = 3, unless otherwise noted. Selectivity index: SI = IC (hMAO-A)/IC (hMAO-B). The experimentally measured IC50
5
0
50
hMAO-B values were converted to the respective inhibition constants (K
i
) using Cheng-Prusoff equation: K
i
m
= IC g5 0/(1 + [S]/K )
d
e
f
with [S] = 150 µM and K = 118.8 µM. Data are from ref. [37]. Data are from ref. [39]. Data are from ref. [46]. Data are from
m
ref. [40]. h = human, na = non-applicable.
9

ACCEPTED MANUSCRIPT
2
.2.1.1. Evaluation of inhibitory activities at human MAOs
In the present study we aimed to perform a comparative study within the series of indazole-5-
carboxamide (designated subclass I, compounds 11–16) and (indazole-5-yl)methanimine (subclass II,
7–22) derivatives with respect to: (i) inhibition of both MAO-A and MAO-B isoforms, (ii) inhibitory
1
activities against both hAChE and hBuChE enzymes, (iii) toxicological effects, and (iv) their
photophysical and quantum-chemical investigation of isomerism and photochemical stability.
For evaluation of SARs of the compounds in Table 1, rasagiline and safinamide were used as
reference for irreversible and reversible MAO-B inhibitors, respectively. The selective irreversible
inhibitors clorgyline and selegiline were also used as positive controls in the MAO-A and MAO-B
assays, respectively. In general, all tested compounds are selective and potent inhibitors of human
MAO-B with IC values ranging from low nanomolar to even picomolar potency (hMAO-B). With the
5
0
exception of the N1-methoxyethyl-substituted subclass I compound 16 and its subclass II analog 22,
neither of the remaining carboxamide derivatives (compounds 11–15) or their methanimine analogs
(17–21) exhibited noticeable inhibition of MAO-A at the highest tested concentration of 10 µM.
In this study, compounds 11–16 (subclass I) were included to better compare the effects of a spacer
modification on inhibition potency and selectivity at both MAO isoforms with those of 17–22 (subclass
II). Therefore, in order to investigate the role of such small modification within both series of
compounds (e.g., carboxamides vs. methanimines), we retained the 3,4-dihalo-substitution at the phenyl
ring and the C5-substituted indazole moiety unit during further exploration. Replacement of the
carboxamide spacer in subclass I compounds 11–16 by an amine function resulted in potent and
selective IMAO-B (subclass II compounds 17–22). Nevertheless, methanimines 17–22 displayed almost
equally high affinity toward MAO-B compared to their carboxamide analogs 11–16.
The 3,4-dichlorophenyl indazole-5-carboxamide 11 was the most potent and selective IMAO-B
within the indazole N-unsubstituted subclass I compounds (IC = 0.586 nM; SI = 17065), while its
5
0
methanimine analog 17 represents the most active and selective IMAO-B within the subclass II series
1
0

ACCEPTED MANUSCRIPT
(
IC = 0.612 nM; SI = 16340). Compounds 11 and 17 exhibited almost equally improved hMAO-B
5
0
inhibitory activity when compared to the standard IMAO-B, being ∼ 9-fold more potent than selegiline
and safinamide. Further structural modifications of 11 and 17 were performed by replacement of the 4-
Cl or the 3-Cl atoms at the phenyl ring with one fluorine atom (compounds of subclass I 12, 13 and
subclass II 18, 19). The respective 3-chloro-4-fluoro- or 4-chloro-3-fluoro-substituted carboxamides
(
compounds 12 and 13) were almost equipotent and selective inhibitors of hMAO-B (12, IC = 0.68
50
nM, SI = 14727; 13, IC50 = 0.67 nM, SI = 14,970), comparable to the inhibitory activity of 11. In
contrast, the presence of 3-chloro-4-fluoro (compound 18, IC = 2.19 nM, SI = 4672) and 4-chloro-3-
5
0
fluoro substituents (19, IC = 1.91 nM, SI = 5236) at the phenyl ring of subclass II compounds led to a
50
decrease in inhibitory activity and selectivity against hMAO-B, when compared to the potency and
selectivity of the parent methanimine 17 and the respective carboxamide analogs 12 and 13 (subclass I).
It is worth mentioning that the respective 3-fluoro-4-chlorophenyl-substituted derivatives in both series
(e.g., carboxamide 13 and methanimine 19) were slightly more active at hMAO-B enzyme than their 4-
chloro-3-fluorophenyl-substituted analogs 12 and 18, respectively.
Next, the phenyl ring of 11 and 17 was replaced by a bioisosteric 5,6-dichloropyridine residue in
order to enhance water solubility of both lipophilic series of IMAO-B. The resulting carboxamide and
methanimine derivatives 14 (subclass I) and 20 (subclass II) are less active than the respective 3,4-
dichloro-phenyl analogs 11 and 17, but they are still potent IMAO-B (14, IC50 = 5.42 nM; 20, IC50
=
1
.28 nM). Interestingly, methanimine 19 is the only compound within subclass II IMAO-B exhibiting
higher inhibitory activity against hMAO-B than its carboxaminde analog (i.e., subclass I compound 14).
In comparison to 14, compound 20 is ~4.2- and ~4.9-fold more potent against human and rat MAO-B,
respectively. Compared to selegiline and safinamide, methanimine 20 displays a ~4.3- and ~4.6-fold
increase in inhibitory potency towards hMAO-B, respectively, while carboxamide 14 is similarly potent
as both standard IMAO-B.
1
1

ACCEPTED MANUSCRIPT
Introduction of a methyl group at the indazole N1 position in 11 and 17 resulted in compounds 15
and 21. The N1-methylated caroxamide 15 was found to be the best IMAO-B of all compounds under
study (human IC = 0.386 nM, rat IC = 1.32 nM, SI = 25907), being almost 2- and >13-fold more
5
0
50
potent than its precursor 11 and safinamide, respectively. The respective N1-methylated methanimine
compound 21) provided to be weaker IMAO-B than the precursor 17 and the subclass I analog 15.
However, compound 21 (IC50 = 1.03 nM) was 5-fold more potent against hMAO-B than safinamide
IC50 = 5.18 nM). The elongation of the N1-alkyl substitution by introducing of a larger methoxyethyl
(
(
group in compounds of subclass I (carboxamide 16) and subclass II (methanimine 22) generally
provided potent and selective IMAO-B. Both compounds are similarly potent against MAO-B (16, IC50
=
1.08 nM; 22, IC50 = 1.29 nM), displaying a 4- and 5-fold increase in inhibitory activity toward
hMAO-B than safinamide, respectively. As mentioned above, the N1-methoxyethyl-substituted
derivatives 16 and 22 are the only representatives in these series showing inhibitory activity against
hMAO-A enzyme (16, IC = 2870 nM; 22, IC = 4072 nM).
50
50
In addition, the respective IC50 values for all compounds under study were compared with the
corresponding inhibitory constants K , which were obtained and calculated from the hMAO-B enzyme
kinetic experiments (Table 1). A good agreement between the IC and K values (low nM range) for all
i
5
0
i
compounds could be observed. The estimated K values reveal of a competitive mode of inhibition (for
i
competitive inhibitors: K ≤ IC /2, if S ∼ K ) [45,47].
i
50
m
2
.2.1.2. Evaluation of mechanism of monoamine oxidase B inhibition
Due to the preferable safety profile compared to irreversible MAO inhibition, we were particularly
interested in development of reversible IMAO-B [39]. Therefore, time-dependent studies with the most
potent N1-methyl-substituted compound 15 (indazole-5-carboxamide, subclass I) and its (indazole-5-
yl)methanimine analog 21 (subclass II) were performed to investigate whether both representative
compounds are reversible or irreversible MAO-B inhibitors (Fig. 3). The mode of interaction with the
active site of hMAO-B by 15 and 21 and standard MAO-B inhibitors selegiline (irreversible, Irr) and
1
2

ACCEPTED MANUSCRIPT
safinamide (reversible, Rev) was measured after the first 15 min in the presence of low concentration of
p-tyramine (10 µM) and over 300 min after increasing the substrate concentration. A continuous
enhancement on enzymatic residual activity could be detected for compounds 15, 21 and safinamide
after increasing the concentration of p-tyramine at 1.0 mM, while no significant elevation in
fluorescence was measured across the time for selegiline. The experiments are in agreement with the
observed correlation between the IC50 and K
i i
values of 15 and 21 at hMAO-B (IC50 = 0.39 nM and K =
0
.17 nM for 15; IC50 = 1.03 nM and K = 0.46 nM for 21) revealing a reversible MAO-B inhibition.
i
Fig. 3. Time-dependent inhibition (reactivation) of hMAO-B enzyme by the standard irreversible
IMAO-B selegiline (Irr, 30 nM), reversible IMAO-B (Rev, 50 nM), and compounds 15 and 21 (both at
1
.0 nM). The substrate concentration was increased from 10 µM to 1.0 mM during a period of 300 min.
The remaining hMAO-B enzyme activity is expressed as % of control sample used in the experiment.
The values are the mean ± SD (n = 4).
In order to investigate the type of hMAO-B inhibition of 15 and 21, Michaelis-Menten kinetics
experiments were performed (Fig. 4). Accordingly, the initial rates of the deamination reaction of p-
tyramine (at six different concentrations) catalyzed by hMAO-B in the absence (no inhibitor) and in the
presence of different concentrations of 15 and 21 (with the respective inhibitor at 0.5, 1.0 and 5.0 nM)
were measured. From the obtained Lineweaver-Burk plots, representing the reciprocal hMAO-B
enzyme activity vs. the reciprocal p-tyramibe concentration, we observed that both compounds acted as
competitive IMAO-B. The Lineweaver-Burk plots are linear and intersected at the y-axis together with
1
3

ACCEPTED MANUSCRIPT
the plot for the uninhibited hMAO-B (without inhibitor), which is in agreement with our previously
findings [48]. The hMAO-B enzyme binding affinities (inhibition constants K ) obtained form the
i
respective Dixon plots (K = 0.22 nM for 15; K = 0.63 nM for 21) correlate very well with the
i
i
determined IC and the calculated K values for 15 (IC = 0.39 nM, K = 0.17 nM) and 21 (IC = 1.03
5
0
i
50
i
50
nM, K = 0.46 nM), respectively (cf. Table 1). The results obtained from the reactivation and the kinetic
i
experiments of 15 and 21 indicate that indazole-5-carboxamides (compounds 11–16) and (indazole-5-
yl)methanimine derivatives (compounds 17–22) are reversible and competitive MAO-B inhibitors.
Fig. 4. Kinetic studies on the mechanism of hMAO-B inhibition of compounds 15 (A) and 21 (B). The
mode of hMAO-B inhibition was evaluated from the respective double reciprocal Lineweaver-Burk
plots of 1/rate (1/V) vs. 1/p-tyramine substrate concentration in the presence of different concentrations
of the inhibitors (0.5, 1.0 and 5.0 nM).
2
.2.2. Evaluation of cholinesterase inhibitory activity
In order to investigate the activity on other CNS relevant biological targets, all compounds also
were screened for inhibition of hAChE and hBuChE in a concentration range of 10 to 100 µM (Fig. S3
and S4). With exception of the N1-methylated indazole-5-caroxamide 15 at hAChE (IC50 = 78.3 ± 1.7
µM), none of the remaining compounds displayed inhibitory activity below 100 µM range toward both
ChE isoforms (Table 2). Similarly to 15, its N1-methylated (indazole-5-yl)methanimine analog
(
compound 21) was found to be the most active derivative in the series of subclass II compounds (IC =
50
1
18.8 ± 1.1 µM). However, an accurate SAR evaluation of the cholinesterase assays data suggest that
1
4

ACCEPTED MANUSCRIPT
indazole N1 position could be essential for achievement of multi-target active compounds in both series.
Thus, further experimental studies are required in order to find the optimal substitution pattern of the
indazole moiety. The experimental data are in agreement with the results obtained from in silico
evaluation of binding affinities toward both hChEs (see Fig. S7).
Table 2 Cholinesterase activity of the tested compounds.
a
Inhibition % at 100 µM (or IC , µM )
50
Compd.
hAChE
1.6 ± 0.6
hBuChE
11.3 ± 0.5
2
1
1
1
1
1
1
1
1
1
2
2
2
1
2
3
4
5
6
7
8
9
0
1
2
b
b
(
>100, UTC )
(>100, UTC )
2
6.4 ± 2.1
7.21 ± 1.39 b
b
(
>100, UTC )
4.2 ± 1.2
>100, UTC )
(>100, UTC )
1
2.23 ± 0.28 b
b
(
(>100, UTC )
6
.87 ± 1.13 b
9.23 ± 0.21 b
(
>100, UTC )
(>100, UTC )
1
7.9 ± 1.7
11.1 ± 1.4
(>100, UTC )
c
b
(
78.3 ± 1.1)
8
.66 ± 1.98 b
9.58 ± 1.01 b
(
>100, UTC )
(>100, UTC )
1
1.1 ± 1.6
9.01 ± 0.57 b
b
(
>100, UTC )
(>100, UTC )
2
4.6 ± 2.7
4.56 ± 0.52 b
b
(
>100, UTC )
(>100, UTC )
9
.69 ± 0.64 b
6.97 ± 0.78 b
(
>100, UTC )
(>100, UTC )
6
.92 ± 0.76 b
9.64 ± 1.06 b
(
>100, UTC )
(>100, UTC )
1
1.9 ± 0.7
8.57 ± 2.14 b
b
(
>100, UTC )
(>100, UTC )
6
.91 ± 0.74 b
2.27 ± 0.40 b
(
>100, UTC )
(>100, UTC )
7
1.2 ± 2.3
38.3 ± 4.1
(16.5 ± 1.2)
(7.30 ± 0.83)
c
c
Galantamine
(3.03 ± 0.39)
d
d
(
0.80 ± 0.06)
a
b
Data are means ± SD of three independent experiments. Compounds did not reach 50% inhibition at 100 µM, i.e.,
calculation of an IC50 value within the curvature was not possible (>100 µM, UTC = unable to calculate). IC50 values
c
were determined by extrapolating the curvature of log (inhibitor) vs. normalized response with a variable slope in
d
GraphPad Prism 6.0 (GraphPad Softaware) resulting in the respective µM IC50 values. Data are from ref. [49].
1
5

ACCEPTED MANUSCRIPT
2
.3. Evaluation of cytotoxicity
The preliminary cytotoxicity of the parent N-unsubstituted indazole-carboxamide 11 (subclass I)
and methanimine 17 (subclass II) compounds were evaluated by the determination of the cellular
viability in human hepatocarcinoma cells (HepG2 cell line). Cellular viability was estimated after 72
hours incubation period in a concentration range of 0.1 to 50 µM using the methylthiazolyldiphenyl-
tetrazolium bromide (MTT) reduction assay (Fig. 5). In the MTT assay, the yellow tetrazolium salt
(MTT) is reduced in metabolically active cells to an insoluble purple formazan, which is then
spectrophotometrically quantified to provide a direct measure of normal mitochondrial function or
dysfunction (in case of cytotoxic effects of tested compounds) [50]. The HepG2 cell line is often used in
the preclinical safety assessment to predict a drug’s potential to cause hepatotoxicity [51].
In general, the hepatotoxicity profile of compounds 11 (Fig. 6A) and 17 (Fig 6B) followed the same
trend in hepatocarcinoma cells, when compared to the control groups at all tested concentrations. For
both compounds, no pronounced decrease on cellular viability was detected at the highest tested
concentration of 50 µM (11, HepG2: 66.3 ± 6.82%; 17, HepG2: 80.6 ± 8.13%).
Since no major effects were observed in the MTT reduction assay after 72 h incubation period, it
could be concluded that the respective indazole-5-carboxamide (subclass I 11) and (indazole-5-
yl)methanimine (subclass II 17) derivatives did not show a pronounced mitochondrial dysfunction on
HepG2 cells even at the highest tested concentrations of 25 and 50 µM, suggesting no risk of drug-
induced hepatotoxicity and a wide safety window. However, for further preclinical studies, additional
examination of toxicology and genotoxicity of selected drug candidates need to be performed.
1
6

ACCEPTED MANUSCRIPT
Fig. 5. Cytotoxicity profile of compounds 11 (A) and 17 (B) measured on human hepatocarcinoma
HepG2 cells after a 72 h exposure to different concentrations of compounds (0.1 to 50 µM). Untreated
cells were used as positive control and the results are expressed as the mean % of untreated controls ±
SD (n = 4). **** p < 0.0001 vs. control.
2
2
.4. Molecular modeling studies
.4.1. Single X-ray structures of 17 and 20
In order to investigate the binding mode and interactions of the indazole-5-carboxamide derivatives
1–16 and their (indazole-5-yl)methanimine analogs 17–22 within the hMAO-B enzyme active site, the
1
single crystal X-ray structures of the most active subclass II compound 17 and its pyridine-substituted
analog 20 were obtained and then used as crystallographic templates for docking studies (see Section
2
.3.2 and Supporting Information).
The respective single X-ray structures of compounds 17 and 20 with atom numbering are shown in
Fig. 6. Similarly to the crystal structures of subclass I compounds 11 and 15 (cf. Fig. S5) [39], the X-ray
structural analyses of 17 and 20 confirm their almost planar conformation. Relevant experimental details
for the X-ray analysis of both compounds are given in Table S3 and S4 (see Supporting Information).
1
7

ACCEPTED MANUSCRIPT
Fig. 6. Single X-ray structures of the major occupied part of compounds 17 (A) and 20 (B) showing the
atom numbering. The thermal ellipsoids are drawn at the 50% probability level.
For compound 17, the X-ray structure determination reveals serious disorder, however, we decided
to consider the major occupied part (85%) in our docking experiments. Single crystals of 20 were
twinned with a ratio of 62:38, both domains ware taken into account for data integration and refinement.
In both crystal structures, the two fused five- and six-membered rings of the indazole unit are nearly
coplanar to the benzene respectively pyridine ring, defined the mean plane angle amounts to 2.7(2)° in
1
7 and 7.5(2)° in 20. In comparison, the indazole plane and the phenyl ring plane are titled by 3.67(3)°
and 6.01(5)° in 11 and 15, respectively (Fig. S5). The aromatic and indazole units are connected by a
methanimine linkage with torsions (C5-C10-N11-C12) of –177.9(5)° (in 17) and –177.2(6)° (in 20). The
angles are 123.4(5)° (C5-C10-N11) and 118.1(5)° (C10-N11-C12) for 17, and for compound 20
respectively 122.8(6)° (C5-C10-N11) and 121.0(5)° (C10-N11-C12). The plane defined by the
methanimine linker (C5, C10, N1, C12) encloses an angle of 15.6(5)° with the indazole moiety and an
1
8

ACCEPTED MANUSCRIPT
angle of 18.3(5)° with the phenyl ring for 17. For 20, the plane defined by the methanimine linker
encloses an angle of 8.3(4)° with the indazole moiety and an angle of 10.6(6)° with the pyridine unit.
Because of the lack of a carboxamide function able to build N-H…O hydrogen bridges compared to
1
1 and 15, only anti-parallel double chains (hydrogen bridging) can be found between the N1-H1 and
N2 atoms generated by a two-fold screw axis with N-H…N distances of 2.18(1) Å for 17 and 20,
respectively (Fig. S6). Due to the carboxamide function of subclass I compounds, intermolecular
hydrogen bonds for 11 and 15 can be found between H1 and O1 formed by translation along the b axis
forming chains with H...O distances of 2.07(2) Å in 11 and 2.17(3) Å in 15 [39]. The X-ray data for 17
and 20 point toward the importance of the indazole moiety within the observed conformation to build
geometrically sound intermolecular interactions, which are relevant for MAO inhibition.
Moreover, the X-ray analysis of 17 and 20 indicates that the high MAO-B inhibitory activity of N-
unsubstituted (indazole-5-yl)methanimines (subclass II) is due to their stable 1H-indazole tautomeric
forms within an energetically more favourable trans-isomerism (E-isomers) (see also Section 2.5 and
Supporting Information). Similar results were obtained for the tautomerism of N-unsubstituted indazole-
5
-carboxamide derivatives [39,52].
2
.4.2. Docking of compounds 14, 15, 20 and 21 into hMAO-B
Docking experiments were performed with the human model of the MAO-B enzyme. SAR
evaluation of compounds reported herein, including the single X-ray analysis of carboxamides 11 and
5 [39] as well as their methanimine analogs 17 and 20, provided valuable information about the
1
putative binding modes of both subclass I and II compounds within the active site of hMAO-B. To find
a plausible explanation for their high hMAO-B inhibitory potency, computational experiments of
carboxamide derivatives with the lowest (subclass I compound 14) and the highest hMAO-B affinity
(compound 15) were performed utilizing the crystal structure of hMAO-B (PDB code: 2V5Z) [53];
these were compared with the best docking poses for the respective methanimine analogs, e.g., subclass
II compounds 20 and 21.
1
9

ACCEPTED MANUSCRIPT
Following our previously established single X-ray/molecular modeling approach [39,54], we
investigated binding situations and estimated the most significant interactions and desolvation effects
within hMAO-B. At first, binding proposals were computed with LeadIT [55]. Then, the obtained poses
were used for further optimization, as well as semi-quantification of binding and visualization of the
best-computed solutions using the free energy approximation "HYDE", as embedded in SeeSAR (see
Experimental Section and Supporting Information) [56]. Based on our previous experiences [39], the
single X-ray structure geometries (i.e., bond lengths and angles) of carboxamides 11 and 15 as well as
methanimines 17 and 20 were used for docking studies. The results obtained from the molecular
modeling studies provided insight and rationalization the structural features of both subclass I and II
compounds in relation to: (i) the observed preference in inhibitory potency at hMAO-B of 3,4-
dichlorophenyl- versus 5,6-dichloropyridine substituents (15 and 21 vs. 14 and 20), on the one hand
side, and carboxamides (subclass I compounds) versus methanimines (subclass II) on the other, (ii)
investigation of the hydrophobic effects including enthalpic and entropic binding, (iii) the proposed
binding modes of compounds under study within the substrate cavity region of hMAO-B, and (iv)
further exploration of subclass II compounds as CNS drugs.
It must be noted that given the overall property arrangement in the binding pocket of hMAO-B
[hydrophobic "substrate cavity" (close to FAD)]–[hydrophilic "linker region" (on the front of the amide
or methanimine linker)]–[hydrophobic "entrance cavity" (around PRO102 and 104 and PHE103)] and
the close-to-symmetric complementarity of the ligands poses emerged with either the di-halogenated
phenyl- or pyridine-ring pointing towards the FAD cofactor (Fig. 7).
2
0

ACCEPTED MANUSCRIPT
Fig. 7. Overlay representing the overall binding modes of carboxamides 11–16 (A, subclass I
compounds) and methanimines (B, subclass II compounds) 17–22 onto the crystal structure of hMAO-B
(PDB: 2V5Z) with the unoccupied space in the binding site (grey "fog"). The most important protein
residues, water molecules (1180 and 1247) and FAD co-factor are displayed. The same color schema is
used for each ligand in its analog in both subclasses. The respective parent compounds 11 and 17 are
given in magenta.
The re-assessment of the X-ray structure geometries of 11 and 15 with SeeSAR resulted in a similar
conformation of all subclass I compounds 11–16 within hMAO-B. In all carboxamides, the amide linker
is rotated by 180° around the C12–N11–C10-C5 torsion angle (cp. Fig. S5). The overlay of the proposed
binding modes of 11–16 clearly shows that there are no differences in their orientation within the crystal
structure of hMAO-B (cp. Fig. 7A). In contrast, the X-ray analysis of the geometries of 17 and 20 led to
two favorable conformations of methanimines 17–22 – structures that resulted from a rotation around
the C12–N11–C10–C5 torsion angle (i.e., the imine linker) of about 180° (cf. Fig. 6). However, after
docking, the binding modes of methanimines 17–22 show similar orientations within the active site of
hMAO-B as their carboxamide analogs (Fig 7B).
In general, the modeling proposes the formation of a strong intermolecular H-bond between the
indazole N2 and water 1180 in all compounds. This H-bond was almost the same in distance in
carboxamides 14 and 15 (N2---HOH1180 ≈ 1.9 Å); in 20 and 21 it was approximately 1.8 Å and 1.9 Å,
respectively. The carboxamide linker in 14 and 15 is suggested to play an essential role as a hydrophilic
anchor (i.e., H-bond acceptor/donor) for the conformation of the compounds within the substrate cavity
2
1

ACCEPTED MANUSCRIPT
region of hMAO-B [39], interacting favorably with water molecule 1247 (for 14 and 15: CO---
HOH1247 ≈ 1.8 Å) (see Fig. 8A). Because of the inherently missing functionalities of the methanimine
linker, such a second H-bond cannot exist in subclass II compounds 20 and 21 (cf. Fig. 8B).
Furthermore, the binding modes and estimated affinities strongly suggest that the N-unsubstituted
derivatives 14 and 20, as well as the N1-methylated derivatives 15 and 21 occupy the same substrate
cavity space, provided that the ligands investigated herein do not covalently bind with the FAD
cofactor. Compounds 15 and 21 differ from 14 and 20 by the phenyl residue (instead of di-chloro-
substituted pyridine) and the methyl substitution at N1. However, their binding poses show the same
orientation of the indazole moiety within the "substrate cavity" region close to FAD, i.e., around the
TYR398, 435 and 188 and HOH1180. The 5,6-dichloropyridine or 3,4-dihalo-substituted phenyl ring of
the ligands occupies a strongly hydrophobic binding pocket of hMAO-B dominated by hydrophobic
amino acids such as LEU164 and 167, PRO102 and 104, and isoleucin residues (cf. Fig. 7A). In
addition, the computed binding models suggest that for both subclass I and II compounds the relevant
N2---HOH1180 bond is highly coordinated also by protein residues CYS172 and TYR188, whereas for
carboxamide derivatives 14 and 15 the water molecule HOH1247 is surrounded by GLN206, ILE199,
TYR201 and 326. We conclude that the observed H-bonds are essential for the ligand stabilisation of
carboxamides (CO---HOH1247), and that the conformations of all ligands (N2---HOH1180) in the
binding site of hMAO-B are realized through non-covalent interactions.
Given the previously discussed a priori possibility of a flipped binding mode that results in similar
total free energy estimations [39], we nonetheless favor the displayed poses in which the 5,6-
dichloropyridine- (in 14 and 20) or 3,4-dichlorophenyl-ring (in 15 and 21) point toward FAD for three
reasons: (i) in the safinamide co-crystal structure (2V5Z), the fluorine atom occupies the same area as
the halogens in the herein discussed compounds, and (ii) the HOH1180 and 1247 were shown before to
play a pivotal role on the front of FAD and, (iii) the central moiety may act as an amide linker in this
orientation – also contributing to a good agreement with the estimated binding affinities.
2
2

ACCEPTED MANUSCRIPT
2
.4.3. HYDE visual free energy assessment of compounds 14, 15, 20 and 21
In accordance with previous studies on related compounds [39], we also performed HYDE analyses
of the new compounds in order to gain insight into the overall binding thermodynamics, aiming at
explaining the observed differences in their inhibitory potency in hMAO-B (e.g., 14 and 20 vs. 15 and
2
1). The HYDE scoring function as embedded in SeeSAR considers the free energy by computing the
difference between the unbound and bond states. H-bonds (approximate enthalpy) and dehydration
"desolvation", approximate entropy) effects of all non-hydrogen/heavy atoms (HA), contributing to the
(
overall Gibbs free energy (∆G) are computed with respect to their accessibility to water before and after
binding [57,58]. Both the protein (Rec) and the ligand (Lig) energy terms are considered for calculation
of the respective partial contribution to the overall ∆G of each non-hydrogen atom (Fig. 8 and Table
S5).
In this HYDE study, we were particularly interested on the estimation and visualization of the major
ligand-receptor interactions and desolvation effects at (i) the aromatic part of the molecules (3,4-
dichlorophenyl- vs. 5,6-dichloropyridine-substitution) and (ii) the respective linker (e.g., carboxamide
vs. methanimine function). Since the indazole-binding site of the pocket is spatially limited by the FAD
cofactor, we analyzed the contribution of the N1-methyl substituent in 15 and 21 to the overall ∆G
compared to the N-unsubstituted derivatives 14 and 20, respectively (cp. Fig. 8C and 8D). The
desolvation effects for both chlorine atoms at pyridine C5 and C6 in 5,6-dichloropyridine-substituted
compounds 14 and 20 (desolvation of about –15 kJ/mol) is compensated by a desolvation "penalty" at
pyridine nitrogen for both ligands (~2.6 kJ/mol for 14 and ~1.3 kJ/mol for 20) resulting in a HYDE
contribution of the 5,6-dichloropyridine residues of about –18 and –21 kJ/mol for 14 and 20,
respectively (cp. Fig. 8A and 8B). In the case of 3,4-dichlorophenyl-substituted compounds 15 and 21,
the sum of the desolvation effects for both chlorine atoms at phenyl C3 and C4 positions is
approximately –16 (for 15) and –15 kJ/mol (for 21), leading to a partial HYDE score for the 3,4-
dichlorophenyl substituent of about –21 and –20 kJ/mol for 15 and 21, respectively.
2
3

ACCEPTED MANUSCRIPT
Fig. 8. SeeSAR visualization of binding of the N-unsubstituted compounds 14 (A) and 20 (B) and N1-
methylated analogs 15 (C) and 21 (D) to the crystal structure of the hMAO-B-safinamide complex
(PDB: 2V5Z) with HYDE quantification of the most important non-hydrogen atoms contributing to the
binding affinities (binding free energies ∆G) of compounds. HYDE visual affinity assessment: green =
favorable, red = unfavorable and non-colored = no relevant for affinity. For clarity, the most important
protein residues, water molecules (1180 and 1247) and FAD co-factor are shown in the same
orientation, color code and structure representation as in Fig. 8.
With exception of compound 14 that exhibits the lowest predicted hMAO-B potency (i.e.,
approximately –3 kJ/mol lower than calculated for 20), there is no major difference between the 5,6-
dichloropyridine in 20 and the 3,4-dichlorophenyl residues in 15 and 21, when comparing their partial
contribution to the total free binding energy. Further HYDE analyses provide similar contributions of
the most important HA to the free energy for the investigated compounds, including the amide spacers
in 14 and 15 (e.g., N11, C10, and O1) and imine atoms (C10 and N11) in 20 and 21, respectively (see
Figs. S5 and 7). As mentioned above, the flexible carboxamide linkage between two lipophilic moieties
in the indazole-5-carboxamides (subclass I) is crucial not only for the observed high MAO-B inhibitory
potency (NH-C=(O) contribution of ~4.8 and 5.0 kJ/mol for 14 and 15, respectively), but also acts as an
‘anchor’ in the more hydrophobic parts in the pocket [39]. However, our HYDE analysis provides hints
2
4

ACCEPTED MANUSCRIPT
toward why there is similar (or even better for 14 vs. 20) hMAO-B affinity of methanimine compounds
2
0 and 21 compared to their carboxamide analogs 14 and 15: There are "coronas" at their imine nitrogen
and also favorable contributions at the ether C10 atom (see Fig. 9).
Furthermore, the N1-indazole methyl group in 15 and 21 occupies the hydrophobic FAD-subpocket
favorably – yielding a HYDE contribution of approximately –4 kJ/mol in each molecule. The N1-
methylated compounds (15 and 21) exhibit significant preferences towards the hMAO-B enzyme
compared to the N-unsubstituted compounds (14 and 20) due to their higher entropic contributions
(
corresponding to desolvation effects) to the total binding energy of both lipophilic parts, comprising the
,4-dichlorophenyl ring and the N1-indazole methyl group. In the case of both N1-methylated
compounds (15 vs. 21), the additional H-interaction (enthalpic effect) at the carbonyl function (e.g., CO-
-HOH1247) in 15 did not show a higher contribution to the overlall binding energy when compared to
3
-
the imine linker in 21 (NH-C=(O) contribution of ~5.0 kJ/mol vs. ~5.3 kJ/mol of the N=CH linker). The
sum of the desolvation effects for the 3,4-di-Cl-Ph ring and the N1-methyl group in 15 (–25.2 kJ/mol) is
marginally higher than those in 21 (–23.2 kJ/mol). The HYDE analyses of compounds 14, 15 and 20, 21
reproduced their hMAO-B activities, which decrease as follows: 15 (IC50 = 0.386 nM, ∆G = –45.6
kJ/mol) > 21 (IC = 1.03 nM, ∆G = –43.6 kJ/mol) > 20 (IC = 1.28 nM, ∆G = –41.4 kJ/mol) > 14 (IC
50
5
0
50
=
5.42 nM, ∆G = –38.7 kJ/mol).
Finally, we computed the estimated affinities of our compounds (11–22) in SeeSAR (SeeSAR's
HYDE reports K ranges rather than values to avoid an overinterpretation of affinity estimation) at all
i
biological targets of interest (e.g., hMAO-A, hMAO-B, hAChE and hBuChE) (Fig. S7). All final poses
of the compounds were validated (with N = 5 poses after docking) and visually inspected. Reassuringly,
for hMAO-B, the selected poses lie in the respective affinity regions – as confirmed by biological
experiments. However, it should be mentioned that against hMAO-A and hChEs these compounds were
inactive and, therefore, no enzyme kinetics were performed. Overall, there is a good agreement with the
estimated affinity ranges for compounds under study at hMAO-B, e.g., HYDE scores (Ki HYDE ranges in
2
5

ACCEPTED MANUSCRIPT
the low nM range) / K values (in the (sub-)nM area) (cf. Fig. S7B and Table 1). Consistently, the N1-
i
substitution in 15 and 21 appears to be more favorable for hMAO-B inhibition (K = 0.17 and 0.46 nM
i
for 15 and 21 vs. 2.39 and 0.57 nM for 14 and 20).
In conclusion, the results obtained from the docking studies and HYDE analyses suggested that (i)
an imine linker is well-tolerated by hMAO-B; it has a similar contribution to the total binding energy in
methanimines 17–22 as the amide linker in 11–16, (ii) a replacement of the amide or imine linker by
other electron rich groups will also be favorable for hMAO-B inhibition, and (iii) considering the
optimal substituent length of four atoms, a large hydrophilic substitution should make the indazole N1
position less accessible for water, leading to an improvement of affinity toward hMAO-B or other
relevant targets because the desolvation penalty upon binding should be smaller. We esteem these
findings to be crucial for future design of CNS drug.
2
.5. Profiling of physicochemical properties
In the course of our comparative study we further analyzed the photophysical properties of
carboxamide 11 (subclass I compounds) and selected methanimines (subclass II compounds 17 and 20),
as well as the drug-like profile of all compounds 11–22 and reference safinamide. Due to the specific
physicochemical properties of Schiff bases (compounds 17–22), we performed further examination of
1
7 and 20 with focus on: (i) chemical stability and photosensitivity, and (ii) possible isomer (E/Z-
isomerism and tautomerism) formation and distribution by using of time- and solvent-dependent
photophysical experiments (UV-Vis), LC/ESI-MS analysis and quantum-chemical calculations.
2
.5.1. Photophysical evaluation
Electronic absorption spectra of 11, 17 and 20 were measured in different solvents, including dry
DMSO, acetonitrile (ACN) and methanol (MeOH) at a concentration of 50 µM (Table 3). The UV-Vis
spectra of carboxamide 11 exhibited different profile compared to those of both methanimines 17 and
2
0 due to the linker-specific absorbance features of chromophores (e.g., amide vs. imine linker) (see Fig.
2
6

ACCEPTED MANUSCRIPT
S8). The absorption maxima ( max
λ
) of 11 were found to be 265 (in DMSO) and ∼ 262 nm (in ACN and
MeOH), while for 17 and 20 the λmas values lie in the range of 300–325 nm. Compared to 11 and 17,
compound 20 exhibits higher max at the same experimental conditions due to its pyridine moiety. The
largest differences in the absorption maxima between 11 and methanimines 17 (λmax ∼ 322 nm) and 20
λmax = 325 nm) were observed to be 54 nm for 11 vs. 17 and 57 nm for 11 vs. 20, respectively (cp.
Table 3). Furthermore, the compounds showed slight differences in their molar absorptivity (
measured in the respective solvents. For all compounds, the molar absorptivities decrease with
λ
(
ε
)
increasing the solvent polarity, with the lowest
methanol.
ε values observed for methanimines 17 and 20 in
In addition, the UV-Vis experiments of 11, 17 and 20 show that no solvent-depending absorption
changes, neither in polar aprotic conditions (e.g., DMSO and ACN as solvent) nor in methanol, could be
observed. Therefore, the UV-Vis spectra suggest that under these experimental conditions (i)
carboxamide 11 as well as methanimines 17 and 20 exist in their most stable single tautomeric 1H-
indazole form and (ii) no structural changes (e.g., hydrolysis) of 17 and 20 could be detected (cp. Fig.
S8, for details, see Supporting Information). These results were confirmed by quantum-chemical
calculations performed with the uncharged possible tautomeric forms of 17 and 20 (cp. Section 2.5.2).
In order to further investigate the chemical stability of methanimine-based Schiff bases subjected in
this work, time-dependent stability experiments in different solvents with compound 20 were performed
(
Fig. S9). For this purpose, samples of 20 (at 50 µM) in DMSO, ACN or MeOH were exposed to
daylight for 76 days at ambient temperature and the UV-Vis absorption spectra measured across the
time (Fig. S9A-–S9C). The single and superposition UV-Vis spectra of the corresponding building
blocks, e.g., 1H-indazole-5-carbaldehyde (2) and 5,6-dichloropyridin-3-amine (6), measured at 50 µM
2
in MeOH/H O 1:1 were used as reference (Fig. S9D).
2
7

ACCEPTED MANUSCRIPT
a
Table 3 Photophysical data of compounds 11, 12, 17, and 20 (at 50 µM) in different solvents.
Compd.
DMSO
ACN
MeOH
b
c
b
c
b
c
λmax
ε
λmax
ε
λmax
ε
1
1
1
7
0
267.5
321.5
325
19426
19042
15334
262.5
300
19023
19168
18124
262
18486
17627
15063
312.5
314.5
2
313.5
a
b
c
–1
–1
All measurements were performed at 25.0 ± 0.01°C. Unit: nm. Unit: M cm . na = not applicable.
In the experiments with 20 in polar aprotic solvents, no spectral changes can be detected across the
time of 76 days, suggesting that no degradation of 20 occurs in DMSO and ACN and, therefore, these
solvents can be used for preparation of stock solutions of methanimines under study (compounds 17–
2
2). In contrast, the experiments with 20 in methanol clearly showed a decrease of the absorption
maxima accompanied by changes of the absorbance spectra, indicating that a slow degradation of 20 is
occurring over time (cp. Fig. 9C and 9E). Furthermore, we observed that 20 was stable during the first 3
days under these conditions, while a gradually hydrolysis (∼ 68%) of 20 was occurred after 76 days in
methanol (cp. Fig. S8F).
As next, we investigated the ability of imine 20 to undergo a photo-induced [2+2]-cycloaddition
(dimerization) as well as its concentration-dependent stability using different samples of the compound
in dry methanol (1.0 and 10 mM). The samples were then exposed to daylight at ambient temperature
for certain period of time, slowly evaporated either to dryness or to a minimum solvent and analyzed by
LC/ESI-MS or X-ray structural analysis. The results are depicted in Figure S10. In these experiments,
we observed that the imine 20 was relatively stable during the first 24–96 hours at both test
concentrations (1.0 and 10 mM). Only traces of a dimer of 20 were detected by LC/ESI-MS after 96
hours in methanol (at 1.0 mM). Compound 20 was found to be stable under these conditions, while a
slow hydrolysis of 20 was occurred after 62 days (Fig. S10A–10C). Crystallization of 20 was observed
when its sample (10 mM) was kept for 43 hours and slowly evaporated to a minimum solvent (Fig.
S10D and S10F). The structure of crystalline 20 was confirmed by X-ray crystal analysis (cp. Figure 7).
Altogether, the stability experiments demonstrated that the imine 20 is chemically stable and non-
2
8

ACCEPTED MANUSCRIPT
photosensitive for up to 96 hours in diluted as well as concentrated methanol solutions. In polar aprotic
solvents such as DMSO and ACN, compounds 17 and 20 are completely stable.
Moreover, the results obtained from the chemical stability and photosensitivity experiments
suggested that the imines 11–22 are relatively stable under the experimental conditions used in this
work, thus offering further opportunities for structural modifications. However, there are some
limitations to consider. It is known that a fast hydrolysis of imines could be observed in water/alcohol
mixtures and/or under acidic conditions, in which the degradation rate depends on the water/alcohol
ratio. Therefore, it appears to be indispensable to avoid the use of different alcoholic solutions for
performing test samples of these compounds, especially for long-term experiments.
2
.5.2. Evaluation of E/Z isomerism and tautomerism
For a better understanding of the relationship between the bioactivity at the configurational ground
state (e.g., E vs. Z isomers) and the isomeric stability of (indazole-5-yl)methanimines 17–22, the
respective E/Z-geometry and possible tautomeric forms of compounds 17 and 20 was studied by mean
of quantum-chemical calculations [59]. The E/Z-isomers for both compounds formed via a rotation of
the respective 3,4-dichloro-phenyl (for 17) and 5,6-dichloro-pyridine ring (for 20) around the
methanimie linker are illustrated in Figure S11. The possible tautomers (tautomeric indazole forms a–c)
of 17 and 20 are sketched in Fig. S12, while the relative energies of their isomers and conformers in gas
phase and different solvents are collected in Table S7 and S8, respectively (see Supporting Information).
The theoretical calculations were performed using the M06-2X functional [60] with TZVP basis set
[61]. This fitted hybrid meta-GGA functional with 54% Hartree-Fock (HF) exchange has been specially
developed to describe main-group thermochemistry and non-covalent interactions; it has previously
shown very good results for the prediction of tautomeric equilibrium compounds with intramolecular
hydrogen bonds [62]. Solvent effects are described by using the Polarizable Continuum Model (the
integral equation formalism variant, IEFPCM) [63].
2
9