Crystal structures, binding interactions, and ADME evaluation of brain penetrant N-substituted indazole-5-carboxamides as subnanomolar, selective monoamine oxidase B and dual MAO-A/B inhibitors

Article abstract of DOI:10.1016/j.ejmech.2017.01.011

The pharmacological and physicochemical analysis of structurally optimized N-alkyl-substituted indazole-5-carboxamides, developed as potential drug and radioligand candidates for the treatment and diagnosis of Parkinson's disease (PD) and other neurological disorders, is reported. Recent efforts have been focused on the development of subnanomolar potent, selective MAO-B (N1-alkyl-substituted compounds 12a–14a and 15) and dual active MAO-A/B (N2-methylated compounds 12b–14b) inhibitors with nanomolar potency towards MAO-B and moderately active against MAO-A enzyme, respectively. The most promising drug-like derivatives in both series were N-(3-chloro-4-fluorophenyl)-1-methyl-1H-indazole-5-carboxamide (13a, NTZ-1441, IC₅₀hMAO-B 0.662?nM, >15000-fold selective versus MAO-A) and N-(3-chloro-4-fluorophenyl)-2-methyl-2H-indazole-5-carboxamide (13b, NTZ-1442, IC₅₀hMAO-B 8.08?nM, IC₅₀hMAO-A 0.56?μM, SI?=?70). Moreover, compounds 13a and 13b were predicted to cross both the gastrointestinal tract (at pH 2.0, 5.5, and 7,4) and the blood-brain barrier (BBB) in?vitro with appropriate drug-like properties required for CNS active drugs. Combined single X-ray/molecular modeling studies provided insights into the enzyme–inhibitor interactions within both MAO isoforms and the rationale for their inhibitory activity with controlled MAO-A/B selectivity – despite their small structural differences. The binding modes of 12a,b and 13a,b confirmed that the major interactions with hMAO-B were established via the flexible carbonyl group of the carboxamide linkage and the electron-donating nitrogens N1 or N2 of the indazole moiety, allowing further exploration of the alkyl side chain for next step lead optimization efforts.

Full text of DOI:10.1016/j.ejmech.2017.01.011

Accepted Manuscript
Crystal structures, binding interactions, and ADME evaluation of brain penetrant N-
substituted indazole-5-carboxamides as subnanomolar, selective monoamine oxidase
B and dual MAO-A/B inhibitors
Nikolay T. Tzvetkov, Hans-Georg Stammler, Beate Neumann, Silvia Hristova, Liudmil
Antonov, Marcus Gastreich
PII:
S0223-5234(17)30019-3
10.1016/j.ejmech.2017.01.011
EJMECH 9162
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 29 November 2016
Revised Date: 21 December 2016
Accepted Date: 8 January 2017
Please cite this article as: N.T. Tzvetkov, H.-G. Stammler, B. Neumann, S. Hristova, L. Antonov,
M. Gastreich, Crystal structures, binding interactions, and ADME evaluation of brain penetrant N-
substituted indazole-5-carboxamides as subnanomolar, selective monoamine oxidase B and dual MAO-
A/B inhibitors, European Journal of Medicinal Chemistry (2017), doi: 10.1016/j.ejmech.2017.01.011.
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
TABLE OF CONTENTS GRAPHIC
1

ACCEPTED MANUSCRIPT
Crystal structures, binding interactions, and ADME evaluation of brain penetrant
N-substituted indazole-5-carboxamides as subnanomolar, selective monoamine
oxidase B and dual MAO-A/B inhibitors
a,
b
b
c
Nikolay T. Tzvetkov ∗ , Hans-Georg Stammler , Beate Neumann , Silvia Hristova ,
c
d
Liudmil Antonov , Marcus Gastreich
a
b
c
NTZ Lab Ltd., Krasno selo 198, Sofia 1618, Bulgaria
Department of Chemistry, University of Bielefeld, Universitätsstr. 25, 33615 Bielefeld, Germany
Bulgarian Academy of Sciences, Institute of Organic Chemistry, Centre of Phytochemistry, Acad. G.
Bonchev Str., Bl. 9, Sofia 1113, Bulgaria
d
BioSolveIT GmbH, An der Ziegelei 79, 53757 St. Augustin, Germany
KEYWORDS:
ADME; MAO inhibitors; Molecular modeling; Indazole-5-carboxamides; Parkinson’s disease; X-ray
∗
Corresponding author.
E-mail address: ntzvetkov@ntzlab.com (N. T. Tzvetkov). Phone: +49-179-528-4358
1

ACCEPTED MANUSCRIPT
Abstract
The pharmacological and physicochemical analysis of structurally optimized N-alkyl-substituted
indazole-5-carboxamides, developed as potential drug and radioligand candidates for the treatment and
diagnosis of Parkinson’s disease (PD) and other neurological disorders, is reported. Recent efforts have
been focused on the development of subnanomolar potent, selective MAO-B (N1-alkyl-substituted
compounds 12a–14a and 15) and dual active MAO-A/B (N2-methylated compounds 12b–14b)
inhibitors with nanomolar potency towards MAO-B and moderately active against MAO-A enzyme,
respectively. The most promising drug-like derivatives in both series were N-(3-chloro-4-fluorophenyl)-
1
-methyl-1H-indazole-5-carboxamide (13a, NTZ-1441, IC hMAO-B 0.662 nM, >15000-fold selective
50
versus MAO-A) and N-(3-chloro-4-fluorophenyl)-2-methyl-2H-indazole-5-carboxamide (13b, NTZ-
442, IC hMAO-B 8.08 nM, IC hMAO-A 0.56 µM, SI = 70). Moreover, compounds 13a and 13b
1
5
0
50
were predicted to cross both the gastrointestinal tract (at pH 2.0, 5.5, and 7,4) and the blood-brain
barrier (BBB) in vitro with appropriate drug-like properties required for CNS active drugs. Combined
single X-ray/molecular modeling studies provided insights into the enzyme–inhibitor interactions within
both MAO isoforms and the rationale for their inhibitory activity with controlled MAO-A/B selectivity
–
despite their small structural differences. The binding modes of 12a,b and 13a,b confirmed that the
major interactions with hMAO-B were established via the flexible carbonyl group of the carboxamide
linkage and the electron-donating nitrogens N1 or N2 of the indazole moiety, allowing further
exploration of the alkyl side chain for next step lead optimization efforts.
2

ACCEPTED MANUSCRIPT
1. Introduction
Parkinson’s disease (PD) is the second most prevalent chronic neurodegenerative disorder of the
central nervous system (CNS), typically characterized by the deficiency of the neurotransmitter
dopamine (DA) in the nigrostriatal pathway. It results in a progressive impairment of the core motor
functions and can be recognized by the major symptoms of bradykinesia (slowness), rigidity, resting
tremor, and postural instability, as well as by various non-motor symptoms such as depression,
behavioral and cognitive complications that are using to clinical diagnose of PD [1–4]. To date, drugs
approved for the treatment of PD are those acting on the dopaminergic and monoaminergic systems [5–
7]. In particular, a combination of levodopa replacement therapy with DA agonists, DOPA-
decarboxylase inhibitors (DDIs), as well as monoamine oxidase B (MAO-B) and/or catechol-O-
methyltransferase (COMT) inhibitors are often used for the treatment of PD [1,3,8]. Most of the
currently approved medicines have an impact of several motor symptoms especially in early-stage PD,
but do not replace the loss of neuronal cells or stop the disease progression [9,10].
Monoamine oxidases (МАОs, EC 1.4.3.4) are mitochondrial flavoenzymes that are involved in the
oxidative deamination of exogenous and endogenous amines, including neurotransmitters in both the
peripheral nervous system (PNS) and CNS. Two subtypes of MAOs have been identified in mammals,
MAO-A and MAO-B, encoded by distinct genes with opposite orientation on the X-chromosome
(Xp11.23-11.4), where they share ~70% identity of the protein sequence [11,12]. Both MAO isoforms
are essential for the inactivation of monoaminergic neurotransmitters but display regional differences in
enzyme activity, substrate preference and distribution in the human brain, and therefore, different
clinical significances [13,14]. For example, DA is a substrate for both isoforms but it is mainly
metabolized by MAO-B in the substantia nigra, where MAO-B is the main form in glial cells and the
increased MAO-B activity may lead to dopamine deficiency in the human brain, whereas serotonin (5-
hydroxytryptamine, 5-HT) is preferentially degraded by MAO-A and the loss of MAO-A in humans is
associated with aggressive behavior [13]. The expression levels and activity of MAO-B in the human
3

ACCEPTED MANUSCRIPT
brain, rather than those of MAO-A, increase ~4-fold with aging, leading to a higher production of
hydrogen peroxide (H O ) and other reactive oxygen species (ROS), which are associated with
2
2
oxidative stress and neuronal cell death [15,16]. Enhanced activity and overexpression of MAO-B in the
brain is thus observed in patients with Alzheimer’s disease (AD) and PD [14,17,18]. Due to their
affinity for specific neurotransmitters, selective inhibition of MAO enzymes is an established approach
for treating of mental disorders (MAO-A) and PD (MAO-B) [19–26].
For example, selective MAO-A inhibitors such as the irreversible inhibitor clorgyline and the
reversible inhibitor moclobemide are used in the treatment of depression and anxiety (for structures, see
Fig. 1) [19,20]. The irreversible MAO-B inhibitors selegiline and rasagiline are currently applied in PD
monotherapy or in a combination with levodopa in late-stage PD [21,22]. However, irreversible MAO
inhibition may cause pharmacological side effects and safety complications [23]. Recently, safinamide
has been approved as an add-on drug to DA agonists or to levodopa for the treatment of motor
symptoms in patients with early or mid-to late-stage PD [24,25]. Safinamide is a multitarget compound
exhibiting both, dopaminergic properties (selective and reversible inhibition of MAO-B and DA
reuptake) and non-dopaminergic mechanism of action, including selective sodium channel blockade and
calcium channel modulation with consequent inhibition of the excessive glutamate release [26].
Fig. 1. Structures of irreversible (Irr) and reversible (Rev) MAO inhibitors in clinical use.
4

ACCEPTED MANUSCRIPT
Considering the beneficial role of reversible MAO inhibition in neuroprotection and modulation of
various pathophysiological processes associated with PD [27], dual and reversible inhibitors of MAO A
and B, rather than single MAO-B inhibitors, may have increased therapeutic and neuroprotective
potential not only in the treatment of PD but also for AD therapy. Such compounds should inhibit
MAOs selectively in the brain with slight preference for MAO-B than for the A-form compared to the
inhibition of these enzymes in the peripheral tissues [28]. Moreover, it has been observed that brain-
selective MAO drugs may exhibit the combined benefit of antioxidative and neurorestorative activities
with a reduced risk for toxicity and side effects [29].
Because of the promising pharmacological properties and favorable safety profiles associated with
reversible MAO inhibitors (MAOIs), we have been particularly interested in developing such inhibitors.
Recently we have reported the discovery of a series of novel selective, reversible and competitive
MAO-B inhibitors with subnanomolar potency [30,31]. Applying a structure-based drug design strategy,
we performed systematic structural modifications of different heterobicyclic moieties (indazole, indole,
imidazopyridine, and triazopyridine) by introducing a large number of mono- and di-substituted
aromatic rings (phenyl, pyridine and dihydropyridine) or by replacing the type (carboxamide, inverse
amide, methanimine, benzamide, and acetamide) and the position (C5 or C6) of the linker located
between the heterobicyclic motif and the exocyclic aromatic ring (Fig. 2). Thus, several compounds of
the structurally related substance classes of indazole-5-carboxamides, indole-5-carboxamides and
(indazol-5-yl)methanimines were identified in vitro as best-in-class MAOIs that may now serve as
promising lead structures or even as drug, radioligand, and diagnostic candidates, e.g., for the treatment
of PD and other neurodegenerative diseases [32]. Among these new series, indazole-5-carboxamides
were evaluated to be the most promising, selective and reversible inhibitors of human MAO-B, many of
them with MAO-B activity in subnano- or even picomolar range and significantly improved in silico
physicochemical and drug-like properties [32]. Consequently, further exploration has been performed
within the series of indazole-5-carboxamides by introducing a lipophilic methyl and methoxyethyl
5

ACCEPTED MANUSCRIPT
substituent either at the N1- or N2-position of the indazole moiety. The 3,4-dihalogen-substituted
phenyl residue and the carboxamide linker were kept constant (Fig. 2). A similar inhibitor construction
is also present in previously reported 5-carboxamidoindole derivatives with a reversed carboxamide
linker [27,33]. However, the herein presented scaffolds realize very favorable water interactions by an
amide flip and the introduction of an indazole N2 atom, entailing a pronounced affinity increase.
In the present study we report a series of structurally optimized N-alkylated indazole-5-carboxamide
derivatives. All these compounds contain common pharmacophoric features: a hydrophobic 3,4-
disubstituted-phenyl unit (Ph), hydrogen bonding domains (HBD, a carboxamide linker with hybrid
character) and additional electron donor nitrogens (ED) in the indazole moiety. The compounds were
tested at rat and human MAO A and B and, depending on the substitution pattern of the indazole N1- or
N2-position, they may act either as selective MAO-B or dual MAO-A/B inhibitors with reversible mode
of action. To find a plausible explanation for their high affinity and the observed differences in
selectivity, docking experiments were performed using a novel modeling technology platform using X-
ray structures of selected compounds under study. Moreover, we evaluated the drug-like and in vitro
ADME properties of the studied compounds, including their water-solubility, logD, blood-brain-barrier
(BBB) penetration, and gastrointestinal (GI) permeability.
Fig. 2. Generic structure of previously discovered selective MAO-B inhibitors [31] (left) and intended
chemical modifications of the indazole-5-carboxamide scaffold subjected in this work (right).
6

ACCEPTED MANUSCRIPT
2. Results and discussion
2.1. Chemistry
Synthetic routes and chemical structures of the investigated compounds are shown in Scheme 1.
Following our synthetic strategy for the formation of N-alkylated indazole-5-carboxamide derivatives
possessing different alkyl substituents either at the indazole nitrogen atom 1 or 2, respectively, we
started from the commercially available 1H-indazole-5-carboxylic acid (1). The advantage of this
methodology consists in the introduction of the corresponding N-alkyl substituent either in the very last
or in the first synthetic step, namely preparation of 12a,b–14a,b (convergent approach A) and 15
(regioselective approach B) by only two or four steps, respectively.
Scheme 1. Synthetic routes to N1- and N2-substituted derivatives 12a–14a, 12b–14b and 15. Reagents
and conditions: (i) EDC–HCl/DIPEA, DMF, r.t., 3–16 h; (ii) for 9–11: MMS (method 1) or MeI
(
method 2), K CO , DMF, r.t., 3–20 h; (iii) 1) MeOH, c. H SO (10 mol-%), 1–3 h, 65–70 °C, 2) 1-
2
3
2
4
bromo-2-methoxyethane, K CO , DMF, reflux, 9–16 h, 3) 2
M
NaOH, THF–H O (1:1), 30–35 °C, 1–3 h.
2
3
2
7

ACCEPTED MANUSCRIPT
The preparation of indazole-5-carboxamide derivatives 8–11 was performed following a known
procedure [31], but in this work we used optimized reaction conditions leading to 8–11 in higher yields
(
81% yield in average). The carboxylic acid 7 was prepared following a three-step reaction procedure
for details, see Supporting Information). Amide coupling was carried out by reaction of 1 and 7 with
(
3,4-disubstituted anilines 2–5. The subsequent alkylation of 9–11 was performed with methyl
methanesulfonate (MMS, method 1) and methyl iodide (method 2) as alkylating reagents, yielding
regioisomeric mixtures of the corresponding N1- and N2-methylated products 12a,b–14a,b. To improve
the product yields and regioselectivity, the reaction was conducted with several modifications and
improvements [34]. Higher yields of 12a,b–14a,b (71% combined yield in average) were obtained when
using MMS (method 1), whereas higher regioselective formation (preference for N1- vs. N2-substitution
of the indazole moiety, see Table 1) was achieved by methylation of 9–11 with methyl iodide (method
2). The final products 12a,b–14a,b were separated from their N-methylated mixtures by preparative
RP18-HPLC. Product 15 was purified chromatographically. The optimized procedures allow the
introduction of broad structural modifications, e.g., by a variation of the alkylation reagents and the
differently substituted anilines, and therefore, an easy access to compound libraries for rapid evaluation
of structure-activity and structure-selectivity relationships (SARs and SSRs).
In addition, we obtained the X-ray structures of 9, 11, 12a, and 13b, which confirmed the assumed
preference of the 1H-indazole tautomeric form of the N-unsubstituted compounds 9 and 11, as well as
the methylation at the N1-position in 12a and 13a (for more details, see Section 2.3).
Table 1 N1- vs. N2-regioselectivity and yields for the alkylation of 9–11.
a
b
Starting compd.
9
Method
N1/N2 product mixture
12a/12b
N1/N2 ratio
Combined yield (%)
c
1
2.8 : 1
4.7 : 1
71
56
d
2
1
1
0
1
1
2
13a/13b
14a/14b
2.7 : 1
4.5 : 1
72
59
1
2
3.1 : 1
4.3 : 1
72
61
a
c
b
Product ratio was determined by quantitative LC/ESI-MS of the isolated product mixtures. Isolated yields refer to 1.
d
Method 1: alkylating reagent MMS. Method 2: alkylating reagent MeI.
8

ACCEPTED MANUSCRIPT
2.2. Monoamine oxidase studies
The compounds were tested for their inhibitory activities on rat and human MAO-A and MAO-B
using mitochondria-enriched rat liver fractions and microsomes of baculovirus-infected insect cells
BTI-TN-5B1-4) as sources for the rat and human recombinant MAO isoforms, respectively [31]. The
(
evaluation of the enzyme inhibition of the test compounds against MAO-A and MAO-B isoforms was
performed by a fluorescence-based assay [35] with the common substrate p-tyramine measuring the
compounds' effects on the production of hydrogen peroxide (H O ) using the commercial assay kit
2
2
Amplex Red [36]. The two-step reaction for the determination of MAO activity involves an H O -
2
2
production via oxidative deamination of p-tyramine by MAO (step 1) and a subsequent horseradish
peroxidase (HRP)-catalyzed oxidation of the non-fluorescent and highly sensitive 10-acetyl-3,7-
dihydroxyphenoxazine (Amplex Red reagent) by H O to produce a fluorescent product, resorufin (step
2
2
2). The ratio of resorufin production is equal to the ratio of p-tyramine that is oxidized by MAO to the
corresponding 2-(4-hydroxyphenyl)acetaldehyde in the above two-step reaction. To irreversibly block
the respective isoenzyme in rat liver fractions, the inhibitors clorgyline (for MAO-A) and selegiline (for
MAO-B) were applied for determining selective inhibition of the other isoenzyme [31]. These selective
inhibitors, clorgyline and selegiline, were also used as positive controls in the corresponding MAO-A
and MAO-B assays, respectively. Determined in vitro inhibitory potencies (IC values) and selectivity
5
0
toward hMAO-A and hMAO-B (expressed as selectivity index, SI) for all compounds and reference
inhibitors are reported in Table 1. In addition, the kinetic parameters of hMAO-B (K and Vmax) were
m
determined under the above experimental conditions in the presence of different concentrations of
p-tyramine (Fig. S2, Supporting Information).
9

ACCEPTED MANUSCRIPT
Table 2 Monoamine oxidase activity of the tested compounds.
a
IC ± SEM (nM)
5
0
1
b
Compd.
8
R
R
rat MAO-A
human MAO-A rat MAO-B
human MAO-B SI
H
H
>10000
>10000
≥10000
>10000
>10000
>10000
420 ± 24
<10000
≥10000
562 ± 62
>10000
764 ± 40
2870 ± 218
708 ± 34
117 ± 13
>86
9
3,4-Cl
3-Cl, 4-F
3,4-F
H
H
H
>10000
1.43 ± 0.11
2.36 ± 0.17
8.89 ± 0.05
1.32 ± 0.09
22.5 ± 0.6
5.57 ± 0.65
1.36 ± 0.13
20.3 ± 3.0
5.65 ± 0.48
221 ± 21
0.586 ± 0.087
0.679 ± 0.044
1.59 ± 0.16
0.386 ± 0.052
1.44 ± 0.41
1.87 ± 0.12
0.662 ± 0.056
8.08 ± 1.05
1.52 ± 0.18
38.4 ± 1.8
1.08 ± 0.08
nt
17065
14727
6289
25906
292
1
1
1
1
0
>10000
1
>10000
2a
2b
3,4-Cl
3,4-Cl
3,4-Cl
Me
Me
Me
>10000
1740 ± 60
>10000
c
1
1
2a/12b
5348
15106
70
3a
3
3
-Cl, 4-F Me
-Cl, 4-F Me
>10000
(
NTZ-1441)
1
3b
1690 ± 143
>10000
(
NTZ-1442)
1
1
1
4a
4b
5
3,4-F
3,4-F
3,4-Cl
Me
Me
6579
20
1880 ± 93
EtOMe >10000
3.82 ± 0.22
nt
>2657
d
Moclobemide
–
–
–
–
10700 ± 700
361 ± 19
–
e
e
Safinamide
580000
rat brain)
45000
(human brain)
18.0 ± 4.9
5.18 ± 0.04
5000
e
e
(
a
b
c
n = 3, unless otherwise noted. Selectivity index: SI = IC (hMAO-A)/IC (hMAO-B). Tested as a mixture of N1-/N2-
5
0
50
d
e
isomers (N1/N2 ratio, 4.7:1). Data are from ref. 37. Data are from ref. 30. nt = not tested.
10

ACCEPTED MANUSCRIPT
2.2.1. SARs and SSRs of indazole-5-carboxamide derivatives
The main goal of the present study was to evaluate the role and effects of the indazole N1- or N2-
substitution with respect to the inhibition of both MAO-A and MAO-B enzymes. Based on their SARs
and SSRs, the investigated compounds in Table 1 can be divided into three closely related chemical
series as follows: N-unsubstituted indazole-5-carboxamides (designated subclass I, compounds 8–11),
N1-substituted indazole-5-carboxamides (subclass II, 12a–14a and 15), and N2-methyl-substituted
indazole-5-carboxamides (subclass III, 12b–14b). In general, all tested compounds show high inhibition
of MAO-B ranging from low micromolar to picomolar potency. The biological evaluation of subclass I
compounds was included to demonstrate the effect of the lipophilic phenyl residue on MAO inhibitory
activity. Compounds of subclasses II and III were designed to further investigate the effects on
inhibition potency and selectivity at both target MAO enzymes. A few substituents (mainly small
lipophilic methyl groups) were also introduced at the position N1 or N2 of the indazole core to deeply
understand SARs and SSRs of these new series of indazole-5-carboxamides. Evaluation of the
biological data of all N-substituted compounds resulted in the identification of promising specific MAO-
B (subclass II compounds) and dual MAO-A/B inhibitors (compounds of subclass III).
From all tested compounds, the N-unsubstituted compound 8 without substituents at the phenyl ring
shows the lowest MAO inhibitory potency (human MAO-B, IC = 117 nM; rat MAO-B, IC = 708
5
0
50
nM). The introduction of electron-withdrawing groups (Cl and F) at the meta- and para-position of the
phenyl ring in compound 8 resulted in remarkably potent and selective MAO-B inhibitors showing
picomolar potency (subclass I compounds 9–11). The 3,4-dichlorophenyl-substituted indazole-5-
carboxamide 9 was found to be the most potent compound in this series (human MAO-B, IC = 0.586
5
0
nM; rat MAO-B, IC = 1.43 nM), being 9-fold more potent against human MAO-B than the standard
5
0
inhibitor safinamide. The presence of 3-chloro-4-fluoro (compound 10, human, IC = 0.679 nM; rat,
5
0
IC = 2.36 nM) and 3,4-difluoro substituents (compounds 11, human, IC = 1.59 nM; rat IC = 8.89
5
0
50
50
nM) led to highly potent inhibitors of human and rat MAO-B, the potencies of which are comparable to
11

ACCEPTED MANUSCRIPT
those of the parent compound 9. Compared to safinamide, compounds 10 and 11 display a >7- and >3-
fold increase in inhibitory potency toward hMAO-B, respectively.
Following our rational design strategy, we used the 3,4-dihalophenyl-substituted indazole-5-
carboxamides 9–11 as basic structures for further modifications of the indazole N1 and N2 positions. As
a next step, the indazole N1 atom was substituted to obtain the N1-alkyl-substituted regioisomers 12a–
1
4a and 15. The N1-methylated derivatives 12a–14a proved to be extremely potent and selective
inhibitors of MAO-B with slightly increased potency when compared to their N-unsubstituted analogs
–11. Methylation at the indazole N1 position in 9 provided the most potent MAO-B inhibitor
compound 12a) of all compounds investigated (human, IC = 0.386 nM; rat, IC = 1.32 nM; SI =
9
(
5
0
50
25906), being ~2-fold more potent than the parent compound 9. The N1-methylated 3-chloro-4-
fluorophenyl- and 3,4-difluorophenyl derivatives 13a (human, IC = 0.662 nM; rat, IC = 1.36 nM)
5
0
50
and 14a (human, IC = 1.52 nM; rat, IC = 5.65 nM) display equally high activity against rat and
5
0
50
human MAO-B and selectivity versus MAO-A (13a, SI = 15106; 14a, SI = 6579) compared to the
potency and selectivity of the corresponding N-unsubstituted compounds 10 and 11. It should be noted
that for 12a–14a (subclass II compounds) the 3,4-dihalophenyl-substituted phenyl ring has the same
effect on MAO-B inhibitory activity and selectivity as observed for N-unsubstituted indazole-5-
carboxamides 9–11 (subclass I compounds). The IC50 values and SI slightly decrease within both
subclasses in the rank order as follows: 3,4-di-Cl (9, 12a) ≈ 3-Cl, 4-F (10, 13a) > 3,4-di-F (11, 14a).
This observation could be explained with a decrease in the lipophilic character (Cl vs. F) of these
compounds (cp. the modeling studies below). Next, we continued with the modification of the indazole
N1 position in 9 by introducing of a larger methoxyethyl substituent (compound 15). In comparison to
the N1-methyl-substituted indazole-5-carboxamides and the unmethylated lead structure 9, compound
1
5 is similarly potent against MAO-B (human MAO-B, IC = 1.08 nM; rat MAO-B, IC = 3.82 nM)
50 50
and approximately 5-fold more potent than safinamide. Moreover, compound 15 was also identified as
the only example in this series exhibiting notable inhibitory activity at human MAO-A (IC = 2870
5
0
12

ACCEPTED MANUSCRIPT
nM), whereas at rat MAO-A it is essentially inactive (10 µM). The N1-methoxyethyl substitution of the
indazole moiety was found to be preferable for MAO-B inhibition, 15 is still >2657-fold selective for
human MAO-B over human MAO-A. Our results suggest that the active site of human MAO-B in the
substrate cavity region (within a close vicinity to FAD of about 3-4 Å) is able to accommodate
lipophilic substituents with an appropriate length up to four atoms at the indazole N1 position.
Consequently, the indazole N2 position in compounds 9–11 was methylated producing N2-methylated
regioisomers 12b–14b (subclass III). All of them were found to be dual MAO A and B inhibitors with
low to subnanomolar potencies against hMAO-B and moderate nanomolar potencies against hMAO-A.
Compared to their N1-methylated regioisomers 12a–14a, the potency at rat and human MAO-B
decreased, while an increase of the inhibitory activity against rat MAO-A (IC = 1.74, 1.69, and 1.88
5
0
µM) and human MAO-A (IC = 420, 562, and 764 nM) was observed for N2-methylated analogs 12b–
5
0
14b, respectively. Compound 12b was identified as the most potent dual MAO-A/B inhibitor in this
series (human MAO-B, IC = 1.44 nM; human MAO-A, IC = 420 nM) with ~15-fold higher affinity
50
50
for the human than for the rat MAO-B enzyme (rat MAO-B, IC = 22.5 nM), displaying the major
5
0
species differences of all investigated compounds. The N2-methylated compounds 13b and 14b are less
potent than the corresponding N1-substituted isomers 13a and 14a and the unmethylated analogues 10
and 11, but they are still highly potent against human MAO-B (13b, IC = 8.08 nM; 14b, IC = 38.4
5
0
50
nM). Compounds 12b and 13b show MAO-B and MAO-A inhibitory potencies when compared to the
standard reversible inhibitors moclobemide (MAO-A) and safinamide (MAO-B), respectively. Thus, the
N2-methyl substitution of the indazole moiety was evaluated to be beneficial for inhibition of human
MAO-A with preferences for MAO-B over MAO-A isoenzyme (12a, SI = 292; 13b, SI = 70; 14b, SI =
20). Finally, a mixture of both most potent N1- and N2-methylated regioisomers 12a and 12b (4.7:1
ratio) was tested in order to evaluate the combined effect of the indazole N1 and N2 substitution on
MAO-B inhibitory activity and selectivity against MAO-A. The mixture of 12a and 12b showed high
MAO-B inhibitory potency (human MAO-B, IC = 1.87 nM; rat MAO-B, IC = 5.57 nM), being
50
50
13

ACCEPTED MANUSCRIPT
>5000-fold selective for human MAO-B over human MAO-A. On the basis of its high bioactivity,
which was comparable to the single N1- and N2-methylated isomers 12a and 12b, such mixture may as
well be considered for further development of radioligands.
2.2.2. Species differences
Meaningful species differences and clear correlations between rat and human IC50 values were
previously described not only for subclass I compounds, but also for other related classes MAO-B
inhibitors [30]. In order to access the species selectivity, all compounds were tested at rat and human
orthologs of MAO-A and B. As shown in Fig. 3, significant differences in inhibitory potencies
(
expressed as pIC ) at rat and human MAO-B for compounds of all three series could be noticed. A
50
good correlation between rat and human MAO-B inhibitory activity could be observed when all
determined pIC values at rat MAO-B versus pIC values at human MAO-B isoform were plotted.
5
0
50
2
Linear regression gave a slope of 1.10, y-intercept equal to –1.47, and R = 0.93 (Fig. S3, Supporting
Information). Compounds of subclass I and II were virtually inactive at MAO-A at the highest
concentration tested (10 µM). Therefore, species differences at MAO-A could only be observed for the
N2-methylated indazole-5-carboxamide derivatives 12b–14b (subclass III compounds). Overall, the
potency of the compounds at human MAO isoforms was noticeably higher than at their rat orthologs.
The N2-methylated derivatives 12b and 14b showed the highest deviation between the pIC values at
5
0
rat and human MAO-B (~1.2 log units for 12b) and the pIC values at rat and human MAO-A (>1.4 log
5
0
units for 14b), respectively. Considering that a similar trend in species differences was observed for
structurally related compounds [32], it will be required to determine MAO inhibition of both A and B
isoforms first against the human enzyme and, only for selected compounds, against their rat orthologs
before starting preclinical and in vivo clinical studies.
14

ACCEPTED MANUSCRIPT
Fig. 3. Rat vs. human MAO-B pIC values distribution within the series of indazole-5-carboxamides 8–
5
0
1
5. Almost all compounds are highly potent MAO-B inhibitors (pIC > 8.00).
50
2
2
.3. Crystal structures
.3.1. Crystal structures of 9 and 11
In order to understand the rationale behind the MAO inhibitory and selectivity properties of N-
substituted indazole-5-carboxamides (subclass II and III compounds), our first step was to determine the
single crystal molecular X-ray structures of selected N-unsubstituted precursors 9 and 11 as well as N1-
methylated derivatives 12a and 13a, and to evaluate their specific inter- and intramolecular interactions
within the respective crystal units (see Experimental Section and Supporting Information).
An overview of the X-ray structural analysis of compounds 9 and 11 with atom numbering is shown
in Fig. 4. The crystal structure of 9 confirms its almost planar conformation. The two fused five- and
six-membered rings are forming together the first plane of the molecule, maximum deviation from the
mean plane of the indazole moiety shows C13 with 0.015(1) Å. A second plane, defined by C1-C6 of
the planar 3,4-dichlorophenyl-ring, is tilted by 3.67(3)° against the first one, the nitrogen atom N1
deviates 0.08(1) Å from this plane. The aromatic and indazole units are connected by a carboxamide
15

ACCEPTED MANUSCRIPT
linkage with torsions of 177.7(1)° (C8-C7-N1-C5) and –1.2(2)° (O1-C7-N1-C5). The angles are
121.5(1)° (O1-C7-C8) and 122.7(1)° (N1-C7-O1). The plane defined by the carboxamide linker (C5,
N1, C7, O1, C8) encloses an angle of 32.16(4)° with the indazole moiety and an angle of 33.09(5)° with
the phenyl ring. To investigate the ability of indazole-5-carboxamides to form intermolecular hydrogen
bonds under conditions that mimic the biological targets of interest, e.g., MAO A and B enzyme, we
studied the effect of water and different solvents on the conformational changes and tautomerism of the
indazole moiety in 11. Therefore, crystals suitable for X-ray structural analysis were obtained after slow
recrystallization of 11 from a methanol–water mixture (2:1) as well as methanol and acetonitrile
solvents at room temperature (RT). The X-ray structure of 11 reveals a different rotamer compared to 9
(Fig. 4B). The indazole and phenyl rings are not co-planar but titled by 62.12(4)° (in 9: 3.67°), which is
clearly demonstrated by an overlay of the indazole moieties of 9 and 11 (Fig. 5). A solvent-depending
rotamerization of 11 was not observed. The deviation of C7 atom from the best plane of the indazole
moiety is 0.01(1) Å. The respective torsions (176.4(2)° and –3.1(2)°) and the angles (121.5(1)° and
123.0(1)°) of the carboxamide linkage are similar to those observed in the crystal structure of 9. The
plane defined by the carboxamide linker (C5, N1, C7, O1, C8) encloses an angle of 25.42(4)° with the
indazole moiety and an angle of 36.85(5)° with the phenyl ring. Relevant structural data for all single X-
ray structures in the present work are listed in Table S9 (see Supporting Information).
16

ACCEPTED MANUSCRIPT
Fig. 4. Single X-ray structures of compounds 9 (A) and 11 (B) showing the atom numbering. The
thermal ellipsoids are drawn at the 50% probability level.
Fig. 5. Overlay of compounds 9 and 11 showing the different rotamers. Hydrogen atoms are omitted for
clarity.
In comparison to the crystal structure of 9, the indazole moiety, the phenyl residue, and the carbonyl
group in 11 have another orientation to each other indicating of a rotamerization (rotation around the
carboxamide bond C8–C7), resulted probably by intermolecular interactions in the crystal by hydrogen
bonds (Fig. S3, Supporting Information). Both compounds build a chain by hydrogen bridging between
H1 and O1 by translation of the molecule along the b axis with N-H…O distances of 2.07(2) for 9 and
17

ACCEPTED MANUSCRIPT
2.26(2) Å for 11. Additional hydrogen bridges can be found between N3-H3 and N2 generated by a
two-fold screw axis with N-H…N distances of 2.03(2) Å resp. 2.00(2) Å, thus both molecules form
double chains.
The molecules adopt an antiparallel orientation; the plane-to-plane angle defined by the mean plane
of all atoms of an individual molecule is of 75.9(2)° for 9. For 11, the plane-to-plane angle defined by
the atoms of the indazole moiety was 60.2(1)°. As can be seen from Fig. S3 the different rotamers
shows similar double chains, the difference lies direction of the second N-H…N hydrogen bond
compared to the first one: In 9 they point into opposite directions, in 11 they point into the same
direction. Our results suggest that the carboxamide linker in 9 and 11 play an important role for the
formation of hydrogen bonds due to its conformational and hybrid features (hydrogen bond
acceptor/donor) allowing structural flexibility of indazole-5-carboxamides within the MAO enzymes,
i.e., under physiological conditions (e.g., pH 7.4 at ∼ 37 °C). Moreover, the X-ray analysis of 9 and 11
indicates that the comparably high MAO-B inhibitory activity of N-unsubstituted indazole-5-
carboxamides (subclass I) is due to their stable 1H-indazole and keto tautomeric forms. Otherwise, a
2
H-indazole and/or keto-enol tautomerism would be observed in the case of compound 11, which was
recrystallized in the presence of water (see also Section 2.5 and Experimental Section).
2.3.2. Crystal structures of compounds 12a and 13a
The X-ray structural analysis of compounds 12a and 13a with the atom numbering is shown in Fig.
. Single crystals of 13a were twinned with a ratio of 59:41 by a rotation of 180° around the [001]
6
direction. Similarly to 9, the crystal structures of 12a and 13a confirm their almost planar conformation.
The two fused five- and six-membered rings in the indazole moiety are almost coplanar; no atoms of
these units show a deviation of more than 0.02 Å out of that plane. The deviation of the carbonyl C7 and
the methyl C15 atoms are 0.12(1) Å respectively 0.07(1) Å for 12a and 0.15(1) Å respectively 0.16(1) Å
for 13a. The second plane, defined by the phenyl ring, is tilted by 6.01(5)° (in 12a) and 5.7(1)° (in 13a)
against the first one (Fig. 6).
18

ACCEPTED MANUSCRIPT
Fig. 6. Single X-ray structures of compounds 12a (A) and 13a (B) showing the atom numbering. The
thermal ellipsoids are drawn at the 50% probability level.
The nitrogen atom N1 resides nearly perfectly in the plane of the phenyl ring (deviation 0.02(2) Å
and 0.01(1) Å, respectively). Interestingly, the plane defined by the carboxamide linker is tilted versus
both ends, namely 27.5(1)° to the indazole moiety and 28.3(1)° to the phenyl ring in 12a. The
corresponding values are 29.1(1)° and 30.5(1)°, respectively, in 13a. The bond angles and lengths of the
carboxamide linker in all X-ray structures are almost identical (cp. Table S9). As well as the
intermolecular patterns of 9 and 11 for both structures, intermolecular hydrogen bonds can be found
between H1 and O1 generated by translation along the b axis forming chains with H...O distances of
2.17(3) Å in 12a and 1.99(4) Å in 13a (Fig. S4). Because of the lack of a second N-H donor compared
to 9 and 11, no comparable anti-parallel double chains are formed. This conformation does not prevent
the carboxamide linkage to build geometrically sound intermolecular hydrogen bonds. The
intermolecular interaction geometries of compounds 9, 11, 12a and 13a are collected in Table S10.
19

ACCEPTED MANUSCRIPT
2
2
.4. Molecular Modeling Studies
.4.1. Docking of compounds 12a, 12b, 13a, and 13b into human MAO-B
SAR and SSR evaluation of indazole-5-carboxamide analogs reported herein, including the X-ray
analysis of compounds 9, 11, 12a and 13a, as well as previous docking studies with the structurally
related indole-5-carboxamide derivative NTZ-1010 [39] provided valuable information about the
binding modes determining their MAO-B inhibitory potency. The co-crystal X-ray structures of both
hMAO-A and hMAO-B enzymes with the reversible inhibitors harmine (PDB code: 2Z5X) [40] and
safinamide (PDB code: 2V5Z) [41] provided insight into the structural features of these key biotargets
for the research in the field of MAO inhibition in relation to their therapeutic use in neurological
disorders.
As a next step it occurred therefore natural to select the N1-methylated compounds 12a and 13a as
well as their N2-methyl-substituted derivatives 12b and 13b for molecular modeling experiments. After
a successful applicability check of the methods applied (re-docking and scoring), we investigated their
binding situation and rationalized the most significant interactions and desolvation effects within the
substrate active site of hMAO-B (12a,b and 13a,b) and hMAO-A (12b and 13b). The results shall
provide additional information including (i) an explanation for the observed preference in inhibitory
potency at hMAO-B isoform of N1- versus N2-methyl-substituted regioisomers (12a and 13a vs. 12b
and 13b), on the one hand side, and versus safinamide on the other, (ii) a semi-quantification and
visualization of the hydrophobic effects and enthalpic and entropic binding, (iii) an exploration of the
unoccupied space within the substrate cavity region of hMAO-B for further development of CNS drugs,
and (iv) proposals of binding modes of dual MAO-A/B active compounds 12b and 13b within the
binding site of the human MAO-A isoform. Binding modes were computed with LeadIT [42], free
energy of binding estimates were calculated and visualized with the software SeeSAR [43] using a
novel, yet well-validated computational free energy approximation "HYDE" in combination with single
X-ray analysis [44]. Since biological data showed a species-dependent preference in inhibitory potency
20

ACCEPTED MANUSCRIPT
for human versus rat MAO orthologs, the modeling studies were run on the human model of the MAO-
A and MAO-B enzyme (see Experimental Section). Given their availability, the docking experiments
were performed with the ligands' single X-ray structure geometries (i.e., bond lengths and angles) of
1
2a and 13a without further preparation, leading to an excellent agreement between experiment and
computation (cf. Supporting Information). The selected docking solutions of N2-methylated compounds
2b and 13b were guided using co-visualization with their N1-methylated isomers 12a and 13a within
1
the binding site of the hMAO-B enzyme.
On the basis of the docking model and previous observations [30,39], the binding pocket of human
MAO-B can be divided into three different subpockets inside the protein from the ligands perspective:
(
i) an indazole binding region is anchored in the "substrate cavity" and oriented toward the FAD with a
postulated H-bond 1 between the N2 atom and the water molecule 1180 (designated subpocket I, S1),
ii) a linker region with a postulated H-interaction 2 between the oxygen of the carbonyl group and the
water molecule 1247 (subpocket II, S2), and (iii) a highly lipophilic halogen binding subpocket
hydrophobic subpocket III, S3) located in the "entrance cavity" of the human MAO-B enzyme (Fig. 7
(
(
and 8). Furthermore, the binding modes and estimated affinities strongly suggest that 12a,b and 13a,b
occupy the same substrate cavity space as previously determined for the N-unsubstituted compound 9
[
30], provided that both ligands do not covalently bind with the FAD cofactor. Compounds 12a and 13a
differ from 9 by a methyl substitution at N1 and, therefore, their binding poses show some significant
features with respect to the orientation of the indazole scaffold in 9 within the subpocket S1. The X-ray
analysis of 9 and re-assessing the geometries with SeeSAR resulted in a similar conformation to the
ones of the N1-methyl analogs 12a and 13a, which results from a rotation on the C8-C7–N1-C5 torsion
angle of about 180° (cp. Fig. S6).
21

ACCEPTED MANUSCRIPT
Fig. 7. (A) Binding modes of 12a (magenta sticks) and 13a (orange sticks) overlaid onto the crystal
structure of the hMAO-B/safinamide complex (PDB: 2V5Z). The main surface area of safinamide is
shown as a grey mesh. For clarity, only the relevant residue side chains and the FAD cofactor are shown
in grey sticks; all water molecules except HOH1180 and HOH1247 (red spheres) were removed. The
crystal structure of MAO-B is represented as ribbons. Clipping planes have been used for better
visibility [45]. (B) 2D binding model for 12a and 13a differ only by the halogen substitution (Cl vs. F,
light grey sphere) in the 4-position of the phenyl ring. The postulated H-interactions 1 and 2 are
displayed as dashed lines. The dashed arrow between the FAD cofactor (grey ellipse) and the N1-methyl
substituent (red sphere) indicates a favorable orientation of the indazole moiety within the active
binding site of hMAO-B.
In addition to the N1-methyl substitution, the indazole N2 atom and the carboxamide linker of 12a
and 13a are suggested to play an essential role for their conformation within the substrate cavity region
of hMAO-B (subpocket I and II) interacting favorably with the water molecules 1247 (CO---HOH1247
≈
1.8 Å) and 1180 (N2---HOH1180 ≈ 1.9 Å), respectively (Fig. 7B). Finally, the 3,4-dihalo-substituted
phenyl ring of the ligands occupies a strongly hydrophobic binding pocket in the models (subpocket III)
dominated by hydrophobic effects (see the HYDE analyses below). There are a multitude of
hydrophobic amino acids, notably leucin, isoleucin and proline residues. In summary, we computed that
for this subclass of N1-methyl indazoles the two relevant H-bonds are highly coordinated also by
protein residues (for HOH1247: GLN206, ILE199, TYR201 and 326; for HOH1180: CYS172 and
TYR188), making them more essential for the ligand stabilisation (CO---HOH1247) and conformation
(N2---HOH1180) in the binding site through non-covalent interactions.
22

ACCEPTED MANUSCRIPT
From earlier HYDE analyses it emerged that N2 is the only atom contributing slightly unfavorably
in the N-unsubstituted indazoles [30], it appeared therefore natural that an option for further
optimization of this class of compounds could proceed from, for example, a vector at this very "water-
like" N2 position. This computed proposal is now experimentally backed by the N2-methyl indazole
regioisomers 12b–14b (subclass III compounds), which are not only well tolerated by human MAO-B,
but also showed a moderate activity for the human MAO-A isoenzyme. With the aim to deeper
understand why the N1-methyl-substituted indazoles exhibit higher affinity towards human MAO-B
than their N2-methylated regioisomers (e.g., more than one log unit K difference between 13a and 13b),
i
we compared compounds 12a and 13a with their respective N2-methylated isomers 12b and 13b. For
this purpose, we computed the binding models of 12b and 13b on the structural basis of their N1-
methylated analogues using the new 3D structure editor in SeeSAR (Fig. 8).
Fig. 8. (A) Binding modes of 12b (grey sticks) and 13b (pink sticks) overlaid onto the crystal structure
of the hMAO-B-safinamide complex (PDB: 2V5Z). The main surface area of safinamide is shown as a
grey mesh [45]. (B) 2D binding model for 12b and 13b differ only by the halogen substitution (Cl vs. F,
grey sphere). The dashed arrow between the FAD cofactor (grey ellipse) and the N2-methyl substituent
(red sphere) indicates a favorable orientation of the indazole moiety within the active binding site of
hMAO-B.
Similarly, compounds 12b and 13b share analog binding mode motifs in the active binding site as
observed for 12a and 13a (cp. above) and also comparable desolvation effects. However, there are some
23

ACCEPTED MANUSCRIPT
differences in ligand-receptor interactions and desolvation contributions observed in particular for (i) the
N2-methyl group and (ii) the neighboring N1 position in the indazole moiety: Here again, the indazole-
binding site of the pocket is limited by the FAD cofactor (Fig. 8B).
The N2-methyl group of 12b and 13b is closer to the FAD co-factor than for the N1-methylated
analogs 12a and 13a (distance to the flavin C4 and C4a atoms of about 3.8–4.1 Å). Furthermore, the
N2-indazole methyl group in 12b and 13b occupies the hydrophobic subpocket favorably – yielding a
HYDE contribution of approximately –6 kJ/mol (i.e., approx. –2 kJ/mol lower than for 12a and 13a). In
the case of an indazole N2-methyl substitution, we observed only one H-bond between water 1247 and
the carboxamide linkage (CO---HOH1247 ≈ 1.8 Å) and hydrophobic contacts with the amino acids
ILE199 and TYR326 (for details, see Table S11 in the Supporting Information). In conclusion, the N1-
(12a and 13a) and N2-methyl-substituted indazoles (12b and 13b) exhibit slightly different binding
affinities towards human MAO-B enzyme (e.g., 12a, IC = 0.39 nM; 12b, IC = 1.44 nM; 13a, IC =
5
0
50
50
0
.66 nM; 13b, IC = 8.08 nM) due to their different entropic contributions (hydrophobic/desolvation
50
effects) to the total binding energy and an additional H-interaction (enthalpic effect) at the N2 position
in 12a and 13a. Compared to the N1-methyl substitution, the N2-methyl group has a slightly different
space orientation in the FAD cleft that might be to avoid clash and, therefore, to a regiochemically
unfavorable "switched" conformation within the active site of hMAO-B. Furthermore, it is well known
that unfavorable desolvation effects may well lead to a decreasing in potency, even though the tightness
of fit is good [46].
Regrettably, there are no co-crystal structures of 12a,b and 13a,b available today to confirm these
models experimentally. However, we have good belief in the modeled binding modes for the following
reasons: (i) the cores of 12a,b and 13a,b are in exactly the same binding region as observed for
safinamide in X-ray structures before, and (ii) the trends of affinity are correctly described: The
unsubstituted phenyl analog 8 shows an approximately 177- and 15-fold reduced hMAO-B activity in
the experiment compared to 13a and 13b, respectively. This is in (quantitative) agreement with the
24

ACCEPTED MANUSCRIPT
HYDE predicted contribution of the two halogen atoms of 13a (approx. –14 kJ/mol) and 13b (approx. –
1
3 kJ/mol), corresponding to a contribution of almost three orders in magnitude of the K values (Fig. 9).
i
Fig. 9. SeeSAR visualization of the binding of compounds 12a (A), 13a (B), 12b (C), and 13b (D) to
human MAO-B (PDB: 2V5Z). HYDE visual affinity assessment: green = favorable, red = unfavorable
and non-colored = no relevant for affinity. The ligand, protein residues, water molecules and FAD are
shown in the same orientation, color code and structure representation as in Fig. 7 and 8.
To gain insight into the overall binding thermodynamics of 12a,b and 13a,b to human MAO-B and,
therefore, to find a plausible explanation for their preference in inhibitory potency versus safinamide,
we applied HYDE scoring and visualization as embedded in SeeSAR. HYDE considers H-bonds
(enthalpy) and dehydration ("desolvation", entropic) effects of all atoms using a united atom
computation at the end (i.e., non-hydrogen atoms contain the hydrogen contributions); the atomic
visualization as spheres ("HYDE coronas") reflects both the protein as well as the ligand energy terms.
SeeSAR was applied after dockings (for details, see Fig. S7 and Table S12, Supporting Information) and
entails an RMSD-limited pre-optimization towards clash-free and torsionally relaxed output. Our HYDE
study was particularly focused on the estimation and visualization of the major ligand-receptor
interactions plus desolvation effect estimations at (i) the indazole N1- and N2-methyl function and (ii)
25

ACCEPTED MANUSCRIPT
the 3,4-dihalo-substituted phenyl ring, especially at the phenyl C4 position (chloro- vs. fluoro-
substitution). The indazole-binding site of the pocket is limited by the FAD cofactor (cf. Fig. 7 and 8).
The N1-methyl substitution, which entails a displacement of HOH1180, is computed as responsible for
a slight affinity increase compared to safinamide alongside the effects of the respective 3,4-dihalophenyl
substituents in 12a and 13a. The H-bond interaction between the indazole N2 atom and the water 1180
(∆H of about –18 kJ/mol) is compensated by a desolvation "penalty" at TYR435 for both ligands (~15
kJ/mol) resulting in a HYDE contribution of this bond of about –4 kJ/mol for both ligands (cp. Table
S11). In the case of safinamide, the water molecule 1192 and the residue GLN206 are involved in H-
bond interactions with the NH group in the α-aminoamide polar part – leading to a HYDE score of
2
about –3.7 kJ/mol.
Since the fluorine atom is much smaller than the chlorine atom (cp. 12a), the HYDE model predicts
slightly less affinity for the fluoro-substituted compound 13a (Fig. 9A and 9B). However, the error bars
largely overlap – in other words, the computer cannot distinguish between the energies of the two.
HYDE reveals –15.6 kJ/mol in total for both chlorine atoms at phenyl C3 and C4 positions in 12a,
whereas the scoring for 13a provides a value of –9.7 kJ/mol for the chlorine and –4.4 kJ/mol for the
fluorine at the phenyl C4 position, respectively. The sum of the desolvation effects for both halogens in
13a is approximately –14.1 kJ/mol (i.e., –1.5 kJ/mol lower than calculated for 12a). Further HYDE
analyses provide similar contributions of the most important heavy atoms (e.g., amide spacer, N1, C6,
C3 and methyl carbon atom) to the free energy for both compounds. The flexible carboxamide linkage
between two lipophilic moieties in the indazole-5-carboxamides is crucial not only for the observed high
MAO-B inhibitory potency (NH-C=(O) contribution of ~5–6 kJ/mol; also cp. the preferences for N1-
methyl-substituted indazoles versus safinamide), but also acts like an ‘anchor’, steering the positioning
of the hydrophobic parts in the pocket. Our HYDE analysis provides hints towards why there is better
hMAO-B affinity of the N1-methyl indazole-5-carboxamides compared to safinamide, e.g., "red
corona" at its imine nitrogen and no favorable contributions at the ether carbon atom, versus the overall
26

ACCEPTED MANUSCRIPT
favorable contributions computed and visualized in 12a and 13a (cp. Fig. S6B). The thermodynamic
profile computed by HYDE (that takes explicit, untrained account of enthalpy-entropy compensation
[
47,48 and references therein]) of safinamide to hMAO-B showed almost equal, favorable contributions
of the enthalpic and entropic terms (approx. –20 kJ/mol each) to the total free binding energy (Fig. S7).
Compounds 12a and 13a showed similar thermodynamic profiles with a predominant enthalpic
contribution of –38 kJ/mol and favorable entropic terms of –9.9 and –7.5 kJ/mol, respectively. In
contrast, the free energy of 12b and 13b is rather driven by desolvation (entropic) effects summing up to
a favorable –T∆S value of –25 kJ/mol (cp. Fig. S7). Finally, the results obtained from the HYDE
analyses of compounds 9 and 11–13 reproduced their hMAO-B activities, which decrease in the
following rank ordering: 12a > 13a ≈ 9 > 12b ≈ 11 > 13b.
The results obtained from the docking studies and HYDE analyses of N1- and N2-methyl-
substituted indazole-5-carboxamides prompted us to further explore the potential of this promising class
of MAO-B inhibitors for the development of CNS drug candidates. A detailed examination of their
SARs and thermodynamic profiles suggested that a large hydrophilic substitution should make the
indazole N1 position less accessible for water, entailing a "quenching" (downsizing) of unfavorable
desolvation effects and, therefore entail an improvement of affinity with respect to the N2 alternation
[
32].
In this light, we compared the best docking poses of our compounds (9–15) computed in SeeSAR
with those observed from the biological experiments. All final poses of the compounds were validated
number of docking poses per ligand N = 5) and visually inspected. The estimated affinities (SeeSAR's
(
HYDE reports K ranges rather than values to avoid the illusion of a non-existing accuracy of our
i
computations) of all selected poses lie in the nanomolar (or even better) affinity region – as confirmed
by biological experiments (Table S13, Supporting Information).
Table 3 summarizes the experimental and computed affinity data for safinamide and compounds 9–
1
5 at human MAO-B (K values in the (sub-)nanomolar range). There is a good overall agreement with
i
27

ACCEPTED MANUSCRIPT
the estimated affinity ranges, e.g., HYDE scores / K values (low nM). As expected from a visualization
i
of the situation in SeeSAR, consistently, the N1-substitution appears to be highly more favorable (0.17
and 0.29 nM for N1 vs. 0.64 and 3.57 nM for N2, and 0.48 nM for 15 with N1, too).
Table 3 HYDE estimated affinity ranges and experimental K values for the inhibition of human MAO-
i
B by safinamide and compounds 12–15.
a
b
Compd.
N1-/N2-substitution
Ki HYDE (nM)
0–37
K (nM)
i
1
1
1
1
1
2a
2b
3a
3b
5
N1
N2
N1
N2
N1
0.17
0.64
0.29
3.57
0.48
2.29
1–123
1–103
1–148
<6.05
Safinamide
–
2–179
a
b
Estimated HYDE K range values. The K values were approximated from the experimentally measured IC values
i
i
50
according to the equation by Cheng and Prusoff [49]: K = IC /(1 + [S]/K ) with [S] = 150 µM and K (p-tyramine) =
i
50
m
m
1
18.8 µM.
An essential input towards further optimizing binding of the indazole moiety and exploring the
unoccupied space of the substrate cavity region of hMAO-B was obtained by comparing the
experimental data for the N1-methoxyethyl-substituted compound 15 and the docking experiments of
12a and 13a (Table 3). We esteem these findings to be crucial for future MAO-B inhibitor design,
especially considering the optimal substituent length of four atoms and the type of functional groups
needed. Docking studies and a subsequent HYDE analysis of all target structures are in line with a
proposal of a non-covalent binding mode to human MAO-B, typically estimated for reversible inhibitors
and experimentally confirmed for safinamide and the indazole derivatives 9 and 12a [30–32].
2.4.2. Docking of compounds 12b and 13b into human MAO-A: applicability test in re-docking harmine
As a precaution it should be noted that the unavailability of a MAO-A co-crystal structure with a
comparably large ligand leaves uncertainties for this molecular modeling proposal that uses fixed
protein heavy atom positions. Regrettably, we could not get hold of the simulated receptor structure of
the human MAO-A enzyme alone; to the best of our knowledge it has not yet been disclosed [50]. We
therefore used the co-crystal structure available in the PDB that had been co-crystallized with harmine
28