Synthesis and Pharmacological Characterization of Conformationally Restricted Retigabine Analogues as Novel Neuronal Kv7 Channel Activators

Article abstract of DOI:10.1021/acs.jmedchem.9b00796

Kv7 K⁺ channels represent attractive pharmacological targets for the treatment of different neurological disorders, including epilepsy. In this paper, 42 conformationally restricted analogues of the prototypical Kv7 activator retigabine have been synthesized and tested by electrophysiological patch-clamp experiments as Kv7 agonists. When compared to retigabine (0.93 ± 0.43 μM), the EC₅₀s for Kv7.2 current enhancements by compound 23a (0.08 ± 0.04 μM) were lower, whereas no change in potency was observed for 24a (0.63 ± 0.07 μM). In addition, compared to retigabine, 23a and 24a showed also higher potency in activating heteromeric Kv7.2/Kv7.3 and homomeric Kv7.4 channels. Molecular modeling studies provided new insights into the chemical features required for optimal interaction at the binding site. Stability studies evidenced improved chemical stability of 23a and 24a in comparison with retigabine. Overall, the present results highlight that the N5-alkylamidoindole moiety provides a suitable pharmacophoric scaffold for the design of chemically stable, highly potent and selective Kv7 agonists.

Full text of DOI:10.1021/acs.jmedchem.9b00796

Subscriber access provided by BIU Pharmacie | Faculté de Pharmacie, Université Paris V
Article
Synthesis and pharmacological characterization of conformationally
restricted retigabine analogues as novel neuronal Kv7 channels activators
Carmine Ostacolo, Francesco Miceli, Veronica Di Sarno, Piera Nappi, Nunzio Iraci, Maria Virginia
Soldovieri, Tania Ciaglia, Paolo Ambrosino, Vincenzo Vestuto, Anna Lauritano, Simona Musella,
Giacomo Pepe, Manuela Giovanna Basilicata, Michele Manfra, Diego Romano Perinelli, Ettore
Novellino, Alessia Bertamino, Isabel M. Gomez-Monterrey, Pietro Campiglia, and Maurizio Taglialatela
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b00796 • Publication Date (Web): 09 Dec 2019
Downloaded from pubs.acs.org on December 9, 2019
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the
course of their duties.

Page 1 of 96
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Synthesis and pharmacological characterization of
conformationally restricted retigabine analogues as
novel neuronal Kv7 channels activators
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
†,#
¶,#
‡,#
¶
‡
Carmine Ostacolo, Francesco Miceli , Veronica Di Sarno , Piera Nappi , Nunzio Iraci , Maria
§
‡
¥
‡
¶
Virginia Soldovieri , Tania Ciaglia , Paolo Ambrosino , Vincenzo Vestuto , Anna Lauritano ,
⊥
‡
‡
⸆
Simona Musella , Giacomo Pepe , Manuela G. Basilicata , Michele Manfra , Diego Romano
†
‡
†
‡,*
Perinelli⸆ , Ettore Novellino , Alessia Bertamino , Isabel M. Gomez-Monterrey , Pietro Campiglia ,
¶
,*
Maurizio Taglialatela .
† Department of Pharmacy, University Federico II of Naples, Via D. Montesano 49, 80131, Naples, Italy
¶ Department of Neuroscience, Reproductive Sciences and Dentistry, University Federico II of Naples, Via
Pansini, 5, 80131, Naples, Italy
‡
§
Department of Pharmacy, University of Salerno, Via G. Paolo II 132, 84084, Fisciano, Salerno, Italy
Department of Medicine and Health Science V. Tiberio, University of Molise, Via F. de Sanctis, 86100,
Campobasso, Italy
Department of Science and Technology (DST), University of Sannio, Via Port'Arsa 11, 82100, Benevento
Italy;
¥
⸆
⸆
Department of Science, University of Basilicata, Via dell'Ateneo Lucano 10, 85100 Potenza, Italy
School of Pharmacy, University of Camerino, Via Gentile III da Varano, 62032, Camerino (MC), Italy
⊥ European Biomedical Research Center (EBRIS), Via Salvatore de Renzi, 3, 84125 Salerno (SA), Italy
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSTRACT
+
Kv7 K channels represent an attractive pharmacological targets for the treatment of different
neurological disorders, including epilepsy. In this paper, 42 conformationally restricted analogues of
the prototypical Kv7 activator retigabine have been synthesized and tested by electrophysiological
patch-clamp experiments as Kv7 agonists. When compared to retigabine (0.93±0.43 µM), the EC s
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
for Kv7.2 current enhancements by compound 23a (0.08±0.04 µM) was lower, whereas no change in
potency was observed for 24a (0.63±0.07 µM). In addition, compared to retigabine, 23a and 24a
showed also higher potency in activating heteromeric Kv7.2/Kv7.3 and homomeric Kv7.4 channels.
Molecular modeling studies provided new insights into the chemical features required for optimal
interaction at the binding site. Stability studies evidenced improved chemical stability of 23a and 24a
in comparison with retigabine. Overall, the present results highlight that the N5-alkylamidoindole
moiety provides a suitable pharmacophoric scaffold for the design of chemically-stable, highly-potent
and selective Kv7 agonists.
2
ACS Paragon Plus Environment

Page 3 of 96
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
Epilepsy is one of the most common neurological disorders affecting approximately 1% of the
worlwide population. Although several classes of antiepiletic drugs are available for clinical use,
about one third of patients do not benefit from currently-available medications; thus, drugs with novel
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
mechanisms of actions might significantly reduce this figure.1
+
The K 7 (KCNQ) subfamily of K channels comprises five members, each showing a distinct tissue
v
2
distribution and physiological roles. K 7.2 and K 7.3 are the most abundant K 7 subunits in the
v
v
v
central and peripheral nervous system; heteromeric channels composed of K 7.2 and K 7.3 subunits
v
v
+
are thought to underlie a K current, named M-current (I ), whose activation reduces neuronal
KM
+
excitability by driving the membrane potential to values close to the equilibrium potential for K ions,
thus limiting repetitive firing and causing spike frequency adaptation. The critical role exerted by IKM
in controlling neuronal excitability is also highlighted by the observation that mutations in the genes
encoding for K 7.2 and K 7.3 are responsible for a wide spectrum of epileptic diseases ranging from
v
v
Benign Familial Neonatal Seizures to severe epileptic encephalopathies, and that drugs acting as Kv7
3
activators are potent anticonvulsants in humans. Retigabine is the proptotype Kv7 activator,
4
recognizing a hydrophobic pocket located in the pore domain; within this cavity, H-bond interactions
are established between the carbamate group of retigabine and a tryptophan residue in the channel
5
pore (W236 in Kv7.2, W265 in Kv7.3). Retigabine has been approved as an add-on treatment for
drug-resistant partial onset seizures with or without secondary generalisation in people aged 18 or
older, and has been on the market since 2011; however, because of a low benefit/risk ratio, clinical
use of retigabine has been declining over the years, leading to its discontinuation in 2017. Several
drawbacks are likely responsible for the limited clinical efficacy of retigabine which can be
summarized as follows: a. Poor selectivity for Kv7 subtypes. Indeed, activation of Kv7.4/5 channels
expressed in genitourinary smooth muscle cells seems responsible for one of the most frequently-
reported side effect, namely urinary retention; b. Short half-life. Retigabine is metabolized by phase-II
enzymes (acetylation and N-glucuronidation), with little involvement of the cytochrome P450
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
system. Consequently, retigabine has fast clearance kinetics (t1/2 ~ 6-8 hours), and requires three-
6
times-a-day dosing; c. Poor brain penetration. Because of its limited lipophilicity (Log P= 3.08), the
brain-plasma concentration ratio of retigabine is rather low, thus requiring posological schemes at
7
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
relatively high doses; d. Chemical instability.One of the main clinical concern over retigabine is its
8
tendency to cause retinal and muco-cutaneous blue-gray discoloration; although the mechanism for
this toxic effects remains poorly understood, one hypothesis is that light exposure may cause
photodegradation and oxidation of retigabine’s aniline ring, which may lead to dimer formation,
partly in conjuction with melanin, which would be responsible for the abnormal pigmentation light-
9
exposed tissues. . To possibly overcome some of these limitations, we have designed and synthesized
a small library of 42 novel conformationally-restricted retigabine derivatives, and characterized some
of these compounds with respect to: potency, selectivity, chemical stability, and in vitro
pharmacokinetics. In addition, molecular modeling studies were performed to identify the chemical
features required for optimal interaction between the syntetized compounds and Kv7 channels.
Noteworthy, in view of the previously-mentioned removal from the market of retigabine, and of the
even more recent withdrawal of the marketing authorization for flupirtine, the precursor of retigabine
used as a non-opioid analgesic, several naturally-occurring or synthetic small molecules with
heterogeneous chemical scaffolds acting as Kv7 activators have been recently described.10 In
particular molecules bearing indoline and indole scaffolds and acting as Kv7 activators have been
previously described in a patent by Valeant Pharmaceuticals;11 nonetheless, a comprehensive
description of their structure-activity relationships and a detailed pharmacological characterization of
these molecules in terms of pharmacodynamics, metabolism and photostability is lacking.
4
ACS Paragon Plus Environment

Page 5 of 96
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
RESULT AND DISCUSSION
Design and Chemistry
A new library of structural analogues of the Kv7 agonist retigabine was synthesized introducing a
conformational restriction of retigabine (I) scaffold aimed at increasing the potency and the metabolic
and photochemical stability by blocking the labile 4-NH and removing the 2-NH of benzene-1,2,4-
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
2
triamine scaffold.12 By this strategy, indoline (A), indole (B) and tetrahydronaphthalene (C)
derivatives were designed and synthesized (Figure 1). Moreover, the designed structures allow
extensive and chemically accessible modifications to investigate structure-activity relationships. The
indoline derivatives (series A, Figure 1) were designed maintaining the substituted benzyl group at
the N-1 position while varying the nature of both the R1 chain and the linker X on the amino group
at position C-5. At this position, we have chosen a series of linker groups characterized by different
electrostatic potential, electron density distribution, and N-X rotation energy profiles, such as amide,
sulphonamide, urea, thiourea and guanidine, which could mimic the retigabine pattern HB acceptor-
HB donor-aromatic ring (Figure S1 and Table S1).
H
N
R1
X
N
A
R4
H
R2
N
R1
R3
H
N
O
O
H
N
O
C
N
NH
R1
2
X
H
F
Retigabine
N
B
R2
R4
R2
R3
2
X= CO, SO , CONH, CSNH, CNHNH2
Figure 1. Structures of retigabine and its newly synthesized analogues
The first series of indoline analogues was synthesized according to the following scheme 1.
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Scheme 1: Synthesis of 1-(4-fluorobenzyl)-5-substituted indoline derivatives 4a-b, 5a-b, 6, 7, 8a-i and 9a-c.
H
R
N
O
R
CH2CH3 4a
O
N
H
N
H
N
R
R
4
b
O
C
5a
5b
N
H2
F
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
e
sgen
IPE
R
pho
D
Cl
O
A
F
ri
T
NaHCO3
H2
O
H
RN
H
N
H
N
H
H
NO2 F
NO2
NH2
CS , TEA
N
N
[H2]
HgO, Na SO
2
2
4
S
NH
N
N
N
N
N
(CH3COO)3BH4
H
6
NH4OH
7
H2N
1
2
3
R
SO2 Cl,
F
F
F
F
D
B
U
H
N
R
RCOCl
DIPEA
S
8a
C
8e
N
8h
O
O
H
2
N
R
H
8b
8c
N
R
H2
C
CH3 9a
R
N
8i
O
8f
N
F
O
C
H2
O
9b
CH2
C
8g
8d
H2
H2C
9c
F
Starting from 5-nitroindoline (1) a reductive amination at the N-1 led to the corresponding 1-(4-
fluorobenzyl)-5-nitroindoline (2) that, upon continuous flow hydrogenation to the corresponding
amine (3), was used in the next steps. Reaction of 3 with ethyl or menthyl chloroformate in alkaline
media gave the corresponding carbamate derivatives 4a and 4b. Urea derivatives 5a and 5b were
prepared by one-pot reaction of 3 with triphosgene and 2,2-dimethylpropan-1-amine or
cyclohexylamine, using DIPEA as base. On the other hand, treatment of 3 with CS and 2,2-
2
dimethylpropan-1-amine using TEA as base lead to the thiourea analogue of 5a 1-(1-(4-
fluorobenzyl)indolin-5-yl)-3-neopentylthiourea (6). Derivative 6 was converted to the corresponding
carbodiimmide by reaction with HgO and then reacted with NH OH to give the guanidine derivative
4
7
. Finally, coupling of intermediate 3 with different sulfonyl chlorides at the presence of DBU led to
the sulfonamide derivatives 8a-i, while coupling with acyl chlorides using DIPEA as base led to the
amide derivatives 9a-c. A second series of indoline analogues, investigating the influence of N-1
substituents on the pharmacological activity, was synthesized according to schemes 2 and 3.
Scheme 2: Synthesis of 1-benzyl-5-cyclohexanesulfonamido indoline derivatives 13a-g.
6
ACS Paragon Plus Environment

Page 7 of 96
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
H
R
R1
R2
R3
O
S
NO2
NH2
Cl
N
S
R
R
O
R
13a
13b
13c
13d
13e
13f
O
H
H
H
H
H
F
F
H
F
H
H
H
H
F
H
H
H
O
N
[H2]
N
N
11a-f
12a-f
DBU
CF3
OCH3
CF3
O
R3
H
R3
R3
R3
CF3
R2
R1
R1
R1
R1
CH3 OCH3
H
H4
R2
R2
R2
) B
3
O
O
C
H3
(
C
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
NO2
1) [
R
H
2]
N
2
)
DBU,
R1
R= H R1=H
R= CH3 R1=H 10a
R=H
R1=Boc 10b
1
O
S
H
Cl
H
N
H
N
O
N
S
S
O
(CH COO) BH
S
O O
3
3
4
TFA
O
O
O
NO2
N
N
H
N
O
H
1
3g
Boc
11g
12g
Br
O2N
Br
5-nitroindoline (1) or 2-methyl-5-nitroindoline (10) were derivatized by reductive amination using
differently functionalized aldehydes, giving intermediates 11a-f. Continuous flow hydrogenation of
these intermediates led to the amines 12a-f that were converted to the corresponding
cyclohexylsulfonamides (13a-f). Due to the lability of the 4-bromo-2-nitrobenzyl group to the
hydrogenation conditions, final product 13g was obtained using a different synthetic approach. The
N-1 position of 5-nitroindoline (1) was protected as tert-butoxycarbonyl derivative obtaining the
intermediate tert-butyl 5-nitroindoline-1-carboxylate (10b). The nitro group of compound 10b was
reduced to the corresponding amine (11g) and coupled with cyclohexane sulfonyl chloride, as
previously described, before removal of the Boc protecting group by trifluoroacetic acid (TFA). The
intermediate 12g, thus obtained, underwent to reductive amination using 4-bromo-2-
nitrobenzaldehyde. Functionalization of the N-1 was also attained by reaction of 1 with 4-
chlorobenzoylchlorides, 2-(4-chlorophenyl)acetyl chloride and 1-(2-bromoethyl)-4-fluorobenzene to
give intermediates 14a-b and 17 (Scheme 3). These intermediates were reduced by continuous flow
hydrogenation and converted to the corresponding sulfonamides 16a-b and 19 by coupling with
cyclohexanesulfonyl chloride (scheme 3).
Scheme 3: Synthesis of 1-acyl-5-cyclohexanesulfonamido (16a-b) and 1-alkyl-5-cyclohexanesulfonamido
(19) indoline derivatives.
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
H
NO2
NH2
O
S
Cl
N
S
O
O
N
N
O
R1
n
[H ]
Cl
2
N
O
1
n
4a-b
15a-b
O
O
DBU
O
F
Cl
0
1
16a
16b
n
n
n
R1
DIPEA
NO2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
R1
R1
N
Br
R1
H
1
O
S
H
Cl
F
NO2
H ]
NH2
N
NaH
S
O O
O
[
2
N
N
N
DBU
17
18
19
F
F
F
Indole analogues (B, figure 1) were synthesized according to scheme 4. Intermediate 20b was
obtained starting from 4-hydroxybenzaldehyde, that reacted with methoxymethyl chloride (MOMCl)
to give 4-(methoxymethoxy)benzaldehyde (19a). The protected aldehydes was reduced to the
corresponding alcohol (20a) using NaBH and then subjected to iodination affording 1-(iodomethyl)-
4
4
-(methoxymethoxy)benzene (20b). Nucleophilic displacements of 20 or commercially available 4-
substitued benzyl halides by 5-nitroindole in strong basic medium provided the nitro intermediates
1a-d, which were reduced to the corresponding substituted amino indole derivatives 22a-d, as
2
previously described. Coupling of these intermediates with acyl chlorides afforded the final alkyl
amide derivatives 23a-b, 24a and 25a. Removal of the methoxymethyl group from derivative 23b by
TFA led to 24b. On the other hand, reaction of amine intermediates 22a-d with
cyclohexanesulfonylchloride gave final sulfonamides 23c, 23d and 26a, while reaction with 2-
cyclohexylethane-1-sulfonyl chloride (20d) led to the sulfonamide derivative 27a. For the synthesis
of 28a, intermediate 22a was reacted with chloroacetyl chloride and then subjected to nucleophilic
displacement by n-butylamine. Alkylation of the secondary amine on the side chain by reductive
amination with formaldehyde gave final product 29a.
Scheme 4: Synthesis of 1-benzyl-5-amido (23a-25a, 23b-24b, 28a-29a) and 1-benzyl-5-
cyclohexanesulfonamido (23c-23d, 26a-27a) indole derivatives.
8
ACS Paragon Plus Environment

Page 9 of 96
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
O
H
O
HO
I
O
H
R1
R
MOMCl
NaBH4
(Ph)3P
I2
H
N
R
2
3a
CF3
C
O
O
O
H2
O
N
OH
24a
O
O
CF3
C
H2
19a
20a
20b
CF3
C
25a
23b
24b
H2 12
R1
OCH2OCH3
OH
C
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
H2
TFA
RCOCl
DIPEA
H
H
NO2
[
NH2
N
N
R1
N
N
H
O
O
NO2 20b or
X
N
N
N
N
H2]
1) ClCOCH2Cl, DIPEA
2) nBuNH2, tBuOK
HCOH
NaBH4
N
NaH
H
R1
1 a-d
R1
CF3
CF3
Cl
2
S
22 a-d
28a
29a
O
O
DBU
or
0d
2
R1
F
n
0
H
N
S
n
23c
O
O
O
O
N
Br
Br
HS
S
OCH3 0
23d
26a
27a
S
1
) H2N
NH2
Oxone
KBr, water
CF3
CF3
0
2
2)
DBU
R1
20c
20d
Tetrahydronaphthalene analogues (C, figure 1) were synthesized as depicted in scheme 5. Reaction
of 30 with heptanoyl chloride and DIPEA, gave the amide intermediate 31, which was reduced to the
tetrahydronaphthol derivative 32 using NaBH as reducing agent. Intermediate 32 was subjected to a
4
modified Wittig reaction by formation of the corresponding phosphonium salt, followed by reaction
1
3
with 4-(trifluoromethyl)benzaldehyde, giving the geometric isomers 33a and 33b. Saturation of the
double bond in 33a and/or 33b using ammonium formate and Pd/C as catalyst gave final compound
3
4 as racemic mixture.
Scheme
5:
Synthesis
of
(R,S)-N-(5-(4-(trifluoromethyl)benzyl)-5,6,7,8-tetrahydronaphthalen-2-
yl)heptanamide (34).
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
H
N
5
O
33a
H
H
N
H
H
NH2
Cl
N
N
+
-
CF3
NaBH4
1) (Ph) PH Br
NH4HCO2
O
DIPEA
O
3
5
5
+
O
O
2)
H
H
N
Pd/C
O
34
O
O
OH
5
30
31
32
5
F3C
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
CF3
33b
H
F3C
Effects of the retigabine derivatives on Kv7.2 channels expressed in CHO mammalian cells.
To assess the effects of these novel retigabine derivatives on Kv7.2 currents, and to identify
compounds more active than retigabine we performed whole-cell patch-clamp electrophysiological
experiments in mammalian CHO cells transiently transfected with Kv7.2 cDNA. The effect of the
synthesized compounds was investigated at a single concentration (10 µM), using a ramp protocol in
which Kv7.2 currents were activated by a 3-second voltage ramps from -80 to +20 mV. We used 10
µM of each compound given that this concentration exerts maximal efficacy in the case of retigabine,
the comparator drug (see below). Moreover, most drug screening efforts performed in search of novel
Kv7 modulators use a 10 μM compound concentration.14
Ramp-evoked currents were analysed by measuring the currents at -40 mV (a membrane potential
value close to the activation threshold) and at 0 mV (a membrane potential value at which Kv7.2
conductance is almost completely saturated), and the effect of each compound was expressed as the
ratio between current amplitude in the presence (A , B , I ) and absence (A , B , ICTL) of the drug
2
2
drug
1
1
at these two membrane voltages (Figure 2 a-b, Table S2). Measuring drug effects at both -40 mV and
mV membrane potential values allowed to formulate distinct predictions on the potential effects of
0
these compounds. In fact, the reference compound retigabine is described as a Kv7 activator because
it exerts two main effects on Kv7.2 channel function: one is a leftward shift in the activation gating,
the other being an increase in the maximal current at more depolarized potentials where conductance
is saturated. Therefore, drug-induced current enhancement at -40 mV, a membrane potential value
close to the current activation threshold, is likely to result in membrane hyperpolarization and
1
0
ACS Paragon Plus Environment

Page 11 of 96
Journal of Medicinal Chemistry
1
2
3
inhibition of action potential firing. On the other hand, drug-induced current enhancement at 0 mV,
may reduce sustained neuronal depolarization during trains of action potentials and cause spike
frequency adaptation. Therefore, measurements performed at both membrane potential values are
likely to predict different profiles of mechanism of actions of a Kv7 activator. Using this method,
drugs showing Idrug/ICTL >1 are considered as enhancers, drugs whose Idrug/ICTL is < 1 are blockers and
drugs whose Idrug/ICTL is = 1 are considered ineffective.
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. Effect of synthesized compounds on Kv7.2 current in comparison with retigabine. a-
b. Representation of the ratio between current amplitude in the presence (A , B ) and absence (A ,
2
2
1
B ) of each of the indicatedcompounds (10 µM) calculated at 0 mV (a) and at -40 mV (b). The inset
1
shows the protocol used to evaluate drug activity on Kv7.2 currents. Current scale, 400 pA; time
scale, 0.2 s. Histograms are colored according to the following scheme: blue represents available Kv7
activators, such as retigabine (RET) at 1 and 10 µM, Flupirtine (FLU) at 10 µM, and ICA27243
(ICA43) at 10 µM; red represents indoline derivatives, yellow represents indole derivatives and green
represents tetrahydronaphthalene derivatives. Each data point is expressed as the mean ± standard
error of the mean (SEM) of at least 3 cells recorded in at least 2 independent transfections; the precise
number of cells included in the analysis for each compound is specified in Table S2. * indicates values
significantly different (p < 0.05) from retigabine (10 µM).
1
5
The effect of each newly synthesized compounds was compared to that of retigabine. In this cellular
assay, retigabine showed a ratio Idrug/ICTL at -40 mV of about 6, whereas at 0 mV the Idrug/ICTL ratio
1
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
was about 2 (Figure 2, a-b), suggesting that retigabine is more active at negative potentials than
positive potentials, as previously described.15,16 In order to further validate our screening assay we
also evaluated the effects of a lower concentration (1 µM) of retigabine, as well as those of two
additional Kv7 activators, namely flupirtine (FLU) and ICA-27243 (ICA43), two compounds
showing lower and higher activity, respectively, than retigabine on Kv7.2 channels when the three
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
7
compounds were compared at the 10 µM concentration. The results obtained suggest that our
screening assay is able to discriminate compounds with different activity on Kv7.2 currents, both at
-40 mV and 0 mV (Figure 2).
Of the 42 compounds tested at the single concentration of 10 µM, seven (8a, 13a, 13d, 13g, 23a, 23c,
and 24a) were able to increase Kv7.2 currents more than retigabine (p<0.05) at 0 mV; among these,
only two (23a and 24a) also increased Kv7.2 currents more than retigabine at -40 mV (p<0.05)
(Figure 2 A-B). Given that activity at both -40 mV and 0 mV is predictive of a wider spectrum of
pharmacological effects on neuronal excitability needed for novel anticonvulsant/analgesic
compounds acting as Kv7 activators, 23a and 24a were selected for further analysis.
To compare the kinetics of onset and offset of the Kv7.2-activating effects and the potency of the two
most active compounds 23a and 24a with those of retigabine, time-course experiments at increasing
drug concentrations (0.01-10 µM) were performed. As shown in figure 3a and 3b, retigabine dose-
dependently increased Kv7.2 currents with a fast ON kinetics; indeed, upon application of 10 µM of
retigabine witha fast solution exchange apparatus, maximal steady-state levels were achieved in about
10 seconds. Similarly, currents returned to basal values after about 10 seconds upon retigabine
removal from the bath. By contrast, the kinetics of onset and offset of both 23a and 24a appeared
significantly slower. In particular, upon exposure to 10 µM 23a, more than 200 seconds were needed
to reach maximal steady-state current levels; little (if any) current recovery was observed upon 10
µM 23a removal from the perfusion solution for at least 200 seconds. Similar, although slightly faster
onset and recovery kinetics were observed upon exposure and removal, respectively, of 10 µM 24a.
It seems likely that the slower wash-in and wash-out kinetics shown by 23a and 24a when compared
1
2
ACS Paragon Plus Environment

Page 13 of 96
Journal of Medicinal Chemistry
1
2
3
to retigabine may be ascribed to the enhanced hydrophobicity of both newly-synthesized compounds,
whose logPs were 6.83 and 5.58, respectively, both values significantly higher than that of retigabine
(logP of 3.08). Given the higher hydrophobicity of 23a and 24a when compared to that of retigabine,
the self-aggregating propensity of all three compounds was further evaluated using pyrene
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
8
fluorescence spectroscopy measurements. The results obtained, shown in Figure S2, revealed that
the ratio of the pyrene fluorescence intensity peaks at 372 nm (I) and 383 nm (III) for 23a and 24a
decreased as a function of drug concentration (1-100 M), indicating an increase of hydrophobicity
in the microenvironment sensed by the probe; these data are consistent with the hypothesis that 23a
and 24a show the typical behaviour of amphiphilic self-assembling molecules forming
supramolecular aggregates (micelles) in the nanometric range. By contrast, no change in the pyrene
I/III fluorescence ratio was observed in the presence of retigabine. These findings suggest that the
slow ON/OFF kinetics of compounds 23a and 24a (with the respect to those of retigabine) may
correlate with their increased hydrophobicity. Although we don’t know yet whether this physico-
chemical property translates into significant changes in distribution and metabolism in vivo, and
whether it represents a potential hurdle for in vivo application, it should be mentioned that one of the
major limitation of retigabine is its rather poor brain penetration and short half-life, requiring 3 times-
7
a-day administration. Both these pharmacokinetic drawbacks might be improved by the increased
hydrophobicity shown by 23a and 24a.
The EC s of 23a and 24a on Kv7.2 current activation (calculated as described in the Methods section
5
0
and depicted in Fig. 3C) showed that 23a was about 12 times more potent than retigabine, while no
statistically-significant difference was observed between retigabine and 24a (Figure 3D).
1
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Effect of retigabine, 23a and 24a on ramp-evoked Kv7.2 current. a. Representative
whole-cell current traces from Kv7.2 channels activated by the indicated ramp protocol recordedin
control conditions and upon exposure to 0.01-10 µM retigabine (RET), 23a and 24a. b. Time courses
of Kv7.2 current potentiation by retigabine (RET), 23a and 24a. The steady-state amplitudes of the
peak currents at -40 mV were plotted as a function of time. The bars on top of the plot indicate the
duration of drug application. Current scale, 200 pA; time scale, 0.2 s. c. Protocol used to calculate
V (V -V ) at 20% of current peak value for controls (V ) and after RET exposure (V ). d. Dose-
2
1
1
2
response curve for RET-, 23a-, and 24a-induced V shift.
1
4
ACS Paragon Plus Environment

Page 15 of 96
Journal of Medicinal Chemistry
1
2
3
The voltage ramp-protocol used in the previously-described experiments did not allow us to precisely
define the effect of the drugs on the steady-state properties of the Kv7.2 current; therefore, in the next
experiments, we compared the effect of 0.03 µM retigabine and 23a on Kv7.2 channels activated by
constant voltage pulses (Figure 4 a-b). While 0.03 µM retigabine failed to enhance maximal Kv7.2
currents and to modify their voltage-dependence, the same concentration of 23a doubled Kv7.2
maximal conductance (Figure 4c) and caused a leftward shift of about 18 mV in the half-activation
voltage (V ) of Kv7.2 currents (Figure 4); in fact, the V vaues were -20.1±1.2 mV, -24.4±1.8 mV
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
/2
½
and -38.2±2.0 mV for control (Ctl), 0.03 M RET, and 0.03 M 23a, respectively (Figure 4d).
Figure 4. Effect of retigabine and 23a on Kv7.2 currents activated by constant voltage pulses.
a-b. Representative macroscopic current traces recorded from a CHO cell expressing Kv7.2
homomeric channels in response to the indicated voltage protocol before (left) and after (right)
application of 0.03 µM RET (a) or 23a (b). Current scale, 100 pA; time scale, 0.2 s. c. Kv7.2
current/voltage curves before and after application of 0.03 µM RET or 23a. d. Normalized Kv7.2
conductance/voltage curves before and after application of RET or 23a. Continuous lines represent
fits of the experimental data to a double Boltzmann distribution.
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Small molecules can activate Kv7 channels throught different molecular mechanisms. Among these,
retigabine activates Kv7.2 channels by recognizing a hydrophobic pocket located between the
cytoplasmic parts of the S5 and the S transmembrane domains in the open channel configuration;
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
within this cavity, an H-bond is established between the carbamate group of retigabine and the
aromatic side chain of a tryptophan present at the intracellular end of the S helix (W236 in the Kv7.2
5
sequence). Substitution of this tryptophan with smaller and less hydrophobic leucine largely prevents
the ability of retigabine to activate neuronal Kv7 subunits. 5b, 5c By contrast, a second group of
compounds, belonging to benzamides (ICA-27243 and ICA-069673) appears to recognize a poket
located in the voltage sensing domain (VSD; region between the S and the S transmembrane
1
4
1
9
domains). In particular, two specific aminoacid positions in the S segment of the Kv7.2 VSD (A181
3
and F168) seem to be relevant to confer current sensitivity to activation by benzamides.19b
To define the structural determinants for Kv7.2 current enhancement by 23a and 24a, and in particular
whether these molecules required the presence of W236 in S or of F168 in S , we examined the effect
5
3
of 23a and 24a on retigabine-insensitive Kv7.2 W236L and on ICA-insensitive Kv7.2 F168L mutant
channels. Application of 10 µM 23a or 24a, similarly to retigabine, failed to enhance Kv7.2 W236L
currents; whereas ICA-27243 strongly activated Kv7.2 W236L currents (Figure 5 a-b). By contrast,
application of 10 µM of retigabine, of 23a or of 24a increased Kv7.2 F168L currents, whereas ICA-
2
7243 was ineffective. These results strongly suggest that W236 exerts a critical influence on
retigabine, 23a and 24a binding within the pore hydrophobic pocket (Figure 5 c-d). Consistent with
this hypothesis, 23a (similarly to retigabine itself) failed to affect homomeric Kv7.1 currents where
the 236 position is naturally occupied by a leucine residue (Figure S3).
1
6
ACS Paragon Plus Environment

Page 17 of 96
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. The W236 residue in Kv7.2 exerts a critical influence on 23a and 24a abilty to activate
Kv7.2 currents. a-b. Representative traces of whole-cell ramp-evoked currents from Kv7.2 W236L
(a) and Kv7.2 F168L (b) channels upon applicatio of 10 µM retigabine (RET), ICA 27243 (ICA43),
23a and 24a. c. Quantification of the effects of the indicated compounds on Kv7.2, W236L and Kv7.2
F168L currents; data are expressed as the ratio between current amplitude at -40 mV in the presence
and absence of 10 µM drugs (IDRUG/ICTL). Each data point is expressed as the mean ± standard error
of the mean (SEM) of at least 3 cells recorded in at least 2 independent transfections. * indicates
values significantly different (p < 0.05) from respective controls.
Characterization of the selectivity profile for channels formed by different Kv7 subunits.
Considering that IKM is mainly formed by heteromeric assembly of Kv7.2 and Kv7.3 subunits, in
subsequent experiments we evaluated the activity of 23a and 24a on Kv7.2/Kv7.3 heteromeric
currents. Similarly to Kv7.2, 23a and 24a dose-dependently (0.01-10 µM) increased Kv7.2/Kv7.3
with both compounds showing higher potency when compared to retigabine; also, when compared to
retigabine, both 23a and 24a were more potent in activating homomeric Kv7.4 channels, whereas
1
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
only 24a was more potent than the reference compound in activating homomeric Kv7.5 channels
Table 1).
In was
(
conclusion,
the
selectivity
profile
for
23a
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Kv7.2=Kv7.4=Kv7.2/Kv7.3>Kv7.5>>Kv7.3>>Kv7.1; while for the 24a was Kv7.4>Kv7.5=
Kv7.2=Kv7.2/Kv7.3=Kv7.3>>Kv7.1.
Table 1. EC and induced V shift of retigabine and compounds 23a and 24a over different Kv7
5
0
channels.
EC50 (µM)

V (mV)a
23a
Target channel
Kv7.2
RET
24a
RET
23a
24a
-25.37±1.23
-29.47±2.03
-30.01±3.81
0.93±0.43
0.08±0.04$
0.63±0.07
(
12)
1.72±0.41
8)
(15-20)
(6-9)
0.21±0.14$
0.34±0.06$
Kv7.2/Kv7.3
Kv7.3
-26.81±1.93
-25.45±2.91
-20.00±2.48
(
(6-8)
10.5±6.00†
(7-9)
(9-10)
0.92±0.64
(6)
-39.88±5.13
-26.25±4.60
(30 µM)
-21.40±6.18*
(30 µM)
3.90±1.70
(6)
(30 µM)
Kv7.4
-15.19±1.97
-32.85±2.97**
-24.43±5.39
1.98±0.84
0.10±0.02$
(6-7)
0.08±0.20$
(6)
(
7)
0.80±0.24
3)
Kv7.5
-21.95±5.40
-25.6±4.9
-19.7±3.6
1.07±0.12
(4)
0.35±0.14
(4)
(
The number of recordings is indicated in brackets. a All three drugs were used at 10 µM concentration, unless otherwise indicated;
p<0.05 vs V of RET on Kv7.3 channels; ** p<0.05 vs V of RET on Kv7.4 channels; $ p<0.05 vs EC50 of RET; † p<0.05 vs EC50
*
of 23a on Kv7.2 channels
Molecular Modeling
We used pharmacophore modelling to get insight into the chemical features of our novel Kv7 channel
activators. In particular, using Phase 20 on an active set (composed by molecules showing more than
4
Idrug/Ictl at -40 mV) and on an inactive set (Table S3), we obtained the pharmacophore model A
(Figure 6a).
1
8
ACS Paragon Plus Environment

Page 19 of 96
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 6. Pharmacophore models developed on the basis of our novel Kv7 channel activators. Pharmacophore
features are color-coded as follows: Hydrophobic (HY – cyan), Aromatic (Ar – orange), Hydrogen Bond Donor
(HBD – green), Hydrogen Bond Acceptor (HBA – red). Panel a) Model A with compounds 24a and 8a
depicted in magenta and cyan sticks, respectively. Panel b) Model B with compound 24a depicted in magenta
sticks. Panel c) Model C with compound 8a depicted in cyan sticks. Panel d) Model D with compounds 24a
and 8a depicted in magenta and cyan sticks, respectively. Retigabine is depicted as reference, in every panel,
in green sticks.
The model is composed by 5 pharmacophore features, a hydrogen bond acceptor (HBA), a hydrogen
bond donor (HBD), a hydrophobic region (HY1) and two aromatic regions (Ar1 and Ar2). Despite
the evident influence of the R1 substituent (see figure 1) on the activity, model A does not present
any feature focused on these substituents. We thought this may be due to the different geometries of
the sulfonamide and amide moieties of the molecules belonging to the active set. We thus developed
two separate models for amides and sulfonamides. Removing the sulfonamide derivatives from the
active set we obtained model B (Figure 6b). On the opposite, by keeping sulfonamides only we
obtained model C (Figure 6c).
As evident from their superposition (Figure S4), the two models share features Ar1, Ar2, HY1, HBA
and HBD, with almost identical coordinates (RMSD=0.51Å). We superimposed models B and C to
1
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
model A to get model D, (Figure 6d) that inherits Ar1, Ar2, HY1, HBA and HBD from model A,
HY2 from model B and HY3 from model C. The hydrogen bond interaction with W236, whose
5
a
importance has been proven for other channel modulators as well as for 23a in this study, is
represented in the models by the HBA feature. The finding of the two additional hydrophobic
locations HY2 and HY3 could explain the great variety of substitutions tolerated at position 5, as well
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
as the possibility of linking the HY2/3 substituent at 5-NH with geometrically different linkers.
2
Indeed, despite the most potent molecules herein reported have an amide HBA-HBD moiety, the
sulfonamide one is of great interest for this and for future investigations. It can mimic the HBA-HBD-
Ar pattern of model D in different conformations and address the substituent at position 5 toward
HY2 or HY3, due to its larger HBA surface and to the rotation/energy profile of the bond connecting
HBA and HBD (Figure S1). For instance, compounds 8a and 23a are able to fit, in different
conformations, both HY2 and HY3 (Figure 7). The sulfonamide derivative docks the substituent
preferentially into HY3 (Figure 7b) but in a slightly higher energy conformation (ΔEintra= 1.49
kcal/mol) the substituent can be addressed towards HY2. (Figure 7a) On the opposite, the amide
compound 23a preferentially addresses its aliphatic chain toward HY2, (Figure 7a) but, in another
low energy conformation (ΔEintra= 1.10 kcal/mol), due to the flexibility of the n-hexane moiety, the
chain can fit into HY3. (Figure 7b).
2
0
ACS Paragon Plus Environment

Page 21 of 96
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7. Compounds 23a (green sticks) and 8a (cyan sticks) fitting feature HY2 (panel a) and HY3 (panel b).
Pharmacophore features are color-coded as follows: Hydrophobic (HY – cyan), Aromatic (Ar – orange),
Hydrogen Bond Donor (HBD – green), Hydrogen Bond Donor (HBD – green).
HY2 and HY3 could probably indicate a whole, larger and flexible pocket on Kv7.2, where more
flexible compounds could bind in different conformations, interacting with more protein residues.
This may be achieved by the flexibility of the substituent in the case of 23a, or of the sulfonamide
linker in the case of 8a. Homology modeling of the Kv7.2 219-325 region is in agreement with the
existence of two hydrophobic sites separated by W236, this critical residue appears indeed
surrounded, on two sides, by several hydrophobic residues that could be the counterpart of HY1 and
2
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
HY2/3 on the channel protein (Figure 8). Addressing the binding mode of our inhibitors, and their
optimization by structure-based studies, is our next goal, although previously reported studies 5a, 21
have already provided very interesting clues on the binding mode of retigabine and other Kv7
channels modulators.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. Kv modulators binding region modeled by homology with the CryoEM structure of Xenopus
KCNQ1 channel (PDB id: 5vms). W236 (magenta sticks) is surrounded by several hydrophobic residues (cyan
sticks) that could host chemical features HY1 and HY2/3.
2
2
ACS Paragon Plus Environment

Page 23 of 96
Journal of Medicinal Chemistry
1
2
3
Structure-activity relationship analysis
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Compounds 5a, 6, and 7 show the activity rank that would be expected on the basis of their potential
hydrogen bonding strength with the W236 side chain, with the urea moiety of 5a showing higher
efficacy than the guanidine of 7 and the thiourea of 6.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The compounds bearing non-branched substituents at 5-NH give clues about the HY2/3 feature.
2
While in the amide series a six membered chain gives an optimal activity (23a, 23b, 24b), in the
sulfonamide series, it does not (8e). Substitutions of a methylene group of 23a by nitrogen, as in
compounds 28a and 29a, increase hydrophilicity of the chain that would fall into HY2 and result
detrimental to the activity. Compounds with longer (25a) or bulkier (8f-8g) chains show a loss in
activity that could be due to steric clashes with the target protein. Even the substitution of the
cyclohexyl substituent of 8a with an aromatic ring (8b), is detrimental to the activity. Interesting SAR
clues are also evidenced for the substituents at N-1 position. Derivatization of the N-1 by acyl groups,
with its subsequent change in geometry and misfitting of HY1, is responsible for a markedly reduced
activation of the channel (compounds 16a and 16b). On the other side, N-1 benzyl derivatives are
among the most active compounds (compounds 8a, 23a-d and 24a-b), while the substitution of the
benzyl moiety with a phenylethyl group is not tolerated (compound 19). Both electron-withdrawal
and electron-donating groups were used to decorate N-1 benzyl moiety. In accordance with previous
reports, electron-withdrawal substituents exert positive effects on the channel activation (compounds
9
a
1
3g, 23a, 24a). Nevertheless, we found that also some electron-donating groups such as hydroxy
and alkoxy groups are suitable substituents in the para position of the N-1 benzyl moiety (compounds
3d, 23b, 23d and 24b). In addition, we found some differences in efficacy between the indoline and
1
the indole series when some specific para substituents were used to decorate the N-1 benzyl group.
For instance, the p-trifluoromethyl group is effective in improving efficacy for the indole series
(
compound 23a) while penalizing the activity for the indoline series (compound 13c). At the opposite,
the p-methoxy substituent is responsible for a remarkable increase in efficacy in the indoline series
compound 13b) when compared to fluorine, but decreases efficacy in the indole series (compound
(
2
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
3d). These differences are actually difficult to be explained by the data we collected so far; synthesis
of more derivatives and future target-based study could provide further clues. It can be speculated
that in our model, the N-1 benzyl moiety falls within the AR2/HY1 region, thus an electron
withdrawing group might be able to improve hydrophobic interactions in these regions. On the other
side, N-1 benzyl substituents, might be also involved in more specific interactions in the HY1 region,
such as hydrogen or halogen bonds, justifying the apparent conflicting results. However, the para
position appears to be the preferred position for the N-1 benzyl substituents. Multiple substitution in
different positions are well tolerated, but do not significantly influence efficacy (derivatives 13a-b,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
3c-e and 13g).
Retigabine and its conformationally restricted analogues 4a and 9a are able to fit six out of the seven
chemical features required by the pharmacophore model. Despite this, the activity of retigabine is
significantly higher. This may be explained by the lack of a primary amine in compounds 4a and 9a,
corresponding to the 2-NH of retigabine, that probably exerts supplementary interactions at the
2
receptor site that 4a and 9a do not properly compensate with additional hydrophobic interactions at
HY2/3 as the most potent compound of these series do.
Finally, tetrahydronaphthalene derivatives 33a, 33b and 34, were totally inactive when tested over
Kv7.2 channels. Considering that these molecules bear the substituents conferring enhanced efficacy
in the A and B chemiotypes, the lack of activity has to be attributed to the tetrahydronaphthalene
scaffold. This hypothesis is consistent with molecular modeling data showing that the different central
core and the double bond of 33a and 33b indeed hamper the fitting within the Ar2-HY1 region.
Photostability assays
Retigabine chemical photoinstability is a well-known characteristic of the molecule, likely leading to
some important adverse events observed during chronic long-term therapy, such as blue-gray
discololoration of skin, mucosae, and, possibly, retina8,9a,22. Two main impurities, corresponding to
C5-C5 and C5-N retigabine dimers, likely obtained by oxidative self-coupling of two quinone diimine
2
4
ACS Paragon Plus Environment

Page 25 of 96
Journal of Medicinal Chemistry
1
2
3
23
tautomers, have been foud in retigabine batches. Retigabine dimer formation, which is favoured in
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
3
aqueous media under UV radiation, is reponsible for the change in color of retigabine aqueous
solutions from colourless to red/dark blue when left on air, and for the previously mentioned retinal
and muco-cutaneous side effects.8,9a More recently, phenazinium, phenazine and benzimidazole
dimers of retigabine with high melanin affinity have been identified by matrix-assisted laser
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
4
desorption/ionization imaging mass spectrometry in corneal tissue. Given these premises, all
synthesized derivatives were subjected to photostability assays. Initially, an UV quartz cuvette
containing an aqueous solution of retigabine (100 g/mL) was irradiated in a black-box UV analyzer
at wavelenghts of 254 nm (UV-B) and 365 nm (UV-A) for 6 hours. At predetermined intervals (2, 4
and 6 hours) aliquots were withdrawn and analyzed by HPLC. The molecule showed high
photostability under these experimental conditions with negligible degradation after 6 hours (about
5
%, data not shown). By contrast, exposure to visible light led to a remarkable and significant
photodegradation of retigabine; as a matter of fact, about 47,33±4,03% of retigabine was degraded
after 6 hours of drug exposition to artificial daylight lamp. Quantitatively similar results were
obtained upon direct exposure of retigabine to natural solar light (43.5±4.3% degradation at 6 hours).
We then decided to further investigate the photodegradative behaviour of retigabine by analysing the
corresponding by-products. As shown in figure S5 the HPLC profile of photodegraded retigabine
batches is mainly characterized by the presence of two additional peaks with a retention time of 5.98
and 14.39 minutes. Running HPLC/MS analysis of these peaks we were able to detect the mass
spectra only for the late eluting peak (14.39 minutes), putatively identified as the retigabine adduct
ethyl (2,4-bis((4-fluorobenzyl)amino)phenyl)carbamate (Figure S6). The peak eluting at 5.98
minutes was undetectable by mass spectrometry, in accordance with a previous report showing the
2
2
occurrence of non-identified retigabine impurity at shorter retention times. Thus, we decided to
analyse the unknown peak using a GC/MS method, leading to the identification of 4-
fluorobenzaldehyde as the main by-product in retigabine photodegradation (Figure S7). The result
was further confirmed by comparison with the HPLC trace of pure 4-fluorobenzaldehyde (Figure S8)
2
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
and is consistent with a photooxidative C-N bond cleavage mechanism widely described in
2
5
literature. The proposed mechanism justifies the reason why visible light is more effective in
inducing retigabine photodegradation25c and corroborate the hypothesis concerning the molecular
structure of the peak eluting at 14.39 minutes. Considering that Groseclose and Castellino 24 found
that the typical absorbance of retigabine photoinduced dimers is 550 nm, we decided to set the same
wavelength for the UV detection, evidencing some additional byproducts with a retention time
between 7.00 and 10.00 minutes. HPLC/MS/MS analysis of the most abundant peak lead to the
putative identification of a dimer with a mass peak of 601.2357 m/z and a serie of MS/MS peaks
roughly corresponding to some of the previously reported phenazinium dimers (Figure S9).24
Thereafter, the most active synthesized derivatives, namely derivatives 8a, 13a, 13d, 23a and 24a,
were subjected to the photostability assay. Results obtained are reported in Table 2.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 2: Photoinduced degradation of retigabine and its analogues under solar lighting. Results are
expressed as percentage of degradation±SD
Time (h)
Derivative
2
4
6
Retigabine
9.65±6.93%
45.29±7.35%
33.46±6.50%
37.51±7.25%
1.36±0.89%
6.46±1.78%
0.96±0.64%
25.16±7.31%
86.87±10.54%
81.72±1.12%
84.66±4.12%
5.82±2.05%
15.78±1.97%
6.62±2.17%
47.33±4.03%
≥99%
8
a
1
3a
3d
23a
3c
24a
≥99%
1
≥99%
9.15±2.91%
23.11±1.84%
8.45±1.99%
2
Our results indicate that the N-(1-benzyl)indolin-5-yl)sulfonamide derivatives 8a, 13a and 13d are
even less stable than retigabine. At 6h degradation is almost complete and only traces of the starting
molecules are detectable by HPLC analysis. This is likely due to the sulfonyl activation of the p-
quinone diimine intermediate26 representing the starting product in the degradative pattern for
retigabine-like by-products. We perfomed HPLC/MS analysis for derivative 8a to further investigate
2
6
ACS Paragon Plus Environment

Page 27 of 96
Journal of Medicinal Chemistry
1
2
3
the photodegradation pattern for this series of indolin-5-yl sulfonamides (Figure S10), but no
significant mass peaks were identified. On the other side, GC/MS analysis revealed, also for 8a, the
presence of 4-fluorobenzaldehyde as the main photooxidation by-product, evidencing a similar
photogradative pathway with retigabine (Figure S10). Increased photostability was attained turning
to the N-(1-(4-fluorobenzyl)-1H-indol-5-yl)sulfonamide scaffold, as for 23c, probably due to the
improved delocalization of the -electrons of the indole ring. Finally, derivatives 23a and 24a,
obtained by replacement of the sulfonamide with a carboxamide moiety, were the most photostable
retigabine analogues, showing less than 10% of degradation after the 6 hours time-course.
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
In vitro metabolism
In vitro metabolism assays were performed using the S9 fraction of human liver microsomes as
model. Compounds 23a and 24a were selected for this study due to the potency and efficacy shown
as Kv 7.2 agonists and to their inherent photostability. The phase I metabolism of these derivatives
was initially investigated, in comparison with retigabine. In our assay, only a very small extent
(
3.11±0.21%) of retigabine undergoes phase I metabolism, a result consistent with available data.6
Instead, derivative 23a was found to undergo extensive (48.71±2.47 %) phase I metabolism. This is
27
likely due to the fact that indole moiety is naturally prone to cytochrome metabolism , and the long
and lipophilic side chain further decrease the metabolic stability of the compound. 28 In fact, when
the lipophilicity of the side chain is decreased, as for derivative 24a, an improved Phase I metabolic
stability is attained (17.30±1.74% of degradation). Next, retigabine and the more stable to Phase I
metabolism 24a underwent to UDP-glucuronsyltransferases (UGT) and N-acyl transferase (NAT)
activity evaluation. Retigabine was metabolized in 11.41±2.73% in UGT activity assays and in
12.68±1.94% in NAT activity assays, in agreement with the literature 12b, 29. Compound 24a displayed
increased stability in both assays, being metabolized by 7.42±1.81% in UGT activity assays and
3
.41±0.62% in NAT activity assays. Collectively, these results highlight a good in vitro metabolic
2
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 96
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
stability for compound 24a, especially during Phase II metabolism, making 24a a potential candidate
for in vivo metabolic and pharmacokinetic studies.
CONCLUSIONS
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Activation of Kv7 channels effectively suppresses neuronal hyperexcitability, a common feature of
different neuropsychiatric disorders such as epilepsy,30 neuropathic pain,31 amyotrophic lateral
sclerosis 32 and many others. For these reasons, identification of Kv7 channels modulators showing
improved pharmacokinetic and pharmacodynamic properties over existing compounds represents an
attractive research topic in medicinal chemistry; this issue is even more pressing in the light of the
recent market withdrawal of retigabine and flupirtine, the only two clinically approved drugs acting
as Kv7 activators. In fact, the analgesic flupirtine was found to cause drug-induced liver injury in rare
but fatal cases, whereas prolonged use of the antiepileptic retigabine led to muco-cutanous and retinal
discoloration. Several efforts have been previously made in search for the structure-activity
relationships regulating retigabine binding and actions at Kv7.2 and Kv7.3 channels. Results obtained
so far describe the retigabine binding site as a central hydrophilic core flanked by two hydrophobic
pockets. The most important chemical interaction evidenced is represented by a hydrogen bond
between the carbamate group of retigabine as donor and a specific, highly conserved, triptophan
residue (W265 in Kv7.3, W236 in Kv7.2) as acceptor. This H-bond has been considered as critically
responsible to confer retigabine its ability to control Kv7 channel gating 5a, 21, 34 and is the reason why
retigabine is inactive over cardiac Kv7.1 channels, in which this specific triptophan is substituted by
a leucine residue. Previous investigations also provided evidence that the use of electron-withdrawing
group in the para position of benzylamino substituents at C-4 of retigabine is able to increase potency
33
9a
and efficacy over Kv7.2/7.3 heterotetramers, while introduction of a fluorine atom in position 3 of
3
5
the benzene-1,2,4-triamine core modulated selectivity among Kv7 channel subtypes. Nevertheless,
a systematic investigation of the structural determinants required for Kv7 channel modulation is still
missing. In the present paper we have investigated the Kv7-opening ability of a series of constrained
2
8
ACS Paragon Plus Environment

Page 29 of 96
Journal of Medicinal Chemistry
1
2
3
retigabine analogues carrying specific substitutions in critical positions, thus generating a
pharmacophore model. Our results further highlight the pivotal role of the H-bonding with W236 in
Kv7.2 channel activation, but also reveal the possibility of using H-bond donor groups different from
carbamate, such as amide and sulfonamide. It is worth of note that the replacement of the benzyl
group at the N-4 position of retigabine (N-1 position in our derivatives) is hardly tolerated. When
different substituents are used, the efficacy of the compound is always decreased in our series.
Moreover, we found that the N-1 benzyl moiety tolerate multiple substitutions, regardless of the
electron-withdrawing or electron-donating properties of the substituents. Previous results have led to
the hypothesis of the existence of a narrow hydrophobic pocket accommodating the N-1 benzyl
moiety exerting a specific interaction in the para-position. In addition, by the present data we may
suppose the existence of a larger hydrophobic pocket (HY2 and HY3 region of the pharmacophore
model), able to accommodate a wide range of hydrophobic substituents. Notably, some substituents
at the position C-1 of retigabine (corresponding to the C-5 position in our derivatives) were also
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
9
a
studied by Kumar et al., who demonstrated that while lipophilic isopropyl chain improved potency
by about 2/3 times, more hydrophilic substituents were not tolerated, suggesting the existence of a
hydrophobic pocket. In agreement with this hypothesis, our results also highlight that changes in
lipophilicity at C-5 also modify the kinetics of onset and offset of Kv7.2 channel activation; in fact,
when compared to retigabine (log P=3.08), more lipophilic 23a (logP=6.83) and 24a (logP=5.58)
both displayed slower kinetics. These results are in good agreement with recently published studies
3
6
concerning flupirtine analogues. Mostly important, we found that the type of substituents in position
C-5 of the indole series influences selectivity between Kv7.2 and Kv7.3 channels. In particular, long
and flexible carbon chains as in derivative 23a are required for Kv7.2 selectivity. Furthermore, both
indole derivatives 23a and 24a showed selectivity for Kv7.4 channels. Finally, we found that some
of the most active derivatives show a dramatic improvement in photostability in comparison with
retigabine. Further investigations are needed to optimize metabolic properties of synthesized
2
9
ACS Paragon Plus Environment