Flupirtine and retigabine as templates for ligand-based drug design of KV7.2/3 activators

Article abstract of DOI:10.1039/c9ob00511k

Drug induced liver injury (DILI) and tissue discoloration led to the recent discontinuation of the therapeutic use of the closely related drugs flupirtine and retigabine, respectively. Experience gained with these drugs strongly suggests that heterotetramer, voltage-gated potassium channels 2 and 3 (K_V7.2/3) are valid targets for effective treatment of pain and epilepsy. Because the adverse effects are not related to the mechanism of action, it appears promising to investigate chemical modifications of these clinically validated, drug-like leads. In the present retro-metabolic drug design study, a series of 43 compounds were synthesized and characterized with regard to K_V7.2/3 opening activity and efficacy. The most active compound 22d displays excellent potency (EC₅₀ = 4 nM) and efficacy (154%) as a K_V7.2/3 opener. Limited aqueous solubility hampered toxicity testing at concentrations higher than 63 μM, but this concentration was nontoxic to two hepatocellular cell lines (HEP-G2 and TAMH) in culture. The slightly less active but more soluble compound 25b (EC₅₀ = 11 nM, efficacy 111%) showed an improved toxicity/activity ratio compared to flupirtine by three orders of magnitude and represents an attractive lead structure for the development of safer analgesics and antiepileptics.

Full text of DOI:10.1039/c9ob00511k

View Article Online
View Journal
Organic &
Biomolecular
Chemistry
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: A. S. Surur, C.
Bock, K. Beirow, K. Wurm, L. Schulig, M. K. Kindermann, W. Siegmund, P. J. Bednarski and A. Link, Org.
Biomol. Chem., 2019, DOI: 10.1039/C9OB00511K.
This is an Accepted Manuscript, which has been through the
Volume 14 Number
1 7 January 2016 Pages 1–372
Royal Society of Chemistry peer review process and has been
accepted for publication.
Organic &
Biomolecular
Chemistry
www.rsc.org/obc
Accepted Manuscripts are published online shortly after
acceptance, before technical editing, formatting and proof reading.
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the
author guidelines.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
standard Terms & Conditions and the ethical guidelines, outlined
in our author and reviewer resource centre, still apply. In no
event shall the Royal Society of Chemistry be held responsible
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.
ISSN 1477-0520
COMMUNICATION
Takeharu Haino et al.
Solvent-induced emission of organogels based on tris(phenylisoxazolyl)
benzene
rsc.li/obc

Please do not adjust margins
Organic & Biomolecular Chemistry
Page 1 of 11
View Article Online
DOI: 10.1039/C9OB00511K
Organic & Biomolecular Chemistry
ARTICLE
Flupirtine and retigabine as templates for ligand-based drug
design of KV7.2/3 activators
a
a
a
a
a
Abdrrahman Shemsu Surur , Christian Bock , Kristin Beirow , Konrad Wurm , Lukas Schulig ,
Markus K. Kindermann a, Werner Siegmund b, Patrick J. Bednarski a, and Andreas Link* a
Received 00th January 20xx,
Accepted 00th January 20xx
DOI: 10.1039/x0xx00000x
Drug induced liver injury (DILI) and tissue discoloration led to the recent discontinuation of the therapeutic use of the closely
related drugs flupirtine and retigabine, respectively. Experience gained with these drugs strongly suggests that heterodimer,
voltage-gated potassium channels 2 and 3 (KV2/3) are valid targets for effective treatment of pain and epilepsy. Because the
adverse effects are not related to the mechanism of action, it appears promising to investigate chemical modifications of
these clinically validated, drug-like leads. In the present retro metabolic drug design study, a series of 43 compounds were
synthesized and characterized with regards to KV7.2/3 opening activity and efficacy. The most active compounds 22d displays
excellent potency (EC50 = 4 nM) and efficacy (154%) as an KV7.2/3 opener. Limited aqueous solubility hampered toxicity
testing at concentrations higher than 63 µM, but this concentration was nontoxic to two hepatocellular cell lines (HEP-G2
and TAMH) in culture. The slightly less active but better soluble compound 25b (EC50 = 11 nM, efficacy 111%) showed an
improved toxicity/activity ratio compared to flupirtine by three orders of magnitude and represents an attractive lead
structure for the development of safer analgesics and antiepileptics.
www.rsc.org/
Voltage-gated “M-type” (KCNQ, KV7) K+ channels are low- with 1 excluded such an investigation.9 Last year, the European
threshold channels that regulate spike-frequency adaptation, medical agency EMA recommended the withdrawal of 1 from
bursting, and hyper excitatory states.1,2 KV7 channels are the market.10
composed of a homomeric or heteromeric assembly of
subunits. The heterodimer of KV7.2 and KV7.3 is the most
abundant and a predominant contributor of the so-called M-
current (IM).3 Recently, attention has focused on M-channels
because neural dysfunction is considered a causative factor in
diseases with unmet medical needs including tinnitus, neonatal
epilepsy and pain. Chronic tinnitus affects 5-15% of the world
population and channelopathy of KV7.2/3 is considered a main
cause in the development of tinnitus.4 Concerning epilepsy, a
25% reduction in Kv7.2/3 appears to shift neuronal excitability
to epileptogenic levels in early infancy.5 An underlying feature
of neuropathic pain is sustained excitability of nociceptive
sensory fibres, where IM is one of the major factors modifying
tonic excitability of neurons.6,7
Flupirtine (1) is an atypical analgesic that alleviates pain with a
mode of action divergent from NSAIDs and opioids, identified as
KV7.2/3 opener post hoc (Figure 1). Moreover, recent results
indicated 1 as an extremely effective treatment option for
neonatal seizures in rats and provide the basis for a possible trial
in newborn humans.8 However, rare but severe adverse drug
reactions, namely drug induced liver injury (DILI) associated
a. Pharmaceutical and Medicinal Chemistry, Institute of Pharmacy, University
Figure 1: Chemical structures of flupirtine (1), retigabine (2), elusive metabolites (3a-d
of Greifswald, Friedrich-Ludwig-Jahn-Str. 17, 17487 Greifswald, Germany. E-
and 4a-d), dimers (5e-h) of 3c,d, and 4c,d, and hypothetical, analogous but improbable
dimers (5a-d) of 3a,b, 4a,b.
mail: link@uni-greifswald.de.
b. Center of Drug Absorption and Transport (C_DAT) Greifswald, Germany.
† Footnotes relating to the title and/or authors should appear here.
Electronic Supplementary Information (ESI) available: [details of any
supplementary information available should be included here]. See
The precise nature behind the molecular cause of DILI in
patients treated with 1 is still unknown, but evidence is
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 1
Please do not adjust margins

Please do not adjust margins
Organic & Biomolecular Chemistry
Page 2 of 11
ARTICLE
Journal Name
mounting that 1 is not the noxious agent as such but can form oxidation while maintaining and even improving oVnietwhAertiKclVe7O.n2li/n3e
toxic metabolites in patients.11 For instance, cysteine- and N- opening activity.18 For example, in a previous communication
DOI: 10.1039/C9OB00511K
acetyl cysteine-conjugates of 1 were identified in human urine we reported that thioether analogues of flupirtine, which are
(not shown), evidence for the formation of reactive, more difficult to oxidize than flupirtine and retigabine, are still
electrophilic intermediates.12 This led to the hypothesis that effective as KV7.2/3 channel opening agents.19 In the present
oxidation of 1 could result in the formation of highly reactive work, we have used classical medicinal chemistry strategies like
and thus elusive para- or ortho-azaquinone diimines (3a and 4a) variation of stereo-electronic effects and extension or
in vivo. Nucleophilic attack of the azaquinone diamine by alteration of substitution patterns to design new and improved
glutathione followed by the metabolic transformation to KV7.2/3 channels openers that are more resistant to oxidation
cysteine or N-acetyl cysteine conjugates (i.e., mercapturic acids) than the clinically validated lead structures 1 and 2. It was
would explain the presence of these metabolites in urine of envisioned that by avoiding an increase in the number of atoms,
some patients. In analogy to the notorious N- the desirable biopharmaceutical properties of 1 and 2 such as
acetylbenzoquinoneimine (NAPQI) formed after overdosing orally bioavailability could be maintained.
with acetaminophen, the azaquinone diimine metabolites 3a Here we describe the synthesis, physicochemical
and 4a may affect both the pharmacological activity and liver characterization and evaluation as openers of Kv7.2/3 channels
injury of 1.
43 drug-like analogues, some of which show dramatic increases
Structural optimizations of 1 led to the development of the in channel opening potency. A reported relationship between
desazaflupirtine derivative retigabine (2) with improved anti- the oxidation potential and activity of flupirtine analogues was
seizure efficacy in rat maximal electroshock seizure test, reinvestigated, and the factor hydrophobicity was also
marketed as Trobalt®. Retigabine (2) is a unique anti-epileptic investigated.
due to the novel mechanism of action, which created optimism Design cycle one: The commencement of the ligand design
to be able to tackle epileptogenesis. In addition, 2 was found to envisioned easily oxidizable analogues to test the previous
prevent the behavioural signs of tinnitus in mice. Unfortunately, hypothesis that oxidizability at lower potential is a requirement
2 was reported to cause bluish pigmentation of ocular tissues, for KV7.2/7.3 channel opening activity of flupirtine and
fingernails, toenails, and skin around eye and lips. The use of 2 analogues.20 At the same time, large increases in lipophilicity of
thus came to halt eight years after approval.13–16 While the the derivatives was initially regarded as undesirable. Analogues
molecular basis of the observed bluish pigmentation is not 7a-g were designed in this fashion, which all have a different
completely understood, a very recent study revealed a substituent in place of the 4-fluorobenzyl amino substituent of
conserved distribution of 2 and dimers to melanin-containing 1. The synthesis of these compounds is outlined below (Scheme
ocular tissues and proposed these dimers as causative agents of 1).
this adverse effect.17 The oxidation of 2 to quinone diimines 3c In addition to the anodic peak potentials (Epa) that characterize
and 4c and its N-acetylated decarbethoxy metabolite to 3d and the ease of oxidation of analogues 7a-g, lipophilicity was also
4d could be the initial step in a cascade of reactions that investigated as a critical physicochemical property (Table 1).
proceed only in melanin-containing tissues.
With the exclusion of 7f, which was more resistant to oxidation
than 1, all other analogues turned out to be oxidizable at lower
potentials than 1 (Ep.a. = 350 mV). Lipophilicity was assessed as
the LogD7.4 by using a reversed-phase HPLC method (see
supporting material), which is useful for comparing close
structural analogues.
Results and discussion
Since metabolic oxidation of both flupirtine and retigabine are
likely to play a central role in their respective toxicities, we have
been working to identify analogues that lack the capacity of
O
O
O
N
O
N
O
N
NH
O
O
O
a
b
c
Cl
N
Cl
Cl
N
NH2
N
NH2
N
NH2
5
6a-k
7a-k
HN
HN
HN
O
N
H
N
HN
O
HN
N
HN
N
N
N
O
N
O
HO
F
F
F
F
7a
7b
7c
7d
7e
7f
7g
7h
7i
7j
7k
Scheme 1. Synthesis of analogues 7a-k; a) aq. NH3 (25%), 2-propanol, 35 °C, 5 days; b) 6a & 6i: amine reagent, triethylamine, 2-propanol, reflux, overnight-3 days; for
6b-h: amine reagents, triethylamine, 2-propanol, microwave (110 - 120 °C, 15 - 30 min); c) 7a-j: H2, Pd/C, 2-propanol/pyridine, rt, 2.5 – 28.5 h, then triethylamine (only
7h-i), acylating agent, 0 °C - rt, 2 – 4 h; 7k: iron powder, NH4Cl, 2-propanol/water, reflux, 4 h, then triethylamine, acylating agent, 0 °C - rt, 2 h.
2 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 3 of 11
Organic & Biomolecular Chemistry
Please do not adjust margins
Journal Name
ARTICLE
Table 1: Synthesized analogues: anodic peak potential (Epa), potency, efficacy and toxicity
View Article Online
DOI: 10.1039/C9OB00511K
Entry
1
2
7a
7b
7c
7d
7e
7f
7g
7h
7i
7j
Epa [mV][a]
350
205
286
331
321
306
346
381
346
208
217
467
412
499
504
494
238
230
316
351
266
351
386
371
386
366
385
255
220
281
230
220
225
227
232
197
235
206
210
243
422
316
253
373
222
LogD7.4
2.96
2.57
3.12
1.68
2.02
2.80
1.22
1.74
1.41
3.23
2.80
2.62
3.26
3.21
n.d.
3.19
4.73
3.08
3.76
5.36
3.74
2.49
1.82
3.21
3.21
2.64
3.44
3.62
3.30
3.58
3.71
2.38
3.53
3.30
3.58
3.11
3.98
3.86
3.62
2.80
4.32
4.02
4.29
3.58
n.d.
EC50 [M][b]
0.918
0.147
> 10
Efficacy [%]
LD25 [M][c]
103
> 400
> 500
> 500
> 1000
505
> 1000
> 1000
> 1000
134
Tox/Act.[c,d]
LD25 [M][e]
Tox/Act.[d,e]
100
134
-
-
-
-
-
-
-
-
-
-
-
112
74
269
172
> 500
> 1000
330
> 1000
> 1000
> 1000
128
99
> 20
> 125
198
139
76
123
> 40
60
n.d. [f]
94
220
144
66
42
81
1,830
-
-
-
-
-
-
-
-
-
-
-
184
68
13
-
-
-
-
-
> 10
> 10
> 10
> 10
> 10
> 10
> 10
> 10
> 10
> 10
1.074
2.039
5.688
> 10
-
-
-
-
-
-
163
> 20
117
121
125
144
15
> 40
66
n.d.[f]
164
111
> 500
147
62
500
> 16
> 31
> 63
68
> 63
> 500
> 16
25
7k
-
113
61
25
-
11a
11b
11c
11d
11e
11f
11g
11h
12a
12b
12c
12d
12e
12f
12g
12h
12i
12j
12k
12l
16a
16b
21a
21b
22a
22b
22c
22d
22e
25a
25b
25c
118
124
92
-
> 10
-
-
-
0.133
1.865
0.109
3.26
> 10
> 10
> 10
12.92
2.13
0.36
0.30
129
121
119
125
-
496
n.d. [f]
1505
34
-
-
-
39
451
n.d. [f]
862
68
-
-
-
39
-
-
84
75
134
116
119
-
500
> 16
> 31
30
100
2,391
-
-
-
0.023
> 10
> 10
2,956
-
-
-
385
4211
-
55
> 63
> 500
> 16
45
> 32
> 125
71
-
-
> 10
0.065
0.0038
> 10
0.073
> 10
104
115
-
155
-
692
16
> 125
> 125
49
56
> 125
9
28
> 15
212
-
973
-
-
-
-
45
> 10
-
> 63
> 125
51
54
> 30
231
4.814
0.0042
0.0037
0.026
0.0113
5.817
81
154
165
125
111
160
2,143
7,568
12,143
14,594
18,761
20,442
> 50
> 100
[a] Determined with 1.0 mM compound in 100 mM TRIS-buffer (pH 7.4); [b] EC50- and LD25- are means of 4–5 independent determinations; [c] determined using TAMH
cells; [d] LD25divided by EC50; [e] determined using HEP-G2 cells;[f] displayed abnormal hepatotoxicity profile (see supporting information).
The physicochemical parameter LogD7.4 was quantified because pharmacophore and thus important for activity, we tested this
of its general importance for activity and toxicity. The common assumption with molecules 7h-k. The extended and shortened
feature among analogues 7a-g is the exchange of the aromatic analogues, 7h and 7i were synthesized for this purpose. For the
4-fluorophenyl substituent for less planar, sp3-rich alicyclic amide and ether analogues 7j and 7k, respectively, additional
moieties. As planned, analogues 7a-g displayed lower LogD7.4 synthetic steps not described in scheme 1 were necessary (for
values.
details see chemical synthesis). Interestingly, none of the
Analogues 7h-k differ from 1 by variation of the methylamino compounds of this design cycle possessed KV7-channel opening
linker while retaining the rest of the molecular scaffold. Because activity.
the methylamino group has been thought to be part of the core
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 3
Please do not adjust margins

Please do not adjust margins
Organic & Biomolecular Chemistry
Page 4 of 11
ARTICLE
Journal Name
Design cycle two: Primary aromatic amines may affect the (Figure 2). Keto-enol tautomerization is a likely exVpielwanAartitciloe nOnflioner
DOI: 10.1039/C9OB00511K
activity of a drug molecule positively owing to possible this deviation in redox behavior.
hydrogen bonding or other intermolecular interactions with the
active site. The stable conformer of 1 generated in silico displays
NO2
OH
NO2
OH
NO2
O
an intramolecular hydrogen bond between the primary amino
group and the carbonyl oxygen, which may be important for
sterical pre-organization and binding to the target.19 This idea is
supported by a report that removal of the primary amino group
from 1 is detrimental to activity.20 On the other hand, primary
aromatic amines can result in the formation of toxic metabolites
such as hydroxyl amines and quinone imines. Likewise, a
reactive hydroperoxide metabolite of 1 has been reported.21
Therefore, it was important to know if the primary amino
groups of 1 and 2 are part of the pharmacophore or not. Various
strategies can be used to assess the pharmacological
significance of such primary amino groups. Methyl groups were
used to hinder hydrogen bonding by replacing the amino for this
group, which has a similar steric demand but is incapable of
hydrogen bonding (11a-11c) (Scheme 2). Secondary and tertiary
amines (11d and 11h) as well as the corresponding phenol and
methoxy congeners 11e-11g were also evaluated. Surprisingly,
11a retained complete activity compared to 1 (Table 1)
although it was previously known that omission of a substituent
at this position leads to a loss of activity.20 Thus, replacement
strategies seem promising. The methyl substituent is suited to
avoid the formation of a hydroxyl amine species and could as
well hamper formation of electrophiles 4a-b, minimizing
possible hepatotoxicity associated to these reactive
metabolites. The slightly diminished in vitro hepatotoxicity of
11a may be explained by restricting the formation of ortho-
quinone diimines 4a-b; the residual toxicity of 11a could be
attributed to the formation of para-quinone diimines 3a-b, as
this oxidation is not blocked by methylation.
c
d
Cl
N
N
H
N
N
H
N
F
F
9a
10f
10g
e
b
O
O
NO2
R
NO2
R
NH
R
a
c
e
Cl
N
N
H
N
N
N
H
5
R = NH2
R1
R1
8a R = NHCH3
10a-e
11a-h
8b R = OCH(CH3)2
11a R = CH3
11b R = CH3
11c R = CH3
11d R = NHCH3
11e R = OH
R1 = 4-F
R1 = H
R1 = 3-OCH3
R1 = 4-F
R1 = 4-F
R1 = 4-F
c
NO2
CH3
Br
N
11f R = OCH3
11g R = OCH(CH3)2 R1 = 4-F
11h R = morpholino R1 = 4-F
Scheme 2: Synthesis of 11a-h; a) for 5: see scheme 1; for 8a: methylamine,
triethylamine, acetonitrile, 0 °C, 40 min, then rt, 30 min – 3 h; for 8b: 2-propanol,
NaH, toluene (anhydrous), argon, 0 °C, then rt, overnight; b) 9a: conc. H2SO4,
NaNO2 (aq.), stepwise addition; c) for 10a-d: amine reagents, triethylamine, 2-
propanol, microwave (120 °C, 15
– 30 min); for 10e-f: amine reagent,
triethylamine, 2-propanol, reflux, overnight; d) 10g: K2CO3, dimethylformamide
(anhydrous), iodomethane, rt, 2 h; e) for 11a-d: Pd/C, H2, 2-propanol, rt, 4 h -
overnight, then triethylamine, ethyl chloroformate, rt, 2 - 5 h; for 11e-f: Pd/C,
pyridine, H2, rt, 3.5 - 22.5 h, then 0 °C, ethyl chloroformate, 2.5 - 3 h; for 11g: iron
powder, NH4Cl, 2-propanol/H2O, reflux,
chloroformate, 0 °C, 3 h.
3
h, then triethylamine, ethyl
4.0
3.0
1
11e
25b
2.0
1.0
0.0
In addition to 11a, 2-deamino-2-methyl-flupirtines 11b-c also
have the amino group replaced by methyl but on top the
electronic features of the benzylamine moiety have been
modified. In 11b, the 4-fluoro substituent was omitted and
replaced with an electron donating methoxy group in 11c. Both
analogues 11b and c were found to be active but less so than
11a. To gain further insight into steric factors, analogues 11e-h
were synthesized. Compound 11d contains a methylated,
secondary amino group in position 2. This minimal extension
abolished the activity and increased the hepatotoxic property in
TAMH cells, which is in line with the notion that the formation
of quinone diimines is not suppressed by this modification. The
loss of activity however was surprising because a far greater
modification to a tertiary dimethylamino group reportedly
preserved activity.20 In the case of 11e, the primary aromatic
amino group was exchanged for a phenolic hydroxyl group.
Such a moiety manifests both hydrogen bond acceptor and
donor characteristics. In contrast to 11a, 11e was found
inactive. This observation undermines the idea that hydrogen
bonding by the primary amino group of 1 is essential. Moreover,
inactivity and ease of oxidation of 11e indicated that a low
oxidation potential is not predictive of KV7.2/3 opening activity
(Figure 2). Inspection of the cyclovoltammogram of 11e reveals
the unique absence of reduction signals in the reverse sweeps
-1.0
-2.0
-0.5
0
0.5
1
Voltage (V)
Figure 2: Cyclovoltammograms of 1, 11e and 25b
N
H
N
NH2
F
12a-i
NH2
NO2
H2N
NH2
NH2
O
O
O
O
O
O
NH
12a
NH
12b
NH
12c
NH
NH
NH
12f
12d
12e
F
F
F
NH
NH
NH
NH
O
O
O
O
O
S
NH
NH
N
N
O
O
12g
12h
12i
12j
12k
12l
Figure 3: The chemical structures of 12a-i
4 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 5 of 11
Organic & Biomolecular Chemistry
Please do not adjust margins
Journal Name
ARTICLE
Design cycle three: The next strategy was introduction of either
smaller or bulkier substituents in lieu of the
O
O
View Article Online
DOI: 10.1039/C9OB00511K
NH
N
NH2
O
O
ethoxycarbonylamino side chain (Figure 3).
A shorter
a
b
N
H
N
H
N
aminomethylcarbonylamino chain in 12b proved detrimental to
activity. The introduction of additional phenyl rings in both
stereochemically possible orientations (R)- in 12c and (S)- in 12d
likewise resulted in a loss of activity. In fact, 12d displayed
higher hepatotoxicity than 1. The 4-(amino)benzamido
derivative 12e opened KV7.2/3 channels while showing a low
hepatotoxicity profile in TAMH cells. The corresponding 4-nitro
congener 12f increased the channel opening activity further but
also led to increased hepatotoxicity. The urea analogue 12g
exhibited a greater than two-fold increase in potency compared
to 1 but the liver cell toxicity could not be excluded above
31 µM.
F
O
O
F
O
13
14
c
O
NH
NH
N
O
d
N
H
NH2
N
H
F
F
O
15a-b
16a-b
F
F
F
Previous work has shown that by remodeling the ethoxy residue
of 1 with fluoroaromatic substituents can lead to increases in
activity.22,23 In our case, the 4-fluorophenyl substituent in 12h
led to slightly improved activity and hepatotoxicity properties
while the 3,5-difluorobenzyl substituent in 12i led to a clear
boost in potency (EC50 = 23 nM).
a
b
Scheme 3: Synthesis of 16a-b; a) 13: 2, phthalic anhydride, toluene, reflux, 6 h; b)
14: hydrobromic acid (32% in acetic acid), rt, 64 h; c) 15a-b: carboxylic acid, HATU;
diisopropylethylamine, dimethylformamide, rt - 40 °C, 6 - 18 h; d) 16a-b: hydrazine
hydrate , methanol/2-propanol, rt, 17 h
NO2
Cl
NO2
Cl
NO2
Cl
NO2
R
Analogues lacking a carbonyl group in the side chain were
synthesized to judge the importance of the carbonyl moiety of
1. This was the case with compound 12j, which was suspected
responsible for inactivity. This idea is rationalized as the
carbonyl oxygen of 2 has been implicated in hydrogen bond
formation with the indole NH of Trp265.24 Trp265 is considered
crucial as its repositioning completely abolished the activity of
2. Analogue 12k has a cyclic side chain with no hydrogen bond
donor and showed no hepatotoxicity up to 500 µM but no
activity as well. Therefore, a phenyl ring was annulated in 12l to
increase the bulkiness of the dioxopyrrolidinyl substituent of
12k, anticipating a shift towards activity while preserving the
better hepatotoxicity profile. Compound 12l, however, was
devoid of activity. The cyclization in 12k-l and the resulting loss
of hydrogen bond donating –NH group may in part be
responsible for the inactivity. Lemmerhirt et. al. reported a
similar loss of activity following a methyl substitution of the
ethoxycarbonylamino side chain.20 Thus, the secondary amide
group was then preserved in the rest of the analogues described
here.
It seemed worthwhile to explore alterations in the
ethoxycarbonylamino side chain of 2 with substituents that
have already proven to be beneficial for KV7.2/3 activity.
Subsequently, 4-fluorophenyl (16a) and 3,5-difluorobenzyl
(16b) residues were selected (Scheme 3). Compared to 1 or 2,
both analogues uphold activity by several folds. The 3,5-
difluorobenzyl residue in 16b (EC50 = 3.8 nM) evidently resulted
in a surge in activity. The LogD7.4 of 16a and 16b are 3.30 and
3.58, respectively, which indicates the substituents did not alter
the hydrophobicity of the analogues to a great extent.
Design cycle four: cysteine- and N-acetylcysteine conjugates of
1 were isolated from human urine, as mentioned earlier.12 The
conjugations occurred at C-5 of pyridine and 21a was
synthesized by introducing a methyl substituent at the exact
position (Scheme 4).
a
b
c
N
N
Cl
N
Cl
N
O
R = NH2
17
18
19a
19b
R = morpholino
d
O
O
NH
R
NO2
R
e
N
H
N
N
H
N
F
F
R = NH2
R = NH2
20a
21a
21b
R = morpholino
R = morpholino
20b
Scheme 4: Synthesis of 21a-b; 17: urea-hydrogen peroxide, trifluoroacetic
anhydride, dichloromethane, ice-cold water bath, 30 min, then rt, 36 h; b) 18:
phosphorous oxychloride, reflux, 5 h; c) for 19a: aq. NH3 (25%), 2-propanol, 35 °C,
5 days; for 19b: amine reagents, triethylamine, acetonitrile, 0 °C, 40 min, then rt,
3 h; d) for 20a: amine reagents, triethylamine, dimethyl sulfoxide, microwave (100
°C, 2.5 h); for 20b: amine reagents, triethylamine, 2-propanol, reflux, 24 h; e) for
21a-b: Pd/C, H2, 2-propanol, rt - 40 °C, overnight - 24 h, then triethylamine,
acylating agent, rt, 1 h - until completion.
A methyl substituent in position C-5 was considered to act as
metabolic blocker to avoid the formation of conjugates and
perhaps indirectly prevent the liver injury caused by 1.
However, 21a was found inactive and offers no improvement in
toxicity compared to the hepatotoxicity of 1. The primary amino
group of 21a was exchanged for a morpholine ring, yielding 21b,
which showed very promising biological activity. Analogue 21b
(EC50 = 73 nM), together with 11h, demonstrated a positive
impact of a morpholino substituent at position 2 of the central
ring.
Design cycle five: the last strategy was to consolidate
substituents that increases in activity into one structure. The
ethoxycarbonylamino side chain of 7a was replaced with the
benzyloxycarbonylamino and 3,4-difluorobenzoylamino side
chains, respectively, in the corresponding derivatives 22a and
22b (Figure 4) Both oxidize at lower CV potentials but exhibited
no activity.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 5
Please do not adjust margins

Please do not adjust margins
Organic & Biomolecular Chemistry
Page 6 of 11
ARTICLE
Journal Name
limiting factor for the evaluation of hepatotoxicity since the
highest concentration that could be reached was only 15 M.
F
F
View Article Online
DOI: 10.1039/C9OB00511K
F
F
O
O
O
O
O
R1
NH
NH
NH
NH
N
R
11h R = 4-fluorobenzyl R1 = ethoxyl
N
H
N
NH2
N
H
N
NH2
N
H
N
NH2
N
H
N
25a R = cyclohexyl
R1 = 3,5-difluorobenzyl
O
22a
22b
22c
c
O
O
NO2
N
NH
O
NH
R
24b R = 4-fluorobenzyl
24a R = cyclohexyl
N
H
N
N
H
N
N
H
N
O
O
F
F
22d
22e
b
NO2
Cl
NO2
N
NO2
N
+
a
Figure 4: The chemical structure of analogues 22a-e
Cl
N
Cl
N
N
N
O
O
O
23a
23b
The analogue 22c was synthesized as an analogue of inactive
22b, whereby the cyclohexylamino moiety was replaced by a
smaller cyclopropylamino group. Unpredictably, 22c was the
first active analogue identified with an alicyclic moiety in place
of the 4-fluorobenzylamino substituent. Consequently, doubt
surfaced that the early generalization that the 4-fluorophenyl
and methylamino linker are mandatory for activity was not
correct. On closer inspection, analogue 22c as well as other
active analogues with alicyclic moiety in the western part of the
molecule that have been published, however need to have in
addition an aromatic side chain in the eastern part of the
structure.22,23,25 Thus, a reversal of the binding mode that forces
the aromatic side chain into the pocket occupied by the 4-
fluorobenzylamino tail of 1 has to be taken into account to
explain the activity of 22c. The inactivity of analogues 7a-g with
no aromatic side chain is consistent with this explanation.
Analogue 11f displayed a seven-fold increase in potency
compared to 1. The methoxy substituent of 11f were swapped
for bulky ether moieties with only a subtle aliphatic variation of
the ethoxycarbonylamino side chain in order to limit the
increase in hydrophobicity to an acceptable level. Analogue 22d
has a benzyloxy group at C-2 and butanamide side chain, which
led to a dramatic 200-fold increase in potency with respect to 1.
An isopropyoxy substituent led to the even more potent
analogue 22e. Although both compounds turned out to be more
hepatotoxic than 1 in vitro, the increase in toxicity is more than
compensated by the superior channel opening activity, yielding
better safety indices compared to 1 in Hep-G2 cells. For
example, 22d and 22e have safety indices of 12,143 and 14,594,
respectively, versus a ratio of 81 in the case of 1 (table 1).
Activity favoured substitutions for the ethoxycarbonylamino
side chain and the primary amino group were put together in
the design of 25a-c (Scheme 5) before the favorable influence
of the replacement of 4-fluorobenzyl by cyclopropyl in the
western part was unraveled. Analogue 25a was thus equipped
with cyclohexylamino and morpholino substituents for the 4-
fluorobenzylamino and primary amino group, respectively. The
substituents sought to, in addition to activity, convey ease
oxidation and improve water solubility. As predicted, 25a was
exceedingly potent (EC50 = 26 nM). However, the fall in solubility
associated with increased lipophilicity (LogD7.4 = 4.29) was a
c
O
R
NH
N
R
N
N
O
O
R= F
R= OCH3
25b
25c
Scheme 5: Synthesis of 11h and 25a-c; a) 23a-b: amine reagents, triethylamine,
acetonitrile, 0 °C, 40 min, then rt, 30 min – 3 h; b) for 24a: amine reagents,
triethylamine, 2-propanol, reflux, overnight - 24 h; for 24b: amine reagent,
triethylamine, dimethyl sulfoxide, microwave (100 °C, 2 h); c) 11h & 25a-c: Pd/C,
H2, ethyl acetate/ 2-propanol, rt – 40 °C, overnight, then triethylamine, acylating
agent, 0 °C – 40 °C, 1 - 4 h
The next systematic strategy was to exchange the
cyclohexylamino moiety of 25a for still another morpholino ring
to further increase water solubility. Accordingly, the
dimorpholino analogue 25b was synthesized, which showed a
less lipophilic character (LogD7.4 = 3.58), better solubility, and
improved activity (EC50 = 11.3 nM) compared to 25a. Besides
the impressive 80-fold increase in potency compared to 1
(Figure 5), 25b also displayed safe hepatoxicity profiles in both
TAMH (LD25 = 212 mM) and HEP-G2 (LD25 = 231 mM) cell lines
(Figure 6). Safety indices of 18,761 and 20,442 with the TAMH
and HEP-G2 cells, respectively, were calculated for 25b. An
attempt to identify the oxidation products of 25b following
exposure to two equivalent m-chloroperbenzoic acid failed,
illustrating the desired resistance to oxidation.
Figure 5: The dose-response curve of the reference flupirtine (1) and 25b
The molecular orbital calculation (Figure 7) for 25b predicted
the lone pairs of electrons on the carbonyl oxygen atom as the
highest occupied molecular orbital (HOMO). This indicates the
6 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 7 of 11
Organic & Biomolecular Chemistry
Please do not adjust margins
Journal Name
ARTICLE
oxidation of 25b switches the formation of metabolites away tissues to unknown compounds in vivo triggers an immune
View Article Online
DOI: 10.1039/C9OB00511K
from reactive intermediates that are toxic to liver cells.
response that could lead to hepatotoxicity seen in a small
number of patients receiving 1 but none in case of 2.26
The balance of lipophilicity, aqueous solubility biological activity
and hepatotoxicity profiles makes 25b the fruitful conclusion of
this project. In a continued effort to identify improved
substituents for the ethoxycarbonylamino residue, the 3,5-
methoxybenzyl group, reported to improve activity as well as
solubility to a higher extent than 3,5-difluorobenzyl,27 was
identified. The target molecule 25c bearing this 3,5-
methoxybenzyl group instead of 3,5-difluorobenzyl in 25b,
however, showed reduced channel opening activity (EC50 = 5.82
M). A possible explanation was suggested by a conformation
analysis of 25b and 25c. The stable conformation calculated for
25b revealed that the 3,5-difluorophenyl ring lies close to the 2-
morpholino substituent but its replacement with a bulky 3,5-
dimethoxy substituent in 25c forces a shift in the conformation
away from the favoured one of 25b, which may lead to lower
biological activity.
Figure 6: Concentration-toxicity curves of 25b in TAMH and HEP-G2 liver cell line after
48 h
X-ray crystal structures of KV7.2/3 heteromers are not available,
yet. Therefore, ligand-based drug design approaches instead of
structure-based design was the method of choice in this study.
To design analogues that could fill the empty therapeutic niche
effectively, the surrounding of the pharmacophore of 1 was
partitioned into four major parts: the 4-fluorophenyl tail, the
methylamino linker, the primary amine substituent and the
ethyl carbamate side chain.
Conclusions
Figure 7: The highly occupied molecular orbital (HOMO) of 25b
An important consideration during retro metabolic drug design
is to consider not only the structure–activity relationships that
lead to increased pharmacological potency but also structure-
metabolism relationships that can affect toxicity. The new
analogues of flupirtine and retigabine were designed in five
iterative cycles to improve the desired pharmacological activity
while reducing adverse effects on liver cells. The present study
confirms that the structures of 1 and 2 are amendable to such a
retro metabolic drug design approach, which successfully led to
dramatic improvements in the desired biological activities. The
replacement of the primary amino group with a morpholino or
bulky ether substituents resulted in a dramatic improvement in
terms of channel opening potencies. Moreover, substituted
aromatic groups in place of the ethoxycarbonylamino side chain
also improved the activities of analogues. An amide moiety at
position 3 of the central pyridine was found to be the single
most important structural feature crucial for potent activity.
The most active derivatives 16b, 22d and 22e are characterised
by nanomolar activities (i.e., EC50 values of ca. 4 nM) and
superior efficacy of 154% (22d) on the Kv7.2/3 channel.
Moreover, the slightly less active compound 25b (potency 11
nM, efficacy 111%) showed an outstanding toxicity/activity
index (2∙104), which can be rationalized by its resistance to
oxidation. This molecular feature avoids the formation of
quinone diimines that are believed to be the causative agents
of fatal liver failure in the case of flupirtine, or give rise to
dimerization and tissue discoloration in the case of retigabine.
Based on MS data, Groseclose and Castellino proposed a
plausible mechanism for the formation of dimers of 2 by the
reaction of 3c with a second para-quinone-diimine resulting
from cleavage of the urethane and acylation of the N atom that
was unsubstituted in 2 yielding a regioisomer of 5f. The
postulated formation of the analogous regioisomer of 5e from
3c and a second para-quinone-diimine with a carbamate
requires that the only primary N atom of 2 would be
ethoxycarbonylated in vivo, this seems less plausible.
Therefore, we suggest the mechanism as shown in figure 1,
initiated by attack of para-quinone-diimines 3c,d on the
opposite side of the reaction partner in its ortho-quinone
diimine form 4c,d. Subsequent oxidation of the
hydrophenazines to phenazinium cations and dealkylation to
phenazines is reported.17 Remarkably, the absence of
detectable levels of these dimers in plasma indirectly explains
the safe clinical hepatotoxicity profile of 2. A comparable
formation of hitherto unknown aza-phenazinium salts in the
case of the highly analogous 1 is improbable. An initial attack of
3a-b at nitrogen atoms of 4a-d would be compulsory for
formation of flupirtine dimers 5a-d but is unlikely.
Consequently, metabolism of
1 disallows generation of
phenazinium dimers, explaining the absence of tissue
discoloration even after prolonged intake of 1. Recently, DILI
associated to 1 was reported to involve class II human leucocyte
genotype. It seems possible, that polymerization in other
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 7
Please do not adjust margins

Please do not adjust margins
Organic & Biomolecular Chemistry
Page 8 of 11
ARTICLE
Journal Name
using palladium under hydrogen atmosphere and then
Experimental
View Article Online
subsequently carbamoylated to yield anaDloOgIu: 1e0s.170a39-g/C. 9OB00511K
The structural details of all synthesized analogues and
intermediates, together with synthetic procedures and
structural characterizations are provided in the supporting
material. Analogues 7h-k have different substitution patterns in
place of the methylamino linker of 1. Among those analogues,
slight modifications of Scheme 1 were implemented to
synthesize 7j-k. The synthesis of 7j required a further amination
of the 6-chloro of intermediate 5 and the resulting compound
was coupled with 4-fluorobenzoyl chloride, reduced and
carbomylated to yield 7j. The synthesis of 7k, on the other hand,
encompasses a pre-reduction of 4-fluorobenzaldehyde to 4-
benzyl alcohol, which was then coupled with 5.
In design cycle 2, substitution or replacement of the primary
amine of 1 was employed to synthesize analogues 11a-h. The
synthesis started with the treatment of 2,6-dichloro-3-
nitropyridine with different amino or alcoholic reagents
(Scheme 2). Following the substitution of the 2-chloro moiety, a
4-fluorobenzyl group was coupled on the other end of the
pyridine.
Analogues 12a-l in design cycle 3 have different groups in place
of the ethoxycarbonylamino side chain of 1 and were
synthesized by varying the acylating agents at the end of the
sequence outlined in Scheme 1. Among 12a-l, pretreatments
were necessary to protect the amino group in analogues 12b-d.
The amino group of the starting amino acids were protected
using phthalic anhydride, which followed by conversion of the
carboxylic acid to acyl chloride using thionyl chloride. A
deprotection after the final acylation procedure yielded
analogues 12b-d. In addition, analogues 12k-l required an
additional step following the final acylation step shown in
Scheme 1; an eventual treatment with sulfuric acid was
necessary to effect a ring closure.
Varying the ethoxycarbonylamino side chain was also extended
to lead 2 (Scheme 3). Retigabine (2) was the starting material
for preparation of 16a-b. Initially, the primary amino group was
protected using phthalic anhydride. The ethyl carbamate was
then cleaved using hydrobromic acid, which was trailed by the
coupling of fluoro-substituted aromatic groups and removal of
the protecting group.
Synthetic procedures, melting points, NMR and HRMS data is
summarized in the Supporting Information. All commercial
reagents and solvents were used without further purification.
Reaction control via TLC was performed on silica gel 60 F254
aluminium plates from Merck by using UV-light ( = 254 nm and
366 nm) for visualization. NMR spectra were recorded with a
Bruker Avance III instrument at 400 MHz (1H-NMR) and 101
MHz (13C-NMR), respectively. The samples were measured in
DMSO-d6 solution (unless stated otherwise) with
tetramethylsilane as an internal standard at 25 °C. HRMS were
obtained with an LCMS-IT-TOF (Shimadzu) by using electrospray
ionization. IR spectra were recorded on a Bruker optics Alpha
FT-IR spectrometer equipped with a diamond single ATR
probehead. Melting points were measured automatically with
the Melting Point M-565 (Büchi). The compound purities were
determined by HPLC with DAD detection (100% method at 220
nm). Flash chromatography was performed (unless stated
otherwise) on a glass column using silica gel 60 from Carl Roth
with a particle size of 20−45 µm. The flow rate was adjusted to
approximately 25 mL/min by applying pressure with
compressed air.
Oxidation potentials of the compounds were obtained by using
a 797 VA Computrace device from Metrohm. Working
electrode: glassy carbon, reference electrode: Ag/AgCl/KCl 3 M,
auxiliary electrode: Pt. The compound (10 µmol) was
suspended in 10 mL of TRIS buffer (100 mmol/L, pH 7.4) and
sonicated for 3 min. The mixture was poured into the glass
sample holder and degassed by nitrogen for 3 min at room
temperature. Starting at –0.5 V the voltage was increased to
1.0 V with a speed of 0.1 V/s, while the current was recorded in
0.05 V steps. Once 1.0 V was reached, the voltage was inverted
until –0.5 V with the same speed to obtain a full cycle. Five of
those cycles were recorded but only the first sweep was used to
determine the anodic peak potential (Epa) as it gave the
strongest signal. The consecutive cycles showed lower currents,
most likely because of polymerization of the oxidized
compounds on the surface of the working electrode. Thus, the
graphite surface had to be polished after every measurement
with an alumina oxide suspension in water which was
distributed on a patch of cloth. The Epa was determined
automatically with the software 797 VA Computrace 1.3.1.
The lipophilicity of the compounds at pH 7.4 was determined by
using a RP-HPLC method. First, a calibration curve was
established by plotting the log(retention factor) (logk ′ ) of
several references against their literature logP. With the help of
the resulting logarithmic equation, the LogD7.4 values of our
compounds could be calculated after determination of their
logk′. More details are given in the Supporting Information.
In design cycle four, analogues 21a-b were synthesized as the 5-
methyl substituted analogues of 1 following the path outlined
in Scheme 4. Additional synthetic procedures – oxidation of the
pyridine nucleus and subsequent chlorination using
phosphorous oxychloride - were essential before starting the
amination of the pyridine ring.
Compounds 22a-e are analogues of 1 with at least two different
substitutions in place of the primary aromatic amino group,
ethoxycarbonylamino side chain or the 4-fluorobenzyl tail. The
general Scheme 1 was followed for the synthesis of those
analogues with the exception that the primary amino group was
not fully preserved. Moreover, analogues 25a-c have
simultaneous modifications of the primary amino group, the 4-
fluorobenzyl amino and the ethoxycarbonylamino side chain
and were synthesized as per Scheme 5.
Chemical synthesis
Analogues 7a-g synthesized by replacing the 4-fluorobenzyl side
group of 1 with various alicyclic groups. The synthesis was
initiated by amination of 2,6-dichloro-3-nitropyridine (Scheme
1). This was followed by coupling of 5 with various amino
reagents to synthesize 6a-g. The intermediates were reduced by
8 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 9 of 11
Organic & Biomolecular Chemistry
Please do not adjust margins
Journal Name
ARTICLE
Biological Evaluations
Cell culture
absorbance values for each treatment condition VtihewroAurtigclhe Otnhlianet
measured for control wells expressed inD%OI(:b1o0.t1h03c9o/Crr9eOcBte00d5f1o1Kr
background without cells).
The TAMH cells were a gift of Dr Sidney D. Nelson (University of
Washington, Seattle, USA) and cultured in T75 flasks with
Dulbecco's Modified Eagle's Medium/Nutrient Mixture F-12
(DMEM/F12 1:1 mix) (PAN Biotech), supplemented with 5%
PANEXIN NTA (PAN Biotech), 10 mM nicotinamide (Sigma
Aldrich) and 10 μg/mL gentamicin sulfate (PAN Biotech). The
HEP-G2 cell were purchased from the Leibniz-Institute-
The LD25 is the estimated concentration that reduces the T/Ccorr
to 75% and was determined by plotting (log)concentration of
test compound versus T/Ccorr and performing linear regression
analysis in Microsoft Excel (2013). Determined values are the
mean of at least three independent experiments ± standard
deviation (SD).
Deutsche Sammlung von Mikroorganismen und Zellkulturen KV7.2/3 channel opening assay
GmbH (Braunschweig, Germany) and were cultured in T75 flask
HEK293 cells were detached from the flask by TrypLE Express
(Sarstedt) with RPMI 1640 (PAN Biotech), supplemented with
10% heat-inactivated fetal bovine serum (FBS) (Sigma Aldrich)
and 1% penicillin/streptomycin (PANBiotech). The KV7.2/3
transformed HEK293 cells were obtained from SB Drug
Discovery (Glasgow, UK) and cultured in a T75 flask with
minimum essential medium with non-essential amino acids
(MEM) (Thermo Fisher Scientific), supplemented with 10%
heat-inactivated fetal bovine serum (FBS) (Sigma Aldrich), 2mM
L-glutamine (Sigma Aldrich), 4 μg/mL blasticidin S-HCl (Merck),
1% penicillin/streptomycin (Thermo Fisher Scientific) and 0.78
mg/mL G418 sulfate (Carl Roth). All cell lines were incubated at
37 °C in a humidified incubator with 95% air/5% CO2.
(Thermo Fisher Scientific) when ~80% confluent and suspended
in MEM media. After centrifugation the cell pellet was
resuspended in MEM media. Cells were counted with an EVE™
Automated Cell Counter and 60,000 cells/well/100 L MEM
media were seeded into black walled 96 well plates with clear
bottom (4titude).
Cells were allowed to attach for 24 h in a humidified incubator
at 37 °C, forming a near confluent monolayer. FLIPR® Potassium
Assay Kit (Molecular Devices, USA) was used to determine
KV7.2/3 channel opening activity of test compounds according
to manufacturer's instructions. Briefly, an equal amount of
Loading Buffer containing 5 mM probenecid (Santa Cruz
Biotechnology, USA) (pH 7.4) was added to wells and incubated
for 1 h at room temperature in the dark. Serial dilution series of
test compounds (5-9 concentrations) were prepared in DMSO
(BioScience-grade) (Carl Roth, Germany) and incubated with
cells for another 30 min at room temperature in the dark.
Control wells contained 1% DMSO. Plates were placed in an
Infinite® F200 Pro plate reader (Tecan, Switzerland) and
background fluorescence (baseline) for each well was bottom
read at excitation/emission wavelength: 485 nm/535 nm for
approximately 20 s. Then, 50 L Stimulus Buffer (c(K+) = 25 mM,
c(Tl+) = 15 mM) was added to each well and fluorescence
intensity was recorded immediately every 2.5 - 3 s continuously
for further 2 min.
The recorded data were transferred to Microsoft Excel (2013).
The ratio readout of maximal fluorescence intensity change of
a compound's concentration at a specific time point was divided
through the average of fluorescence intensities of baseline
(ΔF/F) and analogous determined value for control was
subtracted (corr. ΔF/F). At the chosen specific time point the
sum of ΔF/F of all concentrations belonging to a dilution series
of a compound became maximal. Corr. ΔF/F were plotted versus
log(concentration) of a compound, fitted to a four parameter
logistic equation and EC50 values were calculated according to
the software GraphPad Prism 6 (La Jolla, California, USA).
Indicated EC50 values are the mean of at least three
independent experiments ± standard deviation (SD).
MTT cell viability assay
TAMH cells detached from the flask when ~90% confluent with
Trypsin-EDTA (Sigma Aldrich), suspended in the presence of
0.25 mg/mL soybean trypsin inhibitor (Sigma Aldrich) in
DMEM/F12. After centrifugation, the cell pellet was
resuspended in DMEM/F12 media. Cells were counted with an
EVE™ Automated Cell Counter (NanoEnTek, S. Korea) and
20,000 cells/well/200 L DMEM/F12 media were seeded into
96 well plates (Sarstedt). Plates were incubated until cells
became confluent (approximately 48 h after seeding).
The HEP-G2 cells were detached from the flask by Trypsin-EDTA
(Sigma Aldrich) when ~70% confluent and suspended in RPMI
1640. Cells were counted with an EVE™ Automated Cell Counter
and 15,000 cells/well/100 µL RPMI 1640 media were seeded
into 96 well plates (Sarstedt). Cells were allowed to attach for
24 h in a humidified incubator at 37 °C.
In triplicate, the medium was removed in both cell lines and
replaced by 200 L/well of respective media containing a serial
dilution series of test compounds (5-9 concentrations) prepared
in DMSO (BioScience-grade) (Carl Roth). The background signal
was determined using controls with 1% DMSO and wells lacking
cells. Untreated wells containing cells and 1% DMSO in the
respective media were used as control. Following 24 h exposure
to the test compounds, medium was aspirated off, 125 L of the
respective medium, supplemented with 20% (V/V) of 2.5
mg/mL
solution
of
Emax value indicates the intrinsic activity of a compound relative
to flupirtine. It was determined by calculating the difference of
lowest and highest corr. ΔF/F value of an obtained sigmoidal
curve of a compound when corr. ΔF/F were plotted versus
log(concentration) and related to Emax of flupirtine maleate that
was defined as 100%. Emax value of a compound is the mean of
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide
(MTT) (Alfa Aesar) in phosphate buffered saline (PBS), was
added to wells and plates were incubated 2.5-4 h in a
humidified incubator at 37 °C. Absorbance was measured at λ =
570 nm with SpectraMax 190 microplate reader (Molecular
Devices). Cell viability (T/Ccorr) was obtained by dividing
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 9
Please do not adjust margins

Please do not adjust margins
Organic & Biomolecular Chemistry
Page 10 of 11
ARTICLE
Journal Name
15. S. V. Mathias and B. W. Abou-Khalil, Epilepsy Behav. Case
at least three independent experiments ± standard deviation
(SD).
View Article Online
DOI: 10.1039/C9OB00511K
Repo., 2017, 7, 61–63.
16. S. Jankovic and I. Ilickovic, Expert Opin. Drug Discov.,
2013, 8, 1429–1437.
17. M. R. Groseclose and S. Castellino, Chem. Res. Toxicol.,
2019, 53, 294-303.
Conflicts of interest
There are no conflicts to declare.
18. C. Bock, K. Beirow, A. S. Surur, L. Schulig, A. Bodtke, P. J.
Bednarski and A. Link, Org. Biomol. Chem., 2018, 16,
8695–8699.
Acknowledgements
19. A. S. Surur, K. Beirow, C. Bock, L. Schulig, M. K.
Kindermann, A. Bodtke, W. Siegmund, P. J. Bednarski and
A. Link, ChemistryOpen, 2019, 8, 41–44.
20. C. J. Lemmerhirt, M. Rombach, A. Bodtke, P. J. Bednarski
and A. Link, ChemMedChem, 2015, 10, 368–379.
21. K. Methling, P. Reszka, M. Lalk, O. Vrana, E. Scheuch, W.
Siegmund, B. Terhaag and P. J. Bednarski, Drug Metab.
Dispos., 2009, 37, 479–493.
AL and PJB are recipients of grants DFG LI 765/7-1 and DFG BE
1287/6-1 by the Deutsche Forschungsgemeinschaft. We wish to
acknowledge the excellent technical assistance of Anne
Schüttler, Maria Hühr and Michael Eccius.
The urine samples originate from healthy volunteers of a clinical
study group (male and female, 20–32 years, 51.5–102 kg, height
162–197 cm, BMI 19.4–26.7 kg/m2), who gave informed written
consent according to current international and national
22. C. Jessen, W. Dalby-Brown and D. Strøbæk,
Tetrahydropyran-aminopyridine derivatives and their
medical use (WO 2010/026104 A1).
regulations
(EudraCT2007-007483-17;
ClinicalTrials.gov
NCT01676246).
23. W. Dalby-Brown, C. Jessen and D. Strøbæk, Substituted
pyridine derivatives and their medical use (WO
2010/094644 A1).
24. R. Y. Kim, M. C. Yau, J. D. Galpin, G. Seebohm, C. A.
Ahern, S. A. Pless and H. T. Kurata, Nat. Commun., 2015,
6, doi: 10.1038/ncomms9116
25. W. Dalby-Brown, C. Jessen and D. Strøbæk, Substituted
pyridine derivatives and their medical use (WO
2010/094645 A1).
26. P. Nicoletti, A. N. Werk, A. Sawle, Y. Shen, T. J. Urban, S.
A. Coulthard, E. S. Bjornsson, I. Cascorbi, A. Floratos, T.
Stammschulte, U. Gundert-Remy, M. R. Nelson, G. P.
Aithal and A. K. Daly, Pharmacogenet. Genomics, 2016,
26, 218–224.
References
1.
2.
3.
D. A. Brown and G. M. Passmore, Br. J. Pharmacol., 2009,
156, 1185–1195.
P. Delmas and D. A. Brown, Nat. Rev. Neurosci., 2005, 6,
850–862.
A. P. Stewart, J. C. Gómez-Posada, J. McGeorge, M. J.
Rouhani, A. Villarroel, R. D. Murrell-Lagnado and J. M.
Edwardson, J. Biol. Chem., 2012, 287, 11870–11877.
S. Li, V. Choi and T. Tzounopoulos, Proc. Natl. Acad. Sci.
U. S. A., 2013, 110, 9980–9985.
4.
5.
6.
M. A. Rogawski, Trends Neurosci., 2000, 23, 393–398.
K. Rose, L. Ooi, C. Dalle, B. Robertson, I. C. Wood and N.
Gamper, Pain, 2011, 152, 742–754.
27. C. Bock, A. S. Surur, K. Beirow, M. K. Kindermann, L.
Schulig, A. Bodtke, P. J. Bednarski and A. Link,
ChemMedChem, 2019, 10.1002/cmdc.201900112.
7.
8.
9.
E. Cisneros, C. Roza, N. Jackson and J. A. López-García,
Front. Cell. Neurosci., 2015, 9, 10.3389/fncel.2015.00470
D. Sampath, R. Valdez, A. M. White and Y. H. Raol,
Neuropharmacology, 2017, 123, 126–135.
European medicines agency, Assessment report for
flupirtine containing medicinal products, 2013.
10. European medicines agency, Withdrawal of pain
medicine flupirtine endorsed, 2018. Press
release 23/03/2018, downloaded 01/03/2019 from
https://www.ema.europa.eu/en/news/withdrawal-pain-
medicine-flupirtine-endorsed.
11. C. Bock and A. Link, Future Med. Chem., 2019, 11, 337–
355.
12. E. Scheuch, K. Methling, P. J. Bednarski, S. Oswald and W.
Siegmund, J. Pharm. Biomed. Anal., 2015, 102, 377–385.
13. Y. Ihara, Y. Tomonoh, M. Deshimaru, B. Zhang, T. Uchida,
A. Ishii and S. Hirose, PLoS ONE, 2016, 11, e0150095.
14. M. J. Gunthorpe, C. H. Large and R. Sankar, Epilepsia,
2012, 53, 412–424.
10 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Please do not adjust margins
Organic & Biomolecular Chemistry
Page 11 of 11
View Article Online
DOI: 10.1039/C9OB00511K
Organic & Biomolecular Chemistry
ARTICLE
Puzzling stability: Molecular jigsaw pieces of residues characterized in light of activity, lipophilicity, stability
against oxidation, and hepatotoxicity were combined to yield flupirtine analogue 25b.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 11
Please do not adjust margins