Synthesis and potassium KV7 channel opening activity of thioether analogues of the analgesic flupirtine

Article abstract of DOI:10.1039/c8ob02530d

Flupirtine, an opener of neuronal voltage gated potassium channels (K_V7.2/3), has been used as a therapeutic alternative for pain treatment in patients refractory to NSAIDs and opioids. Because flupirtine is associated with rare but fatal drug-induced liver injury that may result from the formation of toxic metabolites upon metabolic oxidation, we synthesized novel derivatives with the goal of identifying equally active and ultimately safer K_V7.2/3 channel openers. Four thioether analogues were designed to lack a nitrogen atom that would be a prerequisite for the formation of toxic para-quinone diimines, and form sulfoxide and sulfone metabolites instead. K_V7.2/3 channel opening activity and hepatotoxicity data of twelve novel flupirtine analogues, four thioethers and their respective sulfoxide and sulfone metabolites are reported.

Full text of DOI:10.1039/c8ob02530d

Organic &
Biomolecular Chemistry
View Article Online
View Journal | View Issue
COMMUNICATION
Synthesis and potassium KV7 channel opening
activity of thioether analogues of the analgesic
flupirtine†
Cite this: Org. Biomol. Chem., 2018,
16, 8695
Received 12th October 2018,
Accepted 29th October 2018
Christian Bock, Kristin Beirow, Abdrrahman S. Surur, Lukas Schulig, Anja Bodtke,
DOI: 10.1039/c8ob02530d
Patrick J. Bednarski and Andreas Link
*
rsc.li/obc
Flupirtine, an opener of neuronal voltage gated potassium chan- (DILI).3 Since the hepatotoxicity of 1 is an obstacle to a future
nels (KV7.2/3), has been used as a therapeutic alternative for pain use as an analgesic, the search for analogues with a better
treatment in patients refractory to NSAIDs and opioids. Because safety profile while retaining or even improving the KV7.2/3
flupirtine is associated with rare but fatal drug-induced liver injury channel opening potency would seem a worthwhile drug-
that may result from the formation of toxic metabolites upon design goal.
metabolic oxidation, we synthesized novel derivatives with the
In previous work, we designed flupirtine analogues that
goal of identifying equally active and ultimately safer KV7.2/3 would prohibit the formation of azaquinone diimines like 2
channel openers. Four thioether analogues were designed to lack a and 3 by introducing methyl groups onto the oxidizable N
nitrogen atom that would be a prerequisite for the formation of atoms.4 We determined the oxidative potentials of the
toxic para-quinone diimines, and form sulfoxide and sulfone
metabolites instead. KV7.2/3 channel opening activity and hepato-
toxicity data of twelve novel flupirtine analogues, four thioethers
and their respective sulfoxide and sulfone metabolites are
reported.
The centrally acting analgesic flupirtine (1) is believed to act by
opening voltage gated potassium channels (KV7.2/3) in the
membranes of nerve cells of the CNS, leading to a hyperpolar-
ization of the membrane potential, thus suppressing pain
mediation and repetitive neuronal firing (as observed in con-
vulsions). Until recently, the drug was regarded as safe for
short-term use and thus an attractive alternative to NSAIDs
and opiates in pain control.1 However, fatal drug-induced liver
injury in a small number of patients has led to its removal in
2018 from the European market. It was recently shown that 1
can be metabolized to cysteine and mercapturic acid conju-
gates in humans, evidence that 1 is oxidized in vivo to reactive
azaquinone diimines 2 and 3, which can react covalently with
glutathione (Fig. 1).2 Subsequent depletion of glutathione levels
could result in the covalent interaction with other nucleophiles
such as liver proteins and cause drug-induced liver injury
Pharmaceutical and Medicinal Chemistry, Institute of Pharmacy,
University of Greifswald, Friedrich-Ludwig-Jahn-Str. 17, 17487 Greifswald, Germany.
E-mail: link@uni-greifswald.de
†Electronic supplementary information (ESI) available: Analytical data for all
synthesized compounds and methods describing the cyclovoltammetry, log D7.4
determinations, in vitro hepatotoxicity and Kv7.2/3 channel opening experi-
Fig. 1 Structure of KV channel opener flupirtine (1), elusive metabolites
2 and 3 and product 4 of phase II drug metabolism.
ments. See DOI: 10.1039/c8ob02530d
This journal is © The Royal Society of Chemistry 2018
Org. Biomol. Chem., 2018, 16, 8695–8699 | 8695

View Article Online
Communication
Organic & Biomolecular Chemistry
N-methylated analogues electrochemically by cyclic voltamme-
try and attempted to correlate these results with both the
KV7.2/3 (KCNQ2/3) opening activity, determined by an estab-
lished assay with HEK293 cells overexpressing these channels,
and the in vitro hepatotoxicity, determined by a cytotoxicity
assay with the mouse hepatocyte cell line TAMH. In the initial
test set of alkylated derivatives, a link was found between the
oxidation potentials of the compounds and their EC50 values
for potassium channel opening activity but not between oxi-
dation potential and cytotoxicity in TAMH cells.4
Here, we report another approach to modify chemical stabi-
lity towards oxidation by replacement of the benzylic NH of 1
with bioisosteric, divalent S. It was envisioned that the oxi-
dative pathway of the sulfides might be directed away from oxi-
dations to azaquinone intermediates and towards oxidations
to sulfoxides5 and sulfones, which would be expected to be
less hepatotoxic. This altered oxidative pathway could help to
attenuate hepatotoxicity while retaining or even improving on
the KV7.2/3 opening activity.
The synthesis of the novel thioether analogues 6a–d of 1
started with the trisubstituted pyridine 5 (Scheme 1). In
analogy to Artemov et al., the thiol was deprotonated6 and
reacted with the respective benzyl bromides at room tempera-
ture to yield the corresponding sulfides. Palladium-catalysed
reduction under a hydrogen atmosphere failed to convert the
nitro group quantitatively under reported conditions.7 Clean
conversion was accomplished using iron powder in 2-propa-
nol/water with NH4Cl. The intermediates were transformed to
6a–d by reaction with ethyl chloroformate at 40 °C. Sulfoxides
7a–d were obtained by oxidation of 6a–d with meta-chloroper-
oxybenzoic acid (mCPBA).8 To end the reaction at the sulfoxide
step, we used only a small excess of the oxidizing agent and
added it dropwise at 0 °C. Sulfones 8a–d were synthesized in
an analogous fashion by oxidation of 6a–d with at least two
equivalents of mCPBA.
Fig. 2 Concentration–response curves of 1 (as maleate salt) and 6d
determined after 30 min exposure via thallium-flux assay in HEK293
cells expressing KV7.2/3; data were normalised to control (1% DMSO);
values are the mean SD (n ≥ 3); EC50 values calculated from log(con-
centration)-response curves which were fitted to a four-parameter
logistic equation.
Table 1 Comparisons of anodic peak potentials (Epa) of 6a–d and
metabolites 7a–d, and 8a–d, log D7.4 values, LD50 values after 24 h in
TAMH cells and EC50 values towards KV7.2/3 channels in HEK293 cells
Entry R1
R2 Epa a [mV] Log D7.4
LD50 c [µM] EC50 c [µM]
b
1
336
464 10
5
2.96
3.54
3.45
3.66
3.91
2.64
2.50
2.80
3.20
2.91
2.75
3.07
3.42
487 51
>40
>250
>30
0.9 0.1
1.3 0.3
1.8 0.4
0.8 0.1
0.2 0.0
>10
>10
>10
>10
>10
6a
6b
6c
6d
7a
7b
7c
7d
8a
8b
8c
8d
F
H
F
CF3
F
H
F
CF3
F
H
H
H
F
H
H
H
F
H
H
H
F
455
460
459
653
645 10
639 11
626
717 10
709
719
711
3
3
3
5
>20
>125
>1000
>250
>100
>125
>500
>125
>30
3
5
8
3
>10
>10
>10
F
CF3
H
a Measured as 1 mM solutions in 100 mM TRIS-buffer pH 7.4, shown
are mean values SD from 3–4 independent measurements. b Log D7.4
values were determined using an HPLC method (n = 2). c EC50- and
LD50-values are means and standard deviations of 4–5 independent
determinations.
After structural confirmation (i.e., 1H-NMR, 13C-NMR, MS)
of the compounds 6–8, their anodic peak potentials (Epa) were
measured by cyclic voltammetry4 and lipophilicity (log D7.4) by
an RP-HPLC method. The potential for hepatotoxicity of the
synthesized analogues was evaluated in a transgenic mouse
hepatocyte (TAMH) model by means of the MTT cytotoxicity for 1 and 6d. The end-points for all these studies are collected
assay as described previously.4 The KV7.2/3 channel opening in Table 1. By replacing the benzylic nitrogen of 1 for sulfur,
activity was assessed by a fluorescence-based cellular thallium the anodic peak potentials, and thus the resistance to oxi-
flux assay with HEK293 cells overexpressing these channels.9 dation, were increased from 336 to ca. 460 mV (Table 1). As
Fig. 2 shows the representative concentration-response curves expected, a further increase in the peak potentials was
Scheme 1 Synthesis of thioethers 6a–d and putative sulfoxide and sulfone metabolites rac-7a–d and 8a–d. (i) Alkyl bromide, NaOH, DMF, r.t., 3 h.
(ii) Fe, NH4Cl, 2-propanol/water, reflux, 5 h. (iii) Ethyl chloroformate, triethylamine, 0 °C, 2 h, r.t., 5 h. (iv) mCPBA, DCM, 0 °C, 1–5 h. (v) mCPBA (>2
eq.), DCM, 0 °C, 2–7 h.
8696 | Org. Biomol. Chem., 2018, 16, 8695–8699
This journal is © The Royal Society of Chemistry 2018

View Article Online
Organic & Biomolecular Chemistry
Communication
achieved when the sulfide was modified to either a sulfoxide
or a sulfone. In contrast, substitutions in the benzylic ring had
no influence on the anodic peak potentials.
The log D7.4 of the sulfide 6a increased relative to 1 by 0.60.
On the other hand, oxidation of the sulfide to the sulfoxide
reduced lipophilicity by ca. 10-fold. Further oxidation to the
sulfone led to a slight increase in lipophilicity relative to the
sulfoxides. As expected, the introduction of fluorine led to
increases in log D7.4 values by values of ca. 0.15 for each F
atom. The LD50 for 1, obtained by the MTT assay after exposing
the TAMH mouse hepatocyte cell line for 24 h, was 487 µM.
However, the limiting solubility of the new compounds hin-
dered our ability to determine their LD50 values in this assay.
Nonetheless, it can be seen in Table 1 that the LD50 values of
none of the compounds were reached even at the highest
soluble concentrations. According to thallium-flux KV7.2/3
channel-opening assay, the most potent compound from this
study was the thioether 6d, with an EC50 value of 0.24 µM,
leading to a significant left-shift of the concentration-response
curve and a higher maximal effect compared to 1 (Fig. 2). As
proof of principle, the small loss in activity associated with the
exchange of said NH of 1 by S in 6a from EC50 0.9 µM to
1.3 µM could be more than compensated by the introduction
of a second F into 6c (EC50 = 0.8 µM) or replacement of F by
CF3 in the benzylic moiety in 6d (EC50 = 0.2 µM), presumably
due to stronger lipophilic interactions.
While all the sulfides showed activity in this assay, none of
the sulfoxides or sulfones were active at concentrations below
10 µM. Two possibilities for these differences in activity are:
(1) the higher anodic peak potentials of the sulfoxides and sul-
fones relative to the sulfides make them less active than the
sulfides; indeed, 1 with the lowest peak potential in the series
of p-fluorophenyl derivatives is the most potent (compare 1
with 6a). We observed a similar relationship with the group of
N-methylated analogues of 1.4 (2) The reduced lipophilicity of
the sulfoxides and sulfones relative to the sulfides makes them
less active; however, the log D7.4 of 1 is 2.96, which is similar to
that of the inactive sulfoxide 8a (2.91). Thus, there does not
appear to be a simple relationship between activity and
lipophilicity.
Fig. 3 Predicted binding pose of compound 6a and unfavourable inter-
actions of sulfone 8a (lower right corner). The docking was performed
by using an induced fit with the force field based MM/GBVI scoring
method as implemented in MOE (see the ESI†).
to cross-check the proposed binding mode. The assumption
that 6c and 6d are more active than 1 and 6a due to additional
lipophilic contacts is in accordance with this docking pose.
However, especially another hydrophobic pocket on the
bottom side leaves room for larger substituents in position 3
of the pyridine ring and could be exploited in the future. As
can be seen from Fig. 3, sulfide 6a adopts a conformation that
avoids steric clashes in the region of the sulfur atom. In con-
trast, the two additional oxygen atoms of the sulfones 8a–d
should lead to an unfavourable interaction in this bridging
region. This rationalization was confirmed by negative results
for the sulfones 8a–d (Table 1). While this is obvious for 8a–d,
it is not so straightforward to explain the low activity for
the sulfoxides 7a–d. At least the R-enantiomers of the
chiral sulfoxides 7a–d lacking an oxygen atom in the critical
position might be anticipated to avoid such unfavourable
interactions.
To rationalize further the finding that the sulfide analogues
are active while the sulfoxides and sulfones are not, we
explored a molecular modelling approach. Because the crystal
structures of the KV7.2/3 channel are not available, we chose to
visualize the supposed binding mode by means of a homology
Conclusions
model based on a heteromeric approach and published experi- Thioethers 6c and d are more potent as KV7.2/3 channel
mental data (Fig. 3).10,11 While these modelling results are not openers compared to the clinically effective drug flupirtine (1).
suited to predict a new generation of KV channel openers, it Thus the development of thio-analogues of known drug 1 may
can be deduced from the anticipated binding mode in Fig. 3 result in more effective KV channel openers by helping to sep-
that the sulfur atom will not alter the orientation of the arate pharmacological activity from toxicity. Putative metab-
benzylic moiety relative to 1 and thus, that this aromatic olites resulting from hepatic or intracellular oxidation, namely
binding motif will be buried in the hydrophobic cleft formed sulfoxides 7a–d and sulfones 8a–d, are of interest for the pre-
by amino acid residues L272, L314, L311 and F344 as the diction of toxicity of oxidative metabolites and thus were made
upper boundary of the presumed binding site. We have taken available for biological evaluation. These sulfoxides and sul-
into account the experimentally determined important resi- fones were inactive as KV7.2/3 channel openers, and impor-
dues11 as well as other derivatives described in the literature12 tantly, avoid the formation of reactive azaquinone diimines.
This journal is © The Royal Society of Chemistry 2018
Org. Biomol. Chem., 2018, 16, 8695–8699 | 8697

View Article Online
Communication
Organic & Biomolecular Chemistry
Synthesis of sulfone 8a
Experimental
6a (1 mmol, 0.32 g) was dissolved in 20 mL DCM and cooled
to 0 °C. mCPBA (2 eq., 2 mmol, 0.63 g of a 55% mixture with
water) was dissolved in 20 mL DCM and added dropwise over a
period of 1 hour. The solution was stirred for further 45 min at
0 °C. The resulting suspension was filtered and the residue
was washed with DCM. The filtrate was washed with 3 ×
100 mL sat. NaHCO3 solution and 100 mL brine. The organic
phase was evaporated and purified by flash chromatography
(puriFlash® 450 system from Interchim, 25 g silica column,
mobile phase: n-hexane/ethyl acetate, gradient: 1 CV 10% ethyl
acetate, 10 CV 10–80% ethyl acetate, 5 CV 80% ethyl acetate).
The solid was washed with small quantities of diethyl
ether and dried under reduced pressure at 40 °C. Pink solid
Synthesis of thioether 6a
6-[(4-Fluorobenzyl)thio]-3-nitropyridin-2-amine (5 mmol, 1.4 g)
was suspended in 30 mL water and 50 mL 2-propanol. Iron
powder (3 eq., 15 mmol, 0.84 g) and NH4Cl (4.5 eq.,
22.5 mmol, 1.2 g) were added and the suspension was refluxed
for 5 hours. The resulting dark brown mixture was filtered and
evaporated to dryness. The residue was diluted with 150 mL
water and washed three times with 60 mL ethyl acetate. The
combined organic layers were washed with 150 mL brine and
dried over Na2SO4. Et3N (1.4 eq., 7 mmol, 0.97 mL) was added
and the solution was cooled to 0 °C. A solution of ethyl chloro-
fomate (1.2 eq., 6 mmol, 0.57 mL) in 5 mL ethyl acetate was
added dropwise. After 2 hours additional ethyl chlorofomate
(1,2 eq., 6 mmol, 0.57 mL) was added at r.t., 5 hours later,
ethyl chlorofomate (0.6 eq., 3 mmol, 0.29 mL) and Et3N
(0.7 eq., 3.5 mmol, 0.49 mL) were added. After 2 hours the
reaction was completed and the solution was washed with
200 mL water and 200 mL saturated brine. After drying with
Na2SO4 and evaporation to dryness, the resulting solid was
recrystallized from DCM. The product was dried at 40 °C
under reduced pressure. Brown solid (m.p.: 140–142 °C). Yield:
0.75 g (47%). Compound purity: 100.0%. Rf = 0.83 (cyclo-
hexane/ethanol/Et3N 6 : 2 : 2). 1H-NMR [ppm]: δ = 1.22 (t, 3H,
J = 7.1 Hz), 4.09 (q, 2H, J = 7.1 Hz), 4.29 (s, 2H), 5.94 (s, 2H),
6.43 (d, 1H, J = 8.0 Hz), 7.07–7.12 (m, 2 H), 7.40–7.45 (m, 3H),
8.57 (m, 1H). 13C-NMR [ppm]: δ = 14.5, 32.6, 60.3, 109.3, 115.0
(m.p.: 168–171 °C). Yield: 0.09
g (25.5%). Compound
purity: 100.0%. Rf = 0.41 (n-hexane/ethyl acetate). 1H-NMR
[ppm]: δ = 1.26 (t, 3H, J = 7.0 Hz), 4.16 (q, 2H, J = 7.0 Hz),
4.61 (s, 2H), 6.64 (s, 2H), 7.00 (d, 1H, J = 8.0 Hz), 7.13–7.17
(m, 2H), 7.22–7.25 (m, 2H), 7.95 (d, 1H, J = 7.8 Hz), 9.05 (s,
1H). 13C-NMR [ppm]: δ = 14.4, 56.1, 60.9, 111.5, 115.2 (d, 2C,
2JC,F = 22 Hz), 123.3, 124.8 (d, 4JC,F = 3 Hz), 127.4, 133.0 (d, 2C,
3JC,F = 8 Hz), 147.3, 151.3, 153.9, 160.9 and 163.3 (d, 1JC,F = 245
Hz). IR [cm−1]: ˜ν = 1232, 1462, 1511, 1634, 1714, 2979, 3305,
3445. HRMS: [C15H16N3O4FS + H]+ req.: 354.0918, found:
354.0930.
Conflicts of interest
2
3
(d, 2C, JC,F = 21 Hz), 115.2, 130.7 (d, 2C, JC,F = 8 Hz), 131.8,
4
135.0 (d, JC,F = 3 Hz), 150.2, 152.8, 154.4, 159.9 and 162.3
There are no conflicts of interest to declare.
1
(d, JC,F = 243 Hz). IR [cm−1]: ˜ν = 1454, 1678, 2984, 3145, 3260,
3403. HRMS: [C15H16N3O2FS + H]+ req.: 322.1020, found:
322.1027.
Acknowledgements
AL and PJB are recipients of grants DFG LI 765/7-1 and DFG
BE 1287/6-1 by the Deutsche Forschungsgemeinschaft. C. B.
and K. B. are funded by these grants.
Synthesis of sulfoxide 7a
6a (0.5 mmol, 0.17 g) was dissolved in 10 mL DCM and cooled
to 0 °C. meta-Chloroperoxybenzoic acid (mCPBA, 1.1 eq.,
0.55 mmol, 0.17 g of a 55% mixture with water) was dissolved
in 10 mL DCM and added dropwise within 30 min. The solu-
tion was stirred for 3 hours, while the product precipitated as
a white solid. The suspension was evaporated to dryness and
purified by flash chromatography (mobile phase: DCM/MeOH
95 : 5). The product was washed with diethyl ether, sat.
NaHCO3 solution and water and dried under reduced pressure
at 40 °C. Light pink solid (m.p.: 202–204 °C). Yield: 0.11 g
(65.1%). Compound purity: 100.0%. Rf = 0.58 (n-hexane/ethyl
Notes and references
1 A. S. Nair, Anesth. Analg., 2018, 126, 1425; L. Puljak, Curr.
Med.
Res.
Opin.,
2018,
DOI:
10.1080/
03007995.2018.1528217; S. Kaplan, B. Ehlken and
X. Hamann, Curr. Med. Res. Opin., 2018, DOI: 10.1080/
03007995.2018.1499507.
2 E. Scheuch, K. Methling, P. J. Bednarski, S. Oswald and
W. Siegmund, J. Pharm. Biomed. Anal., 2015, 102, 377–
385.
1
acetate 5 : 5). H-NMR [ppm]: δ = 1.25 (t, 3H, J = 7.1 Hz), 3.99
(d, 1H, J = 13.0 Hz), 4.14 (q, 2H, J = 7.0 Hz), 4.32 (d, 1H, J =
13.0 Hz), 6.38 (s, 2H), 6.72 (d, 1H, J = 7.9 Hz), 7.05–7.12 (m, 4H),
7.82 (d, 1H, J = 7.5 Hz), 8.86 (s, 1H). 13C-NMR [ppm]: δ = 14.5,
3 K. Konishi, T. Fukami, T. Ogiso and M. Nakajima, Biochem.
Pharmacol., 2018, 155, 242–251.
2
57.6, 60.7, 108.7, 114.9 (d, 2C, JC,F = 21 Hz), 120.5, 126.6 (d,
4 C. J. Lemmerhirt, M. Rombach, A. Bodtke, P. J. Bednarski
and A. Link, ChemMedChem, 2015, 10, 368–379.
5 J. D. Turner, R. Sharma, G. Al Jayoussi, H. E. Tyrer,
J. Gamble, L. Hayward, R. S. Priestley, E. A. Murphy,
J. Davies, D. Waterhouse, D. A. N. Cook, R. H. Clare,
4JC,F = 3 Hz), 129.5, 132.2 (d, 2C, JC,F = 8 Hz), 152.0, 154.2,
3
154.6, 160.7 and 163.0 (d, JC,F = 244 Hz). IR [cm−1]: ˜ν = 1219,
1
1466, 1532, 1731, 2978, 3213, 3294, 3347. HRMS:
[C15H16N3O3FS + H]+ req.: 338.0969, found: 338.0978.
8698 | Org. Biomol. Chem., 2018, 16, 8695–8699
This journal is © The Royal Society of Chemistry 2018

View Article Online
Organic & Biomolecular Chemistry
Communication
A. Cassidy, A. Steven, K. L. Johnston, J. McCall, L. Ford,
J. Hemingway, S. A. Ward and M. J. Taylor, Proc. Natl. Acad.
Sci. U. S. A., 2017, 114, E9712–E9721.
6 V. A. Artemov, V. L. Ivanov, A. V. Koshkarov,
A. M. Shestopalov and V. P: Litvinov, Chem. Heterocycl.
Compd., 1998, 34, 96–101.
9 C. D. Weaver, D. Harden, S. I. Dworetzky, B. Robertson and
R. J. Knox, SLAS Discovery, 2004, 9, 671–677.
10 W. Lange, J. Geißendörfer, A. Schenzer, J. Grötzinger,
G. Seebohm, T. Friedrich and M. Schwake, Mol. Pharmacol.,
2009, 75, 272–280.
11 V. Barrese, F. Miceli, M. V. Soldovieri, P. Ambrosino,
F. A. Iannotti, M. R. Cilio and M. Taglialatela, Clin.
Pharmacol., 2010, 2, 225–236.
7 F. K. Hansen and D. Geffken, J. Heterocycl. Chem., 2012, 49,
321–328.
8 N. Barry, N. Brondel, S. E. Lawrence and A. R. Maguire, 12 I. Dorange and B.-M. Swahn, Annu. Rep. Med. Chem., 2011,
Tetrahedron, 2009, 65, 10660–10670.
46, 53–65.
This journal is © The Royal Society of Chemistry 2018
Org. Biomol. Chem., 2018, 16, 8695–8699 | 8699