Discovery of (S,E)-3-(2-fluorophenyl)-N-(1-(3-(pyridin-3-yloxy)phenyl) ethyl)-acrylamide as a potent and efficacious KCNQ2 (Kv7.2) opener for the treatment of neuropathic pain

Article abstract of DOI:10.1016/j.bmcl.2013.08.092

Acrylamide (S)-6, a potent and efficacious KCNQ2 (Kv7.2) opener, demonstrated significant activity in two models of neuropathic pain and in the formalin test, suggesting that KCNQ2 openers may be useful in the treatment of neuropathic pain including diabetic neuropathy.

Full text of DOI:10.1016/j.bmcl.2013.08.092

Bioorganic & Medicinal Chemistry Letters 23 (2013) 6188–6191
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of (S,E)-3-(2-fluorophenyl)-N-(1-(3-(pyridin-3-yloxy)
phenyl)ethyl)-acrylamide as a potent and efficacious KCNQ2 (Kv7.2)
opener for the treatment of neuropathic pain
⇑
Yong-Jin Wu , Charles M. Conway, Li-Qiang Sun, Frederic Machet, Jie Chen, Ping Chen, Huan He,
Clotilde Bourin, Vincenzo Calandra, Joseph L. Polino, Carl D. Davis, Karen Heman, Valentin K. Gribkoff,
Christopher G. Boissard, Ronald J. Knox, Mark W. Thompson, William Fitzpatrick, David Weaver,
David G. Harden, Joanne Natale, Steven I. Dworetzky, John E. Starrett Jr.
Research and Development, Bristol-Myers Squibb Company, 5 Research Parkway, Wallingford, CT 06492, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Acrylamide (S)-6, a potent and efficacious KCNQ2 (Kv7.2) opener, demonstrated significant activity in
two models of neuropathic pain and in the formalin test, suggesting that KCNQ2 openers may be useful
in the treatment of neuropathic pain including diabetic neuropathy.
Received 25 April 2013
Accepted 26 August 2013
Available online 8 September 2013
Ó 2013 Elsevier Ltd. All rights reserved.
Keywords:
KCNQ2
Kv7.2
Neuropathic pain
Diabetic neuropathy
Acrylamide
An estimated 4 million people suffer from neuropathic pain in
the United States.1 Neuropathic pain is characterized by spontane-
ous pain, mechanical and thermal hyperalgesia (exaggerated re-
sponse to a painful stimulus) and allodynia (pain response to an
innocuous stimulus). Common types of neuropathic pain include
diabetic neuropathy, postherpetic neuralgia, trigeminal neuralgia,
cancer-related neuropathy, and phantom limb pain. Diabetic neu-
ropathy, a microvascular complication of diabetes, is one of the
most common forms of neuropathic pain. It develops in 28–55%
of patients with diabetes mellitus.1 The commonly prescribed
drugs for the treatment of neuropathic pain include gabapentin,
pregablin, carbamazepine and lamotrigine. Unfortunately, these
treatments only provide pain relief for certain patient pupulation
(30–50%), and frequently the relief is only partial (40–50%).2 There-
fore, there is an urgent need for the discovery of more effective
anti-neuropathic pain medications with improved responsive rate.
One potential approach to treat neuropathic pain is to dampen
neuronal excitability through activation of KCNQ2/KCNQ3 (Kv7.2/
Kv7.3) potassium channels.3,4 These channels are the molecular
correlate of the M-current, which serves to stabilize the membrane
potential and control neuronal excitability. Activation of KCNQ2/
KCNQ3 channels hyperpolarize neuronal membranes, thus inhibit-
ing the induced hyperexcitability. Indeed, several KCNQ2/KCNQ3
openers such as retigabine5,6 (a FDA approved anticonvulsant
drug) and compound 17demonstrated significant analgesic activity
on tactile allodynia in the Chung-lesioned rats, indicating the ther-
apeutic potential of KCNQ2 openers for the treatment of neuro-
pathic pain.
Our initial medicinal chemistry efforts led to the discovery of
acrylamide (S)-2 (Chart 1) as a moderately active KCNQ2 opener
with oral activity in a rat model of migraine.8,9 Subsequently, we
identified acrylamide (S)-3 (Chart 1) with increased potency and
efficacy.10 Our efforts were then directed at the KCNQ2 openers
with improved potency over acrylamide (S)-3 while maintaining
its efficacy, and we found that optimization of both parameters
proved to be challenging. For example, compounds (S)-411 and
(S)-512 were more potent than (S)-3, but they were less efficacious.
Thus, we undertook extensive structure–activity relationship stud-
ies, and these studies led to two biaryl ether acrylamides (S)-6 and
(S)-7 with the desired potency and efficacy. This Letter describes
the synthesis and in vivo activity of these acrylamides in various
animal models of neuropathic pain.
Scheme 1 describes the synthesis of acrylamides (S)-6 and (S)-7.
The commercially available (S)-1-(3-methoxyphenyl)ethanamine
(8) was treated with di-tert-butyl dicarbonate and triethyl amine
⇑
Corresponding author. Tel.: +1 203 677 7485; fax: +1 203 677 7702.
E-mail address: yong-jin.wu@bms.com (Y.-J. Wu).
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.08.092

Y.-J. Wu et al. / Bioorg. Med. Chem. Lett. 23 (2013) 6188–6191
6189
Table 1
O
O
Whole cell patch-clamping data (À40 mV)
NH
NH2
O
a,b
Compd
EC50 (nM)
E/Eref
N
S
O
N
N
N
H
(S)-2
(S)-3
3200 600
60 10
1.4 0.1
1.8 0.1
1.1 0.1
0.85 0.01
2.2 0.1
NA
2.9 0.1
NA
1.6 0.1
H
F3CO
F
(S)-4
20
1
1
retigabine
(S)-5
(S)-6
(R)-6
(S)-7
(R)-7
Retigabine
10 0.1
2.4 0.1
>10,000
0.9 0.1
>10,000
1300 40
F
F
O
O
O
N
O
N
N
H
N
H
(S)-3
(S)-2
NA: not available.
N
F
a
Cl
O
O
These values are the mean SEM (n = 2–5).
The efficacy of the test compound relative to ( )-4.
b
O
O
N
H
N
H
(S)-4
(S)-5
opening efficacy was determined from the concentration depen-
dent increase in the baseline current measured prior to compound
addition. Compound effects were expressed as the percent of com-
pound free control current using current responses measured at
À40 mV. The data at À40 mV were used for preliminary struc-
ture–activity relationship purpose as this membrane potential is
most likely related to physiological conditions. The EC50’s shown
in Table 1 were calculated from the concentration response rela-
tionship of the test compounds. The efficacy of the test compound
was normalized to the efficacy of ( )-4 (reference compound) to
provide an E/Eref ratio (Table 1). The ratio was calculated by divid-
ing the maximum current amplitude produced by the test com-
pound by the maximum current amplitude produced by the
reference compound. The benefit of using an E/Eref value is that it
allows ranking of compound efficacies independent of the amount
of basal KCNQ2 current expression. Compounds which give a high-
er E/Eref ratio are more efficacious KCNQ2 channel openers.
Whole-cell patch-clamp data demonstrates that both acryla-
mides (S)-6 and (S)-7 are very potent and efficacious openers of
KCNQ2 channels. As shown in Table 1, (S)-6 has an improved po-
tency over (S)-3 and retigabine by a factor 25- and 542-fold,
respectively. Compound (S)-7 is even twofold more potent than
(S)-6 with an IC50 in the 1 nM range. The E/Eref values of acrylamide
(S)-6, (S)-3 and retigabine are 2.2, 1.8 and 1.6, respectively. Thus,
compound (S)-6 is 22% and 38% more efficacious in opening KCNQ2
channels than (S)-3 and retigabine, respectively. Again, (S)-7 is 33%
more efficacious than (S)-6. In contrast with (S)-6 and (S)-7, (R)-6
and (R)-7 exhibited no appreciable KCNQ2 opener activity when
F
O
O
O
F
N
O
N
H
N
H
N
F
(S)-6
(S)-7
Chart 1.
to give the NHBoc derivative 9, which underwent concurrent de-
methylation and Boc deprotection with boron tribromide to pro-
vide (S)-3-(1-aminoethyl)phenol (10) in good yield. This amino-
phenol can also be obtained directly from amine
8 upon
exposure to boron tribromide, but the yield was significantly low-
er. Boc protection of the free amine in 10 gave the protected deriv-
ative 11, and the copper(I)-promoted O-arylation of 11 with 3-
bromopyridine and 4-bromopyridine furnished the biaryl ethers
12a and 12b, respectively. After deprotection of the Boc group,
the free amines 13a and 13b were coupled with 2-fluorocinnamic
acid and 2,5-difluorocinnamic acid under typical conditions to give
(S)-6 and (S)-7, respectively. Compounds (R)-6 and (R)-7 were pre-
pared in the same fashion as their opposite enantiomers except
that (R)-1-(3-methoxyphenyl)ethanamine was used as the starting
material.
The effects of test compounds on outward potassium currents
were determined using whole-cell patch-clamp recordings from
HEK 293 cells stably expressing cloned mouse (m) KCNQ2 channels
as described previously.9 Baseline currents were evoked by step-
ping from À80 mV holding potential to À40 mV. Test compounds
were rapidly applied to cells by local superfusion, and channel
tested at 10 lM. This high enantioselectivity is consistent with
what we observed with other acrylamides previously reported.8–12
Both (S)-6 and (S)-7 were evaluated in the Chung model of
peripheral neuropathy. In this model, the L5 and L6 spinal nerves
of male Sprague–Dawley (SD) rats were tightly ligated with a sur-
gical suture following the method of Kim and Chung.13 Two to five
weeks post surgery, these rats were placed on the unrestrained
mesh floor, and their hind paws were touched with graded von
Frey filaments until paw withdrawal occurred. The paw lift thresh-
old (grams) was obtained as a measure of the sensitivity of the
hind paw to mechanical stimulation. The control rat group which
underwent the same surgery but without the spinal nerve ligation
provided a paw withdrawal threshold of 15 g. In our studies, gaba-
pentin, a well known neuropathic pain medication, was used as a
positive control. Test compounds were injected via tail vein. Tactile
allodynia assessment was repeated 15, 30, 60 and 90 min following
injection. As shown in Figure 1, rats given a 10 mg/kg intravenous
(iv) dose of compound (S)-6 or 100 mg/kg of gabapentin had signif-
icantly higher mechanical thresholds than vehicle-treated controls.
The attenuation of neuropathic pain behavior induced by (S)-6
peaked 15 min after iv injection, while the impact of gabapentin
was steadier. A 3 mg/kg iv dose of (S)-6 was also evaluated, but
i
O
ii
O
OH
HN
Boc
H2N
H2N
10
8
9
OH
O
v
iii
iv
HN
Boc
HN
N
Boc
12a
12b (4-pyridyl)
(3-pyridyl)
11
O
vi
H2N
(S)-6 and (S)-7
N
13a
(3-pyridyl)
13b (4-pyridyl)
Scheme 1. Reagents and conditions: (i) Boc2O, Et3N, CH2Cl2, 0 °C, 100%; (ii) BBr3,
CH2Cl2, À78 °C, 100%; (iii) Boc2O, Et3N, CH2Cl2, 0 °C, 68%; (iv) 3- or 4-bromopyr-
idine, K2CO3, CuI, 115 °C, 48 h, 48%; (v) 4 N HCl, dioxane, 50 °C, 4 h, 89%; (vi) 2-
fluorocinnamic acid for (S)-6, 2,5-difluorocinnamic acid for (S)-7, EDACÁHCl, Et3N, rt,
76–79%.

6190
Y.-J. Wu et al. / Bioorg. Med. Chem. Lett. 23 (2013) 6188–6191
Figure 1. Effects of (S)-6 at 3 and 10 mg/kg (iv) and gabapentin at 100 mg/kg (iv) in
the Chung model. ⁄P <0.05, ⁄⁄P <0.01 versus PEG-400, Dunnett’s test. n = 8–10 per
group.
Figure 2. Effects of (S)-6 at 3 and 10 mg/kg (iv) and gabapentin at 200 mg/kg (iv) in
the diabetic neuropathy model. ⁄P <0.05, ⁄⁄P <0.01 versus PEG-400, Dunnett’s test.
n = 8–10 per group.
no robust activity was observed (Fig. 1). Unlike (S)-6, a 10 mg/kg iv
doses of (S)-7 resulted in only marginal reversal of mechanical
thresholds relative to vehicle-treated controls (data not shown).
To understand the peak effect of (S)-6 at 15 min and inactivity
of (S)-7, we undertook preliminary brain exposure studies. A
1 mg/kg iv administration of (S)-6 to male SD rats provided total
brain concentrations of 2.0 and 0.56 lM at 15 min and 1 h post
iv injection. These data appeared to be consistent with the peak ef-
fect observed at 15 min time point and much reduced effect at 1 h
time point in the Chung model (Fig. 1). Even though the total brain
drug levels for (S)-6 at 1 h post 10 mg/kg iv injection were substan-
tially higher than the IC50, its attenuation of neuropathic pain
behavior in the Chung model was reduced (Fig. 1), suggesting that
the drug has a high brain-bound fraction. The total brain concen-
Figure 3. Effects of (S)-6 at 10 mg/kg (iv) and gabapentin at 100 mg/kg (iv) in both
pain phases in the formalin test. Values represent average of flinches in phase I (0–
9 min) and II (10–60 min). ⁄⁄P <0.01 versus PEG-400, Dunnett’s test. n = 10 per
group.
trations of (S)-7 with 1 mg/kg iv dose were 0.16 lM at 15 min
curred at 1 min following injection of formalin and resulted from
direct stimulation of nociceptors. The second phase of pain re-
sponse (9–60 min period) occurred after a period of sensitization
(quiescent period) during which inflammation took place. Pain re-
sponse scoring involved counts of spontaneous paw flinches for a
60 min period by means of computer-controlled data collection.
Test compounds were injected via tail vein 2 min before (Fig. 3, left
graph) or 8 min after (Fig. 3, right graph) formalin injection. As
shown in the left graph of Figure 3, compound (S)-6 at 10 mg/kg
(iv) significantly reduced paw flinches in the first phase of the for-
malin test. In contrast, no reduction of paw flinches was observed
with gabapentin at 100 mg/kg (iv). Both (S)-6 at 10 mg/kg (iv) and
gabapentin at 100 mg/kg (iv) significantly decreased painful
behaviors in the second phase of the formalin test (Fig. 3, right
graph), thus indicating an effect of these compounds on central
sensitization.
We also evaluated (S)-6 in a locomotor activity assay to deter-
mine whether the results obtained in the pain models were influ-
enced by a possible change in spontaneous locomotor activity.
Thus, male SD rats were administered with compound (S)-6 at
10 mg/kg iv dose, which was shown to be efficacious in all the pain
models described as above. The spontaneous activity of naïve ani-
mals was measured by determining the length of time that a given
rat stayed on the rotating rod. We chose both 5 and 15 min after iv
injection because at both time points the anti-allodynic peak ef-
fects occurred in the Chung model. As shown in Figure 4, com-
pound (S)-6 did not decrease normal rat motor performance at
both time points as compared with the vehicle-treated control
group. Chlordiazepoxide (CDP, Librim™), a well known psycho-
and 42 nM, 12-fold less than those obtained with (S)-6. The poor
brain penetration of (S)-7 presumably resulted in lack of significant
activity in the Chung model. Free fraction data and/or CSF concen-
trations are needed to further understand the PK–PD relationship.
Compound (R)-6, an inactive KCNQ2 opener, was also tested in
the Chung model. As expected, it failed to reverse neuropathic pain
behavior at 10 mg/kg (iv) (data not shown), thus providing support
to the KCNQ2 channel opening of (S)-6 as the mechanism of action
responsible for its reduction of tactile allodynia in the Chung
model.
Acrylamide (S)-6 was also investigated in the diabetic neuropa-
thy model.14 The diabetic rats were prepared by an intraperitoneal
(ip) injection of streptozotocin (STZ, 75 mg/kg). STZ is an antibiotic
that selectively kills pancreatic b-cells that produce insulin, thus
leading to increased blood glucose and neuropathic pain behaviors.
Diabetes was confirmed three weeks after STZ injection by mea-
surement of tail vein blood glucose levels. The hypersensitivity of
the diabetic rats was significantly reversed by compound (S)-6 at
10 mg/kg (iv) and gabapentin at 200 mg/kg (iv) (Fig. 2). As opposed
to the reduction of tactile allodynia in the Chung model, the peak
reversal was more prolonged (15–60 min), and the 3 mg/kg dose
was active. These data suggest that the diabetic neuropathy model
appears somewhat more sensitive than the Chung model to mod-
ulation by this mechanism.
Finally, we subjected compound (S)-6 to the formalin test, a
well established model for persistent pain. In this model, a 2% solu-
tion of formalin was injected subcutaneously into the dorsal skin of
rat hind paw to produce a biphasic pain response over a test period
of 60 min. The initial pain response (phase I, 0–9 min period) oc-

Y.-J. Wu et al. / Bioorg. Med. Chem. Lett. 23 (2013) 6188–6191
6191
(poor brain exposure) suggests that activation of KCNQ2 channel in
the CNS produced by (S)-6 is the mechanism of action responsible
for its anti-allodynic effect. However, application of selective
KCNQ2 blockers and/or antisense experiments are required in or-
der to complete the validation. Validation should also include iden-
tification of KCNQ heteromers (likely KCNQ2/3) involved in
neuropathic pain pathophysiology.
References and notes
1. Montfort, E. G.; Amy Witte, A. P.; Ward, K. US Pharm. 2010, 35, HS8–HS15.
2. Backonja, M.; Woolf, C. J. Oncologist 2010, 15, 24–29.
3. Gribkoff, V. K. Expert Opin. Ther. Targets 2008, 12, 565–581.
4. Munro, G.; Dalby-Brown, W. J. Med. Chem. 2007, 50, 2576–2582.
5. Gunthorpe, M. J.; Large, C. H.; Sankar, R. Epilepsia 2012, 53, 412–424.
6. Blackburn-Munro, G.; Jensen, B. S. Eur. J. Pharmacol. 2003, 460, 109.
7. Fritch, P. C.; McNaughton-Smith, G.; Amato, G. S.; Burns, J. F.; Eargle, C. W.;
Roeloffs, R.; Harrison, W.; Jones, L.; Wickenden, A. D. J. Med. Chem. 2010, 53,
887–896.
Figure 4. Effect of (S)-6 at 10 mg/kg (iv) and chlordiazepoxide (CDP) at 40 mg/kg
(iv) on the spontaneous locomotor activity in SD male rats. Results are expressed as
the total time (sec) spent on the rotating rod S.E.M. ⁄P <0.01 versus PEG-400,
Dunnett’s test. n = 11–12 per group.
8. Wu, Y.-J.; Boissard, C. G.; Greco, C.; Gribkoff, V. K.; Harden, D. G.; He, H.;
L’Heureux, A.; Kang, S. H.; Kinney, G. G.; Knox, R. J.; Natale, J.; Newton, A. E.;
Lehtinen-Oboma, S.; Sinz, M. W.; Sivarao, D. V.; Starrett, J. E., Jr.; Sun, L.;
Tertyshnikova, S.; Thompson, M. W.; Weaver, D.; Wong, H. S.; Zhang, L.;
Dworetzky, S. I. J. Med. Chem. 2003, 46, 3197–3200.
9. Wu, Y.-J.; He, H.; Sun, L.; L’Heureux, A.; Chen, J.; Dextraze, P.; Starrett, J. E., Jr.;
Boissard, C. G.; Gribkoff, V. K.; Natale, J.; Dworetzky, S. I. J. Med. Chem. 2004, 47,
2887–2896.
10. Wu, Y.-J.; Boissard, C. G.; Chen, J.; Fitzpatrick, W.; Gao, Q.; Gribkoff, V. K.;
Harden, D. G.; He, H.; Knox, R. J.; Natale, J.; Pieschl, R. L.; Starrett, J. E., Jr.; Sun,
L.; Thompson, M.; Weaver, D.; Wu, D.; Dworetzky, S. I. Bioorg. Med. Chem. Lett.
2004, 14, 1991–1995.
11. Wu, Y.-J.; Sun, Li.; He, H.; Chen, J.; Starrett, J. E., Jr.; Dextraze, P.; Daris, J.;
Boissard, C. G.; Pieschl, R. L.; Gribkoff, V. K.; Natale, J.; Knox, R. J.; Harden, D. G.;
Thompson, M. W.; Fitzpatrick, W.; Weaver, D.; Wu, D.; Gao, Q.; Dworetzky, S. I.
Bioorg. Med. Chem. Lett. 2004, 14, 4533–4537.
12. L’Heureux, A.; Martel, A.; He, H.; Chen, J.; Sun, L.; Starrett, J. E., Jr.; Natale, J.;
Dworetzky, S. I.; Knox, R. J.; Harden, D. G.; Weaver, D.; Thompson, M. W.; Wu,
Y.-J. Bioorg. Med. Chem. Lett. 2005, 15, 363–366.
13. Kim, S. H.; Chung, J. M. Pain 1992, 50, 355–363.
14. Tsech, G. H.; Allen, T. J. Nephrology (Carlton) 2007, 12, 261–266.
15. Weaver, C.D. PCT Int. Appl. 2002, WO 0231508A1.
16. Benzen, B. H.; Schmitt, N.; Calloe, K.; Brown, W. D.; Grunnet, M.; Olesen, S.
Neuropharmacology 2006, 51, 1068–1077.
17. For a relatively selective KCNQ2 and KCNQ4 channel opener with moderate
potency, see: Yu, H.; Wu, M.; Townsend, S. D.; Zou, B.; Long, S.; Daniels, J. S.;
McManus, O. B.; Li, M.; Lindsley, C. W.; Hopkins, C. R. ACS Chem. Neurosci. 2011,
2, 572–577.
stimulant drug, was used as our positive control, and it signifi-
cantly decreased the locomotor activity at 40 mg/kg (iv). Thus,
we concluded that the anti-allodynic effects produced by com-
pound (S)-6 were not confounded by locomotor effects.
Preliminary studies using a thallium(I) influx assay15 showed
that both (S)-6 and (S)-7, like other members of acrylamides re-
ported previously from this laboratory,8–12 also activated other
members of the KCNQ family.16,17 Further electrophysiological
characterization is required in order to understand the activation
of other KCNQ channels.
Compound (S)-6 was identified as a potent and efficacious
KCNQ2 opener. It exhibited superior activity to gabapentin in
phase 1 of the formalin test, indicating a possible role of KCNQ2
in acute pain and thus potential analgesic effects of KCNQ2 open-
ers. The anti-hyperalgesic effect of (S)-6 in phase 2 of the formalin
test correlates with the implication of KCNQ2 in central sensitiza-
tion hypothesis. This compound also demonstrated significant
activity in two models of neuropathic pain: a spinal nerve ligation
model (Chung model) and a diabetic neuropathy model. The lack of
anti-allodynic effects of both (R)-6 (inactive enantiomer) and (S)-7