Synthesis and structure-activity relationship study of substituted caffeate esters as antinociceptive agents modulating the TREK-1 channel

Article abstract of DOI:10.1016/j.ejmech.2014.01.049

The TWIK-related K⁺ channel, TREK-1, has recently emerged as an attractive therapeutic target for the development of a novel class of analgesic drugs. It has been reported that TREK-1 -/- mice were more sensitive than wild-type mice to painful stimuli, suggesting that activation of TREK-1 could result in pain inhibition. Here we report the synthesis of a series of substituted caffeate esters (12a-u) based on the hit compound CDC 2 (cinnamyl 3,4-dihydroxyl-α-cyanocinnamate). These analogs were evaluated for their ability to modulate TREK-1 channel by electrophysiology and for their in vivo antinociceptive activity (acetic acid induced-writhing assay) leading to the identification a series of novel molecules able to activate TREK-1 and displaying potent analgesic activity in vivo.

Full text of DOI:10.1016/j.ejmech.2014.01.049

European Journal of Medicinal Chemistry 75 (2014) 391e402
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Synthesis and structureeactivity relationship study of substituted
caffeate esters as antinociceptive agents modulating the TREK-1
channel
,
,
,
,e
Nuno Rodrigues a b, Khalil Bennis b c, Delphine Vivier a b, Vanessa Pereira d
,
Franck C. Chatelain f, Eric Chapuy d e, Hemantkumar Deokar b c, Jérôme Busserolles d
,
,
,
,e
Florian Lesage f, Alain Eschalier d g, Sylvie Ducki b
,
e
,
,c,
*
a Clermont Université, Université Blaise Pascal, Institut de Chimie de Clermont-Ferrand, BP 10448, F-63000 Clermont-Ferrand, France
b CNRS, UMR6296, ICCF, F-63171 Aubière, France
c Clermont Université, ENSCCF, Institut de Chimie de Clermont-Ferrand, BP 10448, F-63000 Clermont-Ferrand, France
d Clermont Université, Université d’Auvergne, Pharmacologie Fondamentale et Clinique de la Douleur, BP 10448, F-63000 Clermont-Ferrand, France
e Inserm, UMR1107, Neuro-Dol, F-63001 Clermont-Ferrand, France
f Labex ICST, Institut de Pharmacologie Moléculaire et Cellulaire, UMR CNRS 7275, Université de Nice Sophia Antipolis, F-06560 Valbonne, France
g CHU Clermont-Ferrand, Service de pharmacologie, F-63003 Clermont-Ferrand, France
a r t i c l e i n f o
a b s t r a c t
The TWIK-related Kþ channel, TREK-1, has recently emerged as an attractive therapeutic target for the
development of a novel class of analgesic drugs. It has been reported that TREK-1 ꢀ/ꢀ mice were more
sensitive than wild-type mice to painful stimuli, suggesting that activation of TREK-1 could result in pain
inhibition. Here we report the synthesis of a series of substituted caffeate esters (12aeu) based on the hit
Article history:
Received 16 October 2013
Received in revised form
13 January 2014
Accepted 19 January 2014
Available online 31 January 2014
compound CDC 2 (cinnamyl 3,4-dihydroxyl-a-cyanocinnamate). These analogs were evaluated for their
ability to modulate TREK-1 channel by electrophysiology and for their in vivo antinociceptive activity
(acetic acid induced-writhing assay) leading to the identification a series of novel molecules able to
activate TREK-1 and displaying potent analgesic activity in vivo.
Keywords:
TREK-1 channel
Pain
Ó 2014 Elsevier Masson SAS. All rights reserved.
Antinociceptive
Caffeate esters
Structureeactivity relationship study
QSAR
Analgesic agents
1. Introduction
associated with nociception [5,6]. The functionality of the TREK-1
channels results from their sensitivity to a variety of stimuli. They
Ion channels are major drug targets whose study can contribute
to the understanding of the physiopathology of pain and the evo-
lution of therapeutic analgesics [1e3]. The TREK-1 (TWIK-related
Kþ channels 1) channel belongs to the two pore-domain potassium
(K2P) channel family and was first identified in 1998 [4]. Human
TREK-1 is highly expressed in the peripheral sensory system,
particularly in small dorsal root ganglion (DRG) neurons that are
are gated by membrane stretching (mechano-activation), pH
(intracellular acidosis), temperature (heat), polyunsaturated fatty
acids (PUFAs) such as arachidonic acid, and some general volatile
anesthetics (chloroform, diethyl ether, and nitrous oxide). They can
also be down-modulated by PKA/PKC phosphorylation pathways
and tonically inhibited by the actin cytoskeleton [7]. Structurally,
the TREK-1 channels are transmembrane (TM) proteins with
intracellular N- and C-termini [8]. The pore for Kþ conduction re-
sults from the assembly of two dimers, each containing four helical
transmembrane segments (TM1eTM4), arranged in tandem with
two pore domains (Fig. 1).
* Corresponding author. Clermont Université, ENSCCF, Institut de Chimie de
Clermont-Ferrand, BP 10448, F-63000 Clermont-Ferrand, France. Tel.: þ33 (0)
473407132; fax: þ33 (0)473407008.
Recent studies have reported that mice with a disrupted TREK-1
gene (TREK-1 ꢀ/ꢀ) were more sensitive than wild-type mice to
E-mail address: Sylvie.Ducki@ensccf.fr (S. Ducki).
0223-5234/$ e see front matter Ó 2014 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ejmech.2014.01.049

392
N. Rodrigues et al. / European Journal of Medicinal Chemistry 75 (2014) 391e402
Fig. 1. Topology of TREK-1 channel.
Fig. 3. Antinociceptive effect of compounds 1e5 (10 mg/kg, s.c.) against acetic acid-
induced abdominal writhes in mice (n ¼ 8). Each bar represents the mean ꢁ SEM of
eight experimental values. The percentage (%) of writhes inhibition (as compared to
vehicle) is given above each bar and in Table 1. t-Test: *ꢂ0.05, **ꢂ0.01, ***ꢂ0.001.
heat and mechanical stimulations [9,10], suggesting that TREK-1
channels are important in pain perception. TREK-1 channel could
represent a novel molecular target of interest for the discovery of a
new class of analgesics [11].
13.9% and 33.6% inhibition. This result correlated with the ability of
these agents to activate TREK-1, since 10
increase TREK-1 current by 7-fold compared to 40
These results suggest that the distance between the two aromatic
rings and/or the substitution at the -position of the ester could be
m
M of CDC is required to
mM of CAPE [15].
2. Results and discussion
a
2.1. Identification of hit molecules
an important feature for optimal pharmacological activity.
Although it has been found to possess high analgesic activity,
riluzole 1 is non-specific for TREK-1 activation. The drug has been
widely studied and many pharmacological activities have been
associated with this molecule [19e22]. However no bioactivity data
have been reported on CDC 2, which also displays a good drug-like
profile (Fig. 4A and B) [23]. We further evaluated the analgesic
activity of CDC 2 in the acetic acid-induced writhing assay and
determined an ID50 of 8.0 mg/kg (Fig. 4C). No noticeable sedative or
adverse effect was observed at the doses tested.
A literature survey allowed us to identify several small organic
molecules able to module the TREK-1 channels. Riluzole 1 (Fig. 2), a
neuroprotective agent currently used in the treatment of amyo-
trophic lateral sclerosis [12], was reported as an activator of TREK-1
[13,14]. This activation was reported to be transient followed by an
inhibition, process attributable to an increase in the intracellular
cAMP concentration by riluzole that produces a PKA-dependant
inhibition of TREK-1. Cinnamyl 3,4-dihydroxyl-a-cyanocinnamate
(CDC 2, Fig. 2) and caffeic acid phenylethyl ester (CAPE 3, Fig. 2)
were also reported as TREK-1 activators [15]. Danthi et al. carried
out a structureeactivity relationship (SAR) study but failed to
improve the gating activity of the caffeate esters. Because of the
reported activities of 2 and 3, we also decided to consider the
CAPEeCDC hybrid molecule 4 [16,17]. Finally, Danthi et al. also re-
ported that polyunsaturated fatty acids, such as linoleic acid 5
(Fig. 2), were able to activate TREK-1 channels [18].
The antinociceptive activity of compounds 1e5 (10 mg/kg,
administered subcutaneously) was evaluated by using the acetic
acid-induced writhing test in mice. Compared to the controlled
mice (vehicle), all the compounds were able to inhibit the induced
abdominal writhes (Fig. 3, Table 1). The most active compounds
were found to be riluzole 1 (63.6% inhibition) while linoleic acid 5
showed a moderate analgesic activity (29.6% inhibition). CDC 2
showed potent analgesic activity (50.8% inhibition) while CAPE 3
and CDCeCAPE-hybrid 4, which are structurally similar to 2,
showed low to moderate antinociceptive effect with respectively
2.2. Synthesis of substituted caffeate esters
Since no information was available about the binding of CDC 2 to
the TREK-1 channel, and the 3D structure of TREK-1 was not
available at the time, [8] we decided to perform hit optimization by
conventional structureeactivity relationship (SAR) studies around
the general structure 12, studying the influence of various sub-
stituents (R1eR3, X) on the pharmacological activity of the analogs.
We developed two routes to obtain a-substituted cinnamate esters
12 (Scheme 1). Condensation of aldehyde 9 with carbonyl com-
pound 8, which could be obtained from the carboxylic acid 7 and
synthon 6 (route A), would lead to analogs 12 with variations in R1,
R2 and X. Alternatively esterification of cinnamic acid derivatives 10
with synthon 6, the
a,b-unsaturated acid emanating from a
condensation between aldehyde 9 and carboxylic acid 7 (route B)
would lead to analogs 12 bearing different R3 substituents.
Fig. 2. The chemical structures of compounds 1e5.

N. Rodrigues et al. / European Journal of Medicinal Chemistry 75 (2014) 391e402
393
Table 1
yield under acidic conditions. The corresponding analogs 12aef
and 12heq were obtained by a Knoevenagel condensation under
basic conditions between esters 8ael and the appropriate benzal-
dehydes 9aed in moderate 30% to excellent 99% yields (Scheme
2B). Acetylation of CDC 2 yielded analog 12g (80% yield).
Analogs 12reu were obtained through route B (Scheme 3). A
WadswortheHorner olefination [24] between the MOM-protected
benzaldehyde 9e and the appropriately substituted phosphonates
7b,c afforded, after saponification under basic conditions, the
desired MOM-protected caffeic acids 10cee in 47 and 74% yield
respectively. These acids were coupled to cinnamyl alcohol 6a and
the corresponding esters were deprotected under acidic conditions
to give analogs 12r,s in moderate yields (21 and 46%). Caffeic acid
10e, obtained in 21% yield from phosphonate 7c through an olefi-
nation/reduction/oxidation sequence, was coupled to 6d to yield
analog 12t in 64% yield after deprotection. Finally, 10e was depro-
tected under acid conditions to afford 12u in 40% yield.
Antinociceptive activity of compounds 1e5 (10 mg/kg, s.c.) against
acetic acid-induced abdominal writhes in mice (n ¼ 8).
Compounds
% Writhing inhibition
1
2
3
4
5
63.6 ꢁ 11.1
50.8 ꢁ 5.5
13.9 ꢁ 9.4
33.6 ꢁ 6.5
29.6 ꢁ 8.1
Analogs 12aep were prepared through route A (Scheme 2). The
appropriately substituted alcohols 6aek were reacted with cya-
noacetic acid 7a under coupling conditions (DCC, DMAP, DCM) [17]
to yield esters 8aek in good to excellent yields (35e98%, Scheme
2A). Alcohol 6a was subjected to a bromination/alkylation/addi-
tioneelimination sequence to give ketone 8l (41% yield over three
steps) and diol 8m resulted from the opening of epoxide 8e in 50%
Fig. 4. (A) Drug-like properties of CDC 2. (B) Radar-plot of drug-like properties of CDC 2. The blue and red lines represent “drug-like” property space [23], and the green line
2
ꢀ
represents the predicted values for CDC 2. Log P, log of calculated partition coefficient; MW, molecular weight; PSA, polar surface area in A ; log S, log of predicted solubility score;
nb rot bond, number of rotatable bonds; HBA, number of hydrogen-bond acceptors; HBD, number of hydrogen-bond donors; n atoms, number of heavy atoms. (C) Antinociceptive
dose response of CDC 2 given subcutaneously against acetic acid-induced abdominal writhes in mice (n ¼ 8). Each bar represents the mean ꢁ SEM of eight experimental values. t-
Test: *ꢂ0.05, **ꢂ0.01, ***ꢂ0.001. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Esterification
Condensation
ROUTE A
ROUTE B
when X = O
Knovenagel
Coupling
condensation
O
O
O
O
R2
R2
R1
R1
R1
R1
R1
Y
OH
H
X
6
X
+
+
R3
10
R3
12
R3
R1
9
8
Wittig
condensation
Esterification
O
O
R2
Y
Y
9
+
OH
+
OH
HX
R3
R3
6
7
7
Scheme 1. Retrosynthetic analysis of 12.

394
N. Rodrigues et al. / European Journal of Medicinal Chemistry 75 (2014) 391e402
A
O
O
R2
i.
R2
NC
ii.
HX
6a-k
X
Ph
6a
8e
CN
8a-k (35-98%)
8l (41%)
a, R2=-CH2-CH=CH-Ph, X=O
b, R2=-CH2-CH2-Ph, X=O
O
OH
c, R2=-CH2-CH=CH-Ph, X=S
d, R2=-CH2-CH=CH-Ph, X=NH
e, R2=-CH2-CH-O-CH-Ph, X=O
f, R2=3-phenylprop-2-ynyl, X=O
g,R2= 3-(thiophen-3-yl)prop-2-ynyl, X=O
h, R2=(1H-indol-2-yl)methyl , X=O
i, R2=-CH(Et)-CH=CH-Ph, X=O
j, R2= -Et, X=O
iii.
O
Ph
CN
OH
8m (50%)
k, R2= -tBu, X=O
Conditions: i) α-cyanoacetic acid 7a, DCC, DMAP, DCM, 2-24h, 0 °C-RT. ii) a. CBr4, PPh3, MeCN, 12 h, RT. b.
EtOAc, LDA, CuI, THF, 90 min, -30 °C. c. MeCN, n-BuLi, THF, 12 h, -78 °C to RT; iii) H2SO4, THF, H2O, 4 h, RT.
O
O
R2
R2
B
R1
R1
i.
X
X
CN
CN
12a (50% from 8c + 9a), R1 = OH, R2=-CH2-CH=CH-Ph, X=S
12b (76% from 8d + 9a), R1 = OH, R2=-CH2-CH=CH-Ph, X=NH
12c (76% from 8l + 9a), R1 = OH, R2=-CH2-CH=CH-Ph, X=-CH2
12d (46% from 8a + 9b), R1 = H, R2=-CH2-CH=CH-Ph, X=O
12e (64% from 8a + 9d), R1 = Cl, R2=-CH2-CH=CH-Ph, X=O
12f (80% from 8a + 9c), R1 = Me, R2=-CH2-CH=CH-Ph, X=O
12h (30% from 8m + 9a), R1 = OH, R2=-CH2-CH-(OH)-CH(0H)-Ph, X=O
12i (35% from 8e + 9a), R1 = OH, R2=-CH2-CH-O-CH-Ph, X=O
12j (40% from 8f + 9a), R1 = OH, R2=3-phenylprop-2-ynyl, X=O
12k (41% from 8g + 9a), R1 = OH, R2= 3-(thiophen-3-yl)prop-2-ynyl, X=O
12l (40% from 8h + 9a), R1 = OH, R2=(1H-indol-2-yl)methyl , X=O
12m (99% from 8i + 9a), R1 = OH, R2=-CH(Et)-CH=CH-Ph, X=O
12n (96% from 8b + 9a), R1 = OH, R2=-CH2-CH2-Ph, X=O
12o (71% from 8j + 9a), R1 = OH, R2= -Et, X=O
8a-m
O
R1
R1
H
9a, R1 = OH
9b, R1 = H
9c, R1 = Cl
9d, R1 = Me
12p (48% from 8k + 9a), R1 = OH, R2= -tBu, X=O
iii.
12q (90% from 12q), R1 = OH, R2= H, X=O
O
AcO
AcO
ii.
O
2
CN
Ph
12g (80%)
Conditions: i.. 9a-d, piperidine, DCM, 12 h, RT (except for 12a: DABCO, THF, 2 days, RT). ii. Ac2O, TEA, DCM,
4 h, 0 °C. iii) TFA, 1 h, RT
Scheme 2. Preparation of 12aeq.
2.3. Pharmacological evaluation
X ¼ CH2) resulted in a diminished ability of these compounds to
increase TREK-1 current density. We note that this correlated with a
decrease in pain inhibition (from 50.8% for 2 to 26.5, 12.3 and 15.0%
for 12aec respectively).
TREK-1 was expressed in Xenopus oocytes and the effect of
compounds 12aeu (20
mM) on channel activity was monitored
using the double microelectrode voltage-clamp technique as pre-
viously described [25]. Most compounds were able to activate
TREK-1 currents at 20 mM and in all cases, the activation was rapidly
reversed following washout of the tested compounds (Table 2). The
antinociceptive activity of compounds 12aeu (10 mg/kg adminis-
tered subcutaneously) was initially evaluated by using the acetic
acid induced-writhing test in mice [26,27]. Apart from two com-
pounds (12l and 12r), the other analogs inhibited the induced
abdominal constrictions in the range 5e50%.
Next, we varied the substitution R1 on the aromatic ring (12de
g). Substitution of the hydroxyl groups in CDC 2 (R1 ¼ OH) by
hydrogen atoms (12d, R1 ¼ H), electron-donating groups (12e,
R1 ¼ CH3) or electron-withdrawing atoms (12f, R1 ¼ Cl) led to
insoluble analogs which could not be evaluated by electrophysi-
ology and their analgesic activity remained moderate (18e31%).
CDC 2 (R1 ¼ OH) was acetylated to give analog 12g (R1 ¼ OAc)
which resulted in a total loss of TREK-1 activation with retention of
a moderate analgesic activity (29%).
The replacement of the ester function of CDC 2 (X ¼ O) by a
Our attention then turned towards the R2 side-chain (12heq).
Oxidation of the alkene 2 (R2 ¼ eCH2eCH]CHePh) into the
thioester (12a, X ¼ S), an amide (12b, X ¼ NH) or a ketone (12c,

N. Rodrigues et al. / European Journal of Medicinal Chemistry 75 (2014) 391e402
395
O
O
O
O
R1
R1
R1
R1
i. or ii
(EtO)2P
H
OH
OEt
+
R3
R3
10c (47% from 7b + 9e), R1 = -OMOM, R3 = CH3
10d (74% from 7c + 9e), R1 = -OMOM, R3 = CH2-NHBoc
10e (21% from 7c + 9f), R1 = -OTBDMS, R3 = CH2-NHBoc
7b, R3 = CH3
7c, R3 = CH2-NHBoc
9e, R1 = -OMOM
9f, R1 = -OTBDMS
O
R1
R1
iii. or iv.
Ph
X
Ph
10c-e
+
HX
R3
12r (46% from 10c + 6a), R1 = -OH, R3 = CH3, X = O
12s(21% from 10d + 6a), R1 = -OH R3 = CH2-NH3+Cl-, X = O
12t (60% from 10e + 6d), R1 = -OH, R3 = CH2-NH3+TFA-, X = NH
6a, X = O
6d, X = NH
O
HO
HO
v.
OH
NH3+ Cl-
12u (40%)
10e
Conditions: i. 10c-d : a. NaH, THF, RT, 2 days; b. LiOH, H2O, MeOH, RT, 12 h. ii. 10e : a. NaH, THF, 4 h, RT.
b. LAH, THF, 4 h, 0 °C. c. MnO2, THF, 18 h, RT. d. NaH2PO4.H20, NaClO2. iii. 12r-s: a. 6a, DCC, DMAP, DCM, 2
h, RT b. 10% HCl, MeOH, 30 min, reflux. iv. 12t : a. 6d, DCC, DMAP, DCM, 2 days, RT. b. TBAF, THF, 3 h, RT.
c. TFA, DCM, 30 min, RT. v. HCl, MeOH, RT, 2 days.
Scheme 3. Preparation of 12reu.
corresponding diol 12h or epoxide 12i, chain contraction into 12n
(R2 ¼ eCH2eCH2ePh) or chain extension to give 12m (R2 ¼ e
CH(Et)eCH]CHePh) led to no improvement in channel activation
while the analgesic effect was maintained for 12i and 12m. Further
variation of the R2 side-chain included substitution into an alkyne
chain (12j,k, R2 ¼ eCH2eCH^CHeAr) which retained a potent
ability to activate TREK-1 channel accompanied by only a moderate
anti-nociceptive activity. Rigidification of the phenylethylene side-
chain of 2 into an indole 12l was detrimental, suggesting that a
certain degree of flexibility is required to fit into the binding site. On
the other hand, simplification of CDC 2 into the corresponding ethyl
ester 12o or even a carboxylic acid 12q resulted in either increase or
retention of TREK-1 or analgesic activity.
The objective was to develop ligand-based pharmacophore hy-
potheses (by collecting the common features distributed in a 3D
space which represent groups in a molecule that could participate
in important interactions with its target) and derive an-atom based
three-dimensional quantitative structure activity relationship (3D-
QSAR) using PHASE (Pharmacophore Alignment and Scoring En-
gine) [28,29] module from Schrodinger Suite 2010 (to identify
atomic features which are responsible for biological activity of
novel TREK-1 channel activators).
A total of six pharmacophore hypotheses were generated upon
completion of the pharmacophore identification process including
AAADD, AAADR, AAARR, AADDR, AADRR, ADDRR. We only selected
hypotheses whose scores ranked in the top 1%. Each hypothesis
alignment was then used for the atom-based 3D-QSAR model
development. The pharmacophore hypothesis with one hydrogen-
bond acceptor (A), two hydrogen-bond donors (D) and two aro-
matic features (R) named ADDRR (Fig. 5) yielded a statistically
significant atom-based 3D-QSAR model (Table 4) with an excellent
correlation (R2 ¼ 0.885) for the training set (Fig. 6A) and an
excellent predictive power (Q2 ¼ 0.808) for test set (Fig. 6B).
Structural insights into the activity can be gained by visualizing
the 3D-QSAR model applied to individual molecules for each
chemical characteristic (Fig. 7). In the context of hydrogen-bond
donor (D), the 3D-QSAR model clearly indicates that the intro-
duction of alcohols on the ester alkyl chain (red cubes (in web
version)) is not favorable for activity (Fig. 7A, compounds 2 and
12h). This is confirmed when these molecules are compared in
terms of hydrophobicity (H) (Fig. 7B, molecules 2 and 12h), sug-
gesting that lipophilicity (blue cubes (in web version)) is preferred.
The model also suggests that the spatial orientation of the catechol
Analogs 12reu varied at the
a
-position of the carbonyl (R3). We
-methyl (12r),
replaced the -nitrile of the caffeic ester 2 by an
a
a
reduced into the corresponding methylammonium salt 12s only to
find out that these analogs were only moderately able to activate Kþ
currents. One compound (12r) even worsened the writhes induced
by acetic acid while the other (12s) retained some ability to reduce
the number of writhes induced by acetic acid. Finally among the
two compounds (12teu) possessing two structural modifications
(R2, R3) compared to the hit compound 2, the simplified caffeic acid
12u was able to increase TREK-1 current by almost 3-fold and
displayed the most promising anti-nociceptive activity.
2.4. Pharmacophore identification and atom-based 3D-QSAR
Fifteen molecules with determined TREK activity (Table 3) were
selected and divided into a training set (10 molecules) and a test set
(5 molecules). The training set molecules were selected in such a
way that they contained information in terms of both their struc-
tural features and biological activity ranges. The logarithm of the
determined RTREK-1 (pRTREK-1) was used in this study. The highly
active molecules (pRTREK-1 > 0.3), moderately active (0 < pRTREK-
moiety (R4) in relation to the
a,b-unsaturated ester (A1) is a critical
feature for high activity (blue cubes (in web version)) (Fig. 7B, an-
alogs 2, 12r, 12k). In fact, when the molecules are visualized in the
context of their electron withdrawing density (W) (Fig. 7C, com-
pounds 2, 12c, 12h, 12l), we observed that this feature is important
and depends strongly on the spatial orientation of the nitrile (C^N)
< 0.3), and inactive molecules (pRTREK-1 < 0) were included to
1
spread out the range of activities.

396
N. Rodrigues et al. / European Journal of Medicinal Chemistry 75 (2014) 391e402
Table 2
Pharmacological effect of compounds 2, 12aeu. a) Modulation of TREK-1 currents expressed in Xenopus oocytes as measured by the double microelectrode voltage-clamp
technique. RTREK-1 is the ratio of the currents measured after and before the application of the compounds 2, 12aeu (20 M). All the currents were recorded at 0 mV and
m
at the steady state. b) Antinociceptive effect of compounds 2, 12aeu (10 mg/kg, s.c.) against acetic acid-induced abdominal writhes in mice (n ¼ 8). % Inhibition is the difference
between the number of writhes observed in presence and absence of compounds 2, 12aeu (10 mg/kg, s.c.).
Compounds
X
O
R1
R2
R3
RTREK-1
% Inhibition
2
OH
CN
2.64
50.8***
12a
12b
12c
12d
12e
12f
S
OH
OH
OH
H
CN
CN
CN
CN
CN
CN
CN
1.41
1.58
1.71
NSa
26.5*
12.3*
15.0*
25.1*
18.4*
30.7*
29.0*
NH
CH2
O
O
Me
Cl
NSa
O
NSa
12g
O
OAc
0.99
12h
12i
12j
O
O
O
OH
OH
OH
CN
CN
CN
0.75
1.47
2.21
24.1**
49.7**
22.0*
12k
O
OH
CN
2.38
5.0*
12l
O
O
OH
OH
CN
CN
1.00
1.04
<0b
12m
45.3*

N. Rodrigues et al. / European Journal of Medicinal Chemistry 75 (2014) 391e402
397
Table 2 (continued )
Compounds
X
O
R1
R2
R3
RTREK-1
% Inhibition
12n
OH
CN
1.28
17.0*
12o
12p
12q
O
O
O
OH
OH
OH
Et
t-Bu
H
CN
CN
CN
2.76
1.17
1.23
21.8*
43.4
31.8**
12r
12s
O
O
OH
OH
CH3
1.48
1.27
<0b
CH2NHþ3 Clꢀ
36.4**
12t
NH
NH
OH
OH
CH2NHþ3 TFAꢀ
CH2NHþ3 Clꢀ
1.05
50.0*
12u
H
2.87
50.1***
t-Test: *ꢂ0.05, **ꢂ0.01, ***ꢂ0.001.
a
NS: not determined because of poor solubility of compound.
<0. Compound resulted in increased number of writhes compared to vehicle.
b
and ester (C]O) functions, suggesting the S-trans conformation
adopted by the -unsaturated ester is preferred (Fig. 7C, com-
pound 2, blue cubes (in web version) for C^N and C]O).
the other hand, compounds 12i and 12t display potent analgesic
activity (w50%) but are unable to activate the Kþ channel (RTREK-
1 <1.5), suggesting their pharmacological activity is independent or
partially dependent on TREK-1 targeting. Finally one analog 12u
proved to be the most potent with high ability to activate TREK-1
channels (RTREK-1 ¼ 2.87) and potent analgesic activity (50.1%
a,b
3. Conclusion
writhes inhibition) at 10 mg/kg. This
a-methyl ammonium
We have identified CDC 2 as a lead compound for the devel-
opment of novel analgesic agents targeting TREK-1 channels. A
structureeactivity relationship study led us to prepare 21 analogs
(12aeu), bearing various R1, R2, R3 and X moieties. We developed
two synthetic routes to access the desired molecules in 3e12 steps
in 3e77% overall yields. Route A provided us with three analogs
bearing different X atoms (eOe 2, eSe 12a, eNHe 12b, eCH2e
12c), four analogs with various aromatic substitutions R1 (eOH 2, e
H 12d, eCH3 12e, eCl 12f, eOAc 12g), and ten analogs bearing R2
side-chains (12a, 12heq), while route B led us to three analogs with
various R3 substitution (eCN 2, eCH3 12r, eCH2eNHþ3 12seu). We
note that compounds 12j, 12k and 12o activate TREK-1 channel
with RTREK-1 > 2 but display little analgesic activity (5e22%), sug-
gesting that they may not be able to reach their target in vivo. On
substituted caffeic acid could be a potential metabolite of the lead
compound 2. Moreover, this SAR study allowed us to generate a
highly predictive atom-based 3D-QSAR model (PHASE) which
consists in a five-point pharmacophore hypothesis with one
hydrogen-bond acceptor (A), two hydrogen-bond donors (D) and
two aromatic features (R). The developed atom-based 3D-QSAR
model can provide insights into the structural requirement of novel
caffeic acid analogs as activators of TREK-1 channels.
4. Experimental section
4.1. Chemistry
Solvents were distilled prior to use. Other reagents were used as
received. Thin layer chromatography analysis (TLC) was performed
on pre-coated aluminum backed silica plates Kieselgel 60 F254
(Merck). Spots were visualized under UV light (254 nm) before
revealing in an ethanolic solution of phosphomolybdic acid (heat-
ing). Purification by column chromatography was carried out on
silica gel (70e230 mesh). Melting points were measured by a
Reichert plate-heating microscope. 1H and 13C NMR spectra were
Table 3
Compounds in training and test sets with observed activity (pRTREK-1 obs), number of
conformations generated (Nb conf), chemical characteristics (hydrogen-bond
acceptor A, hydrogen-bond donor D, hydrophobic/non-polar H, negative ionic N,
positive ionic P, aromatic R), and predicted activity for pharmacophore hypothesis
ADDRR (pRTREK-1 pred).
Compounds QSAR set pRTREK-1 obs Nb conf
A
D
H
N
P
R
pRTREK-1 pred
recorded on
100.61 MHz respectively. Chemical shifts d
a
Brücker Avance spectrometer at 400.13 and
are reported in ppm
2
Training
Test
Test
Training
Test
0.422
0.149
0.199
0.233
ꢀ0.004
61
133
48
122
46
271
96
47
45
54
60
98
5
5
4
4
7
7
6
5
5
5
5
5
4
4
4
2
2
3
2
0
4
2
2
2
3
2
2
2
5
5
1
4
1
1
1
0
0
1
1
0
2
1
1
1
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
2
2
2
2
2
2
2
2
2
3
2
2
2
2
2
0.18
0.08
0.19
0.20
0.02
0.01
0.11
0.23
0.28
0.04
0.09
0.35
0.35
0.12
0.02
12a
12b
12c
12g
12h
12i
12j
12k
12l
12m
12n
12r
12s
12u
relative to solvent residual signal. The coupling constants J are
given in Hertz (Hz). The abbreviations used for signal descriptions
are as follows: s, singlet; d, doublet; t, triplet; q, quartet; m,
multiplet. Electron Impact Mass Spectra (EI-MS) were obtained on a
spectrometer Hewlett Packard 5989B at 70 eV. High Resolution
Electro-spray Ionization Mass Spectra (HR-ESI-MS) were obtained
from UBP-Start (Clermont-Ferrand, FRANCE). High performance
liquid chromatography (HPLC) conditions for purity analysis [col-
Training ꢀ0.125
Training
Test
0.167
0.344
0.377
0.000
0.017
0.107
0.167
0.104
0.021
Training
Test
Training
Training
Training
Training
Training
umn: Poroshell 120 EC-C18 2.7
m
m (3.0 ꢃ 50 mm); temperature
column: 40 ꢄC; injection: 10
mL, 1 mL/min; solvent used H2O:MeOH
49
79
179
mixture with 0.1% TFA (t ¼ 0.95% H2O:5% MeOH, t ¼ 1e18 min
gradient from 95% H2O:5% MeOH to 5% H2O:95% MeOH, 1 min

398
N. Rodrigues et al. / European Journal of Medicinal Chemistry 75 (2014) 391e402
Fig. 5. A. Pharmacophore hypothesis A1D2D3R4R5. B. Distances (in A) between the pharmacophoric sites of the ADDRR hypothesis. C. Angles (in ꢄ) between the pharmacophoric
sites of the ADDRR hypothesis. D. All molecules aligned with pharmacophore hypothesis ADDRR. E. The best matched molecule 12j with pharmacophore hypothesis ADDRR.
ꢀ
isocratic 5% H2O:95% MeOH), UV detection at 254 nm and
280.5 nm].
Rf ¼ 0.50 (SiO2, 10% EtOAc/cyclohexane). 1H NMR (400 MHz, MeOD-
d4)
d
8.15 (s, 1H), 7.67 (d, J ¼ 2.1 Hz, 1H), 7.44 (d, J ¼ 7.3 Hz, 2H), 7.39
(dd, J ¼ 2.0, 8.4 Hz,1H), 7.32 (t, J ¼ 7.4 Hz, 2H), 7.25 (t, J ¼ 7.3 Hz,1H),
6.85 (t, J ¼ 9.0 Hz, 1H), 6.77 (d, J ¼ 15.9 Hz, 1H), 6.41 (dt, J ¼ 6.3 and
15.9 Hz, 1H), 4.93 (dd, J ¼ 1.0 and 6.4 Hz, 2H). 13C NMR (101 MHz,
4.1.1. (E)-Cinnamyl 2-cyano-3-(3,4-dihydroxyphenyl)acrylate 2
To
a solution of cinnamyl 2-cyanoacetate 8a (82.4 mg,
MeOD-d4) d 164.5,156.6,153.2,147.1,137.6,135.8,129.7,129.2,128.2,
0.41 mmol, 1.1 eq) in anhydrous DCM under Ar atmosphere was
added 3,4-dihydroxybenzaldehyde 9a (50 mg, 0.373 mmol, 1 eq)
and piperidine (cat). The mixture was left to stirred at room tem-
perature until total consumption of cinnamyl 2-cyanoacetate (TLC).
The mixture was then quenched by the addition of a saturated
aqueous solution NH4Cl (20 mL) and extracted with EtOAc
(3 ꢃ 8 mL). The combined organic fractions were dried (Na2SO4).
After removal of the solvent under reduced pressure, chromatog-
raphy purification of the residue on silica gel with cyclohexane/
EtOAc gave CDC 2 as an orange powder (119.8 mg, 99%). Mp 164 ꢄC.
127.7, 125.1, 123.7, 117.8, 117.3, 116.7, 98.4, 67.7. HR-ESI-MS calcd for
C
19H15NO4 (M þ Naþ) ¼ 344.0899; found 344.0902.
4.1.2. (E)-Phenethyl 3-(3,4-dihydroxyphenyl)acrylate 3
To a solution of (E)-4-(3-oxo-3-phenethoxyprop-1-enyl)-1,2-
phenylene diacetate 11a (474 mg, 1.29 mmol, 1 eq) in anhydrous
THF (15 mL) was added pyrrolidine (cat). After stirring at room
temperature for 30 min, water (10 mL) was added and the reaction
mixture was extracted with EtOAc (20 mL). The combined organic
layers were washed with 1 N HCl (7 ꢃ 20 mL), brine. After drying
(Na2SO4) and concentration under reduced pressure, CAPE 3 was
obtained as a yellow powder (36.7 mg, 75%). Mp 125.5 ꢄC. Rf ¼ 0.75
(SiO2, 40% EtOAc/cyclohexane). 1H NMR (400 MHz, Acetone-d6)
Table 4
Summary of atom-based 3D-QSAR model for the pharmacophore hypothesis ADDRR
which consists of five pharmacophoric sites: one hydrogen-bond acceptor (A), two
hydrogen-bond donors (D) and two aromatic features (R).
d
8.36 (br s, 1H), 8.08 (br s, 1H), 7.6 (d, J ¼ 15.0 Hz, 1H) 7.40 (dd,
J ¼ 9.0, 2.0 Hz, 1H), 7.35e7.28 (m, 5H), 7.25 (d, J ¼ 2.0 Hz, 1H), 7.20
(d, J ¼ 9.0 Hz, 1H), 6.30 (d, J ¼ 15.0 Hz, 1H), 4.35 (t, J ¼ 7.0 Hz, 2H),
Training set
Test set
Q2
2.90 (t, J ¼ 7.0 Hz, 2H). 13C NMR (101 MHz, Acetone-d6)
d 167.4,
R2
SD
0.058
F
P
RMSE
Pearson-R
0.914
148.9, 148.4, 145.9, 139.4, 129.9, 129.4, 127.7, 127.3, 122.6, 116.5,
115.6, 115.3, 65.4, 35.9.
0.885
79.7
4.453e-00.6
0.807
0.0345

N. Rodrigues et al. / European Journal of Medicinal Chemistry 75 (2014) 391e402
399
Fig. 6. Graphs of observed versus predicted activity pRTREK-1 of (A) training set, (B) test set.
4.1.3. (E)-Cinnamyl 3-(3,4-dihydroxyphenyl)acrylate 4
ester was obtained as a yellow powder (181 mg, 98%). Mp 145.5 ꢄC.
To a solution of the 4-((E)-3-(cinnamyloxy)-3-oxoprop-1-enyl)-
1,2-phenylene diacetate) 11b (236.4 mg, 0.622 mmol, 1 eq) in
anhydrous THF (7.5 mL) was added pyrrolidine (cat). After stirring
at room temperature for 30 min, the reaction was worked up by
addition of H2O (10 mL) and extracted with EtOAc (20 mL). The
organic layer was washed with 1 N HCl (7 ꢃ 20 mL), brine. After
drying (Na2SO4) and concentration under reduced pressure, the
Rf ¼ 0.75 (SiO2, 40% EtOAc/cyclohexane). 1H NMR (400 MHz,
Acetone-d6)
d
7.60 (d, J ¼ 16.0 Hz, 1H), 7.40 (m, 5H), 7.05 (d,
J ¼ 2.0 Hz, 1H), 6.96 (dd, J ¼ 6.0, 2.0 Hz, 1H), 6.79 (d, J ¼ 6.0 Hz, 1H),
6.70 (d, J ¼ 16.0 Hz, 1H), 6.40 (dt, J ¼ 16.0, 6.0 Hz, 1H), 6.30 (d,
J ¼ 16.0 Hz, 1H), 4.86 (br s, 2H), 4.82 (d, J ¼ 6.0 Hz, 2H). HR-ESI-MS
calcd for C18H17O4 (M þ Hþ) ¼ 297.1089; found 297.0932. The NMR
data are in agreement with the literature [30].
Fig. 7. Pictorial representation of the cubes generated by the QSAR model. Blue cubes indicate favorable regions, while red cubes indicate unfavorable region for the activity. Atom-
based 3D-QSAR model visualized in the context of A. Hydrogen-bond donor (D) feature. B. Hydrophobicity (H) and C. Electron-withdrawing density (W). (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of this article.)

400
N. Rodrigues et al. / European Journal of Medicinal Chemistry 75 (2014) 391e402
4.1.4. Cinnamyl 2-cyanoacetate 8a
peritoneal cavity of the animal. This activates nociceptors directly
and/or produces inflamed viscera (subdiaphragmatic organs) and
subcutaneous (muscle wall) of tissues [9]. The number of induced
abdominal writhes was determined during 20 min after the in-
jection of acetic acid. The inhibition of abdominal writhes (anal-
gesic effect) is determined by comparing the number of cramps
induced in the presence and absence of compound. Additional dose
response was performed for compound 2 (CDC) at the following
doses: 2.5, 5 and 10 mg/kg (s.c.).
To a solution of commercially available cyanoacetic acid 7a
(157.6 mg, 1.85 mmol, 1.1 eq) and cinnamyl alcohol 6a (192.07 mg,
1.43 mmol, 1 eq) in anhydrous DCM (8 mL) under Ar at 0 ꢄC were
added DCC (383 mg, 1.85 mmol, 1.1 eq) and DMAP (cat.). The re-
action mixture was stirred for 4 h at 0 ꢄC. The precipitate was
filtered and the filtrate was concentrated under reduced pressure.
Chromatographic purification of the residue (SiO2, 50% EtOAc/
cyclohexane) gave the ester 8a as a yellow liquid (296.7 mg, 98%).
Rf ¼ 0.5 (SiO2, 30% EtOAc/cyclohexane). 1H NMR (400 MHz, CDCl3)
d
7.40e7.30 (m, 5H), 6.68 (d, J ¼ 14.8 Hz, 1H), 6.27 (dt, J ¼ 15.9,
4.2.2. Electrophysiology
6.7 Hz, 1H), 4.66 (dd, J ¼ 6.8, 1.0 Hz, 2H), 3.49 (s, 2H). 13C NMR
Xenopus oocytes deprived of their follicular cells are injected
with 50 ng cRNA encoding TREK-1 channel [25]. Eighteen to 24 h
after injection, TREK-1 currents are recorded by the technique of
the two-microelectrode voltage-clamp technique. In a perfusion
chamber of 0.3 mL, the oocyte is impaled with two microelectrodes
standards (1e2.5 M resistance) filled with 3 M KCl solution and
maintained at a voltage clamp with a Dagan TEV 200 amplifier in a
solution standard ND96 (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl2,
2 mM MgCl2, 5 mM HEPES, pH 7.4 with NaOH). Then a solution of
(101 MHz, CDCl3)
113.1, 67.4, 24.8.
d 162.9, 135.9, 135.7, 128.7, 128.5, 126.8, 121.4,
4.1.5. (E)-3-(3,4-Diacetoxyphenyl)acrylic acid 10b
To a cold solution of caffeic acid 10a (460 mg, 2.55 mmol, 1 eq)
and DMAP (7.78 mg, 0.055 mmol, 0.025 eq) in pyridine (1.10 mL)
was added acetic anhydride (603 mL, 6.38 mmol, 2.5 eq) under Ar.
The mixture was stirred at room temperature for 2 h and poured
over crushed ice. The solution was acidified (pH < 2) and extracted
with EtOAc (3 ꢃ 10 mL). The combined extracts were dried (Na2SO4)
and concentrated under reduced pressure. Acid 10b was obtained
as a white powder (629.7 mg, 93%). Mp 206 ꢄC. Rf ¼ 0.43 (SiO2, 80%
compound [20
mM, compounds were dissolved in DMSO then
diluted in extracellular recording solution (The DMSO concentra-
tion in this solution having to be kept below 5%, some compounds
were not soluble at concentrations exceeding 10 mM)] is perfused
for 15 min, followed by rinsing with ND96 for 6 min. Stimulation of
the preparation, data acquisition and analysis were performed us-
ing pClamp software.
EtOAc/cyclohexane). 1H NMR (400 MHz, DMSO-d6)
d 12.50 (br s,
1H), 7.30 (m, 3H), 7.31 (d, J ¼ 14.0 Hz, 1H), 6.50 (d, J ¼ 14.0 Hz, 1H),
2.31 (s, 3H), 2.29 (s, 3H). The NMR data are in agreement with the
literature [31].
4.3. Molecular modeling
4.1.6. (E)-4-(3-Oxo-3-phenethoxyprop-1-enyl)-1,2-phenylene
diacetate 11a
4.3.1. Pharmacophore modeling
To a suspension of diacetylcaffeic acid 10b (151 mg, 0.572 mmol,
3 eq) in anhydrous toluene (3 mL) were sequentially added 2-
Pharmacophore modeling was carried out using PHASE [28,29]
running on Red Hat Linux WS 3.0. A set of 15 novel analogs with
determined bioactivity (Table 3) was taken for the development of
ligand-based pharmacophore hypothesis and atom-based 3D-QSAR
model. The logarithm of the measured RTREK-1 [pRTREK-1 ¼ Log
(RTREK-1)] was used in this study. These 15 compounds were divided
into a training set (10 compounds) and a test set (5 compounds).
The training set molecules were selected in such a way that they
contained information in terms of both their structural features and
biological activity ranges. In order to assess the predictive power of
the model, a set of 5 compounds was arbitrarily set aside as the test
set. The test compounds were selected in such a way that they truly
represent the training set.
phenylethanol 6b (23.6
mL, 0.197 mmol, 1 eq), DCC (113.8 mg,
0.551 mmol, 2.8 eq) and DMAP (24.1 mg, 0.197 mmol, 1 eq) under
Ar. The reaction mixture was heated at 90 ꢄC for 3 h and diluted
with EtOAc (10 mL). The mixture was washed with saturated
aqueous NaHCO3 (10 mL) and brine (10 mL). After drying (Na2SO4)
and concentration under reduced pressure, the residue was puri-
fied by flash chromatography (SiO2, 30% EtOAc/cyclohexane) to
afford yellow oil (36.7 mg, 51%). Rf ¼ 0.46 (40% EtOAc/cyclohexane).
1H NMR (400 MHz, CDCl3):
d
7.60 (d, J ¼ 15.0 Hz, 1H), 7.40 (dd,
J ¼ 10.0, 2.0 Hz, 1H), 7.35e7.25 (m, 5H), 7.25 (d, J ¼ 2.0 Hz, 1H), 7.20
(d, J ¼ 8.0 Hz, 1H), 6.30 (d, J ¼ 15.0 Hz, 1H), 4.35 (t, J ¼ 7.0 Hz, 2H),
2.90 (t, J ¼ 7.0 Hz, 2H), 2.80 (s, 6H). 13C NMR (101 MHz, CDCl3):
d
168.2, 168.1, 166.6, 143.0, 142.5, 137.9, 133.4, 129.0, 128.6, 126.7,
4.3.2. Generation of common pharmacophore hypothesis (CPH)
The common pharmacophore hypothesis (CPH) was carried out
by PHASE. All molecules were built in Maestro. All ligands were
prepared using LigPrep with the OPLS_2005 force field. Confor-
mational space was explored through combination of Monte-Carlo
Multiple Minimum (MCMM)/Low Mode (LMOD) with maximum
number of conformers 1000 per structure and minimization steps
100. Each minimized conformer was filtered through a relative
126.5, 124.0, 122.8, 119.3, 62.3, 35.3, 20.8, 20.7.
4.2. Biological assays
4.2.1. Acetic acid writhing assay
All experiments were performed on 20e24 g male CD1 mice
(Janvier, France). All mice were housed in grouped cages in a
temperature-controlled environment with food and water ad libi-
tum. The behavioral experiments were performed blind to the
treatment, in a quiet room, by the same experimenter taking great
care to avoid or minimize discomfort of the animals. Animals were
randomly divided in groups of 8 mice each and each animal was
used only once per compound and euthanized. All animal pro-
cedures were approved by the local Animal Ethics Committee and
experiments were performed according to the guidelines provided
by the European Community guiding in the care and use of animals
(86/609/CEE). Animals were pretreated with the compound
(10 mg/kg, s.c.) or vehicle (5% Tween 80 in saline (0.9%)) 15 min
before injection of acetic acid (0.6% solution, 10 mL/kg) into the
ꢀ
energy window of 50 kJ/mol and redundancy check of 2 A in the
heavy atom positions. Common pharmacophoric features were
then identified from a set of variants (a set of feature types that
define a possible pharmacophore) using a tree-based partitioning
algorithm with maximum tree depth of four with the requirement
that all actives must match. After applying default feature defini-
tions to each ligand (Table 3), common pharmacophores containing
ꢀ
three to six sites were generated using a terminal box of 1 A. Scoring
of pharmacophore with respect to activity of ligand was conducted
using default parameters for site, vector and volume terms. In
atom-based QSAR, a molecule is treated as a set of overlapping van
der Waals spheres. Each atom (and hence each sphere) is placed

N. Rodrigues et al. / European Journal of Medicinal Chemistry 75 (2014) 391e402
401
into one of six categories according to a simple set of rules: hy-
TBDMS tert-butyldimethylsilyl
drogens attached to polar atoms are classified as hydrogen bond
donors (D); carbons, halogens, and CeH hydrogens are classified as
hydrophobic/non-polar (H); atoms with an explicit negative ionic
charge are classified as negative ionic (N); atoms with an explicit
positive ionic charge are classified as positive ionic (P); non-ionic
atoms are classified as electron-withdrawing (W); and all other
types of atoms are classified as miscellaneous (X). For purposes of
QSAR development, van der Waals models of the aligned training
set molecules were placed in a regular grid of cubes, with each cube
allotted zero or more ‘bits’ to account for the different types of
atoms in the training set that occupy the cube. This representation
gives rise to binary-valued occupation patterns that can be used as
independent variables to create partial least-squares (PLS) QSAR
models. Atom-based QSAR models were generated for all hypoth-
TEA
TFA
TREK
TWIK
triethylamine
trifluoroacetic acid
TWIK-related Kþ channels
tandem of pore domains in a weak inward rectifying
Kþ channel
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.ejmech.2014.01.049.
References
ꢀ
eses using the training set using a grid spacing of 1.0 A. The best
QSAR model was validated by predicting activities of the test set
molecules.
[1] E. Bourinet, A. Alloui, A. Monteil, C. Barrere, B. Couette, O. Poirot, A. Pages,
J. McRory, T. Snutch, A. Eschalier, J. Nargeot, Silencing of the Cav3.2 T-type
calcium channel gene in sensory neurons demonstrates its major role in
nociception, EMBO J. 24 (2005) 315e324.
[2] T.R. Cummins, P.L. Sheets, S.G. Waxman, The roles of sodium channels in
nociception: implications for mechanisms of pain, Pain 131 (2007) 243e257.
[3] D. Juilius, A.I. Basbaum, Molecular mechanisms of nociception, Nature 413
(2001) 203e210.
[4] A.J. Patel, E. Honore, F. Maingret, F. Lesage, M. Fink, F. Duprat, M. Lazdunski,
A mammalian two pore domain mechano-gated S-type Kþ channel, EMBO J.
17 (1998) 4283e4290.
[5] F. Maingret, A.J. Patel, F. Lesage, M. Lazdunski, E. Honore, Mechano- or acid
stimulation, two interactive modes of activation of the TREK-1 potassium
channel, J. Biol. Chem. 274 (1999) 26691e26696.
[6] F. Maingret, E. Honore, M. Lazdunski, A.J. Patel, Molecular basis of the voltage-
dependent gating of TREK-1, a mechano-sensitive Kþ channel, Biochem. Bio-
phys. Res. Commun. 292 (2002) 339e346.
Author contributions
A.E. and S.D. initiated the project. A.E., J.B., F.L., K.B., S.D. super-
vised the project, participated in data analysis and manuscript
preparation. N.R., D.V. synthesized, purified and characterized the
molecules, under the direction of K.B. and S.D. N.R., F.C.C. conducted
the electrophysiology experiments under the guidance of F.L. V.P.,
E.C. conducted behavioral experiments under the guidance of J.B.
H.D. conducted the modeling studies under the guidance of S.D.
[7] A. Dedman, R. Sharif-Naeini, J.H.A. Folgering, F. Duprat, A. Patel, E. Honore, The
mechano-gated K-2P channel TREK-1, Eur. Biophys. J. 38 (2009) 293e303.
[8] W. Treptow, M.L. Klein, The membrane-bound state of K2P potassium chan-
nels, J. Am. Chem. Soc. 132 (2010) 8145e8151.
[9] A. Alloui, K. Zimmermann, J. Mamet, F. Duprat, J. Noel, J. Chemin, N. Guy,
N. Blondeau, N. Voilley, C. Rubat-Coudert, M. Borsotto, G. Romey,
C. Heurteaux, P. Reeh, A. Eschalier, M. Lazdunski, TREK-1, a Kþ channel
involved in polymodal pain perception, EMBO J. 25 (2006) 2368e2376.
[10] J. Noel, K. Zimmermann, J. Busserolles, E. Deval, A. Alloui, S. Diochot, N. Guy,
M. Borsotto, P. Reeh, A. Eschalier, M. Lazdunski, The mechano-activated Kþ
channels TRAAK and TREK-1 control both warm and cold perception, EMBO J.
28 (2009) 1308e1318.
Acknowledgments
We would like to thank the Regional Council of Auvergne
(Conseil Régional d’Auvergne) and the European Fund for Regional
Economic Development (FEDER) for a doctoral scholarship for NR;
the Novartis Training Scheme for funding a fellowship for DV; the
French Ministry of Higher Education and Research (MESR) for a
doctoral scholarship for DV; the French National Research Agency
(ANR) for funding of the project TREK-ANALGESIA (ANR-12-EMMA-
0017-01); LabEx ICST (ANR-11-LABX-0015-01); the Fondation pour
la Recherche Médicale (Equipe FRM 2011 to FL); the Société Fran-
çaise d’Etude et de Traitement de la Douleur (SFETD), programme
“Groupes de Projets Scientifiques” 2010.
[11] (a) J. Noël, G. Sandoz, F. Lesage, Molecular regulations governing TREK
and TRAAK functions, Channels 5 (2011) 402e409;
(b) S.N. Bagriantsev, K.H. Ang, A. Gallardo-Godoy, K.A. Clark, M.R. Arkin,
A.D. Renslo, D.L. Minor Jr., A high-throughput functional screen identifies
small molecule regulators of temperature- and mechano-sensitive K2P
channels, ACS Chem Biol. 16 (2013) 1841e1851.
[12] A. Doble, The pharmacology and mechanism of action of riluzole, Neurology
47 (1996) S233eS241.
[13] F. Duprat, F. Lesage, A.J. Patel, M. Fink, G. Romey, M. Lazdunski, The neuro-
protective agent riluzole activates the two P domain Kþ channels TREK-1 and
TRAAK, Mol. Pharmacol. 57 (2000) 906e912.
[14] A. Cadaveira-Mosquera, S.J. Ribeiro, A. Reboreda, M. Perez, J.A. Lamas, Acti-
vation of TREK currents by the neuroprotective agent riluzole in mouse
sympathetic neurons, J. Neurosci. 31 (2011) 1375e1385.
[15] S. Danthi, J.A. Enyeart, J.J. Enyeart, Caffeic acid esters activate TREK-1 potas-
sium channels and inhibit depolarization-dependent secretion, Mol. Phar-
macol. 65 (2004) 599e610.
[16] H. Cho, M. Ueda, M. Tamaoka, M. Hamaguchi, K. Aisaka, Y. Kiso, T. Inoue,
R. Ogino, T. Tatsuoka, Novel caffeic acid derivatives: extremely potent in-
hibitors of 12-lipoxygenase, J. Med. Chem. 34 (1991) 1503e1505.
[17] L.H. Hu, H.B. Zou, J.X. Gong, H.B. Li, L.X. Yang, W. Cheng, C.X. Zhou, H. Bai,
F. Gueritte, Y. Zhao, Synthesis and biological evaluation of a natural ester
sintenin and its synthetic analogues, J. Nat. Prod. 68 (2005) 342e348.
[18] S. Danthi, J.A. Enyeart, J.J. Enyeart, Modulation of native TREK-1 and Kv1.4 Kþ
channels by polyunsaturated fatty acids and lysophospholipids, J. Membr.
Biol. 195 (2003) 147e164.
[19] C.A. Zarate Jr., H.K. Manji, Riluzole in psychiatry: a systematic review of the
literature, Expert Opin. Drug Metab. Toxicol. 4 (2008) 1223e1234.
[20] C. Pittenger, C. Vladimir, M. Banasr, M. Bloch, J.H. Krystal, G. Sanacora, Riluzole
in the treatment of mood and anxiety disorders, CNS Drugs 22 (2008) 761e
786.
Abbreviations
CAPE
CDC
caffeic acid phenethyl ester
cinnamyl 1e3,4-dihydroxy-a-cyanocinnamate
DABCO 1,4-diazabicyclo[2.2.2]octane
DCC
N,N0-dicyclohexylcarbodiimide
DCM
DEAD
DIAD
dichloromethane
diethyl azodicarboxylate
diisopropyl azodicarboxylate
DMAP 4-dimethylaminopyridine
HBA
HBD
K2P
LAH
LDA
hydrogen-bond acceptor
hydrogen-bond donor
two-pore domain potassium channel
lithium aluminum hydride
lithium diisopropylamide
mCPBA meta-chloroperoxybenzoic acid
MOM
MW
methoxymethyl
molecular weight
[21] M.C. Obinu, M. Reibaud, V. Blanchard, S. Moussaoui, A. Imperato, Neuro-
PSA
polar surface area
protective effect of riluzole in
a primate model of Parkinson’s disease:
PTSA
QSAR
TBAF
para-toluenesulfonic acid
quantitative structureeactivity relationship
tetrabutylammonium fluoride
behavioral and histological evidence, Mov. Disord. 17 (2002) 12e19.
[22] D.L. Ginsberg, Riluzole for treatment-resistant obsessive-compulsive disorder,
Prim. Psychiatry 12 (2005) 33e34.

402
N. Rodrigues et al. / European Journal of Medicinal Chemistry 75 (2014) 391e402
[23] T.J. Ritchie, P. Ertl, R. Lewis, The graphical representation of ADME-related
molecule properties for medicinal chemists, Drug Discov. Today 16 (2011)
65e72.
[28] S.L. Dixon, A.M. Smondyrev, S.N. Rao, PHASE: a novel approach to pharma-
cophore modeling and 3D database searching, Chem. Biol. Drug Des. 67 (2006)
370e372.
[24] T. Wa˛seka, J. Olczakb, T. Janecki, New, simple and versatile synthesis of pro-
[29] S.L. Dixon, A.M. Smondyrev, E.H. Knoll, S.N. Rao, D.E. Shaw, R.A. Friesner,
tected
a
-alkylidene-
b
-amino acids from activated vinylphosphonates, Synlett
PHASE: a new engine for pharmacophore perception, 3D QSAR model
10 (2006) 1507e1510.
development, and 3D database screening: 1. Methodology and preliminary
results, J. Comput. Aided Mol. Des. 20 (2006) 647e671.
[30] T. Nagaaoka, A.H. Bansjota, I.S. Tezuka, S. Kadota, Selective antiproliferative
activity of caffeic acid phenethyl ester analogues on highly liver-metastatic
murine colon 26-L5 carcinoma cell line, Bioorg. Med. Chem. 10 (2002)
3351e3359.
[31] S. Saito, S. Kurakane, M. Seki, E. Takai, T. Kasai, J. Kawabata, Radical scavenging
activity of dicaffeoyloxycyclohexanes: contribution of an intramolecular
interaction of two caffeoyl residues, Bioorg. Med. Chem. 13 (2005) 4191e4199.
[25] M. Fink, F. Duprat, F. Lesage, R. Reyes, G. Romey, C. Heurteaux,
M. Lazdunski, Cloning, functional expression and brain localization of a
novel unconventional outward rectifier Kþ channel, EMBO J. 15 (1996)
6854e6862.
[26] H. Miranda, F. Sierralta, G. Pinardi, Neostigmine interactions with non-
steroidal anti-inflammatory drugs, Br. J. Pharmacol. 135 (2002) 219e222.
[27] E. Siegmund, R. Cadmus, G. Lu, A method for evaluating both non-narcotic and
narcotic analgesics, Proc. Soc. Exp. Biol. Med. 95 (1957) 729e731.