Towards a KCC2 blocker pharmacophore model

Article abstract of DOI:10.1016/j.bmcl.2012.04.097

A multi-disciplinary approach was used to identify the first pharmacophore model for KCC2 blockers: several physico-chemical studies such as XRD and NMR were combined to molecular modelling techniques, SAR analysis and synthesis of constrained analogues in order to determine a minimal conformational space regrouping few potential bioactive conformations. These conformations were further compared to the conformational space of a different series of KCC2 blockers in order to identify the common pharmacophoric features. The synthesis of more potent analogues in this second series confirmed the usefulness of this KCC2 blocker pharmacophore model.

Full text of DOI:10.1016/j.bmcl.2012.04.097

Bioorganic & Medicinal Chemistry Letters 22 (2012) 3978–3982
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Towards a KCC2 blocker pharmacophore model
Florence Lebon a, , Cécile Pégurier a, , Marie Ledecq a, Benoit Mathieu a, Nathalie Bosman a,
Anne Frycia a, Sébastien Lengelé a, Kashinath Dhurke b, Ananda Kumar Kanduluru b, Stéphane Meunier b,
Alain Wagner b, Christian Wolff a, Laurent Provins a
⇑
,
⇑
,
a UCB Pharma, UCB NewMedicines, Chemin du Foriest, B-1420 Braine-L’Alleud, Belgium
b Laboratoire des systèmes chimiques fonctionnels, UMR 7199 associé au CNRS, Faculté de Pharmacie, Université de Strasbourg, 74 route du Rhin, BP 24, F-67401 Illkirch, France
a r t i c l e i n f o
a b s t r a c t
Article history:
A multi-disciplinary approach was used to identify the first pharmacophore model for KCC2 blockers:
several physico-chemical studies such as XRD and NMR were combined to molecular modelling tech-
niques, SAR analysis and synthesis of constrained analogues in order to determine a minimal conforma-
tional space regrouping few potential bioactive conformations. These conformations were further
compared to the conformational space of a different series of KCC2 blockers in order to identify the com-
mon pharmacophoric features. The synthesis of more potent analogues in this second series confirmed
the usefulness of this KCC2 blocker pharmacophore model.
Received 22 March 2012
Revised 19 April 2012
Accepted 20 April 2012
Available online 30 April 2012
In memory of Dr. Patrice Talaga
Ó 2012 Elsevier Ltd. All rights reserved.
Keywords:
KCC2
Pharmacophore
Bioactive conformations
COOH
KCC2 is a neuronal-specific electroneutral potassium-chloride
co-transporter belonging to a larger family of cation-chloride co-
transporters (CCC’s) which play important roles in a variety of
physiological processes. KCC2 acts as a key modulator of inhibitory
neurotransmission in the brain and in the spinal cord. Its potential
role in pathologies such as epilepsy or neuropathic pain triggered
the interest of several groups over the last few years.1–5 Loop
diuretics, such as furosemide, constitute a major class of drugs
known to interact with KCC2 but without the adequate selectivity
allowing using them as relevant pharmacological tools.
Despite the recent identification of selective KCC2 blockers
chemical series,6,7 no information is available regarding the struc-
tural requirements for binding to KCC2. In this Letter, we wish to
report our strategy to derive the bioactive conformation of our
hit compound 1 as well as the first pharmacophore model for
KCC2 blockers by taking advantage of various sources of structural
information such as molecular modelling techniques, crystallo-
graphic or NMR data. The hypotheses formulated along the study
were assessed through SAR analysis as well as the design and syn-
thesis of constrained analogues.
N
O
O
N
SO2NH2
O
Cl
O
(R)-1
Figure 1. Furosemide reference compound and hit compound 1.
hit with an IC50 of 0.3
l
M (Fig. 1). Interestingly, compound (S)-1 is
M, Table 1). We made the
100-fold less potent on KCC2 (IC50 = 50
l
assumption that, for all analogues in this series, the active enantio-
mer has the same configuration as (R)-1, although only racemic
mixtures were prepared. We reported earlier6 the structure–activity
relationships (SAR) around our hit compound focusing on the
modulation of the two aryls and the N-acyl moieties. In order to
identify the pharmacophoric features required for binding to the
KCC2 transporter, several analogues were prepared where the ester
function is replaced or which contain structural rigidification.
An indoline-based analogue 2 was prepared by deprotonation
with LiHMDS at À70 °C and alkylation with benzyl bromide
(Scheme 1). The ketone analogues 4–5 were obtained following
the synthetic route depicted in Scheme 2 from the methyl ester
intermediate reported previously.6 The aldehyde was prepared
after a two-step procedure consisting of a LiBH4 reduction of the
ester followed by a Dess–Martin oxidation. The addition of a
Grignard reagent then provided a primary alcohol that could be
High throughput screening (HTS) of our corporate compound
collection using a Rb+ flux assay8 on the KCC2 co-transporter led to
the identification of benzyl 1-acetyl-2-benzylprolinate (R)-19 as a
⇑
Corresponding authors.
E-mail addresses: florence.lebon@ucb.com (F. Lebon), cecile.pegurier@ucb.com
(C. Pégurier).
These authors contributed equally to this work.
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.04.097

F. Lebon et al. / Bioorg. Med. Chem. Lett. 22 (2012) 3978–3982
3979
described by Duggan et al.11 using a ring-closing metathesis fol-
lowed by a catalytic hydrogenation (Scheme 6).
Table 1
KCC2 activities for ester compounds 1–2
a
When compound 1 was identified, very little information was
available on the pharmacophoric features required to observe
binding to KCC2. As no other families of KCC2 blockers were known
at the time, we decided to probe the bioactive conformation of
compound 1 by systematically assessing each torsion angle of
the molecule (Fig. 2) using molecular modelling techniques, crys-
tallographic as well as NMR data and confronting those to SAR
data. As the project progressed, additional families were identified
which were used to derive a KCC2 pharmacophore described in
this Letter.
Compound
Structure
IC50 (lM)
1
0.8
0.3
O
O
O
N
O
O
O
O
We first focused on torsion angle T1. Within the 2-benzylproli-
nate family, we observed a loss of KCC2 activity upon replacement
of the ester moiety by an amidic group (compound 3, Table 2). On
the other hand, compound 4, in which the ester is replaced by
(R)-1
(S)-1
2
N
O
C(O)CH2 was synthesized and displays a potency of 0.3
addition, compound 6, in which the ester was replaced by CH2O
only provides an activity of 20 M. These data suggest that the car-
lM. In
l
50
bonyl of the ester moiety is critical for activity and that the loss of
potency of the amide 3 is not due to the loss of an H-bond acceptor
but may rather be due to a conformational constraint. Indeed, the
conformational space of compounds 1 and 3 were computed12 and
showed the presence of an intramolecular H-bond between the NH
of the amide and the carbonyl of the acetamide in the case of com-
pound 3 that constrains its conformational space compared to
compound 1. Some regions occupied by the benzyl ester group of
1 which was previously described as critical to the KCC2 activity6
are not sampled by compound 3. These regions are characterised
by a torsion angle T1 of ꢀ90°, 180° and 270°. Taking this feature
into account, we designed several constrained bioisosteres of com-
pound 1 in order to further explore T1. Compounds 7 and 8 were
synthesized and tested against KCC2. Compound 7 in which
N
O
0.2
O
N
O
O
a
Rb+ flux assay on cloned rat KCC2.8
finally oxidized with the Dess–Martin reagent. The ether analogue
6 was easily obtained (Scheme 3) after alkylation of the previous
alcohol intermediate with benzyl bromide. The oxazoline 7 was
synthesized from the coupling of the 1-acetyl-2-benzylproline6
with (R)-2-phenylglycinol followed by cyclisation in the presence
of Deoxofluor™ (Scheme 4).10 The oxazole 8 was prepared from
the condensation of the primary amide with the 2-bromoacetophe-
none under microwave heating with poor yield (Scheme 5). A con-
strained analogue 9 was obtained from a known intermediate
T1 = 115° displays an IC50 of 1.3 lM which is in a comparable po-
tency range as compound 1. On the contrary, compound 8 which
possess a torsion angle T1 ꢀ 180° is 10 times less potent. Torsion
angle T1 ꢀ 90° was further considered as a potential pharmaco-
phoric constraint.
A search in the Cambridge Structural Database suggests that a
torsion angle T3 of 180° is strongly favored. The trans conformation
of the acetamide carbonyl is indeed described to be favored by an
c
a, b
OH
O
O
N
N
N
H
O
O
O
O
O
2
Scheme 1. Reagents and conditions: (a) CH3COCl, K2CO3, CH3CN, rt, 35%; (b) benzylbromide, K2CO3, KI cat., CH3CN, 80 °C, 70%; (c) LiHMDS, THF, À70 °C, benzylbromide, 64%.
R
R
R
R
R
R
R
a, b
d
ClMg
O
N
N
N
N
O
O
c
OH
O
O
O
O
O
R=H, 4
R=F, 5
Scheme 2. Reagents and conditions: (a) LiBH4, THF, 70 °C, 65%; (b) Dess–Martin reagent, CH2Cl2, rt, quant; (c) THF, 0 °C, 70% for R = H, 74% for R = F; (d) Dess–Martin reagent,
CH2Cl2, rt, 82% for R = H, 80% for R = F.

3980
F. Lebon et al. / Bioorg. Med. Chem. Lett. 22 (2012) 3978–3982
a
b
O
OH
O
N
N
N
O
O
6
O
O
Scheme 3. Reagents and conditions: (a) LiBH4, THF, RT, 62%; (b) benzylbromide, NaH, nBu4NI cat., THF, rt, 30%.
OH
H2N
H
b
a
OH
OH
N
N
+
N
N
N
O
O
O
O
O
O
7
Scheme 4. Reagents and conditions: (a) EDCI, HOBt, DIPEA, CH2Cl2, rt, 65%; (b) deoxofluor™, CH2Cl2, À20 °C, 60%.
a
OH
NH2
b
N
N
N
N
O
O
O
O
O
O
8
Scheme 5. Reagents and conditions: (a) EDCI, HOBt, NH4OH aq, THF, 92%; (b) 2-bromoacetophenone, propionitrile, 120 °C, microwaves, 3 h, 2%.
OBn
a, b, c
d, e, f
OBn
OBn
N
N
O
N
O
O
O
O
O
O
9
Scheme 6. Reagents and conditions: (a) LiHMDS, THF, À70 °C, benzylbromide, 22%; (b) TFA, CH2Cl2, 86%; (c) acryloyl chloride, NEt3, CH2Cl2, 95%; (d) Hoveyda-Grubbs catalyst
2nd gen. (10 mol %), CH2Cl2, 0 °C, 83%; (e) H2, Pd/C, EtOH, 95%; (f) benzylbromide, DIPEA, CH3CN, reflux, 65%.
tained, this compound being one of the only analogue of this series
22
19
16
18
17
that could be crystallized in our hands. In this structure, the trans
conformation of the amide carbonyl was also observed (Fig. 3).
In order to assess torsion angles T2 and T4, the ROESY NMR data
obtained for compound 5 was further exploited, this ketone
analogue was chosen because it affords more protons useful for
measuring atom interdistances than the ester analogues. Correla-
tions observed for compound 5 were converted into three distance
intervals based on the relative intensities of the signals (Table 4).
Intramolecular distances were calculated for the 42 low energy
conformers of the conformational space of compound 5. A set of
9 conformers complied with NMR constraints and presented
torsion angles T1 and T3 of respectively 90° and 180°. These con-
formers display torsion angle values for T2 and T4 in the following
ranges: 180° < T2 < 299° and T4 ꢀ 20° or ꢀ140°. In the crystal
structure of compound 2, these torsion angles are T2 = 189.5°
and T4 = 21.4°, close to a low energy NMR and SAR compliant con-
former of 5 (Fig. 3). At this stage of the project, using molecular
modelling as well as spectroscopic studies on the only KCC2 block-
ers series known to us at the time, we were able to restrict the bio-
active conformations of our series to less than a dozen conformers
where T1 ꢀ 90°, T3 ꢀ 180°, 180° < T2 < 299° and T4 ꢀ 20° or
24
25
12
11
21
4
3
T4
7
23
T1= C6-O7-C8-C9
T2= C5-C2-C6-O10
T3= C5-N1-C13-O14
T4= C5-C2-C11-C12
9
5
O
2
20
8
T1
6
N1
T2
T3
O
13
10
15
O14
(R)-1
Figure 2. Chemical structure of benzyl N-acetyl-2-benzylprolinate (R)-1 featuring
the torsion angles considered for the study.
n?p⁄ electronic effect with the proximal ester group as described
by Hinderaker and Raines.13 In order to test this hypothesis a con-
strained analogue, compound 9, in which T3 equals 180° was syn-
thesized and was shown to display an IC50 comparable to
compound 1 (Table 3). This amide trans conformation was further
confirmed by the ROESY NMR data generated on compound 5. In-
deed, the spectra obtained for this compound only displays one set
of signals that could be attributed to the trans rotamer (Table 4,
correlation 15–5). Moreover, the crystal structure14 of compound
2, a close analogue of 1 with an indoline core scaffold, was ob-

F. Lebon et al. / Bioorg. Med. Chem. Lett. 22 (2012) 3978–3982
3981
Table 2
KCC2 activities for bioisoster compounds 3–8
R1
R2
N
O
3-8
a
Compound
R1
R2
IC50
100
(lM)
36
4
5
H
H
F
CONHCH2Ph
Figure 3. Crystallographic structure of compound 2 (cyan), a close analogue of 5
from the indoline family. Torsion angles are T1 = 215.8°, T2 = 189.5°, T3 = 172.9°
and T4 = 21.4°. These torsion angles are close to a low energy NMR and SAR
compliant conformer (grey) of 5 where T1 = 85.9°, T2 = 223.4°, T3 = 178.8° and
T4 = 22.4°.
COCH2CH2Ph
COCH2CH2Ph (p-F)
CH2OCH2Ph
0.3
0.2
20
6
H
N
*
7
8
H
H
1.3
O
Table 5
N
*
KCC2 activities for coumarin compounds 10–11
12.6
a
Compound
Structure
IC50
(l
M)
O
a
Rb+ flux assay on cloned rat KCC2.8
O
O
N
O
N
10
4
Table 3
KCC2 activities for constrained analogue 9
O
a
Compound
Structure
IC50
(
lM)
N
O
N
11
0.6
Br
O
9
1
O
a
Rb+ flux assay on cloned rat KCC2.8
N
O
O
ꢀ140°. Torsion angles T2 and T4 needed to be further refined in or-
der to obtain a pharmacophore model for KCC2 blockers.
Following the identification of a novel coumarin-based chemi-
cal series of KCC2 blockers by HTS (Table 5), we chose the rather
rigid compound 10 to deduce the common pharmacophoric fea-
tures with compound 1. A pharmacophore model was derived
using DISCOtech implemented within the SYBYL 7.2 software pack-
age.15 The conformational space of compound 10 was computed
and used as input of DISCOtech, alongside the previously identified
plausible bioactive conformations of compound 1. Several
pharmacophore hypotheses were generated around the possible
compound superimpositions. Among them;one model was in
agreement with the known SAR. It involves one of the lowest en-
ergy conformer of compound 1 (0.4 kcal/mol from the global min-
imum) characterised by the following torsion angles: T1 ꢀ 90°,
T2 ꢀ 195°, T3 ꢀ 180° and T4 ꢀ 20°. Four common pharmacophoric
features were pointed out: 2 H-bond acceptors and 2 aromatic
rings. In this model, the amidic oxygen atom of compound 10 cor-
responds to the acetamide of compound 1 while the coumarin car-
bonyl oxygen atom is aligned on compound 1 ester function. The
benzyl ring and the coumarin moiety in compound 10 are aligned
respectively on the benzyl ester and the 2-benzyl group of com-
pound 1 (Fig. 4).
a
Rb+ flux assay on cloned rat KCC2.8
Table 4
ROESY NMR data for compound 5
26
26
F
F
22
19
16
18
17
22
19
16
18
17
F27
F27
24
24
12
11
21
25
12
21
4
25
3
4
3
23
11
7
23
9
7
5
5
9
2
20
2
8
20
6
N1
8
6
N1
O
13
O
10
13
10
15
O14
15
O14
5
5
Protons
Relative intensitiesa
(ROESY)
H–H interatomic
distances (Å)
5a–7
11a–7
3a–11a
11b–3b
15–5a
15–5b
0.6
0.6
0.3
0.11b
0.9
1.1
0.8
1
3.7–5
3.7–5
3.7–5
3.7–5
2.5–3.7
2.5–3.7
2.5–3.7
2.5–3.7
2.5–3.7
2.5–3.7
2.5–3.7
<2.5
3a–11b
3b–7
Previously reported SAR6 indicated that the KCC2 affinity is in-
creased when the 2-benzyl group of compound 1 is substituted in
para with a bulky substituent. This corresponds to the 7-position of
the coumarin (yellow arrow) according to the pharmacophore
model. This hypothesis was confirmed by compound 11 where
adding a bromine substituent led to a seven-fold increase in
potency.
7–20/21
8–20/21
11a–16/17
11b–16/17
1
1 (ref)
1.1
1.4
a
200 ms Spinlock;measured in aceton-d6.
Measured in CDCl3.
b

3982
F. Lebon et al. / Bioorg. Med. Chem. Lett. 22 (2012) 3978–3982
Pacico for the testing of compounds on KCC2. The authors are
grateful to Sandrine Rome and Bernadette Norberg (FUNDP, Na-
mur) for the crystallographic support.
References and notes
1. Song, L.; Mercado, A.; Vazquez, N.; Xie, Q.; Desai, R.; George, A. L.; Gamba, G.;
Mount, D. B. Mol. Brain. Res. 2002, 103, 91.
2. (a) Rivera, C.; Voipio, J.; Payne, J. A.; Ruusuvuori, E.; Lahtinen, H.; Lamsa, K.;
Pirvola, U.; Saarma, M.; Kaila, K. Nature 1999, 397, 251; (b) Blaesse, P.;
Airaksinen, M. S.; Rivera, C.; Kaila, K. Neuron 2009, 61, 820.
3. De Koninck, Y. Curr. Opin. Pharmacol. 2007, 7, 93.
4. Gamba, G. Physiol. Rev. 2005, 85, 423.
5. Hesdorffer, D. C.; Stables, J. P.; Hauser, W. A.; Annegers, J. F.; Cascino, G. Ann.
Neurol. 2001, 50, 458.
6. Pégurier, C.; Bosman, N.; Collart, P.; Delporte, M. L.; Leclercq, K.; Lengelé, S.;
Kanduluru, A. K.; Meunier, S.; Pacico, N.; Vadali, L. R.; Wagner, A.; Wolff, C.;
Provins, L. Bioorg. Med. Chem. Lett. 2010, 20, 2542.
7. Delpire, E.; Days, E.; Lewis, L. M.; Mi, D.; Kim, K.; Lindsley, C. W.; Weaver, C. D.
Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 5383.
Figure 4. Superimposition of compound 1 (C atoms in white) and 10 (C atoms in
cyan). The common pharmacophoric features are HB acceptors (green spheres) and
aromatic rings (orange spheres).
8. The rat isoform of KCC2 was cloned and stably expressed in a human liver
adenocarcinoma cell line (SK-Hep). The KCC2 assay consists in the activation of
the co-transporter by the alkylating agent N-ethylmaleimide in the presence of
the test compound and to induce the Rb+ influx by adding 5 mM RbCl to the
extracellular medium. The Rb+ influx is stopped after 10 min by several
washing steps and the amount of intracellular Rb+ is determined by atomic
absorption spectroscopy.
In conclusion, a multi-disciplinary approach was used to iden-
tify the potential bioactive conformations of our lead series. Sev-
eral constrained benzyl prolinate analogues were synthesized
and tested in order to evaluate the importance of the torsion angles
on KCC2 activity. T1 ꢀ 90° and T3 ꢀ 180° were rapidly considered
as key pharmacophoric constraints. NMR ROESY experiments were
then performed to further assess torsion angles T2 and T4. X-ray
diffraction experiments on one analogue have further confirmed
the NMR observations. Finally, the first KCC2 pharmacophore mod-
el was derived by the superimposition of a few plausible bioactive
conformations of the benzyl prolinate compound 1 and a rigid li-
gand 10, belonging to a structurally-unrelated chemical family.
9. (R)-1 was confirmed as the active enantiomer of
1 by a synthesis from
commercially available (2R)-benzylproline.
10. Phillips, A. J.; Uto, Y.; Wipf, P.; Reno, M. J.; Williams, D. R. Org. Lett. 2000, 2,
1165.
11. Duggan, H. M. E.; Hitchcock, P. B.; Young, D. W. Org. Biomol. Chem. 2005, 3,
2287.
12. Conformational spaces were computed using an internal procedure described
in Moureau, F.; Neuwels, M.; Dogimont, C.; Goldstein, S.; Massingham, R. Int. J.
Pept. Res. Ther. 1998, 5, 155. Only conformations with an energy lower than
5 kcal/mol were considered.
13. Hinderaker, M. P.; Raines, R. T. Protein Sci. 2003, 12, 1188.
14. CCDC 801742 contains the supplementary crystallographic data for this paper
under the name OCOKAU. These data can be obtained free of charge from The
Cambridge Crystallographic Data Centre via http://www.ccdc.cam.ac.uk/
data_request/cif.
Acknowledgements
This work was financed in part by the Walloon Region (Bel-
gium) under the Convention n°5296. The authors thank Nathalie
15. SYBYL 7.2. Tripos Associates, Inc: St Louis, MO, 2006