Discovery of heterocyclic nonacetamide synaptic vesicle protein 2A (SV2A) ligands with single-digit nanomolar potency: Opening avenues towards the first SV2A positron emission tomography (PET) ligands

Article abstract of DOI:10.1002/cmdc.201300482

The role of the synaptic vesicle protein 2A (SV2A) protein, target of the antiepileptic drug levetiracetam, is still mostly unknown. Considering its potential to provide in vivo functional insights into the role of SV2A in epileptic patients, the development of an SV2A positron emission tomography (PET) tracer has been undertaken. Using a 3D pharmacophore model based on close analogues of levetiracetam, we report the rationale design of three heterocyclic non-acetamide lead compounds, UCB-A, UCB-H and UCB-J, the first single-digit nanomolar SV2A ligands with suitable properties for development as PET tracers. Avenues towards imaging... of synaptic vesicle protein 2A (SV2A) in vivo were opened after the rationale discovery of the first non-acetamide SV2A ligands, displaying single-digit nanomolar potency, using a 3D pharmacophore model. An extensive profiling of the most potent compounds allowed the identification of three leads (see figure) as potential candidates for positron emission tomography (PET) ligand development.

Full text of DOI:10.1002/cmdc.201300482

CHEMMEDCHEM
COMMUNICATIONS
DOI: 10.1002/cmdc.201300482
Discovery of Heterocyclic Nonacetamide Synaptic Vesicle
Protein 2A (SV2A) Ligands with Single-Digit Nanomolar
Potency: Opening Avenues towards the First SV2A
Positron Emission Tomography (PET) Ligands
Joꢀl Mercier,*[a] Laurence Archen,[a] Vꢁronique Bollu,[a] Stꢁphane Carrꢁ,[a] Yves Evrard,[a]
Eric Jnoff,[a] Benoꢂt Kenda,[a] Bꢁnꢁdicte Lallemand,[a] Philippe Michel,[a] Florian Montel,[a]
Florence Moureau,[a] Nathalie Price,[a] Yannick Quesnel,[a] Xavier Sauvage,[b] Anne Valade,[a]
and Laurent Provins[a]
The role of the synaptic vesicle protein 2A (SV2A) protein,
target of the antiepileptic drug levetiracetam, is still mostly un-
known. Considering its potential to provide in vivo functional
insights into the role of SV2A in epileptic patients, the devel-
opment of an SV2A positron emission tomography (PET) tracer
has been undertaken. Using a 3D pharmacophore model
based on close analogues of levetiracetam, we report the ra-
tionale design of three heterocyclic non-acetamide lead com-
pounds, UCB-A, UCB-H and UCB-J, the first single-digit nano-
molar SV2A ligands with suitable properties for development
as PET tracers.
(racetams) in both rodent and human brain. A remarkable and
strong correlation between the SV2A affinity of several series
of racetams and their antiseizure potency in the audiogenic
mouse model of epilepsy has been reported,[3–6] therefore sug-
gesting a major role played by SV2A in the mechanism of
action of levetiracetam.
SV2 proteins are 12-transmembrane-region glycoproteins
present on the synaptic vesicles of all neural cells and are ho-
mologous with the major facilitator superfamily (MFS), a large
and broadly distributed family of transporters. SV2A is the
most widely expressed among the three SV2 isoforms (SV2A,
SV2B and SV2C) and is ubiquitously distributed in the
brain.[7–12] Although its exact molecular function is still unclear,
it is believed to play a role in the exocytosis of neurotransmit-
ters and to act as a modulator of vesicle fusion.
Epilepsy, one of the most common chronic neurological disor-
ders, effects about 1% of the population worldwide and is
characterized by recurrent spontaneous seizures caused by dis-
turbances in the electrical activity of the brain. Epilepsy has
a tremendous impact on the individuals afflicted with this dis-
ease, as well as on their families.[1] Chronic pharmacotherapy is
the primary treatment for epileptic patients, aiming at main-
taining them seizure-free. However, despite the recent intro-
duction of second- and third-generation anticonvulsants, re-
fractory epilepsy is still a significant unmet medical need. In
a prospective study, Kwan and Brodie[2] show that fewer than
50% of patients with epilepsy become seizure-free after begin-
ning an initial antiepileptic medication and that more than
30% of patients cannot be adequately treated with existing
antiepileptic drugs because of either lack of efficacy or poor
tolerability.[1]
Synaptic vesicle protein 2A (SV2A) has been identified as the
unique brain-specific protein binding site of levetiracetam
(Keppraꢃ; compound A in Scheme 1) and structural analogues
[a] J. Mercier, L. Archen, V. Bollu, S. Carrꢀ, Y. Evrard, Dr. E. Jnoff, Dr. B. Kenda,
B. Lallemand, Dr. P. Michel, Dr. F. Montel, Dr. F. Moureau, N. Price,
Dr. Y. Quesnel, Dr. A. Valade, Dr. L. Provins
Global Chemistry, UCB NewMedicines, UCB Pharma
Chemin du Foriest, B-1420 Braine-L’Alleud (Belgium)
E-mail: joel.mercier@ucb.com
[b] Dr. X. Sauvage+
Cyclotron Research Center, University of Liꢁge
Allꢀe du 6 Aoꢂt, Bldg. B30, Sart Tilman, 4000 Liꢁge (Belgium)
[+] Present Address: INEOS Olefins & Polymers Europe, Brussels (Belgium)
Scheme 1. Acetamide and nonacetamide synaptic vesicle protein 2A (SV2A)
ligands.
ꢄ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 693 – 698 693

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
As part of the strategy aimed at a better understanding of
the role and function of SV2A, the design of a suitable SV2A
positron emission tomography (PET) tracer allowing 1) to pro-
vide critical information on the target engagement of our
drugs in clinical development, and 2) to characterize for the
first time the distribution and expression of SV2A in epileptic
patients, was considered of primary importance.
Positron emission tomography (PET) is a powerful nuclear
imaging technique allowing the in vivo quantification of
a radioligand labelled with a short-lived positron-emitting iso-
tope, typically [18F] or [11C]. There have been multiple reviews
on the benefits of PET imaging in the support of central nerv-
ous system (CNS) drug discovery efforts.[13–15]
Given the proven clinical efficacy of levetiracetam and the
encouraging phase IIb results of brivaracetam (structure B in
Scheme 1),[16–18] a second-generation SV2A modulator currently
under clinical evaluation, as potent antiepileptic drugs with an
excellent tolerability, we initiated an extensive medicinal
chemistry program aiming at the identification of new ligands
with improved potency and suitable properties for develop-
ment as SV2A PET tracers.
As previously reported,[19] a systematic exploration of the
pyrrolidinone acetamide scaffold (racetam) mainly revealed
that 4-substitution of the pyrrolidinone ring by small hydro-
phobic groups or replacement by an indolone bicycle (struc-
ture C in Scheme 1) improves both the in vitro SV2A affinity
and the in vivo antiseizure potency. This strategy led to the se-
lection of the phase III clinical candidate brivaracetam (struc-
ture B in Scheme 1).
In the absence of available structural information on SV2A,
a ligand-based drug design approach has been used to further
explore the chemical diversity accommodated by the levetira-
cetam binding site present on SV2A. In order to define the
SV2A pharmacophore, the conformational space of six diverse
SV2A ligands with pIC50 values ranging between 4.5 and 7
were compared using DISCO.[20] The very good potency of sev-
eral prototypical nitrogen-containing heterocyclic derivatives
(e.g., imidazole) revealed that the acetamide side chain is
acting as a hydrogen-bond acceptor rather than donor. So,
among the different pharmacophore hypothesis generated,
only those exhibiting the hydrogen-acceptor character of the
side chain and its projected acceptor site on the protein were
retained. We then reviewed all those models in light of known
structure–activity relationships. The only model consistent with
all pieces of information incorporates the key element of two
hydrogen acceptors, represented by the carbonyl of the pyrro-
lidinone pointing towards the side chain and either the sp2 ni-
trogen of the heterocycle or the acetamide carbonyl moiety
(Figure 1). This suggested that various small heteroaromatic
rings could provide the key hydrogen-bond acceptor feature
and advantageously replace the carboxamide.
Figure 1. Synaptic vesicle protein 2A (SV2A) pharmacophore; superimposi-
tion of amidic and heterocyclic derivatives.
around the pyrrolidinone ring and identified a few heteroaro-
matics displaying unprecedented single-digit nanomolar SV2A
potency. In particular, imidazole, pyrazole and pyridine deriva-
tives proved to be the heterocycles of choice that led to
a quantum leap in terms of SV2A potency.
Capitalizing on our experience in the preparation of pyrroli-
dinone rings as well as on a large library of available building
blocks, we decided to prepare our compounds by disconnec-
tion at the methylene linker between the pyrrolidinone and
the heteroaromatic ring (Scheme 2). The general procedures
for the preparation of all the compounds described herein
have been reported earlier.[21–23] The 4-substituted pyrrolidi-
nones were obtained from the corresponding aldehydes via
a Wittig–Horner reaction followed by the addition of nitrome-
thane and a subsequent reduction–cyclization step as depicted
in Scheme 3. The choice of catalyst for the reduction of the
nitro group is dependent upon the nature of the R1 substitu-
ent as well as on the targeted building block. Raney Nickel
was used as the standard catalyst for compounds with halo-
gen-containing R1 groups, affording the cyclized pyrrolidinone
after reaction. In order to obtain the intermediate amino-ester,
reduction was performed using palladium on charcoal in the
Using this information, we undertook a rather systematic ex-
ploration of small 5- or 6-membered heteroaromatic scaffolds
displaying a hydrogen-bond acceptor in the appropriate posi-
tion. More than 500 compounds were prepared by both com-
binatorial and medicinal chemistry teams, and their evaluation
confirmed the known structure–activity relationships (SAR)
Scheme 2. Disconnections at the methylene linker.
ChemMedChem 2014, 9, 693 – 698 694
ꢄ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
pursue the SAR exploration with racemic mixtures. It eventual-
ly led to the identification of several (poly)fluorinated phenyls
as the best compromise between SV2A potency and metabolic
stability (data not shown).
These substitutions have been further combined with the 5-
imidazolyl and 4-pyrazolyl heterocycles also leading to very
potent compounds (compounds 9–14 in Table 1). Interestingly,
1-N-methylation of the 5-imidazolyl heterocycle is slightly ben-
eficial to the SV2A binding potency. By contrast, only the N-un-
substituted 4-pyrazolyl motif is tolerated, affording comparable
potency as the imidazolyl ring.
Scheme 3. Preparation of the pyrrolidinone building blocks. Reagents and
conditions: a) (Carbethoxymethylene)triphenylphosphorane, THF, 208C
(~80%); b) DBU, CH3NO2, 208C (~90%); c) H2 (20 psi), raney Ni, EtOH, 208C,
then toluene, reflux (80–90%); d) H2 (40 psi), 12n HCl, Pd-C, EtOH, 208C
(~75%).
Besides the 5-membered heterocycles, 3- and 4-pyridyl rings
have also been prepared (Scheme 5B), leading to similar re-
sults. Indeed, polyfluorinated phenyls in the R1 position, in par-
ticular the (+)-enantiomer of the 3,4,5-trifluorophenyl group
(compounds 17, 19, 21, 23 in Table 2), afforded low-nanomolar
SV2A binders. Interestingly, the 4-pyridyl heterocycle is slightly
less sensitive to substitution in the R1 position (c.f., com-
pounds 15, 17 and 18 with 16, 19 and 20 in Table 2). This un-
expected profile made the 4-pyridyl entity particularly interest-
ing for pursuing further exploration.
presence of one molar equivalent of concentrated hydrochloric
acid to prevent cyclization.
Compounds 1–8 bearing a N-linked imidazole ring were all
prepared from intermediates G via formylation, followed by
the activation of the hydroxy group and final nucleophilic sub-
stitution (Scheme 4). Depending on the availability and accessi-
bility of the building blocks, compounds 9–24 bearing a C-
linked heterocycle were prepared either by direct alkylation of
the corresponding benzylic halide (Scheme 5A) or by reductive
amination of the corresponding carbaldehyde (Scheme 5B).
The synthesis of compounds bearing an imidazole or a pyrazole
ring required the preliminary protection of the free NH group,
using classical protecting groups, in order to run the reductive
amination with good yield.
Scheme 4. Preparation of N-linked imidazoles. Reagents and conditions: a) aq
formaldehyde, NaOH, EtOH, 608C (~80%); b) carbonyl-diimidazole, CH3CN,
208C (~80%).
As anticipated from the 3D pharmacophore superimposi-
tions (Figure 1), compounds 1–
24 have shown a high binding
potency to SV2A. Capitalizing on
the ease of preparation of the 1-
imidazolyl feature, several key R1
substituents have been prepared
to quickly demonstrate a very
good transferability of SAR from
the acetamide subseries (levetir-
acetam and brivaracetam).[19]
Indeed, as exemplified by
compounds 1–8, small linear
alkyl chains, in particular n-
propyl, as well as substituted
phenyls provided compounds
with the best SV2A potency ob-
tained so far, achieving for the
first time single-digit nanomolar
potency (e.g., compound 3).
Both enantiomers have been
prepared for selected R1 substi-
tutions
(compounds 1–4
in
Scheme 5. Preparation of C-linked heteroaromatics. Reagents and conditions: a) NaH, n-Bu4NI, THF, 208C (~75%);
b) Et3N, NaBH4, MeOH, 208C, then 658C (~45%), or c) NaBH3CN, AcOH, MeOH, 208C then 658C (~45%); d) R3 =H:
AcOH, MeOH, 208C (~80%); R3 =Me: CH3I, CH3CN, then AcOH, MeOH, 208C (~80%); e) n-Bu4NF, THF, 858C
Table 1) and provided no signifi-
cant SV2A affinity difference be-
tween enantiomers allowing to (~15%).
ꢄ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 693 – 698 695

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
Table 1. In vitro pharmacological data for compounds 1–14.
Table 2. In vitro pharmacological data for compounds 15–24.
Compd
R1
R2
Enantiomer
pIC50 (SV2A)[a]
Compd
R1
R2
Enantiomer
Rac.
pIC50 (SV2A)[a]
1
n-Pr
n-Pr
Ph
À
8.0Æ0.1
15
n-Pr
n-Pr
7.2Æ0.1
2
3
4
5
6
7
8
+
7.4[b]
16
17
18
19
20
Rac.
+
6.4Æ0.1
7.9Æ0.1
7.3Æ0.1
8.1Æ0.1
6.5Æ0.1
À
8.1Æ0.1
8.1Æ0.1
7.8[b]
(3,4,5-F)-Ph
(3,4,5-F)-Ph
(3,4,5-F)-Ph
(3,4,5-F)-Ph
Ph
+
À
(3,4,5-triF)-Ph
(3-F)-Ph
Rac.
Rac.
Rac.
Rac.
+
7.8Æ0.1
7.9[b]
À
(3,5-diF)-Ph
(2,3,5-triF)-Ph
21
22
(3,4,5-F)-Ph
(3,4,5-F)-Ph
+
7.8Æ0.1
6.8Æ0.1
8.0Æ0.0
À
9
n-Pr
Rac.
6.8[b]
23
24
(3,4,5-F)-Ph
(3,4,5-F)-Ph
+
8.2Æ0.1
7.1Æ0.1
10
11
12
n-Pr
Rac.
À
7.3Æ0.1
7.0Æ0.2
7.9Æ0.1
À
(3,5-diF)-Ph
(3,5-diF)-Ph
[a] Human synaptic vesicle protein 2A (SV2A) pIC50 values are Àlog of the
unlabeled drug concentration that inhibits [3H]levetiracetam or an ana-
logue as described in Ref. [4–6]. All values represent the mean of two in-
dependent determinations. Each pIC50 value lies within 0.2 log units of
the mean.
+
13
14
(2,3,5-triF)-Ph
(2,3,5-triF)-Ph
À
7.5Æ0.1
7.7Æ0.1
Various substituents have been introduced in positions 2
and 3 of the 4-pyridyl heterocycle. Although several kind of
substituents are tolerated in both positions (data not shown),
the best SV2A potency is achieved by the addition of small lip-
ophilic groups in position-3 (compounds 21 and 23 Table 2).
Bearing in mind our objective to use our new high-affinity
SV2A ligands in our PET tracer development program, we de-
cided to investigate whether some of the aforementioned
compounds could be suitable candidates.
+
[a] Human synaptic vesicle protein 2A (SV2A) pIC50 values are Àlog of the
unlabeled drug concentration that inhibits [3H]levetiracetam or an ana-
logue as described in Ref. [4–6]. All values represent the mean of two in-
dependent determinations. Each pIC50 value lies within 0.2 log units of
the mean. [b] Binding performed on rat cortex homogenate as described
in Ref. [4–6].
There are multiple review articles[24–27] dealing with the
design of PET ligands, and these typically cite high nonspecific
protein binding and poor brain penetration among the most
ꢄ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 693 – 698 696

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
Table 3. In vivo blood–brain barrier (BBB) permeation data for compounds 12, 21 and 23.[a]
Compd
pIC50
(SV2A)
Solubility[b]
LogD[c]
CNS PET
MPO[d]
Clint[e]
[mLminÀ1 mgÀ1 proteinÀ1
Papp/ER[f]
[nmsÀ1
Free B/P
ratio[g]
Fu [%][h]
rBrain
[mgmLÀ1
]
]
]
rPlasma
hPlasma
12
21
23
7.9
7.8
8.2
>1.5
0.3
0.4
1.4
2.3
2.5
5.4
4.6
4.6
20
12
16
383/1.2
707/0.7
323/0.8
0.6
1.6
1.6
64
36
33
12
8
4.5
69
34
31
[a] Parameters obtained after single i.v. (at 0.1 mgkgÀ1) dosing to male Wistar rats (n=2). All experiments were approved by the local ethics committee for
animal experimentation according to Belgian law; [b] Measured at pH 7.4 at 258C after 24 h; [c] Measured from octanol/water partition coefficient at 258C
and pH 7.4; [d] Calculated according to Ref. [27]; [e] Intrinsic clearance (Clint) in human microsomes; [f] A-B apparent permeability measured on CACO-2
cells/Efflux ratio (ER) defined as the ratio of apparent permeabilities: B-A/A-B; [g] Ratio of free brain concentration versus free plasma concentration at
20 min post-administration in rats; [h] Human (h) and rat (r) plasma protein binding and rat brain tissue binding were determined by equilibrium dialysis
for 4 h at 378C. Samples were analyzed by LC/MS-MS.
frequent causes for attrition. In an attempt to improve the suc-
cess rate, this highly complex process has too often been re-
duced to focusing on a very narrow lipophilicity range as
a way to ensure more suitable properties.
opment as PET tracers and their ability to provide meaningful
PET imaging data will be reported in subsequent papers, as
UCB-A (12), UCB-H (21)[29] and UCB-J (23).
In contrast to what is usually reported in the literature, we
decided to perform a more comprehensive profiling in order
to select the best candidates. Several compounds were there-
fore further profiled in physicochemical and drug metabolism
and pharmacokinetic (DMPK) studies, leading to the selection
of compounds 12, 21 and 23 (Table 3). These compounds are
among the most potent SV2A ligands reported to date (IC50 =
6–15 nm) affording a Bmax/Kd ratio (Bmax/Kd =20–50; SV2A Bmax
~300 nm in human brain)[28] in a suitable range to measure
a specific signal in vivo. In addition, they also provide the best
overall absorption, distribution, metabolism, and excretion
(ADME) profile. Indeed, all three compounds not only display
a measured logD value in the appropriate range, but they also
show high permeability, no P-glycoprotein (P-gp) efflux liability
and very good CNS PET multiparameter optimization (MPO)
score.[27] This adequate in silico and in vitro profile translates
into excellent brain penetration (free B/P ratio >1 for com-
pounds 21 and 23; Table 3) as measured in an i.v. rat blood–
Acknowledgements
The authors would like to thank Dr. B. Mathieu, Mrs. G. Longfils,
Mrs. A. Descamps for lipophilicity and solubility measurements,
Dr. V. Pinilla and her team for chromatographic purifications, Dr.
R. Dieden, Mr. A. Fauconnier, Mr. J. Claessens, and Mrs. C. Derwa
for analytical assistance, and Dr. E. Jigorel for in vitro PK meas-
urements. Part of this work was funded by the Walloon Region
Public–Private Partnership NEUROCOM (Belgium), convention no.
716747.
Keywords: epilepsy
·
imaging agents
·
levetiracetam
·
positron emission tomography (PET)
· synaptic vesicle
proteins · tracers
[1] W. Lçscher, H. Klitgaard, R. E. Twyman, D. Schmidt, Nat. Rev. Drug Dis-
covery 2013, 12, 757–776 and references therein.
[2] P. Kwan, M. J. Brodie, N. Engl. J. Med. 2000, 342, 314–319.
[3] H. Klitgaard, A. Matagne, J. Gobert, E. Wꢅlfert, Eur. J. Pharmacol. 1998,
353, 191–206.
[4] B. Fuks, M. Gillard, P. Michel, B. A. Lynch, P. Vertongen, P. Leprince, H.
Klitgaard, P. Chatelain, Eur. J. Pharmacol. 2003, 478, 11–19.
[5] M. Gillard, B. Fuks, P. Michel, P. Vertongen, R. Massingham, P. Chatelain,
Eur. J. Pharmacol. 2003, 478, 1–9.
[6] B. A. Lynch, N. Lambeng, K. Nocka, P. Kensel-Hammes, S. M. Bajjalieh, A.
Matagne, B. Fuks, Proc. Natl. Acad. Sci. USA 2004, 101, 9861–9866.
[7] D. Reigada, J. Diez-Perez, P. Gorostiza, A. Verdaguer, I. G. de Aranda, O.
Pineda, J. Vilarrasa, J. Marsal, J. Blasi, J. Aleu, C. Solsona, Proc. Natl. Acad.
Sci. USA 2003, 100, 3485–3490.
brain barrier (BBB) permeation study performed at 0.1 mgkgÀ1
.
Plasma protein binding and rat brain tissue binding have also
been measured as a surrogate of in vivo nonspecific binding.
Despite a significantly higher binding to brain tissue compared
with plasma proteins, the data obtained are consistent with
what is reported for successful PET ligands (i.e., brain Fu
>5%).[27]
In addition to having a suitable ADME profile, good PET
ligand candidates must allow the ready incorporation of a posi-
tron-emitting isotope. Although their preparation might not
be straightforward, compounds 12, 21 and 23 bear suitable
functional groups for radiolabelling either through methylation
with [11C]iodomethane or through nucleophilic aromatic substi-
tution with [18F]fluoride.
[8] R. Janz, Y. Goda, M. Geppert, M. Missler, T. C. Sꢅdhof, Neuron 1999, 24,
1003–1016.
[9] T. Xu, S. M. Bajjalieh, Nat. Cell Biol. 2001, 3, 691–698.
[10] A. E. Schivell, R. H. Batchelor, S. M. Bajjalieh, J. Biol. Chem. 1996, 271,
27770–27775.
[11] R. A. Pyle, A. E. Schivell, H. Hidaka, S. M. Bajjalieh, J. Biol. Chem. 2000,
275, 17195–17200.
In conclusion, we have used a 3D pharmacophore model
generated from diverse SV2A ligands to identify a new series
of heterocyclic analogues with drastically improved potency.
An extensive profiling of the most potent SV2A ligands identi-
fied allowed the identification of three lead compounds (12,
21 and 23) as candidates for PET ligand development, opening
up avenues towards the in vivo imaging of SV2A. Their devel-
[12] S. M. Bajjalieh, G. D. Frantz, J. M. Weimann, S. K. McConnell, R. H. Schel-
ler, J. Neurosci. 1994, 14, 5223–5235.
[13] C.-M. Lee, L. Farde, Trends Pharmacol. Sci. 2006, 27, 310–316.
[14] D. F. Wong, J. Tauscher, G. Grꢅnder, Neuropsychopharmacology 2009, 34,
187–203.
[15] N. Seneca, Drugs Future 2011, 36, 601–613.
[16] A. Matagne, D. G. Margineanu, B. Kenda, P. Michel, H. Klitgaard, Br. J.
Pharmacol. 2008, 154, 1662–1671.
ꢄ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 693 – 698 697

CHEMMEDCHEM
COMMUNICATIONS
www.chemmedchem.org
[17] P. von Rosenstiel, Neurotherapeutics 2007, 4, 84–87.
[18] W. Van Paesschen, E. Hirsch, M. Johnson, U. Falter, P. von Rosenstiel, Epi-
lepsia 2013, 54, 89–97.
[19] B. M. Kenda, A. C. Matagne, P. E. Talaga, P. M. Pasau, E. Differding, B. I.
Lallemand, A. M. Frycia, F. G. Moureau, H. V. Klitgaard, M. R. Gillard, B.
Fuks, P. Michel, J. Med. Chem. 2004, 47, 530–549.
[23] B. Kenda, P. Michel, Y. Quesnel, A. Ates, L. Turet, J. Mercier, (UCB Pharma
S.A., Brussels, Belgium), PCT Int. Pub. WO 2006/128693, 2006.
[24] D. J. McCarthy, C. Halldin, J. D. Andersson, M. E. Pierson, Annu. Rep. Med.
Chem. 2009, 44, 501–513.
[25] M. Laruelle, M. Slifstein, Y. Huang, Mol. Imaging Biol. 2003, 5, 363–375.
[26] V. W. Pike, Trends Pharmacol. Sci. 2009, 30, 431–440.
[27] L. Zhang, A. Villalobos, E. M. Beck, T. Bocan, T. A. Chappie, L. Chen, S.
Grimwood, S. D. Heck, C. J. Helal, X. Hou, J. M. Humphrey, J. Lu, M. B.
Skaddan, T. J. McCarthy, P. R. Verhoest, T. T. Wager, K. Zasadny, J. Med.
Chem. 2013, 56, 4568–4579.
[28] M. Gillard, P. Chatelain, B. Fuks, Eur. J. Pharmacol. 2006, 536, 102–108.
[29] F. Bretin, G. Warnock, M. A. Bahri, J. Aerts, N. Mestdagh, T. Buchanan, A.
Valade, F. Mievis, F. Giacomelli, C. Lemaire, A. Luxen, E. Salmon, A. Seret,
A. Plenevaux, EJNMMI Res. 2013, 3, 35.
[30] SYBYL, v.6.1, Tripos Associates, St.Louis, MO, USA, 1994.
[31] M. Clark, R. D. Cramer III, M. Van Opdenbosch, J. Comput. Chem. 1989,
10, 982.
[20] Molecular structures were constructed using the SYBYL software.[30]
Energy minimizations were performed with the Tripos force field (5ꢆ
10À–3 kcalmolÀ1 ꢇ, gradient convergence).[31] A representative sample of
the conformational space was generated by CONFEX (CONFormational
EXploration), an in-house written procedure[32] which uses the DGEOM
algorithm,[33] run until convergence was reached. Briefly, for the latter
a sample of conformers (À200) was first generated and analyzed, dis-
carding conformers too high in energy or redundant (based on RMS
values obtained from the superposition of heavy atoms; C2 and C3
symmetries were taken into consideration). The above procedure was
repeated until no new conformers (according to the above-established
topological criteria) were obtained five times in a row. After energy
minimizations at a semi-empirical level (AM1-Gaussian 92; 3ꢆ10À–4 a.u.,
gradient convergence),[34] redundant structures were again eliminated
as above. These low-energy sets of conformers form the input to the
DISCO program,[35] which uses calculated ligand points and receptor
site points to find superposition rules for each structure (pharmaco-
phore mapping). Solutions chosen included a set of elements found in
all structures and inter element distances that matched within a desired
tolerance.
[32] CONFEX, M. Neuwels, E. Vandewater, UCB Pharma, Belgium.
[33] DGEOM, QCPE Program No. 590, J. M. Blaney, G. M. Crippen, A. Dearing,
J. S. Dixon, QCPE Bull. 1990, 10, 37.
[34] Gaussian 92, M. Frisch, J. Foreswan, J., A. Frisch, Gaussian Inc., Pitts-
burgh, PA, USA.
[35] Y. C. Martin, M. G. Bures, E. A. Danaher, J. De Lazzer, J. Lico, E. A. Pavlik,
J. Comput.-Aided Mol. Des. 1993, 7, 83.
[21] B. Kenda, P. Michel, Y. Quesnel, (UCB Pharma S.A., Brussels, Belgium),
PCT Int. Pub. WO 2005/054188, 2005.
Received: November 25, 2013
[22] B. Kenda, P. Michel, Y. Quesnel, A. Ates, L. Turet, J. Mercier, (UCB Pharma
S.A., Brussels, Belgium), PCT Int. Pub. WO 2006/128692, 2006.
Published online on January 20, 2014
ꢄ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 693 – 698 698