Synergism of virtual screening and medicinal chemistry: Identification and optimization of allosteric antagonists of metabotropic glutamate receptor 1

Article abstract of DOI:10.1016/j.bmc.2009.05.072

We report the identification of novel potent and selective metabotropic glutamate receptor 1 (mGluR1) antagonists by virtual screening and subsequent hit optimization. For ligand-based virtual screening, molecules were represented by a topological pharmacophore descriptor (CATS-2D) and clustered by a self-organizing map (SOM). The most promising compounds were tested in mGluR1 functional and binding assays. We identified a potent chemotype exhibiting selective antagonistic activity at mGluR1 (functional IC₅₀ = 0.74 ± 0.29 μM). Hit optimization yielded lead structure 16 with an affinity of K_i = 0.024 ± 0.001 μM and greater than 1000-fold selectivity for mGluR1 versus mGluR5. Homology-based receptor modelling suggests a binding site compatible with previously reported mutation studies. Our study demonstrates the usefulness of ligand-based virtual screening for scaffold-hopping and rapid lead structure identification in early drug discovery projects.

Full text of DOI:10.1016/j.bmc.2009.05.072

Bioorganic & Medicinal Chemistry 17 (2009) 5708–5715
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Synergism of virtual screening and medicinal chemistry: Identification and
optimization of allosteric antagonists of metabotropic glutamate receptor 1
Tobias Noeske a,b, Dina Trifanova c, Valerjans Kauss c, Steffen Renner a, Christopher G. Parsons a,
a,d,
Gisbert Schneider b, , Tanja Weil
*
*
a Merz Pharmaceuticals GmbH, Altenhöfer Allee 3, D-60438 Frankfurt, Germany
b Johann Wolfgang Goethe-University, Institute of Organic Chemistry and Chemical Biology ZAFES/CMP, Siesmayerstraße 70, D-60323 Frankfurt, Germany
c Latvian Institute of Organic Synthesis, 21 Aizkraukles, Riga LV 1006, Latvia
d National University of Singapore, 3 Science Drive 3, Singapore 117543, Singapore
a r t i c l e i n f o
a b s t r a c t
Article history:
We report the identification of novel potent and selective metabotropic glutamate receptor 1 (mGluR1)
antagonists by virtual screening and subsequent hit optimization. For ligand-based virtual screening, mol-
ecules were represented by a topological pharmacophore descriptor (CATS-2D) and clustered by a self-
organizing map (SOM). The most promising compounds were tested in mGluR1 functional and binding
assays. We identified a potent chemotype exhibiting selective antagonistic activity at mGluR1 (functional
Received 28 November 2008
Revised 27 May 2009
Accepted 28 May 2009
Available online 23 June 2009
IC50 = 0.74 0.29 lM). Hit optimization yielded lead structure 16 with an affinity of Ki = 0.024 0.001 lM
Keywords:
mGluR1
Metabotropic
Antagonist
Self-organizing map
Virtual screening
and greater than 1000-fold selectivity for mGluR1 versus mGluR5. Homology-based receptor modelling
suggests a binding site compatible with previously reported mutation studies. Our study demonstrates
the usefulness of ligand-based virtual screening for scaffold-hopping and rapid lead structure identifica-
tion in early drug discovery projects.
Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction
of mGluRs have been found to bind within the HD.5 The allosteric
binding site is considered to have a higher ‘druggability’ than the
G-protein coupled receptors (GPCRs) represent the largest
family of cell-surface receptors involved in signal transmission.1
They are divided into three families displaying distinct differences
concerning their sequence similarity and their mechanism of
receptor activation: Family 1—for example, rhodopsin/b-adrener-
gic receptors, family 2—for example, secretin receptors and family
3—for example, metabotropic glutamate receptors (mGluRs).2
mGluRs have been subdivided into three groups based on sequence
identity, signal transduction and pharmacology: Group I consists of
mGlu1 and mGlu5, group II comprises mGlu2 and mGlu3, and
group III holds mGlu4, 6, 7 and 8.3 mGluRs are characterized by a
large extracellular domain (ECD) containing a Venus flytrap mod-
ule for orthosteric agonist binding.4 After agonist binding to the
ECD, signal transduction proceeds via the heptahelical domain
(HD), which is composed of seven transmembrane helices linked
to each other by alternating extracellular and intracellular loops
and an intracellular domain (ICD) including the C terminus and
the G-protein interaction sites. Allosteric agonists and antagonists
orthosteric binding site6–8 and therefore, several attempts to iden-
tify novel allosteric modulators of different mGuRs have been
described.6,9 Negative modulation of mGluR1 has been suggested
as a potential therapeutic approach for a variety of indications,
for example, different pain states,3 depression10 and several forms
of cancer.11 Therefore, there is a vivid interest in the identification
of mGluR1 inhibitors suitable for in vivo experiments and several
potent mGluR1 antagonists have been identified.12–15 So far, most
known mGluR modulators were discovered applying high through-
put screening technologies. However, recently several successful
attempts applying ligand-based in silico approaches for the identi-
fication of hit compounds interacting with mGluRs have been
reported.15–17 Ligand-based approaches are fast and allow virtual
screening of large databases of hundreds of thousands of com-
pounds. Such methods are often based on the concept of molecular
similarity that states that molecules with similar features are likely
to exhibit similar biological responses.18,19 Recently, negative
allosteric modulators of mGluR1 and mGluR5 have been identified
applying ligand-based virtual screening methods such as homology
model-based virtual screening,18 alignment-free topological phar-
macophore descriptors,15 neural network ensembles16 or pharma-
cophore models20 that served as valuable starting points for hit
optimization approaches.
* Corresponding authors. Tel.: +49 6979824873; fax: +49 6979824880, tel.: +65
65161377; fax: +65 6779 1691 (T.W.).
E-mail addresses: G.Schneider@chemie.uni-frankfurt.de (G. Schneider),
chmweilt@nus.edu.sg (T. Weil).
0968-0896/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2009.05.072

T. Noeske et al. / Bioorg. Med. Chem. 17 (2009) 5708–5715
5709
In this article, we report the identification of novel selective
metabotropic glutamate receptor 1 (mGluR1) antagonists with
nanomolar affinities by virtual screening and subsequent hit
optimization. For ligand-based virtual screening, molecules were
represented by a topological pharmacophore descriptor (CATS-
2D) and clustered by a self-organizing map (SOM). Homology-
based receptor modelling suggests a binding site compatible
with previously reported mutation studies.15,20 We have focused
the hit optimization process on activity optimization as well as
achieving selectivity versus mGluR5, which represents the struc-
turally closest analogue sharing a high sequence identity with
mGluR1. mGluR1 antagonists often display limited selectivity
versus the mGluR5 receptor20 and it has been shown that the
combined administration of an mGluR1 and an mGluR5 antago-
nists resulted in the occurrence of more pronounced side-
effects.21 Therefore, selectivity versus the mGluR5 receptors is
considered crucial for the development of mGluR1 antagonists.
In addition, due to their binding site within the transmembrane
domain, the combination of high functional activity and accept-
able water-solubility for in vivo and in vitro studies is an addi-
tional key concern.
Twenty eight screening candidates were manually selected
from the top-ranking 60 compounds based on structural novelty
with respect to already known mGluR1 antagonists, ordered and
tested in both binding and functional assays. Details on the con-
duction of the assays were included in Section 4. We obtained a
screening result hit rate of >20% at a ligand concentration of
10
l
M: One compound was ‘highly active’ (<1
l
M), five com-
M), seven com-
M), and for 15
M). Bioactivity
pounds (18%) were ‘moderately active’ (1–15
pounds (25%) exhibited ‘low activity’ (15–40
compounds (54%) we observed ‘no activity’ (>40
l
l
l
data for molecules with propenone as a common structural moiety
are given in Table 1. Structures and data of the remaining com-
pounds from this study can be found in the Supplementary data
in S-Table 1. Noteworthy, all propenone-derivatives were inactive
at rat mGluR5 (rmGluR5).
2.2. Structure–activity relationship
Comparing hits 3 and 4 shows a fourfold preference of a
6-methyl substituent over a 7-methyl substituent at the quinoline
group. 1 versus 5—an analogue of 3 but with benzene (1) or a
p-methoxybenzene ring (5) instead of the thienyl-substituent—
showed a 12-fold deterioration for the addition of a methoxy-
group at the para-position of the benzene ring. A similar trend is
observed comparing 7 containing both unfavourable substituents
and 1 leading to a more than 20-times decreased antagonistic
activity (Table 1). We tried to explain the interaction of 1 with
rmGluR1 by means of our own homology model of rmGluR1
(Fig. 2a) that was published recently.15,20 The model suggests the
2. Results and discussion
2.1. Virtual screening
The self-organizing map (SOM) principle was introduced by
Kohonen in 1982,22 and has been applied to a variety of tasks in
chemistry and chemical biology since that time.23,24 We have
recently reported a combination of molecule encoding and cluster-
ing to predict compound selectivity,25 which was realized by a
topological pharmacophore descriptor26 and the SOM algo-
rithm.27,28 Here we employed this concept for virtual screening of
a large compound library for potential allosteric mGluR1 antago-
nists. First, we defined ‘drug-like chemical space’ using a SOM pro-
jection of drug-like compounds.25 Each molecule was encoded by a
150-dimensional descriptor (CATS-2D)26 giving scaled frequencies
of pairs of potential pharmacophore points (PPP) in the molecular
graph. The underlying dataset was the COBRA (v.3.12) collection of
pharmacologically active molecules.27
The SOM projection is topology-preserving, that is, molecules
which are located close to each other on the map, are also adjacent
in the original high-dimensional space. Here, a map with either
100 or 225 ‘neurons’ (clusters), each of which contains molecules
having certain pharmacophore features in common. In the next
step, a subset of 357 reference compounds including 213 published
and Merz in-house structures of allosteric mGluR1 antagonists was
mapped onto the trained SOMs. In both maps we visualized the
distribution of the reference molecules (Fig. 1). Considering that
these SOMs have a toroidal topology, both of them (10 ꢀ 10 and
15 ꢀ 15 grid) reveal one large cluster, one smaller cluster and sev-
eral distributed ‘activity islands’ containing reference compounds.
Neurons 8/7 and 6/6 displayed the highest density of mGluR1
antagonists (Fig. 1a).
presence of three hydrogen-bond donors (R6613.29/N74745.51
,
S6683.36 and T8157.39) in the allosteric antagonist binding pocket
of mGluR1, flanked by two hydrophobic regions (region I:
V7535.43, P7565.46, V7575.47; region II: L6402.58, P6452.63, V8197.43).
Ligand 1 was manually docked into the homology model to best
explain the observed structure–activity data, followed by energy
minimization within the receptor pocket according to a procedure
that has been published recently.15,20 Since no activity was found
in human mGluR1 (hmGluR1) for molecules 1–7 (data not shown),
the ligands were also assumed to interact with V7575.47, which
represents the single amino acid difference between rat and hu-
man receptors within the ligand-binding region15: V7575.47 (5.47
according to the numbering scheme of Ballesteros29) in rmGluR1
versus L757 in hmGluR1.15 Another possibility could be that the
amino acid species difference led to an alternate conformation of
the binding site. Given that many known ligands bind to both spe-
cies,15,20 the binding site difference should likely be local around
the mutation, and would also justify to place a part of the ligand
near to V7575.47. The predicted binding mode for 1 is shown in
Figure 2.
In the hypothesized binding mode of 1 the ligand interacts with
the hydrogen-bonding cluster of R6613.29 and N74745.51 via the lin-
ker carbonyl oxygen. The quinoline-nitrogen hydrogen-bond
acceptor interacts with T8157.39. Notably, the linker-conformation
orients both potential hydrogen-bonding groups in the same direc-
tion, towards the N-terminal domain of the receptor. According to
our receptor model, the para-methoxy group of the benzene ring of
1 would sterically interfere with the hydrophobic cluster at TM5, in
particular with V7575.47. At the opposite end of the binding site, a
7-methyl substituent at the quinoline of 1 would interfere with the
Finally, the Asinex Gold Collection 2003 (194,563 compounds)
was projected onto the SOMs. Again, the maps were coloured
according to compound density (Fig. 1b). Since we were interested
in only those Asinex compounds that contain pharmacophoric
features similar to known mGluR1 antagonists, we focused on neu-
rons 8/7 and 6/6 exclusively (Fig. 1): Molecules that are co-located
are assumed to have similar pharmacological activity.25,27,28 Neu-
ron 8/7 (10 ꢀ 10 map) contained 1864 molecules, and 6/6
(15 ꢀ 15 map) contained 749 molecules. All Asinex compounds
that were retrieved by both maps were selected and ranked
according to their distance to the cluster centroids.
hydrophobic cluster at TM2, most likely with P6452.63
.
The predicted binding mode is in agreement with the binding
pocket of EM-TBPC30 (1-Ethyl-2-methyl-6-oxo-4-(1,2,4,5-tetrahy-
dro-benzo[d]azepin-3-yl)-1,6-dihydro-pyrimidine-5-carbonitrile)
that has been determined by a mutation study.30 All residues iden-
tified to influence ligand binding (V7575.47, W7986.48, F8016.51
,
Y8056.55, T8157.39) are in proximity to 1 in our model. N747A45.51

5710
T. Noeske et al. / Bioorg. Med. Chem. 17 (2009) 5708–5715
Figure 1. Self-organizing map showing the distribution of mGluR1 reference molecules (N = 357) (a) and distribution of the candidate screening compounds (N = 194,563) (b)
in topological pharmacophore space (CATS-2D descriptor). Colouring reflects relative compound density.
was found to increase the activity of EM-TBPC,30 which is in agree-
ment with the lack of a corresponding hydrogen-bonding partner
on that ligand.
Since a,b unsaturated ketones are generally known to be good
Michael acceptors, limited metabolic stability might occur due to
an interaction with for example, cysteine groups. Therefore, the vi-
nyl group in 8–12 was converted into a cyclopropyl ketone group,
which has similar steric requirements and polarity but lacks the
potentially active site for Michael addition reactions (Scheme 1).
In this way, cyclopropylketones 13–16 were synthesized according
2.3. Activity optimization via the synthesis of structural
analogues
to a Corey–Chaykovsky reaction where the a,b-unsaturated ketone
The putative allosteric binding site of rmGluR1 (Fig. 2a and b)
reveals several hydrophobic amino acids, for example, V757,
V753, V819, L640 and V664. Therefore, the replacement of the phe-
nyl, thiophenyl or furyl substituents of 1–7 by an alkyl or cyclic al-
kyl group might lead to an improved binding affinity for rmGluR1.
Consequently, a small library of alkylpropenone derivatives was
synthesized as shown in Scheme 1. The synthesis of alkylprope-
none derivatives 8–12 is a one-step reaction based on a condensa-
tion reaction of an aldehyde with an alkylketone.31 Scheme 1
shows the reaction of quinoline-3-carbaldehyde (17) with 1-ada-
mantan-1-yl-ethanone (18) in the presence of sodium hydroxide
to give 1-adamantan-1-yl-3-quinolin-3-yl-propenone (10) in high
yield. A detailed description of the synthesis of all propenone
derivatives 8–12 of Table 2 is given in Section 4.
was reacted with trimethyl sulfoxonium iodide (Corey–Chaykosky
reagent), which was generated from dimethylsulfoxide and methyl
iodide to give the respective cyclopropyl ketone in good yields.32,33
A description of the synthesis is given in Section 4.
The attachment of a cyclopropylpropenone group gave 8 dis-
playing no affinity for rmGluR1. However, in case a tertbutylprope-
none group (9) was introduced as a substituent, a moderately
active rmGluR1 antagonist was gained (Table 2). Highly active
rmGluR1 antagonists were obtained after the introduction of a
bulky adamantanylpropenone group (10–17). Here, antagonist 10
revealed a high affinity for mGluR1 with a Ki of 0.02
a considerable interaction with rmGluR5 was detected (Ki of
0.807 M) thus resulting in only 30-fold selectivity for rmGluR1
versus rmGluR5. It was shown previously that dual inhibition of
lM. However,
l

T. Noeske et al. / Bioorg. Med. Chem. 17 (2009) 5708–5715
5711
Table 1
Structure and in vitro functional (IC50) and binding (Ki) results for propenone-based screening compounds
Chemical structure
ID
IC50
(
lM) rmGluR1
Ki (
lM) rmGluR1
Ki (l
M) rmGluR5
O
1
1.7 ( 0.15)
8.5 ( 0.62)
0.74 ( 0.29)
2.97 ( 0.40)
20.5* ( 5.28)
>40
9.18 ( 1.08)
>100
>100
>100
>100
>100
>100
>100
Cl
N
O
2
3
4
5
6
7
>40
Cl
N
O
S
S
9.93 ( 1.25)
17.23 ( 2.14)
>40
Cl
N
N
N
N
N
O
O
Cl
Cl
O
O
O
>40
O
Cl
Cl
39.30* ( 16.3)
>40
O
Results are the mean values of at least two independent experiments. Asterisks indicate values were estimated from single concentrations. Standard errors are given in
parentheses.
both mGluR1 and mGluR5 leads to more pronounced side effects in
vivo, for example, cognitive impairment.21 Therefore, the reduction
of mGluR5 activity is crucial to diminish those side effects. The
presence of a 2,3-dihydro-benzo[1,4]dioxine or a 4-methoxy-3-
methyl-phenyl group in 11 and 12, respectively, gave rmGluR1
antagonists with reduced affinities and lower selectivities versus
rmGluR5.
It is noteworthy that the conversion of the propenone group of
10 into a cyclopropylketone group (13–16) yielded an rmGluR1
antagonist with an improved selectivity profile. Compound 13
was achieved by converting the propenone group of 10 into a
cyclopropylketone group resulting in a about sixfold reduction of
rmGluR1 binding affinity but no affinity for rmGluR5 up to
the basis of SAR for series of coumarines15 and tetrahydroquinoli-
nones.20 Within the coumarine series, a carbonyl-linked adamantyl
group was predicted to interact with the hydrophobic cluster at
TM5, similar to the optimized ligands 10–16. For the tetrahydro-
quinolinone series, several ligands were found with piperazine or
tetrahydropiperidine substituents that might well overlap with
the morpholine group of 16. The comparable rmGluR1 affinity of
10 and 16 indicates that the tertiary amino group of the morpho-
line ring is most likely not in a charged state, which is also well
in agreement with the predominantly hydrophobic binding site
in mGluR1, and in particular with the interaction of the morpho-
line substituent with the second hydrophobic cluster at TM2 and
TM7.
10 lM. Replacement of the quinoline group of 13 by a pyridyl
group (14) significantly reduced both the affinity for rmGluR1
and the selectivity versus rmGluR5. In case the pyridyl group car-
ried an o-methoxy (15) the binding affinity was recovered in com-
bination with an improved selectivity. A superior in vitro profile
was achieved after attaching an o-morpholino group to the pyridyl
ring of 14. In contrast to all other rmGluR1 antagonists reported
herein, 16 combined high affinity for rmGluR1 as well as selectivity
versus mGluR5 and an improved water-solubility due to the pres-
ence of the morpholino group thus allowing its use for in vivo
experiments. In addition, a high affinity for the human mGluR1
receptor (hmGluR1) was gained (Table 2).
3. Conclusion
In a prospective virtual screening study, we have compiled a
small subset from a large and diverse compound library yielding
a substantial hit rate. Some molecules showed functional activity
at rmGluR1 at the low micromolar and even nanomolar level.
Economic chemical optimization resulted in potent and selective
lead candidates. The tight interplay of virtual screening and medic-
inal chemistry was a key to success.
Based on pharmacological results, we hypothesize a binding
mode for the propenone hits, that was well in agreement with
mutational data on other mGluR1 ligands.30 Knowledge of a poten-
tial binding side facilitated the design of novel mGluR1 antagonists
The results for the optimized ligands agree well with the prope-
none binding mode postulated for 1 in our homology model, and
with two earlier studies where we predicted binding modes on

5712
T. Noeske et al. / Bioorg. Med. Chem. 17 (2009) 5708–5715
Figure 2. Potential binding mode of 1 in the allosteric binding site of rmGluR1; (a) overview, (b) detailed view.
with a markedly improved activity and selectivity profile. In partic-
ular, lead candidate 16 combines high mGluR1 affinity, functional
activity, selectivity versus mGluR5 as well as an improved water-
solubility, which enables its use for in vivo investigations.
4.2. (E)-1-Cyclopropyl-3-quinolin-3-yl-propenone (8)
Yield 52%; mp 101–102 °C; Rf = 0.27 [hexane/EtOAc (2:1)]; 1H
NMR (CDCl3) d 1.04 and 1.22 (both m, both 2H, CH2CH2); 2.30
(m, 1H, cyclopropyl CH); 7.11 (d, 16.2 Hz, 1H, @CH); 7.59 and
7.77 (both t, 7.4 and 6.6 Hz, both 1H, 60,70-CH); 7.76 (d, 16.0 Hz,
1H, @CH); 7.86 and 8.12 (both d, 8.0 Hz, both 1H, 50,80-CH); 8.29
(d, 2.2 Hz, 1H, 40-CH) and 9.13 (d, 2.2 Hz, 1H, 20-CH). Anal. Calcd
for C15H13NO: C, 80.7; H, 5.9; N, 6.3. Found: C, 80.6; H, 5.8; N, 6.2.
4. Experimental
Melting points were determined in capillary tubes on a Gallenk-
amp melting point apparatus and are uncorrected. Microanalyses
were performed on a Carlo Erba Instrument EA1108. 1H NMR spec-
tra were recorded on Varian Mercury 200BB spectrometer for solu-
tions in CDCl3 with TMS as internal standard. The solvents used
were of commercial grade and were distilled before use. Flash
chromatography was carried out on Kieselgel (35–70 lm). TLC
analyses were performed on Kieselgel 60 F254 plates (Merck),
UV-detection. Starting materials were from commercial sources.
4.3. (E)-4,4-Dimethyl-1-quinolin-3-yl-pent-1-en-3-one (9)
Yield 45%; mp 146–148 °C; Rf = 0.34 [hexane/EtOAc (2:1)]; 1H
NMR (CDCl3) d 1.27 (s, 9H, 3CH3); 7.33 (d, 15.0 Hz, 1H, @CH);
7.59 and 7.76 (both t, 7.2 Hz, both 1H, 60,70-CH); 7.83 (d, 15.2 Hz,
1H, @CH); 7.86 and 8.11 (both d, 7.8 and 8.0 Hz, both 1H, 50,80-
CH); 8.27 (d, 2.2 Hz, 1H, 40-CH) and 9.13 (d, 2.2 Hz, 1H, 20-CH). Anal.
Calcd for C16H17NOꢁ0.25H2O: C, 78.8; H, 7.2; N, 5.7. Found: C, 78.5;
H, 7.0; N, 5.7.
4.1. General procedure for the preparation of propenones
To
a solution of aldehyde (1 mmol) and methylketone
4.4. (E)-1-Adamantan-1-yl-3-quinolin-3-yl-propenone (10)
(1.1 mmol) in ethanol (5 ml) was added 1 N NaOH (1.3 mmol).
The mixture was stirred at room temperature for 1–13 days. The
resulting precipitate was filtered off, washed with cold ethanol
and dried under reduced pressure. If there were no precipitate
the solution was evaporated under reduced pressure to dryness
Yield 69%; mp 169–171 °C; Rf = 0.34 [hexane/EtOAc (2:1)]; 1H
NMR (CDCl3) d 1.78 (s, 6H, adamantane CH2); 1.93 (s, 6H, adaman-
tane CH2), 2.12 (br s, 3H, adamantane CH); 7.37 (d, 16.2 Hz, 1H,
@CH); 7.58 and 7.75 (both t, 6.6 and 7.2 Hz, both 1H, 60,70-CH);
7.83 (d, 16.2 Hz, 1H, @CH); 7.87 and 8.11 (both d, 8.2 Hz, both
1H, 50,80-CH); 8.27 (d, 2.4 Hz, 1H, 40-CH) and 9.13 (d, 2.2 Hz, 1H,
and purified by flash chromatography using
hexanes–EtOAc or CH2Cl2–MeOH as eluent.
a
mixture of
O
O
O
O
a), 18
b), NaH
H
NaOH
N
DMSO
N
N
10
13
17
O
O
S
c), MeI
S+
I
19
20
Scheme 1. Synthesis strategy for alkylpropenone (8–12) and cyclopropyl ketone derivatives (13–16). Reagents and conditions: (a) EtOH, aq NaOH (1 N), rt, 1–13 days, 29–
88%; (b) NaH in DMSO, rt, 12 h, 70–90%; (c) MeI, 85 °C, 24 h, closed ampoule.

T. Noeske et al. / Bioorg. Med. Chem. 17 (2009) 5708–5715
5713
Table 2
Structure and in vitro results for the alkylpropenone and cyclopropyl ketone derivatives
Chemical structure
ID
Ki (l
M) rmGluR1
Ki (l
M) rmGluR5
Selectivity Ki (rmGluR5)/Ki (rmGluR1)
O
8
>100
>100
—
N
N
O
O
9
5.23 ( 0.98)
30.59
5.9
10
0.021 ( 0.005)
0.807 ( 0.05)
29.9
N
O
O
O
11
12
13
14
15
16*
0.641 ( 0.05)
0.294 ( 0.10)
0.137 ( 0.05)
7.04 ( 1.80)
6.23 ( 072)
5.15 ( 0.38)
>100
9.72
O
O
17.51
>730
3.26
O
N
O
22.94 ( 1.50)
18.4 ( 0.63)
27.78 ( 0.38)
N
O
O
0.304 ( 0.02)
0.024 ( 0.001)
60.5
N
O
1158
N
N
O
Bold values highlight the high affinity of 16 for rmGluR1 and its selectivity versus rmGluR5.
*
hmGluR1 FLIPR ꢂ 0.11 + ꢂ0.02
lM (repetitions: n = 4).
20-CH). Anal. Calcd for C22H23NOꢁ0.25H2O: C, 82.1; H, 7.4; N, 4.4.
7.59 ppm (d, 15.4 Hz, 1H, @CH). Anal. Calcd for C21H26O2ꢁ0.66H2O:
Found: C, 82.1; H, 7.3; N, 4.4.
C, 78.3; H, 8.5. Found: C, 78.3; H, 8.5.
4.5. (E)-1-Adamantan-1-yl-3-(2,3-dihydro-benzo[1,4]dioxin-6-
4.7. General procedure for the preparation of cyclopropanones
yl)-propenone (11)
By the method of Corey and Chaykovsky,32 sodium hydride (60%
dispersion in oil, 1.3 mmol) was added to a solution of trimethyl-
sulfoxonium iodide (1.3 mmol) in DMSO (4 ml). After stirring for
15 min, propenone (1 mmol) solution in DMSO (2 ml) was added
drop wise and it was stirred overnight. The reaction mixture was
quenched with water (10 ml), extracted with methylene chloride
(3 ꢀ 10 ml), dried (MgSO4) and evaporated under reduced pressure
to give a crude product, which was purified by flash chromatogra-
phy using a mixture of hexanes–EtOAc as eluent.
Yield 29%; mp 125–127 °C; 1H NMR (CDCl3) d 1.75 (6H, s, ada-
mantane CH2); 1.87 (d, 6H, adamantane CH2); 2.07 (s, 3H, adaman-
tane CH); 4.28 (s, 4H, CH2CH2); 6.85 (d, 8.0 Hz, 1H, 70-CH); 7.00 (d,
1H, @CH); 7.09 (d, 8.0 Hz, 1H, 80-CH); 7.11 (s, 1H, 50-CH) and 7.56
(d, 15.2 Hz, 1H, @CH). Anal. Calcd for C21H24O3: C, 77.7; H, 7.5.
Found: C, 77.3; H, 7.6.
4.6. (E)-1-Adamantan-1-yl-3-(4-methoxy-3-methyl-phenyl)-
propenone (12)
4.8. Adamantan-1-yl-(2-quinolin-3-yl-cyclopropyl)-methanone
hydrochloride (13)
Yield 41%; mp 138–140 °C; Rf = 0.53 [hexane/EtOAc (4:1)]; 1H
NMR (CDCl3) d 1.76 (br s, 6H, adamantane CH2); 1.89 (br s, 6H, ada-
mantane CH2), 2.09 (br s, 3H, adamantane CH); 2.24 (s, 3H, CH3);
3.86 (s, 3H, OCH3); 6.81 (d, 8.0 Hz, 1H, 50-CH); 7.02 (d, 15.4 Hz,
1H, @CH); 7.38 (d, 8.0 Hz, 1H, 60-CH); 7.40 (s, 1H, 20-CH) and
Dry HCl solution in Et2O was added, solvent was evaporated
under reduced pressure, the residue was triturated with Et2O,
filtered and dried; yield 83%; mp 178–181 °C; 1H NMR (CDCl3) d

5714
T. Noeske et al. / Bioorg. Med. Chem. 17 (2009) 5708–5715
1.46–1.56 and 1.80–1.84 (both m, both 1H, cyclopropyl 3-CH2);
1.74 (s, 6H, adamantane CH2); 1.88 (s, 6H, adamantane CH2);
2.08 (s, 3H, adamantane CH); 2.56–2.73 (m, 2H, cyclopropyl 1,2-
CH); 7.77 and 7.92 (both t, 8 Hz, both 1H, 6,7-CH); 7.96 (d, 9 Hz,
1H, 5-CH); 8.32 (s, 1H, 4-CH); 8.56 (d, 9 Hz, 1H, 8-CH) and 8.73
(d, 1.5 Hz, 1H, 2-CH). Anal. Calcd for C23H25NOꢁHClꢁH2O: C, 71.6;
H, 7.3; N, 3.6. Found: C, 71.7; H, 7.0; N, 3.6.
MV Scintillation Cocktail (Packard Bioscience, Groningen, The
Netherlands) was added. After 14–16 h radioactivity was counted
in a MicroBetaÒTrilux (Perkin Elmer Life Sciences GmbH, Rodgau-
Jügesheim, Germany).
4.13. mGluR5 binding assay—[3H]MPEP assay
Cortical membranes were prepared according to Parsons et al.35
Incubations were started by adding [3H]-MPEP (50.2 Ci/mmol,
5 nM, Tocris) to vials with 125–250 lg protein (total volume
4.9. Adamantan-1-yl-(2-pyridin-3-yl-cyclopropyl)-methanone
hydrochloride (14)
0.5 ml) and various concentrations of the agents. The incubations
were continued at room temperature for 60 min (equilibrium
was achieved under the conditions used). Non-specific binding
Dry HCl solution in Et2O was added, solvent was evaporated
under reduced pressure, the residue was triturated with Et2O,
filtered and dried; yield 99%; mp 90–94 °C; 1H NMR (CDCl3) d
1.38–1.48 (m, 1H, cyclopropyl CH2); 1.66–1.80 (m overlapped by
m, 7H, cyclopropyl CH2, adamantane CH2); 1.84–1.85 (m, 6H, ada-
mantane CH2); 2.08 (s, 3H, adamantane CH); 2.56–2.64 (m, 2H, 2
cyclopropyl CH); 7.84 (dd, 8 and 8 Hz, 1H, 5-CH) 8.19 (d, 8 Hz,
1H, 4-CH); 8.47 (s, 1H, 2-CH) and 8.57–8.62 ppm (m, 1H, 1-CH).
Anal. Calcd for C19H23NOꢁHClꢁH2O: C, 68.0; H, 7.8; N, 4.2. Found:
C, 68.7; H, 7.7; N, 4.1.
was defined by the addition of unlabelled MPEP (10 lM). Incuba-
tions were terminated using a Millipore filter system. The samples
were rinsed twice with 4 ml of ice cold assay buffer over glass fibre
filters (Schleicher & Schuell) under a constant vacuum. Following
separation and rinse the filters were placed into scintillation liquid
(5 ml Ultima Gold) and radioactivity retained on the filters was
determined with
a conventional liquid scintillation counter
(Hewlett Packard, Liquid Scintillation Analyser).
4.10. Adamantan-1-yl-[2-(6-methoxy-pyridin-3-yl)-
cyclopropyl]-methanone (15)
4.14. rmGluR1 assay for the determination of accumulation of
[3H]-inositol phosphates
Yield 90%; mp 106–108 °C; 1H NMR (CDCl3) d 1.25 (td, 7.5 and
3.7 Hz, 1H, cyclopropyl-CH2); 1.54–1.63 (m, 1H, cyclopropyl-
CH2); 1.64–1.73 (m, 6H, adamantane CH2); 1.85 (s, 6H, adamantane
CH2); 2.05 (s, 3H, adamantane CH); 2.32 (t, 7 Hz, 2H, cyclopropyl
CH); 3.91 (s, 3H, OCH3); 6.68 (d, 8 Hz, 1H, 5-CH); 7.29 (dd, 8 and
2 Hz, 1H, 4-CH) and 7.99 (d, 3 Hz, 1H, 2-CH). Anal. Calcd for
C20H25NO2: C, 77.1; H, 8.1; N, 4.5. Found: C, 77.3; H, 8.3; N, 4.4.
Cererebellar granule cells obtained from P8 postnatal Sprague
Dawley rats were cultured in basal Eagle medium (BEM) supple-
mented with 10% foetal calf serum and 2 mM glutamine on 96 well
plates—for further details see—Noeske et al. (2007). After 6 DIV the
BEM was replaced completely with inositol free DMEM (MP
Biomedicals, Eschwege, Germany) containing [3H]-myo-inositol
(Perkin Elmer Life Sciences GmbH, Rodgau-Jügesheim, Germany)
at a final concentration of 0.5
further 48 hours. The DMEM in each well was replaced with
100 l Locke’s buffer (plus 20 mM LiCl, pH 7.4) and incubated for
lCi/100 ll/well and incubated for a
4.11. Adamantan-1-yl-[2-(6-morpholin-4-yl-2-pyridin-3-yl)-
cyclopropyl]-methanone (16)
l
15 min at 37 °C. Locke’s buffer was replaced with agonists/antago-
nists/putative mGluR1 ligands in Locke’s buffer and incubated for
45 min. These solutions were then replaced with 100 ll 0.1 M
Yield 91%; mp 139–141 °C; Rf = 0.33 [hexane/EtOAc (2:1)]; 1H
NMR (CDCl3) d 1.20–1.27 (m, 1H, cyclopropyl CH2); 1.55–1.65 (m,
1H, cyclopropyl CH2); 1.64–1.82 (m, 6H, adamantane CH2); 1.86
(s, 6H, adamantane CH2); 2.05 (s, 3H, adamantane CH); 2.24–2.32
(m, 2H, cyclopropyl CH); 3.44–3.49 and 3.79–3.85 (both m, both
4H, morpholine NCH2 and OCH2); 6.59 (d, 9 Hz, 1H, 5-CH); 7.25
(m overlapped by CHCl3, 1H, 4-CH) and 8.04 (d, 2 Hz, 1H, 2-CH).
Anal. Calcd for C23H30N2O2: C, 75.4; H, 8.3; N, 7.6. Found: C, 75.2;
H, 8.5; N, 7.6.
HCl in each well and incubated for a further 10 min on ice in order
to lyse the cells. The 96 well plates could be frozen at ꢂ20 °C at this
stage until further analysis.
Home made resin exchange columns were prepared as follows.
Empty Bio-Spin Chromatography columns (Biorad Laboratories
GmbH, München, Germany) were plugged with filter paper before
filling with 1.1–1.3 ml of resin (AG1-X8 Biorad, 140-14444) sus-
pended in 0.1 M formic acid (24 g resin per 50 ml acid). The formic
acid was allowed to run out before sealing the syringe tips and fill-
4.12. mGluR1 binding assay—[3H]-2 assay
ing with 200–300 ll of 0.1 M formic acid before storage at 4 °C.
Cerebellar membranes were prepared according to Noeske
et al.15 On the day of assay the membranes were thawed and
washed once more by re-suspension in 50 mM Tris–HCl, pH 7.5
and centrifugation at 48,000g for 20 min. The amount of protein
On the day of assay, columns were washed with 1 ml of 0.1 M
formic acid followed by 1 ml of distilled water. Then the contents
of each assay well were added to one column and washed with
1 ml distilled water followed by 1 ml of 5 mM sodium tetrabo-
rate/60 mM sodium formate. Thereafter, the retained radioactive
inositol phosphates were eluted with 2 ꢀ 1 ml of 1 M ammonium
formate/0.1 M formic acid into 24-well visiplates. Scintillation
liquid (1.2 ml UltimaFlow AF, Perkin Elmer) was added to each
well, the plate sealed and vortexed before radioactivity was deter-
mined by conventional liquid scintillation counting (MicroBe-
taÒTrilux, Perkin Elmer Life Sciences GmbH, Rodgau-Jügesheim,
Germany). Unless otherwise stated, all reagents were obtained
from Sigma.
in the final membrane preparation (250–500
lg/ml) was deter-
mined according to the method of Lowry et al.34
Binding assays were performed at room temperature in quadru-
plicate on 96-well format using fixed concentrations of test com-
pound (10
of 1 nM [3H]-2 and membranes (total volume 0.5 ml) and non-spe-
cific binding was estimated using 30 M of the known mGluR1
lM). The assay was incubated for 1 h in the presence
l
antagonist
(3-ethyl-2-methyl-quinolin-6-yl)-(4-hydroxy-cyclo-
hexyl)-methanone.13 Directly after transferring the reaction
volume onto a 96-well multiscreen plate with glass fibre filter
0.22
l
m (Millipore GmbH, Eschborn, Germany) binding was termi-
Acknowledgements
nated by rapid filtration using a multiscreen vacuum manifold
(Millipore GmbH, Eschborn, Germany). Afterwards, filters were
washed three times with ice-cold assay-buffer and Ultima-GoldTM
Tanja Bauer, Sabine Denk and Christina Wollenburg are thanked
for technical assistance in performing functional mGluR1 assays

T. Noeske et al. / Bioorg. Med. Chem. 17 (2009) 5708–5715
5715
14. Ito, S.; Satoh, A.; Nagatomi, Y.; Hirata, Y.; Suzuki, G.; Kimura, T.; Satow, A.;
Maehara, S.; Hikichi, H.; Hata, M.; Kawamoto, H.; Ohta, H. Bioorg. Med. Chem.
2008, 16, 9817.
15. Noeske, T.; Jirgensons, A.; Stachenkovs, I.; Renner, S.; Jaunzeme, I.; Trifanova,
D.; Hechenberger, M.; Bauer, T.; Schneider, G.; Parsons, C. G.; Weil, T.
ChemMedChem 2007, 2, 1763.
and mGluR5 binding assays. T.N. is grateful to Merz Pharmaceuti-
cals GmbH for a PhD fellowship. This research was supported by
the Beilstein-Institut zur Förderung der Chemischen Wissenschaf-
ten, Frankfurt am Main.
16. Renner, S.; Hechenberger, M.; Noeske, T.; Bocker, A.; Jatzke, C.; Schmuker, M.;
Parsons, C. G.; Weil, T.; Schneider, G. Angew. Chem., Int. Ed. 2007, 46, 5336.
17. Renner, S.; Noeske, T.; Parsons, C. G.; Schneider, P.; Weil, T.; Schneider, G.
ChemBioChem 2005, 6, 620.
18. Radestock, S.; Weil, T.; Renner, S. J. Chem. Inf. Model. 2008, 48, 1104.
19. Bender, A.; Glen, R. C. Org. Biomol. Chem. 2004, 2, 3204.
20. Vanejevs, M.; Jatzke, C.; Renner, S.; Müller, S.; Hechenberger, M.; Bauer, T.;
Klochkova, A.; Pyatkin, I.; Kazyulkin, D.; Aksenova, E.; Shulepin, S.; Timonina,
O.; Haasis, A.; Gutcaits, A.; Parsons, C. G.; Kauss, V.; Weil, T. J. Med. Chem. 2008,
51, 634.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmc.2009.05.072.
References and notes
1. Wess, J. Pharmacol. Ther. 1998, 80, 231.
2. Bockaert, J.; Pin, J. P. Embo J. 1999, 18, 1723.
3. Lesage, A. S. Curr. Neuropharmacol. 2004, 2, 363.
21. Gravius, A.; Pietraszek, M.; Schäfer, D.; Schmidt, W. J.; Danysz, W. Behav.
Pharmacol. 2005, 16, 113.
22. Kohonen, T. Biol. Cybern. 1982, 43, 59.
4. Kunishima, N.; Shimada, Y.; Tsuji, Y.; Sato, T.; Yamamoto, M.; Kumasaka, T.;
Nakanishi, S.; Jingami, H.; Morikawa, K. Nature 2000, 407, 971.
5. Pagano, A.; Ruegg, D.; Litschig, S.; Stoehr, N.; Stierlin, C.; Heinrich, M.;
Floersheim, P.; Prezeau, L.; Carroll, F.; Pin, J. P.; Cambria, A.; Vranesic, I.; Flor,
P. J.; Gasparini, F.; Kuhn, R. J. Biol. Chem. 2000, 275, 33750.
6. Kew, J. N. Pharmacol. Ther. 2004, 104, 233.
7. Williams, D. L., Jr.; Lindsley, C. W. Curr. Top. Med. Chem. 2005, 5, 825.
8. Malherbe, P.; Kratochwil, N.; Mühlemann, A.; Zenner, M. T.; Fischer, C.; Stahl,
M.; Gerber, P. R.; Jaeschke, G.; Porter, R. H. J. Neurochem. 2006, 98, 601.
9. Gasparini, F.; Bilbe, G.; Gomez-Mancilla, B.; Spooren, W. Curr. Opin. Drug Discov.
Dev. 2008, 11, 655.
23. Schneider, G.; Wrede, P. Prog. Biophys. Mol. Biol. 1998, 70, 175.
24. Zupan, J.; Gasteiger, J. Neural Networks in Chemistry and Drug Design; Wiley-
VCH: Weinheim, 1999.
25. Noeske, T.; Sasse, B. C.; Stark, H.; Parsons, C. G.; Weil, T.; Schneider, G.
ChemMedChem 2006, 1, 1066.
26. Schneider, G.; Neidhart, W.; Giller, T.; Schmid, G. Angew. Chem., Int. Ed. 1999,
38, 2894.
27. Schneider, G.; Schneider, P. QSAR Comb. Sci. 2003, 22, 713.
28. Schneider, G.; Schneider, P. In Drug Discovery; Kubinyi, H., Müller, G., Eds.;
Wiley-VCH: Weinheim, 2004; p 341.
29. Ballesteros, J. A.; Weinstein, H. Methods Neurosci. 1995, 25.
30. Malherbe, P.; Kratochwil, N.; Knoflach, F.; Zenner, M. T.; Kew, J. N.; Kratzeisen,
C.; Maerki, H. P.; Adam, G.; Mutel, V. J. Biol. Chem. 2003, 278, 8340.
31. Nielsen, A. T.; Houlihan, W. J. Org. React. 1968, 16, 1.
32. Corey, E. J.; Chaykovsky, M. J. Am. Chem. Soc. 1962, 84, 867.
33. Corey, E. J.; Chaykovsky, M. J. Am. Chem. Soc. 1965, 87, 1353.
34. Lowry, O. H.; Rosebrough, N. J.; Farr, A. L.; Randall, R. J. J. Biol. Chem. 1951, 193,
256.
10. Belozertseva, I. V.; Kos, T.; Popik, P.; Danysz, W.; Bespalov, A. Y. Eur.
Neuropsychopharmacol. 2007, 17, 172.
11. Namkoong, J.; Martino, J. J.; Chen, S. Front. Biosci. 2006, 11, 2081.
12. Zheng, G. Z.; Bhatia, P.; Daanen, J.; Kolasa, T.; Patel, M.; Latshaw, S.; Kouhen, O.
F. E.; Chang, R.; Uchic, M. E.; Miller, L.; Nakane, M.; Lehto, S. G.; Honore, M. P.;
Moreland, R. B.; Brioni, J. D.; Stewart, A. O. J. Med. Chem. 2005, 48, 7374.
13. Mabire, D.; Coupa, S.; Adelinet, C.; Poncelet, A.; Simonnet, Y.; Venet, M.;
Wouters, R.; Lesage, A. S.; Beijsterveldt, L. V.; Bischoff, F. J. Med. Chem. 2005, 48,
2134.
35. Parsons, C. G.; Danysz, W.; Bartmann, A.; Spielmanns, P.; Frankiewicz, T.;
Hesselink, M.; Eilbacher, B; Quack, G. Neuropharmacology 1999, 38, 85.