Scaffold hopping approach towards various AFQ-056 analogs as potent metabotropic glutamate receptor 5 negative allosteric modulators

Article abstract of DOI:10.1016/j.bmcl.2013.09.059

The metabotropic glutamate receptor subtype 5 has evolved into a promising target for the treatment of various diseases of the central nervous system, such as Fragile X and l-DOPA induced dyskinesia. One of the most advanced clinical compound is Novartis'

Full text of DOI:10.1016/j.bmcl.2013.09.059

Accepted Manuscript
Scaffold hopping approach towards various AFQ-056 analogs as potent metab-
otropic glutamate receptor 5 negative allosteric modulators
Holger Kubas, Udo Meyer, Mirko Hechenberger, Kai-Uwe Klein, Patrick Plitt,
Ronalds Zemribo, Harm W. Spexgoor, Sander G.A. van Assema, Ulrich Abel
PII:
S0960-894X(13)01133-5
DOI:
http://dx.doi.org/10.1016/j.bmcl.2013.09.059
BMCL 20911
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
19 August 2013
17 September 2013
21 September 2013
Please cite this article as: Kubas, H., Meyer, U., Hechenberger, M., Klein, K-U., Plitt, P., Zemribo, R., Spexgoor,
H.W., van Assema, S.G.A., Abel, U., Scaffold hopping approach towards various AFQ-056 analogs as potent
metabotropic glutamate receptor 5 negative allosteric modulators, Bioorganic & Medicinal Chemistry Letters
(2013), doi: http://dx.doi.org/10.1016/j.bmcl.2013.09.059
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Graphical Abstract
Scaffold hopping approach towards various
AFQ-056 analogs as potent metabotropic
glutamate receptor 5 negative allosteric
modulators
Leave this area blank for abstract info.
Holger Kubasa, Udo Meyera, Mirko Hechenbergera, Kai-Uwe Kleina, Patrick Plitta, Ronalds Zemribob, Harm W.
Spexgoorc, Sander G.A. van Assemac, and Ulrich Abela
a Merz Pharmaceuticals GmbH, Eckenheimer Landstrasse 100, 60318 Frankfurt am Main, Germany
b Institute of Organic Synthesis, Aizkraukles iela 21, Riga, LV1006, Latvia
c Mercachem B.V., Kerkenbos 1013, 6546 BB Nijmegen, The Netherlands
AFQ-056
X = 18 different core structures
(Novartis)

Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com
Scaffold hopping approach towards various AFQ-056 analogs as potent
metabotropic glutamate receptor 5 negative allosteric modulators
Holger Kubasa,, Udo Meyera, Mirko Hechenbergera, Kai-Uwe Kleina, Patrick Plitta, Ronalds Zemribob,
Harm W. Spexgoorc, Sander G.A. van Assemac, and Ulrich Abela
a Merz Pharmaceuticals GmbH, Eckenheimer Landstrasse 100, 60318 Frankfurt am Main, Germany
b Institute of Organic Synthesis, Aizkraukles iela 21, Riga, LV1006, Latvia
c Mercachem B.V., Kerkenbos 1013, 6546 BB Nijmegen, The Netherlands
ARTICLE INFO
ABSTRACT
Article history:
Received
Revised
Accepted
Available online
The metabotropic glutamate receptor subtype 5 has evolved into a promising target for the
treatment of various diseases of the central nervous system, such as Fragile X and L-DOPA
induced dyskinesia. One of the most advanced clinical compound is Novartis’ AFQ-056
(Mavoglurant), which served us as a template for a scaffold hopping approach, generating a
structurally diverse set of potent analogs. Both the limited aqueous solubility and the relatively
poor metabolic stability of AFQ-056 were improved with hexahydrocyclopenta[c]pyrrole
derivative 54a, which proved to be a valuable candidate for further development.
Keywords:
Metabotropic glutamate receptor
mGluR5
2013 Elsevier Ltd. All rights reserved.
Allosteric modulator
Scaffold hopping
AFQ-056
Glutamate is the most prominent excitatory neurotransmitter
in the brain and crucial for memory formation, regulation, and
learning.1 It functions via ionotropic glutamate receptors and via
G-protein coupled metabotropic glutamate receptors (mGluR),
which belong to the GPCR family C. mGluR5 together with
mGluR1 belongs to the group I of mGluRs. It is centrally
expressed in the limbic cortex, hippocampus, amygdala, basal
ganglia, and thalamus, as well as peripherally in the skin and the
vagal nerve system,2 which reflects the multitude of
pharmacological effects anticipated for mGluR5 modulators. As
a classical mGlu receptor, the mGluR5 is structurally composed
of a large extracellular N-terminal domain, which harbors the
orthosteric binding site, and of a seven transmembrane (7TM) -
helical domain, which accommodates the allosteric MPEP
binding site.3 It couples positively to phospholipase C via Gq,
eventually leading to an increase of intracellular calcium.4
Orthosteric inhibitors of mGluR5 were shown to be of limited
use as leads for drug discovery as a consequence of their high
compound polarity, poor bioavailability and a low brain
penetration. In contrast, negative allosteric modulators (NAMs)
are in general more lipophilic, and mGluR5 NAMs were shown
to exhibit properties, which are in accordance with the use as
CNS-acting compounds. Since NAMs exert their modulatory
effect in a non-competitive way, the tempo-spatial signaling
pattern of endogenous glutamate is maintained and modulated
rather than being fully suppressed.
mGluR5 NAMs have been clinically tested in Fragile-X
syndrome (FXS),5 anxiety,2 L-DOPA induced dyskinesia (LID),6
gastro-esophageal reflux disease (GERD),7 migraine,8 and chorea
in Huntington´s disease.9 The most advanced clinical candidate is
Novartis´ Mavoglurant (AFQ-056, 1), which is currently being
tested in in a phase 3 clinical trial for the treatment of FXS,10 and
which has also successfully been evaluated in phase 2 clinical
trials for LID, similarly to Addex´ Dipraglurant (ADX48621, 2)11
and Roche´s RG-7090 (3)11, 12 (Figure 1). However, no marketed
mGluR5 drug has yet evolved, and there is a continuing demand
of additional chemical lead structures.
Figure 1. Clinically evaluated mGluR5 negative allosteric modulators.
We here describe the discovery and optimization of novel
potent acetylenic mGluR5 negative allosteric modulators based
on the chemical structure of Novartis’ drug candidate AFQ-056
(1). In contrast to most of the published mGluR5 NAMs
———
 Corresponding author. Tel.: +49-(0)69-1503-478; fax: +49-(0)69-1503-188; e-mail: holger.kubas@merz.de, hkubas@freenet.de

containing a triple bond motif, the bicyclic scaffold presents a
relatively high fraction of sp3 hybridized carbons (Fsp3), which is
often associated with a superior drug-like profile as compared
with ‘flat’ diarylalkynes.13 Several molecular parameters crucial
for a successful development of an oral drug, such as aqueous
solubility and selectivity over relevant off-targets, are supposed
to be significantly improved by adding three-dimensionality to
the molecular structure. Therefore, the fully saturated bicyclic
core of 1 was chosen as a template for a scaffold-hopping
approach, resulting in the identification of various nitrogen-
containing bicyclic and spiro ring systems. The first hit series
around compound 9 was further varied at three different positions
to examine structure-activity-relationships. For all other
identified scaffolds the substitution pattern of AFQ-056 was
maintained to enable a proper comparison with the original. As a
result, seven out of the 18 novel scaffold series produced
compounds with in vitro activities in the low nanomolar range
(IC50 < 100 nM).
Figure 2. Crystal structure of 3-((3-fluorophenyl)ethynyl)-3-endo-
hydroxy-8-azabicyclo[3.2.1]octan-8-yl)(pyrrolidin-1-yl)methanone (I).
Final compounds 14 and 15, containing a heteroaryl residue,
were also synthetically accessible via intermediate product 4,
which was treated with ethynyltrimethylsilane to yield alkyne 13
(Scheme 1). The TMS group was finally replaced by an
appropriate aryl halide using tetrabutylammonium fluoride
(TBAF), a suitable palladium catalyst, and copper(I)iodide under
basic conditions, leading to the formation of the desired products.
N-Derivatized compounds 19–22 were made via Boc-protected
intermediate 16, which was synthesized from nortropinone
employing di-tert-butyl dicarbonate under basic conditions.
Ketone 16 was then reacted with 1-ethynyl-3-methylbenzene in
the presence of n-BuLi to give alkyne derivative 17, which was
subsequently N-deprotected using hydrochloric acid. In the last
step, the formed cyclic amine (as HCl salt) was converted into
the final products via coupling reactions with an appropriate acyl
chloride, ethyl chloroformate, or dimethylcarbamoyl chloride,
respectively (Scheme 1).
At first, a diverse set of nortropane-related analogs of
compound 1 was prepared as outlined in Schemes 1–3, starting
with nortropinone hydrochloride.14 Final products 5–10, 11a,
11b, and 12 were prepared in two steps. The starting material
was first treated with methyl chloroformate in the presence of
triethylamine and DMAP to form the carbamate intermediate 4
(Scheme 1). In a subsequent step, an appropriate ethynylbenzene
was first deprotonated by means of n-butyllithium (n-BuLi) and
then reacted with ketone 4 to provide the target compounds.
Scheme 1. Reagents and conditions: (a) methyl
chloroformate, Et3N, DMAP, THF, rt, 3 h; (b) 1. subst.
ethyne, n-BuLi, THF, -20 oC → 0 oC, 90 min; 2. 4, THF, -78
oC → 0–5 oC, 2 h; (c) aryl halide, Et3N, DMF, CuI,
PdCl2(PPh3)2, TBAF, THF, 60 oC, 2 h → rt, overnight; (d) di-
tert-butyl dicarbonate, Et3N, DMAP, DCM, 0 oC → rt, 16 h;
(e) 1. 1-ethynyl-3-methylbenzene, n-BuLi, THF, -20 oC → 0
oC, 1 h; 2. 16, THF, -20 oC → 0 oC; 1h; (f) 4N HCl in
dioxane, DCM, rt, 1h; (g) R-C(O)Cl, Et3N, DCM, 0 oC → rt,
overnight.
Scheme 2. Reagents and conditions: (a) 1. diphosgene, DCM,
rt, 30 min; 2. azetidine, Et3N, rt, 1 h; (b) 1. 1-ethynyl-3-
methylbenzene, n-BuLi, THF, -20 oC → 0 oC, 90 min; 2. 23,
26, 28, or 30, resp., THF, -78 oC → 0–5 oC, 2 h; (c) diphenyl
cyanocarbonimidate, Et3N, DCM, rt, 2 h; (d) Et2NH/THF,
iPrOH, 80 oC, 3 h; (e) methylsulfonyl chloride, DIPEA,
DCM, rt, 4 h; (f) 2-chloronicotinonitrile, Et3N, THF, reflux, 3
days.
Basically, the ethinylation of nortropinone derivatives could
produce two different isomers, endo or exo, depending on the
side of attack of the alkyne anion. The crystal structure of a
representative derivative (I, Figure 2) confirmed our presumption
that the endo-OH orientation was the favored stereoconfiguration
of the final products. However, in a few cases both isomers were
formed and isolated (see methoxy derivatives 11a and 11b).

The synthesis routes towards additional, diversely N-
substituted products 24, 27, 29, and 31 are depicted in Scheme 2.
The respective precursors were made from nortropinone
hydrochloride. Intermediate 23 was formed in a one-pot-two-step
reaction using diphosgene and azetidine. Diphenyl
cyanocarbonimidate was employed to prepare 25, which was then
further reacted with diethylamine to give cyanoguanidine
derivative 26. N-Sulfonyl analog 28 was synthesized using
methylsulfonyl chloride, and N-arylation of the starting material
with 2-chloronicotinonitrile yielded precursor 30. Ultimately, the
four intermediate ketones were treated with deprotonated 1-
ethynyl-3-methylbenzene to provide the final products, as
described above.
directly treated with deprotonated 1-ethynyl-3-methylbenzene.
Subsequent steps included N-deprotection and conversion to the
final product by treating the amine with ethyl chloroformate.
Alternatively, the starting material was first deprotected,
followed by conversion to the corresponding methyl carbamate,
which was finally subjected to the ethinylation reaction described
above. Respective N-benzylated building blocks
– either
commercially available or synthesized from suitable precursors –
were converted to the corresponding carbamates either directly or
in a two-step-procedure via the amine intermediate. Final
ethinylation was performed according to the previously
delineated procedure.
Introduction of a hydrogen atom in 3-position turned out to be
rather challenging. Eventually, a 4-step-procedure successfully
provided compound 35 (Scheme 3). The synthesis route started
from ketone 4, which was converted to spiro oxirane derivative
32 using trimethylsulfoxonium iodide and potassium tert-
butoxide as base. The formed epoxide ring was opened up by
means of boron trifluoride etherate, and subsequent treatment
with DBU in methanol gave exocyclic ketone 33, which was then
further reacted to alkyne 34 by applying Bestmann-Ohira
conditions.15 Finally, the m-tolyl residue was appended via a
Sonogashira reaction using 1-iodo-3-methylbenzene in the
presence of a suitable palladium catalyst and copper(I)iodide
under basic conditions to furnish product 35.16 3-Methoxy analog
36 was prepared by O-methylation of compound 9 with methyl
iodide and n-BuLi under cryogenic conditions (Scheme 3).
Scheme 4. General synthesis routes towards final products 39,
42, 45, 47, 49, 52a, 52b, 54, 54a, 54b, 57, 60, 62, 65, 68, 70,
70a, 70b, 74, 76, 79, 82, 85, and 87. For details please refer to
the Supplementary data.
Each of the synthesized compounds was tested in a functional
assay to determine the activity on the human mGlu5 receptor
stably expressed in CHO cells. The assay was based on the
detection of intracellularly mobilized calcium using a calcium
sensitive dye. For all compounds which were found to be active
(%inhibition >50%) at 10 µM test concentration follow-up
measurements were performed in form of concentration response
curves to examine their functional potencies. Most active
compounds were further characterized in rat primary astrocytes,
whereby their mGluR5 NAM activities could be demonstrated
also in native tissue.
In general, the in vitro potencies determined in primary cells
(rat primary astrocytes) were found to be in very good agreement
with those obtained from CHO cells containing the human
mGlu5 receptor. Moreover, the high affinities of several
compounds tested in a binding assay (replacement of [3H]-M-
MPEP) proved our assumption that this type of mGluR5 ligands
binds to the MPEP allosteric site (Table 1). All tested compounds
were found to be highly selective over the mGlu1 receptor (data
not shown).
Scheme 3. Reagents and conditions: (a) 1.
trimethylsulfoxonium iodide, KOtBu, THF, N2, reflux, 3 h; 2.
4, THF, rt, 3 days; (b) 1. 32, BF3×Et2O, THF, -3 oC → 5 oC; 3
h, rt, overnight; 2. DBU, MeOH, THF, rt, 1 h; (c) dimethyl 1-
diazo-2-oxopropylphosphonate, K2CO3, MeOH, 0 oC → rt, 2
h; (d) 1-iodo-3-methylbenzene, Pd(PPh3)4, CuI, Et3N, DMF,
N2, 60 oC, 1 h; (e) n-BuLi (1.6M in hexanes), MeI, THF,
DMSO, -78 oC, 1.5 h.
Detailed structure-activity relationships of the synthesized
tropinone derivatives are summarized in Table 1. Within the first
set of ten compounds the residue R2 was systematically modified,
while the methoxycarbonyl group as N-substituent (R1) and the
hydroxyl group as residue R3 were retained. The prototypic
derivative 5 bearing an unsubstituted phenyl ring already showed
moderate to good activity with an IC50 value at the human mGlu5
receptor of 0.298 µM. Moreover, the in vitro potency could be
substantially elevated by adding a substituent to the 3-position,
demonstrated by compounds 6 (3-F, 5-fold increase), 8 (3-Cl, 10-
Preparation of a set of compounds containing various bicyclic
core structures is generically described in Scheme 4 (details are
given in the Supplementary data). Boc-protected amines were

fold), the direct AFQ-056 analog 9 (3-Me, 6-fold), 10 (3-CN, 2-
fold), 11a (3-OMe, 4-fold), and 12 (3-OCHF2, 4-fold). In
contrast, substitution in 4-position (compound 7, 4-F, IC50 = 2.19
µM) and replacement of the benzene ring by pyridine (compound
14, IC50 = 0.818 µM) led to a marked drop of activity. Further
variations of the pyridine residue like bicyclic heteroaryl ring
systems turned out to be even more detrimental, exemplified by
the inactive imidazopyridine derivative 15. Notably, the isolated
exo-OH isomer of 11a, this is 11b (see Scheme 1), was found to
be active only in the micromolar range (data not shown),
providing evidence that the endo-configuration is clearly favored.
In a second step, the N-substitution (R1) was varied, keeping
the original R2 (m-tolyl) and R3 (OH) residues of parent
compound 1. Interestingly, the replacement of the N-
methoxycarbonyl moiety of compound 9 by a simple n-propyl
group caused a total loss of activity (data not shown). This and
the results obtained for N-acyl derivatives 19 and 20 revealed that
a H-bond acceptor function like a carbonyl group in α-position to
the nitrogen seems to be crucial for submicromolar potencies.
However, compared to compound 9 these amides were found to
be 6- and 10-fold less active, respectively. A series of further
amides showed activities in the same range or lower (data not
shown). Switching from amide to ethyl carbamate (21) restored
the 2-digit nanomolar potency of methyl analog 9. Further
variation of the carbamate function yielded reasonably active
urea derivatives 22 (hIC50 = 0.132 µM) and 24 (hIC50 = 0.210
µM). Appending larger azetidine homologs like, for example, a
pyrrolidine ring, did not improve the potency of 24 (data not
shown). In comparison to the urea derivatives, cyanoguanidine
27 turned out to be equipotent, whereas the activity of the more
polar methylsulfonyl analog 29 was ~3-fold lower. Due to the
similar potencies found for 29 and its acetyl analog 19 the
sulfonyl group presented a suitable bioisoster for C=O. Finally,
compound 31, bearing 3-cyanopyridin-2-yl as an N-aryl residue,
reached in vitro results which were comparable with those
obtained for its bioisosteric cyanoguanidine 27.
A third small series comprised of compounds 34 and 35
demonstrated that the hydroxyl group (R3) is not mandatory for
reaching high potencies. Both a simple hydrogen residue and a
methoxy group were shown to be equivalent. However, aqueous
solubility was significantly diminished by these modifications
(data not shown).
In summary, the in vitro activity results obtained for the
nortropane series around compound 9 clearly evinced that small
changes of the original substitution pattern of AFQ-056 were
generally tolerated, indicated by several equipotent compounds
such as 6, 8, 11a, 12, 21, and 35. However, more distinct
modifications of the functionalities often led to a marked drop of
in vitro potencies. This finding applied to the change of the
position of substitution on the phenyl ring, to the replacement of
the phenyl residue by heteroaryl rings, and to specific variations
of the N-substituent. Moreover, it was figured that the hydroxyl
group in 3-position of the bicyclic ring system was not crucial for
achieving high activities as shown with compounds 35 and 36.
Table 1. mGluR5 in vitro potency data of azabicyclo[3.2.1]octane derivatives
Human mGluR5 (CHO)
Rat mGluR5 (rpA)
IC50 (µM)a
Rat mGluR5 (ctx)
Compound ID
R1
-
R2
-
R3
-
IC50 (µM)a
[3H]-M-MPEP Ki (µM)b
1 (AFQ-056)
0.0716±0.00647
0.0516±0.00595
Nt
0.0660±0.00624
Nt
5
6
OH
OH
0.298±0.0248
0.0613±0.0091
0.0678±0.00489
Nt
0.0777±0.00577
Nt
7
OH
2.19
8
OH
OH
OH
OH
0.0274±0.00296
0.0507±0.00579
0.187±0.0282
0.0252
Nt
9
0.0479±0.00583
Nt
0.0710
Nt
10
11a
0.0763±0.00877
0.0740
0.0760

12
14
OH
OH
0.0722±0.0142
0.818
0.0907
Nt
0.0578±0.00612
Nt
15
OH
>100
Nt
Nt
19
20
21
22
24
OH
OH
OH
OH
OH
0.543
Nt
Nt
Nt
Nt
Nt
Nt
0.308±0.00664
0.0478±0.00247
0.132±0.0172
0.210±0.0219
Nt
0.0291
0.113
Nt
27
29
31
OH
OH
OH
0.168±0.0150
0.575
Nt
Nt
Nt
Nt
Nt
Nt
0.170±0.0248
35
36
H
0.0648±0.0140
0.137±0.0264
Nt
Nt
Nt
Nt
OMe
a IC50 values are given as geometric means ±SEM of at least three independent experiments, each performed in quadruplicate (no
SEM-values indicate a lower number of independent experiments).
b
Replacement of [3H]-M-MPEP; Ki values are given as geometric means ±SEM of at least three independent experiments each
performed in triplicates.
Table 2. mGluR5 in vitro potency data of further AFQ-056 analogs
Human mGluR5 (CHO)
IC50 (µM)a
Rat mGluR5 (rpA)
IC50 (µM)a
Rat mGluR5 (ctx)
Compound ID
X
[3H]-M-MPEP Ki (µM)b

1 (AFQ-056)
0.0716±0.00647
0.0516±0.00595
0.0660±0.00624
9
0.0507±0.00579
1.02
0.0479±0.00583
Nt
0.0710
Nt
39
42
45
5.0 (est.)
49 (est.)
Nt
Nt
Nt
Nt
47
49
0.0466±0.0055
3.35
0.0325
Nt
Nt
Nt
52a (exo-OH)
0.0850±0.00613
52 (est.)
0.0387
Nt
Nt
Nt
52b (endo-OH)
54 (rac)
54a (E1)
54b (E2)
0.103±0.0169
0.0748±0.0191
27 (est.)
0.188±0.0274
0.0775
Nt
0.110±0.0547
Nt
Nt
57
60
8.6 (est.)
Nt
Nt
Nt
0.0365±0.00355
0.0389
62
65
68
0.930
1.01
Nt
Nt
Nt
Nt
Nt
Nt
15 (est)
70
0.0511±0.00468
0.0284±0.00574
0.189±0.0355
0.0366
Nt
Nt
Nt
Nt
70a (E1)
70b (E2)
Nt
74
0.180±0.0148
Nt
Nt

76
79
0.227±0.0429
0.306±0.075
Nt
Nt
Nt
Nt
82
85
87
0.813
Nt
Nt
0.340±0.139
0.0863±0.0312
Nt
Nt
0.0881
0.0825±0.0337
a IC50 values are given as geometric means ±SEM of at least three independent experiments, each performed in quadruplicate (no
SEM-values indicate a lower number of independent experiments).
b Replacement of [3H]-M-MPEP; Ki values are given as geometric means ±SEM of at least three independent experiments each
performed in triplicates.
Further modifications and variations of the nortropane core
produced three different compound subseries grouped as bridged
and fused bicyclic ring systems and spiro compounds (Table 2).
For proper comparison with parent compound 1, the original
substitution pattern was not altered. Starting with nortropane
derivative 9, the bridge length within the bicyclic system as well
as the position of the ring-nitrogen was varied. The switch from a
[3.2.1] (9) to a [3.3.1] ring system in compound 39 had a
detrimental effect on the mGluR5 activity, indicated by a 20-fold
drop of the respective IC50. However, when the positions of the
ring nitrogen and the ethynyl substitution were interchanged,
leading to compound 47, the high potency of 9 was fully
restored. Surprisingly, the shrinkage of the 3-carbon-bridge in
compound 47 to a 2- or 1-carbon-bridge produced compounds
(45 and 42, respectively) with only micromolar potencies. The
same effect was observed when an oxygen was incorporated into
the bridge, realized with compound 49 (hIC50 = 3.35 µM). On the
other hand, compound 52a, containing a [2.2.2] ring system,
showed a hIC50 value below 100 nM, suggesting that certain
flexibility in spatial size and orientation is given for the core
structure. However, its OH-endo isomer 52b was found to be
almost inactive. Stereochemistry and particularly the specific
orientation of the alkyne and the hydroxyl residues again seem to
have a critical impact on receptor binding, like already seen with
the compound pair 11a and 11b.
binding. Analog 60, which contains a [6,4] ring system instead of
a [5,5] bicycle, turned out to be one of the most active
compounds. The racemic mixture showed an excellent hIC50
value of 0.0365 µM (respective enantiomers have not been
separated). In contrast, a [5,6] ring system as substructure of
compounds 62 and 65 were less effective, demonstrated by in
vitro potencies around 1 µM. In this case, the position of the
alkyne substitution had no effect on activity. Ring extension to a
[5,7] bicycle like present in compound 68 was even more
derogatory. Additional piperidine-based analogs as a follow-up
of the highly active compound 60 proved to be more suitable,
achieving activities in the two- and three-digit nanomolar range.
In
particular,
one
enantiomer
of
octahydro-1H-
cyclopenta[b]pyridine derivative 70 (70a, absolute configuration
unknown) convinced as the most active substance within the
entire series (hIC50 = 0.0284 µM). Notably, the difference
between the IC50 values measured for the two enantiomers was
only sixfold as compared with a ratio of 360 obtained for 54a and
54b. With difluorinated analog 74 (hIC50 = 0.180 µM) – designed
to improve the relatively low metabolic stability of 70 (vide
infra) – the high potency could not be retained, unfortunately.
The same range was hit with octahydroquinoline derivative 76.
Ultimately, diverse spiro ring systems were introduced into
the structure template. 4-, 5-, and 6-Membered rings were
utilized as second cycle while the N-substituted pyridine ring was
maintained in each case. The four presented examples 79, 82, 85,
and 87 all showed IC50 values in the submicromolar range, with
azaspiro[5.5]undecane derivative 87 (hIC50 = 0.0863 µM) being
the most potent member of this subseries.
The usage of various fused bicyclic scaffolds, structurally
derived from the core structure of AFQ-056 (1), led to the
discovery of several potent analogs. Basically, both the position
of the alkyne substituent as well as the size and orientation of the
rings exerted an effect on activity on target. Although compounds
54 and 57 share the same hexahydrocyclopenta[c]pyrrole core
structure, the determined activities were found to be significantly
divergent. The eightyfold higher activity of 54 is obviously
caused by the different ring attachment point of the ethinyl
moiety and therefore by the discriminative shape of the two
regioisomers, with the one of 54 being preferred. Consequently,
its enantiomers 54a and 54b were isolated and tested separately.
As a result, 54a turned out to be 360-fold more potent as
compared with its enantiomer 54b, highlighting the great impact
of the spatial orientation of the alkyne residue on receptor
In a next step, the most promising lead candidates were further
characterized with respect to solubility in aqueous media and
stability in human and rat liver microsomes (Table 3). While
most of the tested compounds showed a solubility profile
comparable to the data obtained for template compound 1,
particularly analogs 9 and 12, two compounds with significantly
divergent results were identified. On the one hand, the only
compound devoid of the hydroxyl group (35) was found to be
markedly less soluble, as expected. Due to the poor solubility in
the assay medium compound 35 has therefore not been
characterized
further.
On
the
other
hand,

hexahydrocyclopenta[c]pyrrole derivative 54a displayed a more
than twofold higher solubility in both the in vitro assay medium
and water (pH 7.4) as compared with AFQ-056 (1). Moreover,
this compound also displayed an improved profile with respect to
metabolic stability in human and rat liver microsomes. As a
consequence, the fast and extensive metabolism of compound 1
observed in healthy subjects after single dose administration
could potentially be positively modulated with compound 54a.17
In contrast, several structurally different candidates such as 21,
22, 47, 70, and 87 revealed a very poor metabolic stability, which
led to an instant termination of further development. The
observed instability could be reduced either to specific N-
substituents like an ethyloxycarbonyl group in compound 21 – its
methyl analog 9 proved to be markedly more stable – or to the
specific type of the bicyclic scaffold, for example in spiro
compound 87, which ring system might be metabolically opened
up. In general, metabolic stability was found to be noticeably
higher in human than in rat liver microsomes for almost all tested
compounds.
compared with AFQ-056, leveraging a further development and
characterization in vivo to demonstrate safety and efficacy in the
treatment of L-DOPA induced dyskinesia.
Acknowledgments
The authors warmly thank Andrea Baude, Sabine Falk and
Christiane Reinbach for expert technical assistance in in vitro
activity experiments, Dr. Meik Sladek for providing the in vitro
binding data, Michelle Werner for ascertaining metabolic
stability data, and Jens Neubauer and Kathleen Wolff for
solubility determinations.
Supplementary data
Supplementary data (experimental details for the synthesis and
characterization of intermediate and final products) associated
with this article can be found, in the online version, at http://...
References and notes
Table 3. Solubility and metabolic stability data
1.
neurobiology, neurology and psychiatry 2002, 8, 562.
2. Jaeschke, G.; Wettstein, J. G.; Nordquist, R. E.; Spooren, W.
Expert Opinion on Therapeutic Patents 2008, 18, 123.
3. Pin, J. P.; Kniazeff, J.; Liu, J.; Binet, V.; Goudet, C.; Rondard, P.;
Prezeau, L. The FEBS journal 2005, 272, 2947.
4. Crawford, J. H.; Wainwright, A.; Heavens, R.; Pollock, J.; Martin,
D. J.; Scott, R. H.; Seabrook, G. R. Neuropharmacology 2000, 39, 621.
5. Gross, C.; Berry-Kravis, E. M.; Bassell, G. J.
Petroff, O. A. The Neuroscientist : a review journal bringing
Compound
ID
Solubility Aa
[µM]
Solubility Bb
[µg/mL]
CLint (h/r)c
(µL/min/mg)
1
190
440
Nt
26
19
Nt
24
17
24
Nt
Nt
Nt
Nt
Nt
60
Nt
Nt
Nt
51 / 134
33 / 314
69 / 266
68 / 458
67 / 385
41 / 120
428 / 2718
141 / 889
Nt
6
Neuropsychopharmacology : official publication of the American College of
Neuropsychopharmacology 2012, 37, 178.
8
6.
Berg, D.; Godau, J.; Trenkwalder, C.; Eggert, K.; Csoti, I.; Storch,
9
274
466
170
Nt
A.; Huber, H.; Morelli-Canelo, M.; Stamelou, M.; Ries, V.; Wolz, M.;
Schneider, C.; Di Paolo, T.; Gasparini, F.; Hariry, S.; Vandemeulebroecke,
M.; Abi-Saab, W.; Cooke, K.; Johns, D.; Gomez-Mancilla, B. Movement
disorders : official journal of the Movement Disorder Society 2011, 26, 1243.
11a
12
21
22
35
47
52a
54a
60
70
7.
Zerbib, F.; Bruley des Varannes, S.; Roman, S.; Tutuian, R.;
Galmiche, J. P.; Mion, F.; Tack, J.; Malfertheiner, P.; Keywood, C.
Alimentary pharmacology & therapeutics 2011, 33, 911.
8.
Keywood, C.; Wakefield, M.; Tack, J. Gut 2009, 58, 1192.
Nt
9.
Ribeiro, F.; Pires, R. W.; Ferguson, S. G. Mol Neurobiol 2011, 43,
6
1.
10.
Grégoire, L.; Morin, N.; Ouattara, B.; Gasparini, F.; Bilbe, G.;
Nt
155 / 983
71 / 651
28 / 54
Johns, D.; Vranesic, I.; Sahasranaman, S.; Gomez-Mancilla, B.; Di Paolo, T.
Parkinsonism & Related Disorders 2011, 17, 270.
Nt
11.
Rocher, J.-P.; Bonnet, B.; Bolea, C.; Lutjens, R.; Le Poul, E.; Poli,
S.; Epping-Jordan, M.; Bessis, A.-S.; Ludwig, B.; Mutel, V. Current Topics
in Medicinal Chemistry 2011, 11, 680.
470
Nt
12.
13.
Emmitte, K. A. ACS Chemical Neuroscience 2011, 2, 411.
Lovering, F.; Bikker, J.; Humblet, C. Journal of Medicinal
30 / 139
278 / 1100
758 / 734
Chemistry 2009, 52, 6752.
Nt
14.
Abel, U. WO2012152854; 2012.
15.
1996, 521.
16.
46.
17.
Müller, S.; Liepold, B.; Roth, G. J.; Bestmann, H. J. Synlett 1996,
Sonogashira, K. Journal of Organometallic Chemistry 2002, 653,
Walles, M.; Wolf, T.; Jin, Y.; Ritzau, M.; Leuthold, L. A.;
87
76
a
Kinetic solubility in assay medium (0.5% DMSO/TRIS
buffer).
b Thermodynamic solubility in water (pH = 7.4).
c Intrinsic clearance in human and rat liver microsomes.
Krauser, J.; Gschwind, H.-P.; Carcache, D.; Kittelmann, M.; Ocwieja, M.;
Ufer, M.; Woessner, R.; Chakraborty, A.; Swart, P. Drug Metabolism and
Disposition 2013, 41, 1626.
In conclusion, we could show that based on the chemical
structure of Novartis’ clinical mGluR5 NAM AFQ-056
(Mavoglurant) various modifications of the bicyclic core
structure were well tolerated by the receptor. By means of a
scaffold hopping approach a diverse set of potent analogs
structurally derived from fused or bridged bicyclic scaffolds or
spiro compounds was generated. Most notably, compound 54a
not only showed equipotency on the mGlu5 receptor but also an
improved aqueous solubility and metabolic stability profile as

Table 1
Table 1. mGluR5 in vitro potency data of azabicyclo[3.2.1]octane derivatives
Human mGluR5 (CHO)
Rat mGluR5 (rpA)
IC50 (µM)a
Rat mGluR5 (CHO)
Compound ID
R1
-
R2
-
R3
-
IC50 (µM)a
[3H]-M-MPEP Ki (µM)b
1 (AFQ-056)
0.0716±0.00647
0.0516±0.00595
Nt
0.0660±0.00624
Nt
5
6
OH
OH
0.298±0.0248
0.0613±0.0091
0.0678±0.00489
Nt
0.0777±0.00577
Nt
7
OH
2.19
8
OH
OH
OH
OH
OH
OH
0.0274±0.00296
0.0507±0.00579
0.187±0.0282
0.0763±0.00877
0.0722±0.0142
0.818
0.0252
Nt
9
0.0479±0.00583
0.0710
10
11a
12
14
Nt
Nt
0.0740
0.0907
Nt
0.0760
0.0578±0.00612
Nt
15
OH
>100
Nt
Nt
19
20
21
22
OH
OH
OH
OH
0.543
Nt
Nt
Nt
Nt
Nt
0.308±0.00664
0.0478±0.00247
0.132±0.0172
Nt
0.0291
0.113
24
27
OH
OH
0.210±0.0219
0.168±0.0150
Nt
Nt
Nt
Nt

29
31
OH
OH
0.575
Nt
Nt
Nt
Nt
0.170±0.0248
35
36
H
0.0648±0.0140
0.137±0.0264
Nt
Nt
Nt
Nt
OMe
a
IC50 values are given as geometric means ±SEM of at least three independent experiments, each performed in
quadruplicate (no SEM-values indicate a lower number of independent experiments).
b
Replacement of [3H]-M-MPEP; Ki values are given as geometric means ±SEM of at least three independent
experiments each performed in triplicates.

Table 2
Table 2. mGluR5 in vitro potency data of further AFQ-056 analogs
Human mGluR5 (CHO)
Rat mGluR5 (rpA)
IC50 (µM)a
Rat mGluR5 (ctx)
Compound ID
X
IC50 (µM)a
[3H]-M-MPEP Ki (µM)b
1 (AFQ-056)
0.0716±0.00647
0.0516±0.00595
0.0660±0.00624
9
0.0507±0.00579
1.02
0.0479±0.00583
0.0710
Nt
39
42
Nt
Nt
5.0 (est.)
Nt
45
47
49 (est.)
Nt
Nt
Nt
0.0466±0.0055
0.0325
49
3.35
Nt
Nt
52a (exo-OH)
0.0850±0.00613
52 (est.)
0.0387
Nt
Nt
Nt
52b (endo-OH)
54 (rac)
54a (E1)
54b (E2)
0.103±0.0169
0.0748±0.0191
27 (est.)
0.188±0.0274
0.0775
Nt
0.110±0.0547
Nt
Nt
57
60
8.6 (est.)
Nt
Nt
Nt
0.0365±0.00355
0.0389
62
65
0.930
1.01
Nt
Nt
Nt
Nt

68
15 (est)
Nt
Nt
70
0.0511±0.00468
0.0284±0.00574
0.189±0.0355
0.0366
Nt
Nt
Nt
Nt
70a (E1)
70b (E2)
Nt
74
0.180±0.0148
Nt
Nt
76
79
0.227±0.0429
0.306±0.075
Nt
Nt
Nt
Nt
82
85
0.813
Nt
Nt
0.340±0.139
0.0863±0.0312
Nt
Nt
87
0.0881
0.0825±0.0337
a
IC50 values are given as geometric means ±SEM of at least three independent experiments, each performed in
quadruplicate (no SEM-values indicate a lower number of independent experiments).
b
Replacement of [3H]-M-MPEP; Ki values are given as geometric means ±SEM of at least three independent
experiments each performed in triplicates.

Table 3 (amended)
Table 3. Solubility and metabolic stability data
Compound
Solubility Aa
[µM]
Solubility Bb
[µg/mL]
CLint (h/r)c
ID
(µL/min/mg)
1
190
440
Nt
26
19
Nt
24
17
24
Nt
Nt
Nt
Nt
Nt
60
Nt
Nt
Nt
51 / 134
33 / 314
69 / 266
68 / 458
67 / 385
41 / 120
428 / 2718
141 / 889
Nt
6
8
9
274
466
170
Nt
11a
12
21
22
35
47
52a
54a
60
70
87
Nt
6
Nt
155 / 983
71 / 651
28 / 54
Nt
470
Nt
30 / 139
278 / 1100
758 / 734
Nt
76
a Kinetic solubility in assay medium (0.5% DMSO/TRIS buffer).
b Thermodynamic solubility in water (pH = 7.4).
c Intrinsic clearance in human and rat liver microsomes.

Figure 1

Figure 2

Scheme 1

Scheme 2

Scheme 3

Scheme 4

Captions
Figure 1. Clinically evaluated mGluR5 negative allosteric modulators.
Figure 2. Crystal structure of 3-((3-fluorophenyl)ethynyl)-3-endo-hydroxy-8-azabicyclo[3.2.1]octan-8-yl)(pyrrolidin-1-yl)methanone (I).
Scheme 1. Reagents and conditions: (a) methyl chloroformate, Et3N, DMAP, THF, rt, 3 h; (b) 1. subst. ethyne, n-BuLi, THF,
-20 oC → 0 oC, 90 min; 2. 4, THF, -78 oC → 0–5 oC, 2 h; (c) aryl halide, Et3N, DMF, CuI, PdCl2(PPh3)2, TBAF, THF, 60 oC,
2 h → rt, overnight; (d) di-tert-butyl dicarbonate, Et3N, DMAP, DCM, 0 oC → rt, 16 h; (e) 1. 1-ethynyl-3-methylbenzene, n-
BuLi, THF, -20 oC → 0 oC, 1 h; 2. 16, THF, -20 oC → 0 oC; 1h; (f) 4N HCl in dioxane, DCM, rt, 1h; (g) R-C(O)Cl, Et3N,
DCM, 0 oC → rt, overnight.
Scheme 2. Reagents and conditions: (a) 1. diphosgene, DCM, rt, 30 min; 2. azetidine, Et3N, rt, 1 h; (b) 1. 1-ethynyl-3-
methylbenzene, n-BuLi, THF, -20 oC → 0 oC, 90 min; 2. 23, 26, 28, or 30, resp., THF, -78 oC → 0–5 oC, 2 h; (c) diphenyl
cyanocarbonimidate, Et3N, DCM, rt, 2 h; (d) Et2NH/THF, iPrOH, 80 oC, 3 h; (e) methylsulfonyl chloride, DIPEA, DCM, rt, 4
h; (f) 2-chloronicotinonitrile, Et3N, THF, reflux, 3 days.
Scheme 3. Reagents and conditions: (a) 1. trimethylsulfoxonium iodide, KOtBu, THF, N2, reflux, 3 h; 2. 4, THF, rt, 3 days;
(b) 1. 32, BF3×Et2O, THF, -3 oC → 5 oC; 3 h, rt, overnight; 2. DBU, MeOH, THF, rt, 1 h; (c) dimethyl 1-diazo-2-
oxopropylphosphonate, K2CO3, MeOH, 0 oC → rt, 2 h; (d) 1-iodo-3-methylbenzene, Pd(PPh3)4, CuI, Et3N, DMF, N2, 60 oC,
1 h; (e) n-BuLi (1.6M in hexanes), MeI, THF, DMSO, -78 oC, 1.5 h.
Scheme 4. General synthesis routes towards final products 39, 42, 45, 47, 49, 52a, 52b, 54, 54a, 54b, 57, 60, 62, 65, 68, 70,
70a, 70b, 74, 76, 79, 82, 85, and 87. For details please refer to the Supplementary material.
Table 1. mGluR5 in vitro potency data of azabicyclo[3.2.1]octane derivatives
Table 2. mGluR5 in vitro potency data of further AFQ-056 analogs