Discovery of cyclopentane- and cyclohexane-trans-1,3-diamines as potent melanin-concentrating hormone receptor 1 antagonists

Article abstract of DOI:10.1016/j.bmcl.2007.05.034

We herein report the optimization of cyclopentane- and cyclohexane-1,3-diamine derivatives as novel and potent MCH-R1 antagonists. Structural modifications of the 2-amino-quinoline and thiophene moieties found in the initial lead compound served to improve its metabolic stability profile and MCH-R1 affinity, and revealed unprecedented SAR when compared to other 2-amino-quinoline-containing MCH-R1 antagonists.

Full text of DOI:10.1016/j.bmcl.2007.05.034

Bioorganic & Medicinal Chemistry Letters 17 (2007) 4232–4241
Discovery of cyclopentane- and cyclohexane-trans-1,3-diamines
as potent melanin-concentrating hormone receptor 1 antagonists
Fabrizio Giordanetto,a,* Olle Karlsson,b Jan Lindberg,b Lars-Olof Larsson,c
Anna Linusson,d Emma Evertsson,a David G. A. Morgane and Tord Inghardtb,*
aLead Generation, Computational Chemistry, AstraZeneca R&D Mo¨lndal, SE-431 83 Mo¨lndal, Sweden
bMedicinal Chemistry, AstraZeneca R&D Mo¨lndal, SE-431 83 Mo¨lndal, Sweden
cDMPK&BAC, AstraZeneca R&D Mo¨lndal, SE-431 83 Mo¨lndal, Sweden
d
˚
˚
Organic Chemistry, Department of Chemistry, Umea University, SE-901 78 Umea, Sweden
eCVGI Bioscience, AstraZeneca R&D Alderley Park, Macclesfield, Cheshire SK10 4TG, UK
Received 27 March 2007; revised 6 May 2007; accepted 11 May 2007
Available online 17 May 2007
Abstract—We herein report the optimization of cyclopentane- and cyclohexane-1,3-diamine derivatives as novel and potent MCH-
R1 antagonists. Structural modifications of the 2-amino-quinoline and thiophene moieties found in the initial lead compound served
to improve its metabolic stability profile and MCH-R1 affinity, and revealed unprecedented SAR when compared to other 2-amino-
quinoline-containing MCH-R1 antagonists.
Ó 2007 Elsevier Ltd. All rights reserved.
Melanin-concentrating hormone (MCH) is a cyclic neu-
ropeptide mainly expressed in the lateral hypothalamic
area and the zona incerta region of the brain.1,2 MCH
is known to play a key role in the fine regulation of feed-
ing behavior and energy homeostasis.3,4 Particularly, a
single administration of MCH in rats induces food-in-
take,5 while chronic intracerebroventricular (ICV) infu-
sion substantially increases body-weight.6 Interestingly,
mice overexpressing the MCH gene appear to be hyper-
phagic and prone to obesity, and display insulin-resis-
tance,7 whereas transgenic MCH-R1 knockout mice
are lean, resistant to diet-induced obesity, and show
hyperactivity.8 Consequently, selective MCH-R1 antag-
onism could provide a viable treatment for obesity. As a
result of such pharmacological validation, the search for
MCH-R1 antagonists is a highly active field of medici-
nal chemistry research and several MCH-R1 antagonists
have appeared in the literature, as exemplified by 1–6 in
Figure 1.9 We describe herein the discovery of cyclopen-
tane- and cyclohexane-1,3-diamines as novel and highly
potent MCH-R1 antagonists.
High-throughput screening (HTS) of the AstraZeneca
compound collection in a MCH-R1 binding assay and
the subsequent hit-to-lead efforts resulted in the identifi-
cation of 7 as a very potent MCH-R1 antagonist with a
IC50 value of 15 nM. An optimization campaign was
then started to explore the SAR around 7 and to im-
prove its pharmacodynamic profile. Compound 7 can
be dissected into three main parts: a hydrophobic thio-
phene moiety connected to the 2-amino-quinoline sys-
tem through
a
central linker of basic nature.
Interestingly, several groups have previously reported
2-amino-quinoline-containing MCH-R1 antagonists.10–15
However, despite the structural similarities, these
compounds displayed markedly different SAR when
compared to the current chemical series, as detailed
below.
Docking16 of 7 to the homology model of MCH-R1
based on bovine rhodopsin (PDB: 1u19) suggested that
the central basic group interacts with D123, a residue
involved in the recognition of the MCH peptide.17
Moreover, the 2-amino-quinoline portion could estab-
lish hydrogen bonds with the side chain of Q127, whilst
no specific interactions were observed for the thiophene
Keywords: Melanin-concentrating hormone receptor; MCH; MCH-R1
antagonists; Obesity.
*
Corresponding authors. Tel.: +46 31 7065723; fax: +46 31 7763710
(F.G.); tel.: +46 31 7761954; fax: +46 31 7763724 (T.I.); e-mail
addresses: fabrizio.giordanetto@astrazeneca.com; tord.inghardt@
astrazeneca.com
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.05.034

F. Giordanetto et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4232–4241
4233
O
H
F
N
O
NH
O
N
N
O
H
N
O
O
F
N
O
1
2
F
F
Cl
O
O
O
O
N
N
HN
S
N
N
Cl
N
N
OH
N
H
4
3
O
O
N
O
N
HN
O
N
Cl
N
N
H
H
O
O
5
6
Figure 1. Selected MCH-R1 antagonists.9
ring (Fig. 2). Earlier modeling studies highlighted the
important role of D123 in the binding of 2-amino-quin-
oline derivatives.14,15 Interestingly, although Q127 is
contributing to the binding sites described by Clark
et al.,14 and Ulven et al.,15 no evident electrostatic inter-
actions with the corresponding ligands have been
reported.
to restrict the conformational degrees of freedom of the
initial compound. The flexible nature of the alkyl chain
present in 7 could theoretically generate selectivity and
polypharmacology issues, considering the ability of a
highly flexible ligand to fit into several different binding
sites. It was thus felt necessary to identify and lock the
bioactive conformation of the lead compound via struc-
tural rigidification. The results of such campaign are
presented in Table 1.
The central linker was the subject of a thorough struc-
tural investigation. The purpose was 2-fold: first, to ex-
plore its contribution to MCH antagonism and second,
According to the docked pose, the propane-1,3-diamino
moiety of 7 adopts a semi-extended conformation. The
ethane-1,2-diamino derivative 8 was prepared to test
whether a shorter spacer could mimic the predicted con-
formation of 7 and to probe the length requirements for
the central core. Unfortunately, the smaller alkyl chain
ligand (8, IC50 = 3.59 lM) resulted in a 239-fold loss
of binding affinity over 7. When docked, 8 is still able
to establish a salt bridge with D123. However, although
the 2-amino-quinoline system is pointing toward the side
˚
chain of Q127, the corresponding distance (3.7 A) does
not permit a hydrogen bond interaction and the pocket
surrounding the quinoline ring is only partially filled.
These findings could help explain the observed drop in
potency. The addition of a carbon atom to yield
butane-1,4-diamine (9) resulted in a total loss of activity
at the tested concentration. There appears to be some
specific spatial constraints on the central fragment: these
could probably be dictated by the presence of D123 and
Q127 on the receptor side. Following such hypothesis,
an ideal scaffold would lock the positions of the 2-ami-
no-quinoline group and the secondary amine function
Figure 2. Predicted binding mode of 7 (capped sticks). Polar side
chains involved in the binding are rendered as balls-and-sticks.

4234
F. Giordanetto et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4232–4241
Table 1. In vitro binding data for a series of central linkers
so as to provide optimal interactions with Q127 and
D123, respectively.
O
A number of cyclic-1,3-diamine analogues (10–15) were
superimposed onto the docked conformation of 7 to
identify potential alternative linkers. According to the
results, both cyclohexane and cyclopentane-1,3-diamine
derivatives would provide the best three-dimensional
alignments to 7. More specifically, their trans configura-
tions (i.e., 1S,3S or 1R,3R) would be preferred over the
cis forms (i.e., 1S,3R or 1R,3S). Interestingly, the intro-
duction of (1S,3S)-cyclohexane-1,3-diamine to afford 10
(IC50 = 3 nM) resulted in an improvement in MCH-R1
binding affinity compared to the simple propane spacer
found in 7 (IC50 = 15 nM). It is noteworthy that the
opposite enantiomer (1R,3R) found in 11 showed only
moderate MCH-R1 antagonism (IC50 = 411 nM). This
may be due to the presence of the cyclohexane ring in
a ‘forbidden’ region of the binding pocket. The docking
results suggest that it would face the negatively charged
side chain of D123 thus preventing the formation of an
optimal salt bridge. Incorporation of (1S,3S)-cyclopen-
tane-1,3-diamine (12) proved to be somehow less fa-
vored (IC50 = 62 nM), despite the good alignment to 7
and 10, and no obvious penalizing interactions in the
MCH-R1 binding pocket. In both the cyclohexane and
cyclopentane subseries, the cis isomers showed a marked
decrease in potency (results not shown).
R1
N
Compound
R1
h-MCH-R1
a
IC50 (lM)
*
N
H
N
H
7
0.015
S
H
N
*
*
8
3.59
>33
N
H
S
H
N
9
N
H
S
10
0.003
*
N
H
N
H
S
11
0.411
*
*
N
H
N
H
S
S
12
13
0.062
0.035
N
H
N
H
Conformational restriction of the central linker proved
to be a successful strategy to gain binding affinity.
Assuming a chair conformation as the most populated
form of the central scaffold in 10, there is still the possi-
bility of interconversion between the two different chair
forms (i.e., equatorial-axial and axial-equatorial). Intro-
duction of a bicyclo[2.2.1]heptane ring would provide a
completely rigid central spacer and could potentially
lock the bioactive conformation for these analogues.
To our dismay, none of the bicyclo[2.2.1]heptane deriv-
atives synthesized (13–15) demonstrated the sought
improvement in potency over 10. The best result was
displayed by 13 (IC50 = 35 nM), in which the quinoline
ring is in endo of the bridge carbon atom. The bicy-
clo[2.2.1]heptane structure locks the boat conformation
of cyclohexane. If the chair conformation of cyclohex-
ane is indeed the bioactive conformation, then the struc-
tural differences between a rigid boat and a flexible chair
could be held responsible for the loss of binding affinity.
Alternatively, the presence of the additional bridge atom
could cause an imperfect fit to the receptor due to the in-
creased steric hindrance of the linker. This second
hypothesis seems to be favored by both ligand align-
ments and docking results. However, according to the
computational results, the steric hindrance effect is more
evident for both 15 (IC50 = 369 nM) and 14
(IC50 = 5 lM), whereas it is barely noticeable in the case
of 7.
*
*
*
N
H
N
H
S
S
S
14
15
5.01
N
H
N
H
0.369
N
H
N
H
O
16
17
18
0.106
0.119
2.99
*
*
*
N
H
N
H
S
S
S
N
H
N
N
N
H
*
*
19
20
2.30
2.13
N
N
N
H
S
N
H
The 2-amino-quinoline and secondary amine functional-
ities were so far assumed to be essential for potent
MCH-R1 antagonism, based on the predicted molecular
interactions with the receptor. These hypotheses were
challenged in order to gain a better understanding of
S
a Values are means of at least two experiments. Compounds competed
with 125I-MCH for binding at the human MCH1 receptor (h-MCH-
R1) expressed in the HEK293 cell line.

F. Giordanetto et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4232–4241
4235
the SAR around the MCH-R1 antagonists studied here.
Specifically, a pharmacophore model18 was derived by
superimposing the best and most diverse antagonists in
the series (7, 10, and 13), as shown in Figure 3. Three
features were defined: the positively charged nitrogen
(N+) centered on the secondary amine group and two
hydrophobic centroids for the thiophene (Hy1) and
quinoline (Hy2) rings (Fig. 3). This served as a frame-
work to evaluate molecules in their ability to (1) main-
tain a good shape similarity to the best compounds
and (2) provide alternative structural motifs for the
interaction with D123 and Q127. Transformation of
the secondary amine functionality of 7 into amide (16)
Although 10 proved to be a very potent MCH-R1
antagonist (IC50 = 3 nM), its ability to antagonize
MCH-R1 signaling in a functional assay (MCH-R1
GTPcS IC50 = 46 nM) and its metabolic stability (Hu-
man Liver Microsomes intrinsic clearance, HLM Clint
=
128 lL minÀ1 mgÀ1) needed to be improved. In this
context, both ends of the molecule were metabolized
(O-demethylation on the quinoline ring and hydroxylation
of the thiophene moiety), whereas the central cyclohex-
ane-diamine scaffold appeared to be stable.
The quinoline system was subjected to a thorough
analysis in order to identify the best ring assembly and
the corresponding substitution pattern. According to the
structure-based modeling results, there appeared to be
some room for small modifications in either position 5,
6 or 7 of the quinoline ring. Moreover, position 4 would
probably only tolerate very minor replacements,
whereas positions 3 and 8 did not seem to allow for
any beneficial substitutions. A number of ring alterna-
tives were designed in order to challenge the docking re-
sults and to systematically explore the chemical space in
that region. These findings are outlined in Table 2. The
synthesis of a representative compound, 33, is outlined
in Scheme 1.20 As a first comparison, the unsubstituted
quinoline ring of 21 displayed reduced binding and func-
tional activity (IC50 = 68 nM, GTPcS IC50 = 234 nM)
when compared to 10. The 4-methyl-quinoline deriva-
tive 22 still maintained potency but showed no improve-
ment in the GTPcS assay (IC50 = 7 nM, GTPcS
IC50 = 59 nM). On the other hand, simplification of 10
to yield the 6-methoxy-quinoline analogue 23 reduced
binding but improved functional antagonism
(IC50 = 57 nM, GTPcS IC50 = 25 nM), although meta-
afforded
a
weaker antagonist (IC50 = 106 nM).
Although the amide would still be able to establish a
hydrogen bond with D123 through its N–H group, such
polar interaction is not preferred and appears to con-
tribute less to the overall binding affinity. This would
strengthen the hypothesis that securing a salt bridge
with D123 is a key requisite for potency in the current
chemical series. However, it cannot be completely ruled
out that the observed potency loss originates from other
subtle effects such as altered electrostatic properties of
the thiophene ring and rigidification of the linker. A
number of tertiary amines were also prepared to investi-
gate their effect on binding. Surprisingly, the simple
N-methylation of 12 (IC50 = 62 nM) to provide 17
(IC50 = 119 nM) translated in a loss of affinity, while
inclusion of the piperidine-3-methylamine scaffold (18,
IC50 = 2.99 lM) exhibited a more significant drop in
potency.
Modification of the hydrogen bond donor properties of
the 2-amino-quinoline moiety was also carried out to
verify the supposed interaction with Q127. To this end,
compounds lacking a hydrogen bond donor group were
prepared, here exemplified by 19 (IC50 = 2.30 lM) and
20 (IC50 = 2.13 lM), which all displayed poor activity
at the receptor. This is in stark contrast with previous re-
ports disclosing 2-amino-quinoline derivatives as MCH-
R1 antagonists.10–15 These studies clearly demonstrated
that tertiary amino-quinolines afforded MCH-R1 anta-
gonists in the low nanomolar range. The differences in
binding hypotheses and the role of Q127 may help to
rationalize the divergent SAR observed.
bolic stability was not enhanced (HLM Clint
=
295 lL minÀ1 mgÀ1). Additional substitutions in posi-
tion 6 (24–26) always translated in a loss of potency,
suggesting that the methoxy group is probably optimal.
Intriguingly, the potent 2-amino-quinoline-containing
MCH-R1 antagonists published by different laborato-
ries displayed an amide function at the 6-position, anal-
ogously to 26, which carried a variety of bulky
substituents.11–15 The modification of position 4 showed
some very interesting results. Replacement of methyl by
trifluoromethyl (27) or cyano (28) drastically reduced
binding (IC50 = 4.8 and 9.9 lM, respectively). Introduc-
tion of dimethylamino maintained activity (29,
IC50 = 29 nM, GTPcS IC50 = 97 nM), whilst the 4-
methoxy-quinoline compound 30 afforded the sought
improvement in the functional assay (IC50 = 9 nM,
GTPcS IC50 = 8 nM). The observed effects of cyano
and dimethylamino substitutions on binding were cor-
rectly predicted in silico. However, docking failed to
anticipate the loss of potency observed for the trifluo-
romethyl analogue 27. A possible explanation may be
found in the electron-donating and withdrawing
effects that such substitutions exert on the quinoline
nitrogen and which are not fully captured by the
docking model.
Figure 3. Three-dimensional alignment of 7 (pale green), 10 (light
cyan), and 13 (yellow). The pharmacophore features employed in the
search for new spacers are also depicted: positive charge (N+) and
hydrophobic centroids (Hy1 and Hy2).
Interestingly, the 4-methoxy-6-fluoro-quinoline deriva-
tive (31) also resulted in a potent, functional antagonist
(IC50 = 26 nM, GTPcS IC50 = 9 nM), whereas the

4236
F. Giordanetto et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4232–4241
Table 2. In vitro binding and functional data, and HLM intrinsic clearance results for substituted quinolines
R5
R8
R4
N
R3
R6
R7
N
H
N
H
S
Compound
R3
R4
R5
R6
R7
R8
h-MCH-R1
IC50 (lM)
h-MCH-R1 GTPcS
IC50 (lM)
HLM Clint
(lL minÀ1 mgÀ1
)
a
b
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Me
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Cl
0.068
0.007
0.057
0.571
0.847
2.52
0.234
0.059
0.025
NDc
NDc
NDc
NDc
NDc
0.097
0.008
0.09
NDc
0.001
0.042
NDc
NDc
NDc
60
Me
H
H
H
H
MeO
Cl
Me2N
295
NDc
NDc
NDc
68
NDc
30
NDc
NDc
NDc
32
H
H
H
H
H
H
AcN(Me)
H
CF3
CN
Me2N
MeO
MeO
MeO
Me
H
H
4.8
9.92
H
H
H
H
0.029
0.009
0.026
0.231
0.0001
0.054
1.39
H
H
H
F
H
i-PrO
H
H
MeO
H
MeO
H
H
38
H
MeO
H
H
NDc
15
H
H
H
5.1
a Values are means of at least two experiments. Compounds competed with 125I-MCH for binding at the human MCH1 receptor (h-MCH-R1)
expressed in the HEK293 cell line.
b Functional assay measuring [35S]GTPcS accumulation.
c Not determined.
Novozyme 435
+
+
H2N
NH2
H2N
NH2
AcNH
NHAc
AcNH
NH2
40 °C, 4 eq EtOAc
Toluene, 5d
cis/trans
2.7:1
S,S
R,R
R,S
>90% ee
1. Cbz-Cl
2. Flash chromatography, crystallisation from MeOH
5% overall, 100%ee
CbzNH
NHCbz
1. H2, Pd/C
Pd(OAc)2
BINAP, Cs2CO3
THF
70 ˚C, 24h
N
Cl
N
N
O
O
NHCbz
O
N
N
H
N
2.
O
H
Cbz
S
S
33
PolBH3CN
68%
DCM, MeOH, HOAc
100 °C 10 min
66%
Scheme 1.
6-isopropyloxy analogue 32 showed a 10-fold drop in
binding affinity due to the increased steric bulk and
the consequent imperfect fit. A number of 4-methyl-
quinolines were also designed to further investigate the
SAR. Notably, the 7-methoxy substituent provided 33,
a very promising MCH-R1 antagonist with excellent
(IC50 = 0.1 nM, GTPcS IC50 = 1 nM). Interestingly,
walking the methoxy substituent from the 6- to the 7-po-
sition of the quinoline ring significantly improved meta-
bolic stability (HLM Clint = 32 lL minÀ1 mgÀ1) of the
MeO-group. The two major metabolites from 33 were
hydroxylation at the quinoline ring and a N-dealkyla-
tion (cleavage) of the methyl thiophene moiety.
binding
and
functional
antagonism
abilities

F. Giordanetto et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4232–4241
4237
Table 3. In vitro binding functional affinity and HLM intrinsic
clearance data for 37–47
Modifications at positions 3, 5, and 8 to yield 34, 35, and
36, respectively, did not offer any improvement over 33,
as shown in Table 2. The 3-methyl group on 35 forced
the quinoline and cyclohexane rings to adopt a confor-
mation that cannot be accommodated in the binding
pocket, whilst the 8-chloro substituent of 36 was pre-
dicted to spatially impair the hydrogen bond between
the quinoline ring and Q127. This is another striking dif-
ference in SAR between the current series and other
structurally related MCH-R1 antagonists. Specifically,
Vasudevan et al. reported 2-amino-8-alkoxy-quinolines
as potent MCH-R1 antagonists,10 suggesting that the
two series may adopt different binding modes.
R1
O
N
N
H
N
H
Compound R1
h-MCHr1 HLM Clint
IC50 (lM) (lL minÀ1
a
mgÀ1
)
N
37
0.022
0.178
0.089
0.013
0.015
500
*
Overall, docking was successful in predicting the conse-
quences of structural changes at the different positions
of the quinoline ring. This was especially true for mod-
ifications that considerably altered the size and confor-
mation of the ligands. In contrast, subtler variations of
their electronic properties, as discussed for the substitu-
ents at the 4-position, remained difficult to capture.
N
38
NDb
NDb
59
*
N
N
39
N
*
The rigidification of the central linker coupled with the
optimization of the quinoline ring identified 33, as a
MCH-R1 antagonist with a very promising pharmaco-
dynamic profile. Theoretical concerns around the thio-
phene moiety, namely the potential to form reactive
metabolites,21 prompted us to explore safer chemical
alternatives. Two libraries were assembled to investigate
the effects of thiophene modifications on biological affin-
ity and metabolism. The first library was built around
the (1S,3S)-cyclohexane-1,3-diamine scaffold and
designed with the aim to explore the chemical space
compatible with the receptor constraints and with
drug-like physico-chemical properties.22 Table 3 sum-
marizes the binding data obtained for the library.
Cl
40
*
Cl
F
F
41
64
*
O
O
F
Br
42
43
0.028
0.010
NDb
NDb
*
*
S
N
N
Unfortunately, none of the synthesized derivatives dis-
played an improvement in binding affinity over 10.
The best antagonist was the benzo[1,2,5]thiadiazol-4-yl
substituted compound 43 with an IC50 value of 10 nM.
Interestingly, variation of the substitution pattern on
the same ring system to yield 46 resulted in an antago-
nist of comparable potency (IC50 = 16 nM). Introduc-
tion of other fused heterocycles comprising
isoquinoline (37: IC50 = 22 nM) and substituted indoles
(44, 45: IC50’s = 61 and 58 nM, respectively) did not im-
prove potency. Moreover, installing single ring, heteroa-
ryl moieties, exemplified by pyrazine 39 (IC50 = 89 nM)
and 1,3-dimethyl-pyrazole 38 (IC50 = 178 nM), proved
detrimental for potency. These findings suggest that
polarized ring systems are not very well tolerated. This
could be due to the fact that the binding pocket has
mainly a lipophilic character, and burial of polar
moieties would be energetically unfavored. Among the
phenyl-substituted analogues, 3,4-dichloro-phenyl and
3-trifluoromethoxy-phenyl afforded the best compounds
(40, IC50 = 13 nM and 41, 15 nM), whereas the 2-bro-
mo-5-methoxy-phenyl derivative 42 (IC50 = 28 nM)
and the 2-chloro-4-methanesulfonyl-phenyl analogue
47 (IC50 = 50 nM) displayed a potency loss when com-
pared to 10. Regrettably, none of the compounds in
the cyclohexane library displayed an improved meta-
N
44
45
0.061
0.058
41
52
*
*
Br
Cl
N
S
N
O
N
46
47
0.016
0.050
227
269
*
*
N
Cl
O
S
O
a Values are means of at least two experiments. Compounds competed
with 125I-MCH for binding at the human MCH1 receptor (h-MCH-
R1) expressed in the HEK293 cell line.
b Not determined.

4238
F. Giordanetto et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4232–4241
Table 4. In vitro binding and functional affinities, and HLM intrinsic clearance data for 48–55
R1
N
H
O
N
N
H
Compound
R1
h-MCH-R1
a
IC50
(lM)
h-MCH-R1
GTPcS
IC50 (lM)
HLM
Clint
(lL minÀ1 mgÀ1
)
b
48
0.009
0.085
0.035
<12
*
N
F
F
O
49
0.005
12
*
*
N
O
50
0.080
NDc
26
N
N
51
52
53
0.029
0.071
0.033
0.083
NDc
20
*
*
N
N
12
N
N
0.118
<12
*
*
N
N
54
0.020
0.127
<12
N
N
55
0039
NDc
<12
*
N
a Values are means of two experiments. Compounds competed with 125I-MCH for binding at the human MCH1 receptor (h-MCH-R1) expressed in
the HEK293 cell line.
b Functional assay measuring [35S]GTPcS accumulation.
c Not determined.

F. Giordanetto et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4232–4241
4239
bolic profile over 10. Here, compound 44 was the most
stable, with a HLM Clint of 41 lL minÀ1 mgÀ1
mouse,24 to assess their effects on food intake and weight
reduction. Animals were dosed perorally for 5 days with
a daily dose of 20 and 50 lmol/kg, respectively. Disap-
pointingly, we could not observe any statistically signif-
icant body weight reduction in these experiments. We
then examined the pharmacokinetic (PK) characteristics
of the compounds in order to investigate and rationalize
their lack of in vivo efficacy. The PK profile of com-
pound 51 in lean mice is shown in Table 5. Despite
showing excellent oral bioavailability, compound 51
gave only a modest brain exposure (Brain/Plasma Cmax
ratio = 0.13, brain Cmax = 500 nmol/kg). Additionally,
plasma protein binding for compound 51 is high
(fu = 6.5% in mouse plasma) and this indicates a high
degree of protein and/or tissue binding in the brain.25
We therefore speculate that the lack of efficacy in obese
mice can be attributed to an insufficient occupancy of
the MCH receptor.26
.
The second library used the (1S,3S)-cyclopentane-1,3-
diamine, which is synthetically more easily accessible23
when compared to the cyclohexane analogue. Addition-
ally, we observed that cyclopentane derivatives had in
general improved metabolic stability over cyclohexane
compounds (cf 10 and 12, HLM Clint = 128 and
78 lL minÀ1 mgÀ1, respectively). For this library, we
adopted a more focused design around the indole ring
of 48, which showed good potency and low intrinsic
clearance in human liver microsomes, as outlined in
Table 4. The synthesis of compound 51 is shown in
Scheme 2.20
Overall the library was successful in identifying metabol-
ically stable compounds. However, coupling stability
with potency in the low, single-digit nanomolar range
proved more difficult. Substitutions on position 5 of
the indole ring greatly affected the binding ability of
the compounds. 5-Difluoromethoxy afforded the best
antagonist in the library, 49, with an IC50 value of
5 nM and an intrinsic HLM clearance value of 12. Inter-
estingly, the 5-benzyloxy-group on 50 compromised
potency (IC50 = 80 nM), corroborating the finding that
there are steric constraints in the receptor pocket. More
importantly, the introduction of a cyano group on the
indole (53–55) always resulted in very stable antagonists
with HLM clearance values of less than 12, thereby
highlighting the importance of reducing the electron
density on the ring as a means of diminishing oxidative
metabolism.
In summary, optimization of the initial lead compound
7 led to the discovery of several highly potent MCH-R1
antagonists. Structural rigidification of the central
Table 5. Mouse in vivo pharmacokinetic profile of 51
Parametera
Value
CL (mL/min kg)
*
Vss (L/kg)
15
7.6
Fpo (%)
t1/2 (h) plasma
100
8.6
Cmax brain (lmol/kg)
Tmax brain (h)
0.5
0.26
3.78
24
Cmax plasma (lmol/L)
Tmax plasma (h)
Cmax brain/Cmax plasma
0.13
Following the positive pharmacodynamic and metabolic
stability results (Table 4), compounds 49 and 51 were
evaluated in a diet-induced obesity (DIO) model in
a Determined following oral and iv administration of a 20 lmol/kg
nanosuspension dose in C57BL female mice.
EtO
Me2N
NMe2
N
Cl
N
CH3B(OH)2
N
H2, Pd/C
37%
EtO
N
N
O2N
O2N
(Ph)4PPd
57%
H
O2N
OHC
NaBH4
N
1. MeI, DMF
N
N
2. HMTA, TFA
49%
N
O
N
H
N
H
N
51
NH2
O
N
N
73%
H
1. Pd(OAc)2
BINAP, NaOtBu
Tolene, 100 °C
2. TFA, DCM
+
NHBoc
H2N
O
N
Cl
49%
Scheme 2.

4240
F. Giordanetto et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4232–4241
Blackburn, C., et al. 230th American Chemical Society
National Meeting, Washington, DC, USA, 2005: MEDI-
086; (f) Souers, A. J.; Gao, J.; Wodka, D.; Judd, A. S.;
Mulhern, M. M.; Napier, J. J.; Brune, M. E.; Bush, E. N.;
Brodjian, S. J.; Dayton, B. D.; Shapiro, R.; Hernandez, L.
E.; Marsh, K. C.; Sham, H. L.; Collins, C. A.; Kym, P. R.
Bioorg. Med. Chem. Lett. 2005, 15, 2752.
diamine spacer identified (1S,3S)-cyclohexane-1,3-di-
amine and (1S,3S)-cyclopentane-1,3-diamine as the best
scaffolds. Subsequent optimization of the substitution
pattern on the quinoline ring afforded 33
(IC50 = 0.1 nM, GTPcS IC50 = 1 nM), the most active
MCH-R1 antagonist in the present study. Systematic
structural exploration aimed at replacing thiophene
identified 51 as a novel, potent, and metabolically stable
MCH-R1 antagonist. Unfortunately, lack of efficacy in
the DIO model, at therapeutically meaningful doses,
precluded advancement of any member of this lead
series into further development.
10. Vasudevan, A.; Wodka, D.; Verzal, M. K.; Souers,
A. J.; Gao, J.; Brodjian, S.; Fry, D.; Dayton, B.;
Marsh, K. C.; Hernandez, L. E.; Ogiela, C. A.;
Collins, C. A.; Kym, P. R. Bioorg. Med. Chem. Lett.
2004, 14, 4879.
11. Arienzo, R.; Clark, D. E.; Cramp, S.; Daly, S.; Dyke, H.
J.; Lockey, P.; Norman, D.; Roach, A. G.; Stuttle, K.;
Tomlinson, M.; Wong, M.; Wren, S. P. Bioorg. Med.
Chem. Lett. 2004, 14, 4099.
12. Ulven, T.; Little, P. B.; Receveur, J.; Frimurer, T. M.;
Rist, Ø.; Nørregaard, P. K.; Ho¨gberg, T. Bioorg. Med.
Chem. Lett. 2006, 16, 1070.
13. Jiang, J.; Hoang, M.; Young, J. R.; Chaung, D.; Eid, R.;
Turner, C.; Lin, P.; Tong, X.; Wang, J.; Tan, C., et al.
Bioorg. Med. Chem. Lett. 2006, 16, 5270.
Acknowledgments
The authors thank their AstraZeneca coworkers Mia
´
Boethius-Litzen, Susanne Ekehed, Mikael Rehngren,
and Arja Schedwin for providing MCH-R1 binding
and functional data, human CLint data, and
in vivo PK data, respectively, as well as Pernilla
˚
Hakansson for running the mouse diet-induced obes-
ity model.
14. Clark, D. E.; Higgs, C.; Wren, S. P.; Dyke, H. J.; Wong,
M.; Norman, D.; Lockey, P. M.; Roach, A. G. J. Med.
Chem. 2004, 47, 3962.
15. Ulven, T.; Frimurer, T. M.; Receveur, J.; Little, P. B.;
Rist, Ø.; Nørregaard, P. K.; Ho¨gberg, T. J. Med. Chem.
2005, 48, 5684.
References and notes
16. Giordanetto, F.; Lindberg, J.; Karlsson, O.; Inghardt, T.
Abstract of Papers, 232nd National Meeting of the
American Chemical Society, San Francisco, CA, 2006,
COMP-410.
17. Macdonald, D.; Murgolo, N.; Zhang, R.; Durkin, J. P.;
Yao, X.; Strader, C. D.; Graziano, M. P. Mol. Pharmacol.
2000, 58, 217.
1. Bittencourt, J. C.; Presse, F.; Arias, C.; Peto, C.; Vaughan,
J.; Nahon, J. L.; Vale, W.; Sawchenko, P. E. J. Comp.
Neurol. 1992, 319, 218.
2. Skofitsch, G.; Jacobowitz, D. M.; Zamir, N. Brain Res.
Bull. 1985, 15, 635.
3. Saito, Y.; Nothacker, H.-P.; Civelli, O. Trends Endocrinol.
Metab. 2000, 11, 299.
4. Schwartz, M. W.; Woods, S. C.; Porte, D.; Selley, R. J.;
Baskin, D. G. Nature 2000, 404, 661.
5. Qu, D.; Ludwig, D. S.; Grammeltoft, S.; Piper, M.;
Pelleymounter, M. A.; Cullen, M. J.; Mathes, F. W.;
Przypek, J.; Kanerek, R.; Maratos-Flier, E. Nature 1996,
380, 243.
18. Three-dimensional alignments were carried out with
ROCS (OpenEye Scientific Software, 3600 Cerrillos Rd.,
Suite 1107, Santa Fe, NM 87507) using the Shape
Tanimoto function to quantify similarity. An in-house
pharmacophore search method19 was employed to identify
central linkers that satisfied the three pharmacophoric
features described (N+, Hy1, and Hy2) within a root
˚
mean square deviation of 0.75 A.
6. Gomori, A.; Ishiara, A.; Ito, M.; Mashiko, S.; Matsuhita,
H.; Yumoto, M.; Ito, M.; Tanaka, T.; Tokita, S.; Moriya,
M.; Iwaasa, H.; Kanatani, A. Am. J. Physiol. Endocrinol.
Metab. 2003, 284, E583.
7. Ludwig, D. S.; Tritos, N. A.; Mastaitis, J. W.; Kulkarni,
R.; Kokkotou, E.; Elmiqst, J.; Lowell, B.; Flier, J. S.;
Maratos-Flier, E. J. Clin. Invest. 2001, 107, 379.
8. Shimada, M.; Tritos, N. A.; Lowell, B. A.; Flier, J. S.;
Maratos-Flier, E. Nature 1998, 396, 670.
9. (a) Takekawa, S.; Asami, A.; Ishihara, Y.; Terauchi, J.;
Kato, K.; Shimomura, Y.; Mori, M.; Murakoshi, H.;
Suzuki, N.; Nishimura, O.; Fujino, M. Eur. J. Pharmacol.
2002, 438, 129; (b) Borowsky, B.; Durkin, M. M.;
Ogozalek, K.; Marzabadi, M. R.; DeLeon, J.; Lagu, B.;
Heurich, R.; Lichtblau, H.; Shaposhnik, Z.; Daniewska,
I., et al. Nat. Med. 2002, 8, 825; (c) McBriar, M. D.;
Guzik, H.; Xu, R.; Paruchova, J.; Li, S.; Palani, A.;
Clader, J. W.; Greenlee, W. J.; Hawes, B. E.; Kowalski, T.
J.; O’Neill, K., et al. J. Med. Chem. 2005, 48, 2274; (d)
Handlon, A. L.; Al-Barazanji, K. A.; Barvian, K. K.;
Bighan, E. C.; Carlton, D. L.; Carpenter, A. J.; Cooper, J.
P.; Daniels, A. J.; Garrison, D. T.; Goetz, A. S., et al.
228th American Chemical Society National Meeting,
Philadelphia, PA, USA , 2004: MEDI-193; (e) Iyengar,
R.; Souers, A. J.; Lynch, J. K.; Judd, A. S.; Gao, J.;
Freeman, J. C.; Mulhern, M.; Vasudevan, A.; Wodka, D.;
19. Lyne, P. D.; Kenny, P. W.; Cosgrove, D. A.; Deng, C.;
Zabludoff, S.; Wendoloski, J. J.; Ashwell, S. J. Med.
Chem. 2004, 47, 1962.
20. The preparation of representative compounds is described
in: Ray, A. K.; Sigfridsson, E. M.; Linusson, A.; Stina,
M.; Sandberg, P. M.; Inghardt, T.; Svensson, A. M.;
Brickmann, K. WO2004/004726 (compound 9), and in:
Evertsson, E.; Inghardt, T.; Lindberg, J.; Linusson, A.;
Giordanetto, F. WO2005/066132 (compounds 13, 15, 23,
29, 31, 38, 41, 43, 49–55).
21. Dansette, P. M.; Thang, D. C.; El Amri, H.; Mansuy, D.
Biochem. Biophys. Res. Commun. 1992, 186, 1624.
22. A variety of structural descriptors, including logD and
aqueous solubility predictions, served to profile and filter
the chemical library. These were calculated using in-house
developed models implemented in Pipeline Pilot version.
4.5, SciTegic: San Diego, CA.
23. The preparation of tert-butyl[(1S,3S)-3-aminocyclopen-
tyl]carbamate is detailed in Evertsson, E.; Inghardt, T.;
Lindberg, J.; Linusson, A.; Giordanetto, F. WO2005/
066132.
24. Female C57Bl/6J mice were given ad libitum access to
calorie-dense ‘cafeteria’ diet (soft chocolate/cocoa-type
pastry, chocolate, fatty cheese, and nougat) and standard
lab chow for 8–10 weeks until a body weight of 45–50 g
was achieved. Compounds to be tested were then admin-

F. Giordanetto et al. / Bioorg. Med. Chem. Lett. 17 (2007) 4232–4241
4241
istered perorally once daily for a period of 5–12 days, and
the body weights of the mice monitored on a daily basis.
During this period ad libitum access to calorie-dense
‘cafeteria’ diet and standard lab chow was maintained.
25. Wan, H.; Rehngren, M.; Giordanetto, F.; Bergstro¨m, F.;
Tunek, A., J. Med. Chem., submitted for publication.
26. Maurer, T. S.; DeBartolo, D.; Tess, D. A.; Scott, D. O.
Drug Metab. Dispos. 2005, 33, 175.