Discovery of a new class of ghrelin receptor antagonists

Article abstract of DOI:10.1016/j.bmcl.2012.01.014

A series of benzodiazepine antagonists of the human ghrelin receptor GHSR1a were synthesized and their antagonism and metabolic stability were evaluated. The potency of these analogs was determined using a functional aequorin (Euroscreen) luminescent assay measuring the intracellular Ca²⁺ concentration, and their metabolic stability was measured using an in vitro rat and human S9 hepatocyte assay. These efforts led to the discovery of a potent ghrelin antagonist with good rat pharmacokinetic properties.

Full text of DOI:10.1016/j.bmcl.2012.01.014

Bioorganic & Medicinal Chemistry Letters 22 (2012) 2046–2051
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of a new class of ghrelin receptor antagonists
⇑
Jeffrey T. Mihalic , Yong-Jae Kim, Mike Lizarzaburu, Xiaoqi Chen, Jeff Deignan, Malgorzata Wanska,
Ming Yu, Jiasheng Fu, Xi Chen, Alex Zhang, Richard Connors, Lingming Liang, Michelle Lindstrom, Ji Ma,
Liang Tang, Kang Dai, Leping Li
Amgen Inc., 1120 Veterans Boulevard, ASF2-3, South San Francisco, CA 94080, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of benzodiazepine antagonists of the human ghrelin receptor GHSR1a were synthesized and their
antagonism and metabolic stability were evaluated. The potency of these analogs was determined using a
functional aequorin (Euroscreen) luminescent assay measuring the intracellular Ca2+ concentration, and
their metabolic stability was measured using an in vitro rat and human S9 hepatocyte assay. These efforts
led to the discovery of a potent ghrelin antagonist with good rat pharmacokinetic properties.
Ó 2012 Elsevier Ltd. All rights reserved.
Received 21 November 2011
Revised 6 January 2012
Accepted 9 January 2012
Available online 14 January 2012
Keywords:
Ghrelin
GHSR1a
Obesity
Obesity has reached an epidemic scale in the US; researchers
estimate that one third of adults in the United States are over-
weight or obese.1 This disease is responsible for approximately
300,000 deaths a year and is associated with other morbidities
such as heart disease, diabetes, asthma, sleep apnea, and some
types of cancer.2 Although the pathogenesis for obesity is often
caused by several factors, recent studies identified several promis-
ing targets such as the ghrelin receptor to combat this disease.3
Ghrelin is the endogenous ligand for this G-protein coupled recep-
tor. The ligand is a 28-amino acid acylated peptide hormone pro-
duced mainly in the stomach by X/A epithelial cells.4 It was first
identified as the endogenous ligand for the growth hormone secre-
tagogue receptor (GHSR), and later it was recognized to have a role
in weight regulation because of its effect on food intake. This effect
has been shown to be independent of growth hormone stimula-
tion. Ghrelin’s food intake stimulation occurs via its receptor sub-
type GHSR1a in the hypothalamic arcuate nucleus where it
activates NPY and AgRP neurons.5 The hormone’s effect on food in-
take is well established in the literature. For example, infusion of
the hormone has been shown to increase feeding in both mice
and humans.6 Peptide antagonists for the ghrelin receptor have
been shown to decrease food intake in lean mice, mice with in-
duced obesity, and ob/ob mice.7 Moreover, an ICV injection of
anti-ghrelin antibody has been shown to reduce body weight in
rats.8 These studies suggest that a small molecule antagonist for
the ghrelin receptor could be a useful therapeutic agent for treating
obesity.
In 1993 researchers demonstrated that benzoazepine analog 1
(Fig. 1) can block ghrelin induced growth hormone secretion.9 This
was the first report in the literature of a non-peptidic ghrelin
antagonist. Later, a different research group disclosed two classes
of GHSR1a antagonists based on the tetralin and isoxazole carbox-
amide scaffolds (compounds 2 and 3, Fig. 1).10 Compounds 2 and 3
were potent ghrelin receptor antagonists with IC50 values of 16 nM
and 8 nM, respectively, as measured by a blockade of intracellular
Ca2+ mobilization induced by ghrelin. During a high throughput
screen of our compound library, we discovered benzodiazepine 4
(Fig. 1) as a novel antagonist of the GHSR1a receptor. In our aequo-
rin functional assay, benzodiazepine 4 had an IC50 of 8 nM for the
human ghrelin receptor, and 7 nM for the mouse ghrelin receptor.
O
O
NH
Me
O
i-BuO
OMe
N
NEt2
O
N
O
N
H
1
3
2
N
O
N
Cl
O
H
N
O
O
N
Ph
N
NH
O
O
Cl
HN
O
4
NEt2
⇑
Corresponding author.
E-mail address: jmihalic@amgen.com (J.T. Mihalic).
Figure 1. Known ghrelin receptor antagonists.
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2012.01.014

J. T. Mihalic et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2046–2051
2047
O
O
O
R2
O
F
b
OH
a
OH
NO2
O
O
t-Bu
OH
NO2
R1
N
H
R1
R1
O
NO2
8
6
5
7
R2 =
H
c
R2
O
O
R2
OH
d-f
O
O
t-Bu
O
N
O
N
R1
R1
H
O
NH2
N
9
H
O
10a
Scheme 1. Reagents and conditions: (a) Cs2CO3, DMF, 100 °C; (b) EDC or HBTU, amine, Et3N, DCM; (c) Pd/C, H2, EtOH; (d) chloroacetyl chloride, Et3N, DCM; (e) NaH, DMF; (f)
TFA.
Ph
N
O
O
N
O
R1
N
O
a
H
H
N
10b
O
R2
N
R3
OH
O
O
O
d,e
a
O
N
R1
O
O
N
O
N
O
N
O
N
R1
O
O
R1
H
N
10a
N
O
H
O
H
12
13a-h
a
H
R2
N
H
N
O
H
N
N
N
Bn
b
c
O
NH
O
N
O
N
O
O
N
R1
O
O
O
O
N
O
N
H
N
O
H
4, 11a-m
O
H
14
15
Scheme 2. Reagents and conditions: (a) Amine, EDC, Et3N, DCM; (b) Pd/C, H2, EtOH; (c) benzoic acid, EDC, Et3N, DCM; (d) TFA, H2O; (e) amine, NaBH(OAc)3, DCM.
OTf
COOEt
Ph
O
a-c
d
O
O
N
Ph
O
O
O
O
O
19
O
N
16
17
O
18
O
O
e
O
f, g
O
O
HO
H2N
O
20
21
Scheme 3. Reagents and conditions: (a) Ethylene glycol, pTSA, toluene; (b) LAH, THF; (c) TF2O, Et3N, DCM; (d) compound 18, LDA, then compound 17, THF; (e) H2O2, LiOH,
THF, H2O; (f) DPPA, Et3N, BnOH, toluene; (g) H2, Pd/C, MeOH.
This compound was a full antagonist for both ghrelin receptors and
had a Ki of 4.4 nM in a human I125-ghrelin competitive binding as-
say. Unfortunately, 4 suffered from poor metabolic stability; after
incubation with S9-hepatocellular fraction for one hour, only 19%
(rat) and 30% (human) of 4 remained unchanged. Metabolite iden-
tification indicated that debenzylation and aromatic ring oxidation
were the major routes of metabolism. This poor stability was
reflected in the rat PK with a clearance of 8.6 L/h/kg and an oral
bioavailability of 6% (Table 5). Herein we report our studies of
the SAR around structure 4 and describe modifications which
improve overall potency and reduce metabolic liabilities.
The analogs were synthesized as outlined in Schemes 1–5. The
benzodiazapine intermediate 10a was synthesized according
to Scheme 1. Via nucleophillic aromatic substitution, various

2048
J. T. Mihalic et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2046–2051
substituted phenols 5 were condensed with 5-fluoro-2-nitroben-
zoic acid 6 to give the resulting biphenyl ether 7. This was then
condensed with the corresponding amino acid t-butyl ester using
EDC or HBTU coupling conditions to give amide 8. Hydrogenation
over palladium on carbon reduced the nitro group of 8 to aniline
9, which was then acylated and cyclized with chloroacetyl chloride
under basic conditions followed by deprotection of the t-butyl
group with TFA to give benzodiazepine 10a.
Analogs 4 (Tables 1–3), 10b (Table 1), 11a–m (Tables 1 and 2),
13a–h (Tables 1–3), 14 (Table 3), and 15 (Table 1) were synthe-
sized according to Scheme 2. From intermediate 10a, EDC medi-
ated amide formation afforded analogs 4, 11a–m, 10b, and
intermediate 12. Hydrogenation was used to remove the benzylic
group on 4 yielding 14. Intermediate 14 was then coupled with
benzoic acid using EDC to give amide 15. Finally, intermediate 12
was converted to the ketone under acidic conditions and was sub-
sequently reductively aminated with various amines in the pres-
ence of sodium triacetoxyborohydride to give analogs 13a–h.
Scheme 3 outlines the synthesis of ketal intermediate 21. The
synthesis begins with the protection of ketone 16 with ethylene
glycol and p-toluenesulfonic acid followed by the reduction of
the ester group with LAH and conversion of the resulting alcohol
to triflate 17. Using Evans’ asymmetric alkylation chemistry, oxa-
zolidinone 18 was treated with LDA and alkylated with triflate
17 to give intermediate 19.11 The oxazolidinone group was then re-
moved under basic conditions, and the resulting acid 20 was con-
verted to the amine 21 under Curtius rearrangement conditions.12
Analogs 26a,b (Tables 1 and 4) were prepared according to
Scheme 4. First, amine 21 was coupled with acid 7 in the presence
of EDC. Next, using hydrogenation conditions, the nitro group on
O
O
O
a
O
O
O
7
OH
NO2
O
O
R1
N
H
R1
H2N
NO2
22
21
O
b
O
c,d
O
O
O
O
O
N
O
N
H
R1
R1
N
NH2
23
O
H
24
e
H
N
R4
O
O
O
f
O
N
O
N
R1
R1
N
H
N
H
O
O
26a, b
25
Scheme 4. Reagents and conditions: (a) EDC, Et3N, DCM; (b) Pd/C, H2, EtOH; (c) chloroacetyl chloride, Et3N, DCM; (d) NaH, DMF; (e) TFA, H2O; (f) cyclopropylmethyl amine or
benzylamine, -selectride, THF.
L
F
F
a-c
O
CO2Me
I
O
CO2H
NO2
F
F
28
27
d
F
O
CO2Me
R5
e
O
F
O
29 R5
=
-CH2CHO
H2N
21
O
30 R5 = -CH2CH2CHO
O
F
CO2Me
O
R6
NH
F
31 R6= -CH2CH2-
32 R6 = -CH2CH2CH2-
f-h
f-h
H
N
H
N
F
O
O
F
O
N
O
N
F
F
34
33
Scheme 5. Reagents and conditions: (a) TMS–diazomethane, MeOH, ether; (b) Pd/C, H2, EtOH; (c) HNO2, KI, EtOH; (d) (1) ethoxyethyne, BH3, THF; (2) NaOH, Pd(PPh3)4, THF;
or (1) allyl alcohol, Pd(OAc)2 Et3N, THF; (2) TFA; (e) NaBH(OAc)3, DCM; (f) 50–60 °C, DMF; (g) TFA, H2O; (h) cyclopropylmethyl amine, -selectride, THF.
L

J. T. Mihalic et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2046–2051
2049
Table 1
SAR of piperidine ring
O
R7
O
N
R1
N
H
O
Compound
R1
H
R7
GHSRIa (Aeq.)a IC50
(
lM)
Rat S9 % remaining
19
Human S9 % remaining
H
N
4
0.008
30
—
O
O
N
N
Ph
Ph
H
N
11k
11l
H
H
0.314
0.012
—
—
H
N
—
N
N
Ph
Ph
O
O
H
N
11m
13a
H
H
0.196
0.008
—
—
H
N
Bn
72
62
N
O
O
H
N
15
H
>10
—
—
N
Ph
Ph
O
H
N
10b
H
H
>10
95
—
23
—
O
H
N
26a
0.006
Bn
a
Assay values are the means of three experiments. Standard deviation is 30%.
Table 2
SAR of phenoxy ring
H
N
N
R1
O
O
R8
R8=
N
O
N
H
Ph
N
Ph
O
HN
A
B
Compound
R1
R8
GHSRIa (Aeq.)a IC50
(lM)
Rat S9 % remaining
Human S9 % remaining
4
A
A
A
A
A
A
A
A
A
A
A
B
H
4-F
4-Me
2-Me
3-Cl
0.008
0.011
0.007
0.003
0.171
0.028
0.026
0.143
0.079
0.0002
0.001
0.0003
19
—
—
3
—
—
19
10
—
3
30
—
—
5
—
—
26
22
—
5
11a
11b
11c
11d
11e
11f
11g
11h
11i
11j
13h
4-Cl
2-OMe
3-OMe
4-OMe
4-F,2-Me
2,6-Me
2,4-F
—
65
—
70
a
Assay values are the means of three experiments. Standard deviation is 30%.
22 was reduced to aniline 23 which was then acylated with 2-chlo-
roacetyl chloride and cyclized with sodium hydride yielding
benzodiazepine 24. The ketal group was then removed under
acidic conditions, and the resulting ketone 25 was reductively
aminated using cyclopropylmethyl amine or benzylamine and
Analogs 33 and 34 (Table 4) were synthesized as outlined in
Scheme 5. First, the aryl nitrobenzoic acid 27 was esterified with
TMS–diazomethane and then hydrogenated over palladium on
carbon to give the aniline which was further transformed to arylio-
dide ester 28 under Sandmeyer conditions.13 Using palladium-
mediated conditions, ester 28 was converted to compounds 29
and 30. Reductive amination with amine 21 converted these
L-selectride to yield analogs 26a,b in a 12:1 ratio of cis to trans
isomers.

2050
J. T. Mihalic et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2046–2051
Table 3
SAR of benzyl sidechain
R1
O
R9
O
N
O
O
HN
Compound
R1
H
R9
GHSRIa (Aeq.)a IC50
0.008
(
lM)
Rat S9 % remaining
19
Human S9 % remaining
30
H
N
4
N
Ph
H
N
14
H
H
H
H
>10
—
—
NH
N
N
N
N
13a
13b
13c
0.0080
0.0025
0.0165
72
46
48
62
27
19
N
H
Ph
Ph
Ph
N
H
N
H
13d
H
0.0025
28
30
N
H
N
N
N
13e
13f
H
H
H
0.0041
0.313
0.050
—
76
—
N
H
O
—
N
H
13g
71
86
N
H
a
Assay values are the means of three experiments. Standard deviation is 30%.
Table 4
SAR of benzodiazapine ring
R10
NH
Compound
R10
GHSR1a (Aeq.)a IC50
(
lM)
Rat S9 % remaining
27
Human S9 % remaining
68
O
F
O
N
26b
0.020
F
N
O
H
O
F
F
O
O
N
33
0.002
0.007
97
69
92
90
F
F
O
N
34
a
Assay values are the means of three experiments. Standard deviation is 30%.
aldehydes to intermediates 31 and 32. The ketals 31 and 32 were
cyclized upon heating to the corresponding quinolinone and azepi-
none, and then deprotected with acid. The resulting ketones were
treated with cyclopropylmethyl amine and L-selectride to give the
final products 33 and 34.
of rat and human hepatocytes.15 Values are expressed as a percent
of parent remaining after 1 h incubation.
The lead optimization utilized a strategy where the initial struc-
ture 4 was divided into four regions: the valine piperidine linker
(Table 1), the aryl phenoxy region (Table 2), the benzyl substituent
(Table 3), and finally the benzodiazepine ring (Table 4). These re-
gions were initially optimized independently of each other, and
the best structural features were then combined in an effort to im-
prove potency and metabolic stability. Particular attention was fo-
cused on variations that decreased molecular weight, removed or
The analogs were tested in a functional aequorin (Euroscreen)
luminescent assay measuring the intracellular Ca2+ concentration
to evaluate overall potency. IC50 values are presented in
lM con-
centrations.14 The metabolic stability for selected compounds
was determined in an in vitro metabolic assay using S9 fractions

J. T. Mihalic et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2046–2051
2051
Table 5
against the ghrelin receptor as demonstrated by the excellent
activity of alkyl analogs 13d (2.5 nM) and 13e (4.1 nM). Smaller
lipophilic substitutions were also tolerated such as the cyclopropyl
analog 13g which had an IC50 of 50 nM. Increasing polarity was
detrimental to the activity as demonstrated by the THP analog
13f which showed a 76-fold loss in potency compared to the cyclo-
hexyl analog 13e. Substitution in this area of the molecule had a
significant impact on metabolic stability. The methylcyclopropyl
analog 13g had the best stability with rat and human S9 of 71%
and 86%, respectively.
S9 stability and pharmacokinetic parametersa,b
Compds
CL (L/h/kg)
%F
AUC PO (
l
g h/L)
S9 % remainingc
Rat
Human
4
33
8.6
0.6
6
39
132
618
19
97
30
92
a
b
c
n = 3 Sprague-Dawley rats per study.
Dosed at 2 mg/kg po.
Percent of parent remaining after 1 h incubation with rat and human S9.
In Table 4, the best attributes from the left and right hand sides
of the molecule were combined with various modifications to the
central core structure. The difluorophenoxy and cyclopropylmethyl
amine appendages were chosen for their marked improvement in
overall metabolic stability while maintaining a respectable level
of potency and lower molecular weight. When these sidechains
were combined with the benzodiazepine core, this resulted in
the structure 26b which had an IC50 of 20 nM. This activity could
be further improved by replacing the diazepine-2,5-dione in 26b
with the corresponding benzoazepin-1-one lacking the lower
amide functionality as in structure 33. When the seven member
ring was replaced with a six member ring (34), this resulted in only
a minimal loss of potency compared to 33. Additionally, metabolic
stability was markedly improved with these structures. Analog 33
had the best metabolic stability with values in rat and human S9
fractions being 97% and 92%, respectively
Analog 33 was further evaluated in rat PK where it was shown
to have a clearance of 0.6 L/h/kg and a bioavailability of 39%. These
values were a marked improvement over the initial lead structure
4 (Table 5).
In conclusion, a new class of ghrelin antagonists was identified.
Through a systematic approach, key SAR interactions were discov-
ered and metabolic liabilities were removed throughout the struc-
ture giving rise to analog 33. This structure demonstrated excellent
potency towards the ghrelin receptor and showed improved phar-
macokinetic properties.
ameliorated metabolically liable groups, and reduced the number
of hydrogen donors and acceptors.
The SAR surrounding the valine and piperidine groups is out-
lined in Table 1. Removal of the valine isopropyl sidechain 11k
or the enantiomer 11m caused a significant drop in potency,
whereas changing the isopropyl group to the cyclopropyl group
11l showed potency similar to the lead structure 4. The regio-iso-
meric piperidine analog, namely the 4-(benzylamino)piperidin-1-
yl amide 13a, showed potency similar to the lead structure (1-ben-
zylpiperidin-4-yl amide) 4. Similarly, removal of the amide linkage
altogether also resulted in an equipotent analog, 26a. On the other
hand, removal of the basic amine within the piperidine ring re-
sulted in analog 15 and the cyclohexyl analog 10b which showed
complete loss of activity. This demonstrated that the basic amine
was essential for potency against the ghrelin receptor. Metabolic
stability data was not obtained for all of the structures within this
table. However, it should be noted that the regio-isomeric piperi-
dine amide 13a showed a marked improvement in both rat and hu-
man S9 fractions with values of 72% and 62%, respectively,
compared to the lead structure 4.
Table 2 summarizes the SAR of the phenoxy region of the
molecule. In general, small electron withdrawing groups main-
tained or improved the potency, whereas larger electron donating
groups as in the analogs 11f–h generally reduced the potency.
Substitution in the meta position was less tolerated as shown
by the analogs 11d and 11g. Disubstituted analogs
4-fluoro-2-methylphenoxy 11i and 2,6-dimethylphenoxy 11j
showed an improvement in potency. ortho Substitutions on the
phenoxy ring are believed to restrict the rotation of the ring
which allows the compounds (11i and 11j) to bind in a favorable
conformation resulting in enhanced potency in the functional as-
say. Metabolic stability overall remained similar to the lead com-
pound 4 with S9 values ranging from 3% to 26%. The 2-methyl
substituted analogs 11c and 11i demonstrated a decrease in over-
all metabolic stability, with S9 values ranging from 3% to 5%.
Metabolic identification showed that the 2-methyl group in these
structures underwent significant oxidation in the presence of
both rat and human S9. In a similar series using the regio-iso-
meric piperidine linker, structures containing a 2,4-difluorophen-
oxy substitution such as 13h were able to ameliorate this
metabolic liability while maintaining similar potency.
References and notes
1. Flegal, K. M.; Carroll, M. D.; Ogden, C. L.; Curtin, L. R. JAMA 2010, 303(3), 235.
2. Gale, S.; Castracane, D.; Mantzoros, C. J. Nutr. 2004, 134(2), 295.
3. Gura, T. Science 2003, 299, 846.
4. Kojima, M.; Hosoda, H.; Date, Y.; Nakazato, M.; Matsu, H.; Kangawa, K. Nature
1999, 402, 656.
5. Hagemann, D.; Meier, J. J.; Gallwitz, B.; Schmidt, W. E. Z. Gastroenterol. 2003, 41,
929.
6. (a) Tschop, M.; Smiley, D.; Heiman, M. Nature 2000, 407, 908; (b) Wren, A. M.;
Seal, L. J.; Cohen, M. A.; Brynes, A. E.; Frost, G. S.; Murphy, K. G.; Dhillo, W. S.;
Ghatei, M. A.; Bloom, S. R. J. Clin. Endocrinol. Metab. 2001, 86, 5992.
7. Asakawa, A.; Inui, A.; Kaga, T.; Katsuura, G.; Fujimiya, M.; Fujino, M. A.; Kasuga,
M. Gut 2003, 52(7), 947.
8. Murakami, N.; Hayashida, T.; Kuroiwa, T.; Nakahara, K.; Ida, T.; Mondal, M. S.;
Nakazato, M.; Kojima, M.; Kangawa, K. J. Endocrinol. 2002, 174, 283.
9. Cheng, K.; Chan, W. W.; Butler, B.; Wei, L.; Schoen, W. R.; Wyvratt, M. J.; Fisher,
M. H., Jr; Smith, R. G. Endocrinology 1993, 132, 2729.
10. (a) Zhao, H.; Xin, Z.; Liu, G.; Schaefer, V. G.; Falls, H. D.; Kaszubska, W.; Collins,
C. A.; Sham, H. L. J. Bioorg. Med. Chem. Lett. 2004, 47, 6655; (b) Liu, B.; Liu, G.;
Xin, Z.; Serby, M. D.; Zhao, H.; Schaefer, V. G.; Falls, H. D.; Kaszubska, W.;
Collins, C. A.; Sham, H. L. Bioorg. Med. Chem. Lett. 2004, 14, 5223.
11. Evans, D. A. Aldrichim. Acta 1982, 15, 23.
12. Jessup, P. J.; Petty, C. B.; Roos, J.; Overman, L. E. Org. Synth. Coll. 1988, 6, 95.
13. Hodgson, H. H. Chem. Rev. 1947, 40(2), 251.
14. The aequorin assay for ghrelin is described in detail in the following patent:
Amgen Inc. WO2006020959, 2006.
15. Wu, W.; McKown, L. Optimization in Drug Discovery: In Vitro Methods. In:
Methods in Pharmacology and Toxicology; Humana Press Inc.: Totowa, NJ, 2004;
163.
The SAR surrounding the benzyl group is outlined in Table 3. Re-
moval of the benzyl group as seen in structure 14 resulted in com-
plete loss of activity compared to the lead analog 4. Introducing
substitution in the benzylic position such as in analog 13b with
an IC50 of 2.5 nM proved to be beneficial over analog 13a which
had an IC50 value of 8 nM. The corresponding gem-dimethyl analog
13c showed twofold loss in potency with an IC50 of 16.5 nM com-
pared to 13a. The aromatic ring was not essential for potency