Discovery of novel 5-oxa-2,6-diazaspiro[3.4]oct-6-ene derivatives as potent, selective, and orally available somatostatin receptor subtype 5 (SSTR5) antagonists for treatment of type 2 diabetes mellitus

Article abstract of DOI:10.1016/j.bmc.2017.06.007

Somatostatin receptor subtype 5 (SSTR5) has emerged as a novel attractive drug target for type 2 diabetes mellitus. Starting from N-benzyl azetidine derivatives 1 and 2 as in-house hit compounds, we explored the introduction of a carboxyl group into the t

Full text of DOI:10.1016/j.bmc.2017.06.007

Accepted Manuscript
Discovery of novel 5-oxa-2,6-diazaspiro[3.4]oct-6-ene derivatives as potent,
selective, and orally available somatostatin receptor subtype 5 (SSTR5) antag-
onists for treatment of type 2 diabetes mellitus
Hideki Hirose, Takeshi Yamasaki, Masaki Ogino, Ryo Mizojiri, Yumiko
Tamura-Okano, Hiroaki Yashiro, Yo Muraki, Yoshihide Nakano, Jun Sugama,
Akito Hata, Shinji Iwasaki, Masanori Watanabe, Tsuyoshi Maekawa, Shizuo
Kasai
PII:
S0968-0896(17)30952-5
DOI:
http://dx.doi.org/10.1016/j.bmc.2017.06.007
BMC 13790
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
9 May 2017
5 June 2017
6 June 2017
Please cite this article as: Hirose, H., Yamasaki, T., Ogino, M., Mizojiri, R., Tamura-Okano, Y., Yashiro, H., Muraki,
Y., Nakano, Y., Sugama, J., Hata, A., Iwasaki, S., Watanabe, M., Maekawa, T., Kasai, S., Discovery of novel 5-
oxa-2,6-diazaspiro[3.4]oct-6-ene derivatives as potent, selective, and orally available somatostatin receptor subtype
5
(SSTR5) antagonists for treatment of type 2 diabetes mellitus, Bioorganic & Medicinal Chemistry (2017), doi:
http://dx.doi.org/10.1016/j.bmc.2017.06.007
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.

Discovery of novel 5-oxa-2,6-diazaspiro[3.4]oct-6-ene derivatives as
potent, selective, and orally available somatostatin receptor subtype 5
(SSTR5) antagonists for treatment of type 2 diabetes mellitus
a
Hideki Hirose *, Takeshi Yamasaki , Masaki Ogino , Ryo Mizojiri , Yumiko Tamura-Okano ,
b
a
b
b
b
Hiroaki Yashiro , Yo Muraki , Yoshihide Nakano , Jun Sugama , Akito Hata , Shinji Iwasaki ,
b
c
a
b
b
a
Masanori Watanabe , Tsuyoshi Maekawa , Shizuo Kasai
a
a
a
Research Division Medicinal Chemistry Laboratory, SCOHIA PHARMA, Inc., 26-1,
Muraoka-Higashi 2-Chome, Fujisawa, Kanagawa 251-8555, Japan
b
Research Division, Takeda Pharmaceutical Co., Ltd., Shonan Research Center, 26-1,
Muraoka-Higashi 2-Chome, Fujisawa, Kanagawa 251-8555, Japan
c
Global Procurement, Takeda Pharmaceutical Co., Ltd., 26-1, Muraoka-Higashi 2-Chome, Fujisawa,
Kanagawa 251-8555, Japan
*Corresponding author. Tel.: +81 466 32 1592; fax: +81 466 29 4471.
E-mail: hideki.hirose@scohia.com (H. Hirose).
Abstract
1

Somatostatin receptor subtype 5 (SSTR5) has emerged as a novel attractive drug target for type 2
diabetesmellitus. Starting from N-benzyl azetidine derivatives 1 and 2 asin-house hit compounds, we explored
the introductionof a carboxyl group into the terminal benzene of1 to enhance SSTR5 antagonistic activityby
the combination of the substituents at the 3-position of the isoxazoline. Incorporationof a carboxyl group atthe
4-position of the benzene ring resulted in a significant enhancement in potency, however, the 4-benzoic acid
derivative 10c exhibited moderate human ether-a-go-go related gene (hERG) inhibitoryactivity.Asubsequent
optimization studyrevealed that replacement of the 4-benzoic acid with an isonipecotic acid
dramatically reduced hERG inhibition (5.6% inhibition at 30 µM) by eliminating π–related interaction
+
with hERG K channel, which resulted in the identification of
1
4
-(2-((2,6-diethoxy-4'-fluorobiphenyl-4-yl)methyl)-5-oxa-2,6-diazaspiro[3.4]oct-6-en-7-yl)piperidin-
-carboxylic acid 25a (hSSTR5 / mSSTR5 IC50 = 9.6 / 57 nM). Oral administration of 25a in
high-fat diet fed C57BL/6J mice augmented insulin secretion in a glucose-dependent manner and
lowered blood glucose concentration.
Keywords: somatostatin, SSTR5, SSTR5 antagonist, anti-diabetic drug, hERG inhibition, OGTT
Abbreviations: SSTR5, somatostatin receptor subtype 5; OGTT, oral glucose tolerance test; hERG,
human ether-a-go-go related gene; DM, diabetes mellitus; HFD, high-fat diet; NCS,
2

N-chlorosuccinimide; WSC, 1-ethyl-3-(3-(dimethylamino)propyl) carbodiimide hydrochloride;
3 4
HOBt, 1-hydroxybenzotriazole; Et N, triethylamine; MS4A, molecular sieve 4A; LiAlH , lithium
n
3
aluminium hydride; ADDP, 1,1'-(azodicarbonyl)dipiperidine; P( Bu) , tributylphosphine; SAR,
3
structure-activity relationship; ; HBA, hydrogen bond acceptor; HBD, hydrogen bond donor; Fsp ,
3
fraction of sp carbons; CHO, Chinese hamster ovary; HLM, human liver microsome; MLM, mouse
liver microsome; KO, knock out.
1. Introduction
Diabetes mellitus (DM) is a disease characterized bypathologicallyelevated blood sugar levels due to insulin
secretion failure or insulin resistance, and is known to be a risk factor for various severe complications, such as
1
,2
diabetic neuropathy, retinopathyand nephropathy. The developmentof DM is affected byvarious
1
environmentalfactors, such as lack of exercise, overeating, and obesity, and genetic factors. The number of
diabetic patients will increase as the numberof obese people increases in the future. DM is classified into an
insulin-dependent type 1 diabetes mellitus (T1DM) and non-insulin-dependent type 2 diabetesmellitus (T2DM),
and a majorityof patients with DM have T2DM. T2DMis a disease caused bydecrease in the amountofβ cell
insulin secretion in response to glucose aswell asdecrease in insulin sensitivityin the peripheraltissues. T2DM
is treated with drug therapyin addition to diets and exercise therapy. Examplesof typicaldrugs presentlyused
for therapyinclude insulin, insulin analogues, glucagon-like peptide-1 (GLP-1) analogues, sulfonylurea agents,
3
biguanide agents, α-glucosidase inhibitors, PPARγ agonists, DPP-4 inhibitors, and SGLT-2 inhibitors. However
3

3
some patients withT2DM have insufficient glycemiccontrol and failto reach target HbA1c levels. To improve
glycemic control for these patients, it is of considerable importance to develop novel anti-diabetic drugs that act
1,3
via novel mechanisms.
Somatostatin (SST) is a peptide hormone comprising14 or 28 amino acids. It is widelydistributed throughout
the body, such as the hypothalamus, pancreatic islets ofLangerhans, and intestinal mucosa, and itplays an
important role in gastrointestinal motility, secretion ofdigestive juice, and regulation of glucose or lipid
4
,5
metabolism. Inparticular, somatostatin inhibits production or secretionof various hormones, proliferation
factors, and physiologicallyactive substances, such as growth hormone, thyroid stimulating hormone, prolactin,
insulin, glucagon, gastrin, secretin, peptideYY(PYY), gastric inhibition polypeptide (GIP), GLP-1,
6–11
cholecystokinin (CCK), vasoactive intestinal peptide (VIP), and oxyntomodulin. Furthermore, somatostatin
12,13
also acts as a paracrine hormone in the islets of Langerhans or the digestive tract mucosa. Thus, somatostatin
has various endocrine system, exocrine system, and nervous system impacts.
SST acts via five distinct somatostatin receptors (SSTR1–SSTR5), which are characterized as
14–16
seven-transmembrane G-protein coupled receptors.
Among them, SSTR5 is involved in the regulation of
insulin and incretin secretion. It is also reported that SSTR5 knockout mice fed high-fat diets display
1
1,17
anti-diabetic and anti-obese phenotypes. In contrast, SSTR2 is involved in the regulation of glucagon
18
secretion, and its inhibition results in elevationofblood-glucose level. Therefore, SSTR5 selective antagonists
19–23
are expected to improve hyperglycemia, hyperinsulinemia, and obesity.
On the basis of this background, we
initiated a program to develop novel SSTR5 antagonists for the treatment ofT2DM and obesity, and identified
4

N-benzyl azetidine derivatives1 and 2 with moderate SSTR5 antagonistic activityfrom high-throughput
screening (HTS) of our in-house compound library(Figure 1). In this article, we will report the design and
synthesis of the 5-oxa-2,6-diazaspiro[3.4]oct-6-ene derivatives as new SSTR5 antagonists as well as
the pharmacological effect of representative compound 25a. In the course of this report, we will also
discuss the elimination of hERG inhibition of 4-benzoic acid derivative 10c by modification at the
benzyl position of N-benzyl azetidine or modification of the benzene ring in the 4-benzoic acid
moiety.
OEt
OEt
N
N
O
HO2C
O
N
N
OEt
OEt
F
F
1
2
hSSTR5 / mSSTR5 IC50 = 2800 / 980 nM
hSSTR5 / mSSTR5 IC50 = 3200 / 220 nM
Figure 1. HTS hit compounds 1 and 2.
2. Chemistry
Scheme 1 describes the synthesis of spiroisoxazoline derivatives 10a–h, and 11 with various
substituents at the 3-position of the isoxazoline ring. The synthesis was started from the reaction of
aldehydes 3a–h with hydroxylamine, followed by chlorination by using N-chlorosuccinimide (NCS)
to furnish intermediates 5a–h. 1,3-Dipolar cycloaddition of 5a–h and tert-butyl
3
-methyleneazetidine-1-carboxylate gave spiroisoxazoline derivatives 6a–h, which were subjected to
5

Boc deprotection under acidic conditions to yield azetidine hydrochlorides 7a–h. Subsequent
reductive amination of 7a–h with 2,6-diethoxy-4'-fluorobiphenyl-4-carbaldehyde (8) provided
N-benzyl azetidine derivatives 9a–h. Hydrolysis reaction of 9a–f gave carboxylic acid derivatives
10a–f. Palladium-catalyzed carbonylative cross coupling reaction of 9g and 9h provided methyl ester
derivatives 9g' and 9h', which were subsequently hydrolyzed to yield carboxylic acid derivatives 10g
and 10h. Carboxylic acid derivative 10c was converted to the corresponding carboxamide 11.
N
H
OH
N
OH
NBoc
a
b
c
R
A
CHO
R
A
R
A
R
A
Cl
N O
3
3
a-f R = CO2Me
g,h R = Br
4a-f R = CO2Me
4g,h R = Br
5a-f R = CO2Me
5g,h R = Br
6a-f R = CO2Me
6g,h R = Br
OHC
OEt
OEt
8
F
OEt
N
NH
d
e
g
R
A
R
A
O
N
O
HCl
N
OEt
F
7
7
a-f R = CO2Me
g,h R = Br
9
9
9
a-f R = CO2Me
g,h R = Br
g',h' R = CO2Me
f
OEt
OEt
O
N
N
h (for 10c)
HO2C
A
H2N
O
N
O
N
OEt
OEt
F
F
1
0a-h
11
Compound
a
b
c
d
e
f
g
h
Cl
Me
R
R
R
A
N
R
R
R
R
R
R
Scheme 1. Synthesis of spiroisoxazoline derivatives 10a–h, and 11. Reagents and conditions: (a)
hydroxylamine hydrochoride, NaHCO
3
, MeOH, rt, 35–88%; (b) NCS, DMF, rt; (c) tert-butyl
6

3
3
-methyleneazetidine-1-carboxylate, Et N, THF, reflux; (d) 4 M HCl/AcOEt, rt, 25–84% over 3
steps; (e) 8, NaBH(OAc)
3
, Et
3
N, THF, rt, 42–86%; (f) CO (0.5 MPa), PdCl
2
2 2 3
(dppf)·CH Cl , Et N,
o
o
MeOH, DMF, 90 C, 66–80%; (g) 1 M NaOH aq, EtOH, THF, 50 C, 66–89%; (h) WSC, HOBt
ammonia salt, DMF, rt, 66%.
The synthesis of benzyl alcohol derivative 12 and phenylacetic acid derivative 14 is described in
Scheme 2. Reduction of the methyl ester of 9c gave benzyl alcohol 12. Mitsunobu reaction of 12
with acetone cyanohydrin furnished phenylacetonitrile 13, which was subsequently hydrolyzed to
yield the desired compound 14.
4
Scheme 2. Synthesis of spiroisoxazoline derivatives 12 and 14. Reagents and conditions: (a) LiAlH ,
o
n
3
THF, 0 C to rt, 64%; (b) acetone cyanohydrin, ADDP, P( Bu) , THF, rt, 75%; (c) 1 M NaOH aq,
o
EtOH, 180 C, microwave irradiation, then 6 M HCl, 75%.
7

Synthesis of compounds 17 and 20 is depicted in Scheme 3. Alkylation of 7c with benzyl chloride
derived from 1-(2,6-diethoxy-4'-fluorobiphenyl-4-yl)ethanol (15) and subsequent hydrolysis reaction
afforded the desired compound 17. Compound 7c was subjected to a condensation reaction with
2,6-diethoxy-4'-fluorobiphenyl-4-carboxylic acid (18) to give amide 19, which was hydrolyzed to
yield the desired compound 20.
Scheme 3. Synthesis of spiroisoxazoline derivatives 17 and 20. Reagents and conditions: (a) (i) 15,
o
o
2 3
SOCl , toluene, rt, (ii) Et N, DMF, 70 C, 35% over 2 steps; (b) 1 M NaOH aq, EtOH, THF, 50 C,
3
59%; (c) 18, WSC, HOBt, Et N, DMF, rt, 75%; (d) 1 M NaOH aq, EtOH, THF, rt, 76%.
Synthesis of isonipecotic acid derivative 25a and pyrrolidine-3-carboxylic acid derivative 25b is
illustrated in Scheme 4. Intermediate 22 was synthesized by 1,3-dipolar cycloaddition of tert-butyl
8

3
N
-methyleneazetidine-1-carboxylate and 1,1-dibromoformaldoxime (21) and then an S Ar reaction
with cyclic amines afforded 23a and 23b. The desired compounds 25a and 25b were prepared by
Boc deprotection of 23a and 23b and hydrolysis after reductive amination with 8.
Scheme 4. Synthesis of spiroisoxazoline derivatives 25a and 25b. Reagents and conditions: (a)
o
, AcOEt, rt to 50 C, 79%; (b) ethyl
tert-butyl 3-methyleneazetidine-1-carboxylate, NaHCO
3
o
, DMF, 130 C, 80% (for 23a); (c) methyl
piperidine-4-carboxylate, Na
2
CO
3
o
, DMF, 130 C (for 23b); (d) formic acid, then 8,
pyrrolidine-3-carboxylate hydrochloride, Na
2
CO
3
NaBH(OAc)
3
, Et
3
N, THF, rt, 79% (for 24a), 11% over 2 steps (for 24b); (e) 2 M NaOH aq, EtOH, 70
o o
C, 78% (for 25a); (f) 2 M NaOH aq, EtOH, 70 C, then 1 M HCl, 97% (for 25b).
Scheme 5 describes the synthetic steps to prepare azetidine-3-carboxylic acid derivative 25c. After
the removal of the Boc group of intermediate 22, reductive amination of azetidine hydrochloride 26
9

N
with 8 provided N-benzyl azetidine 27. An S Ar reaction of ethyl azetidine-3-carboxylate with 27
gave intermediate 28, which was subsequently hydrolyzed to yield the desired compound 25c.
Scheme 5. Synthesis of spiroisoxazoline derivative 25c. Reagents and conditions: (a) 4 M
HCl/AcOEt, AcOEt, rt, 60%; (b) 8, NaBH(OAc)
3
, Et
3
N, THF, rt, 84%; (c) ethyl
n
o
, BuOH, 120 C; (d) 2 M NaOH aq, EtOH, 70 C,
o
azetidine-3-carboxylate hydrochloride, Na
then 2 M HCl, 13% over 2 steps.
2
CO
3
Scheme 6 describes the synthesis of benzaldehyde 8, benzyl alcohol 15, and benzoic acid 18.
Alkylation reaction of 4-bromo-3,5-dihydroxybenzoic acid (29) with iodoethane, followed by
palladium-catalyzed coupling reaction with (4-fluorophenyl)boronic acid gave biphenyl intermediate
31. The hydrolysis reaction of 31 provided benzoic acid 18. Reduction of the ethyl ester of 31
furnished alcohol 32, which was then oxidized to give benzaldehyde 8. Reaction of 8 with
1
0

methylmagnesium bromide furnished benzyl alcohol 15.
Scheme 6. Synthesis of benzaldehyde 8, benzyl alcohol 15, and benzoic acid 18. Reagents and
o
, DMF, 60 C, 82%; (b) (4-fluorophenyl)boronic acid, Palladium
conditions: (a) iodoethane, K
2
CO
3
o
, toluene, water, 90 C, 87%; (c) 2M NaOH aq, MeOH, THF,
acetate, tricyclohexylphosphine, K
3
PO
4
o
o
6
4 3
0 C, 89%; (d) LiAlH , THF, 0 C; (e) Sulfer trioxide-pyridine complex, Et N, DMSO, rt, 84% over
2
steps; (f) MeMgBr (1M in THF solution), THF, rt, 77%.
3. Results and discussion
The antagonistic activities of the compounds synthesized in this study against human SSTR5
11

(
hSSTR5) and mouse SSTR5 (mSSTR5) were evaluated in an assay of their functional ability to alter
cAMP levels in Chinese hamster ovary (CHO) cells stably expressing hSSTR5 and mSSTR5, with
somatostatin 14 (SST14) serving as a ligand.
N-Benzyl azetidine derivatives 1 and 2 were identified as hit compounds with moderate SSTR5
antagonistic activity via in-house high-throughput screening (HTS). The core structure of the hit
compounds was 5-oxa-2,6-diazaspiro[3.4]oct-6-ene (a unique spiro ring system), and the substituent
at the 1-position of the azetidine ring was a (2,6-diethoxy-4'-fluorobiphenyl-4-yl)methyl group. Since
the substituent at the 3-position of the isoxazoline ring of the hit compounds was a phenyl group or a
propionic acid moiety, we hypothesized that a polar group could be introduced through a
hydrophobic linker to the 3-position of the isoxazoline ring, and effective interactions of the
substituent at the 3-position with the binding site of SSTR5 would enhance SSTR5 antagonistic
activity. Hence, we attempted to introduce a carboxyl group into the benzene ring at the 3-position of
the isoxazoline ring and replace the carboxyl group with other polar groups to discover a lead
compound, such as model compound I (Figure 2).
1
2

Figure 2. Design of model compound I from HTS hit compounds 1 and 2.
Table 1 summarizes the SSTR5 antagonistic activity of compounds with a variety of substituents at
the 3-position of the isoxazoline ring. Initially, we examined the effect of placing a carboxyl group
on the terminal benzene ring. The benzoic acid derivatives from most to least potent were 10a
(
ortho) < 10b (meta) < 10c (para). Notably, the 4-benzoic acid derivative 10c exhibited dramatically
enhanced SSTR5 antagonistic activity, with IC50 values of 9.0 and 33 nM for hSSTR5 and mSSTR5,
respectively. In contrast, replacement of the 4-benzoic acid with a 4-phenyl acetic acid (14) caused a
25-fold decline in potency. These results suggested that the 4-position of the terminal benzene ring
was the best substitution position for the carboxyl group. In addition, we assumed that the significant
enhancement in potency resulted from the optimal positioning of the carboxyl group on the terminal
benzene ring as well as from π–related and/or hydrophobic interactions of the benzene ring with the
binding site of SSTR5.
Subsequently, we examined the effect of replacing the carboxyl group with other polar groups to
confirm the interaction mode of the carboxyl group of 10c with SSTR5. On the assumption that the
1
3

carboxyl group would form hydrogen bonds with putative amino acid residue(s) of SSTR5, we
synthesized compound 11 with a carboxamide moiety as HBD (hydrogen bond donor) or HBA
(
hydrogen bond acceptor) and compound 12 with a hydroxymethyl group as HBD. As a result, these
replacements led to profound loss of potencies. Notably, replacement with a neutral hydroxymethyl
group (12) caused a significant decline in potency in comparison with a weakly acidic carboxamide
(
11). These results suggested that putative basic amino acid residue(s) in the binding site of SSTR5
interact with the carboxyl group of 10c by electrostatic interaction.
Replacement of the benzene ring at the 3-position of the isoxazoline ring with a pyridine ring (10d)
caused a 7-fold decline in hSSTR5 antagonistic activity in comparison with 10c. Since this
modification displayed limited reduction in potency in comparison with carboxamide derivative 11
and hydroxymethyl derivative 12, we suggested that some hydrophilic structures would be tolerable
for the neighborhood of the 3-position of the isoxazoline ring. Through our initial structure-activity
relationship (SAR) study on a substituent at the 3-position of the isoxazoline ring, we successfully
discovered 4-benzoic acid derivative 10c as a lead compound with excellent SSTR5 antagonistic
activity.
Table 1
1
4

SSTR5 antagonistic activity: SAR of the 3-position of isoxazoline ring
a
IC50 (nM)
Compound R
b
c
hSSTR5
>10000
mSSTR5
1
1
1
1
0a
0b
0c
>10000
5
50
260
(
340–910) (170–390)
9
.0
33
(
4.5–18)
(13–81)
2
20
74
d
4
(
120–390) (37–150)
1
50
150
11
(
110–220) (110–210)
1
000
430
1
1
2
HO
(530–2100) (250–720)
6
1
20
0d
(
36–110)
(6.5–62)
a
IC50 values were presented as the means of duplicate experiments with 95% confidence intervals in
parentheses.
b
Antagonistic activity against human SSTR5.
c
Antagonistic activity against mouse SSTR5.
d
Hydrochloride salt.
1
5

Because 4-benzoic acid derivative 10c was found to exhibit moderate hERG inhibitory activity
65% inhibition at 30 µM) in our ADME/Tox study, elimination of the hERG inhibition associated
(
with potential lethal arrhythmia was extremely important for the development of SSTR5 antagonists
2
4,25
as anti-diabetic drugs.
Accordingly, our lead optimization study was directed toward addressing
the issue of undesirable hERG inhibition.
It was reported that a number of existing hERG inhibitors possessed a basic amine moiety as a
+
positively charged group, and cation–π interaction of the basic amine with hERG K channel affected
26–28
hERG inhibitory activity.
On the basis of this report, we initially assumed that the basic amine of
the azetidine ring caused the hERG inhibition of 10c. Therefore, we attempted to reduce hERG
inhibitory activity of 10c by following two synthetic strategies: introduction of a methyl group at the
alpha-position of the basic amine to prevent cation–π interaction by steric hindrance, or removal of
basicity by amidation of the azetidine ring (Table 2).
Addition of a methyl group (17) on the benzyl position or amidation (20) of the azetidine resulted in
significant reduction of hERG inhibitory activity regardless of change in lipophilicity, suggesting
that these modifications would be effective for the removal of cation–π interaction of the basic amine
+
in 10c with the hERG K channel. In contrast, given the significantly decreased potency of these
compounds, we concluded that the basic amine of the azetidine ring played an important role in
SSTR5 antagonistic activity as well as hERG inhibitory activity. Consequently, further modification
1
6

of the benzyl position for reduction of hERG inhibition was thought to be difficult because of the
narrow SAR.
Table 2
SSTR5 antagonistic activity and hERG inhibitory activity: SAR of benzyl position of N-benzyl
azetidine
a
IC50 (nM)
d
hERG inhibition
e
Compound
R
LogD7.4
b
c
hSSTR5
mSSTR5
33
(% inhibition)
9.0
1
1
0c
7
65
14
3.3
(
4.5–18) (13–81)
95
2000
3.6
2.8
(
46–200) (1200–3500)
2
0
>10000
>10000
7.9
a
IC50 values were presented as the means of duplicate experiments with 95% confidence intervals in
parentheses.
b
Antagonistic activity against human SSTR5.
c
Antagonistic activity against mouse SSTR5.
d
Percent inhibition of hERG inhibitory activity at 30 µM compound concentration.
e
29
The log D value at pH 7.4.
1
7

+
Subsequently, we focused on the interaction of the benzoic acid with hERG K channel. Although
introduction of a carboxyl group into a hERG inhibitor was generally reported to attenuate hERG
3
0,31
inhibition by electron repulsion,
the 4-benzoic acid derivative 10c still showed moderate hERG
inhibitory activity, unfortunately. In contrast, after an in-depth analysis of hERG inhibition by
analogous compounds, we recognized that the 2-benzoic acid derivative 10a exhibited a significant
reduction of hERG inhibitory activity (Figure 3). Because the introduction of a carboxyl group at the
2
-position of the terminal benzene ring creates torsion between the benzene and the isoxazoline ring,
the 2-carboxyl group was thought to alter the conformation of the benzene ring as well as change the
position of the carboxyl group from that of the 4-benzoic acid. In addition, given that the aromatic
+
26–28
rings in existing hERG inhibitors interact with hERG K channel in π–related manner,
we
surmised that the reduced hERG inhibition of 10a was due not only to the electron repulsion of the
2
-carboxyl group but also to the change in benzene ring conformation. Accordingly, we attempted to
reduce the hERG inhibition by introducing substituents at the 2-position of the 4-benzoic acid of 10c
to change its conformation.
Figure 3. Profiles of 2-benzoic acid derivative 10a.
1
8

To validate the effects of the 2-substituents against hERG inhibition of 10c, a chlorine atom (10g)
and a methyl group (10h) were introduced at the 2-position of the 4-benzoic acid (Table 3). These
modifications attenuated hERG inhibition relative to 10c (10g: 40% inhibition, 10h: 18% inhibition
at 30 µM), and decreased SSTR5 antagonistic activity (5-fold for 10g and 15-fold for 10h). Because
these reductions in hERG inhibition were not attributable to the electron-withdrawing or
electron-donating effects of the 2-substituent, we hypothesized that the torsion between the benzene
+
and isoxazoline had disrupted the interactions within hERG K channel.
According to the calculation of the dihedral angle of simplified models 10c', 10g', and 10h' by
32
using MOE software, the 4-benzoic acid and the isoxazoline ring of 10c' positioned on the same
plane, however, the introduction of a chlorine atom (10g') or a methyl group (10h') gave a torsion
with the dihedral angle of 12 or 30 degrees, respectively (Figure 4). Relating torsion to hERG
inhibition, hERG inhibitory activity was well correlated with dihedral angle, and 10h had the largest
dihedral angle and the lowest hERG inhibitory activity. From these results, we suspected that a
+
π–related interaction occurred between the terminal benzene ring and hERG K channel and that
torsion between the benzene and isoxazoline could effectively prevent the interaction. Consequently,
we found that the interaction with the terminal benzene ring was one of the key contributors to the
hERG inhibition of 10c. Regarding the balance between SSTR5 antagonistic activity and hERG
inhibition, we found that the modification of the terminal benzene ring resulted in more limited
1
9

reduction in potency in comparison with that of the benzyl position. Accordingly, we focused on
further structural modification of the terminal benzene ring to reduce hERG inhibitory activity
without a loss of potency.
Table 3
SSTR5 antagonistic activity and hERG inhibitory activity: SAR of substituents at the 2-position of
the 4-benzoic acid
Compound
a
IC50 (nM)
b
hSSTR5 / mSSTR5
4
6 (24–90) / 31 (19–52)
0
140 (71–270) / 21 (14–32)
c
d
hERG inhibition
4
18
(
% of inhibition)
e
LogD7.4
3.5
3.5
a
IC50 values were presented as the means of duplicate experiments with 95% confidence intervals in
parentheses.
b
Antagonistic activity against human SSTR5.
c
Antagonistic activity against mouse SSTR5.
d
Percent inhibition of hERG inhibitory activity at 30 µM compound concentration.
e
29
The log D value at pH 7.4.
2
0

Me
Me
Me
O
N
HO
O
N
1
0c'
=
0 degree
Cl
O
N
HO
O
N
1
0g'
=
11.7 degree
Me
O
N
HO
O
N
1
0h'
=
29.6 degree
Figure 4. Dihedral angles between the benzene and the isoxazoline.
+
To eliminate the π–related interaction of the terminal benzene with hERG K channel, we replaced
a benzene ring in the 4-benzoic acid with a saturated ring (Table 4). As a result, the hERG inhibitory
activity of cyclohexane derivatives 10e and 10f, and cyclic amine derivatives 25a–c decreased
dramatically, while maintaining a lipophilicity comparable to that of 10c. These results were
+
consistent with our reasoning that the terminal benzene ring of 10c interacted with hERG K channel
by π–related interaction. Based on the data in Tables 3 and 4, we concluded that structural
modifications, such as torsion of a benzene ring or replacement with a saturated ring, could reduce
the hERG inhibitory activity of 10c without decreasing lipophilicity.
2
1

In terms of the SSTR5 antagonistic activity, replacement of the benzene ring with a
trans-cyclohexane (10e) resulted in potency comparable to that of 10c; however, replacement with a
cis-cyclohexane (10f) significantly reduced potency. These results suggested that the carboxyl group
at the equatorial-position of 10e (as observed in 10c) is preferable to that at the axial-position of 10f.
In addition, the isonipecotic acid derivative 25a displayed potent SSTR5 antagonistic activity as
mentioned above, with IC50 values of 9.6 and 57 nM for hSSTR5 and mSSTR5, respectively.
Conversion into a pyrrolidine (25b) or an azetidine (25c) exhibited significant decline in hSSTR5
antagonistic activities, probably because their carboxyl groups were unable to effectively interact
with the basic amino acid residue(s). Consequently, we showed that replacement of the 4-benzoic
acid with the isonipecotic acid dramatically reduces hERG inhibitory activity while maintaining
potent SSTR5 antagonistic activity. The isonipecotic acid derivative 25a was selected as a promising
compound for further in vivo evaluation.
Table 4
2
2

SSTR5 antagonistic activity and hERG inhibitory activity: SAR of saturated rings at the 3-position of
the isoxazoline ring
a
IC50 (nM)
d
hERG inhibition
c
mSSTR5 (% inhibition)
e
Compound
LogD7.4
3.2
b
hSSTR5
3
0
38
1
1
2
2
2
0e
15
(
13–67)
(18–81)
130
210
0f
33
3.3
(
80–210)
(120–350)
9.6
57
5a
5b
5.6
3.0
(
4.6–20)
(33–99)
190
48
f
17
2.9
(
140–270) (33–69)
200
120
f
5c
8.1
2.9
(
150–260) (61–240)
a
IC50 values were presented as the means of duplicate experiments with 95% confidence intervals in
parentheses.
b
Antagonistic activity against human SSTR5.
c
Antagonistic activity against mouse SSTR5.
d
Percent inhibition of hERG inhibitory activity at 30 µM compound concentration.
2
3

e
f
29
The log D value at pH 7.4.
Hydrochloride salt.
Because inhibition of SSTR2 was known to induce the elevation of blood-glucose level by
glucagon secretion, we examined the selectivity of 25a for SSTRs1–4 (Table 5). The binding
affinities of 25a to SSTRs1–4 were low (i.e. less than 15% inhibition at 10 µM), indicating that 25a
is a highly selective SSTR5 antagonist.
Table 5
a
Selectivity of 25a over SSTRs1–4
hSSTR1
hSSTR2
hSSTR3
14
hSSTR4
10
Compound
5a
(
% inhibition) (% inhibition) (% inhibition) (% inhibition)
2
11
8
a
Inhibitory activities on SSTRs1–4 were evaluated at a drug concentration of 10 µM by radioligand
3
3
binding assay.
Prior to the in vivo evaluation, the selected compound 25a was evaluated for metabolic stability,
pharmacokinetic properties in mice, and solubility (Tables 6 and 7). Compound 25a exhibited good
metabolic stability toward both human and mouse microsomes. Pharmacokinetic screening in mice
indicated that 25a was orally available with acceptable plasma exposure. In addition, 25a exhibited
excellent solubility at pH 6.8 (260 µg/ml). Indeed, 25a displayed a 200-fold higher solubility than
2
4

10c (1.3 µg/ml), suggesting that the replacement with an isonipecotic acid also contributed to the
improvement of solubility. This result was connected with increase of molecular flexibility explained
3
by the fraction of sp carbons (Fsp = the number of sp hybridized carbons/total carbon count), and
3
3
34
3
3
5a showed larger Fsp value than 10c (Fsp values: 10c; 0.31, 25a; 0.51). On the basis of these
2
results, the isonipecotic acid derivative 25a was found to be suitable for in vivo evaluation.
Table 6
Microsomal clearance and solubility of 25a
a
in vitro CLint
b
Solubility at pH6.8
Compound (µL/min/mg)
(
µg/ml)
HLM MLM
<10 19
2
5a
260
a
Human/mouse liver microsomal clearance.
b
The second fluid for the disintegration test is described in the Japanese Pharmacopoeia 15th edition
(
pH6.8).
Table 7
a
Pharmacokinetic profiles of 25a
iv (0.1 mg/kg)
b
po (1 mg/kg)
e
Cmax
h
F
c
d
f
Tmax AUC0-8 h
g
Compound CLtotal
Vss
MRTiv
1.7
(
%)
(
mL/h/kg) (mL/kg) (h)
(ng/mL) (h)
(ng∙h/mL)
25a
1761
3052
74.8
2.0
332
58
a
ICR mouse (male, 8 weeks, n = 3).
b
c
Total clearance.
Volume of distribution at steady state.
2
5

d
e
Mean residence time.
Maximal plasma concentration.
f
Time of maximal concentration.
g
h
Area under the plasma concentration–time curve (0–8 h).
Bioavailability.
Compound 25a was examined for its effect on postprandial hyperglycemia in an oral glucose
tolerance test (OGTT) performed in high-fat diet (HFD) fed C57BL/6J mice. Compound 25a (100
mg/kg), Alogliptin (30 mg/kg), and Glibenclamide (10 mg/kg) were administered orally 1 h before
the oral glucose load (5 g/kg), and blood glucose and insulin concentrations were monitored over 2 h.
As shown in Figure 3, the maximum efficacy of 25a was superior to that of Glibenclamide and
comparable to that of Alogliptin (Figure 5A and 5B). In addition, 25a augmented insulin secretion in
a glucose-dependent manner and displayed a blood glucose-lowering effect, indicating its
anti-diabetic efficacy in vivo (Figure 5C and 5D). Furthermore, 25a did not evoke hypoglycemia
induced by the administration of Glibenclamide before the oral glucose load.
2
6

Figure 5. Effects of 25a on glucose excursion and insulin secretion measured during oral glucose
tolerance tests in HFD fed C57BL/6J mice. Compound 25a (100 mg/kg), Alogliptin (30 mg/kg), and
Glibenclamide (10 mg/kg) were orally administered 1 h before the oral glucose load (5 g/kg). (A)
and (C) show time–dependent changes in blood glucose and insulin levels after a 5 g/kg glucose
challenge, respectively. (B) represents the area under the curve (AUC) of blood glucose levels shown
in (A). (D) represents the AUC of plasma insulin levels shown in (C). Data are presented as mean ±
SD (n = 6). (*) p < 0.05 versus vehicle by Dunnett’s test. (**) p < 0.01 versus vehicle by Dunnett’s
test. (***) p < 0.001 versus vehicle by Dunnett’s test.
To evaluate its dose-dependent effect on the OGTT, 25a (1, 3, 10, 30 mg/kg) was orally
2
7

administered to HFD fed C57BL/6J mice. As shown in Figure 6, the blood glucose-lowering effect
of 25a was dose-dependent detectable from 1 mg/kg and reached a maximum at 10 mg/kg. In
addition, 25a did not display blood glucose-lowering effect in OGTT in SSTR5 KO mice, indicating
that the efficacy of 25a depended on its SSTR5 antagonistic activity (data not shown). From these in
vivo evaluations, we concluded that this SSTR5 antagonist is potentially an attractive anti-diabetic
drug without hypoglycemic risk.
Figure 6. Dose-dependent effect of 25a on glucose excursion measured during the oral glucose
tolerance test in HFD fed C57BL/6J mice. Compound 25a (1, 3, 10, 30 mg/kg) was orally
administered. (A) shows time–dependent changes in blood glucose levels after a 5 g/kg glucose
challenge. (B) represents the area under the curve (AUC) of blood glucose levels shown in (A). Data
are presented as mean ± SD (n = 5). (*) p < 0.05 versus vehicle by Williams’s test. (**) p < 0.01
versus vehicle by Williams’s test.
4. Conclusion
2
8

Aiming to develop novel anti-diabetic drugs, we conducted synthetic studies in an effort to obtain
novel and potent SSTR5 antagonists. Starting from N-benzyl azetidine derivatives1 and 2 as in-house hit
compounds, we introduced a carboxyl group into theterminalbenzene ring of1 to enhance SSTR5 antagonistic
activitythrough the combination of a phenyl group and a propionic acid moietyas the substituents at the
3-position of the isoxazoline ring.As a result, incorporationof a carboxyl group at the 4-position of the benzene
ring resulted in significant enhancement in potency, and the 4-carboxygroup efficientlyinteracted with basic
amino acid residue(s) in the binding site of SSTR5. Since the 4-benzoic acid derivative 10c exhibited moderate
hERG inhibitoryactivityin theADME/Tox study, ouroptimization studyaddressed the issue of this undesirable
hERG inhibition. Structural modifications were focused on the benzyl position of the N-benzyl azetidine or the
benzene ring of the 4-benzoic acid.Among these modifications, increasing the torsion between the benzene and
the isoxazoline ring, or replacement of the benzenering with a saturated ring were effective strategies for
+
eliminating π–related interaction of the terminal benzene ring with hERG K channel. Notably,
replacement of the 4-benzoic acid with an isonipecotic acid dramatically reduced hERG inhibition
while maintaining SSTR5 antagonistic potency, which led to the discovery of a novel and potent
SSTR5 antagonist 25a. The introduction of an isonipecotic acid also significantly improved
solubility at pH 6.8 by increasing molecular flexibility. In addition, 25a displayed oral bioavailability
in mice with excellent plasma exposure. Oraladministration of 25a in HFD fed C57BL/6J mice
augmented insulin secretion in a glucose-dependent manner to lower blood glucose without
2
9