Discovery of a potent glucokinase activator with a favorable liver and pancreas distribution pattern for the treatment of type 2 diabetes mellitus

Article abstract of DOI:10.1016/j.ejmech.2018.06.060

Glucokinase (GK) is an enzyme that plays an important role as a glucose sensor while maintaining whole body glucose homeostasis. Allosteric activators of GK (GKAs) have the potential to treat type 2 diabetes mellitus. To identify novel GKAs, a series of compounds based on a thiophenyl-pyrrolidine scaffold were designed and synthesized. In this series, compound 38 was found to inhibit glucose excursion in an oral glucose tolerance test (OGTT) in mice. Optimization of 38 using a zwitterion approach led to the identification of the novel GKA 59. GKA 59 exhibited potent blood glucose control in the OGTT test as well as a favorable safety profile. Owing to low pancreatic distribution, compound 59 primarily activates GK in the liver. This characteristic could overcome limitations of other GKAs, such as hypoglycemia, increased plasma triglycerides, and loss of efficacy.

Full text of DOI:10.1016/j.ejmech.2018.06.060

Accepted Manuscript
Discovery of a potent glucokinase activator with a favorable liver and pancreas
distribution pattern for the treatment of type 2 diabetes mellitus
Hiroki Fujieda, Masakazu Kogami, Masao Sakairi, Noriyasu Kato, Mitsuhiro Makino,
Naoki Takahashi, Toshiyuki Miyazawa, Satoko Harada, Tokuyuki Yamashita
PII:
S0223-5234(18)30551-8
10.1016/j.ejmech.2018.06.060
EJMECH 10530
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 26 March 2018
Revised Date: 22 June 2018
Accepted Date: 23 June 2018
Please cite this article as: H. Fujieda, M. Kogami, M. Sakairi, N. Kato, M. Makino, N. Takahashi, T.
Miyazawa, S. Harada, T. Yamashita, Discovery of a potent glucokinase activator with a favorable liver
and pancreas distribution pattern for the treatment of type 2 diabetes mellitus, European Journal of
Medicinal Chemistry (2018), doi: 10.1016/j.ejmech.2018.06.060.
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.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Discovery of a Potent Glucokinase Activator with a
Favorable Liver and Pancreas Distribution Pattern for
the Treatment of Type 2 Diabetes Mellitus
Hiroki Fujieda,* Masakazu Kogami, Masao Sakairi, Noriyasu Kato, Mitsuhiro Makino,
Naoki Takahashi, Toshiyuki Miyazawa, Satoko Harada, Tokuyuki Yamashita
Drug Discovery Laboratories, Sanwa Kagaku Kenkyusho Co., Ltd., 363 Shiosaki,
Hokusei-cho, Inabe-shi, Mie 511–0406, Japan
Corresponding author:
Hiroki Fujieda h_fujieda@skk-net.com
Abstract
Glucokinase (GK) is an enzyme that plays an important role as a glucose sensor while
maintaining whole body glucose homeostasis. Allosteric activators of GK (GKAs) have
the potential to treat 2 diabetes mellitus. To identify novel GKAs, a series of
compounds based on a thiophenyl-pyrrolidine scaffold were designed and synthesized.
In this series, compound 38 was found to inhibit glucose excursion in an oral glucose
tolerance test (OGTT) in mice. Optimization of 38 using a zwitterion approach led to
the identification of the novel GKA 59. GKA 59 exhibited potent blood glucose control
in the OGTT test as well as a favorable safety profile. Owing to low pancreatic
distribution, compound 59 primarily activates GK in the liver. This characteristic could
overcome limitations of other GKAs, such as hypoglycemia, increased plasma
triglycerides, and loss of efficacy.
Keywords
Type 2 diabetes; Glucokinase; Glucokinase activator; thiophenyl-pyrrolidine scaffold;
hERG; Zwitterion
1

ACCEPTED MANUSCRIPT
1
.
Introduction
Diabetes is one of the most common chronic diseases worldwide. According to the
International Diabetes Federation, diabetes affected approximately 425 million people
in 2017, and this number is expected to rise to 629 million by 2045. The increasing
prevalence of diabetes is driven by a variety of factors, including diet, urbanization, and
obesity. Type 2 diabetes mellitus (T2DM) is the predominant form of the disease and
constitutes the majority of adult cases worldwide [1,2]. T2DM is typically characterized
by hyperglycemia, insulin resistance, abnormally elevated hepatic glucose production,
and inadequate glucose-stimulated insulin secretion (GSIS) from pancreatic β-cells [3].
Insufficient blood glucose control increases the risk of vascular complications such as
coronary artery disease, peripheral arterial disease, stroke, nephropathy, neuropathy, and
retinopathy [4]. Although various oral anti-hyperglycemic agents are available,
monotherapy or combination regimens are frequently inadequate for maintaining blood
glucose levels in the long term. Furthermore, many of these agents exhibit side effects
such as hypoglycemia, weight gain, gastrointestinal side effects, and genitourinary
infection [5,6]. Therefore, an unmet need exists for more effective therapies offering
improved efficacy and safety for the management of diabetes.
GK, also called hexokinase IV or D, is a glycolytic enzyme that converts glucose to
glucose-6-phosphate. GK has unique characteristics compared with those of the
ubiquitously expressed hexokinase isoforms I–III, and is predominantly expressed in the
pancreas and liver. In pancreatic β-cells, GK regulates the threshold for GSIS. In
hepatocytes, GK regulates glycogen synthesis and hepatic glucose production, and its
activity is controlled by glucokinase regulatory protein (GKRP) [7–9]. Considering that
the activation of GK in the pancreas and liver could lead to increased insulin secretion
from pancreatic β-cells and glucose uptake in the liver, allosteric activators of GK
(GKAs) have emerged as attractive targets for the treatment of T2DM [10].
Following the initial discovery of RO0281675 [11,12], several GKAs, including phenyl
acetamides, benzamides, and imidazolylacetamides, have been reported (Figure 1)
[
13–18].
2

ACCEPTED MANUSCRIPT
Figure 1. Representative structures of glucokinase activators from the literature and the
structure of our lead compound 7.
To date, several GKAs have advanced to clinical studies and have demonstrated an
ability to lower blood glucose levels in both healthy subjects and patients with T2DM.
However, in these clinical studies hypoglycemia, increased plasma triglycerides (TG),
and loss of efficacy within several months, were determined to be the main obstacles to
developing GKAs as therapeutic agents for T2DM [16,19]. It is believed that
hypoglycemia is caused by the overstimulation of islet β-cell GK, while increased
plasma TG is the result of overstimulation of liver GK. However, the cause of the loss
of efficacy in the short term remains to be elucidated [20]. To mitigate these limitations,
various types of GKAs have been discovered and have progressed to human clinical
trials [18]. Although most were terminated at Phase I or Phase II, some GKAs
(including vTv Therapeutics’s TTP399 and Hua Medicine’s HMS5552) remain
currently active and have shown positive results in recent clinical trials [21–23]. These
two compounds exhibit characteristics that are unique among GKAs. TTP399 is a
liver-selective GKA that does not affect the inhibition of GK by GKRP in the liver. The
manufacturer suggests that disruption of the GK-GKRP interaction leads to
hypoglycemia and dyslipidemia. HMS5552 has an amino acid-based structure, which is
markedly different from that of other GKAs. The development of GKA as a clinical
agent for the treatment of T2DM therefore requires the identification of a new type of
GKA with characteristics different from those of other GKAs. Herein, we report the
discovery of a new type of hepatoselective GKA with a unique structure, robust GK
3

ACCEPTED MANUSCRIPT
activation potency, and benign safety profile. In addition, a favorable distribution
pattern in the pancreas and liver was observed. To the best of our knowledge, an orally
active GKA exhibiting such characteristics has not been disclosed to date.
In a previous report, the discovery of novel small molecule 7, using molecular
modeling-aided medicinal chemistry approaches, was described [24]. This compound
has a characteristic structure with an amide portion connecting the phenyl thiophene
ring and cyclopentathiazole. GKAs with this type of structure are poorly represented in
the literature. Therefore, compound 7 was selected as the lead compound for our
discovery series.
2
. Results and discussion
. 1. Chemistry
2
The compounds prepared for this study are shown in Tables 1–6. Syntheses were
carried out as outlined in Schemes 1–4. The route used for the synthesis of compounds
(16, 18–21) is described in Scheme 1. Commercially available thiophene-3-carboxylic
acid 8 was converted to thiophene methyl ester 10 by methylation of the 2-position of
thiophene with methyl iodide and esterification under acidic conditions. Treatment of
thiophene 10 with N-bromosuccinimide (NBS) and subsequent Suzuki-Miyaura
cross-coupling with 2-formylphenylboronic acid gave aldehyde 12. Aldehyde 12 was
converted to carboxylic acid 13 by pinnick oxidation, and subsequent amide coupling of
acid 13 with 2-aminoethanol provided ester 14. Hydrolysis of ester 14 under alkaline
conditions
,6-dihydro-4H-cyclopenta[d]thiazol-2-amine afforded alcohol 16. Alcohol 16 was
converted to iodide 17 via Appel reaction using I and PPh , followed by S 2 reaction
gave
carboxylic
acid
15.
Amidation
of
15
with
5
2
3
N
of 17 with NaCN to afford nitrile 19. Nitrile 19 was converted to tetrazole 20 using
Bu SnN and to amide 21 by hydrolysis of 19 using a platinum-containing catalyst
3
3
prepared as previously described [25]. The reaction of iodine 17 with dimethylamine
gave amine 18.
4

ACCEPTED MANUSCRIPT
Scheme 1. Synthesis of phenyl thiophenyl derivatives; Reagents and conditions: (a) (i)
diisopropylamine, n-BuLi, THF, 0 °C, (ii) MeI, −60 °C to rt, 85%; (b) SOCl , MeOH,
2
reflux, 96%; (c) NBS, DMF, rt, 96%; (d) 2-formylphenylboronic acid, Pd(PPh ) ,
3
4
K CO , DME, reflux, 74%; (e) NaClO , Na HPO , 2-methyl-2-butene, t-BuOH, H O, rt,
2
3
2
2
4
2
6
8%; (f) 2-aminoethanol hydrochloride, EDCI, HOBt, DIPEA, DMF, rt, 96%; (g) 5 N
aqueous NaOH, THF, MeOH, 50 °C, 41%; (h)
5
,6-dihydro-4H-cyclopenta[d]thiazol-2-amine, EDCI, HOBt, DIPEA, DMF, 50 °C,
3
9%; (j) I , PPh , imidazole, THF, 0 °C, 86%; (k) NaCN, DMSO, rt, 21%; (l) Bu SnN ,
2
3
3
3
toluene, 130 °C, 48%; (m) [PtH(PMe OH)(PMe O) H], EtOH, H O, reflux, 55%; (n)
2
2
2
2
Me NH, THF, rt, 72%.
2
5

ACCEPTED MANUSCRIPT
Scheme 2 shows the preparation of compounds 26a, 26b, 29a, and 29b. Ketones 22a
and 22b were prepared from ester 11 in two steps: a Grignard exchange reaction of 11
with i-PrMgBr, and reaction of the resulting Grignard reagent with N-Boc-protected
cyclic amide. Intramolecular reductive amino alkylation of ketones 22a and 22b gave
pyrrolidine 23a and piperidine 23b respectively. After acetylation of amine 23a and 23b,
amide 24a and 24b were hydrolyzed under alkaline conditions followed by coupling
with 5,6-dihydro-4H-cyclopenta[d]thiazol-2-amine to give 26a and 26b. A
Cu-catalyzed coupling reaction of 11 with 2-pyrrolidone and 2-piperidone afforded
esters 27a and 27b [26]. Hydrolysis of esters 27a and 27b and subsequent amide
coupling with 5,6-dihydro-4H-cyclopenta[d]thiazol-2-amine gave 29a and 29b,
respectively.
6

ACCEPTED MANUSCRIPT
Scheme 2. Synthesis of heterocycloalkyl thiophenyl derivatives; Reagents and
conditions: (a) (i) i-PrMgBr, THF, −40 °C, (ii) 1-Boc-2-pyrrolidone or
1
-Boc-2-piperidone, −40 °C to rt, 51–62%; (b) (i) TFA, CH Cl , 0 °C to rt, (ii)
2 2
NaBH CN, conc. HCl, i-PrOH, 0 °C, 63–70%; (c) acetyl chloride, pyridine, CHCl ,
3
3
0
°C to rt, 85–98%; (d) LiOH·H O, MeOH, H O, 50 °C, 80–98%; (e)
2 2
5
,6-dihydro-4H-cyclopenta[d]thiazol-2-amine, EDCI, HOBt, DIPEA, DMF, rt,
1–43%; (f) 2-pyrrolidone or 2-piperidone, CuI, N,N’-dimethylethylenediamine, K CO ,
2
2
3
toluene, reflux, 17–33%; (g) LiOH·H O, THF, H O, 50 °C, 80–87%.
2
2
The routes for the synthesis of compounds 33a–33c, 34d, 34e, 38, and 42 are shown in
Scheme 3. Compounds 33a–33c, 34d, and 34e were synthesized from ester 23a
obtained in Scheme 2. Treatment of 23a with (Boc) O in the presence of triethylamine
2
in CH Cl followed by hydrolysis of the ester group by aqueous LiOH gave
2
2
Boc-carboxylic acid 30. Next, amidation of 30 with a corresponding thiazole amine
afforded amides 31a–31c. Removal of the Boc group of 31a–31c using trifluoroacetic
acid followed by coupling with the corresponding carboxylic acid afforded 34a–34e.
Deprotection of the Boc group of 34a–34c under acidic conditions gave 33a–33c. To
investigate the most efficient chiral-resolving agent for optical resolution of racemic
amine 23a, commercially available acidic resolving agents were used. Optical resolution
of racemic ester 23a was accomplished through diastereomeric salt formation with
(−)-dibenzoyl-L-tartaric acid monohydrate (96.6% de), followed by salt decomposition,
Boc protection and hydrolysis, to give 36. Next, (+)-enriched amine 23a was recovered
as a condensate by evaporating the methanolic mother liquor of the first resolution
followed by salt decomposition through treatment with saturated aqueous NaHCO3.
Optical resolution of liberated (+)-enriched amine 23a was accomplished through
diastereomeric salt formation with (+)-dibenzoyl-D-tartaric acid monohydrate.
Compounds 37 and 41 were synthesized in a similar manner to that shown for the
synthesis of 34e by the reaction of chiral carboxylic acids 36 and 40. Treatment of 37
and 41 with 4 N HCl in 1,4-dioxane afforded hydrochloride salts 38 and 42,
respectively. The crystal structure of the resolved tartaric acid salt 35 was determined by
X-ray crystallographic analysis. As a result, the absolute configuration at the chiral
center in the pyrrolidine ring of 35 was confirmed to be the (S)-configuration (see
Supplementary data).
7

ACCEPTED MANUSCRIPT
Scheme 3. Synthesis of compound 26a analogs; Reagents and conditions: (a) (i)
(
Boc) O, Et N, CH Cl , rt, (ii) LiOH·H O, MeOH, H O, 50 °C, 96% (2 steps); (b)
2 3 2 2 2 2
corresponding thiazoleamine, EDCI, HOBt, DIPEA, DMF, 50 °C, 48–88%; (c) TFA,
CH Cl , 0 °C to rt, 27–97%; (d) corresponding carboxylic acid, EDCI, HOBt, DIPEA,
2
2
DMF, rt, 27–95%; (e) (−)-dibenzoyl-L-tartaric acid monohydrate, MeOH, rt, 48% in
theory; (f) (i) saturated aqueous NaHCO , CHCl , rt; (ii) (Boc) O, Et N, CH Cl , rt; (iii)
3
3
2
3
2
2
LiOH·H O, MeOH, H O, 50 °C, 87–94%; (g) (i) 2-amino-5-chlorothiazole
2
2
hydrochloride, EDCI, HOBt, DIPEA, DMF, 50 °C, (ii) TFA, CH Cl , 0 °C to rt, (iii)
2
2
8

ACCEPTED MANUSCRIPT
N,N-dimethylglycine, EDCI, HOBt, DIPEA, DMF, 49–58% (3 steps); (h) 4 N HCl in
1
,4-dioxane, CH Cl , 0 °C to rt, 70–97%; (j) saturated aqueous NaHCO , CHCl , rt;
2 2 3 3
a
(k) (+)-dibenzoyl-D-tartaric acid monohydrate, MeOH, rt, 45% in theory. (+)-enriched
enantiomer.
The route for the synthesis of compounds 47a–48b, 51, 53a–53e, and 58a–59 is shown
in Scheme 4. Compounds 47a–48b and 51 were synthesized from amine 31c.
Condensation of amine 31c with the corresponding carboxylic acid, followed by
deprotection of the Boc group, gave amines 46a and 46b. Amines 46a and 46b were
treated with mesyl chloride to give mesylated compounds 47a and 47b. Furthermore,
amines 46a and 46b were treated with acetyl chloride to give acetylated compounds 48a
and 48b, respectively. Chloroacetylation of amine 31c, followed by treatment with
glycine tert-butyl ester and subsequent removal of the tert-butyl ester using
hydrochloric acid, afforded carboxylic acid 51. Fluorothiazole derivative 46c was
synthesized in a similar manner as that for the chlorothiazole counterpart 46b.
Treatment of amines 46b and 46c with bromo ester derivatives or tert-butyl acrylate
followed by deprotection of the ester group gave carboxylic acids 53a–53e. Chiral
isomers 58a, 58b, and 59 were synthesized from chiral acid 36 obtained in Scheme 3.
Carboxylic acid 36 was converted to amines 56a–56c in a four-step sequence:
condensation with 2-amino-5-fluorothiazole, deprotection of the Boc group, and
condensation with the corresponding N-Boc amino acid, followed by deprotection of the
Boc group. Compounds 58a, 58b, and 59·HCl were synthesized in a similar manner to
that shown for the synthesis of 53a–53e from 46b and 46c by the reaction of amines
5
6a–56c. Finally, we sought to establish a method for isolation of the zwitterion of
9·HCl. As a result, zwitterion product 59 was crystallized directly from the reaction
5
mixture with 84% yield after pH adjustment to the isoelectric point of pH 5.8 with 1 N
aqueous NaOH.
9

ACCEPTED MANUSCRIPT
1
0

ACCEPTED MANUSCRIPT
Scheme 4. Synthesis of 47a–48b, 51, 53a–53e, and 58a–59; Reagents and conditions:
(
a) (i) (COCl) , toluene, 0 °C, (ii) 2-amino-5-fluorothiazole hydrochloride,
2
N,N-diethylaniline, toluene, rt, 76–78% (2 steps); (b) TFA, CH Cl , 0 °C to rt, 79–99%;
2
2
(c) glycine derivatives, EDCI, HOBt, DIPEA, DMF, rt, 90–99%; (d) mesyl chloride,
pyridine, CH Cl , 0 °C to rt, 69–81%; (e) acetyl chloride, pyridine, CH Cl , 0 °C to rt,
2
2
2
2
7
4–85%; (f) chloroacetyl chloride, 2,6-di-tert-butylpyridine, CH Cl , 0 °C to rt, 83%;
2 2
(g) glycine tert-butyl ester hydrochloride, DIPEA, DMF, 50 °C, 68%; (h) 4 N HCl in
1
1
,4-dioxane, CH Cl , 0 °C to rt, or 1 N aqueous NaOH, MeOH, rt, then 4 N HCl in
2 2
,4-dioxane, 0 °C to rt, 68–99%; (j) tert-butyl ester derivatives, DIPEA, DMF or
tert-butyl acrylate, benzyltrimethylammonium hydroxide in MeOH, DMF, rt to 60 °C,
2–77%; (k) corresponding Boc-amino acid, EDCI, HOBt, DIPEA, DMF, rt, 82–99%;
l) 4 N HCl in 1,4-dioxane, CH Cl or 4 N HCl in AcOEt, CH Cl , 0 °C to rt, 71–99%;
2
(
(
2
2
2
2
m) 1 N aqueous NaOH, MeOH, H O, rt, 84%.
2
11

ACCEPTED MANUSCRIPT
2
. 2.
Identification of a pyrrolidine-thiophene-based glucokinase activator
Since compound 7 had potent glucokinase activity in 5 mM glucose, its
glucose-lowering effects were assessed by OGTT in mice. Unfortunately, compound 7
did not exhibit glucose-lowering effects in our OGTT study at 30 mg/kg (p.o.), and
had poor bioavailability following oral dosing (Cmax < 1 µM, data not shown). To
convert 7 to the orally bioavailable GKA, we considered two strategies: the first was to
significantly improve the GK activation potency of 7, and the second was to improve
the physicochemical properties of 7 affecting bioavailability. In our previous work, the
nitrogen atom in the thiazole ring, amide NH, and carboxyl oxygen of terminal
carbamoyl group of 7 were expected to form hydrogen bonds with amino acid residues
in the GK active site. Given the existing interactions of compound 7 with the GK active
site, we chose a strategy to improve the properties of 7 while keeping maintaining these
interactions (Figure 2).
Figure 2. First design of hydrophilic analogs from lead compound 7.
The results of our mouse OGTT study demonstrated that the clearance rate of 7 was not
rapid, suggesting that metabolic stability of 7 was not the cause of poor bioavailability.
Furthermore, the membrane permeability of 7 was considered adequate (Log D value at
pH 7.4 was 3.23) [27]. Therefore, we hypothesized that improvements in solubility
would lead to increased plasma exposure. From our previous modeling-aided studies,
the cavity of the GKA binding site in which the carbamoyl group of 7 binds is
composed of hydrophilic amino acid residues, and accommodates the introduction of
functional groups. Therefore, we introduced various polar groups with hydrogen bond
donors and acceptors to this position. Table 1 details the structure–activity relationship
(
SAR) of polar groups at the 2-position of the phenyl ring, and the effect of conversion
1
2

ACCEPTED MANUSCRIPT
of the phenyl ring in 7 into a saturated hetero ring. GK activation potency of compounds
were determined in a biological assay monitoring the rate of glucose 6-phosphate
formation using G6PDH/NADP coupling. Compounds were characterized for relative
GK activation using compound 1 (RO0281675) as a reference, and to determine batch
variability. Aqueous solubilities were determined experimentally using a nephelometric
assay in phosphate buffered saline (pH 7.4). All compounds with a basic amino moiety
were used in the free form unless otherwise stated. As shown in Table 1, hydroxyl ethyl
derivative 16 and dimethylamino derivative 18 preserved GK activation potency.
Unfortunately, 16 and 18 did not show an improvement in solubility relative to that of 7.
Tetrazole derivative 20 and amide derivative 21 showed improvement in solubility, but
had decreased GK activation potency compared with that of 7. Although the
introduction of polar groups in this part of the molecule was well tolerated, a promising
compound with good solubility and GK activation potency was not identified. Therefore,
we sought an alternative scaffold to replace the phenyl core. Lovering and colleagues
found previously that carbon bond saturation correlates with solubility [28], we
therefore designed molecules with a saturated hetero ring adjacent to the thiophene ring
instead of the phenyl group (26a, 26b, 29a, and 29b). Although the piperidine and
piperidone derivatives 26b and 29b exhibited no improvement in solubility relative to
that of 7, the pyrrolidine and pyrrolidone derivatives 26a and 29a showed more than a
two-fold improvement in solubility compared with that of 7. These differences were
thought to derive from differences in hydrophilicity and molecular weight. Pyrrolidine
derivatives have a lower Log D value at pH 7.4 and a lower molecular weight than
piperidine derivatives. Since 26a was approximately four-fold more soluble than 7 and
maintained GK activation potency to some extent, we focused our attention on the
synthesis of pyrrolidine derivatives.
1
3

ACCEPTED MANUSCRIPT
Table 1.
SAR for analogs of compound 7
a
d
Activation at 5 mM glucose
Solubility
Compound
R
b
c
Fold
Relative
(µM; pH 7.4)
7
6.23 ± 0.03
3.89 ± 0.08
4.31 ± 0.04
0.84 ± 0.01
2.56 ± 0.03
2.53 ± 0.02
1.13 ± 0.00
2.41 ± 0.01
0.699 ± 0.004
0.373 ± 0.011
0.428 ± 0.005
−0.019 ± 0.001
0.187 ± 0.004
0.187 ± 0.002
0.017 ± 0.001
0.168 ± 0.002
0.235 ± 0.004
1.95
16
18
20
21
0.98
1.95
31.3
3.91
7.81
0.98
3.91
0.98
2
6a
2
6b
2
9a
2
9b
2.96 ± 0.04
a
b
Values are the mean ± SEM for n = 3. Glucokinase activity at 10 µM relative to
c
DMSO control. Glucokinase activity at 10 µM relative to 10 µM of 1 as a standard
d
activator (DMSO:0; 1:1). Aqueous solubility was determined experimentally using a
nephelometric assay in PBS (pH 7.4).
1
4

ACCEPTED MANUSCRIPT
The optimization of the pyrrolidine derivative 26a is shown in Table 2. The amino ethyl
derivative 33a was more potent than 26a, although the solubility of 33a was low.
Considering that the hydrophobicity of cyclopentathiazole moiety was extremely high,
the non-substituted thiazole analog 33b was prepared. The GK-activation potency of
3
3b decreased, although the solubility of 33b improved. Next, the aminomethyl
derivative 33c and the dimethylaminomethyl derivative 34d were synthesized. These
derivatives were comparable to 26a in solubility, with good GK activation potency. The
Cl derivative 34e was also evaluated. Introduction of a Cl group at the 5-position of the
thiazole ring resulted in decreased solubility compared with that of 34d, but 34e showed
strong GK activation potency.
1
5

ACCEPTED MANUSCRIPT
Table 2.
SAR for analogs of compound 26a
a
d
Activation at 5 mM glucose
Solubility
1
2
Compound
R
R
b
c
Fold
Relative
(µM; pH 7.4)
2
3
6a
3a
2.53 ± 0.02
3.46 ± 0.02
2.86 ± 0.02
3.67 ± 0.01
4.70 ± 0.04
0.187 ± 0.002
0.274 ± 0.002
0.175 ± 0.002
0.293 ± 0.001
0.380 ± 0.004
0.598 ± 0.005
7.81
0.98
15.6
7.81
7.81
3.91
3
3b
3
3c
N
N
3
4d
O
O
3
4e
6.81 ± 0.05
a
b
Values are the mean ± SEM for n = 3. Glucokinase activity at 10 µM relative to
c
DMSO control. Glucokinase activity at 10 µM relative to 10 µM of 1 as a standard
d
activator (DMSO:0; 1:1). Aqueous solubility was determined experimentally using a
nephelometric assay in PBS (pH 7.4).
1
6

ACCEPTED MANUSCRIPT
Thus, with a promising compound showing good GK activation potency and improved
solubility identified, chiral isomers of 34e were synthesized and assessed for GK
activation potency to confirm the effect of the pyrrolidine chiral center on GK activity
(Table 3). Both isomers 38 and 42 showed GK activation potency, although 38 showed
stronger potency than 42. Therefore, the biologically active form was confirmed to be
the (S)-configuration. Next, the effect of the compounds on insulin secretion was
evaluated in MIN-6 cells, a mouse pancreatic beta-cell line [29]. Compound 38 showed
a good insulin secretory effect in this cell line.
1
7

ACCEPTED MANUSCRIPT
Table 3.
The effect of the chiral center of compound 34e
a
e
Activation at 5 mM glucose
Insulin secretion Solubility
Compound
R
from MIN-6
(µM; pH
b
c
Fold
Relative
a,d
cells (relative)
7.4)
3
4e
6.81 ± 0.05 0.598 ± 0.005
1.62 ± 0.27
1.61 ± 0.21
1.15 ± 0.06
3.91
7.81
1.95
f
3
8
10.4 ± 0.1
0.650 ± 0.010
f
4
2
4.99 ± 0.12 0.381 ± 0.011
a
b
Values are the mean ± SEM for n = 3. Glucokinase activity at 10 µM relative to
c
DMSO control. Glucokinase activity at 10 µM relative to 10 µM of 1 as a standard
d
activator (DMSO:0; 1:1). Insulin secretion activity at 10 µM relative to 10 µM of 1 as a
e
standard activator (DMSO:0; 1:1).
Aqueous solubility was determined
f
experimentally using a nephelometric assay in PBS (pH 7.4). Hydrochloride salt.
Given the promising in vitro GK activation potency and effect on insulin secretion, 38
was evaluated for the ability to reduce glucose levels in rodents. The effect of 38 on
blood glucose was tested using OGTT in non-diabetic (C57BL/6j) mice. Animals were
subjected to an overnight fast. Compound 38 was orally administered 15 min prior to
the administration of 2 g/kg of glucose. Blood glucose was measured at 0, 20, 40, 60, 90,
and 120 min after glucose administration. Compound 38 decreased blood glucose levels
dose-dependently in C57BL/6j mice at 10 and 30 mg/kg (Figure 3). As expected,
1
8

ACCEPTED MANUSCRIPT
compound 38, which demonstrated good solubility, exhibited high bioavailability in this
study (Cmax > 9 µM at 30 mg/kg).
1
9

ACCEPTED MANUSCRIPT
Figure 3. Oral glucose tolerance test (OGTT) of 38 in male C57BL/6j mice. Effect on
(a) blood glucose level and (b) incremental blood glucose AUC during OGTT. Data are
expressed as (a) mean ± SEM (n = 8) and (b) mean + SEM (n = 8). The data was
analyzed using Dunnett's multiple comparison test. *, ***: p < 0.05, p < 0.001 vs.
vehicle control.
2
0

ACCEPTED MANUSCRIPT
For progression of compound 38 as a preclinical development candidate, the
cardiovascular safety profile was evaluated. Unfortunately, 38 appeared to bind to the
human ether-a-go-go related gene (hERG) channel with high affinity (IC < 0.3 µM).
5
0
Inhibition of the α-subunit of IKr channels, which is encoded by hERG, can cause
delayed ventricular cell repolarization. This phenomenon is observed by electro
cardiogram as a prolongation of the QT interval, and represents a major hurdle in drug
development [30]. As it was crucial that our GK activation candidate compounds did
not exhibit this property, it was necessary to reduce the hERG binding ability of 38.
2
. 3. Reduction of hERG inhibitory activity to discover 59
To reduce the hERG inhibitory activity of 38, an extensive survey of the medicinal
chemistry literature was conducted. Some effective strategies to minimize hERG/IKr
activity have been previously reported [31]. Among these, we selected two modification
strategies: the attenuation of pKa and the formation of zwitterions (Figure 4).
2
1

ACCEPTED MANUSCRIPT
Figure 4. Strategy to reduce the inhibitory activity of hERG
Table 4 shows the result of the modification of 38. First, we reduced the terminal
nitrogen pKa. It is reported that π-cation interactions between aromatic residues within
the cavity of the hERG channel and the basic amine moiety in the ligand play a role in
binding affinity. Therefore, reducing the protonation of molecules at physiological pH
by lowering the pKa of a basic nitrogen could disrupt any putative π-cation interactions
between ligand and hERG [32, 33]. To confirm the effect of this strategy, compounds
whose nitrogen atom’s basicity was weakened by a sulfonyl group (47a and 47b) or
acetyl group (48a and 48b) were assessed. Within this series, N-methyl derivatives (47b
and 48b) showed a significant drop in GK potency, but N-H derivatives (47a and 48a)
showed moderate GK activation potency. Furthermore, these N-H derivatives showed a
marked decrease in hERG binding affinity.
Next, we turned our attention to the introduction of a zwitterionic characteristic to the
molecule. The zwitterion approach has been widely employed for the attenuation of
hERG activity [34, 35]. Introduction of a carboxylic acid moiety into our potential
template led to the synthesis of the zwitterionic compounds 51 and 53a. These
2
2

ACCEPTED MANUSCRIPT
compounds showed a substantial decrease in hERG binding affinity. Among these, 53a
showed a strong GK activation potency, comparable with 38.
Table 4.
SAR for analogs of compound 38
S
N
O
R
O
HN
S
N
Cl
a
Activation at 5 mM glucose
Insulin
hERG
secretion from
Compound
R
inhibition
b
c
Fold
Relative
MIN-6 cells
a,e
(%)
a,d
(relative)
f ,g
3
8
10.4 ± 0.1
6.21 ± 0.04
3.24 ± 0.03
5.32 ± 0.08
2.57 ± 0.04
4.75 ± 0.02
8.35 ± 0.08
0.650 ± 0.010
0.460 ± 0.004
0.168 ± 0.002
0.395 ± 0.007
0.118 ± 0.003
0.438 ± 0.003
1.61 ± 0.21
0.690 ± 0.226
-
101 ± 1
16 ± 1
-
4
7a
4
7b
4
8a
1.17 ± 0.16
-
3 ± 1
-
4
8b
g
5
1
0.994 ± 0.043
1.12 ± 0.22
28 ± 1
11 ± 2
g
5
3a
0.718 ± 0.008
a
b
Values are the mean ± SEM for n = 3. Glucokinase activity at 10 µM relative to
c
DMSO control. Glucokinase activity at 10 µM relative to 10 µM of 1 as a standard
d
activator (DMSO:0; 1:1). Insulin secretion activity at 10 µM relative to 10 µM of 1 as a
2
3

ACCEPTED MANUSCRIPT
e
standard activator (DMSO:0; 1:1). Percentage of inhibition of hERG potassium channel
f
g
at 30 µM. Chiral isomer. Hydrochloride salt.
Based on these findings, compound 53a was chosen as the starting point for further
optimization (Table 5). We prepared a series of compounds with a 5-F substituted
thiazole ring and various tether lengths of the carboxylic acid. 5-F-substituted thiazole
analog 53b was as potent as its 5-Cl-substituted thiazole counterpart 53a and showed no
hERG binding affinity at 30 µM. However, 53b showed low insulin secretion in MIN-6
cells. The compound whose tether part was extended from ethanoic acid to propanoic
acid (53c) showed moderate GK activation potency, good MIN6 insulin secretion, and
low hERG binding activity. A fine balance was observed between these parameters
Compounds with a longer tether than 53c (53d and 53e) had favorable GK activation
potency relative to 53c, but unfortunately had relatively strong hERG binding affinity.
Compounds 53a and 53c demonstrated favorable GK activation potency, insulin
secretion, and low hERG binding affinity. As a result, the in vivo efficacy of 53a and
5
3c was evaluated by OGTT in normal C57BL/6j mice. Compounds were dosed orally
at 30 mg/kg, 30 min prior to glucose loading. Unfortunately, only 53c showed a
significant glucose lowering effect. Incremental blood glucose AUC0-120 min values of
5
3a and 53c relative to those of vehicle control were 82% and 47%, respectively. To
investigate differences in the hypoglycemic effects of these compounds, plasma
concentrations were determined following oral administration. Compound 53c showed a
more than six-fold higher plasma concentration than 53a [36]. From these results, 53c
was considered the most promising compound in the series. The chiral isomer of this
compound was synthesized and evaluated in vitro.
2
4

ACCEPTED MANUSCRIPT
Table 5.
SAR for analogs of compound 53a
b
Activation at 5 mM glucose
Insulin
secretion
hERG
a
Compound
R
X
from MIN-6 inhibition
c
d
Fold
Relative
b,f
cells
(%)
b,e
(
relative)
5
3a
Cl
F
-CH
2
2
-
-
8.35 ± 0.08 0.718 ± 0.008 1.12 ± 0.22
6.92 ± 0.02 0.710 ± 0.003 0.379 ± 0.091
4.89 ± 0.04 0.466 ± 0.005 1.19 ± 0.06
6.86 ± 0.03 0.610 ± 0.003 1.21 ± 0.17
11 ± 2
− 6 ± 2
17 ± 3
78 ± 2
77 ± 1
5
3b
-CH
5
3c
3d
3e
F
-(CH
2
)
2
3
4
-
-
-
5
F
-(CH
2
)
)
5
F
-(CH
2
6.95 ± 0.01 0.619 ± 0.001 0.807 ± 0.119
a
b
c
Hydrochloride salt. Values are the mean ± SEM for n = 3. Glucokinase activity at
d
1
0 µM relative to DMSO control. Glucokinase activity at 10 µM relative to 10 µM of 1
e
as a standard activator (DMSO:0; 1:1). Insulin secretion activity at 10 µM relative to
f
1
0 µM of 1 as a standard activator (DMSO:0; 1:1). Percentage of inhibition of hERG
potassium channel at 30 µM.
2
5

ACCEPTED MANUSCRIPT
As expected, the chiral isomer of 53c (58a) showed favorable GK activation potency
and promoted insulin secretion in MIN6 cells (Table 6). Unexpectedly, this compound
showed a stronger hERG binding affinity (49% at 30 µM) relative to racemic compound
5
3c. This suggested that the enantiomer of 58a had a lower hERG binding affinity
relative to 53c, and that the carboxylic acid series proved to be highly sensitive to minor
structural changes. Therefore, minor modification of 58a could decrease hERG binding
affinity. Thus, we designed and synthesized two types of ethanoic acid derivatives, 58b
and 59·HCl. As expected, these derivatives showed good GK activation potency with
low hERG binding. 59·HCl demonstrated higher insulin secretion relative to 58b, and
was selected for further study.
2
6

ACCEPTED MANUSCRIPT
Table 6.
GK activity, insulin secretion, and hERG binding of chiral analogs
b
Activation at 5 mM glucose
Insulin
secretion
hERG
a
Compound
R
from MIN-6 inhibition
c
d
Fold
Relative
b,f
cells
(%)
b,e
(
relative)
5
8a
8b
9·HCl
7.26 ± 0.05 0.686 ± 0.005 0.930 ± 0.017
8.80 ± 0.15 0.832 ± 0.016 0.436 ± 0.039
49 ± 1
4 ± 0
5
5
8.60 ± 0.10 0.870 ± 0.011 0.830 ± 0.165
− 1 ± 1
a
b
c
Hydrochloride salt. Values are the mean ± SEM for n = 3. Glucokinase activity at
d
1
0 µM relative to DMSO control. Glucokinase activity at 10 µM relative to 10 µM of 1
e
as a standard activator (DMSO:0; 1:1). Insulin secretion activity at 10 µM relative to
f
1
0 µM of 1 as a standard activator (DMSO:0; 1:1). Percentage of hERG potassium
channel inhibition at 30 µM.
Based on these encouraging results, 59·HCl was evaluated for its glucose-lowering
effect in rodents via an OGTT study in normal C57BL/6J and diabetic ob/ob mice
5
9·HCl induced a significant reduction in blood glucose excursion in a dose-dependent
manner in both C57BL/6J and ob/ob mice, and showed a satisfactory PK profile
Figures 5 and 6; Table 7). Moreover, in a 4 week repeated administration study,
9·HCl decreased blood glucose levels on day 0, 14, and 28 in ob/ob mice (Figure 7).
(
5
Compound 59 also showed low toxicity in a cell toxicity assay using Chinese hamster
ovary cells and rat hepatocyte cells (CC50 >100 µM), and showed no significant
2
7

ACCEPTED MANUSCRIPT
inhibition of the hERG channel in patch clamp experiments using human embryo
kidney 293 cells (IC > 300 µM) (Table 8).
5
0
2
8