Discovery and preclinical development of AR453588 as an anti-diabetic glucokinase activator

Article abstract of DOI:10.1016/j.bmc.2019.115232

Glucose flux through glucokinase (GK) controls insulin release from the pancreas in response to high levels of glucose. Flux through GK is also responsible for reducing hepatic glucose output. Since many individuals with type 2 diabetes appear to have an inadequacy or defect in one or both of these processes, identifying compounds that can activate GK could provide a therapeutic benefit. Herein we report the further structure activity studies of a novel series of glucokinase activators (GKA). These studies led to the identification of pyridine 72 as a potent GKA that lowered post-prandial glucose in normal C57BL/6J mice, and after 14d dosing in ob/ob mice.

Full text of DOI:10.1016/j.bmc.2019.115232

Journal Pre-proofs
Discovery and Preclinical Development of AR453588 as an Anti-diabetic
Glucokinase Activator
Ronald J. Hinklin, Brian R. Baer, Steven A. Boyd, Mark D. Chicarelli, Kevin
R. Condroski, Walter E. DeWolf Jr., John Fischer, Michele Frank, Gary P.
Hingorani, Patrice A. Lee, Nickolas A. Neitzel, Scott A. Pratt, Ajay Singh,
Francis X. Sullivan, Timothy Turner, Walter C. Voegtli, Eli M. Wallace,
Lance Williams, Thomas D. Aicher
PII:
S0968-0896(19)31029-6
https://doi.org/10.1016/j.bmc.2019.115232
BMC 115232
DOI:
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
19 June 2019
14 November 2019
20 November 2019
Please cite this article as: R.J. Hinklin, B.R. Baer, S.A. Boyd, M.D. Chicarelli, K.R. Condroski, W.E. DeWolf Jr.,
J. Fischer, M. Frank, G.P. Hingorani, P.A. Lee, N.A. Neitzel, S.A. Pratt, A. Singh, F.X. Sullivan, T. Turner, W.C.
Voegtli, E.M. Wallace, L. Williams, T.D. Aicher, Discovery and Preclinical Development of AR453588 as an
Anti-diabetic Glucokinase Activator, Bioorganic & Medicinal Chemistry (2019), doi: https://doi.org/10.1016/
j.bmc.2019.115232
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. 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.
© 2019 Published by Elsevier Ltd.

Discovery and Preclinical Development of
AR453588 as an Anti-diabetic Glucokinase
Activator
Ronald J. Hinklin,* Brian R. Baer, Steven A. Boyd, Mark D. Chicarelli, Kevin R. Condroski,
Walter E. DeWolf, Jr., John Fischer, Michele Frank, Gary P. Hingorani, Patrice A. Lee,
Nickolas A. Neitzel, Scott A. Pratt, Ajay Singh, Francis X. Sullivan, Timothy Turner, Walter C.
Voegtli, Eli M. Wallace, Lance Williams and Thomas D. Aicher
Array BioPharma Inc., 3200 Walnut St., Boulder, CO 80301.
ABSTRACT
Glucose flux through glucokinase (GK) controls insulin release from the pancreas in
response to high levels of glucose. Flux through GK is also responsible for reducing hepatic
glucose output. Since many individuals with type 2 diabetes appear to have an inadequacy or
defect in one or both of these processes, identifying compounds that can activate GK could
provide a therapeutic benefit. Herein we report the further structure activity studies of a novel
series of glucokinase activators (GKA). These studies led to the identification of pyridine 72 as a
potent GKA that lowered post-prandial glucose in normal C57BL/6J mice, and after 14d dosing
in ob/ob mice.
1

Graphical abstract
2

1. Introduction
Glucokinase (GK)1 is a member of the hexokinase family of cellular enzymes responsible for
the conversion of glucose to glucose-6-phosphate, the first step in glucose utilization. Its
expression is restricted primarily to liver, pancreatic -cells, enteroendocrine cells and the
hypothalamus.2,3 GK is the rate-limiting enzyme for glucose metabolism, which controls
glucose-stimulated insulin secretion in the pancreas.4 GK interconverts between inactive and
active conformations in response to glucose concentration.5 It effectively acts as the “glucose
sensor” due to its substrate concentration at half-maximal velocity (S0.5) of 7.5 mM glucose
being in the physiological range for blood glucose and its positive cooperativity for glucose (Hill
slope h = 1.7).6,7
Glucokinase activators (GKAs) control the blood glucose concentration by enhancing the
ability of pancreatic -cells to sense glucose and increase insulin secretion in a glucose-
dependent manner. In addition, GKAs act in the liver to improve glucose disposal and decrease
hepatic glucose output.14 Each GKA has the potential to induce a unique conformation of the
allosteric site of the active form of glucokinase, and the resulting complexes may exhibit unique
enzymatic kinetic profiles. Scientists at Hoffman la Roche were the first to describe heterocyclic
amide-based small molecule activators of glucokinase,8 with subsequent disclosures by
AstraZeneca,9,10,11 Banyu (Merck),12 and other companies13,14,15,16 each containing variations on
the heterocyclic amide core.
In recent years, a number of GKAs have progressed into clinical trials.17,18 Reports from the
trials have demonstrated the glucose lowering effect of activating GK. Depending on the
3

compound and dose, hypoglycemia is observed as a common adverse event. In clinical studies
that last months, rather than weeks, the glucose lowering effect has not been durable. By
activating GK at low glucose concentrations, insulin is released too often and further exacerbates
peripheral insulin resistance. Identifying a compound with an improved kinetic profile should
prevent insulin secretion from the -cell at low blood glucose concentrations. The hypothesis is
that better management of kinetics would both prevent hypoglycemia and lead to a more durable
effect.
In a previous report, we disclosed our initial investigations into a new class of GKAs.19
Utilizing structure-based design, a novel recognition motif was proposed and validated with
pyridylthiazole 1. (Figure 1) Subsequent structure-guided optimization resulted in the
identification of 2 as an early in vivo proof-of-concept compound. The results of this early
investigation served as the basis for a broad investigation into heteroaryl-aminopyridines as new
cores for GKAs. Herein, we describe our further advances in the program to afford a compound
with improved potency, pharmacokinetics and in vivo efficacy.
4

N
N
O
O
S
O
N
O
O
N
N
N
N
O
N
HO
N
O
N
H
H
Cl
N
H
O
O
N
N
S
S
O
O
O
O
N
Piragliatin
Roche
PF-04971729
Pfizer
MK-0941
Merck
S
N
N
N
S
N
N
N
N
H
H
O
O
1
2
Figure 1. Representative glucokinase activators previously in clinical development and initial hit
compound 1 and lead compound 2.
2. Chemistry
We designed methods to access 3-aryloxy-2-aminopyridines in an effort to elaborate on the
scaffold of our initial hit compound 1. The chemistry to access the initial analogs 6 and 7
utilized an SNAr reaction of 3-hydroxy-2-chloropyridine, as shown in Scheme 1. Chloropyridine
3 was reacted with 2- or 4-fluorobenzonitrile to afford 4 and 5, respectively. The resulting
chloropyridines 4 and 5 were reacted with 4-methyl-2-aminothiazole under palladium-catalyzed
conditions to afford target aminothiazoles 6 and 7.20
N
S
N
N
a
b
N
N
H
Cl
Cl
O
O
OH
6-7
3
4-5
CN
CN
5

Scheme 1. Reagents and conditions: (a) K2CO3, 2- or 4-fluorobenzonitrile, DMF, 90 °C; (b) 4-
methyl-2-aminothiazole, Pd2dba3, Xantphos, K3PO4, toluene:water, 90 °C.
3-Phenoxypyridines bearing 5-substituents were synthesized via the route in Scheme 2. SNAr
reaction of 2-aminopyridin-3-ol (8) with the electron-deficient 2,4-dibromo-1-fluorobenzene
allowed for the introduction of the phenyl ether to afford 9. Catalytic hydrogenolysis of the two
remaining halogens afforded the phenyl ether 10. The amino moiety allowed selective 5-
bromination of the pyridyl ring of 10 to afford the bromide 11. Condensation of the amino
pyridine 11 with benzoylisothiocyanate afforded acylthiourea 12. Cleavage to the benzoyl group
of 12 followed by alkylation and cyclization with chloroacetone in a Hantzsch synthesis21
afforded aminothiazole 14. The 5-bromo substituent of 14 allowed the introduction of diverse
functional groups. Initial reaction of 14 with MeLi to remove the acidic NH proton, followed by
n-BuLi to facilitate a halogen-metal exchange, and finally quenching with water (to yield 15) or
a variety of disulfides afforded thioethers 17-19. Alternatively, palladium-mediated coupling of
14 with benzylmercaptan directly afforded benzyl thioether 16.22
Br
N
N
N
N
NH2
a
b
c
d
NH2
NH2
NH2
O
O
O
OH
8
9
10
11
Br
Br
Br
Br
Br
R
N
S
O
N
S
N
S
N
S
e
g or h
f
N
N
N
H
N
Ph
N
NH2
N
N
H
H
H
H
O
O
O
O
12
13
14
15-19
Scheme 2. Reagents and conditions: (a) NaH, 2,4-dibromo-1-fluorobenzene, DMF, 90 °C; (b)
H2, 10% Pd/C; (c) Br2; (d) PhC(O)NCS, THF; (e) NaOH, EtOH, 50 °C; (f) chloroacetone, Et3N,
EtOH, 70 °C; (g) i. MeLi, THF, -78 °C ii. BuLi, -78 °C iii. H2O or RS-SR; (h) BnSH, Et2NiPr,
Xantphos, Pd2dba3, dioxane, 95 °C.
6

Utilizing similar chemistry in Scheme 2, but changing the order of introduction of functional
groups, allowed the 4-substituent of the thiazole to be systematically changed via the route in
Scheme 3. In this effort the 5-(4-pyridylthio) group was employed as a substituent on the
pyridine core. Key thiourea intermediate 22 was obtained from bromide 11 by similar
transformations as described above in Scheme 2. Reaction of the thiourea with a variety of alpha-
halo ketones afforded thiazoles 23-30. Hydrolysis of the ester in 30 afforded the acid 31.
Br
S
S
N
N
N
S
O
N
N
a
b
NH2
NH2
N
H
N
Ph
H
O
O
O
11
20
21
c
S
S
N
S
N
S
R
N
N
S
d
N
N
N
NH2
H
H
O
O
23-30
22
O
O
S
N
S
N
S
e
OMe
OH
N
N
N
N
N
N
H
H
O
O
31
30
Scheme 3. Reagents and conditions: (a) i. MeLi, THF, -78 °C ii. n-BuLi, -78 °C iii. 4-pyridyl
disulfide, RT; (b) PhC(O)NCS, THF; (c) NaOH, EtOH, 50 °C; (d) X-CH2C(O)R, Et3N, EtOH,
70 °C; (e) NaOH, THF, H2O.
Replacement of the amino thiazole with an amino thiadiazole functionality was accomplished
through the facile reaction of -isothiocyano oxime O-sulfonate reagents with amines, as
depicted in Scheme 4.23 Oximes 33a-d were cleanly converted to their corresponding -chloro
7

oxime 34a-d with N-chlorosuccinimde. Addition of triethylamine to a solution of 34a-d and
mesyl chloride in ether affords methylsulfonyloxy iminoyl chlorides 35a-d. The intermediate 36
was synthesized in a similar fashion as 20. Reaction of -chloro oxime O-sulfonates 35a-d in
acetonitrile with NaNCS and pyridine at 45 °C forms an intermediate that is further reacted with
36 and cyclized at 60 °C to form the desired amino thiadiazoles 37-40. Deprotection of the Boc-
group in 40 with trifluoroacetic acid, followed by functionalization of the secondary amine
afforded 42-49.
O
c, R =
d, R =
a, R = Me
b, R =
HO
H
HO
Cl
MsO
Cl
O
c
b
a
N
N
N
H
R
R
R
R
O
NBoc
32b-d
33a-d
34a-d
35a-d
S
Br
S
N
N
S
N
N
N
e
d
R
N
N
N
H
NH2
NH2
O
O
O
Ph
Ph
Ph
11
36
37-40
S
N
S
N
f
g
NH
N
S
40
N
N
H
O
Ph
R =
NBoc
41
S
N
N
N
N
R'
N
N
H
O
Ph
42-49
Scheme 4. Reagents and conditions: (a) NH2OHHCl, Na2CO3, THF, H2O; (b) NCS, DMF; (c)
MsCl, Et3N, Et2O; (d) i. MeLi, THF, -78 °C ii. BuLi, -78 °C iii. 1,2-di(pyridin-2-yl)disulfane,
8

RT; (e) i. 35a-d, NaNCS, pyridine, acetonitrile, 45 °C ii. 36, 60 °C; (f) TFA; (g) R’X, DIEA or
R’OH, EDCI, Et3N.
We sought an efficient methodology for the installation of groups other than phenoxy at the 3-
position of the pyridine core, and developed the SNAr sequence shown in Scheme 5. An apparent
strategy to achieve this is an SNAr onto a 3-chloro-2-nitropyridyl compound. However, the nitro
moiety can compete as a leaving group with the chloride. By changing to a 3-fluoro-2-
nitropyridine, the desired fluoride displacement is the major product; however, the synthetic
route to produce the starting material required an exothermic decomposition of 2-nitropyridine-
3-diazonium tetrafluoroborate at 115 °C, which was not amenable to large scale synthesis.24
Recognizing that nitro groups could be displaced in SNAr reactions and that esters or nitriles are
potential synthetic equivalents for an amino moiety via either a Curtius or Hoffman reaction, we
investigated the 3-nitro-2-cyanopyridine as a synthetic intermediate. It was reported that
nucleophiles (NaOMe and fluoride) selectively displaced the nitro moiety to produce 3-
fluoropicolinonitrile and 3-methoxypicolinonitrile.25 Indeed, as will be seen below, diverse
hydroxyheterocycles displace the nitro moiety selectively.
For this initial SAR investigation of the range of tolerated groups at the 3-position, a set of
analogs bearing the 3-(4-piperidinyl)-thiadiazole was chosen. Selective displacement of the nitro
group from the commercially available cyanopyridine 50 with various phenols and heteroaryl
alcohols afforded pyridyl ethers 51-54.26 Subsequent displacement of the 5-bromo group with
pyridine-2-thiol afforded thioethers 55-58. Partial hydrolysis of the nitrile to the corresponding
primary amides 59-62 was accomplished with sulfuric acid under controlled conditions.
Hofmann rearrangement of the resulting amides, using bromine in aqueous base, afforded the 2-
aminopyridines 63-66. The target thiadiazoles 67-70 were prepared as delineated in Scheme 4
9

above, using 35d and sodium isothiocyanate. The piperidinyl group was further elaborated by
acidic cleavage of the Boc-group, followed by acetylation or mesylation to generate amides 71-
74, and sulfonamide 75.
Br
S
S
Br
N
N
N
N
c
a
b
d
N
N
NH2
CN
CN
CN
O
O
O
O
NO2
R1
R1
R1
50
51-54
55-58
59-62
S
S
S
N
N
S
N
N
S
N
e
f,g
R2
N
N
N
N
N
Boc
N
N
NH2
N
N
H
H
O
O
O
R1
R1
R1
63-66
67-70
71-75
Scheme 5. Reagents and conditions: (a) R1OH, NaH, DMF; (b) 2-pyr-SH, NaH, DMF; (c)
H2SO4; (d) Br2, NaOH; (e) i. 35d, NaNCS, pyridine, acetonitrile, 40 °C ii. 63-66, 75 °C; (f) TFA;
(g) Ac2O, Et3N, or MsCl, Et3N.
3. Results and discussion
3.1 Initial hit and protein crystal structure
Upon examination of the crystal structure of 1,19 we hypothesized that the benzyloxy group
could be replaced with an aryl ether to access similar space in the hydrophobic pocket of the
enzyme. Such a modification would remove the benzylic carbon, a potential site of oxidative
metabolism. Due to its straightforward synthesis, the first compound to test this hypothesis was
the 2-cyanophenyl ether 6. As can be seen in Table 1, its ability to promote GK activation was
promising for a relatively unadorned compound. The Vmax for 6, as well as for the 4-cyano
analog 7, was below 100%. A compound which presents a lower S0.5, but a lower Vmax provides
activation of glucose phosphoryolation at low glucose concentrations; however, it would be
predicted to inhibit glucose phosphorylation at high blood glucose concentrations, leading to
10

prolonged postprandial hyperglycemia. (Figure 2) To identify activators with a Vmax > 100%
was an aim of the lead optimization for this series.
Inhibitory
Region
No GKA
+GKA
Activating Region
100
80
60
40
20
0
10
100
1000
[Glucose] (mg/dL)
Figure 2. Graphical representation of a GKA with low S0.5 and Vmax of 75%. At blood glucose
concentrations >250 mg/dL, the activity of GK bound to the GKA is lower than the native
enzyme.
The co-crystal of glucokinase with glucose and 6 was used to elucidate the single crystal X-ray
structure of the complex. (Figure 3) The biaryl ether occupied a similar pocket as the benzyl
ether of 1, however the nitrile group did not appear to make any favorable interactions with the
enzyme.
11

Figure 3. X-ray crystallographic structure of aryl ether 6 in complex with GK at 2.1 Å. (6E0E)
3.2. SAR of 3-O-Aryl GKAs.
An initial investigation of the SAR of analogs of 6 was undertaken to expand on this initial
finding, (Table 1). Removal of the benzylic methylene of 1 to create the biaryl ether linkages led
to improved potencies. Initially, the 2- and 4-nitriles on the phenyl ring were utilized to facilitate
synthesis of the aryl ethers, as shown in Scheme 1, based on the facile SNAr incorporation of the
aryl group. Hence, alternative routes to the aryl ethers were identified, in order to rapidly gain
access to a wider range of aryl ethers. After solving the cocrystal structure of 6 bound to GK,
molecular modeling identified highly lipophilic areas surrounding the nitrile, which appeared to
be a suboptimal substituent for this position on the aryl ring. A different route (Scheme 2) was
utilized to prepare unsubstituted phenyl ethers, which were designed to better complement the
12

hydrophobic environment in this pocket. As anticipated, the EC50 of the phenyl ether 15 was
greater than 5-fold more potent than 6.
Previously, we reported that addition of groups at the 5-position of the pyridyl ring can confer
dramatic improvements in potency.19 The allosteric binding site for GKAs includes a number of
flexible residues, which can adapt to a wide range of small molecule structures. Among these,
Tyr215 can adopt two families of conformations, depending on the steric requirement of the
activator. Surprisingly, in the structure of the GK-6 complex, where 6 presents only a hydrogen
along the vector facing Tyr215, the phenol group of Tyr215 is observed to be rotated away from
the binding site, creating a larger hydrophobic pocket for activators bearing larger groups in this
region. In the case of these 3-aryloxy GKAs, addition of a 5-bromo (14) resulted in a compound
that was equipotent in terms of EC50 and S0.5, but the Vmax of the enzyme-activator complex was
nearly doubled to 145%.
Table 1. Initial SAR of the pyridyl 3- and 5-substituents.
R1
N
S
R3
N
N
H
O
R2
Compound
R1
R2
Bn
R3
EC50 (nM)
5800
119
S0.5 (mM)
3.2
Vmax (%)
79
1
2
H
Me
iPr
S(CH2)2Imid
Bn
0.9
117
83
6
H
H
2-CN-Ph
4-CN-Ph
Ph
Me
Me
Me
Me
Me
2359
3521
419
1.4
7
2.0
84
15
14
16
H
0.9
80
Br
Ph
400
1.0
145
127
BnS-
Ph
157
1.1
13

17
18
19
PhS-
Ph
Ph
Ph
Me
Me
Me
98
72
32
1.3
0.5
0.4
155
151
158
2-Pyr-S-
4-Pyr-S-
Based on the SAR from our previous work,19 wherein thioethers were preferred 5-substituents,
the bromide 14 was elaborated into a small set of thioethers (16-19). These compounds showed
further improvements in the potency, with 19 having an EC50 of 32 nM. Relative to the
previously published compound 2, 19 was nearly 4-fold more potent with an improved Vmax
.
However, with an S0.5 value of 0.4 mM, the compound begins to activate GK below
physiologically relevant glucose concentrations, which may result in severe hypoglycemia. In
this series of compounds, potency and S0.5 correlated, with the most potent compounds being the
least glucose-sensitive activators. Subsequent efforts were to identify SAR where the correlation
did not apply.
3.3. SAR of thiazole substituents
Based on the crystal structure of GK-6, the 4-methyl group on the thiazole points to a solvent-
accessible channel. The SAR of this region are depicted in Table 2. While all of the compounds
were potent, compounds such as 28 and 29 were especially intriguing. Both of these compounds
displayed EC50 < 100 nM and Vmax >100%, however the S0.5 of the compounds was dramatically
higher than many of the other compounds in the series. These data exemplify clear departures
from earlier compounds that exhibited correlation between potency and S0.5, thus demonstrating
that judicious choice of substituents can have a dramatic influence on the kinetics of the enzyme-
activator complexes.
Table 2. The SAR of 4-substituents of the aminothiazole
14

S
N
S
R
N
N
N
H
O
Compound
R
EC50 (nM)
32
S0.5 (mM)
0.4
Vmax (%)
158
ClogP
5.5
19
23
24
25
26
27
28
29
31
Me
Et
20
0.4
153
6.0
CF3
cPr
213
103
58
2.1
115
5.9
1.0
150
5.9
iPr
0.7
135
6.4
iBu
cHex
(CH2)2Ph
112
92
1.1
151
6.9
2.4
117
7.6
47
1.9
150
7.4
(CH2)2CO2H 22
0.4
131
4.3
3.4. In vivo correlation of S0.5 and blood glucose levels
In order to determine how the in vitro enzyme kinetic parameters (particularly S0.5 and Vmax
)
affected anti-hyperglycemic activity, we evaluated many of the compounds in Table 2 in oral
glucose tolerance tests (OGTT) in euglycemic C57BL/6 mice. The compounds were tested at 50
mg/kg in an attempt to determine their maximum blood-glucose-lowering efficacy as well as
their potential to elicit hypoglycemia. Plasma drug concentrations were obtained at the 2.5 h time
point of the OGTT studies, to provide evidence of adequate exposure of the compounds. (see
Supplemental Information) The OGTT data for this set were compared, and an interesting trend
emerged. The glucose concentration at the 2 h time point appeared to be correlated with S0.5
(Figures 4 and 5). The hyperbolic curve fit of these two parameters was surprisingly tight (r2 =
0.91). No such correlation was identified for Vmax. From this data, the team hypothesized that an
15

S0.5 of approximately 1.0 would confer a good balance of efficacy and safety from hypoglycemia
(correlating with minimal blood glucose concentrations of ~80 mg/dL).
Figure 4. Overlay of OGTTs of compounds from Table 2 at 50 mg/kg PO (n=8 mice per time-
point; values represent the arithmetic mean). The glucosemin at 2 h appears to change depending
on the compound.
16

Figure 5. Correlation of the 2 h blood glucose concentration with the measured S0.5 in the
enzymatic assay. Highlighted area is the proposed range with best balance of efficacy and
hypoglycemic risk.
3.5. Thiadiazole SAR
While the above SAR confirmed our hypothesis that there is a strong correlation between S0.5
and hypoglycemic risk (blood glucose levels <60 mg/dL), electron-rich, unsubstituted 4-
pyridines are not generally good drug candidates (high potential for drug-drug interactions via
CYP inhibition). Because the potency, kinetics, and solubility are similar for compounds with
either a S-4-pyridyl or a S-2-pyridyl group (18 vs. 19), we decided to utilize this group. We were
also focused on lowering the ClogP of the series and felt that further investigation of the thiazole
substituents was warranted. In addition, we were cognizant of the propensity for 2-amino
thiazoles to be metabolized into potentially toxic thioureas.27,28,29 By replacing the thiazole with
thiadiazole, the potential for generation of thiourea metabolites is reduced.30
17

As shown in Table 3, inclusion of piperidinyl-derived substituents on the thiadiazole
successfully lowered the ClogP of the series, while keeping the EC50 and Vmax in a desirable
range. One parameter that was difficult to predict was S0.5. Small changes had little effect on
EC50 and Vmax, but dramatically changed the S0.5. For example, changing the piperidinyl
substituent from acetyl (42) to mesyl (43) kept the potency below 100 nM and the Vmax well
above 100%, however, the S0.5 jumped from 0.61 mM to 2.8 mM. These compounds were still
not ideal in terms of physical properties (3.9 < CLogP < 4.7), which correlated with medium to
high predicted human hepatocyte clearance rates for all compounds except the acetyl and mesyl
compounds, 42 and 43 respectively.
Table 3. In vitro data for initial set of substituted aminothiadiazoles.
S
N
S
N
R
N
N
N
H
O
Compound
R
EC50
(nM)
S0.5 (mM) Vmax (%) ClogP
Predicted
Human
Hepatocyte
ER (%)
37
38
Me
78
21
0.44
0.55
148
138
5.5
4.6
69
47
O
39
42
116
49
0.73
0.61
160
144
4.6
3.9
63
30
O
N
O
18

43
44
70
21
2.8
134
151
4.5
5.2
16
72
O
S
N
N
O
N
0.56
O
O
O
O
O
O
45
46
47
48
49
29
36
20
37
69
0.77
0.71
0.48
0.51
0.57
147
143
139
133
138
4.7
4.1
4.6
4.4
4.4
65
66
52
51
62
N
N
N
N
N
OH
N
OH
OH
3.6. Replacement of 3-O-aryl with heterocycles
A potential problem for these compounds was the lipophilic unsubstituted 3-phenoxy group.
An investigation as to whether the replacement of the phenyl with a polar heterocycle would
improve the physical properties and metabolic stability while retaining the potency of the
original lead was initiated. Table 4 lists the SAR of various heterocyclic ethers at the 3-position,
keeping the thiadiazole and 5-position of the pyridine core constant. Unexpectedly, the higher
19

S0.5 values of the phenoxy versions were generally not retained in the more potent compounds in
the heterocyclic series.
One assay that was included after the start of the project was a human serum albumin-shifted
EC50 assay. This was incorporated as a high-throughput method to estimate the plasma protein
binding of the compounds.31 This was a significant differentiating factor in our SAR
investigation. By changing the phenyl group to a 2-methylpyridyl (42 to 72) the standard EC50
remained the same, but the EC50 in the presence of 4% human serum albumin dropped by 5-fold.
Unfortunately, replacing the acetyl for mesyl in this series did not have the same beneficial effect
on S0.5 (42 and 43 vs. 72 and 73). Structural or physical properties that consistently had an effect
on S0.5 were not identified. From this work, 72 was chosen for further investigation.
Table 4. In vitro data for 3-(heteroaryloxy)pyridine analogs
S
N
S
N
R2
N
N
N
N
H
O
R1
Compound R1
R2
EC50
(nM)
4%
S0.5
(mM)
Vmax
(%)
ClogP
Predicted
Human
Hepatocyte
ER (%)
HSA
EC50
(nM)
42
43
71
72
73
74
Ph
Ac
Ms
Ac
49
1640
3345
2710
362
0.61
2.8
144
134
135
115
131
109
3.9
4.5
2.4
2.9
3.5
3.8
30
16
41
45
26
NT
Ph
70
3-Pyr
233
42
1.1
2-Me-3-Pyr Ac
2-Me-3-Pyr Ms
5-quinoline Ac
0.63
0.58
0.42
33
684
9.3
504
20

75
Ac
279
1967
0.97
126
2.0
32
N
N
Figure 6. X-ray crystallographic structure of 72 bound to GK at 1.9 Å. (6E0I)
The structure of 72 bound to GK was solved by x-ray crystallography, confirming the
predicted binding mode of the compound (Figure 6). The aminothiadiazole makes the typical
dual hydrogen bonds with the main chain of Arg 63, while the 2-pyridyl thioether fills the pocket
optionally occupied by the highly mobile side-chain of Tyr 215, making lipophilic contacts with
both Tyr 214 and Tyr 215. The 2-methyl-3-pyridyl ether resides in the lower pocket as shown,
forming a loose pi-stacking interaction with Tyr 214, while directing the 2-methyl group deep
into a hydrophobic cleft in the protein. The acetyl piperidine substituent stretches through a
21

largely hydrophobic tunnel, directing the acetyl group toward a solvent-exposed opening on the
protein surface.
Compound 72 was progressed into a set of secondary in vitro assays to determine its
probability of in vivo success. The compound displayed acceptable stability in predictive
microsomal and hepatocyte metabolism experiments (predicted mouse microsomal clearance: 41
mL/min/kg; predicted mouse hepatocyte clearance: 17 mL/min/kg). It was highly permeable in a
Caco2 cell assay (A-B: 9.3 cm/s; pe ratio: 0.83) and did not display CYP450 inhibition up to 25
M for 5 isoforms (1A2, 2C9, 2D6, 3A4, 2C19). These results provided the team with
confidence to progress the compound into a set of in vivo pharmacology assays.
3.7. Mouse PK of 72
As shown in Figure 7, 72 showed acceptable pharmacokinetic properties in mouse, with an
observed systemic clearance value in agreement with the predicted mouse hepatocyte clearance
(21 vs. 17 mL/min/kg). Gratifylingly, compound 72 demonstrated a good bioavailability (60%)
in mice and exposures that maintained plasma drug concentrations above the human serum
albumin-shifted EC50 for approximately 4 hours.
22

1 mg/kg IV (PEG400:EtOH:H2O, 4:2:4, by volume)
10 mg/kg PO (30% Captisol®, aq., w/v)
4% HSA EC50
10
1
0.1
0.01
0.001
0
1
2
3
4
5
6
7
8
Time (h)
AUCinf
(hr*µg/mL)
Dose
t1/2 (hr)
1.28
CL (mL/min/kg) Vss (L/kg)
1 mg/kg IV
Dose
0.770
21.6
0.746
F (%)
60.3
AUCinf
(hr*µg/mL)
Cmax (µg/mL)
1.67
Tmax (hr)
1.00
10 mg/kg PO
4.65
Figure 7. Mean plasma concentration/time profiles of compound 72 after a single 1 mg/kg
intravenous (IV; diamonds) or 10 mg/kg oral (PO; squares) dose of compound 72 in solution to
male CD-1 mice (n=3 per time-point; values represent the arithmetic mean ± SD). The dashed
midline represents the EC50 value for compound 72 as determined by the 4% HSA EC50 assay.
3.8. Efficacy of 72 in mouse models of diabetes
23

A dose-response oral glucose tolerance test (OGTT) study of compound 72 performed in
C57BL/6 mice (Figure 8) indicated that 10 mg/kg effectively reduced the peak blood glucose at
15 minutes, and significantly reduced the total glucose area-under-the-curve (AUCglucose) by
23%. The efficacy of 72 at 10 mg/kg was greater than the positive control sitagliptin at 30 mg/kg
in this study. The 30 mg/kg dose dramatically blunted the glucose excursion, and the 2 h average
blood glucose remained above 70 mg/dL, our threshold for overt hypoglycemia in these
euglycemic mice. This compound was therefore determined to be effective with a relatively low
potential for hypoglycemia and was advanced into a diabetic animal model.
Figure 8. Dose response OGTT study of compound 72 in normoglycemic C57BL/6 mice (n=8
mice per time-point; values represent the arithmetic mean ± SD)
Compound 72 was studied in male diabetic ob/ob mice, dosed orally once-daily for 14 days,
with OGTT performed on day 14 and non-fasted blood glucose measured over the course of the
24

study. There were three criteria in which we were interested: lowering of plasma glucose in the
OGTTs; lowering of fasted plasma glucose before OGTT; and lowering of non-fasted plasma
glucose throughout the study. As can be seen in Figures 9 and 10, the 10 and 30 mg/kg doses
were effective at all three critical criteria. At 10 mg/kg, compound 72 lowered the fasted blood
glucose from the control animals on day 14 (-0.5 hr time-point) as well as the AUC of the
OGTT. Also shown in Figures 9 and 10 are the untreated non-diabetic control C57BL/6 mice.
The 10 and 30 mg/kg doses controlled the fasted (t=-0.5 h in figure 9) and non-fasted blood
glucose concentrations (Figure 10) to those of the non-diabetic controls. The 30 mg/kg dose had
an average 2 h blood glucose level of 66 mg/dL, which was just below our 70 mg/dL threshold
for hypoglycemia. However, no adverse effects were observed. Sitagliptin (30 mg/kg) was
utilized as a positive control.
25

Figure 9. Compound 72 studied in ob/ob mouse, OGTT on day 14 (n=8 mice per time-point;
values represent the arithmetic mean ± SD).
Figure 10. Non-fasted blood glucose values over the course of the 14d ob/ob study mice (n=8
per time-point; values represent the arithmetic mean ± SD). Day 0 blood glucose concentrations
were high in ob/ob mice due to stress response from initial handling.
4. Conclusions
We have reported on the discovery and optimization of a new series of non-amide based GKAs,
through iterative structure-based design. By changing the 3-substituent of the pyridine core from
the original benzyl ether lead to an aryl ether, the potency and GK activation kinetics of the
series were improved. Optimization of three different regions of the molecules led to the
synthesis of acetylpiperidine 72, which possessed a good balance of enzymatic, physicochemical,
and pharmacokinetic properties. Compound 72 was progressed through the later stage assays of
the project and demonstrated anti-hyperglycemic activity in a dose-ranging 14 day ob/ob mouse
26

study. Further optimization of this series to improve glucose-dependence and pharmacokinetics
will be reported in future publications.
5. Experimental Section
5.1 Chemistry
Reagents were purchased from commercial sources and used without further purification.
Nuclear magnetic resonance (NMR) spectra were measured in the indicated solvent with
tetramethylsilane (TMS) or the residual solvent peak as the internal standard on a Varian 400
MHz spectrometer. Chemical shifts () are in parts per million. LC-MS experiments were
performed on an Agilent 1100 HPLC coupled with an Agilent MSD mass spectrometer using
ESI as ionization source. Peaks were detected by UV absorbance at 220 and 254 nm, and MS full
scan was applied to all experiments. All compounds disclosed in this paper were confirmed to be
>95% purity via this method.
5.1.1. 2-(2-(4-Methylthiazol-2-ylamino)pyridin-3-yloxy)benzonitrile (6)
2-(2-Chloropyridin-3-yloxy)benzonitrile (0.667 g, 2.89 mmol), 4-methylthiazol-2-amine (0.300
g, 2.63 mmol) and potassium phosphate (0.614 g, 2.89 mmol) were dissolved in toluene (7 mL)
and nitrogen bubbled through the solution for 5 minutes. Pd2dba3 (0.060 g, 0.0657 mmol) and
4,5-bis(diphenylphosphino)-9,9-dimethyl-9H-xanthene (0.041 g, 0.0723 mmol) were added and
the reaction further bubbled through with nitrogen for 5 minutes. Degassed water (2 mL) was
added and the reaction heated to 90 °C overnight under nitrogen. The reaction was cooled to
room temperature and partitioned between water and ethyl acetate, the organic layer was dried
over magnesium sulfate, filtered and concentrated. The residue was purified over silica gel
(elution 10-40% EtOAc in hexanes) to afford 2-(2-(4-methylthiazol-2-ylamino)pyridin-3-
yloxy)benzonitrile (0.385 g, 46.6% yield) as yellow solid. 1H NMR (CDCl3)  ppm 8.23 (dd, J =
27

5.1, 1.6 Hz, 1H), 7.72 (dd, J = 7.6, 1.6 Hz, 1H), 7.53 (m, 1H), 7.24 (td, J = 7.6, 1.0 Hz, 1H), 7.20
(dd, J = 8.0, 1.6 Hz, 1H), 6.92-6.84 (m, 2H), 6.44 (q, J = 1.1 Hz, 1H), 2.33 (d, J = 1.1 Hz, 3H);
MS (ES+) m/e 309.0 (M+H)+.
5.1.2. 4-(2-(4-Methylthiazol-2-ylamino)pyridin-3-yloxy)benzonitrile (7)
Synthesized via the same method as 6. 1H NMR (CDCl3)  ppm 8.23 (dd, J = 4.9, 1.6 Hz, 1H),
7.65 (m, 2H), 7.25 (dd, J = 7.8, 1.4 Hz, 1H), 7.05 (m, 2H), 6.92 (dd, J = 7.8, 4.9 Hz, 1H), 6.44
(q, J = 1.0 Hz, 1H), 2.32 (d, J = 1.0 Hz, 3H); MS (ES+) m/e 309.0 (M+H)+.
5.1.3. 5-Bromo-N-(4-methylthiazol-2-yl)-3-phenoxypyridin-2-amine (14)
A 250 mL round-bottomed flask was charged with 1-(5-bromo-3-phenoxypyridin-2-yl)thiourea
(5.0 g, 15.42 mmol), 1-chloropropan-2-one (1.7 ml, 21.6 mmol), triethylamine (3.7 ml, 27.0
mmol), and EtOH (100 mL) and heated to 70 °C overnight. The reaction was cooled to room
temperature, poured into water, extracted with CH2Cl2 (2 x 125 mL), dried over sodium sulfate,
filtered and concentrated. The residue was purified over silica gel (elution 15% EtOAc in
hexanes) to afford 5-bromo-N-(4-methylthiazol-2-yl)-3-phenoxypyridin-2-amine (5.0 g, 89.5%
yield) as a white solid. 1H NMR (d6-DMSO)  ppm 8.68 (br s, 1H), 8.13 (m, 1H), 7.42 (m, 2H),
7.24 (m, 1H), 7.12 (m, 1H), 7.07 (m, 1H), 7.04 (m, 1H), 6.44 (q, J = 1.1 Hz, 1H), 2.34 (d, J = 1.1
Hz, 3H); MS (apci) m/e 362.2, 364.2 (M+H)+.
5.1.4. N-(4-Methylthiazol-2-yl)-3-phenoxypyridin-2-amine hydrochloride (15)
5-Bromo-N-(4-methylthiazol-2-yl)-3-phenoxypyridin-2-amine (0.125 g, 0.345 mmol) was
dissolved in THF (30 mL) and cooled to -78 °C. 1.6M MeLi (0.270 ml, 0.431 mmol) was added
slowly and stirred for 10 min. 2.5M butyllithium (0.022 g, 0.345 mmol) was added and stirred
for 15 min. The reaction was quenched with aqueous ammonium chloride, extracted with
CH2Cl2, dried over sodium sulfate, filtered and concentrated. The residue was purified over silica
28

gel (elution 10-20% EtOAc in hexanes) to afford freebase. The residue was dissolved in CH2Cl2
(2 mL) and 2M HCl (0.5 mL) in ether was added and concentrated to afford N-(4-methylthiazol-
1
2-yl)-3-phenoxypyridin-2-amine hydrochloride (0.068 g, 61.6% yield). H NMR (d6-DMSO) 
ppm 8.20 (dd, J = 4.8, 1.2 Hz, 1H), 7.46 (t, J = 7.6 Hz, 2H), 7.40 (dd, J = 8.0, 1.2 Hz, 1H), 7.23
(m, 1H), 7.15 (m, 3H), 6.89 (br s, 1H), 2.33 (s, 3H); MS (apci) m/e 284.2 (M+H)+.
5.1.5. 5-(Benzylthio)-N-(4-methylthiazol-2-yl)-3-phenoxypyridin-2-amine hydrochloride (16)
A 10 mL round-bottomed flask was charged with 5-bromo-N-(4-methylthiazol-2-yl)-3-
phenoxypyridin-2-amine (70 mg, 0.193 mmol), 4,5-bis(diphenylphosphino)-9,9-dimethyl-9H-
xanthene (5.6 mg, 0.009 mmol), Pd2dba3 (4.4 mg, 0.005 mmol), N-ethyl-N-isopropylpropan-2-
amine (0.067 ml, 0.39 mmol), phenylmethanethiol (0.025 ml, 0.21 mmol) and dioxane (2 mL).
The reaction was purged with nitrogen and heated to 100 °C overnight. The solvent was removed
and the residue was purified over silica gel (elution 15% EtOAc in hexanes) to afford free base.
The residue was dissolved in CH2Cl2 (5 mL) and 2M HCl in ether (1 mL) was added and the
solvent removed to afford 5-(benzylthio)-N-(4-methylthiazol-2-yl)-3-phenoxypyridin-2-amine
1
hydrochloride (79.4 mg, 92.9% yield) as a white solid. H NMR (d6-DMSO)  ppm 8.08 (m,
1H), 7.43 (m, 2H), 7.28-7.15 (m, 7H), 6.98 (m, 2H), 6.74 (br s, 1H), 4.12 (s, 2H), 2.26 (s, 3H);
MS (apci) m/e 406.2 (M+H)+.
5.1.6. N-(4-Methylthiazol-2-yl)-3-phenoxy-5-(phenylthio)pyridin-2-amine hydrochloride (17)
Synthesized via the same method as 15. 1H NMR (d6-DMSO)  ppm 8.20 (d, J = 2.0 Hz, 1H),
7.39 (m, 2H), 7.31 (m, 2H), 7.21 (m, 4H), 7.16 (m, 1H), 7.07 (m, 2H), 6.65 (s, 1H), 2.25 (s, 3H);
MS (apci) m/e 392.2 (M+H)+.
5.1.7. N-(4-Methylthiazol-2-yl)-3-phenoxy-5-(pyridin-2-ylthio)pyridin-2-amine hydrochloride
(18)
29