Investigation of functionally liver selective glucokinase activators for the treatment of type 2 diabetes

Article abstract of DOI:10.1021/jm900839k

Type 2 diabetes is a polygenic disease which afflicts nearly 200 million people worldwide and is expected to increase to near epidemic levels over the next 10-15 years. Glucokinase (GK) activators are currently under investigation by a number of pharmaceu

Full text of DOI:10.1021/jm900839k

6142 J. Med. Chem. 2009, 52, 6142–6152
DOI: 10.1021/jm900839k
Investigation of Functionally Liver Selective Glucokinase Activators for the Treatment
of Type 2 Diabetes§
Gregory R. Bebernitz,*,† Valerie Beaulieu,† Bethany A. Dale,† Richard Deacon,† Alokesh Duttaroy,† Jiaping Gao,†
Melissa S. Grondine,† Ramesh C. Gupta,‡ Mesut Kakmak,† Michael Kavana,† Louise C. Kirman,† Jinsheng Liang,†
Wieslawa M. Maniara,† Siralee Munshi,‡ Sunil S. Nadkarni,‡ Herbert F. Schuster,† Travis Stams,† Irene St. Denny,†
Paul M. Taslimi,† Brian Vash,† and Shari L. Caplan†
†Novartis Institutes for BioMedical Research, Inc., 100 Technology Square, Cambridge, Massachusetts 02139, and ‡Torrent Research Centre,
Village Bhat, Gujarat, India
Received June 9, 2009
Type 2 diabetes is a polygenic disease which afflicts nearly 200 million people worldwide and is expected
to increase to near epidemic levels over the next 10-15 years. Glucokinase (GK) activators are currently
under investigation by a number of pharmaceutical companies with only a few reaching early clinical
evaluation. A GK activator has the promise of potentially affecting both the β-cells of the pancreas, by
improving glucose sensitive insulin secretion, as well as the liver, by reducing uncontrolled glucose
output and restoring post-prandial glucose uptake and storage as glycogen. Herein, we report our
efforts on a sulfonamide chemotype with the aim to generate liver selective GK activators which
culminated in the discovery of 3-cyclopentyl-N-(5-methoxy-thiazolo[5,4-b]pyridin-2-yl)-2-[4-(4-methyl-
piperazine-1-sulfonyl)-phenyl]-propionamide (17c). This compound activated the GK enzyme (RKa =
39 nM) in vitro at low nanomolar concentrations and significantly reduced glucose levels during an oral
glucose tolerance test in normal mice.
Introduction
and insufficient insulin is released to bring glucose levels back
into the normal range resulting in hyperglycemia. A GK
activator should lower the glucose set point at which insulin
is released and thereby work in concert with lower glucose
output by the liver to effectively normalize plasma glucose
concentrations. A number of reviews have been published in
this area.1
Activation of GK both in the liver and the pancreas should
have a profound effect on circulating glucose levels in the
diabetic state. However, there are serious concerns that
targeting the pancreas would not only result in an increased
potential for hypoglycemia but more significantly result in
increased stress on the β-cells, similar to that observed with
long-term treatment with sulfonylureas, and thereby lead to
β-cell failure and ultimately a worsening of the diabetic state.
For this reason, our effort has focused on compounds whose
Excessive hepatic glucose production and inappropriate
glucose sensing by pancreatic β-cells contribute in a major
way to the resulting whole-body insulin resistance and hyper-
glycemia associated with type 2 diabetes. Glucokinase (GKa)
is a high Km hexokinase present in key glucose metabolizing
tissues (liver and pancreas) that catalyzes the phosphorylation
of glucose to glucose-6-phosphate. The enzyme serves as a
substrate switch in these tissues, thus controlling the fate of
glucose disposal subsequent to a meal and thereby plays a
central role in whole-body glucose homeostasis.
In the liver, GK is the rate-limiting enzyme controlling
glucose utilization, and its activation promotes glycogen
synthesis (Figure 1). The balance between the rate of glucose
phosphorylation by GK and dephosphorylation by glucose-6-
phosphatase determines the net glucose flux into or out of this
tissue. In the diabetic state, control of this balance is lost, and
the liver exports glucose continuously even during the fed
state. A GK activator should overcome this effect and allow
the liver to take up glucose and store it as glycogen during the
prandial state. In the pancreatic β-cells, GK serves as the
glucose sensor and controls the insulin response to a specific
glucose load. In the diabetic situation, this sensor is defective,
*To whom correspondence should be addressed. Phone: (617) 871
7302. Fax: (617) 871 7042. E-mail: greg.bebernitz@novartis.com.
§Atomic coordinates for human glucokinase in complex with com-
pound R(-)17c have been deposited in the Protein Data Bank with the
accession code 3IMX.
a Abbreviations: AUC, area under the curve; GK, glucokinase;
GKRP, glucokinase regulatory protein; PAMPA, parallel artificial
membrane permeability assay; OGTT, oral glucose tolerance test;
IVGTT, intravenous glucose tolerance test; S0.5, concentration giving
half-maximal activation for sigmoidal activation curve.
Figure 1. GK function in pancreatic β-cell and liver.
r
pubs.acs.org/jmc
Published on Web 09/11/2009
2009 American Chemical Society

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19 6143
Figure 2. Structures of key compounds.
Scheme 1. Synthetic Route via Method A
effects are minimized in the pancreas and primarily directed to
the liver. The patient benefit of a liver-selective GK activator,
although likely a safer therapy as compared to a pancreas-
targeted agent, has yet to be proven in the clinic.
to their synthetic accessibility via parallel synthesis techniques
as well as their potential for improved physiochemical proper-
ties.
Chemistry. Compounds were prepared as illustrated in
either Scheme 1 or Scheme 2. All amines utilized were
commercially available.
Herein, we will describe efforts toward developing our first
GK activator, which culminated in the discovery of R(-)17c,
as a potential treatment for type 2 diabetes. Most early
published efforts have reported an amidine moiety as a
necessary component of the GK activator binding motif,
which from internal and published crystallographic informa-
tion forms key interactions with the protein backbone of
Arg63. This interaction anchors the compound and allows
additional interactions, which ultimately maintain helix 13 of
the small domain in the desired active conformation of the
protein and prevent the undesired orientation commonly
referred to as the super-open conformation. This super-open
conformation has been described earlier as a nonfunctional
form of the enzyme, which in the case of the liver allows for
interaction with glucokinase regulatory protein (GKRP).2
This interaction of GK with GKRP is suspected to provide
both regulation and protection of GK during times when its
activity is not required.
Synthesis via Method A (Scheme 1). The ethyl ester of
phenyl acetic acid was treated with chlorosulfonic acid under
heating to afford a mixture of sulfonyl chlorides, 3.
The resulting mixture was carried on crude and treated
with the desired amine to afford a mixture of sulfonamides,
which were separated by crystallization or chromatography
to provide the pure para isomer, 4. Examination of this
mixture unexpectedly indicated two regioisomers which were
not an ortho/para mixture but actually an 80:20 mixture of
para/meta.4 The only examples sited in the literature of meta-
chlorosulfonylation have been when additional directing
groups have contributed to and potentially determined the
regiochemical outcome.5 In addition, while Friedel-Crafts
acylation of ethyl phenylacetate does give substantial
amounts of the meta product and very little if any ortho
product because of the size of the acylating complex,6
nitration of methyl phenylacetate generates no meta product
with only ortho and para (o:p ratio 15:85).7 Therefore, it
appears that the substrate, ethyl phenylacetate, is somewhat
variable in its chemistry with regard to electrophilic aromatic
substitution reactions with the acetate unit neither function-
ing as a good activating ortho/para nor a good deactivating
A high throughput screening effort identified a key lead, 1
(Figure 2), which was shown to directly activate GK and as
well contained the requisite amidine core. Utilizing the pyr-
idoaminothiazole as our amidine counterpart, we elected to
push beyond the sulfone chemotype as exemplified by 2a
(RO0280450)3 and others and explore sulfonamides due both

6144 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19
Scheme 2. Synthetic Route via Method B
Bebernitz et al.
meta director with the outcome of higher meta product
under some conditions than one would have expected. Sub-
sequent conversion of 4 to its enolate with LDA and alkyla-
tion with cyclopentylmethyliodide afforded 5. Saponifica-
tion of the ester under standard conditions afforded 6, which
was subjected to traditional peptide coupling with the pyr-
idoaminothiazole, 7, to afford the desired compounds
13-17. Isolation of the R(-) isomer, if desired could be
accomplished using chiral HPLC.
This procedure, although efficient, cost-effective, and
scalable, introduced the major site of diversity early in the
route, and an alternative method which allowed for the
diversification step late in the synthesis was desired.
Synthesis by Method B (Scheme 2). An alternate route
beginning with ethyl p-nitrophenylacetate proved to be both
straightforward and effective, especially with regard to
variation in the sulfonamide. Alkylation of ethyl p-nitrophe-
nylacetate with cyclopentyl methyl iodide as described be-
fore afforded 8. Subsequent saponification of the ester and
peptide coupling with pyridoaminothiazole 7 then resulted in
10. Reduction of the nitro functionality to the aniline 11 was
followed by diazotization. This material was then treated
with sulfur dioxide in acetic acid in the presence of copper
chloride to afford the sulfonyl chloride, 12, in this case as the
pure para-isomer because of the employment of the nitro
group to direct the regiochemical outcome. This material
could then be quenched with various amines or with ammo-
nium hydroxide to afford 13-17 as shown in Scheme 2.
Evaluation of Compounds. Primary in Vitro Assay. The
standard method of testing compounds involved reporting
glucose levels; and (3) compounds which actually showed
inverted activity upon increasing glucose concentrations,
activating glucose phosphorylation at lower glucose concen-
trations but inhibiting phosphorylation at higher glucose
concentrations. Compounds selected as highest priority were
as expected those of class 1, which showed increasing activity
with increasing glucose levels. Compounds of class 3 should
theoretically be less desirable as therapeutic agents as they
would have no effect at higher glucose concentrations where
their efficacy is most needed. The optimum profile would
entail a maximum effect at high glucose levels, thereby
countering hyperglycemia, and little if no effect at lower
glucose levels, thus preventing further glucose uptake when
hypoglycemia could result. Compounds of this type were not
found in our evaluation of this chemotype.
Hepatocyte Assay. Evaluation of potency in a cellular
model was performed not only to validate the compound’s
effect on endogenous GK activity but additionally to serve as
an early permeability filter for selecting compounds for in
vivo evaluation. The rate of glucose metabolism by hepato-
cytes is dependent on GK activity levels.9 Effects on glucose
phosphorylation were assessed in freshly isolated rat
hepatocytes as determined by the release of 3H2O from
[2-3H]glucose. A discrepancy is noted between the in vitro
RKa and the % GK activation in hepatocytes. Given that the
endogenous rat GK protein shares ∼94% amino acid iden-
tity with the recombinant human GK, differences in the
enzyme source are most likely not the cause. Furthermore,
the GK protein (aa 17-466) used for in vitro assays displayed
activity similar to that of the full-length protein (data not
shown), indicating that the N-terminal truncation does not
alter the kinetics of GK. One possibility to explain the
discrepancy in assay values could be due to protein binding
of the compounds in cells; however, neither the primary in
vitro assay nor the hepatocyte culture medium contained a
source of serum albumin. Other factors such as permea-
bility, clearance, as well as metabolism of the compounds
in hepatocytes could contribute to effects on GK activity
(i.e., by reducing or enhancing it) that would not be observed
in the in vitro assay and therefore need to be taken into
account. In addition, it is possible that there is a minimum
threshold of effect required for GK phosphorylation which
S0.5 in addition to a corresponding percent activation (Vmax
)
at a fixed glucose concentration (10 mM). However, evalua-
tion at both multiple compound (0-20 μM) as well as
multiple glucose (300 μM-100 mM) concentrations fol-
lowed by analysis of the data according to Segal et al.8
allowed for S0.5 and Vmax to be reported in a single value,
RKa, which provided a comprehensive single comparator
between compounds (see Table 1). While employing this
matrix evaluation, three specific classes of compounds were
found: (1) compounds which activated GK by decreasing the
S0.5 of glucose with little or no change in Vmax; (2) com-
pounds which showed little if any change in activity across

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19 6145
Table 1. Compound Structures, Synthesis Method, and Activity Data
cell assay
% GK
activationa
PAMPA
log P GI
(rank)c
HLM
half-life
(min)
in vivo
% glucose
reductiond
synthesis
method
cpd #
R1
R2
RKa (nM)
1
489 ( 50
897 ( 170
NA
411 ( 15*b
13a
13b
13c
14a
14b
14c
14d
14e
14f
H
H
propionamide
butyramide
B
-4.4 (M)
-4.1 (H)
-4.4 (M)
-3.3 (H)
-5.9 (L)
23.1
4.8
22% (0.5 h, p = 0.178)
H
B
H
B
NA
295 ( 120
42 ( 6
35 ( 4
73 ( 15
2.2
NC
Me
Et
Pr
Me
Et
A
A
A
A
A
A
A
A
B
221 ( 7*
345 ( 19*
17.8
5.8
Pr
i-Pr i-Pr
Bu
H
Bu
Ph
181 ( 99
583 ( 190
323 ( 95
302 ( 100
305 ( 100
266 ( 9
726 ( 320
377 ( 83
207 ( 60
180 ( 15
892 ( 580
1453 ( 310
61 ( 25
117 ( 58
1860 ( 580
2810 ( 500
263 ( 80
109 ( 11
147 ( 43
66 ( 28
-3.8 (H)
14g
14h
14i
H
3-pyridyl
cyclohexyl
219 ( 5*
78 ( 5*
H
-7 (L)
4.8
H
4-THP
cyclopropyl
benzyl
-4.5 (M)
-5.2 (L)
13.1
19.5
14j
H
A
A
A
B
14k
14l
H
32 ( 6†b
281 ( 75*
H
3-pyridomethyl
4-pyridomethyl
2-pyridomethyl
2-furanylmethyl
2-thiophen-methyl
CH2CH2OMe
CH2CH2NMe2
CH2COOH
CH2CH2COOH
benzyl
-4.6 (M)
-4.3 (H)
<1.5
<1.5
NC
14m
14n
14o
14p
15a
15b
15c
15d
16a
16b
16c
16d
17a
17b
17c
(R-)17c
17d
H
H
B
H
A
A
A
A
B
14 ( 3
-4.9 (M)
-5.1 (L)
-5 (M)
290
25.3
15.7
9
NC
NC
H
H
85 ( 2*
H
-4.8 (M)
-4.4 (M)
-4.3 (H)
-5.4 (L)
-5.1 (L)
-4.4 (M)
-4.1 (H)
-5.2 (L)
H
27.2
22.6
26.2
5.1
26%* (1 h), 20%* (2 h)
H
B
NC
NC
Me
Me
Me
Me
A
B
2-furanylmethyl
CH2CH2OH
1Me-2-imidazoylmethyl
piperadinyl
67 ( 9*
A
B
10
<1.5
6.2
A
A
A
A
A
66 ( 31
153 ( 32
39 ( 8
23 ( 2
50 ( 2
230 ( 16*
281 ( 9*
498 ( 28*
morpholino
NMe-piperazinyl
(R-) NMe-piperazinyl
piperazinyl
-3.9 (H)
9.1
data in text
22%a (0.5 h; p = 0.19), 26%* (2 h)e
-4.9 (M)
<1.5
a Compounds were tested at 30 μM in the primary cultured rat hepatocytes as described in Experimental Section. b †p < 0.05, *p < 0.0001 vs DMSO
control by Student’s t test. Values are the mean ( SEM for n = 4. c PAMPA was assessed at pH 4, 6.8, and 8, and the best value is reported. Ranking of
high, medium, and low are designated according to ref 11. d Compounds dosed orally at 50 mg/kg in microemulsion vehicle with the exception of 15c and 17d,
which were dosed in water. Data is reported as the % reduction from control, (time, hour), *p < 0.05 or p-value stated. e Dosed IVat 10 mg/kg. NC, no change. NA,
not available.
results in the signal being generated (3H20 release), and this is
not met with certain compounds. It should be noted that
when GK activation was greater than 200%, the values were
considered meaningful. Values falling below 200%, although
statistically relevant, were not used as a single criteria
warranting in vivo evaluation.
Physiochemical Properties. Estimation of physiochemical
properties considered critical to understanding in vivo exposure
are routinely evaluated during drug discovery efforts and used
to prioritize compounds warranting further examination in
vivo. Within this chemotype, permeability was evaluated using
a parallel artificial membrane permeability assay (PAMPA)
and compounds with high in vitro permeability were considered
for in vivo evaluation.10 The data is reported as log PGI and is
the highest of the 3 measured values (pH 4, pH 6.8, and pH 8).
A ranking of H (high), M (medium), and L (low) is attached to
these results and corresponds to the estimated fraction ab-
sorbed.11 High throughput solubility measurements were used
to evaluate a compound’s ability to dissolve or at least disperse
in the gut and allow for sufficient passive diffusion through the

6146 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19
Bebernitz et al.
Figure 3. Mean plasma and liver concentrations (0-24 h) after PO dosing of 13a, 13b, and 13c in C57BL6 mice dosed at 50 mg/kg in 0.5%
carboxymethyl cellulose.
gut wall.12 Evaluation of microsomal stability13 of the com-
pound provided a surrogate to in vivo exposure and half-life,
and was used to further select compounds for profiling in vivo.
In Vivo Activity in C57BL6 Mice. Whereas treatment with
a GK activator was not expected to alter basal glucose levels
acutely in normal mice due to tight glucose regulation, a
reduction in the glucose excursion (glucose AUC) during an
oral glucose tolerance test (OGTT) was expected and was
used to differentiate between compounds.
In Vivo Activity in Diet Induced Obese (DIO) Mice. A lipid
driven model of obesity, insulin resistance, and type 2
diabetes was used. Effects observed in this model should be
more profound than in normal mice. Intravenous glucose
tolerance test (IVGTT) was used to assess activity in β-cells.
Substitution of the primary amine of sulfonamide 13a with
a variety of lipophilic dialkyl amines provided compounds
14a-e, which showed up to 20-fold improvement in in vitro
potency over that of 13a. Compounds 14a and 14b were
evaluated in isolated rat hepatocytes and produced 2- and
3-fold increases in glucose turnover, respectively, as measured
by 3H2O release. Whereas 14b looked especially promising, it
showed poor permeability and faster clearance compared to
those of 14a, with no improvement in aqueous solubility.15
Analogues derived from monosubstituted amine derivatives
were also evaluated and as a whole demonstrated improved in
vitro activity as compared to 13a but rarely approached the
activity exhibited by 14b. Although a phenyl substitution (14f)
provided only a modest improvement, 3-pyridyl (14g) im-
proved activity nearly 3-fold over 13a. Compounds with
saturated rings (14h-14j) provided a similar 3-fold improve-
ment in activation over that of 13a but did not achieve the
activity of the disubstituted series 14a-c. A series of benzy-
lated sulfonamide analogues (14k-14p) were prepared, which
in general showed modest improvement compared to that
of 13a. As it was quite clear at this point that solubility was an
important parameter necessary for in vivo activity, basic
amino groups were included in our selections. The most
interesting compound in the group was 14l, which showed a
2-fold improvement over 13a in the in vitro enzymatic assay
and a 3-fold increase in the cellular assay. In addition, 14l
along with its regioisomers, 14m and 14n, showed an improve-
ment in aqueous solubility to nearly 0.01 g/L as a result of
forming a water-soluble salt, although in vitro metabolic
clearance was very rapid. Attempts were made to further
improve on solubility with a selection of neutral, basic, and
acidic solubilizing groups (15a-15d). Whereas neutral (15a)
and basic (15b) groups provided impressive in vitro activity,
aqueous solubility improved very little. However, acidic
groups suchas glycine(15c) and β-alanine(15d) demonstrated
improved intrinsic solubility (>0.57 g/L) at neutral pH,
generally good permeability, and slow in vitro clearance,
although they had only micromolar GK activity in vitro.
Because of the much improved aqueous solubility as their
potassium salts, these compounds were selected for in vivo
evaluation. Surprisingly, we found that 15c demonstrated
significant glucose lowering, although 15d did not. Whereas
this difference could be ascribed to the reduced in vitro
activity of 15d, compared to that of 15c, other unexplored
factors could also be playing a part, such as β-oxidation or
preferential clearance of 15d by glucuronidation. A group
Results and Discussion
Our approach focused on a series of sulfonamides begin-
ning with the simplest primary sulfonamide, 13a.14 This
compound was similar in activity to the external comparator,
2a (Figure 2), but also demonstrated good permeability and
sufficient stability in microsomes (Table 1) to warrant evalua-
tion in vivo. Oraladministrationof13a (50 mg/kg), produceda
modest though nonsignificant reduction in glucose AUC
during an OGTT. It is likely that poor aqueous solubility of
13a (<0.002 g/L) contributed to the low exposures observed
in vivo. A prodrug strategy was briefly investigated using short
chain carboxamides of 13a. Compounds 13b and 13c showed
improved water solubilityastheir sodiumsaltsthan13a (0.011
and 0.026 g/L, respectively). Subsequent evaluation of in vivo
exposure of 13b and 13c showed significantly improved
plasma (∼10-fold Cmax) and liver exposures (∼5-fold) of
the active entity, 13a, when dosed as a prodrug as compared to
that with the administration of the parent drug (13a)
(Figure 3). Whereas this improved exposure should have
translated into improved efficacy during an OGTT in mice,
unexpectedly, 13b showed no effect on glucose AUC. Because
of the inherent difficulties with advancing prodrugs through
development, such as species and patient variation in cleavage
extent and timing, as well as others, this approach was set
aside in preference to a traditional medicinal chemistry in-
vestigation of variations in the sulfonamide moiety. Early in
the investigation of this series, it became clear that exposure
would be the key hurdle and that it would be driven by
intrinsic aqueous solubility, gut dissolution rate, and perme-
ability.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19 6147
ofN-methyl-N-substituted amine variants werealso evaluated
and provided some improvement over monosubstituted sul-
fonamides. In particular, the 2-furanylmethyl analogue 16c
showed a 6-fold improvement in RKa over its counterpart
secondary amine, 14o. Also of note for 16a-c is the inverse
relationship observed between PAMPA permeability and
clearance from microsomes, which limited further exploration
into this series. Three compounds, 14l, 15a, and 16a, were
selected for evaluation to determine whether solubilizing
vehicles could overcome the perceived solubility limiting
exposure of these compounds. Besides the microemulsion16
vehicle, other vehicles such as polyvinylpyrrolidinone (K30)
and various combinations of DMI-ethanol-Tween-PEG
were evaluated, but these did not provide the improvement
hoped for. On the basis of the observation that disubstituted
sulfonamides provided improved intrinsic activity as com-
pared to their monosubstituted analogues, a small group
of cyclic compounds were prepared. As anticipated, com-
pounds 17a-17d provided around 10-fold improvement
in RKa over that of 13a. Of most interest was the N-methyl
piperazine adduct 17c, which showed significantly improved
enzymatic activity (RKa = 39 nM) and also increased glucose
turnover in hepatocytes by nearly 5-fold. Although 17c
exhibited only a modest improvement in kinetic solubility
(0.05 g/L), its corresponding HCl salt had aqueous solubility
greater than 50 g/L and therefore could be dosed as an
aqueous solution in vivo without extensive formulation ef-
forts. Isolation of R(-)17c, the pure R-enantiomer of 17c was
accomplished using chiral chromatography17 and exhibited
an RKa of ∼23 nM (approximately 40-fold more active than
its corresponding S-enantiomer). Although 17c has a poten-
tially racemizable center, it was found to be quite stable and
required a strong base (KOtBu) in order to racemize. This
racemization then allowed, after subsequent chromatogra-
phy, isolation of additional R(-)17c.
Figure 4, the L-shaped binding site which accommodates 2b
has expanded to a larger Y-shaped binding site in the R(-)17c
structure based on the movement of these two regions.
Specific interactions between R(-)17c and GK are shown
in Figure 5. If R(-)17c is dissected into three branches, the
pyridoaminothiazole branch makes hydrophobic interactions
with Val455, Ala456, and Lys459 of the R13 helix, as well as
Pro66 of connecting region I and Ile159 of the large domain.
There is also a bidentate hydrogen bond formed between the
thiazole N atom and the NH group of the amide in R(-)17c
with both the backbone NH and carbonyl oxygen of Arg63
on GK.
The methyl cyclopentane branch makes hydrophobic inter-
actions with residues Tyr214, Met210, Met235, and Val62 of
the large domain. The last branch, which contains the benze-
nesulfonylpiperazine, makes interactions in an expanded
groove, with similar directionality, as previously observed in
the crystal structure of GK in complex with the activator
Figure 4. Comparison of Banyu (2b) activator binding site with the
R(-)17c binding site.
We had initially planned to utilize the binding pocket from
the published X-ray from Banyu18 with their analogue 2b as a
model but were unable to dock our sulfonamides in this same
pocket while maintaining the amidine in what would be
expected to be the preferred orientation. We therefore set
out to cocrystallize R(-)17c, a compound with good activity
and more importantly with acceptable solubility for crystal-
lization studies, with GK protein. These efforts provided an
X-ray crystal structure of human GK in complex with glucose
˚
and the activator R(-)17c, determined to 2.0 A resolution.
The final structure was refined to an R-factor of 19.5%
(Table 3), and includes residues Gly11-Glu93, Gln98-
Cys461, one glucose molecule, one R(-)17c molecule, and
277 solvent molecules. Loop Gly94-Gly97 and the 4 C-
terminal amino acids were not modeled because of poor
electron density. The glucose binding site and the relationship
between the large and small domains is consistent with the
closed form of GK as reported by Banyu.
Figure 5. Binding site interactions for R(-)17c.
The activator R(-)17c is located in the activator binding
site, which is surrounded by the β1 strand and R5 helix of the
large domain, the C-terminal R13 helix of the small domain,
and the GK specific connecting region I (Ser64-Gly72). This
˚
site is approximately 20 A away from the glucose binding site.
In contrast to the Banyu binding cavity, the surface of this
activator binding site has been significantly expanded to allow
the binding of R(-)17c. This expansion is caused by a large
˚
motion of the loop Val91-Lys104 (>7 A rms for the CR of
Gln98) and a smaller motion of connecting region I of
˚
Ser64-Gly72 (∼2.5 A rms for the CR of Glu70). As seen in
Figure 6. Glucose AUC during OGTT in C57BL6 mice.

6148 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19
Bebernitz et al.
Figure 7. Liver and plasma exposure of 17d from oral administration of 17c.
Figure 8. Plasma glucose excursion and plasma insulin AUC during an intravenous glucose tolerance test (IVGTT) in DIO mice following
administration of R(-)17c. R(-)17c was administered once daily orally at 30 and 100 mg/kg for 4-weeks. Data are means ( SEM (n = 8 mice /
group). *p < 0.05, from vehicle control.
2c (Figure 2).19 The phenyl ring packs between Tyr214 of the
large domain and Val455 of the C-terminal R13 helix, with the
sulfonyl orienting the methyl piperazine into a newly formed
hydrophobic pocket which is surrounded by Leu451 and
Val455 of R13 helix, Tyr215, and Trp99. Additionally, there
is a hydrogen bond between the piperazine nitrogen and the
Oγ atom of Ser69 from the GK connecting region I. As has
been described, R(-)17c makes more extensive interactions
with the C-terminal R13 helix than has been previously seen.
These interactions withthe R13helix havebeenproposed tobe
the cause of activation by sterically inhibiting the release of
this helix from the small domain to form the super-open,
inactive conformation.18 Hence, when the activator is bound,
the enzyme remains in the open/closed catalytically active
conformation resulting in its higher affinity for glucose.
Evaluation of 17c in a standard OGTT in normal mice
resulted in a significant reduction of AUC for glucose excur-
sion as compared to that of controls (Figure 6). A manuscript
with the complete evaluation of the interesting pharmacology
of 17c has been submitted for publication.20 Although17c had
rapid in vitro metabolic clearance and was expected to be
rapidly cleared in vivo, it showed pharmacological effects in
vivo for up to 6 h. A possible explanation for this could be the
presence of a long-lived active metabolite. In silico tools
predicted N-demethylation of the piperazine as a likely meta-
bolite. This was confirmed by analysis of the metabolites of
17c following exposure to rat liver microsomes and was
further substantiated in PK studies, which indicated levels
of both 17c and its primary N-demethylated metabolite, 17d,
in the plasma and liver after oral administration of 17c
(Figure 7). Consistent with 17d showing both reduced in vitro
permeability and even faster metabolic clearance compared to
that of 17c, it was found to be inactive when delivered orally.
To further assess whether circulating concentrations of 17d
would be active at target in vivo, 17d was administered via an
intravenous route, and as expected, it showed efficacy at
lowering glucose AUC as did 17c delivered orally. As levels
of 17d in the liver were higher than 17c at 4 h and as 17d
demonstrated intrinsic activity comparable to that of 17c, it is
possible that the efficacy from oral dosing of 17c was extended
because of contributions from 17d.
The primary goal of this approach was to identify a liver
targeted compound which manifested its effects only in the
liver and showed no increase in insulin release from the β-cell
either in the basal state or during a glucose challenge. The
most straightforward way to target or at least prioritize the
liver would be by selecting an agent whose distribution vastly
favored the liver compared to the pancreas. Unfortunately,
R(-)17c actually showed higher exposures in the pancreas as
compared to the liver from PK studies performed in normal
mice (Table 2). As no changes in insulin output were observed
in the pancreas, it is unclear whether sufficient compound
concentrations actually reached the target β-cell, and no
attempt was made to determine the actual exposure of β-cells
to the compound. With regards to the metabolite17d, whereas
its accumulation in the liver may be responsible for a more
prolonged pharmacological response, 17d is unlikely to play a
Table 2. Plasma and Tissue Cmax and AUC for R(-)17c When Dosed at
60 mg/kg to Normal C57BL6 Micea
tissue
Cmax (μM)
AUC0-24h (μM*h)
plasma
liver b
4.3
52.4
511
44.5
218
34.6
pancreas
muscle
3228
167
a Data is from 3 animals with samples pooledfor analysis. b Exposures
of parent drug only.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19 6149
role in other tissues such as the β-cell because of its low
circulating plasma levels (Figure 7).
Whatismost interesting isthatregardless of the exposure of
R(-)17c in the whole pancreas, no indication of a β-cell effect
was observed in pharmacology studies examining insulin
output during a glucose challenge. The results of an intrave-
nous glucose tolerance test (IVGTT), designed to specifically
examine the effects of a compound on the β-cell component of
glucose disposal following a challenge is shown in Figure 8.
Administration of R(-)17c to diet induced obese mice (DIO)
resulted in a dose-dependent decrease in glucose levels
(glucose AUC) over the course of the experiment. Over the
same time, reductions in insulin AUC were also observed,
indicating decreased pancreatic output and the fact that the
effects of R(-)17c on glucose lowering are not being derived
from an effect in the β-cell but rather is the result of im-
proved insulin sensitivity as marked by improved glucose
uptake and reduced glucose output from the liver. In separate
longitudinal studies, it has also been shown that basal insulin
levels are unchanged or reduced while glucose levels trend
downward to normal levels over time due to drug treat-
ment.20 In summary, the GK activator 17c has clearly demon-
strated efficacy on glucose lowering in vivo primarily through
its effects in the liver without any observable impact in the
β-cells.
Figure 9. Matrix evaluation of 17c.
whether this class of GK activators will be efficacious to treat
diabetes as a stand-alone therapy.
Experimental Section
Anhydrous solvents were obtained from commercial sources
and were used as is. Organic solutions were dried over MgSO4.
Flash chromatography was performed on a Biotage SP4 system.
Proton NMR spectra were recorded on a 400 MHz Bruker AV-
400 spectrophotometer using TMS (δ 0.0 PPM) as internal
standard. Analytical and preparative chromatography were per-
formed on a Shimadzu HPLC equipped with Phenomenex Luna
C8(2) columns using from 5 to 100% acetonitrile/water both
modified with 0.5% TFA. LC-MS was performed on a Water
ZQ equipped with a Intersil ODS-3 (C8) column. Microanalysis
was performed at Robertson Microlit Laboratories, Madison,
NJ. All final compounds had purity >95% as determined by
analytical HPLC and elemental analysis.
Biological Methods. Matrix Evaluation of Enzyme Activity.
Recombinant human liver GK (amino acids 17-466) was
expressed as an N-terminal histidine (His)-tagged fusion protein
in E. coli strain BL21 Star (DE3) and was purified by nickel
affinity chromatography. The His-tag was then cleaved off
leaving an N-terminal GPGSM extension and the protein
further purified using mono Q-anion exchange chromatography
followed by Superdex S200 gel filtration (GE Healthcare)
to >95% purity. The protein was diluted in a buffer containing
50 mM Tris-HCl (pH 7.5), 200 mM KCl, 5 mM DTT, 50 mM
glucose, and 20% glycerol. The GK enzyme assay was per-
formed in a 96-well plate with each sample run in triplicate.
Glucose and compound were titrated in a two-dimensional
matrix using varying concentrations of each: 12 glucose con-
centrations (0 and 168 μM to 60 mM) and 8 compound
concentrations (0 and 11 nM to 20 μM) were tested. Data was
expressed as milli optical density (mOD) per min and were
analyzed by fitting the observed rates at various concentrations
of substrate and activator. A representative run for 17c is shown
in Figure 9.
Conclusions
We have optimized a series of sulfonamides with the
generation of a number of GK activators with low nano-
molar RKa. A selection of these compounds additionally
exhibited improved permeability, reduced metabolic clear-
ance rates, and significant activation of glucose phosphor-
ylation in isolated hepatocytes. As most of the early
compounds required special formulation vehicles in order
to observe any in vivo activity, the incorporation of solubi-
lity handles was critical to this series. The key advancement
came with 17a-17d, which combined good intrinsic activity
with significant cell activity and led ultimately to water-
soluble compounds as exemplified by 17c. Administration
of a solution of 17c dissolved in water during an OGTT
provided a dramatic flattening of the glucose excursion
curve and significant in vivo activity. A cocrystal of 17c with
the GK enzyme showed an expanded binding cavity as
compared to that of 2b and additionally predicted a novel
interaction of the piperazine nitrogen with Ser69 which
could be used to design further analogues.
Activation of GK in both the β-cell and liver would address
many of the defects inherent in type 2 diabetics by reducing
liver glucose output as well as increasing the insulin response
to a glucose load. However, our concern was that direct effects
of a compound in the pancreas would not only further stress
the β-cell, leading to early β-cell failure, but also increase the
propensity for hypoglycemia. This insulin-induced hypogly-
cemia could be especially dangerous in later stage diabetics
who suffer from defective glucose sensing and impaired
counterregulatory response. During this effort, we selected
for compounds which demonstrated their effect only in the
liver and either had no effect in the pancreas or actually
reduced insulin levels under fasting conditions as well as
during an OGTT. As a result of this effort, R(-)17c was
identified as a highly water-soluble GK activator which
manifested its activity primarily in the liver. Evaluation of
functional liver selective agents in the clinic should address
GK Activity in Primary Hepatocytes. The rate of glucose
metabolism by hepatocytes is dependent on the GK activity
levels. Effects on glucose phosphorylation can be measured by
3
the release of H20 from [2-3H]glucose in isolated rat hepato-
cytes.21 Hepatocytes were isolated by collagenase perfusion of
the liver of overnight-fasted male Sprague-Dawley rats
(Charles River Laboratories, Wilmington, MA) as previously
described.22 Cells were cultured in 24-well plates at 37 ꢀC in a
humidified atmosphere containing 5% CO2 in serum-free
DMEM medium containing 5 mM glucose and 10 mM dexa-
methasone for 16-20 h prior to use. Glucose phosphorylation
was determined by the release of 3H20 from [2-3H]glucose
essentially as previously described.23 Assays were run in quad-
ruplicate. Briefly, cells were preincubated in serum-free DMEM
containing 5 mM glucose and DMSO (control) or 30 μM
compound for 3 h at 37 ꢀC, followed by a 30 min incubation
with [2-3H]glucose. On termination of the incubations, the
medium was collected into 0.1 M HCl. The amount of glucose

6150 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19
Bebernitz et al.
detritiated in nmol/h per 106 cells was determined, and the data
(mean ( SEM) was expressed as percent increase over the
vehicle control. Differences were considered statistically signifi-
cant by two-tailed Student’s t-test (p < 0.05).
Table 3. Crystallographic Data and Refinement Information for
R(-)17c
parameters
space group
unit cell
R(-)17c
P212121
a = 45.5 A
Effect on Glucose Excursion during an OGTT. Fifteen week
old male C57BL6 mice (n = 8/group), housed two per cage with
normal chow diet and reversed light cycle, were fasted for 3 h.
Body weights were then measured and 0-h blood samples
taken. Plasma glucose values were determined using a glucose
meter, and plasma samples (10 μL) were saved for insulin
assay. Animals were grouped with means of plasma glucose
matched. Vehicle (microemulsion) or compounds (in micro-
emulsion) at doses of 50 mg/kg for the reference compound,
2a, or 72 mg/kg (a molar equivalent dose to 2a) for 17c were
administered by oral gavage in a volume of 5 mL/kg. Glucose
was administered orally 30 min after vehicle/compound admin-
istration at 1 g/kg (5 mL/kg). Blood samples were taken at 0.5, 1,
2, and 4 h post-dosing for glucose and insulin measurements.
Effect on Glucose Excursion during IVGTT in DIO Mice.
Male C57BL6 mice maintained on high fat diet were adminis-
tered orally with vehicle (dH2O) or compound at 30 to 100 mg/
kg dose once daily for 8 weeks. An intravenous glucose tole-
rance test was conducted after 8-weeks of treatment, and
plasma glucose excursion (left panel) and plasma insulin AUC
(0-120 min; right panel) were calculated.
˚
˚
b = 90.3 A
˚
c = 116.2 A
R = b = γ = 90ꢀ
29.06-2.00
32450
˚
resolution range (A)
total observations
unique reflections
completeness (%)a
I/σa
6490
96.8 (82.1)
8.1 (1.7)
0.106 (0.413)
0.195/0.245
3493
a,b
Rsym
Rcryst/Rfree
protein atoms
c
heterogen atoms
solvent molecules
average B-factor (A )
49
277
2
˚
29.0
rms deviations from ideal values
˚
bond lengths (A)
0.010
1.5
bond angle (ꢀ)
a Numbers in parentheses are for the highest resolution shell. b Rsym
=
Σ|Ih - ÆIhæ/ΣIh over all h, where Ih is the intensity of reflection h. c Rcryst
and Rfree = Σ||Fo| - |Fc||/Σ|Fo|, where Fo and Fc are observed and
calculated amplitudes, respectively. Rfree was calculated using 5% of
data excluded from the refinement.
X-ray Crystallization. Human GK (17-466) with an N-term-
inal GPGSM extension from cloning was concentrated to
10 mg/mL in 20 mM Tris, pH 7.5, 50 mM NaCl, and 5 mM
TCEP and incubated for 3 h at 4 ꢀC with a final concentration of
20 mM glucose and 300 μM R(-)17c. The GK/glucose/R(-)17c
complex was centrifuged for 30 min at 4 ꢀC at 20,000g, and
the supernatant was used for crystallization. Three hundred
nanoliters of complex was mixed with 300 nL of well solution
(100 mM HEPES, pH 6.0, 20% PEG 3350, and 200 mM NaCl)
and incubated at 4 ꢀC. Hexagonal rod shaped crystals were
observed at approximately 30 days. The crystals were cryopro-
tected with the addition of 20% glycerol and 100 μM R(-)17c
for 5 min at 4 ꢀC prior to flash freezing in liquid nitrogen.
Structure Determination. Diffraction data for the GK/glu-
cose/R(-)17c complex were collected on a Saturn 92 CCD
room temperature (RT) over 2 h and then quenched onto ice,
and extracted with ethyl acetate. The organic layer was dried
over magnesium sulfate, filtered, and concentrated to afford a
crude yellow oil (32.6 g, 81% yield) consisting of 3 as a mixture
of regio-isomers. This mixture was used directly in the next step.
[4-(4-Methyl-piperazine-1-sulfonyl)-phenyl]-acetic Acid Ethyl
Ester (4). To a cooled solution of (3) (32.6 g, 124 mmol) in
200 mL of dichloromethane was added dropwise a solution
1-methyl piperazine (15 g, 150 mmol) and DIPEA (30 mL) in
200 mL of dichloromethane. The reaction mixture was stirred at
RT for 3 h and then concentrated under reduced pressure. The
residue was partitioned between ethyl acetate and saturated
brine solution. The organic layer was dried over magnesium
sulfate, filtered, and concentrated to afford 18 g of a crude
yellow oil. The oil was recrystallized from MTBE, separating
out the 3-regioisomer to afford pure 4 as white crystals in 25%
yield. 1H NMR (400 MHz, CDCl3) δ 1.3 (t, J = 7.4, 3H), 2.3 (s,
3H), 2.5 (m, 4H), 3.0 (br s, 4H), 3.7 (s, 2H), 4.2 (q, J = 7.4, 2H),
7.4 (d, J = 8.3, 2H), 7.7 (d, J = 7.3, 2H); MS 327 [M þ 1]þ.
3-Cyclopentyl-2-[4-(4-methyl-piperazine-1-sulfonyl)-phenyl]-
propionic acid ethyl ester (5). A solution of 4 (15 g, 0.046 mol) in
a mixture of THF (60 mL) and DMTP (10 mL) was cooled to
-78 ꢀC under nitrogen. The resulting solution was stirred at
-78 ꢀC for 45 min and to this was added LDA (25.6 mL, 6.40 g,
0.059 mol, 25% solution in THF/Hexane). A solution of
iodomethylcyclopentane (11.60 g, 0.055 mol) in a mixture of
DMTP (12 mL) and THF (20 mL) was added over a period of 15
min at -78 ꢀC and reaction mixture stirred at -78 ꢀC for 3 h
further, followed by stirring at 25 ꢀC for 12 h. The reaction
mixture was then quenched by the dropwise addition of satu-
rated aqueous ammonium chloride solution (50 mL) and is
concentrated in vacuo. The residue was diluted with water
(50 mL) and extracted with ethyl acetate (3 ꢀ 100 mL). The
organic solution was washed with a saturated aqueous sodium
chloride (2 ꢀ 150 mL), dried over sodium sulfate, filtered and
concentrated in vacuo. Column chromatography over silica gel
(60 - 120 mesh), using 50% ethyl acetate in hexane as an eluent
˚
detector using Cu KR radiation (λ = 1.5418 A) from a Rigaku
FR-E microfocus rotating anode X-ray generator at 45 kV and
45 mA, with Varimax optics. The data were measured from a
single crystal, and the reflections were indexed, integrated, and
scaled using the HKL2000 package.24 The space group of the
GK/glucose/R(-)17c complex was P212121 with 1 complex in
the asymmetric unit. Molecular replacement was performed
using PHASER,25 with the initial model of GK/glucose/
2-amino-4-fluoro-5-[(1-methyl-1h-imidazol- 2-yl)sulfanyl]-N-
(1,3-thiazol-2-yl)benzamide complex (PDB entry 1V4S) with
water, sodium ion, glucose, and inhibitor molecules removed.
Structure determination was achieved through iterative cycles of
both positional and simulated annealing refinement using
CNX26 and model building using COOT.27 Individual B-factors
were refined using an overall anisotropic B-factor refinement
along with bulk solvent correction. The glucose, sodium ion,
solvent and R(-)17c molecules were built in later rounds of
refinement. Data collection and refinement statistics are shown
in Table 3.
Chemical Methods. Representative compounds are used to
provide experimental details of Method A and Method B. Struc-
tural identification for additional compounds prepared using
these methods may be found in the Supporting Information.
Synthesis Using Method A with 17c As an Example. 3-Cyclo-
pentyl-N-(5-methoxy-thiazolo[5,4-b]pyridin-2-yl)-2-[4-(4-methyl-
piperazine-1-sulfonyl)-phenyl]-propionamide (17c). (4-Chloro-
sulfonyl-phenyl)-acetic Acid Ethyl Ester (3). To a cooled solution
of chlorosulfonic acid (150 mL) was added dropwise a solution
of phenylacetic acid ethyl ester (25 g, 152 mmol) in dichloro-
methane (25 mL). The reaction mixture was allowed to warm to
1
afforded 5 as a white solid in 70% yield. H NMR (400 MHz,
CDCl3) δ 0.9-2.1 (m, 11H), 1.2 (t, J = 7.1, 3H), 2.3 (s, 3H), 2.5
(br s, 4H), 3.0 (br s, 4H), 3.6 (m, 1H), 4.1 (q, J = 7.1, 2H), 7.5
(d, J = 8.3, 2H), 7.7 (d, J = 8.3, 2H); MS 409 [M þ 1]þ.

Article
3-Cyclopentyl-2-[4-(4-methyl-piperazine-1-sulfonyl)-phenyl]-
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19 6151
hexane/MTBEafforded 8 as an orangeoil in66%yield. 1H NMR
(400 MHz, CDCl3) δ 1.0-1.1 (m, 2H), 1.2 (t, 3H, J = 7.2),
1.4-1.8 (m, 5H), 1.8-1.9 (m, 2H), 2.1-2.25 (m, 2H), 3.74 (t, 1H,
J = 7.8), 4.1 (m, 2H), 7.51 (d, 2H, J = 8.8), 8.19 (d, 2H, J = 8.8);
LC/MS 290 (M - 1).
Propionic Acid (6). A solution of 5 (14 g, 0.034 mol) in methanol/
water (30 mL/10 mL) and sodium hydroxide (4.11 g, 0.10 mol)
was stirred at 60 ꢀC for 8 h in an oil bath. The methanol was
then removed in vacuo at 45-50 ꢀC. The residue was diluted
with water (25 mL) and extracted with ether (1 ꢀ 40 mL).
The aqueous layer was acidified to pH 5 with 3 N aqueous
hydrochloric acid solution. The precipitated solid was collected
by vacuum filtration and washed with water (20 mL), followed
by isopropyl alcohol (20 mL). Finally, the solid cake was washed
with 100 mL of hexane and dried under vacuum at 40 ꢀC for 6 h
3-Cyclopentyl-2-(4-nitro-phenyl)-propionic Acid (9). Com-
pound 8 (3.6 g, 12.3 mmol) was dissolved in 25 mL of methanol,
aqueous NaOH (0.70 g, 17.5 mmol in 4 mL of water) was added,
and the mixture was stirred at RT overnight. The methanol was
removed under reduced pressure, and the residue was diluted
with 100 mL of water and extracted with ether. The aqueous
layer was then acidified with 1 N HCl and then extracted with
ethyl acetate. The combined ethyl acetate layers were dried over
anhydrous magnesium sulfate, filtered, and reduced under
vacuum to a crude orange oil. The crude oil was triturated with
100 mL of hexane/10-15 mL of ether to produce 9 as a solid in
1
to give 6 as a white solid in 85% yield. H NMR (400 MHz,
CDCl3) δ 1.1-2.0 (m, 11H), 2.4 (s, 3H), 2.7 (br s, 4H), 3.1 (br s,
4H), 3.6 (m, 1H), 7.5 (d, J = 8.3, 2H), 7.6 (d, J = 8.3, 2H); MS
381 [M þ 1]þ.
5-Methoxy-thiazolo[5,4-b]pyridin-2-ylamine (7). A solution of
6-methoxy-pyridin-3-ylamine (5.0 g, 0.0403 mol) in 10 mL of
acetic acid was added slowly to a solution of potassium thio-
cyanate (20 g, 0.205 mol) in 100 mL of acetic acid at 0 ꢀC
followed by the addition of a solution of bromine (2.5 mL,
0.0488 mol) in 5 mL of acetic acid. The reaction was stirred for
2 h at 0 ꢀC and then allowed to warm to RT. The resulting solid
was collected by filtration and washed with acetic acid, then
partitioned between ethyl acetate and saturated aqueous sodium
bicarbonate. The insoluble material was removed by filtration,
and the organic layer was evaporated and dried to afford 7 as a
1
72% yield. H NMR (400 MHz, CDCl3) δ 1.0-1.1 (m, 2H),
1.4-1.8 (m, 5H), 1.8-1.9 (m, 2H), 2.1-2.25 (m, 2H), 3.74 (t, 1H,
J = 7.8), 7.51 (d, 2H, J = 8.8), 8.19 (d, 2H, J = 8.8); LC/MS 218
(-CO2, M - 1), 279 (M þ NH4þ).
3-Cyclopentyl-N-(5-methoxy-thiazolo[5,4-b]pyridin-2-yl)-2-(4-
nitro-phenyl)-propionamide (10). Compound 9 (7.5 g, 28.5 mmol)
was dissolved in 25 mL of thionyl chloride and a drop of DMF
and the mixture stirred at RT for 5-6 h. The excess of thionyl
chloride was removed under reduced pressure. The residue was
then taken up in DCM and added dropwise to a solution of 7
(5.2 g, 28.5 mmol) in 25 mL of pyridine. The reaction mixture
was stirred for 5 h before being evaporated to remove the
pyridine. The residue was partitioned between ethyl acetate
and brine, and extracted with ethyl acetate. The combined
organic layers were washed with saturated sodium bicarbonate,
brine, dried over anhydrous magnesium sulfate, filtered, and
then reduced to an orange-brown solid. This was then vacuum-
dried to afford 10 as a foam in 95% yield. 1H NMR (400 MHz,
CDCl3) δ 1.0-1.1 (m, 2H), 1.4-1.8 (m, 5H), 1.8-1.9 (m, 2H),
2.1-2.25 (m, 2H), 3.6 (t, 1H, J = 7.8), 4.01 (s, 3H), 6.8 (d, 1H,
J = 8.8), 7.4 (d, 2H, J = 8.6), 7.8 (d, 1H, J = 8.8 Hz), 8.19 (d,
2H, J = 8.6 Hz), 9.3 (s, 1H); LC/MS 427 (M þ 1).
1
yellow solid in 88% yield. H NMR (400 MHz, DMSO-d6) δ
ppm 3.83 (s, 3 H) 6.69 (d, J = 8.72 Hz, 1 H) 7.44 (s, 2 H) 7.61 (d,
J = 8.59 Hz, 1 H); LC-MS (M þ H = 182.2, M - H = 180.2)
3-Cyclopentyl-N-(5-methoxy-thiazolo[5,4-b]pyridin-2-yl)-2-[4-
(4-methyl-piperazine-1-sulfonyl)-phenyl]-propionamide (17c). A
solution of 6 (5 g, 0.013 mol) in DCM (250 mL) was cooled to
0 ꢀC and then charged with HOBt hydrate (2.66 g, 0.019 mol),
followed by EDCI hydrochloride (6 g, 0.031 mol). The reaction
mixture was stirred at 0 ꢀC for 5 h. After that, a solution of
7 (2.36 g, 0.013 mol) and DIEA (8 mL, 0.046 mol) in a mixture of
DCM (60 mL) and DMF (20 mL) was added dropwise over 30
min. Reaction temperature was maintained at 0 ꢀC for 3 h, then
at RT (28 ꢀC) for 3 days. The reaction was then diluted with 60
mL of water, and the organic layer was separated and washed
with saturated sodium bicarbonate solution (2 ꢀ 50 mL)
followed by washing with water (2 ꢀ 50 mL) and saturated
sodium chloride aqueous solution (1 ꢀ 150 mL). Finally, the
organic layer was dried over sodium sulfate, filtered, and
evaporated. The crude product was purified using column
chromatography over silica gel (60-120 mesh), using 40% ethyl
acetate in hexane as an eluent to afford 17c as a white solid in
65% yield. 1H NMR (400 MHz, CDCl3) δ 0.9-2.1 (m, 11H), 2.2
(s, 3H), 2.5 (br s, 4H), 3.1 (br s, 4H), 3.7 (m, 1H), 4.0 (s, 3H),
6.8 (d, J = 8.8, 1H), 7.5 (d, J = 8.3, 2H), 7.7 (d, J = 8.3, 2H), 7.8
(d, J=8.8, 1H), 8.6 (s, 1H); MS 617 [M þ 1]þ.) Anal. (C26H33-
N5O4S2) C,H,N.
Synthesis Using Method B with 13a As an Example. 3-Cyclo-
pentyl-N-(5-methoxy-thiazolo[5,4-b]pyridin-2-yl)-2-[4-(4-methyl-
piperazine-1-sulfonyl)-phenyl]-propionamide (13a). 3-Cyclopen-
tyl-2-(4-nitro-phenyl)-propionic Acid Ethyl Ester (8). To a 1 L
round-bottom flask containing 250 mL of 9:1 THF/DMPU
at -78C were added under nitrogen 11 mL (78.6 mmol) anhy-
drous DIEA followed byrapid addition of32mLof2.5 M n-BuLi
in hexanes. After 10 min at -78 ꢀC, a solution of 15.4 g (74 mmol)
of p-nitrophenylacetic acid ethyl ester in 100 mL of 9:1 THF/
DMPU was added dropwise over 30 min. A deep purple solution
resulted, and the reaction mixture was stirred at -78 ꢀC for
30 min, and then cyclopentyl methyl iodide (17.6 g, 78 mmol) in
50 mL of 9:1 THF/DMPU was added. The reaction was stirred
while warming slowly to RT overnight. The mixture was poured
into 1 L of 1 N HCl and extracted twice with MTBE.
The combined MTBE extracts were washed with brine, dried
over anhydrous magnesium sulfate, filtered, and reduced to an
orange oil. Flash chromatography over silica eluting with 4:1
2-(4-Amino-phenyl)-3-cyclopentyl-N-(5-methoxy-thiazolo[5,4-
b]pyridin-2-yl)-propionamide (11). Compound 10 (12 g, 28.2
mmol) was diluted with 160 mL of ethanol and 150 mL of acetic
acid. Eight grams of iron powder (325 mesh, 0.14 mol) was
added and the mixture heated to reflux. Once reflux began, the
mixture was stirred vigorously, and heating was discontinued
and the mixture allowed to cool slowly. The solvents were
removed, and the residue was treated with 250 mL of water.
Saturated sodium bicarbonate was then added to bring the
mixture to a pH of 8-9. The mixture was extracted with ethyl
acetate, washed with brine, dried, and evaporated to give an
orange solid which was triturated from hexane. The resulting
1
solid was collected by filtration to afford 11 in 93% yield. H
NMR (400 MHz, CDCl3) δ 1.0-1.1 (m, 2H), 1.4-1.8 (m, 5H),
1.8-1.9 (m, 2H), 2.1-2.25 (m, 2H), 3.6 (t, 1H, J = 7.8), 3.98
(s, 3H), 6.7 (d, 1H, J = 8.8), 6.8 (d, 2H, J = 8.6), 7.2 (d, 2H, J =
8.6), 7.8 (d, 1H, J = 8.8); LC/MS 397 (M þ 1).
4-[2-Cyclopentyl-1-(5-methoxy-thiazolo[5,4-b]pyridin-2-ylcar-
bamoyl)-ethyl]-benzenesulfonyl chloride (12). Compound 11
(2.0 g, 5.1 mmol) was dissolved in 50 mL of acetic acid and
20 mL of concentrated HCl, and the mixture was cooled to 0 ꢀC.
A solution of 0.35 g (5.1 mmol) of NaNO2 in 5 mL of water was
added dropwise, and the mixture was stirred for 30 min. The
resulting yellow solution was then added to 180 mL of a solution
prepared by bubbling 74 g of sulfur dioxide gas into 740 mL of
glacial acetic acid followed by the addition of 30 g of CuCl2 in
35-40 mL of water. The resulting mixture was filtered through
filter paper to obtain a clear green solution, and the mixture was
stirred at RT overnight. The resulting mixture was poured onto
500 g of ice, and the precipitated solids were collected by
filtration, washed with water and then dissolved in ethyl acetate,
washed with brine, dried over anhydrous magnesium sulfate,

6152 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19
Bebernitz et al.
filtered, and evaporated to afford a yellow foam. This material
was flash chromatographed over silica eluting with 7:3 hexane/
ethyl acetate to afford 12 as a stable yellow foam in 75% yield.
1H NMR (400 MHz, CDCl3) δ 1.0-1.1 (m, 2H), 1.4-1.8 (m,
5H), 1.8-1.9 (m, 2H), 2.1-2.25 (m, 2H), 3.7 (t, 1H, J = 7.8),
4.01 (s, 3H), 6.8 (d, 1H, J = 8.8), 7.5 (d, 2H, J = 8.6), 7.8 (d, 1H,
J = 8.8), 8.19 (d, 2H, J = 8.6), 9.3 (s, 1H); LC/MS 480 (M þ 1).
3-Cyclopentyl-N-(5-methoxy-thiazolo[5,4-b]pyridin-2-yl)-2-
[4-(4-methyl-piperazine-1-sulfonyl)-phenyl]-propionamide (13a).
Compound 12 (210 mg, 0.4374 mmol) was dissolved in acetone
(2 mL) and added to a solution of ammonium carbonate (168
mg, 1.7496 mmol) in water (5 mL). The reaction was stirred at
RT for 1 h and then partitioned between 1 N hydrochloric acid
and ethyl acetate. The organic layer was separated, dried over
anhydrous magnesium sulfate, filtered, and concentrated to
afford the desired product as an off white foam (198 mg, 98%
(9) Agius, L.; Peak, M.; Newgard, C. B.; Gomez-Foix, A. M.; Guino-
vart, J. J. Evidence for a role of glucose-induced translocation of
glucokinase in the control of hepatic glycogen synthesis. J. Biol.
Chem. 1996, 271, 30479–30486.
(10) Avdeef, A. The rise of PAMPA. Expert Opin. Drug Metab. Toxicol.
2005, 1 (2), 325–342; compounds are measured at 5-10 μM in 96-well
filter plates coated with a hexadecane membrane. Transport is measured
at 3 pH values in line with the gastrointestinal tract, and permeability is
quantified by LC/MS .
(11) log PGI values of <4.4 correspond to a high rating (estimated
> 75% absorbed). log PGI values of >4.5 and <5 correspond to a
medium rating (estimated 35-75% absorbed). log PGI values of
>5.1 correspond to a low rating (estimated <35% absorbed).
(12) Zhou, L.; Yang, L.; Tilton, S.; Wang, J. Development of a high
throughput equilibrium solubility assay using miniaturized shake-
flask method in early drug discovery. J. Pharm. Sci. 2007, 96 (11),
3052–3071; the HT-equilibrium solubility assay utilizes a novel min-
iaturized shake-flask approach and streamlined HPLC analysis, and is a
fast, reliable, and cost-effective screening tool for solubility assessment
in early drug discovery.
1
yield). H NMR (400 MHz, DMSO-d6) δ ppm 1.14 (dt, J =
19.96, 8.08 Hz, 2 H) 1.07 - 1.20 (m, 1 H) 1.44 (dd, J = 7.20, 4.67
Hz, 2 H) 1.51 - 1.62 (m, 2 H) 1.57 (d, J = 4.55 Hz, 2 H) 1.66 - 1.76
(m, 1 H) 1.70 (d, J = 7.33 Hz, 1 H) 1.83 (ddd, J = 13.58, 7.07,
6.88 Hz, 1 H) 3.90 (s, 3 H) 4.05 (t, J = 7.58 Hz, 1 H) 6.90 (d, J =
8.84 Hz, 1 H) 7.30 (s, 2 H) 7.59 (d, J = 8.34 Hz, 2 H) 7.81 (d, J =
8.59 Hz, 2 H) 8.02 (d, J = 8.59 Hz, 1 H) 12.63 (s, 1 H); MS e/z
(ES) 461.58 (M þ 1þ, 100%). Anal. (C21H24N4O4S2 0.5EtOAc)
C, H, N.
(13) Lau, Y. Y.; Krishna, G.; Yumibe, N. P.; Grotz, D. E.; Sapidou, E.;
Norton, L.; Chu, I.; Chen, C.; Soares, A. D.; Lin, C. C. The use of in
vitro metabolic stability for rapid selection of compounds in early
discovery based on their expected hepatic extraction ratios. Pharm.
Res. 2002, 19, 1606–1610; metabolic clearance using liver microsomes
with or without specific enzyme cofactors (NADPH, UDPGA). Com-
pounds are incubated at 1 μM in 96-well assay plates with liver
microsomes þ selected cofactors. Clearance is measured at 4 time
points, and % test compound remaining vs. time is used to calculate
clearance.
(14) Racemic materials were employed including racemic 2a for com-
Supporting Information Available: Low resolution MS, MP,
1H NMR, and elemental analysis are provided for target
compounds. This material is available free of charge via the
Internet at http://pubs.acs.org.
parison.
(15) Equilibrium solubility for all compounds in this article were
<0.003 g/L unless otherwise stated.
(16) Microemulsion vehicle is defined as Cremophor RH40 (43%), corn
oil-monodiglyceride (35.75), propylene glycol (10.6%), and
ethanol(10.6%). This vehicle is then diluted with water in a 2-part
microemulsion/3 part water as the compound is dissolved in it.
(17) Separation was accomplished on a proprietary chiral column
similar to the ChiralPak IA column using simulated moving bed
chromatography with 50:40:10 n-hexane/ethyl acetate/ methanol.
(18) Kamata, K.; Mitsuya, M.; Nishimura, T.; Eiki, J.; Nagata, Y.
Structural basis for allosteric regulation of the monomeric allos-
teric enzyme human glucokinase. Structure 2004, 12, 429–438.
(19) Efanov, A. M.; Barrett, D. G.; Brenner, M. B.; Briggs, S. L.;
Delaunois, A.; Durbin, J. D.; Giese, U.; Guo, H.; Radloff, M.; Gill,
G. S.; Sewing, S.; Wang, Y.; Weichert, A.; Zaliani, A.; Gromada, J.
A novel glucokinase activator modulates pancreatic islet and
hepatocyte function. Endocrinology 2005, 146, 3696–3701.
(20) Duttaroy, A.; Caplan, S.; Bebernitz, G.; Chau, M.; Crowe, H.;
Dubiel, D.; Gao, J.; Joly, E.; Kavana, M.; Madiraju, M.; Taslimi,
P.; Zhang, X.; Prentki, M.; Gromada, J. The novel glucokinase
activator LBX192 stably reverses diabetes and obesity, prevents
liver steatosis and increases energy expenditure in diet-inducted
obese mice, Diabetes, submitted.
(21) Katz, J.; Wals, P. A.; Rognstad, R. Glucose phosphorylation,
glucose-6-phosphatase, and recycling in rat hepatocytes. J. Biol.
Chem. 1978, 253, 4530–4536.
(22) Berry, M. N.; Friend, D. S. High-yield preparation of isolated rat
liver parenchymal cells: a biochemical and fine structural study.
J. Cell Biol. 1969, 43, 506–520.
(23) Agius, L.; Stubbs, M. Investigation of the mechanism by which
glucose analogues cause translocation of glucokinase in hepato-
cytes: evidence for two glucose binding sites. Biochem. J. 2000, 346,
413–421.
References
(1) Matschinsky, F. M. Assessing the potential of glucokinase activa-
tors in diabetes therapy. Nat. Rev. Drug Discov. 2009, 8, 399–416.
Sarabu, R.; Berthel, S. J.; Kester, R. F. K.; Tilley, J. W. T. Glucokinase
activators as new type 2 diabetes therapeutic agents. Expert Opin.
Ther. Pat. 2008, 18, 759–768. Grimsby, J.; Berthel, S. J.; Sarabu, R.
Glucokinase activators for the potential treatment of type 2 diabetes.
Curr. Top. Med. Chem. 2008, 8, 1524–1532.
(2) Van Schaftingen, E.; Veigha da Cunha, M. In Frontiers in Diabetes;
Matschinsky, F. M., Magnuson, M. A., Eds.; Karger Publishers: Basel,
Switzerland, 2004; Vol. 16, pp 193-207.Van Shaftingen, E. Short-term
regulation of glucokinase. Diabetologia 1994, 37, S43–47.
(3) Grimsby, J.; Sarabu, R.; Corbett, W. L.; Haynes, N. E.; Bizzarro,
F. T.; Coffey, J. W.; Guertin, K. R.; Hilliard, D. W.; Kester, R. F.;
Mahaney, P. E.; Marcus, L.; Qi, L.; Spence, C. L.; Tengi, J.;
Magnuson, M. A.; Chu, C. A.; Dvorozniak, M. T.; Matschinsky,
F. M.; Grippo, J. F. Allosteric activators of glucokinase: Potential
role in diabetes therapy. Science 2003, 301, 370. Grimsby, J.;
Matschinsky, F. M.; Grippo, J. F. In Frontiers in Diabetes;
Matschinsky, F. M., Magnuson, M. A., Eds.; Karger Publishers: Basel,
Switzerland, 2004; Vol. 16, pp360-378.
(4) Utilizing standard chlorosulfonic acid conditions (-0 ꢀC to RT)
gave the para/meta mixture, while reaction at lower temperatures
(-78 ꢀC) resulted in only sulfonic acid, which upon warming
resulted in the same 80:20 mixture. Heating the reaction to 75 ꢀC
as well did not alter the product ratio but instead decomposed.
Fluorosulfonic acid as well as standard nitration methods gave
identical outcomes.
(24) Otwinowski, Z.; Minor, W. Processing of X-ray Diffraction Data
Collected in Oscillation Mode. Methods in Enzymology; Carter,
C.W., Jr., Sweet, R. M., Eds., Academic Press: New York, 1985; Vol.
276, Macromolecular Crystallography, part A, pp 307-326.
(25) Storoni, L. C.; McCoy, A. J.; Read, R. J. Likelihood enhanced
fast rotation functions. Acta Crystallogr., Sect. D 2004, 59, 1145–
1115.
(5) Aono, T.; Kishimoto, S.; Araki, Y.; Noguchi, S. Indancarboxylic
acids, I. Electrophilic substitution reactions of 1-indancarboxylic
acid and synthesis of 6-substituted 1-indancarboxylic acids as
potential antiinflammatory agents. Chem. Pharm. Bull. 1977, 25
(12), 3198.
(6) Morgan E. D. Synthesis of p-alkylphenylacetic acids. Tetrahedron
1967, 23, 1735-1738; o:m:p ratio 4:38:58.
(26) Brunger, A. T.; Adams, P. D.; Clore, G. M.; DeLano, W. L.; Gros,
P.; Grosse-Kunstleve, R. W.; Jiang, J. S.; Kuszewski, J.; Nilges, N.;
Pannu, N. S.; Read, R. J.; Rice, L. M.; Simonson, T.; Warren, G. L.
Crystallography and NMR system (CNS): A new software system
for macromolecular structure determination. Acta Crystallogr.,
Sect. D 1998, 54, 905–921.
(7) Hajipour, A. R.; Ruoho, A. E. Nitric acid in the presence of P2O5
supported on silica gel- a useful reagent for the nitration of
aromatic compounds under solvent-free conditions. Tetrahedron
Lett. 2005, 46, 8307–8310.
(8) Segal I. H. Enzyme Kinetics: Behavior and Analysis of Rapid
Equilibrium and Steady State Enzyme Systems. In Rapid Equilib-
rium Partial and Mixed-Type Inhibition, Wiley classics library ed.,
Wiley: New York, 1993; pp 189-192.
(27) Emsley, P.; Cowton, K. Coot: Model-Building Tools for Molecular
Graphics. Acta Crystallogr., Sect. D 2004, 60, 2126–2132.