Orally active aminopyridines as inhibitors of tetrameric fructose-1,6-bisphosphatase

Article abstract of DOI:10.1016/j.bmcl.2011.04.044

A novel sulfonylureido pyridine series exemplified by compound 19 yielded potent inhibitors of FBPase showing significant glucose reduction and modest glycogen lowering in the acute db/db mouse model for Type-2 diabetes. Our inhibitors occupy the allosteric binding site and also extend into the dyad interface region of tetrameric FBPase.

Full text of DOI:10.1016/j.bmcl.2011.04.044

Bioorganic & Medicinal Chemistry Letters 21 (2011) 3237–3242
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Orally active aminopyridines as inhibitors of tetrameric
fructose-1,6-bisphosphatase
Paul Hebeisen a, Wolfgang Haap a, Bernd Kuhn a, Peter Mohr a, Hans Peter Wessel a, Ulrich Zutter a,
Stephan Kirchner b, Armin Ruf a, Jörg Benz a, Catherine Joseph a, Rubén Alvarez-Sánchez a, Marcel Gubler a,
Brigitte Schott a, Agnes Benardeau a, Effie Tozzo a, Eric Kitas a,
⇑
a F. Hoffmann-La Roche Ltd, Discovery Research Basel, CH-4070 Basel, Switzerland
b F. Hoffmann-La Roche Ltd, Non Clinical Safety, CH-4070 Basel, Switzerland
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 23 February 2011
Revised 8 April 2011
Accepted 12 April 2011
Available online 20 April 2011
A novel sulfonylureido pyridine series exemplified by compound 19 yielded potent inhibitors of FBPase
showing significant glucose reduction and modest glycogen lowering in the acute db/db mouse model for
Type-2 diabetes. Our inhibitors occupy the allosteric binding site and also extend into the dyad interface
region of tetrameric FBPase.
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
Fructose-1,6-bisphosphatase
FBPase
Allosteric inhibition
Structure–activity studies
X-ray crystallography
Genotoxicity
Glucose reduction
db/db Mouse model
Targeting fructose-1,6-bisphosphatase (FBPase) is a potentially
viable approach for glucose control in Type-2 diabetes (T2D).
Numerous publications have recently appeared describing FBPase
inhibitors1 and, furthermore, molecules from Metabasis Inc. have
been advanced to human clinical trials.2 Besides its role in gluco-
neogenesis control, FBPase has also been implicated in glucose
sensing and in regulating insulin secretion in b-cells.3 We have pre-
viously described our aminothiazole class of small molecule inhib-
itors of FBPase which were shown to significantly lower fasting
glucose levels in a transgenic mouse model of T2D.4 During the
optimization phase we investigated the liability of advanced mol-
ecules to be oxidized to thioureas which have been associated with
lipidosis. Although, we concluded that the very low level of thio-
urea metabolite measured in this study not to be a potential risk,
we also identified six-membered (hetero)aromatic bioisosteres of
aminothiazoles as potent FBPase inhibitors, which are the subject
of this paper.
second binding region, at the dyad interface of tetrameric FBPase.4
In the interface region, there are direct -stacking interactions
p
between the terminal thiazole rings of two adjacent ligands as well
as several strong van der Waals interactions between highly polar-
izable ligand substituents, such as –Br, –SMe, or –Cl and the side
chains of Met18A/C (Fig. 1, left panel). Activity in the thiazole ser-
ies could be increased by up to a factor of 80 by these specific sub-
stitutions. In this paper, we will focus on the replacement of the
aminothiazole motif while trying to satisfy the main recognition
motifs (p-stacking of aromatic ring, polarizable substituent close
to Met18) and discuss the evolving SAR.5 The structure–activity
relationship of the (hetero)aromatic sulfonylureido moiety occu-
pying the AMP binding site paralleled that of the aminothiazole
series and could be transferred to a considerable extent to the
new series reported here. We mainly use 3-Cl substituted phenyl
and 5-(2-methoxy-ethyl)-4-methyl substituted thiophenyl (Table
1), which have been identified as high affinity fragments in our
previous publication.4 While both meta-Cl in the phenyl and
b-methyl in the thiophene series occupy the back pocket in the
As previously outlined, our series occupy the allosteric AMP
binding site and reach through the sulfonylureido linker into a
AMP binding site, the methoxyethyl substituent in the
a-position
of the thiophene ring makes additional non-polar interactions with
Leu30 and Val160 of human liver FBPase (X-ray structure not
shown) further enhancing the binding affinity in this series.
⇑
Corresponding author. Tel.: +41 61 6887618; fax: +41 61 6886459.
E-mail address: eric_a.kitas@roche.com (E. Kitas).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.04.044

3238
P. Hebeisen et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3237–3242
Gly 26C
Met 18A
Thr 27C
Met 18C
Met 18A
Met 18C
Thr 27A
Gly 26A
Figure 1. Left: X-ray structure of human liver FBPase with compound 1 (cyan). The meta-chlorophenyl ring occupies the AMP binding region while the bromothiazole moiety
is involved in interactions at the dyad interface, including contacts to a neighboring ligand (green). Solvent-accessible surfaces of the two protein chains A and C are colored
white and gold, respectively. Right: X-ray structure of FBPase dyad interface region in complex with two ligands 10 (cyan, green). Hydrogen bonds of the urea-substituted
pyridines are depicted with red, dashed lines.6
Table 1
FBPase inhibition: N-heterocycle analoguesa
H
H
N
R2
O
N
O
S
R1
O
Entry
R1
R2
HL IC50
0.33
(l
M)
ML IC50
5.79
(lM)
mEC50
49
(lV)
Cl
N
S
1
Br
Cl
Cl
Cl
2
3
0.64
0.35
12
84
41
Cl
HN
5.0
Br
N
Cl
N
N
N
4
5
0.33
0.29
3.3
4.5
23
13
Br
N
Br
N
S
S
6
7
0.14
0.82
1.7
6.1
O
7
5
Br
N
N
8
51
O
Br

P. Hebeisen et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3237–3242
3239
Table 1 (continued)
Entry
R1
R2
HL IC50
(l
M)
ML IC50
(lM)
mEC50
(lV)
O
NH
N
S
S
S
HN
O
8
0.08
1.0
5
Br
O
NH
N
HN
S
O
O
9
0.17
0.22
1.6
2.6
8
O
NH
N
HN
10
12
O
H2N
N
S
S
11
12
0.13
0.20
1.5
3.7
8
O
O
Br
F
F
NH
F
N
14
Br
N
H2N
S
S
S
13
14
1.8
0.4
9
7
O
O
O
S
H2N
N
N
30
97
O
H2N
15
16
17
0.76
0.60
0.42
6
8
F
F
F
H2N
N
N
N
S
S
F
F
3.5
8.9
O
F
F
F
F
H2N
F
F
4
4
HO
O
F
H2N
S
F
F
18
19
0.67
0.53
3.5
2.7
5.9
O
F
N
S
7
O
F
F
F
(continued on next page)

3240
P. Hebeisen et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3237–3242
Table 1 (continued)
Entry
R1
R2
HL IC50
0.35
(l
M)
ML IC50
(lM)
mEC50
8.3
(lV)
N
S
20
21
4
HO
O
F
F
F
N
N
S
0.52
1.7
7.3
8.3
5.1
O
F
F
F
F
S
22
19
F
F
O
a
HL IC50 and ML IC50 values represent potency of FBPase inhibitors against human liver and mouse liver FBPase, respectively, as determined in enzyme inhibition assays
described previously.9 mEC50 values are a measure of efficacy of FBPase inhibitors in cultivated primary mouse hepatocytes to suppress glucose production.10
1) SO3:DMF
O
O
S
O
2) SOCl2
Mg, Et2O
51%
O
O
+
Br
S
O
S
S
O
S
Cl
1,2 -dichloroethane
88%
O
12.7 g
25 g (1.5 eq)
23
24
NH3, THF
98%
CF3
CF3
NH3
52%
O
+
S
O
S
NH2
H2N
N
N
Cl
O
26
25
O
Cl
72%
O
CF3
N
O
O
S
O
S
N
N
O
H
H
19
Scheme 1. Outlining sulfonamide approach for the preparation of FBPase inhibitor 19.
The dichlorophenyl analog 2 was a first example of a thiazole
replacement with modest enzymatic activity against both human
and mouse isoforms of FBPase although exhibiting poor efficacy
in the mouse hepatocyte assay (Table 1). Due to the strong, positive
impact of –Br substitution on FBPase binding, we performed a vir-
tual screen of available building blocks that would position the Br
atom close to the Met18 side chains. From this exercise, 6-Br-
substituted indoles and subsequently indazoles could be identified.
Compound 6 showed both impressive activities in the enzymatic as
indazole ring with Cl reduced the AMES activity but abolished
FBPase inhibition as well. Replacing the indazole by imidazopyri-
dine 7 removed the AMES activity but yielded also only modest
inhibition of the enzyme.
With the help of molecular modeling and searches in the small
molecule X-ray Cambridge Structural Database8 we identified
urea-substituted pyridines as potential bioisosteres of indazoles.
These replacements (Table 1, compounds 8–10) were well toler-
ated provided that the terminal nitrogen atom carried at least
one hydrogen atom. The hypothesis that, in the binding conforma-
tion, this hydrogen atom was involved in an intramolecular hydro-
gen bond was nicely confirmed by X-ray crystallography (Fig. 1,
right panel). The crystal structure further revealed a hydrogen
bond between the proximal nitrogen atom of the urea substituent
with the O@C backbone of Thr27 from a neighboring subunit (li-
gand A with Thr27C and ligand C with Thr27A, respectively), and
well as the cellular assay, with a mouse EC50 of 6.1 lV. As part of the
multi-dimensional optimization process, compounds were rou-
tinely submitted to safety assays such as the AMES test for genotox-
icity where positive results were obtained for FBPase inhibitors
bearing the indazole moiety.7 Typically, a greater than two-fold
induction of mutant colonies was found in the TA98 strain after
metabolic activation. Blocking the vulnerable 5-position on the

P. Hebeisen et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3237–3242
3241
Table 2
ADME profile and mouse PK/PD parameters of selected aminopyridines
FBPase inhibitors
16
18
19
20
21
Mouse EC50
Cl(human)a
(
l
M)
8.9
2.7
Unstable
Unstable
5.9
nd
31
>0.49
1.7
8.3
28
10
>0.57
0.7
6799
8.9
36f
8
5.1
(l
l/min/mg protein)
10.0
12.1
0.53
4.0
4091
17.4
31f
Unstable
Unstable
>0.62
1.4
Cl (mice)a
(
l
l/min/mg protein)
Solubility (mg/mL)
0.16
PAMPAb Pe (Â10À6/s)
ndc
Plasma levels at 6 h (ng/mL)
Liver-to-plasma ratio
nd
2794
10.5
38f
nd
nd
nd
Glucose reduction after 6 hd (%)
Liver glycogen reduction after 6 h (%)
16e
29f
18
Increased by 21
33f
5
a
b
c
d
e
f
Microsomal intrinsic clearance.
PAMPA is a prediction assay for oral absorption.13
nd = not determined measurement.
Acute glucose lowering in db/db mouse model [dose = 100 mg/kg po].
Acute glucose lowering in db/db mouse model [dose = 200 mg/kg po].
Significantly different from vehicle (p <0.05).
nicely illustrates the planarity of the urea-substituted pyridine
resulting in extended -stacking of both heterocycles. Thus, de-
Acknowledgements
p
spite having formally two more hydrogen-bonding groups, the ure-
ido analog 8 acts as a bioisostere of the bicyclic indazole 6.
Simplifying the structure to the aminopyridine 11 maintained
strong FBPase inhibition but induced again AMES activity. Amine
alkylation however, as in 12, fixed this flaw. The mutagenicity po-
tential was also consistently overcome when 4-Br replacements
were identified such as thiomethoxy in 13, methoxy- in 14 and
the CF3 group in compound 15. The 4-trifluoromethyl aminopyri-
dines were further pursued and gave efficacious compounds not
requiring the amine functionality such as 19 and 20. The esterase
labile acetates 18 and 21 were prepared to explore the potential
modulation of pharmacokinetic properties of the parent alcohols
17 and 20, respectively. Interestingly, no significant loss of
in vitro efficacy was observed following this transformation.
Our general and straightforward synthetic approach to the ami-
nopyridine series can be exemplified with the CF3-substituted
derivative 19 (Scheme 1). 2-Bromo-3-methyl-thiophene was trans-
formed to the corresponding Grignard reagent which was treated
with toluene-4-sulfonic acid 2-methoxyl ethyl ester affording com-
pound 23 in a modest 51% yield. Formation of the sulfonyl chloride
24 using the DMF–sulfur trioxide complex followed by chlorina-
tion of the resultant sulfonate with thionyl chloride was achieved
in an overall 88% yield. Ammonolysis yielded sulfonamide 25
which was finally coupled to the phenyl-carbamate derivative of
6-methyl-4-trifluoromethyl-pyridin-2-ylamine 26 using triethyl-
amine as base.
The excellent technical assistance of Betty Hennequin, Patrick
Studer, Lilli Anselm, Stefan Thomi, Tamara Codilupi, Annie Sellam,
Edith Brandt, Raymonde Schmitt, Anthony Vandjour and Stefan
Masur is gratefully acknowledged.
References and notes
1. Dang, Q.; Kasibhatla, S. R.; Xiao, W.; Liu, Y.; Da Re, J.; Taplin, F.; Reddy, K. R.;
Scarlato, G. R.; Gibson, T.; van Poelje, P. D.; Potter, S. C.; Erion, M. D. J. Med.
Chem. 2010, 53, 441; Tsukada, T.; Kanno, O.; Yamane, T.; Tanaka, J.; Yoshida, T.;
Okuno, A.; Shiiki, T.; Takahashi, M.; Nishi, T. Bioorg. Med. Chem. 2010, 18, 5346;
Rudnitskaya, A.; Borkin, D. A.; Huynh, K.; Torok, B.; Stieglitz, K. ChemMedChem
2010, 5, 384; Heng, S.; Gryncel, K. R.; Kantrowitz, E. R. Bioorg. Med. Chem. 2009,
17, 3916.
2. Dang, Q.; Liu, Y.; Cashion, D. K.; Kasibhatla, S. R.; Jiang, T.; Taplin, F.; Jacintho, J.
D.; Li, H.; Sun, Z.; Fan, Y.; DaRe, J.; Tian, F.; Li, W.; Gibson, T.; Lemus, R.; van
Poelje, P. D.; Potter, S. C.; Erion, M. D. J. Med. Chem. 2011, 54, 153.
3. Zhang, Y.; Xie, Z.; Zhou, G.; Zhang, H.; Lu, J.; Zhang, W. J. Endocrinology 2010,
151, 4688.
4. Kitas, E.; Mohr, P.; Kuhn, B.; Hebeisen, P.; Wessel, H. P.; Haap, W.; Ruf, A.; Benz,
J.; Joseph, C.; Huber, W.; Sanchez, R. A.; Paehler, A.; Benardeau, A.; Gubler, M.;
Schott, B.; Tozzo, E. Bioorg. Med. Chem. Lett. 2010, 20, 594.
5. Structure–activity relationship (SAR) was built using IC50 data of the human
and mouse liver enzymes. Efficacy data on mouse hepatocytes were collected
in parallel and used for selecting compounds for subsequent in vivo
experiments in mouse models.
6. Crystallographic data were collected on beam line X10SA at the Swiss Light
Source and coordinates were deposited with the PDB-codes 2y5l and 2y5k for
compounds 1 and 10, respectively.
7. Genotoxicity is estimated with the use of the AMES microsuspension assay,
where the readout parameter is the increase in the number of revertant
colonies (mutation frequency) of treated compared to untreated control in five
different Salmonella triphimurium tester strains.
A series of 4-trifluoromethyl aminopyridines with promising
properties (high aqueous solubility and PAMPA permeability) and
good efficacy in primary hepatocytes were further profiled in the
acute, db/db mouse model for T2D.11 Compounds were p.o. admin-
istered (100 mg/kg or 200 mg/kg) to 15 week-old db/db mice 4 h
after food removal (in the morning). Surprisingly and in contrast
to our previously described aminothiazole series, aminopyridines
16, 19, and 20 that were in vivo tested showed high liver partition-
ing. Furthermore, all the selected FBPase inhibitors 16, 19–21, ex-
cept 18, decreased significantly blood glucose levels in comparison
to vehicle at 6 h post-dosing (overall, reduction reached 29–38%
compared to vehicle, p <0.05, see Table 2).12 Liver glycogen was
significantly decreased after treatment with compound 19.
Looking at derisking the potential for toxic metabolite forma-
tion from our original aminothiazole FBPase inhibitor series, we
proceeded to successfully identify N-heterocycle isosteres that
maintained strong in vitro activity against this important enzyme
in the gluconeogenesis pathway. Furthermore, robust glucose
reduction in an acute mouse model for T2D was demonstrated
with a set of CF3-substituted aminopyridines.
8. Allen, F. Acta Crystallogr., Sect. B 2002, 58, 380.
9. Hebeisen, P.; Kuhn, B.; Kohler, P.; Gubler, M.; Huber, W.; Kitas, E.; Schott, B.;
Benz, J.; Joseph, C.; Ruf, A. Bioorg. Med. Chem. Lett. 2008, 18, 4708.
10. Primary mouse hepatocytes were seeded in 96-well tissue culture plates
(25,000 cells per well) in 100
fetal bovine serum (FBS). After 24 h of cultivation at 37 °C, the medium was
replaced with 100 l of RPMI medium without glucose (Invitrogen) containing
ll of DMEM medium (Invitrogen) containing 10%
l
1% FBS. The cells were incubated for 24 h at 37 °C for starvation and depletion
of intracellular glycogen. After removal of medium and washing with
phosphate-buffered saline (PBS, Invitrogen), the cells were incubated with
FBPase inhibitors for 6 h at 37 °C in 60 ll of PBS containing 0.1% bovine serum
albumin (BSA), 20 mM lactate and 2 mM pyruvate. Glucose produced and
released by the cells was determined in the culture supernatant using the
glucose oxidase assay (Amplex Red kit, Molecular Probes).
11. Fifteen week-old male db/db mice on C57BLKS background were kept under
controlled-fed-conditions and then block randomized based on blood glucose
levels. Bodyweight was measured in the morning 4 h after food removal (post-
prandial conditions) and the mice were then immediately orally dosed with
4 mL/kg vehicle (0.5% Tween 80, 0.5% hydroxyethylcellulose) (0.3% Tween 80)
or FBPase inhibitors at 100 mg/kg or 200 mg/kg. The FBPase inhibitor CS-917
was used as the positive control.2 Blood glucose was monitored with a hand
held glucose monitoring device (AccuCheck, Roche Diagnostics) at times 0, +2,
+4 and +6 h post dose by tail tip bleed. Full blood (N = 8 per group) was
collected (on EDTA). After the last blood glucose measurement, mice were
sacrificed by decapitation immediately after the 6 h time-point and blood and

3242
P. Hebeisen et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3237–3242
liver were collected for determining compound exposure and liver glycogen
12. Animals were randomized according to body weight and glucose level
measured at baseline before compounds administration. Livers were of
normal appearance (normal color, no hypertrophy, no increase in weight
compared to vehicle-treated mice).
13. Kansy, M.; Fischer, H.; Kratzat, K.; Senner, F.; Wagner, B.; Parrilla, I. Helv. Chim.
Acta 2000, 447.
analysis. Glucose (Glc) results are expressed as % lowering versus the vehicle
group at the same time-point. Glucose data were analyzed using the software
JMP6 for Windows (Version 5.01, SAS institute Inc., SAS Campus Drive, Cary, NC
27513). Analysis of Variance ANOVA (alpha 0.05 and 0.01) followed by
Dunnett’s test (comparison vs control).
Glc (mM) 0 h
Glc (mM) +2 h
Glc (mM) +4 h
Glc (mM) +6 h
Vehicle
Mean
SEM
Mean
SEM
22.73
0.73
22.49
0.75
31.16
0.63
26.99
1.78
26.58
1.09
19.04
1.41
23.84
0.98
16.24
1.70
À31
16 (100 mg/kg)
Compd 16 versus vehicle (%)
18 (200 mg/kg)
Mean
SEM
22.46
0.72
27.59
1.29
23.31
0.75
20.09
0.90
Compd 18 versus vehicle (%)
19 (100 mg/kg)
À16
Mean
SEM
22.55
0.85
21.91
1.24
16.60
1.35
14.64
1.42
À38
Compd 19 versus vehicle (%)
20 (100 mg/kg)
Mean
SEM
22.53
0.77
23.68
1.10
18.86
1.54
16.79
1.47
À36
Compd 20 versus vehicle (%)
21 (100 mg/kg)
Mean
SEM
22.48
0.79
24.66
0.80
18.76
1.26
14.95
0.86
À29
Compd 21 versus vehicle (%)
CS-917 (200 mg/kg)
Mean
SEM
22.69
0.76
29.59
1.61
21.25
1.32
18.61
1.73
À22
CS-917 versus vehicle (%)