Sulfonylureido thiazoles as fructose-1,6-bisphosphatase inhibitors for the treatment of Type-2 diabetes

Article abstract of DOI:10.1016/j.bmcl.2009.11.093

Sulfonylureido thiazoles were identified from a HTS campaign and optimized through a combination of structure-activity studies, X-ray crystallography and molecular modeling to yield potent inhibitors of fructose-1,6-bisphosphatase. Compound 12 showed favo

Full text of DOI:10.1016/j.bmcl.2009.11.093

Bioorganic & Medicinal Chemistry Letters 20 (2010) 594–599
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Sulfonylureido thiazoles as fructose-1,6-bisphosphatase inhibitors
for the treatment of Type-2 diabetes
*
Eric Kitas , Peter Mohr, Bernd Kuhn, Paul Hebeisen, Hans Peter Wessel, Wolfgang Haap, Armin Ruf,
Jörg Benz, Catherine Joseph, Walter Huber, Ruben Alvarez Sanchez, Axel Paehler, Agnes Benardeau,
Marcel Gubler, Brigitte Schott, Effie Tozzo
F. Hoffmann-La Roche Ltd, Discovery Research Basel, 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:
Sulfonylureido thiazoles were identified from a HTS campaign and optimized through a combination of
structure–activity studies, X-ray crystallography and molecular modeling to yield potent inhibitors of
fructose-1,6-bisphosphatase. Compound 12 showed favorable ADME properties, for example, F = 70%,
and a robust 32% glucose reduction in the acute db/db mouse model for Type-2 diabetes.
Ó 2009 Elsevier Ltd. All rights reserved.
Received 14 October 2009
Revised 17 November 2009
Accepted 17 November 2009
Available online 22 November 2009
Keywords:
Fructose-1,6-bisphosphatase
Allosteric inhibition
Structure–activity studies
X-ray crystallography
Glucose reduction
db/db Mouse model
Aminothiazole bioactivation
Clinical studies have demonstrated that excessive hepatic glu-
cose production (HGP) is responsible for increased fasting blood
glucose and significantly contributes to postprandial hyperglyce-
mia.1 HGP comprises glucose derived from breakdown of glycogen
(glycogenolysis) and glucose synthesized from 3-carbon precursors
(gluconeogenesis: GNG). A large number of radioisotope studies
and several experiments using 13C NMR spectroscopy have shown
that GNG can account for up to 100% of the glucose produced by
the liver in the non-absorptive state and that the GNG flux is exces-
sive in Type-2 diabetic (T2D) patients.2 Therefore targeting GNG
represents an attractive approach for the control of fasting blood
glucose in T2D patients.
Fructose-1,6-bisphosphatase (FBPase) is a key regulatory en-
zyme of the hepatic GNG pathway and has recently appeared as
a target for efficient and safe glycemic control in T2D. It is a tetra-
meric enzyme that catalyzes the hydrolysis of fructose-1,6-bis-
phosphate to fructose-6-phosphate and inorganic phosphate.
FBPase is regulated synergistically by fructose-2,6-bisphosphate
(F2,6P2), a substrate site inhibitor, and AMP, which binds alloster-
ically.3 Moreover, a second allosteric binding site at the subunit
interface has been identified by Wright et al.4 Based on structural
analysis of all three putative FBPase inhibition sites, we judged
the allosteric AMP-binding cavity to be the most druggable. Simi-
larly, the phosphonate-inhibitor CS-917 from Metabasis Inc. was
also developed as an AMP-binding site inhibitor and was shown
in Phase II trials to lower plasma glucose levels in human diabetics.
We took the challenge to generate novel AMP-binding site inhibi-
tors that do not require a prodrug approach as used by the phos-
phonate-inhibitor CS-917.5
A HTS campaign was designed to differentiate competitive
inhibitors from allosteric ones.6 The AMP site binders amongst
the latter were identified by surface plasmon resonance (Biacore),
based on competition with reference compounds.7 The human li-
ver FBPase crystal structures of different AMP-site inhibitor hit
classes were subsequently solved. The complex structures were
also used in the selection of an aminothiazole (AT) hit class repre-
sented by 1 for further optimization.
The co-crystal structure of 1 with human liver FBPase was
solved to a resolution of 2.2 Å. Figure 1 (left panel) reveals that
the p-tolyl fragment occupies the AMP adenine binding pocket,
while the sulfonylurea motif serves as a linker to reach the dyad
interface region of two monomers of tetrameric FBPase. In this
partly solvent-exposed interface site the two bromothiazole moie-
ties of neighboring ligands interact through
p-stacking. A similar
binding mode with non-covalent or covalent linkage of two sym-
metry-related FBPase binding sites has been found previously for
sulfonamides9 and bissulfonylureas,6 respectively.
* 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 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.11.093

E. Kitas et al. / Bioorg. Med. Chem. Lett. 20 (2010) 594–599
595
Figure 1. Co-crystal structure of human liver FBPase with 1 (left panel) revealing the binding mode of the sulfonylureido thiazoles (cyan and green). Both the adenine binding
site of AMP and the dyad interface region are occupied by the ligand. Protein surfaces of two adjacent FBPase monomers are colored white and gold, respectively. The binding
mode of AMP (magenta, PDB id = 1fta) is shown for comparison. The right panel shows a close-up view of the AMP binding pocket of human liver FBPase with 4. Dashed red
lines indicate the H-bonding network of the sulfonylurea motif involving both protein residues and a conserved water molecule.8
Table 1 (continued)
We extensively explored the SAR for this hit class by consider-
ing variations at both ends and by looking for isosters of the sulfo-
nylurea (SU) motif. SAR was built using IC50 data of the human and
mouse liver enzymes.10 Efficacy data on mouse hepatocytes were
collected in parallel and used for selecting compounds for subse-
quent in vivo experiments in mouse models.11
Compound
R
HL IC50
M)
HL IC50 +F2,
ML EC50
V)
12
(l
6P2
(lM)
(l
F3CO
8
1.0
0.38
>100
74
9
10
11
0.69
2.1
0.29
0.72
0.09
Table 1
FBPase inhibition: phenyl-substitutions for the adenine pocketa
H2N
H
H
N
N
N
>100
63
O
R
S
S
O
O
Br
0.15
Compound
R
HL IC50
HL IC50 +F2,
ML EC50
V)
12
(l
M)
6P2
(l
M)
(l
AMP
0.44
0.14
—
12
13
0.13
0.32
0.09
0.09
10
62
F
1
0.92
0.38
0.17
118
23
N
O
2
3
0.31
0.35
N
S
a
The HL IC50 +F2,6P2 parameter measures the synergistic effect of combining
fructose-2,6-bisphosphate (F2,6P2), a substrate site inhibitor, and our allosteric
FBPase inhibitors in the enzyme assay; ML EC50 values are a measure of efficacy on
mouse hepatocytes. Assay conditions have been previously described.6
0.14
19
4
5
6
0.23
6.4
0.10
3.5
32
SAR around the adenine binding pocket was in agreement with
co-crystal structures where meta-mono-substituents at the phenyl
sulfonamide ring were favored, and small substituents (Me, Cl, Et)
filling a pocket in the back of the adenine binding site provided an
affinity gain of 3–4-fold (Fig. 1, right panel). Larger meta-substitu-
ents as in 9, or the more polar amino derivative 10, led to loss of
affinity. Furthermore, larger substituents, if optimally placed on
phenyl, for example, 11, 12, were well tolerated and led to in-
creased inhibition of FBPase (Table 1).
>100
49
Cl
0.33
0.11
F
F
O
7
0.21
0.10
24
We focused on 12 for comprehensive in vitro and in vivo char-
acterization, and its synthesis was undertaken on a multigram

596
E. Kitas et al. / Bioorg. Med. Chem. Lett. 20 (2010) 594–599
O
NH2
H
O
S
S
N
H
O
O
SO2Cl
N
b)
a)
O
S
c)
O
Br
F
N
Br
F
N
16
14
15
e)
d)
H
H
O
O
S
N
N
N
O
O
S
S
O
N
H
N
NH2
O
S
H
Br
F
N
F
N
17
12
Scheme 1. Reagents and conditions: (a) tBuNH2, DIEA, 84%; (b) 2-fluoropyridine-5-boronic acid, Pd(PPh3)4, Na2CO3, 1,4-dioxane, 86%; (c) 90% TFA, 81%; (d) PhOC(O)Cl, Et3N,
then 2-amino-5-bromothiazole 1.5 equiv, MsOH 1.5 equiv, 68%; (e) as for step (d) but using thiourea as amine.
scale as outlined in Scheme 1. 4-Bromo-3-methylsulfonyl chloride
was converted to the aryl tert-butyl sulfonamide 14. This interme-
diate underwent Suzuki coupling reaction with 2-fluoropyridine-
5-boronic acid to furnish 15 in 86% yield. tert-Butyl group removal
proceeded smoothly using trifluoroacetic acid. The resultant sul-
Table 2 (continued)
Compound
R
HL IC50
M)
HL IC50 +F2,
ML EC50
V)
(
l
6P2 (
lM)
(l
27
0.31
0.21
12
S
S
O
O
Table 2
28
29
30
31
0.07
0.04
0.07
0.05
0.03
19
FBPase inhibition: heteroaromatic-substitutions for the adenine pocket
O
O
O
H
H
N
N
N
O
R
S
S
O
S
S
O
0.15
0.06
0.06
34
9
Br
N
Compound
R
HL IC50
M)
HL IC50 +F2,
ML EC50
V)
(
l
6P2 (lM)
(l
N
2
0.31
0.44
0.17
23
92
18
0.19
0.08
S
S
Cl
Br
S
9
N
F
19
0.23
87
Cl
32
33
0.09
0.39
0.05
0.19
18
14
S
20
21
0.15
0.04
0.07
0.03
24
13
S
O
N
S
O
34
35
0.07
0.10
—
—
14
22
23
0.22
0.31
0.12
—
10
42
O
S
S
N
HO
HO
O
6.3
N
24
0.09
0.06
5.5
S
O
O
25
26
0.53
0.09
0.35
0.04
27
S
fonamide 16 was activated using phenoxy carbonyl chloride and
coupled to 2-amino-5-bromothiazole resulting in the desired
FBPase inhibitor 12. Thiourea 17 was prepared as a reference com-
pound for subsequent metabolite studies (see Table 5).
3.2
S
O

E. Kitas et al. / Bioorg. Med. Chem. Lett. 20 (2010) 594–599
597
Table 3
Besides phenyl sulfonamides our SAR showed that optimally
substituted thiophene sulfonamides were well tolerated and
led to even improved activity (Table 2). A small b-substituent,
for example, methyl, again increased potency by occupying the
back pocket in the adenine binding site discussed above. Large
FBPase inhibition: aminothiazole substitutions for the dyad interface
H
H
N
R2
O
N
S
R1
O
O
a-substituents (methoxyalkyl, substituted aromatic or heteroaro-
Compound
24
R1
R2
HL IC50
M)
ML EC50
(lV)
matic) further improved potency and yielded strong inhibitors
such as 21, 26 and 30. The isobutyl substituent of 21 occupies
the same binding region as the identical substituent in CS-917
profiting from additional hydrophobic protein–ligand interac-
tions5 (X-ray structures not shown). Last but not least, biaryls
such as benzothiophene 32 and indole 34 also yielded potent
inhibitors.
(l
N
S
0.09
5.5
S
Br
O
F
F
N
S
O
36
0.17
14
Br
As can be seen from Figure 1, right panel, the sulfonylurea linker
shows an exquisite H-bonding network to the protein and a tightly
bound water molecule, and has an optimal geometry to protrude
into the rather narrow channel linking the adenine and dyad inter-
face binding sites. Not surprisingly, changes to the sulfonylurea
moiety (e.g., to acyl urea or acyl sulfonamides, as well as N-alkyla-
tions) led to complete loss of affinity. As observed for the cova-
lently linked bissulfonylureas,6 the phosphate recognition pocket
of the AMP-binding site is not occupied by the ligand but by a
water molecule, which interacts through four hydrogen bonds with
one of the ligand S@O groups as well as Leu30, Lys112, and Tyr113.
This is an elegant indirect way to fulfill the binding requirements
of the very polar phosphate recognition pocket without the need
for strongly acidic ligand functionalities, such as phosphates or
phosphonates.
N
S
37
38
39
40
41
42
0.18
0.18
0.07
0.08
0.26
0.41
6.5
2.4
2.8
5.3
9.2
8.1
S
S
S
S
S
S
S
S
S
S
S
O
O
O
N
S
N
S
HO
HO
HO
HO
HO
N
S
N
S
The SAR and X-ray structure analysis of the aminothiazole res-
idue occupying the dyad interface region (Fig. 1, left panel) re-
vealed favorable
p-stacking between heteroaromatic residues
N
S
from ligands bound to adjacent AMP binding pockets as well as
strong van der Waals interactions involving 5-substituted polariz-
able atoms (e.g., Br, S) of adjacent ligands and Met18A and C side
chains. These favorable interactions could also be extended to
bicyclic aromatic moieties as exemplified by 44. Small, 4-alkyl
substituents were also tolerated and, if combined with hydroxy-
ethyl-substituted thiophene for optimally filling the adenine pock-
et, resulted in potent inhibitors of FBPase like 39 and 40 (Table 3).
Finally, it became evident that both binding regions, that is, the
allosteric AMP site and interface region contribute significantly
to the binding of the sulfonylureas, as both 45 and 46 were inactive
in the FBPase inhibition assay. The R2 substituent in compound 46
was previously shown to be a favorable motif for the interface
site.6
N
S
F
43
44
0.3
10
15
F
F
O
N
S
S
HO
0.04
Cl
45
46
H
>100
>100
—
—
The necessary 5-alkoxy- and 5-thioalkoxy-2-aminothiazoles
were prepared as summarized in Scheme 2. Bromination of A
H3C
(CH2)6CONHSO3CH3
R3
N
H2N
S
R3
Br
R3
O
a)
b)
N
S
N
S
H2N
H2N
R4
A
B
C
c)
O
O
R3
R3
S
d)
O
O
e)
N
N
S
R3
N
N
N
H
H2N
H
S
R4
R4
S
S
D
E
F
Scheme 2. Reagents and conditions: (a) Br2, 20% H2SO4, 0 °C, 50–70%; (b) NaOR4, 3.5 equiv, HOR4, 0 °C, 25–50%; (c) BOC2O, 1.1 equiv, DMAP, 0.1 equiv, NaHCO3, 2.5 equiv,
tBuOH, 40 °C, 70–80%; (d) nBuLi, 2.1 equiv, THF / hexane, À78 °C, then R4SSR4, 2.1 equiv, À78 °C to rt, 70–90%; (e) 4 M HCl in 1,4-dioxane, rt, 14 h, 90–100%.

598
E. Kitas et al. / Bioorg. Med. Chem. Lett. 20 (2010) 594–599
Table 4
alcohol as solvent already at 0 °C into the rather unstable build-
ADME profile and mouse PK/PD parameters of 12
ing blocks C. The sulfur-analogues, on the other hand, were
preferably synthesized relying on South’s method.14 Mono-
BOC-protection under standard conditions afforded D in decent
yield which was deprotonated twice with nBuLi and then
quenched with a symmetrical disulfide to give E. Smooth re-
moval of the BOC group produced key intermediate F. Final
assemblage was accomplished as discussed above (Scheme 1,
step d).
Parameter
Value
Mouse EC50
Cl (human)a
(
(
l
l
M)
10
11.4
5.1
0.1
0.1
70
8.1
220,407
0.6
L/min/mg protein)
Cl (mice)a
(lL/min/mg protein)
Cl (mice)b (mL/min/kg)
Vss (L/kg)
Fc (%)
T1/2 (h)
Cmax plasma (ng/mL)
Liver-to-plasma ratio
Glucose reduction after 6 hd (% vs vehicle)
Having identified potent in vitro inhibitors also showing good
cellular efficacy in mouse hepatocytes we selected 12 for further
in vivo profiling, as it exhibited good aqueous solubility (465 lg/
32
a
b
c
mL), high membrane permeability (Pe (Pampa)15 = 3.6 Â 10À6 cm/
s) and fairly good metabolic stability in microsomes (Table 4). An
in vivo pharmacokinetic (PK) study in mice showed low clearance
and very low volume of distribution resulting in high plasma expo-
sure, excellent oral bioavailability and a long half-life. This good
pharmacokinetic profile resulted in a robust glucose lowering ef-
fect upon acute treatment in db/db mice (Table 4 and Fig. 2). Addi-
tional in vivo experiments with two other ATs (27 and 36) also
showed a significant decrease of glucose levels compared to
vehicle.16
Microsomal intrinsic clearance.
iv Bolus at 2.5 mg/kg.
Oral gavage at 4.8 mg/kg.
d
Acute glucose lowering in db/db mouse model [dose = 100 mg/kg po].
Effect of compound 12 on Blood Glucose
24
vehicle
compound 12
22
Aminothiazoles suffer from the liability of being oxidized to
thioureas, and the latter have been associated with liver toxic-
ity.17 We have therefore extensively assessed our AT-containing
inhibitors for their metabolic reactivity.18 Formation of thiourea
toxophore was determined following compound 12 administra-
tion (100 mg/kg po, db/db mice) in an acute efficacy model for
blood glucose lowering. In vitro incubations in mouse liver micro-
somes as well as in vivo samples (plasma and liver taken at 7 h)
showed very low content of thiourea metabolite 17 compared to
parent as assessed by LC–MS/MS using an authentic standard.
Similar observations were made from rat PK experiments (Table
5). Based on the minor formation of 17 in vivo and in vitro, the
potential toxicity of thiourea metabolite was not considered to
be a major issue for our FBPase inhibitor class. Nevertheless, AT
replacements were identified in the lead optimization phase
which maintained FBPase inhibition and will be described in a la-
ter publication.
20
18
16
14
12
10
0
2
4
6
Time (hrs)
Figure 2. Glucose lowering in db/db mice with compound 12 at 100 mg/kg po
versus vehicle.
The potential for developing FBPase inhibitors not requiring a
prodrug strategy has been exemplified by our sulfonylureido series
where robust in vivo activity could be demonstrated in the db/db
mouse model for Type-2 diabetes.
in aqueous sulfuric acid13 yielded 5-bromo-analogue B which
could be transformed directly by nucleophilic aromatic substitu-
tion with freshly prepared sodium alkoxide in the corresponding
Table 5
Assessing bioactivation of 12: thiourea content in hepatic tissue and plasma
S
O
O
Br
O
O
S
O
O
P450
S
N
N
H
NH2
S
H
N
N
H
N
H
F
N
F
N
17
thiourea toxophore
12
in vivo active at
100 mg/kg po db/db mice;
rat/human GSH-adduct clean16
Sample
% Metabolite 17 (relative to parent)
Rat plasma (single-dose PK profiles)
Rat liver (single-dose PK @ 8 h)
Mouse plasma (PD study 100 mg/kg @ 7 h)
Mouse liver (PD study 100 mg/kg @ 7 h)
Liver microsomes (mouse)
0.8
<1.2
0.8
<0.9
<1.2
<0.5
<0.3
Liver microsomes (rat)
Liver microsomes (human)

E. Kitas et al. / Bioorg. Med. Chem. Lett. 20 (2010) 594–599
599
blood glucose over time of mice assigned to the vehicle was significantly lower
than that of mice treated with the compound (p = 0.0014). Additional data for
compounds 27 and 36 also showed significant decrease of glucose (Glc) levels
compared to vehicle:
Acknowledgements
The excellent technical assistance of Betty Hennequin, Patrick
Studer, Stefan Buerli, Thomas Burger, Roland Keller, Rudolf Minder,
Lilli Anselm, Stefan Thomi, Tamara Codilupi, Annie Sellam, An-
thony Vandjour and Stefan Masur is gratefully acknowledged. Dr.
René Wyler is thanked for his encouragement and support.
Vehicle
Glc (mM)
0 h
Glc (mM)
+2 h
Glc (mM)
+4 h
Glc (mM)
+6 h
Mean
Sem
24.74
0.93
20.56
1.05
19.07
1.58
17.99
1.47
Compound 27 (100 mg/kg)
Supplementary data
Mean
Sem
% vs vehicle
24.00
0.84
14.87
0. 39
À17
11.90
0.67
À34
12.54
1.37
À30
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2009.11.093.
References and notes
Vehicle
Glc (mM)
0 h
Glc (mM)
+2 h
Glc (mM)
+4 h
Glc (mM)
+6 h
1. DeFronzo, R. A. Diabetes 1988, 37, 667.
2. Magnusson, I .; Rothman, D. L.; Katz, L. D.; Shulman, R. G.; Shulman, G. I. J. Clin.
Invest. 1992, 90, 1323.
3. Maryanoff, B. E.; Reitz, A. B.; Tutwiler, G. F.; Benkovic, S. J.; Benkovic, P. A.;
Pilkis, S. J. J. Am. Chem. Soc. 1984, 106, 7851.
Mean
Sem
22.13
1.11
21.10
1.38
19.24
1.64
16.99
1.16
Compound 36 (100 mg/kg)
Mean
Sem
% vs vehicle
21.57
1.00
18.33
1.54
8
15.09
1.38
À11
12.57
1.31
À26
4. Wright, S. W.; Carlo, A. A.; Carty, M. D.; Danley, D. E.; Hageman, D. L.; Karam, G.
A.; Levy, C. B.; Mansour, M. N.; Mathiowetz, A. M.; McClure, L. D.; Nestor, N. B.;
McPherson, R. K.; Pandit, J.; Pustilnik, L. R.; Schulte, G. K.; Soeller, W. C.;
Treadway, J. L.; Ing-Kae Wang, I.-K.; Bauer, P. H. J. Med. Chem. 2002, 45, 3865.
5. Erion, M. D.; van Poelje, P. D.; Dang, Q.; Kasibhatla, S. R.; Potter, S. C.; Reddy, M.
R.; Reddy, K. R.; Jiang, T.; Lipscomb, W. N. Proc. Natl. Acad. Sci. 2005, 102, 7970.
6. 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.
7. Huber, W.; Mueller, F. Curr. Pharm. Des. 2006, 12, 3999.
Vehicle
Glc (mM)
0 h
Glc (mM)
+2 h
Glc (mM)
+4 h
Glc (mM)
+6 h
8. Crystallographic data were collected on beam line X10SA at the Swiss Light
Source and coordinates were deposited with the PDB-codes 2WBD and 2WBB
for compounds 1 and 4, respectively.
9. von Geldern, T. W.; Lai, C.; Gum, R. S.; Daly, M.; Sun, C.; Fry, E. H.; Abad-
Zapatero, C. Bioorg. Med. Chem. Lett. 2006, 16, 1811.
10. IC50 values are listed as averages of at least two independent experiments.
11. For all our FBPase hit series, we consistently found lower enzymatic activities
(5–100-fold) for mouse liver FBPase compared to the human form. This could
not be rationalized by the few sequence differences between human and
mouse binding sites. It appears that the mouse enzyme generally requires
higher concentrations of allosteric AMP site binders to achieve the same
allosteric response as the human enzyme. Moreover, mouse EC50 values are
influenced also by membrane permeability in contrast to human IC50 values.
12. The cooperative effect of fructose-2,6-phosphate (F2,6P2) substrate site
binding was clearly evident in this series.
Mean
Sem
22.20
1.01
19.83
1.57
18.39
1.39
17.66
1.72
Compound 12 (100 mg/kg)
Mean
Sem
% vs vehicle
22.23
1.11
18.75
1.14
À5
16.09
2.23
À13
11.96
1.32
À32
17. Kalgutkar, A. S.; Obach, R. S.; Maurer, T. S. Curr. Drug Metab. 2007, 8, 407;
Kalgutkar, A. S.; Gardner, I.; Obach, R. S.; Shaffer, C. L.; Callegari, E.; Henne, K.
R.; Mutlib, A. E.; Dalvie, D. K.; Lee, J. S.; Nakai, Y.; O’Donnell, J. P.; Boer, J.;
Harriman, S. P. Curr. Drug Metab. 2005, 6, 161.
13. Ochiai, E.; Nagasawa, H. Ber. Dtsch. Chem. Ges. 1939, 72B, 1470.
14. South, M. S.; Van Sant, K. A. J. Heterocycl. Chem. 1991, 28, 1017.
15. For a description of the PAMPA (parallel artificial membrane permeability
assay) model, a prediction assay for oral absorption: Kansy, M.; Fischer, H.;
Kratzat, K.; Senner, F.; Wagner, B.; Parrilla, I. Helv. Chim. Acta 2000, 447.
16. Old male (13 week) db/db mice on C57BLKS background were block randomized
based on blood glucose levels and then orally dosed with 4 mL/kg vehicle (0.3%
Tween 80) or compound 12 at 100 mg/kg after 4 h food removal. Blood glucose
was monitored with a hand held glucose monitoring device (Roche Diagnostics)
at times 0, 2, 4 and 6 h post dose by tail tip bleed. Mice were sacrificed by
decapitation immediately after the 6 h time-point and blood and liver were
collected for compound exposure and liver glycogen analysis. Glucose results are
expressed as a % lowering vs the vehicle group at the same time-point. The
statistical analysis of the blood glucose concentration given the experimental
design was based on mixed models analysis (random intercepts and random
slopes multilevel models). Two models were fitted. The random intercepts
proved to be a good representation of the data over random slopes model (based
on graphical and information criteria (AIC)). After verifying the model
assumptions, hypothesis testing and confidence interval estimation of fixed
effects (group, time and their interaction) were carried out, accounting for
random effects. Statistical analysis was performed in SAS v8. Blood glucose of
mice treated by the compound decrease at a rate of 1.67 (glucose units) per hour
(95% CI: 1.28, 2.07) significantly greater than the one observed in the vehicle
group: 0.75 (glucose units) per hour (95% CI: 0.38, 1.12).The rate of change of
18. Two bioactivation pathways for aminothiazoles resulting in acute liver toxicity
through covalent binding have been described.17 The first is associated with
formation of the known toxophore thiourea as discussed in the text. The
second pathway involves the formation of reactive thiazolium which could be
inferred from glutathione (GSH) trapping experiments. In parallel to the
development of the SAR, GSH-adduct formation was assessed for interesting
compounds. Within our lead series it became evident that 4- or 5- substitution
of the AT ring blocked adduct formation which on the other hand was clearly
seen for the 4,5-unsubstituted compound 47, an inactive compound in the
FBPase assay, viz.
O
glutathione (GSH)a
S
O
O
S
N
N
N
GSH-adduct (MW= 592)
H
H
47
(MW =297.4)
aStandard test conditions: 10
l
M test compound, 1 mg/mL protein, 5 mM GSH,
1 mM NADPH, 60 min incubation.