Discovery of potent and specific fructose-1,6-bisphosphatase inhibitors and a series of orally-bioavailable phosphoramidase-sensitive prodrugs for the treatment of type 2 diabetes

Article abstract of DOI:10.1021/ja074871l

Excessive glucose production by the liver coupled with decreased glucose uptake and metabolism by muscle, fat, and liver results in chronically elevated blood glucose levels in patients with type 2 diabetes. Efforts to treat diabetes by reducing glucose production have largely focused on the gluconeogenesis pathway and rate-limiting enzymes within this pathway such as fructose-1,6-bisphosphatase (FBPase). The first potent FBPase inhibitors were identified using a structure-guided drug design strategy (Erion, M. D.; et al. J. Am. Chem. Soc. 2007, 129, 15480-15490) but proved difficult to deliver orally. Herein, we report the synthesis and characterization of a series of orally bioavailable FBPase inhibitors identified following the combined discoveries of a low molecular weight inhibitor series with increased potency and a phosphonate prodrug class suitable for their oral delivery. The lead inhibitor, 10A, was designed with the aid of X-ray crystallography and molecular modeling to bind to the allosteric AMP binding site of FBPase. High potency (IC₅₀ = 16 nM) and FBPase specificity were achieved by linking a 2-aminothiazole with a phosphonic acid. Free-energy perturbation calculations provided insight into the factors that contributed to the high binding affinity. 10A and standard phosphonate prodrugs of 10A exhibited poor oral bioavailability (0.2-11%). Improved oral bioavailability (22-47%) was achieved using phosphonate diamides that convert to the corresponding phosphonic acid by sequential action of an esterase and a phosphoramidase. Oral administration of the lead prodrug, MB06322 (30, CS-917), to Zucker Diabetic Fatty rats led to dose-dependent inhibition of gluconeogenesis and endogenous glucose production and consequently to significant blood glucose reduction.

Full text of DOI:10.1021/ja074871l

Published on Web 11/28/2007
Discovery of Potent and Specific
Fructose-1,6-Bisphosphatase Inhibitors and a Series of
Orally-Bioavailable Phosphoramidase-Sensitive Prodrugs for
the Treatment of Type 2 Diabetes
Qun Dang,† Srinivas Rao Kasibhatla,†,‡ K. Raja Reddy,† Tao Jiang,†,§
M. Rami Reddy,† Scott C. Potter,† James M. Fujitaki,† Paul D. van Poelje,†
Jingwei Huang,| William N. Lipscomb,| and Mark D. Erion*,†
Contribution from the Departments of Medicinal Chemistry, Biochemistry, and Molecular
Modeling, Metabasis Therapeutics, Inc., La Jolla, California 92037, and Department of
Chemistry and Chemical Biology, HarVard UniVersity, Cambridge, Massachusetts 02138
Received July 3, 2007; E-mail: erion@mbasis.com
Abstract: Excessive glucose production by the liver coupled with decreased glucose uptake and metabolism
by muscle, fat, and liver results in chronically elevated blood glucose levels in patients with type 2 diabetes.
Efforts to treat diabetes by reducing glucose production have largely focused on the gluconeogenesis
pathway and rate-limiting enzymes within this pathway such as fructose-1,6-bisphosphatase (FBPase).
The first potent FBPase inhibitors were identified using a structure-guided drug design strategy (Erion, M.
D.; et al. J. Am. Chem. Soc. 2007, 129, 15480-15490) but proved difficult to deliver orally. Herein, we
report the synthesis and characterization of a series of orally bioavailable FBPase inhibitors identified
following the combined discoveries of a low molecular weight inhibitor series with increased potency and
a phosphonate prodrug class suitable for their oral delivery. The lead inhibitor, 10A, was designed with the
aid of X-ray crystallography and molecular modeling to bind to the allosteric AMP binding site of FBPase.
High potency (IC50 ) 16 nM) and FBPase specificity were achieved by linking a 2-aminothiazole with a
phosphonic acid. Free-energy perturbation calculations provided insight into the factors that contributed to
the high binding affinity. 10A and standard phosphonate prodrugs of 10A exhibited poor oral bioavailability
(0.2-11%). Improved oral bioavailability (22-47%) was achieved using phosphonate diamides that convert
to the corresponding phosphonic acid by sequential action of an esterase and a phosphoramidase. Oral
administration of the lead prodrug, MB06322 (30, CS-917), to Zucker Diabetic Fatty rats led to dose-
dependent inhibition of gluconeogenesis and endogenous glucose production and consequently to significant
blood glucose reduction.
Introduction
pancreatic function, and exacerbation of peripheral insulin
resistance, leaving the majority of treated patients with fasting
Type 2 diabetes mellitus (T2DM) is a growing worldwide
health concern that is expected to afflict over 366 million people
by 2030.1 Patients with diabetes exhibit chronically elevated
blood glucose levels and, as a consequence, are at increased
risk of long-term complications arising from glucose-mediated
damage to organs such as the eyes, kidneys, nerves, heart, and
blood vessels. Numerous drug classes are available that lower
glucose levels and reduce diabetic complications.2 Unfortunately,
the glucose-lowering effect usually diminishes over time due
to poor compliance, suboptimal dose titration, decreased
and postprandial glucose levels above the target levels.3
Consequently, there remains a large need for new, more
effective, and longer-acting drugs to treat T2DM.
Blood glucose levels are elevated in T2DM due to reduced
glucose metabolism by tissues such as muscle, liver, and fat,
coupled with increased endogenous glucose production by the
liver.4 Several popular classes of oral diabetes therapies on the
market (e.g., sulfonylureas and PPAR-gamma agonists) lower
glucose by increasing glucose metabolism either via enhanced
insulin secretion or improved insulin sensitivity. The only drug
† Metabasis Therapeutics.
(3) Saydah, S. H.; Fradkin, J. J.; Cowie, C. C. J. Am. Med. Assoc. 2004, 291,
335-342.
‡ Present address: Biogen Idec, 5200 Research Place, San Diego, CA
92122.
(4) For articles correlating increased endogenous glucose production and
elevated glucose levels, see (a) DeFronzo, R. A. Diabetes 1988, 37, 667-
687. (b) Kolterman, O. G.; Gray, R. S.; Griffin, J.; Burstein, P.; Insel, J.;
Scarlett, J. A.; Olefsky, J. M. J. Clin. InVest. 1981, 68, 957-969. (c) Jeng,
C. Y.; Sheu, W. H.; Fuh, M. M.; Chen, Y. D.; Reaven, G. M. Diabetes
1994, 43, 1440-1444. (d) Maggs, D. G.; Buchanan, T. A.; Burant, C. F.;
Cline, G.; Gumbiner, B.; Hsueh, W. A.; Inzucchi, S.; Kelley, D.; Nolan,
J.; Olefsky, J. M.; Polonsky, K. S.; Silver, D.; Valiquett, T. R.; Shulman,
G. I. Ann. Intern. Med. 1998, 128, 176-185.
§ Present address: Genomic Institute of the Novartis Research Founda-
tion, 10715 John Jay Hopkins Drive, San Diego, CA 92121.
| Harvard University.
(1) (a) Wild, S.; Roglic, G.; Green, A.; Sicree, R.; King, H. Diabetes Care
2004, 27, 1047-1053. (b) DeFronzo, R. A. Med. Clin. North Am. 2004,
88, 787-835.
(2) Moller, D. E. Nature 2001, 414, 821-827.
9
10.1021/ja074871l CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 15491-15502
15491

A R T I C L E S
Dang et al.
Chart 1. Known FBPase Inhibitors
that acts primarily by reducing endogenous glucose production
is metformin.5 Metformin lowers glucose in the absence of
weight gain and therefore is often used to treat T2DM despite
well-recognized safety concerns associated with its use in certain
diabetic patient populations and its overall high incidence of
gastrointestinal intolerance.6
In an effort to find more effective glucose-lowering drugs,
numerous investigators have evaluated various drug targets
within the liver for their potential to reduce endogenous glucose
production.7 Most of the work has focused on enzymes in the
gluconeogenesis (GNG) pathway,8 since this pathway is re-
sponsible for the excessive glucose production found in T2DM.9
Phosphoenolpyruvate carboxykinase (PEPCK)10 and glucose
6-phosphatase (G6Pase)11 were the primary targets in the 1970s
and 1980s. In contrast, we targeted fructose-1,6-bisphosphatase
(FBPase),12 an enzyme whose expression is regulated by the
hormones insulin and glucagon and whose activity is endog-
enously inhibited by fructose-2,6-bisphosphate and AMP through
their interactions with the substrate and allosteric binding sites,
respectively. FBPase is an attractive target not only because it
is a rate-controlling enzyme within the GNG pathway but also
because it functions only within the GNG pathway, unlike both
PEPCK and G6Pase.8c Moreover, adults who are genetically
deficient in FBPase activity exhibit relatively normal clinical
profiles provided they control their diet and avoid prolonged
fasting.13
mimic that could not be significantly improved through synthesis
of structural analogs. More recently, efforts were made to
identify lead compounds by screening large compound librar-
ies.16 Noncompetitive inhibitors binding at or near the AMP
site16a-c (e.g., 1) and uncompetitive inhibitors (e.g., 2) binding
at a site with unknown physiologic function near the subunit
interface16d-f were identified. All showed only modest inhibitory
potency with no reports of glucose lowering activity in animal
models.
Our strategy was to target the allosteric AMP binding site,
which is considerably less hydrophilic than the substrate binding
site and completely insensitive to inhibitor-induced FBPase
precursor buildup. As outlined in the previous report,17 AMP
binding sites are traditionally difficult drug targets due to their
high dependence on electrostatic binding interactions and their
common use by enzymes within important anabolic and
catabolic pathways. Using a structure-guided drug design
strategy, we identified a lead compound (3) and showed that it
is a potent and specific FBPase inhibitor capable of entering
rat hepatocytes.17
Herein, we describe our efforts to find an orally bioavailable
FBPase inhibitor that lowers glucose in diabetic animal models.
Since 3 exhibited poor oral activity and known phosphonate
prodrugs of 3 (e.g., 4) failed to achieve adequate oral bioavail-
ability, we re-examined the structure of the FBPase-3 complex
with the goal of identifying a new inhibitor series with properties
more likely to produce orally bioavailable compounds. This led
to compounds with reduced molecular weight, increased po-
tency, increased solubility, and decreased protein binding. These
findings coupled with the identification of a phosphonate
Previous attempts to discover inhibitors of FBPase failed to
identify suitable drug candidates (Chart 1). In the 1970s,
substrate analogs were evaluated and found to exhibit low
inhibitory activity.14 A nucleoside analog, 5-aminoimidazole-
4-carboxamide riboside, was shown in the 1990s to lower
glucose after conversion to the corresponding 5′-monophosphate
(ZMP) and binding to the allosteric AMP binding site.15 This
compound, however, proved to be a weak, nonspecific AMP
(5) (a) Stumvoll, M.; Nurjhan, N.; Perriello, G.; Dailey, G.; Gerich, J. E. N.
Engl. J. Med. 1995, 333, 550-554. (b) DeFronzo, R. A.; Goodman, A. M.
N. Engl. J. Med. 1995, 333, 541-549. (c) Hundal, R. S.; Krssak, M.;
Dufour, S.; Laurent, D.; Lebon, V.; Chandramouli, V.; Inzucchi, S. E.;
Schumann, W. C.; Petersen, K. F.; Landau, B. R.; Shulman, G. I. Diabetes
2000, 49, 2063-2069. (d) Natali, A.; Ferrannini, E. Diabetologia 2006,
49, 434-441.
(6) For reviews see: (a) Howlett, H. C.; Bailey, C. J. Drug Saf. 1999, 20,
489-503. (b) Dunn, C. J.; Peters, D. H. Drugs 1995, 49, 721-749.
(7) For reviews, see: (a) McCormack, J. G.; Westergaard, N.; Kristiansen,
M.; Brand, C. L.; Lau, J. Curr. Pharm. Des. 2001, 7, 1451-1474. (b) Barf,
T. Mini ReV. Med. Chem. 2004, 4, 897-908. (c) Kurukulasuriya, R.; Link,
J. T.; Madar, D. J.; Pei, Z.; Richards, S. J.; Rohde, J. J.; Souers, A. J.;
Szczepankiewicz, B. G. Curr. Med. Chem. 2003, 10, 123-153.
(8) For reviews on the gluconeogenesis pathway and its regulation, see: (a)
Pilkis, S. J.; Granner, D. K. Annu. ReV. Physiol. 1992, 54, 885-909. (b)
Granner, D.; Pilkis, S. J. Biol. Chem. 1990, 265, 10173-10176. (c) In
addition to their roles in GNG, PEPCK is also important for cellular energy
homeostasis and the red-ox state while G6Pase participates in the coversion
of hepatic glycogen to glucose.
(15) (a) Vincent, M. F.; Marangos, P. J.; Gruber, H. E.; Van den Berghe, G.
Diabetes 1991, 40, 1259-1266. (b) Vincent, M. F.; Erion, M. D.; Gruber,
H. E.; Van den Berghe, G. Diabetologia 1996, 39, 1148-1155.
(16) (a) Wright, S. W.; Carlo, A. A.; Danley, D. E.; Hageman, D. L.; Karam,
G. A.; Mansour, M. N.; McClure, L. D.; Pandit, J.; Schulte, G. K.;
Treadway, J. L.; Wang, I. K.; Bauer, P. H. Bioorg. Med. Chem. Lett. 2003,
13, 2055-2058. (b) von Geldern, T. W.; Lai, C.; Gum, R. J.; Daly, M.;
Sun, C.; Fry, E. H.; Abad-Zapatero, C. Bioorg. Med. Chem. Lett. 2006,
16, 1811-1815. (c) Lai, C.; Gum, R. J.; Daly, M.; Fry, E. H.; Hutchins,
C.; Abad-Zapatero, C.; von Geldern, T. W. Bioorg. Med. Chem. Lett. 2006,
16, 1807-1810. (d) Wright, S. W.; et al. J. Med. Chem. 2002, 45, 3865-
3877. (e) Wright, S. W.; Hageman, D. L.; McClure, L. D.; Carlo, A. A.;
Treadway, J. L.; Mathiowetz, A. M.; Withka, J. M.; Bauer, P. H. Bioorg.
Med. Chem. Lett. 2001, 11, 17-21. (f) Choe, J. Y.; Nelson, S. W.; Arienti,
K. L.; Axe, F. U.; Collins, T. L.; Jones, T. K.; Kimmich, R. D.; Newman,
M. J.; Norvell, K.; Ripka, W. C.; Romano, S. J.; Short, K. M.; Slee, D. H.;
Fromm, H. J.; Honzatko, R. B. J. Biol. Chem. 2003, 278, 51176-51183.
(17) Erion, M. D.; Dang, Q.; Reddy, M. R.; Kasibhatla, S. R.; Huang, J.;
Lipscomb, W. N.; van Poelje, P. D. J. Am. Chem. Soc. 2007, 129, 15480-
15490.
(9) (a) Magnusson, I.; Rothman, D. L.; Katz, L. D.; Shulman, R. G.; Shulman,
G. I. J. Clin. InVest. 1992, 90, 1323-1327. (b) Gastaldelli, A.; Baldi, S.;
Pettiti, M.; Toschi, E.; Camastra, S.; Natali, A.; Landau, B. R.; Ferrannini,
E. Diabetes 2000, 49, 1367-1373. (c) Wajngot, A.; Chandramouli, V.;
Schumann, W. C.; Ekberg, K.; Jones, P. K.; Efendic, S.; Landau, B. R.
Metabolism 2001, 50, 47-52.
(10) Newgard, C. B.; Moore, S. V.; Foster, D. W.; McGarry, J. D. J. Biol. Chem.
1984, 259, 6958-6963.
(11) Herling, A. W.; Burger, H. J.; Schwab, D.; Hemmerle, H.; Below, P.;
Schubert, G. Am. J. Physiol. 1998, 274 (6 Pt 1), G1087-G1093.
(12) Benkovic, S. J.; deMaine, M. M. AdV. Enzymol. Relat. Areas Mol. Biol.
1982, 53, 45-82.
(13) Gitzelmann, R.; Steinmann, B.; Van den Berghe, G. The Metabolic and
Molecular Basis of Inherited Disease; Scriver, C. R., Beaudet, A. L., Sly,
W. S., Valle, D., Eds.; McGraw Hill: New York, 1995; Vol. 1, pp 905-
934.
(14) Pilkis, S. J.; McGrane, M. M.; Kountz, P. D.; el-Maghrabi, M. R.; Pilkis,
J.; Maryanoff, B. E.; Reitz, A. B.; Benkovic, S. J. Biochem. Biophys. Res.
Commun. 1986, 138, 159-166.
9
15492 J. AM. CHEM. SOC. VOL. 129, NO. 50, 2007

Fructose-1,6-Bisphosphatase Inhibitors
A R T I C L E S
Scheme 1 . Drug Design Strategy
Scheme 2 a
prodrug strategy that enables good oral bioavailability led to
the discovery of MB06322 (CS-917), a compound that markedly
lowers glucose levels in animals18,19 and in humans.20
Results
Inhibitor Design. The initial lead inhibitor, 3, lowered
glucose levels in fasted normal rats and in several diabetic
animal models after systemic administration (data not shown)
but failed to lower glucose after oral administration (e100 mg/
kg) due to low oral bioavailability (∼1%). Standard phosphonate
prodrugs21 of 3, such as the bispivaloyloxymethyl prodrug 4,
increased the oral bioavailability (∼2.4%), but not to acceptable
levels. Low oral bioavailability was attributed to a variety of
structural features commonly associated with low absorption,22
including high molecular weight (∼610), low solubility (<10
µg/mL), as well as aqueous and metabolic instability. Conse-
quently, major changes to either the inhibitor structure and/or
the prodrug strategy appeared necessary to achieve the desired
oral bioavailability.
Efforts to redesign the inhibitor focused on structural
modifications that retained the binding interactions shown
previously to contribute significantly to the overall binding
affinity, i.e., hydrogen bonds with the phosphonate, amino
group, and imidazole nitrogen.17 Since the 6-membered ring in
the benzimidazole base contributed little to the overall binding
affinity17 while adding significantly to the molecular weight and
poor aqueous solubility, we modeled compounds in which the
benzimidazole was replaced with five- and six-membered
heterocyclic compounds containing an amino group and a
furanylphosphonate (Scheme 1). Importantly these studies
showed that the distance between the amino and PO32- groups
was still within the range required to simultaneously interact
with the phosphate and purine binding sites (4.0-5.5 Å). Low
energy conformations also showed that hydrogen bonds between
the amino group and FBPase residues V17 and T31 were
possible as were favorable interactions with the imidazole
nitrogen and furanyl oxygen.
a Conditions: (a) CuBr2, EtOH; (b) thiourea, EtOH; (c) formamide, P4S10,
benzene; (d) NBS, CHCl3; (e) PhB(OH)2, Pd(PPh3)4, DME-EtOH, NaHCO3,
80 °C; (f) urea (for oxazole 13), or selenourea (for selenazole 14) or
guanidine-HCl (for imidazole 15), t-BuOH, 80 °C; (g) TMSBr, CH2Cl2.
and 5-aminoimidazoles.23 Alternatively, 2-aminoimidazoles were
reported to exhibit good stability24 and based on modeling
studies were expected to form the same binding site interactions
if the furanylphosphonate and alkyl substituents were attached
at the 4- and 5-positions, respectively (Scheme 1).
Synthesis of FBPase Inhibitors. Five-membered heterocyclic
phosphonates were synthesized from readily accessible 2-acyl-
furan-5-phosphonate diesters as outlined in Scheme 2. The
thiazole analogs 9-12 were prepared from compounds 5-8 by
R-bromination followed by cyclization with thiourea. Reaction
of the R-bromide of 6 with urea, selenourea, and guanidine gave
oxazole (13), selenazole (14), and imidazole (15) phosphonate
diesters, respectively. The 2-desamino analogs 16 and 17 were
prepared from compounds 5 and 6, respectively, by R-bromi-
nation followed by cyclization with in situ generation of
thioformamide. The C5 phenyl thiazole analog 18 was prepared
from 9 via bromination followed by a Suzuki coupling with
phenyl boronic acid. The pyridine phosphonate diester 19 was
synthesized using a Stille coupling of 5-diethylphosphono-2-
tributylstannylfuran and 2-amino-6-chloropyridine followed by
bromination of the 2-aminopyridine and a Suzuki coupling with
phenyl boronic acid (Scheme 3). Conversion of the phosphonate
diethyl esters (9-19) to the corresponding phosphonic acids
(9A-19A) was accomplished using TMSBr. The dimethyl ester
20 was prepared from 10A by treating the corresponding
phosphonodichloridate with methanol in the presence of diiso-
One potential limitation in the design strategy that required
attention early on was the well-recognized instability of electron-
rich azoles, including amino substituted pyrroles and both 4-
(18) 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. U.S.A. 2005, 102, 7970-7975.
(23) (a) Lawson, A. J. Chem. Soc. 1956, 307-310. (b) McLaughlin, M.;
Mohareb, R. M.; Rapoport, H. J. Org. Chem. 2002, 68, 50-54. (c) Dang,
Q.; Liu, Y.; Erion, M. D. J. Am. Chem. Soc. 1999, 121, 5833-5834.
(24) No instability has been reported on the numerous natural products that
contain a 2-aminoimidazole moiety. For example, (a) Bromoageliferin,
Huigens, III, R. W.; Richards, J. J.; Parise, G.; Ballard, T. E.; Zeng, W.;
Deora, R.; Melander, C. J. Am. Chem. Soc. 2007, 129, 6966-6967. (b)
Mauritiamine, Tsukamoto, S.; Kato, H.; Hirota, H.; Fusetani, N. J. Nat.
Prod. 1996, 59, 501-503. (c) Zoanthoxanthins, Xu, Y.; Yakushijin, K.;
Horne, D. A. Tetrahedron Lett. 1999, 33, 4385-4388.
(19) van Poelje, P. D.; Potter, S. C.; Chandramouli, V. C.; Landau, B. R.; Dang,
Q.; Erion, M. D. Diabetes 2006, 55, 1747-1754.
(20) Triscari, J.; Walker, J.; Feins, K.; Tao, B.; Bruce, S. R. Diabetes 2006, 55
(Suppl 1), 444P.
(21) Krise, J. P.; Stella, V. J. AdV. Drug DeliVery ReV. 1996, 19, 287-310.
(22) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J. AdV. Drug
DeliVery ReV. 1997, 23, 3-25.
9
J. AM. CHEM. SOC. VOL. 129, NO. 50, 2007 15493

A R T I C L E S
Scheme 3 a
Dang et al.
a Conditions: (a) Pd(PPh3)4, p-xylene, 170 °C; (b) Br2, AcOH; (c) Ph-
B(OH)2, Pd(PPh3)4, DME-EtOH, NaHCO3, 80 °C; (d) TMSBr, CH2Cl2.
Scheme 4 a
Figure 1. The 2F0-Fc electron density map is contoured at 1 σ and
superimposed on the refined model of the human FBPase-10A complex
at 2.0 Å resolution.
nM) led to a 312-fold improvement in inhibitory potency and
correspondingly to a compound 63- and 5.6-fold more potent
than AMP and 3, respectively.
a Conditions: (a) SOCl2, pyridine, (CH2Cl)2; (b) MeOH, i-Pr2NEt; (c)
NaOH.
Evidence that 10A interacts with the binding site using many
of the same interactions predicted by the initial molecular
modeling studies was indirectly derived from the inhibitory
potencies of closely related analogs (Table 1). Favorable
hydrogen bond interactions by the 2-amino group and van der
Waals interactions by the 5-i-butyl group were suggested by
the 28- and 31-fold reduction in inhibitory potencies observed
for the corresponding 10A analogs, 17A and 9A, respectively,
as well as by the additive loss of inhibitory potency found for
the analog (16A) that contains neither group. Replacement of
the furan with a thiophene (12A) led to a 344-fold decrease in
potency indicating that like 3, 10A likely forms a weak but
favorable electrostatic interaction between the furanyl oxygen
and the amido hydrogen of L30. Consistent with previous
series,17 the phosphonate proved to be essential for binding as
evidenced by the inability of the dimethylester 20, the monom-
ethyl ester 21, or the corresponding carboxylate (22) to inhibit
FBPase.
Less apparent from analysis of the binding interactions was
the dependence of the inhibitory potency on the heterocyclic
base. Replacement of N1 in the imidazole analog (15A) with
sulfur (10A), selenium (14A), and oxygen (13A) led to a 312-,
128-, and 42-fold improvement in potency, respectively. Analo-
gously, replacement of the nitrogen with -CHdCH- and the
alkyl group with phenyl produced a compound (19A) with a
six-membered heterocycle (pyridyl) and good inhibitory potency
(IC50 ) 42 nM).
Table 1. FBPase Inhibitor Structure-Activity Relationships
HL IC50
(nM)a
RL IC50
(nM)b
cpmd
X
Y
Q
R2
R5
AMP
ZMP
3
-
-
-
S
S
S
-
-
-
O
O
O
S
O
O
O
O
O
O
O
O
O
O
-
-
-
1000
12 000 310 000
90
450
16
20 000
-
-
-
-
H
-
-
NH2
370
13 000
61
9A
PO3H2
PO3H2
PO3H2
PO3H2
PO3H2
PO3H2
PO3H2
PO3H2
PO3H2
PO3H2
PO3H2
PO(OMe)2
10A
11A
12A
13A
14A Se
15A NH
16A
17A
18A
NH2 iBu
NH2 nPr
30
NDc
ND
S
O
NH2 nPr >10 000d
NH2 iBu
NH2 iBu
NH2 iBu
H
H
120
39
5000
5500
500
14
2000
150
ND
ND
3500
500
1600
ND
ND
S
S
S
H
iBu
NH2 Ph
NH2 Ph
19A CHdCH
42
20
21
22
S
S
S
NH2 iBu >10 000e
PO(OH)(OMe) NH2 iBu >10 000f
CO2H NH2 Ph >10 000
ND
a Inhibition of human liver FBPase. b Inhibition of rat liver FBPase. c ND
denotes not determined. d Thirty percent at 10 µM. e Greater than 3 mM.
f Greater than 400 µM.
propylethylamine (Scheme 4). Hydrolysis of 20 with NaOH
gave the monomethyl ester 21.
Enzyme Specificity. The lead inhibitor 10A exhibited high
specificity for the AMP binding site of FBPase based on its
lack of activity against five AMP binding enzymes, namely
human erythrocyte AMP deaminase, rabbit muscle glycogen
phosphorylase, human AMP-activated protein kinase, rabbit
muscle adenylate kinase, and rabbit liver phosphofructokinase.18
X-Ray Crystallography. The three-dimensional structure of
the human FBPase-10A inhibitor complex was solved at 2.0
Å resolution (Figure 1, Table 2). The structure showed that 10A
binds to the AMP binding site and forms the predicted
interactions with the binding site residues (Figure 2, Table 3).
FBPase Inhibitor Structure-Activity Relationships. Syn-
thesis and evaluation of substituted 5- and 6-membered hetero-
cyclic bases attached to the 5-position of 2-phosphonfuran led
to the discovery of a new class of potent FBPase inhibitors
(Table 1). The imidazole analog (15A) showed modest inhibitor
potency (IC50 ) 5 µM, Table 1), suggesting that most of the
binding interactions observed in the benzimidazole series (e.g.,
3)17 were likely retained despite removal of one aryl ring, the
translocation of the amino group to the 5-membered heterocycle,
and the rearrangement of the heterocycle nitrogen atoms.
Replacement of the imidazole with a thiazole (10A, IC50 ) 16
9
15494 J. AM. CHEM. SOC. VOL. 129, NO. 50, 2007

Fructose-1,6-Bisphosphatase Inhibitors
A R T I C L E S
Table 2. Data Processing and Refinement Statisticsa
finding. Analysis of the relative differences in solvation free
energies (∆∆Gsol ) ∆Gaq - ∆Ggas) suggested that the main
reason for the differences in potency relate to the energy costs
required to desolvate the heterocycle with the imidazole and
oxazole requiring 1.5 and 1.2 kcal/mol more to desolvate than
the thiazole, respectively. In addition, the transformation of the
i-butyl thiazole to a phenyl-substituted pyridine (10A f 19A)
using the thread method gave a calculated result (1.5 ( 1.2
kcal/mol) consistent with the experimental finding (2.0 kcal/
mol).
FEP calculations were also conducted to evaluate the
contribution of various binding site interactions, including those
by the furanyl oxygen, 2-amino group, and 5-alkyl group (Table
4). The thiazole analog with a thiophene spacer (12A) exhibited
reduced binding affinity (2.7 ( 0.7 kcal/mol) relative to the
furan spacer due to decreased enathalpic energy, which was
partially offset by its reduced desolvation energy (3.4 kcal/mol,
10A f 12A). Consistent with the experimental findings,
removal of the 2-amino group, the alkyl side chain, or both led
to significantly reduced binding affinities of 1.7, 1.5, and
3.1 kcal/mol, respectively, confirming the importance of the
binding interactions made by these groups with the AMP binding
site.
Phosphonic Diamides. The oral bioavailability of 10A was
less than 4% in rats presumably due to its high negative charge
at physiological pH. Standard phosphonate prodrugs21 such as
the acyloxyalkyl29 (23), p-acetoxybenzyl30 (24), S-acyl-2-
thioethyl31 (25), and phenyl32 (26) esters generally led to some
improvement (Table 5) but not to oral bioavailabilities above
15%. Modest oral bioavailability was also observed with
prodrugs (e.g., 27) that cleave following a cytochrome P450
3A-catalyzed oxidation of the ring methine.33 As part of these
efforts, we also explored the potential of phosphonic amides as
prodrugs. Aryl phosphonamidates were prepared based on cell
studies with nucleoside monophosphates,34 which showed that
the intermediate mono-amide generated after esterase-catalyzed
cleavage of the carboxylic ester and subsequent expulsion of
phenol, was converted to the diacid via a phosphoramidase-
catalyzed cleavage of the P-N bond.35 The corresponding
prodrug of 10A, 28, exhibited moderate aqueous instability (pH
) 7.0, t90 ) 14 h), rapid esterase-catalyzed conversion to the
FBPase−10A
space group
P212121
cell dimensions
a ) 67.9 Å, b ) 84.2 Å,
c ) 280.7 Å
data collection and processing
wavelength (Å)
resolution (Å)
0.9764
2.0 (2.07-2.00)
464 199 (5,598)
100 278
91.5 (51.8)
4.6 (1.6)
17.0 (37.5)
7.2 (1.3)
4
reflections
unique reflections
completeness (%)
multiplicity
Rsymb (%)
mean I/σ
number of molecules in
asymmetric unit
refinement statistics
c
Rworking
Rfree
20.4 (31.9)
24.4 (32.9)
21.0
0.0055
1.24
d
average B-factor (Å2)
R.M.S. bond (Å)
R.M.S. angle (deg)
number of protein atoms
number of water molecules
9858
1020
a
b
Values in the parentheses apply to the outer resolution shell. Rsym
)
∑
∑i |Ii (hkl) - I (hkl) |/∑hkl ∑i Ii (hkl), where Ii (hkl) is the intensity
hkl
of an individual reflection, and I (hkl) is the mean intensity obtained from
the multiple observations of symmetry-related reflections taken from one
c
crystal. Rworking ) ∑hkl (|Fo(hkl) - Fc (hkl) |)/∑hkl(|Fo (hkl)|), calculated
d
from reflections in the working set. Rfree) ∑hkl (|Fo(hkl) - Fc (hkl) |)/
∑
hkl(|Fo (hkl)|), calculated from reflections in the test set, which is a random
5% data set selected from the whole data set.
As depicted by modeling studies conducted in advance of both
the synthesis and the X-ray structure, 10A forms a full
complement of interactions with the phosphate binding site. In
addition, the amino group formed hydrogen bonds with the side
chain hydroxyl of T31 and the backbone carbonyl of V17
whereas the thiazole nitrogen formed a favorable electrostatic
interaction with T31. Furthermore, the C5 i-butyl group resided
within van der Waals contact of a hydrophobic surface formed
by the sidechains of M177, L30, and V160.
Free-Energy Calculations. A series of post hoc free-energy
perturbation (FEP) calculations25 provided additional insight into
the binding site interactions of the lead inhibitor series as well
as other factors that contribute to the observed inhibitory
potencies.26 Calculations comparing close structural analogs
were performed using a single topology and a total of 479 ps
of molecular dynamics (MD) simulation whereas calculations
comparing analogs with more substantial structural changes were
performed using the thread method27 and a total of 938 ps of
MD simulation (see Methods). The results from the FEP
calculations using the FBPase model28 (Table 4) were consistent
with the experimental findings in that the binding affinity for
the imidazole analog (15A), regardless of the NH tautomer
evaluated, was at least 2.8 kcal/mol weaker relative to the
corresponding thiazole analog, 10A. The oxazole was intermedi-
ate in its inhibitory potency (10A f 13A, ∆∆Gbind ) 1.7 (
0.7 kcal/mol), which was also consistent with the experimental
(28) (a) Reddy, M. R.; Erion, M. D. J. Am. Chem. Soc. 2001, 123, 6246-6252.
(b) Erion, M. D.; van Poelje, P. D.; Reddy, M. R. J. Am. Chem. Soc. 2000,
122, 6114-6215. (c) Reddy, M. R.; Erion, M. D.; Agarwal, A. In ReViews
in Computational Chemistry; Lipkowitz, K. B., Boyd, D. B., Eds.; Wiley-
VCH Inc.: New York, 2000; Vol. 16, pp 217-304.
(29) Farquhar, D. S.; D. N.; Kattesch, N. J.; Saunders, P. P. J. Pharm. Sci.
1983, 72, 324-325.
(30) (a) Mitchell, A. G.; Thompson, W.; Nicholls, D.; Irwin, W. J.; Freeman,
S. J. Chem. Soc.. Perkins Trans. I 1992, 2345-2353. (b) Dang, Q.; Liu,
Y.; Rydzewski, R. M.; Brown, B. S.; Robinson, E.; van Poelje, P. D.; Colby,
T. J.; Erion, M. D. Bioorg. Med. Chem. Lett. 2007, 17, 3412-3416.
(31) Peyrottes, S.; Egron, D.; Lefebvre, I.; Gosselin, G.; Imbach, J. L.; Perigaud,
C. Mini ReV. Med. Chem. 2004, 4, 395-408.
(32) De Lombaert, S.; Erion, M. D.; Tan, J.; Blanchard, L.; el-Chehabi, L.; Ghai,
R. D.; Sakane, Y.; Berry, C.; Trapani, A. J. J. Med. Chem. 1994, 37, 498-
511.
(33) Erion, M. D.; Reddy, K. R.; Boyer, S. H.; Matelich, M. C.; Gomez-Galeno,
J.; Lemus, R. H.; Ugarkar, B. G.; Colby, T. J.; Schanzer, J.; van Poelje, P.
D. J. Am. Chem. Soc. 2004, 126, 5154-5163.
(34) (a) McGuigan, C.; Pathirana, R. N.; Balzarini, J.; De Clercq, E. J. Med.
Chem. 1993, 36, 1048-1052. (b) Perrone, P.; Luoni, G. M.; Kelleher, M.
R.; Daverio, F.; Angell, A.; Mulready, S.; Congiatu, C.; Rajyaguru, S.;
Martin, J. A.; Leveque, V.; Le Pogam, S.; Najera, I.; Klumpp, K.; Smith,
D. B.; McGuigan, C. J. Med. Chem. 2007, 50, 1840-1849. (c) McGuigan,
C.; Pathirana, R. N.; Mahmood, N.; Devine, K. g.; Hay, A. J. AntiViral
Res. 1992, 311-321.
(35) Saboulard, D.; Naesens, L.; Cahard, D.; Salgado, A.; Pathirana, R.;
Velazquez, S.; McGuigan, C.; De Clercq, E.; Balzarini, J. Mol. Pharmacol.
1999, 56, 693-704.
(25) (a) Zwanzig, R. J. J. Chem. Phys. 1954, 22, 1420-1426. (b) Tembe, B.
L.; McCammon, J. A. Comput. Chem. 1984, 8, 281-283. (c) Kollman, P.
A. Chem. ReV. 1993, 93, 2395-2417.
(26) For reviews of FEP calculations in drug design and lead optimization, see:
(a) Jorgensen, W. L. Science 2004, 303, 1813-1818. (b) Free Energy
Calculations in Rational Drug Design; Reddy, M. R., Erion, M. D., Eds.;
Kluwer/Plenum Press: New York, 2001.
(27) Singh, U. C.; Benkovic, S. J. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 9518-
9523.
9
J. AM. CHEM. SOC. VOL. 129, NO. 50, 2007 15495

A R T I C L E S
Dang et al.
Figure 2. Stereodiagram derived from the X-ray structure of the human FBPase-10A. Putative hydrogen bonds are shown as dotted lines when the distance
between the hydrogen bond acceptor and donor is less than 3.3 Å.
Table 3. Binding Site Interaction Distances
Table 4. Relative Free Energies
hydrogen bond
AMP/10A residue
FBPase
−
AMP
FBPase
Modelb
−
10A
X-raya
X-rayc
NH2
NH2
N
O (furanyl)
5′O (ribose)
3′O (ribose)
3′O (ribose)
OP
OP
OP
OP
OP
OP
OP
OdC (V17)
HO (T31)
HO (T31)
HN (L30)
HO (Y113)
HO (Y113)
HNê (R140)
HN (E29)
HN (G28)
HO (T27)
HN (T27)
HNꢀ (K112)
HO (Y113)
HN (L30)
2.98
3.23
3.54
NA
3.08
2.89
3.46
4.59
-
3.05
3.00
4.08
4.38
-
2.98
2.53
2.95
2.78
2.89
2.73
2.88
2.91
2.70
3.19
4.24
7.27
-
-
-
-
relative free energies (kcal/mol)
∆∆Gbind
2.79
2.89
3.18
3.01
2.78
2.56
-
2.76
3.30
3.03
2.53
2.62
2.69
-
transformation (S1
f
S2)a
∆∆Gsol
∆∆Gbind
(exp)
10A f 13Ab S f O
-1.2 ( 0.5 1.7 ( 0.7
-2.3 ( 0.6 3.0 ( 0.7
1.2
3.4
3.4
2.0
2.1
3.5
10A f 15Ab S f NH (tau1)c
10A f 15Ab S f N; N f (tau2)d -1.4 ( 0.6 2.8 ( 0.7
10A f 9Ae
iBu f H
-0.9 ( 0.6 1.5 ( 0.7
3.6 ( 0.6 1.7 ( 0.7
2.6 ( 0.7 3.1 ( 0.8
10A f 17Ab NH2 f H
10A f 16Ae iBu f H and
NH2 f H
base to P distance
NH2 to P distance
5.28
7.40
5.19
7.31
a Distances in angstroms for X-ray structure of human FBPase-AMP
complex; NA ) not applicable. b Distances in angstroms for computer model
of human FBPase-10A complex. c Distances in angstroms for X-ray
structure of human FBPase-10A complex (2.0 Å resolution).
10A f 19Ae iBu f Ph and
S f HCdCH
-1.3 ( 1.1 1.5 ( 1.2
1.4 ( 0.5 2.7 ( 0.7
0.6
3.4
11A f 12Ab O f S
a Complexes generated from FBPase-10A structure. b Single topology.
c Tautomer 1 (tau1). d Tautomer 2 (tau2). e Double topology.
phosphonamidic acid and low oral bioavailability (Table 5).
Remarkably, the diamides 29-31 displayed good aqueous
stability (pH ) 7.0, t90 > 2 days) and, in general, were converted
much more slowly to the phosphonamidic acid by esterases.
Most importantly, the diamides 29-31 exhibited significantly
better oral bioavailabilities (22-47%) than the other phospho-
nate prodrug classes (Table 5).
Diamide Prodrug Synthesis. Prodrugs 23-28 were prepared
using methods previously described in the literature (see
Supporting Information). Diamide phosphonate prodrugs 29-
31 were prepared by converting 10A to its corresponding dich-
loridate using thionyl chloride and pyridine followed by con-
densation with the corresponding amino acid ester (Scheme 5).
Diamide Prodrug Cleavage Mechanism. Although the
cleavage mechanism was assumed to be similar to that reported
for the monoamidate prodrugs,35 the difference in leaving group
(R-amino acid vs phenol) led us to conduct several studies
monitoring the production of prodrug cleavage products.
Incubation of MB06322 (30) with rat hepatocytes (Figure 3) as
well as rat liver S9 preparations (data not shown) led to a time-
dependent loss of the prodrug and the simultaneous production
of 10A along with an additional product. LC-MS analysis of
the prodrug cleavage products suggested that the new product
was the phosphonamidic acid 33 (Scheme 6), which was
subsequently confirmed by comparing the HPLC retention time
with an independently synthesized synthetic standard. Incubation
of MB06322 with purified pig liver carboxylesterase also led
to 33, but not 10A, suggesting that a carboxylesterase catalyzed
the initial step in the prodrug cleavage mechanism. Intermediate
33 presumably arises from the esterase-cleavage product, 32,
which rapidly proceeds to 33 via a nonenzymatic, unimolecular
reaction involving carboxylate-assisted elimination of an amino
acid and transient formation of a 5-membered cyclic compound
(Scheme 6). 33 was demonstrated to be a prodrug cleavage
intermediate based on its conversion to 10A in the presence of
rat liver S9. The conversion rate (0.019 nmoles/min/mg protein)
was similar to the rate for prodrug MB06322 suggesting that
the second step was the slow step in the production of the
phosphonic acid 10A. The phosphoramidase that catalyzes the
latter step was isolated and shown to be distinct from the
phosphoramidase35 catalyzing the final step in the conversion
of previously reported aryl phosphonamidates based on its
molecular weight (20 vs 50-100 kD) and preferred substrates
(J. Stebbins, personal communication).36
9
15496 J. AM. CHEM. SOC. VOL. 129, NO. 50, 2007

Fructose-1,6-Bisphosphatase Inhibitors
A R T I C L E S
Table 5. FBPase Inhibitor Prodrug SARa
cmpd
X
Y
MW
cLogPb
t90c (hours)
ECRd (nmole/min/mg)
%F e
3
4
-
-
381.3
609.6
302.3
530.6
598.6
506.6
454.5
497.4
477.5
472.5
500.6
528.6
3.72
6.81
1.60
5.01
4.88
3.09
5.24
4.95
5.17
3.65
4.16
5.11
NA
ND
NA
<4
<4
8
ND
ND
14
NA
ND
NA
>300
>300
>300
ND
ND
55
0.9
2.4
<4
11
-
-
10A
23
24
25
26
27
28
29
30
31
OH
OH
O-POM
O-PAB
O-SATE
OPh
-OCH(3-Br-Ph)CH2CH2O-
OPh
Gly-OEt
Ala-OEt
Ala(Me)-OEt
O-POM
O-PAB
O-SATE
OPh
8
ND
0.2
11
12
26
Ala-OEt
Gly-OEt
Ala-OEt
Ala(Me)-OEt
36
>48
>48
43
11
8
22
47
a Abbreviations: NA, not applicable; ND, not determined; POM, pivaloyloxymethylene; PAB, para-acetoxybenzyl; SATE, S-acyl-2-thioethyl. b Calculated
using ADME Boxes V. 3.5. c Prodrug stability at 37 °C in 100 mM potassium phosphate (pH 7) wherein t90 represents the time (in hours) required for 10%
loss in the prodrug concentration. d ECR, esterase conversion rate for prodrugs incubated with rat liver S9 (2 mg/mL) at 37 °C. e Oral bioavailability determined
by measuring urinary excretion of 10A following oral administration of the prodrug vs i.v. administration of 10A.
Scheme 5 a
Scheme 6 . Proposed Diamide Prodrug Cleavage Mechanism
a Conditions: (a) SOCl2, pyridine, (CH2Cl)2; (b) glycine ethyl ester or
L-alanine ethyl ester or 2-methylalanine ethyl ester, i-Pr2NEt, CH2Cl2,
0 °C.
Metformin was also evaluated for its ability to inhibit glucose
production by rat hepatocytes based on reports indicating that
metformin inhibits gluconeogenesis in humans5 and rodents,37
albeit by an indirect mechanism that remains poorly defined.38
Metformin inhibited glucose production from either lactate (IC50
) 2200 ( 300 µM) or glycerol (IC50 ) 3400 ( 1200 µM)
with a potency >1000-fold less than that of MB06322 (Figure
4A). Moreover, metformin and MB06322 differed in their effect
Figure 3. Time course for conversion of prodrug MB06322 to 10A in
freshly isolated rat hepatocytes.
Inhibition of Glucose Production from Rat Hepatocytes.
Concentration-dependent inhibition of the gluconeogenesis
pathway was demonstrated in studies measuring glucose pro-
duction by hepatocytes freshly isolated from fed male Zucker
Diabetic Fatty (ZDF) rats (Figure 4A). The inhibitory potency
of MB06322 was independent of whether glucose production
(36) The major phosphoramidase activity responsible for hydrolysis of the
monoamidate of 10A was purified to near homogeneity from rat liver. In
addition to catalyzing the cleavage of the phosphonamidic acid 33 to the
phosphonic acid 10A and alanine, the enzyme readily cleaved AMP-
phosporamidate to AMP and ammonia. Previously reported aryl phospho-
ramidates of the 5′-monophosphate of AZT (ref 34c) and PMEA (Keith,
K. A.; Hitchcock, M. J. M.; Lee, W. A. Holy, A.; Kern, E. R. Antimicrob.
Agents Chemother. 2003, 47, 2193-2198) were not hydrolyzed by this
phosphoramidase preparation. Some conversion may occur by chemical
hydrolysis, which is fast at low pH (Lee, W. A.; He, G. X.; Eisenberg, E.;
Cihlar, T.; Swaminathan, S.; Mulato, A.; Cundy, K. C. Antimicrob. Agents
Chemother. 2005, 49, 1898-1906).
was from lactate (IC50 ) 1.3 ( 0.1 µM) or glycerol (IC50
)
1.2 ( 0.8 µM) and therefore consistent with the expected
mechanism, i.e., inhibition of gluconeogenesis via inhibition of
FBPase.
(37) Shaw, R. J.; Lamia, K. A.; Vasquez, D.; Koo, S. H.; Bardeesy, N.; Depinho,
R. A.; Montminy, M.; Cantley, L. C. Science 2005, 1642-1646.
9
J. AM. CHEM. SOC. VOL. 129, NO. 50, 2007 15497

A R T I C L E S
Dang et al.
purine binding sites despite their structural dissimilarity. Most
of the interactions were preserved by modest perturbations of
the amino acid sidechains and slight differences in the orienta-
tion of the base. Hydrogen bonds to the 5′-oxygen of AMP and
its ribose were replaced in 3 and 10A by an electrostatic
interaction with the furanyl oxygen and by van der Waals
interactions between the alkyl substituent(s) and a hydrophobic
surface adjacent to the region occupied by the ribosyl moiety
of AMP. Despite the large structural differences between 10A
and AMP and the corresponding differences in binding site
interactions, 10A proved to be a fully functional AMP mimic
capable of binding to the allosteric AMP binding site and
inducing the same protein conformational change as AMP that
converts the catalytically active R-form to the inactive T-form.40
Furthermore, 10A reached the same maximal inhibition as AMP
(∼99%) and like AMP achieved this over a relatively narrow
concentration range (IC10 ) 4 nM and IC90 ) 60 nM). Last,
like AMP, 10A inhibited FBPase synergistically with fructose
2,6-bisphosphate.41,18
Figure 4. Metabolic effects of MB06322 and metformin in male ZDF rat
hepatocytes isolated in the fed state. (A) Inhibition of glucose production
from 14C-lactate. (B) Intracellular lactate/pyruvate ratio with 1 mM glycerol
as the carbon source. All parameters were assessed in duplicate samples
following 30 min of preincubation with drug followed by 1 h of incubation
with substrate. Average results from three separate cellular preparations
are shown. *p < 0.05 compared to controls, student’s t-test.
on cellular redox potential39 with metformin resulting in an
increase in the lactate/pyruvate ratio at concentrations required
to inhibit glucose production (Figure 4B). In addition, as
expected for an inhibitor of an energy-consuming process,
MB06322 treatment resulted in increased energy charge and
ATP levels, whereas metformin showed decreased ATP levels
and no change in energy charge due to its concomitant effects
on the cellular redox potential (Table 6).
In ViWo Activity. Acute administration of MB06322 or
metformin to fasting ZDF rats resulted in significant glucose
lowering (Figure 5A-B) without evidence of hypoglycemia even
at doses 10-fold above the minimum efficacious dose. Glucose
lowering was linked to inhibition of gluconeogenesis and
endogenous glucose production for MB06322 but not met-
formin. Gluconeogenesis was evaluated by measuring 14C-
glucose following intravenous administration of 14C-bicarbonate
to fasted ZDF rats (Figure 5C). MB06322 at doses of 30 and
300 mg/kg inhibited de noVo glucose synthesis, ca. 30 and 85%,
respectively, whereas metformin at a dose (1000 mg/kg) that
significantly lowered glucose (-155 mg/dL) failed to inhibit
the incorporation of 14C-bicarbonate into glucose. Consistent
with these findings, MB06322 at the minimum efficacious dose
(30 mg/kg) reduced endogenous glucose production to a
significantly greater extent than vehicle (-45.3 vs -17.1%, p
< 0.05) whereas metformin (1000 mg/kg) had no effect (-4.3%,
NS) (Figure 5D).
In contrast to AMP and closely related AMP analogs, 10A
discriminated between the AMP site of FBPase and AMP
binding sites of other enzymes. High FBPase specificity was
achieved by exploiting interactions available in the FBPase site
that are unique to FBPase and by avoiding binding interactions
commonly used by other AMP binding sites.42 More specifically,
10A forms favorable electrostatic interactions with the thiazole
nitrogen, amino group, and furanyl spacer oxygen as well as
van der Waals interactions with the i-butyl group. In contrast,
AMP in sites other than FBPase commonly forms hydrogen
bonds with the pyrimidine nitrogens, N1 and/or N3, as well as
with either or both of the ribosyl hydroxyls. The high specificity
of 10A was also attributed to differences in the preferred AMP
binding conformation, which as defined by the torsion angles
corresponding to the glycosidic bond (ø ) -153°), ribose pucker
[γ (O5′-C5′-C4′-C3′) ) 64°] and 5′-phosphate orientation [â
(P-O5′-C5′-C4′) ) -140°] can vary widely across AMP
binding sites. Accordingly, few sites other than FBPase are
expected to readily accommodate molecules that link the base
2-
to the PO3 group by a furanyl spacer group.17
One finding not readily apparent from the structure or initial
modeling studies was the high dependence of inhibitory potency
on the 5-membered heterocycle. The 2-aminoimidazole analog
substituted with an alkyl group at C5 represented the closest
structural analog to the initial lead inhibitor 3 that lacked the
second aryl ring. Replacement of the imidazole in 15A with an
oxazole (13A) or a thiazole (10A) led, respectively, to a 42-
and 312-fold improvement in relative binding affinity. Post hoc
FEP calculations attributed the improvement to decreased
desolvation costs. FEP calculations also supported the experi-
mental observations suggesting that the substituents on the
Discussion
Analysis of the FBPase complexes containing AMP,40 3,17
and 10A highlighted the common binding interactions and those
unique to 10A and therefore likely contributing to its increased
potency and/or specificity (Figure 6). Each inhibitor achieved
the full complement of interactions within the phosphate and
(38) Potential glucose lowering mechanisms for metformin. Hepatic effects: (a)
Argaud, D.; Roth, H.; Wiernsperger, N.; Leverve, X. M. Eur. J. Biochem.
1993, 213, 1341-1348. (b) Cleasby, M. E.; Dzamko, N.; Hegarty, B. D.;
Cooney, G. J.; Kraegen, E. W.; Ye, J. M. Diabetes 2004, 53, 3258-3266.
(c) Perriello, G.; Misericordia, P.; Volpi, E.; Santucci, A.; Santucci, C.;
Ferrannini, E.; Ventura, M. M.; Santeusanio, F.; Brunetti, P.; Bolli, G. B.
Diabetes 1994, 43, 920-928. (d) Zhou, G.; Myers, R.; Li, Y.; Chen, Y.;
Shen, X.; Fenyk-Melody, J.; Wu, M.; Ventre, J.; Doebber, T.; Fujii, N.;
Musi, N.; Hirshman, M. F.; Goodyear, L. J.; Moller, D. E. J. Clin. InVest.
2001, 108, 1167-1174. Non-hepatic effects: (e) Radziuk, J.; Bailey, C.
J.; Wiernsperger, N. F.; Yudkin, J. S. Curr. Drug Targets Immune Endocr.
Metabol. Disord. 2003, 3, 151-169.
(40) (a) Ke, H. M.; Liang, J. Y.; Zhang, Y. P.; Lipscomb, W. N. Biochemistry
1991, 30, 4412-4420. (b) Ke, H. M.; Zhang, Y. P.; Lipscomb, W. N. Proc.
Natl. Acad. Sci. U.S.A. 1990, 87, 5243-5247. (c) Gidh-Jain, M.; Zhang,
Y.; van Poelje, P. D.; Liang, J. Y.; Huang, S.; Kim, J.; Elliott, J. T.; Erion,
M. D.; Pilkis, S. J.; Raafat el-Maghrabi, M.; Lipscomb, W. N. J. Biol.
Chem. 1994, 269, 27732-27738. (d) Iversen, L. F.; Brzozowski, M.;
Hastrup, S.; Hubbard, R.; Kastrup, J. S.; Larsen, I. K.; Naerum, L.; Norskov-
Lauridsen, L.; Rasmussen, P. B.; Thim, L.; Wiberg, F. C.; Lundgren, K.
Protein Sci. 1997, 6, 971-982.
(41) Van Schaftingen, E.; Hers, H. G. Proc. Natl. Acad. Sci. U.S.A. 1981, 78,
2861-2863.
(39) (a) El-Mir, M. Y.; Nogueira, V.; Fontaine, E.; Averet, N.; Rigoulet, M.;
Leverve, X. J. Biol. Chem. 2000, 275, 223-228. (b) Owen, M. R.; Doran,
E.; Halestrap, A. P. Biochem. J. 2000, 348 (Pt 3), 607-614.
(42) (a) Moodie, S. L.; Mitchell, J. B.; Thornton, J. M. J. Mol. Biol. 1996, 263,
486-500. (b) Moodie, S. L.; Thornton, J. M. Nucleic Acids Res. 1993, 21,
1369-1380.
9
15498 J. AM. CHEM. SOC. VOL. 129, NO. 50, 2007

Fructose-1,6-Bisphosphatase Inhibitors
A R T I C L E S
Table 6. Adenylate Levels and Energy Charge in Isolated ZDF Rat Hepatocytes
[drug]
M)
ATPa
ADPa
AMPa
(
µ
(µ
mol/g cells)
(
µ
mol/g cells)
(µ
mol/g cells)
energy chargeb
control
metformin
0
1.35 ( 0.12
1.24 ( 0.05
1.06 ( 0.05c
1.76 ( 0.08c
1.60 ( 0.07
0.37 ( 0.03
0.33 ( 0.02
0.31 ( 0.02
0.39 ( 0.02
0.34 ( 0.01
0.12 ( 0.01
0.10 ( 0.01
0.11 ( 0.01
0.11 ( 0.01
0.10 ( 0.01
0.833 ( 0.003
0.841 ( 0.005
0.825 ( 0.008
0.864 ( 0.006c
0.867 ( 0.005c
3000
10 000
1
MB06322
10
a Adenylates were measured after 1.5 h of drug or vehicle exposure in duplicate samples in three different cell preparations. b Energy charge is defined
as: ([ATP] + 1/2 [ADP])/([ATP] + [ADP] + [AMP]). c p < 0.05 vs control: ANOVA, Dunnett’s post hoc test.
Figure 5. (A-B) Blood glucose lowering following oral administration
of (A) MB06322 (30 or 300 mg/kg) or (B) metformin (100 or 1000 mg/
kg) relative to vehicle controls. All animals were dosed at 7:00 a.m. and
food was withdrawn for the remainder of the protocol. n ) 8/group. *p <
0.05 vs vehicle, repeated measures ANOVA with Dunnett’s post hoc test.
(C) De noVo glucose synthesis from 14C-bicarbonate in fasting male ZDF
rats following acute administration of MB06322 (30 or 300 mg/kg) or
metformin (1000 mg/kg). Results are expressed as % inhibition relative to
vehicle-treated animals. n ) 4-9/group. (D) Effects of vehicle, MB06322
(single dose, 30 mg/kg), or metformin (single dose, 1000 mg/kg) treatment
on endogenous glucose production in male ZDF rats. Black bars indicate
baseline values whereas the blue bars indicate post-treatment values. n )
6 or 7/group. *p < 0.05 vs vehicle, ANOVA with Dunnett’s post hoc test.
Figure 6. Overlay of human FBPase complexes containing AMP (yellow),
3 (green), and 10A (red). The average rms deviation was 0.44 Å for CR
and 0.90 Å for the sidechains between the complexes with AMP and 10A.
The average rms deviation was 0.22 Å for CR and 0.85 Å for the sidechains
between the complexes with 3 and 10A.
cleavage mechanism for aryl phosphonamidates,35,44 cleaves to
the phosphonamidic acid 33 via an intramolecular reaction
involving formation of a hydrolytically unstable cyclic inter-
mediate (Scheme 6). The rate of ring closure is likely related
to the lability of the leaving group, which could explain the
slower conversion rates associated with phosphonic diamides
relative to aryl phosphonamidates (Table 5).45 Similarly, conver-
sion rates for the diamide prodrug class are slow relative to
acyloxyalkyl esters, which, after ester cleavage, breakdown
rapidly to the corresponding mono-phosphonic acid and form-
aldehyde.29 Accordingly, the improved oral bioavailability
associated with the phosphonic diamides may relate to the
reduced formation of the phosphonamidic acid in the gas-
trointestinal tract and the inability of the negatively charged
mono phosphonic acid to undergo appreciable absorption.
In addition to the improved oral bioavailability, the phos-
phonic diamide prodrug class exhibits several other potential
advantages over standard phosphonate prodrugs. First, phos-
phonate prodrugs such as acyloxyalkyl esters and substituted
5-membered heterocycle, that is, C5 i-butyl group, C2 amino,
and C4 furanyl oxygen, contributed to the overall binding
affinity. Moreover, results from FEP calculations also agreed
with the experimental finding demonstrating that the 5-mem-
bered heterocycle could be replaced with a 6-membered
heterocycle, namely a phenyl-substituted pyridine, without loss
of binding affinity.
Oral delivery of this new FBPase inhibitor series required
the development of novel phosphonate prodrugs, since standard
phosphonate prodrugs21 failed to achieve oral bioavailabilities
of more than 15%. Phosphonic diamides were prepared despite
previous efforts showing poor oral bioavailability with another
phosphonic acid43 and modest oral bioavailability for aryl
phosphonamidate 28. The result was a significant improvement
in oral bioavailability (22-47%). Mechanistic studies suggest
the diamide prodrug class undergoes an initial esterase-catalyzed
conversion to a transient intermediate such as 32 (could also
be biscarboxylate), which, analogous to the previously described
(44) Jacobsen, N. E.; Bartlett, P. A. J. Am. Chem. Soc. 1983, 105, 1613-1619.
(45) An analysis of phosphonamidates as angiotensin converting enzyme
inhibitors showed that phenol is a much better leaving group than L-alanyl-
L-proline at physiological pH: Galardy, R. E.; Grobelny, D. A J. Med.
Chem. 1985, 28, 1422-1427.
(43) Serafinowska, H. T.; Ashton, R. J.; Bailey, S.; Harnden, M. R.; Jackson,
S. M.; Sutton, D. J. Med. Chem. 1995, 38, 1372-1379.
9
J. AM. CHEM. SOC. VOL. 129, NO. 50, 2007 15499

A R T I C L E S
Dang et al.
benzyl esters often show significant hydrolytic instability and
therefore require additional formulation development and special
packaging strategies to achieve adequate shelf life.46 The
diamide class showed not only good aqueous stability (t90 > 2
days) but also stability across a broad pH range, which benefits
oral absorption by reducing prodrug loss during transit through
the gastrointestinal tract.47 Second, the diamide class generates
nontoxic byproducts, for example, alanine and ethanol, and
therefore is associated with a lower toxicological risk than other
prodrug classes. Last, the diamides that comprise identical amino
acid esters avoid introducing a new stereocenter at phosphorus
and the associated synthetic complications.
(22-47%) than the standard phosphonate prodrugs but also
avoided several well-known limitations commonly associated
with these prodrugs, including byproduct toxicity and aqueous
instability. Most importantly, these efforts led to MB06322, a
prodrug of 10A that demonstrated that inhibition of gluconeo-
genesis via inhibition of FBPase results in dose-dependent
inhibition of endogenous glucose production and a reduction
in blood glucose levels in the ZDF rat. Given the contribution
of endogenous glucose production to hyperglycemia in human
diabetes, FBPase inhibitors may represent a promising new class
of anti-diabetic agents.
Experimental Procedures
Conversion of MB06322 to 10A led to dose-dependent
inhibition of GNG in isolated hepatocytes and in the ZDF rat.
Inhibition of GNG in the ZDF rat was associated with decreased
endogenous glucose production (EGP) and a dose-dependent
reduction in blood glucose levels. These results not only
demonstrate the importance of GNG to both EGP and the
hyperglycemic state of the ZDF rat, but also to the magnitude
in which both EGP and glucose levels can be reduced by FBPase
inhibitors. Metformin, the only marketed drug for diabetes
whose glucose lowering activity is primarily attributed to
reduced EGP,5 inhibited GNG in rat hepatocytes, but only at a
concentration >1000-fold higher than the IC50 for MB06322.
Moreover, unlike MB06322, inhibition by metformin was
accompanied by effects on the cellular redox state. In the ZDF
rat, metformin at doses that significantly lowered blood glucose
failed to inhibit GNG or reduce EGP after oral administration
for a single day (Figure 5) or 14 days (data not shown). These
results suggest that in the rat, metformin lowers glucose by a
mechanism other than inhibition of EGP.
General Methods. All moisture-sensitive reactions were performed
under a nitrogen atmosphere using oven-dried glassware and anhydrous
solvents purchased and used without further manipulation. TLC was
performed on Merck silica gel 60 F254 TLC glass plates and visualized
with UV light or a suitable stain. Flash chromatography was performed
on 230-400 mesh EM Science silica gel 60. Melting points were
recorded on a Thomas-Hoover capillary melting point apparatus and
are uncorrected. 1H NMR spectra were recorded on Varian Gemini or
Mercury spectrometers operating at 200 or 300 MHz, respectively. Mass
spectrometry data were determined at Mass Consortium Corp. (San
Diego, CA) or on a Perkin-Elmer Sciex API2000 LC-MS system.
Elemental microanalyses were performed by NuMega Resonance Labs,
Inc. (San Diego, CA) or by Robertson Microlit Laboratories (Madison,
NJ).
[5-(4-Methylpentanoyl)furan-2-yl]phosphonic Acid Diethyl Ester
(6). A solution of the 2-furan-2-yl-2-[(3-methyl)butyl]-1,3-]dioxolane
(see Supporting Information, 300 g, 1.43 mol) in anhydrous THF (700
mL) was treated with N,N,N′,N′-tetramethylethylene-diamine (TMEDA)
(215.5 mL, 1.428 mol), cooled to -45 °C, and treated with a solution
of n-butyl lithium (2.5 M, 600 mL) in hexane via an addition funnel
at a speed that maintained the reaction temperature between -45 and
-40 °C. After the addition was complete, the reaction solution was
stirred at 0-2 °C for 1 h. In a separate flask, a solution of diethyl
chlorophosphate (228 mL, 1.11 mol) in anhydrous THF (700 mL) was
cooled to -45 °C and treated with the above generated furan anion
solution (cooled to -45 °C) via a cannula at a rate that maintained the
reaction temperature between -45 and -40 °C. After the addition was
complete, the reaction was allowed to warm to room temperature and
stir for 2 h. The solvent was then removed in Vacuo, and the residue
partitioned between ethyl acetate (1.1 L) and water (880 mL). The layers
were separated, and the aqueous phase was extracted with ethyl acetate
(800 mL). The organic extracts were then combined, washed with brine
(1 L), dried (MgSO4), filtered, and concentrated in Vacuo to give crude
{5-[2-[(3-methyl)butyl]-1,3-]dioxolan-2-yl]-furan-2-yl}phosphonic acid
diethyl ester as a dark-red oil (488 g). The ketal was then dissolved in
methanol (856 mL), treated with hydrochloric acid (1 N, 215 mL),
and heated at 50 °C. After 12 h, the methanol was removed in Vacuo,
and the resulting dark oil partitioned between ethyl acetate (850 mL)
and saturated sodium bicarbonate solution (430 mL). The layers were
separated, and the aqueous phase was extracted with ethyl acetate (400
mL). The organic extracts were combined, washed with brine (850 mL),
dried (MgSO4), filtered, and concentrated in Vacuo to a dark-red oil
(414 g), which was purified by vacuum distillation (126-145 °C/ 0.10
Conclusions
Transformation of the initial FBPase inhibitor lead,17 3, into
a drug candidate suitable for testing the FBPase concept in
diabetic animal models and ultimately in human clinical trials
proved to be a major endeavor. The low oral bioavailability of
the bisPOM prodrug of 3, that is, 4 (2.4%), was attributed to
its high MW, low solubility, and inherent susceptibility to
esterase cleavage in the gastrointestinal tract. Accordingly,
substantial improvement in the oral bioavailability required
changes in both the inhibitor structure and prodrug strategy.
The FBPase inhibitor 10A emerged from efforts to reduce the
molecular weight of 3 using structural information and molecular
modeling studies to guide analog selection and synthesis. 10A
binds to the AMP binding site and elicits potent FBPase
inhibition (IC50 ) 16 nM) despite its structural dissimilarity to
AMP. Oral delivery of 10A was accomplished by successfully
using a new phosphonate prodrug class, namely the phosphonic
diamide, which not only achieved greater oral bioavailabilities
(46) Stability studies on phosphonate prodrugs: (a) Cundy, K. C.; Fishback, J.
A.; Shaw, J.; Lee, M. L.; Soike, K. F.; Visor, G. C.; Lee, W. A. Pharm.
Res. 1994, 11, 839-843. (b) Benzaria, S.; Pelicano, H.; Johnson, R.; Maury,
G.; Imbach, J.; Aubertin, A.; Obert, G.; Gosselin, G. J. Med. Chem. 1996,
39, 4958-4965. (c) Cundy, K. C.; Sue, I.; Visor, G. C.; Marshburn, J.;
Nakamura, C.; Lee, W. A.; Shaw, J. J. Pharm. Sci. 1997, 86, 1334-1338.
(d) Shaw, J.; Louie, M. S.; Krishnamurthy, V. V.; Arimilli, M. N.; Jones,
R. J.; Bidgood, A. M.; Lee, W. A.; Cundy, K. C. Drug Metab. Dispos.
1997, 25, 362-366. (e) Yuan, L.; Dahl, T. C.; Oliyai, R. Pharm. Res. 2000,
17, 1098-1103. (f) Meier, C.; Gorbig, U.; Muller, C.; Balzarini, J. J. Med.
Chem. 2005, 48, 8079-8086.
1
Torr) to give 6 as a pale-orange oil (112 g, 61%). H NMR (CDCl3):
δ 7.19 (m, 2H), 4.22 (m, 4H), 2.87 (t, 2H, J ) 6.6 Hz), 1.61 (m, 3H),
1.39 (m, 6H), 0.97 (d, 6H, J ) 6.6 Hz).
2-Amino-5-isobutyl-4-[2-(5-diethylphosphono)furanyl]thiazole (10).
A suspension of [5-(4-methyl-pentanoyl)-furan-2-yl]phosphonic acid
diethyl ester (6) (500 g, 1.65 mol) and copper(II) bromide (838 g, 3.76
mol) in anhydrous ethanol (5.5 L) was heated to reflux under nitrogen
for 1 h. The reaction was then cooled to room temperature and filtered,
and the solid was washed with ethanol (0.5 L). The combined filtrate
(47) Phosphonamidates were reported to exhibit good stability across the pH
range of 5-8.5: Jacobsen, N. E.; Bartlett, P. A. J. Am. Chem. Soc. 1981,
103, 654-657.
9
15500 J. AM. CHEM. SOC. VOL. 129, NO. 50, 2007

Fructose-1,6-Bisphosphatase Inhibitors
A R T I C L E S
and washing was concentrated under reduced pressure to an oil, which
was dissolved in ethyl acetate (4 L) and quenched with a saturated
sodium bicarbonate solution (2.8 L). The resulting mixture was filtered
through a Celite pad (washed with ethyl acetate, 0.5 L), and the layers
were separated. The aqueous phase was extracted with ethyl acetate
(2.5 L), and the organic phases were combined and washed with a
saturated solution of ammonium chloride, which was subsequently re-
extracted with dichloromethane (2 × 30 mL). The organic extracts were
combined, washed with brine (1.2 L), dried (MgSO4), filtered, and
concentrated in Vacuo to give the crude [5-(2-bromo-4-methyl-
pentanoyl)-furan-2-yl]phosphonic acid diethyl ester as a greenish-yellow
oil (587 g, 93%). The bromoketone (1086 g, 2.848 mol) was then
dissolved in anhydrous ethanol (4 L), treated with thiourea (219 g, 2.88
mol), and heated at reflux. After 1 h, the reaction mixture was cooled
to room temperature, concentrated under reduced pressure, neutralized
with saturated sodium bicarbonate (3.5 L), and extracted with ethyl
acetate (4 L) followed by dichloromethane (1.5 L), ethyl acetate (2 L),
and dichloromethane (0.7 L). The organic extracts were combined,
washed with brine (2 L), dried (MgSO4), and filtered, and the filtrate
was concentrated under reduced pressure to give a light-orange solid.
The crude solid was crystallized from ethyl acetate to give compound
10 as a yellow solid (738 g, 72%). White needles (EtOAc-hexane).
) 6.9 Hz), 1.79 (m, 1H), 1.24-1.10 (m, 12H), 0.90 (d, 6H, J ) 7.2
Hz). [MH]+ calcd for C21H33N4O6PS: 501. Found: 501. Anal. calcd
for C21H33N4O6PS: C: 50.39, H: 6.65, N: 11.19. Found, C: 50.30,
H: 6.46, N: 11.07.
FBPase Crystal Growth. The hanging drop method was used to
grow crystals of the FBPase-10A complex at room temperature. Each
drop contained human FBPase purified to homogeneity using the
procedure previously described17 (2 µL, 24 mg/mL) in a solution (pH
7.2) containing 2.7 mM 10A, 5 mM sodium malonate, 1 mM
dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 0.1 mM EDTA,
and 2 mM fructose-1,6-bisphosphate that was mixed with 2 µL of 6%
(vol/vol) polyethylene glycol 8,000 (PEG-8K)/0.1 M Tris-HCl (pH 8.2),
and equilibrated against 1 mL 6% PEG-8K/0.1 M Tris-HCl (pH 8.2).
The following day, the resulting crystals were soaked in 20% glycerol/
80% crystallographic buffer (6% PEG-8K/0.1 M Tris-HCl, pH 8.2)
(vol/vol) for about 5 s and then plunged into liquid nitrogen (77 K)
and stored until data collection.
Data Collection and Analysis. The diffraction data were collected
at 100 K on beamline F-1 of the Cornell High-Energy Synchrotron
Source (CHESS). The crystal was in the space group P212121. One
tetramer was present in the asymmetric unit.
A search model was obtained from the PDB code 1FTA.40c Molecular
replacement was then carried out by using AMORE.48 The model of
10A was generated by CS Chem3D Pro (http://www.camsoft.com) and
modified according to the electron density in the simulated annealing
omit map, and the conformation of the model was optimized by energy
minimization in CS Chem3D Pro. The structure of human liver FBPase
complexed with 10A was refined by using the CNS_SOLVE package.49
The model was then visualized and refined against 2Fo-Fc and Fo-Fc
maps by using the program O.50 The stereochemical properties of the
intermediate and the final structures were examined by PROCHECK.51
In the final structure, 88.1% residues were in the most favored regions,
11.2% residues were in additional allowed regions, and 0.7% residues
were in generously allowed regions.
Computational Details. All molecular dynamics, molecular me-
chanics, and free-energy perturbation calculations were carried out with
the AMBER program using an all-atom force field52 and the SPC/E
model potential53 to describe water interactions. Procedures for generat-
ing the computer model of the FBPase inhibitor complexes and for
developing all force field parameters for nonstandard residues are as
described in the previous manuscript.17 Five- and six-membered
heterocycles substituted with 2-furanyl phosphonic acid were evaluated
computationally using the three-dimensional structure of the human
FBPase-ZMP complex (W. N. Lipscomb, unpublished results). The
solvated inhibitor and inhibitor-FBPase complex were energy minimized
using a four-stage protocol and the BORN module of the AMBER
program.
1
mp 121-123 °C. H NMR (CDCl3): δ 7.18 (m, 1H), 6.59 (m, 1H),
6.18 (bs, 2H), 4.18 (m, 4H), 2.78 (d, 2H, J ) 6.6 Hz), 1.85 (m, 1H),
1.33 (t, 6H, J ) 7 Hz), 0.98 (d, 6H, J ) 6.6 Hz). [MH]+ calcd for
C15H23N2O4PS: 359. Found: 359. Anal. calcd for C15H23N2O4PS: C:
50.27, H: 6.47, N: 7.82. Found, C: 50.25, H: 6.31, N: 7.79.
2-Amino-5-isobutyl-4-[2-(5-phosphono)furanyl]thiazole (10A). A
solution of 10 (593 g, 1.65 mol) in anhydrous dichloromethane (3.5 L)
was cooled to 5 °C and treated with TMSBr (831 g, 5.42 mol). After
stirring at room temperature for 16 h, the reaction mixture was
evaporated to dryness to yield a residue that was subsequently dissolved
in dichloromethane (2 L) and reconcentrated under reduced pressure.
The resulting solid was dissolved and concentrated twice from methanol
(2 × 2 L) to afford a foam, which was then partitioned between sodium
hydroxide (2 M, 1.8 L) and ethyl acetate (1 L). The layers were
separated and the aqueous phase was extracted with ethyl acetate (2 ×
1 L). The aqueous phase was concentrated slightly under reduced
pressure to remove residual ethyl acetate and diluted with water (0.8
L) and methanol (1.25 L). The resulting orange solution was filtered,
and the filtrate was treated with concentrated hydrochloric acid (12
M, 220 mL) until pH ) 1. The resulting solid was collected via
filtration, washed with water (2 × 0.5 L), and dried under vacuum to
1
give 10A as an off-white solid (476 g, 95%). mp 200-204 °C. H
NMR (DMSO-d6): δ 6.97 (m, 1H), 6.74 (m, 1H), 2.74 (d, 2H, J ) 6.6
Hz), 1.76 (m, 1H), 0.88 (d, 6H, J ) 6.6 Hz). [MH]+ calcd for
C11H15N2O4PS: 303. Found: 303. Anal. calcd for C11H15N2O4PS: C:
43.71, H: 5.00, N: 9.27. Found, C: 43.64, H: 4.78, N: 9.13.
Free-energy calculations were conducted as previously described.28
Prior to the mutation, the system was minimized using the four-stage
protocol followed by an equilibration period using 20 ps of MD
simulation and a 2 fs time step. Mutations suitable for the single
topology method were conducted using 51 windows with each window
comprising 3 ps of equilibration and 6 ps of data collection or a total
2-Amino-5-isobutyl-4-[5-[N,N′-bis[(S)-1-ethoxycarbonyl]ethyl]-
phosphonodiamido]-2-furanyl]thiazole (30). A suspension of 10A (50
g, 165 mmol) in anhydrous (ClCH2)2 (1 L) was treated with SOCl2 (50
mL, 685 mmol) followed by anhydrous pyridine (1.3 mL, 16 mmol).
The resulting mixture was heated to reflux under nitrogen. After 3 h,
the reaction solution was cooled and evaporated to give a dark-orange
semisolid (70.3 g), which was then dissolved in anhydrous (ClCH2)2
(0.5 L), cooled to -22 °C, and treated with a solution of L-alanine
ethyl ester (96.3 g, 822 mmol) in anhydrous CH2Cl2. After stirring for
2 h at room temperature, the solution was evaporated to dryness, and
the resulting residue partitioned between ethyl acetate (1 L) and water
(400 mL). The layers were separated, and the organic phase was washed
with brine (300 mL), dried (MgSO4), filtered, and evaporated to give
a brown-orange sludge, which was purified by flash chromatography
(SiO2, 2 kg, 13 cm ID column, 2, 3, and 4% methanol in EtOAc,
gradient elution) to give compound 30 as a pale-orange solid (45.9 g,
55%). mp 165-166.5 °C. 1H NMR (DMSO-d6): δ 6.95-6.90 (m, 3H),
6.50 (m, 1H), 5.07-4.88 (m, 2H), 4.06-3.71 (m, 6H), 2.77 (d, 2H, J
(48) Navaza, J. Acta Cryst. 1994, A50, 157-163.
(49) 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. Acta Cryst.
1998, D54, 905-921.
(50) Jones, T. A.; Zou, J-Y.; Cowan, S. W.; Kjeldgaard, M. Acta Cryst. 1991,
A47, 110-119.
(51) Laskowski, R. A.; MacArthur, M. W.; Moss, D. S.; Thornton, J. M. J.
App. Cryst. 1993, 26, 283-291.
(52) (a) Weiner, S. J.; Kollman, P. A.; Case, D. A.; Singh, U. C.; Ghio, C.;
Alagoha, G.; Profeta, S., Jr.; Weiner, P. K. J. Am. Chem. Soc. 1984, 106,
765-784. (b) Singh, U.C.; Weiner, P. K.; Caldwell, J. K.; Kollman, P. A.;
AMBER Version 3.0; University of California at San Francisco: San
Francisco, CA, 1986.
(53) (a) Berendsen, H. J. C.; Grigera, J. R.; Straatsma, T. P. J. Phys. Chem.
1987, 91, 6269-6271. (b) Reddy, M. R.; Berkowitz, M. J. Chem. Phys.
1988, 88, 7104-7110.
9
J. AM. CHEM. SOC. VOL. 129, NO. 50, 2007 15501

A R T I C L E S
Dang et al.
of 479 ps of MD simulation. The double topology method27 was used
for mutations involving larger structural changes. In this case, calcula-
tions entailed two stages with each stage using 51 windows and each
window comprising 3 ps of equilibration and 6 ps of data collection or
a total of 938 ps of MD simulation. Each mutation followed the
doublewide sampling procedure, and the results reported are based on
the averages from the backward and forward simulations of the
mutation. Errors were estimated for each window by dividing the
window statistics into four groups (in both forward and backward
directions) and computing the standard deviation for the indicated free-
energy change. The root-mean-square of these window errors are
reported as a measure of the statistical uncertainty in the result for each
mutation.
Hepatocyte Studies. Hepatocytes from freely feeding male ZDF
rats were prepared using the method of Berry and Friend54 as modified
by Groen et al.55 Hepatocytes were suspended at 15 mg/mL (gluco-
neogenesis) or 60 mg/mL (all other assays) in Krebs bicarbonate buffer
supplemented with 20 mM glucose. Reactions were performed in a
shaking water bath (37 °C).
Hepatocyte Uptake and Metabolism of Phosphonic Diamides.
Compound 30 was added to a concentration of 100 µM, and aliquots
of the cell suspension were removed at various time points over 90
min. The cell suspensions were centrifuged for 5 min (Eppendorf
microfuge, 16 000× g, room temperature). The supernatant was
removed and the cell pellet extracted by addition of 60% methanol in
water (v/v). The methanolic extracts were clarified by centrifugation
and analyzed for 30, 33, and 10A by HPLC as described in the
Supporting Information.
tail artery and vein catheters under brief halothane anesthesia.57 Food
was withheld for the remainder of the protocol. After a 4-hour recovery
period, MB06322 (30 or 300 mg/kg), metformin (1000 mg/kg), or
vehicle was administered orally followed by an intravenous bolus of
14C-bicarbonate (0.4 µCi/g) 3 h later. After 20 min, blood was obtained
from the inferior vena cava under halothane anesthesia, diluted 1:12
in water, and processed for 14C-glucose content.
The rate of EGP was determined using standard tracer dilution
techniques.58 Male ZDF rats (10-11 weeks old) were fasted at 8 am
and instrumented with tail catheters as described above. Food was
withheld for the remainder of the protocol. A 16 µCi priming dose of
6-[3H]-glucose was administered intravenously at 1:00 p.m. and
followed by a maintenance infusion at a rate of 8 µCi/h. Neither blood
glucose nor insulin was clamped to better simulate physiological
conditions. Baseline arterial blood samples to evaluate the specific
activity of blood glucose were obtained at 4:00 and 4:30 p.m. At 5:00
p.m., rats were divided into 3 glucose-matched groups and vehicle,
MB06322 (30 mg/kg) or metformin (1000 mg/kg), was administered
orally. Additional arterial blood samples were obtained 1.5 and 3 h
following administration of the test agent. The specific activity of
glucose in arterial blood samples was determined as described.58 The
rate of EGP was calculated by means of the Steele equation59 to
compensate for the delivery of tracer to a changing systemic glucose
pool.
Statistical Analysis. Results are expressed as means ( SEM.
Statistical significance was determined with the student’s t-test or
analysis of variance (ANOVA; one-way or repeated measures) with
Dunnett’s post hoc test as appropriate. Statview software (SAS Institute,
Cary, NC) was employed.
Metabolic Effects. Hepatocytes were preincubated with 30 (0.3-
30 µM) or metformin (0.3-10 mM) for 30 min prior to the addition of
10 mM 14C-lactate (0.25 µCi/mol) or 1 mM 14C-glycerol (0.25 µCi/
mol). For analysis of glucose production, the supernatants were
deproteinated with equal volumes of 3 M Ba(OH)2 and 3 M ZnSO4
and incubated with or without hexokinase (2.8 U/mL; Sigma, St. Louis,
MO) in the presence of 10 mM ATP (1 h, 37 °C). Anion exchange
resin was added to each reaction mixture in a 1:2 ratio (wt/vol), and
the slurry was incubated with shaking for 1 h (room temperature).
Following removal of the resin by centrifugation, radioactivity in the
supernatants was assessed by scintillation counting. 14C-glucose produc-
tion was calculated from the difference in counts between non-
hexokinase-treated and hexokinase-treated (nonspecific) samples. Half-
maximal inhibitory (IC50) values were determined by 4-parameter
logistic regression with use of Systat software (Point Richmond, CA).
Cell pellets were assayed for intracellular lactate and pyruvate as
described.56 For adenylate measurements, hepatocyte suspensions were
spiked directly into acetonitrile (2:3, vol/vol), clarified by centrifugation,
and analyzed via liquid-chromatography tandem mass spectroscopy
(LC-MS/MS) (see Supporting Information).
Acknowledgment. We thank Kevin Fan, Yan Liu, Dan
Cashion, Zhili Sun, Joe Kopcho, William Schulz, and Jay DaRe
for their assistance in compound synthesis; Tim J. Colby, Kathy
Effenberger, Michael J. Estes, Meredith Pankop, and Emily
Topczewski for characterization of the compounds in the
biological assays; Dr. Sridhar Prasad for graphical representa-
tions of the binding interactions; and Dr. Scott Hecker for review
of the manuscript.
Supporting Information Available: Analytical data and
procedures for the preparation of compounds 5, 7-9, 11-29,
9A, 11A-19A, 31, and 33. Enzyme assays, methods for
determining solubility, chemical and biological stability, and
oral bioavailability, and analytical methods for prodrugs, prodrug
metabolites, and cellular adenylates. Procedure for developing
force field parameters for nonstandard residues, as well as
procedures for energy minimizations in solvent and the complex
and for the development of the computer models of FBPase
complexed with 10A. Complete ref 16d. This material is
available free of charge via the Internet at http://pubs.acs.org.
ZDF Rat Studies. Acute glucose lowering was measured in ZDF
rats (ca. 10 weeks old) that were divided into glucose-matched groups
at 7:00 a.m. and dosed orally with MB06322 (30 and 300 mg/kg,
suspension) or metformin (100 and 1000 mg/kg, solution) or an equal
volume of vehicle (0.5% carboxymethyl cellulose in water). Food was
withheld for the remainder of the protocol. Blood glucose was measured
at baseline and at regular intervals following treatment using the
methods described in the Supporting Information.
JA074871L
(56) Bergmeyer, H. U. In Methods of Enzymatic Analysis; Academic Press: New
York, 1974; pp 572-579.
(57) Koo, S. H.; Flechner, L.; Qi, L.; Zhang, X.; Screaton, R. A.; Jeffries, S.;
Hedrick, S.; Xu, W.; Boussouar, F.; Brindle, P.; Takemori, H.; Montminy,
M. Nature 2005, 437, 1109-1111.
De noVo glucose production was measured in conscious, fasting,
male ZDF rats of ca. 11 weeks of age. Rats were instrumented with
(58) Rossetti, L.; DeFronzo, R. A.; Gherzi, R.; Stein, P.; Andraghetti, G.; Falzetti,
G.; Shulman, G. I.; Klein-Robbenhaar, E.; Cordera, R. Metabolism 1990,
39, 425-435.
(54) Berry, M. N.; Friend, D. S. J. Cell Biol. 1969, 43, 506-520.
(55) Groen, A. K.; Sips, H. J.; Vervoorn, R. C.; Tager, J. M. Eur. J. Biochem.
1982, 122, 87-93.
(59) Minoura, H.; Takeshita, S.; Yamamoto, T.; Mabuchi, M.; Hirosumi, J.;
Takakura, S.; Kawamura, I.; Seki, J.; Manda, T.; Ita, M.; Mutoh, S. Eur.
J. Pharmacol. 2005, 519, 182-190.
9
15502 J. AM. CHEM. SOC. VOL. 129, NO. 50, 2007