Modified 3-alkyl-1,8-dibenzylxanthines as GTP-competitive inhibitors of phosphoenolpyruvate carboxykinase

Article abstract of DOI:10.1016/S0960-894X(03)00722-4

The first non-substrate like inhibitors of human cytosolic phosphoenolpyruvate carboxykinase (PEPCK) competitive with GTP are reported. An effort to discover orally active compounds that improve glucose homeostasis in Type 2 diabetics by reversibly inhibi

Full text of DOI:10.1016/S0960-894X(03)00722-4

Bioorganic & Medicinal Chemistry Letters 13 (2003) 3607–3610
Modified 3-Alkyl-1,8-dibenzylxanthines as GTP-Competitive
Inhibitors of Phosphoenolpyruvate Carboxykinase
Louise H. Foley,a,* Ping Wang,a Pete Dunten,a Gwendolyn Ramsey,b
Mary-Lou Gublerb and Stanley J. Wertheimerb
aDepartment of Discovery Chemistry, Roche Research Center, Hoffmann-La Roche Inc., Nutley, NJ 07110, USA
bDepartment of Metabolic Diseases, Roche Research Center, Hoffmann-La Roche Inc., Nutley, NJ 07110, USA
Received 29 January 2003; accepted 16 May 2003
Abstract—The first non-substrate like inhibitors of human cytosolic phosphoenolpyruvate carboxykinase (PEPCK) competitive
with GTP are reported. An effort to discover orally active compounds that improve glucose homeostasis in Type 2 diabetics by
reversibly inhibiting PEPCK led to the discovery of 1-allyl-3-butyl-8-methylxanthine (5). We now report modifications at N-1 and
C-8 that improved the in vitro activity of the initial xanthine HTS hit by 100-fold and a developing SAR for this class of inhibitor.
# 2003 Elsevier Ltd. All rights reserved.
The control of glucose production is one of the key
aspects of an anti-diabetic therapy. Increased hepatic
gluconeogenesis is thought to lead to fasting hyperglyce-
mia in patients withType 2 diabetes. Teh cytosolic
phosphoenolpyruvate carboxykinase (PEPCK) enzyme is
known to catalyze the rate-limiting step in gluconeo-
genesis.1 In addition, it is reported that over-expression
of PEPCK in transgenic animals results in fasting
hyperglycemia and impaired glucose tolerance.2ꢀ4
Therefore, reducing the activity of PEPCK might be
expected to lead to a lowering of fasting blood glucose
levels in diabetic patients. The step in glucose synthesis
catalyzed by PEPCK is shown below:
at N-1 and C-8 that improve the in vitro activity by 100-
fold. The trisubstituted xanthines reported here are
novel xanthines and represent the first inhibitors of
human cytosolic PEPCK to compete with the natural
GTP substrate. We present a SAR for N-1, a developing
SAR for C-8 and a suggestion that xanthine’s N-7
hydrogen might be required for activity.
In a searchfor novel inhibitors of PEPCK a HTS was
initiated and hits confirmed and the effect of the com-
pounds on PEPCK’s enzymatic activity determined
using recombinant human cytosolic PEPCK, expressed
and purified from Escherichia coli as a GST-fusion pro-
tein. GTP and manganese dependent PEPCK enzyme
activity catalyzes the decarboxylation of oxaloacetic
acid yielding GDP and phosphoenolpyruvate (PEP).
This reaction was coupled to pyruvate kinase and lac-
tate dehydrogenase catalyzed reactions and the overall
reaction rate was determined by measuring the differ-
ence in the rate of change in absorbance at 340 nM
during a 20-min incubation period. In brief, 2.5 mg of
recombinant human cytosolic GST-PEPCK was added
to a room temperature reaction mixture which con-
tained 0.3 mM GTP, 0.3 mM OAA, 3 mM MgCl2,
0.075 mM MnCl2, 30 mM potassium phosphates, pH
7.6, 1 mM dithiothreitol (DTT), 0.2 mM ADP, 1 mM
NADH, 0.9 units/mL eachof pyruvate kinase and lac-
tate dehydrogenase and 1 mg/mL BSA. Compounds
were dissolved in DMSO, and then added to the above
mixture such that the final concentration of DMSO was
Oxaloacetic acid ðOAAÞþ
À
À!
GTP
phosphoenolpyruvate ðPEPÞ þ GDP þ CO2
Apart from substrate analogues the only reported
reversible inhibitor of PEPCK is 3-mercaptopicolinic
acid (3-MPA). In the enzyme assay used in our studies
3-MPA was found to have an IC50 in the 20 mM range.
In addition, 3-MPA has been reported to be non-com-
5
petitive withrespect to all substrates.
We now describe modifications of the high throughput
screen (HTS) hit, 1-allyl-3-butyl-8-methylxanthine (5),
*Corresponding author at current address: 1724 Pine Valley Dr. #315,
Fort Myers, FL 33907, USA. E-mail: louf1203@aol.com
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00722-4

3608
L. H. Foley et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3607–3610
10%. DMSO at 10% had no effect on the enzyme in this
assay. Individual compounds were run as duplicates per
assay and the results reported as the average of the two
runs. The results given with the standard deviation were
of compounds assayed more than once. Doses were
chosen such that each incubation included a range of
eight data points for the IC50 determination. Assays of
new compounds always included the most active com-
pound(s) prepared up to that point to give an immediate
determination as to whether a specific modification
improved activity. As the activity of compounds
improved the highest concentration tested was
decreased, which allowed for more data points around
the IC50 value. The result of lowering the maximum
concentration was that IC50 values were not always
obtained for less potent compounds. For these com-
pounds, as seen in Table 1, the % inhibition at the
highest concentration tested is reported.
Preparation of the C-8 unsubstituted analogue 8 and
compound 6 involved refluxing 3a with triethylortho-
formate and 3b with triethylorthoacetate, respectively.10
The initial xanthine HTS lead, compound 5, was a
competitive inhibitor with respect to the substrate GTP
but was a weak inhibitor of PEPCK with an IC50 of
225ꢂ25 mM in the forward reaction. Compounds with
this level of inhibition would not normally have been
pursued, however, xanthine 5 represented a drug-like hit
and a structure witha number of modifiable sites. In
addition compounds with a short half-life were desirable
and xanthines appeared to fit that requirement; for
example in humans 1,3-dimethylxanthine (theophyl-
line), a bronchodilator, has a half-life of ꢃ9 h, bio-
availability of 96%, and the 1,3,7-trimethylxanthine
(caffeine) has a half-life ꢃ5 h, is 100% bioavailable, and
bothcompounds are metabolized in the liver. Also later
work11 suggested that the IC50 values reported here
might be as much as 9-fold higher than the true binding
Preparation of the majority of the xanthines reported
here employed the route shown in Scheme 1 and used
methods described by others working in the uracil and
xanthine fields.6ꢀ8 The 6-amino-1-butyl uracil (1)6 was
converted into the required 1,3-disubstituted-5,6-dia-
mino uracils (3a–f) using the method described by Mul-
ler et al.7 Acylation withteh appropriate carboxylic
acid9 used the procedure of Jacobson et al.8 to afford
the acylamino uracils 4a–i. Cyclization of the acylamino
uracils 4a–i to the 1,3,8-trisubstituted xanthines
employed 10% NaOH in a mixture of methanol-water
heated in a 50 ꢁC oil bath. These conditions minimized
cleavage of the N-acetyl bond in compounds 4b,c,f–i,
which was problematic at higher reaction temperatures.
The lability of the N-trifluoroacetyl group in 4e to the
cyclization conditions was used to advantage in the
preparation of the C-8 4-aminobenzyl analogue 13.
Table 1. IC50 values from the enzyme assay for modifications of
compound 5 at N-1 and C-8a
Compd
R1
R8
IC50, mMb
5
8
9
Allyl
Allyl
Allyl
Allyl
Allyl
Allyl
Allyl
Benzyl
Methyl
H
Ethyl
225ꢂ25
39% @ 500c
100
10
11
12
13
14
15
16
17
N-Acetyl-4-aminobenzyl 18.8ꢂ6.2(4)
N-Acetyl-4-aminophenyl 1.7% @ 500c
Benzyl
4-Aminobenzyl
N-Acetyl-4-aminobenzyl 8.15ꢂ1.65(12)
29% @ 25c
37
2-Fluorobenzyl N-Acetyl-4-aminobenzyl 2.11ꢂ0.60(32)
3-Fluorobenzyl N-Acetyl-4-aminobenzyl
2-Methoxyethyl N-Acetyl-4-aminobenzyl
46% @ 25c
36% @ 25c
a1H NMR and MS data for compounds 8–17 are provided in ref 18.
bIC50 inhibitory values from the enzyme assay. Results showing stan-
dard deviation (sd) values were assayed more than once. If multiple
assays were preformed, the number of repetitions is shown in par-
entheses. A single number without sd indicates a single assay with the
average of the duplicate runs reported.
Scheme 1. Route to 3-butyl-1,8-disubstituted xanthines. Reagents and
conditions: (a) (1) HMDS, (NH4)2SO4, heat, (2) 1 equiv RCH2X,
reflux; (b) (1) NaNO2, HOAc/H2O; (2) Na2S2O4, NH4OH at 85–95 ꢁC;
(c) 1 equiv EDCI, imidazole (cat); DMAP (cat), DMF withteh
required carboxylic acid; (d) NaOH in MeOH/H2O in a 50 ꢁC oil bath.
cPercent inhibition at the mM concentration indicated.

L. H. Foley et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3607–3610
3609
constants (Ki values) due to the fact that the PEPCK
assay used a saturating concentration of GTP.12
see Table 1, which gave a>20-fold improvement over
the activity of the original hit 5. Modeling based on the
X-ray structure of xanthine analogue 10 bound to
PEPCK11 suggested that a H-bond from PEPCK’s
Asn533 side chain amide to an ortho-substituent on the
phenyl ring of the N-1 benzyl unit was possible. The 2-
fluorobenzyl analogue 15 was prepared and its assay
indicated that incorporation of the 2-fluorobenzyl moi-
ety improved the activity nearly 4-fold over that of the
corresponding N-1 benzyl analogue 14.
Assay of several xanthines from the Roche Repository
in follow-up (data not shown) suggested that the p-sys-
tem of the allyl group at N-1 was important and also
that a C-8 benzyl group might improve activity.
Additional support for the need for a p-system at N-1
came from the decrease in activity of the N-1 saturated
analogue 6, see below. The importance of the N-7
hydrogen in xanthine binding was suggested by the loss
of activity shown by the N-7 ethyl analogue 7.13
The choice of the fluoro group was based on its small
size and excellent ability to accept H-bonds (van der
Waals’ radii in A, F 1.47; H 1.20; O 1.5216 and electro-
negativity F 4.0 vs O 3.6517). This subsequently proved
to have been a critical choice, as neither the 2-hydroxy
nor a 2-fluorobenzyl with other groups on the phenyl
ring (S. Pietranico, W. Yun, Roche Research Center,
personal communication) were as active as the 2-
fluorobenzyl analogue 15. These and other results sug-
gested the presence of a small N-1 binding pocket.
The possibility that the improved activity of 15 resulted
from the 2-fluoro-group decreasing the electron density
of the N-1 phenyl ring was explored by preparation of
the identical compound with the fluorine at the 3-posi-
tion. Once again the small size of fluorine allowed the
examination of this possibility. The N-1 3-fluorobenzyl
analogue 16, as shown in Table 1, was >10-fold less
active relative to the 2-fluorobenzyl analogue 15 and
thus provided no support for this hypothesis. The
decreased activity of the N-1 methoxyethyl analogue 17,
which contains only a H-bond acceptor, when com-
pared to either the benzyl or 2-fluorobenzyl analogues
14 and 15, respectively, provided additional support for
the importance of a p-system at N-1 for enhanced
binding.
The requirement for C-8 substitution was confirmed by
preparation and assay of several C-8 modifications of
compound 5 (see Table 1). The C-8 unsubstituted ana-
logue 8 was >2-fold less active than the C-8 methyl
analogue 5. Th eC-8 ethyl analogue 9 improved activity
ꢃ2-fold over the C-8 methyl of the initial hit. Incor-
poration of the N-acetyl-4-aminobenzyl unit at C-8 giv-
ing 1-allyl-3-butyl-8-(N-acetyl-4-aminobenzyl)-xanthine
(10) resulted in a>10-fold improvement in activity.
The importance of at least a one-carbon linker between
the xanthine and the phenyl ring at C-8 was demon-
strated by the loss of activity observed with the N-ace-
tyl-4-aminophenyl analogue 11. Although the C-8
phenyl analogue 11 showed only 1.7% inhibition at
500 mM in the PEPCK assay, other 1,3,8-trisubstituted
xanthines with a phenyl group at C-8 have been repor-
ted to be adenosine receptor (AR) antagonists.14 While
the compounds described here have not been assayed
against ARs, a reported comparison of a 1,3-dimethyl-8
benzylxanthine with its C-8 phenyl-analogue in an
A2BAR assay showed that the C-8 benzyl compound
was >74-fold weaker than the corresponding phenyl
analogue.15 This, and the size of the N-1, N-3 groups,
suggested that the compounds described here should be
weak AR inhibitors at best.
In summary we describe here the first GTP-competitive
inhibitors of human cytosolic PEPCK and modifi-
cations to N-1 and C-8 that improved the activity of the
initial HTS hit by >100-fold. Assay data is presented
that suggests the N-7 hydrogen, a p-system at N-1 and a
benzyl unit at C-8 are important for inhibitory activity.
The optimal modification found for N-1 was the 2-
fluorobenzyl group.
References and Notes
Additional modifications at C-8 explored in this com-
pound class, see Table 1, included the unsubstituted
benzyl analogue 12 and the 4-aminobenzyl analogue 13
both of which retained activity but were slightly less
active when compared to the N-acetyl-4-aminobenzyl
analogue 10. These results suggested that substitution
on the phenyl ring of the C-8 benzyl group might be
important. Further modifications at C-8 will be descri-
bed in future papers.
1. Cimbala, A. N.; Lamers, W. H.; Nelson, J. E.; Monahan,
J. E.; Yoo-Warren, H.; Hanson, R. W. J. Biol. Chem. 1982,
257, 7629.
2. Rosella, G.; Zajac, J. D.; Baker, L.; Kaczmarczyk, S. J.;
Andrikopoulos, S.; Adams, T. E.; Proietto, J. Mol. Endocrinol.
1995, 9, 1396.
3. Valera, A.; Pujol, A.; Pelegrin, M.; Bosch, F. Proc. Natl.
Acad. Sci. U.S.A. 1994, 91, 9151.
4. Sun, Y.; Liu, S.; Ferguson, S.; Wang, L. Q.; Klepcyk, P.;
Yun, J. S.; Freidman, J. E. J. Biol. Chem. 2002, 277, 23301.
5. Makinen, A. L.; Nowak, T. J. Biol. Chem. 1983, 258,
11654.
The need for a p-system at N-1, suggested by compound
6, led to the preparation of the N-1 benzyl analogue 14,
6. Papesch, V.; Schroeder, E. F. J. Org. Chem. 1951, 16, 1879.

3610
L. H. Foley et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3607–3610
7. Muller, C. E.; Shi, D.; Manning, M., Jr.; Daly, J. W. J.
Med. Chem. 1993, 36, 3341.
2H), 5.94 (m, 1H), 11.87 (br s, 1H ex); LRMS calcd for
C14H20N4O2 276; Observed M+H 277.
8. Jacobson, K. A.; Gallo-Rodriguez, C.; Melman, N.;
Fischer, B.; Maillard, M.; van Bergen, A.; van Galen, P. J. M.;
Karton, Y. J. Med. Chem. 1993, 36, 1333.
9. N-Acetyl-4-aminophenylacetic acid was prepared using the
procedure for N-trifluoroacetyl-4-aminophenylacetic acid
Janda, K. D.; Ashley, J. A.; Jones, T. M.; McLoed, D. A.;
Scloeder, D. M.; Weinhouse, M. I.; Lerner, R. A.; Gibbs,
R. A.; Benkovic, P. A.; Hilhorst, R.; Benkovic, S. J. J. Am.
Chem. Soc. 1991, 113, 291.
N-[4-(1-Allyl-3-butyl-2,6-dioxo-2,3,6,7-tetrahydro-1H-purin-
8-ylmethyl)-phenyl]-acetamide (10): 1H NMR (DMSO-d6) d
0.87 (t, 3H), 1.26 (m, 2H), 1.61 (m, 2H), 2.00 (s, 3H), 3.94 (t,
2H), 3.97 (s, 2H), 4.44 (br d, 2H), 5.06 (m, 2H), 5.82 (m, 1H),
7.17 (d, 2H), 7.47 (d, 2H), 9.89 (s, 1H ex), 13.4 (s, 1H ex);
HRMS calcd for C21H25N5O3 395.1957; Observed 395.1947.
1-Allyl-3-butyl-8-(N-acetyl-4-aminophenyl)-3,7-dihydropurine-
1
2,6-dione (11): H NMR (DMSO-d6) d 0.93 (t, 3H), 1.33 (m,
2H), 1.71 (m, 2H), 2.08 (s, 3H), 4.06 (t, 2H), 4.51 (br d, 2H),
5.08 (m, 2H), 5.87 (m, 1H), 7.81 (d, 2H), 8.06 (d, 2H), 10.2 (s,
1H ex), 13.69 (s, 1H ex); LRMS calcd for C20H23N5O3 381;
Observed M+H 382.
10. Procedure similar to that described in ref 7 and references
therein.
11. Foley, L. H.; Wang, P.; Dunten, P.; Ramsey, G.;
Gubler, M. L.; Wertheimer, S. J. Bioorg. Med. Chem. Lett.
See following article. doi: 10.1016/S0960-894X(03)00723-6.
12. For a competitive inhibitor the Ki will be lower than the
IC50 by the factor ½1 þ S=KMꢄ, where S is the concentration of
GTP in the assay, and KM is the Michaelis constant of the
enzyme for GTP. In simple terms, the higher the concentration
of GTP in the assay, the higher the measured IC50 for a com-
petitive inhibitor.
13. The N-7 ethyl analogue 7 was obtained as a side product
when 5 was reprepared from the reaction of the diaminouracil
3a with triethylorthoacetate.
14. Doichinova, I. A.; Natcheva, R. N.; Mihailova, D. N.
Eur. J. Med. Chem. 1994, 29, 133.
1
1-Allyl-8-benzyl-3-butyl-3,7-dihydro-purine-2,6-dione (12): H
NMR (DMSO-d6) d 0.87 (t, 3H), 1.26 (m, 2H), 1.61 (m, 2H),
3.94 (t, 2H), 4.03 (s, 2H), 4.44 (d, 2H), 5.04 (m, 2H), 5.80 (m,
1H), 7.23 (m, 5H), 13.42 (s, 1H ex); HRMS calcd for
C19H22N4O2 M+ 338.1742; Observed 338.1742.
1-Allyl-8-(4-aminobenzyl)-3-butyl-3,7-dihydropurine-2,6-
dione (13): 1H NMR (DMSO-d6–D2O) d 0.87 (t, 3H), 1.26 (m,
2H), 1.61 (m, 2H), 3.82 (s, 2H), 3.92 (t, 2H), 4.44 (d, 2H), 5.00
(m, 2H), 5.80 (m, 1H), 6.44 (d, 2H), 6.90 (d, 2H); HRMS calcd
for C19H23N5O2 354.1931; Observed 354.1928.
N-[4-(1-benzyl-3-butyl-2,6-dioxo-2,3,6,7-tetrahydro-1H-purin-
8-ylmethyl)-phenyl]-acetamide (14): 1H NMR (DMSO-d6) d
0.86 (t, 3H); 1.26 (m, 2H); 1.61 (m, 2H); 2.00 (s, 3H), 3.94 (t,
2H), 3.98 (s, 2H), 5.04 (s, 2H), 7.17–7.30 (m, 7H), 7.47 (d, 2H),
9.89 (s, 1H ex), 13.43 (s, 1H ex); HRMS calcd for C25H27N5O3
446.2192; Observed 446.2199.
15. Bruns, R. F. Biochem. Pharmacol. 1981, 30, 325.
16. Pauling, L. The Nature of the Chemical Bond, 3rd Ed.;
Cornell University Press: Ithaca, NY, 1960.
17. Sanderson, R. T. J. Am. Chem. Soc. 1983, 105, 2259.
18. Spectra general information: 1H NMR spectra were
acquired witha Varian Mercury NMR spectrometer at
300 MHz. The samples were dissolved in either CDCl3 or
DMSO-d6 as noted. The chemical shifts are referenced to
CHCl3 at 7.26 ppm or DMSO-d5 at 2.50 ppm; exchangeable
protons are denoted by ex. Highresolution mass spectra were
obtained on a Micromass AutoSpec instrument. The low
resolution mass spectra were run as APCi on a Micromass
Platform II instrument.
1-Allyl-3-butyl-3,7-dihydropurine-2,6-dione (8): 1H NMR
(DMSO-d6) d 0.90 (t, 3H), 1.30 (m, 2H), 1.65 (m, 2H), 4.00 (t,
2H), 4.48 (br d, 2H), 5.06 (m, 2H), 5.85 (m, 1H), 8.06 (s, 1H
ex), 13.6 (s, 1H ex); HRMS calcd for C12H16N4O2 248.1273;
Observed 248.1274.
1-Allyl-3-butyl-8-ethyl-3,7-dihydropurine-2,6-dione (9): 1H
NMR (CHCl3) d 0.96 (t, 3H), 1.40 (t overlapping m, 5H), 1.77
(m, 2H), 2.88 (q, 2H), 4.13 (t, 2H), 4.68 (br d, 2H), 5.20 (m,
N-{4-[1-(2-Fluoro-benzyl)-3-butyl-2,6-dioxo-2,3,6,7-tetra-
1
hydro-1H-purin-8-ylmethyl]phenyl}-acetamide (15): H NMR
(DMSO-d6) d 0.86 (t, 3H), 1.26 (m, 2H), 1.60 (m, 2H), 2.00 (s,
3H), 3.95 (t, 2H), 3.99 (s, 2H), 5.09 (s, 2H), 6.97–7.27 (m, 6H),
7.48 (d, 2H), 9.82 (s, 1H ex), 13.45 (br s, 1H ex); HRMS calcd
for C25H26N5O3F M+ 464.2098; Observed 464.2086.
N-{4-[3-Butyl-1-(3-fluorobenzyl)-2,6-dioxo-2,3,6,7-tetrahydro-
1H-purin-8-ylmethyl]-phenyl}-acetamide (16): 1H NMR
(DMSO-d6) d 0.94 (t, 3H), 1.34 (m, 2H), 1.70 (m, 2H), 2.08 (s,
3H), 4.03 (t, 2H), 4.06 (s, 2H), 5.13 (s, 2H), 7.10–7.60 (m, 8H),
9.97 (s, 1H ex), 13.52 (s, 1H ex); HRMS calcd for
C25H26N5O3F 463.2020; Observed 463.2018.
1-(2-Methoxyethyl)-3-butyl-8-(N-acetyl-4-aminobenzyl)-3,7-
1
dihydropurine-2,6-dione (17): H NMR (DMSO-d6) d 0.89 (t,
3H), 1.28 (m, 2H), 1.62 (m, 2H), 2.01 (s, 3H), 3.32 (s, 3H), 3.47
(t, 2H), 3.95 (t, 2H), 3.98 (s, 2H), 4.05 (t, 2H), 7.18 (d, 2H),
7.49 (d, 2H), 9.91 (s, 1H ex), 13.40 (s, 1H ex); HRMS calcd for
C21H27N5O4 413.2063; Observed 413.2064.