X-ray structures of two xanthine inhibitors bound to PEPCK and N-3 modifications of substituted 1,8-Dibenzylxanthines

Article abstract of DOI:10.1016/S0960-894X(03)00723-6

The analysis of the X-ray structures of two xanthine inhibitors bound to PEPCK and a comparison to the X-ray structure of GTP bound to PEPCK are reported. The SAR at N-1, N-7 and developing SAR at C-8 are consistent with information gained from the X-ray

Full text of DOI:10.1016/S0960-894X(03)00723-6

Bioorganic & Medicinal Chemistry Letters 13 (2003) 3871–3874
X-ray Structures of Two Xanthine Inhibitors Bound to PEPCK
and N-3 Modifications of Substituted 1,8-Dibenzylxanthines
Louise H. Foley,a,* Ping Wang,a Pete Dunten,a,y 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 analysis of the X-ray structures of two xanthine inhibitors bound to PEPCK and a comparison to the X-ray struc-
ture of GTP bound to PEPCK are reported. The SAR at N-1, N-7 and developing SAR at C-8 are consistent withinformation
gained from the X-ray structures of compounds 1 and 2 bound to PEPCK. Representative N-3 modifications of compound 2 that
led to the discovery of 3-cyclopropylmethyl and its carboxy analogue as optimal N-3 groups are presented.
# 2003 Elsevier Ltd. All rights reserved.
Normally gluconeogenesis maintains blood glucose
levels during periods of fasting, however, in the diabetic
patient it contributes to hyperglycemia. Since the cyto-
solic phosphoenolpyruvate carboxykinase (PEPCK)
enzyme is known to catalyze the rate-limiting step in
gluconeogenesis1 its control could provide a new
therapy for type 2 diabetes.
GTP structure. This work led to the discovery of a
hydrophobic N-3 modification of xanthine 2 that
improved the in vitro activity by ꢀ3-fold, providing
a>250-fold improvement over the initial HTS hit and a
sub-micromolar inhibitor of PEPCK.
The X-ray structure of compound 1-PEPCK is shown in
Figure 1.3 Figure 2a and b provide a comparison of the
guanosine portion of the previously determined X-ray
structure of a non-hydrolyzable GTP analogue bound
to PEPCK4 withthe structure of compound 1-PEPCK.
The X-ray structure of compound 1-PEPCK (Fig. 1 and
depicted in Fig. 2b) shows that GTP’s guanosine unit
and the xanthine inhibitor bind in the same cleft on the
surface of PEPCK. The X-ray of compound 1-PEPCK
is in agreement withthe SAR for positions N-1, C-8 and
N-72 and supports a hypothesis that the xanthine inhib-
itors are binding in a different orientation from GTP’s
guanine ring. The X-ray shows the xanthine ring of 1 to
be flipped and rotated relative to the guanine moiety of
GTP. Rotations of other purine antagonists relative to
the natural ATP ligand in adenosine receptors5 and
CDK2 have been reported.6
We previously reported2 the first competitive inhibitors
of PEPCK withrespect to GTP. Compounds 1 and 2,
withIC 50s from an enzyme assay, were the most active
compounds prepared in each N-1 class. The protocol for
the enzyme assay is given in the preceding paper.2
We now present a comparison of the structure of com-
pound 1 bound to human cytosolic PEPCK3 (com-
pound 1-PEPCK) withthe reported structure of a GTP
analogue bound to PEPCK.4 We also report N-3 modi-
fications aimed at picking up H-bonds found in the
Figure 2a and b show the various interactions of GTP
and compound 1 withteh purine-binding site of
PEPCK. Comparison of Figure 2a and b shows that the
xanthine nitrogens at N-1 and N-3 overlap withGTP’s
N-7 and N-9, respectively. Interestingly, the C-6 car-
bonyl interaction withPEPCK is identical in bothGTP
*Corresponding author at present address: 1724 Pine Valley Dr. #315,
Fort Myers, FL 33907, USA. E-mail: louf1203@aol.com
yCorresponding author concerning the X-ray data: Roche Palo Alto,
3431 Hillview Ave., Palo Alto, CA 94304, USA
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00723-6

3872
L. H. Foley et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3871–3874
and compound 1. A water molecule is found to bridge
xanthine’s N-7 hydrogen and PEPCK’s Trp516 and
Trp527, in agreement withthe decreased activity of all
compounds witha substituent at N-7 (ref 2 and data
not shown). Xanthine’s C-2 carbonyl picks up a H-bond
to PEPCK’s Asn292 side chain, an interaction that has
no counterpart in the binding of the GTP ligand.
The X-ray structure of compound 27 is important in that
it shows that compound 2 binds to the protein in the
same orientation as compound 1 with the xanthine core
of each making the same interactions with the protein (a
stereoview is shown in Fig. 3).
As reported,2 replacement of the N-1 allyl group of
compound 1 withbenzyl and 2-fluorobenzyl groups
improved activity by 2- and >8-fold, respectively. The
X-ray of 2 suggests that part of the improvement was
due to phenyl ring stacking on top of the planar peptide
bond between Asn292 and Leu293. In addition to this
favorable p–p stacking, the fluorine in compound 2 is
optimally placed to accept a H-bond from the Asn533
side-chain amide (distance 3.1 A).
The novel p interaction of PEPCK’s Phe525 with the C-
2 amino group of GTP4 is replaced in compound 1 by
an offset p–p interaction with the phenyl ring of the C-8
benzyl unit. This phenyl ring is also in a favorable edge-
to-face interaction with Phe530. PEPCK’s Phe530 and
Phe517 serve the same function with GTP and the xan-
thine ligand, in effect sandwiching the aromatic purine
cores between the two side-chain phenyl rings.
While the X-ray structures of compounds 1 and 2
bound to PEPCK explained the reason for the improved
activity of compounds witha C-8 benzyl group, they
failed to provide an explanation for the improved
activity of xanthines containing a substituent on the
phenyl ring of the C-8 benzyl unit. Future papers will
cover modifications at C-8 that explored this issue.
The allyl group proved to be the most potent of the
early N-1 groups.2 The X-ray suggests the p-system of
the allyl group in 1 may act as a H-bond acceptor, with
the Asn533 side-chain amide as donor, although at 3.5
A the distance is long.
The observation that groups at N-3 extended into the
area of the protein that bound the ribose ring of GTP
(see Fig. 2a and b) resulted in many unsuccessful
excursions aimed at mimicking these interactions via
xanthine modifications at N-3.
Figure 1. Xanthine 1 binding site in PEPCK. Compound 1 in light
blue and the amino acids in the purine-binding site of PEPCK are
shown in green. In both the xanthine and PEPCK, the atoms of oxygen
and nitrogen are shown in red and dark blue, respectively. H-bonds
are shown with dashed lines in orange and a single water molecule is
shown in pink.
Figure 3. Stereoview of xanthine 2 bound to PEPCK. Legend same as
for Figure 1.
Figure 2. (a) Depiction of GTP’s guanosine unit bound to PEPCK. (b) Depiction of xanthine 1 bound to PEPCK. Guanosine portion of GTP in red
and PEPCK light green, H-bonds indicated in orange, water in blue, xanthine 1 in black. Notations (sc) and (bb) indicate PEPCK side chain
groups and backbone groups, respectively.

L. H. Foley et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3871–3874
3873
Table 1. IC50 values from the enzyme assay for representative N-3
modifications of xanthine 2a
Compd
R0
IC50, mMb
2
8
9
Butyl
Tetrahydropyran-2-ylmethyl
Tetrahydrofuran-2-ylmethyl
Furan-3-ylmethyl
Hexyl
2.11Æ0.60 (32)c
4.5
5.7
10
11
12
13
14
15
16
17
18
19
20
21
4.5
8.9
Propyl
iso-Butyl
1.49Æ0.36
2.02Æ0.30 (6)
3.05Æ0.35
1.18Æ0.24 (6)
0.69Æ0.29 (6)
5.45Æ0.07
2.85Æ0.07
2.0
Cyclopentylmethyl
Cyclobutylmethyl
Cyclopropylmethyl
3-Hydroxylpropyl
4-Hydroxylbutyl
–(CH2)2CO2H
–(CH2)3CO2H
–(CH2)3C(O)NH2
1.95Æ0.07
6.6
a1H NMR and HRMS data for compound 16 is found in ref 15.
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-
enthesis. Assays were conducted with each compound run in duplicate.
A single IC50, without SD, indicates a single assay with the average of
the duplicate run indicated.
Scheme 1. Route to 3-alkyl-1-(2-fluorobenzyl)-8-(N-acetyl-4-amino-
benzyl)xanthines. Reagents and conditions: (a) HMDS,
(NH4)2SO4+2-fluorobenzyl bromide, reflux; (b) (1) NaNO2 in HOAc/
H2O, 85–100 ꢁC; (2) Na2S2O4, NH4OH or 53 psi H2, PtO2 EtOH;14 (c)
EDCI, N-acetyl-4-amino-phenylacetic acid,10 imidazole (cat), DMAP
(cat), DMF; (d) R0-halide in DMF, K2CO3, 40 ꢁC; (e) 10% NaOH in
H2O/MeOH heated in a 50 ꢁC oil bath.
cThe IC50 for compound 2 run as a standard and for a direct com-
parison in these assays was 2.0–2.4 nM.
An efficient route for the preparation of N-3 modifi-
cations of xanthine 2 involved the use of the common
intermediate 6, as shown in Scheme 1. The 3-(2-fluoro-
benzyl)-6-amino uracil (4) was prepared from 6-ami-
nouracil (3) and 2-fluorobenzyl bromide using the
method of Muller.8 The alkylated uracil 4 was con-
verted into the diaminouracil 5 as shown in Scheme 1.9
Acylation of the C-5 amino group with N-acetyl-4-amino-
phenylacetic acid10 used the conditions of Jacobson et
al.11 to afford the acyl uracil 6. This acylation of the C-5
amino group of uracil 6 allowed selective N-1 alkylation
(uracil numbering), as noted by Muller et al.12 Reaction
of 6 with a variety of alkyl halides provided the dialky-
lated uracil derivative 7. The cyclization of 7 used aqu-
eous 10% sodium hydroxide solution in methanol with
heating in a 50 ꢁC oil bath, conditions allowing minimal
loss of the C-8 N-acetyl group, to afford the xanthines
8–26.
pylmethyl 16 was ꢀ3-fold more active than 2 providing
the first sub-micromolar xanthine inhibitor.
Returning to compounds withH-bonding capabilities at
N-3, the 3-hydroxylpropyl analogue 17 and 4-hydroxyl-
butyl analogue 18 and the acids 19 and 20 were not as
active as the N-3 cyclopropylmethyl analogue 16. Amides
derived from the N-3 acids 19 and 20, for example com-
pound 21, also provided no improvement in activity.
A Ki of 0.2 mM for the butyric acid analogue 20 (IC50
1.95Æ0.07 mM) binding to PEPCK was measured via
isothermal titration calorimetry (ITC).4 This, as men-
tioned in the previous paper,2 suggested the Kis for the
xanthine inhibitors described here could be as much as
9-fold lower than the reported IC50s.
Table 2 shows several attempts at using the N-3 cyclo-
propyl ring of xanthine 16 as a scaffold for groups that
might provide H-bond acceptors, however, only the car-
boxyl analogue 22 retained the activity of the unsub-
stituted analogue 16. Increasing the hydrophobicity of
the cyclopropylmethyl group by the addition of a methyl
group, compound 26, also provided no advantage.
A representative sample of N-3 modifications prepared
is shown in Table 1. In comparison withthe N-3 butyl
analogue 2, cyclic ethers 8 and 9, furan-3-ylmethyl 10
and longer alkyls, for example hexyl 11, failed to
improve activity. The shorter alkyls, the propyl analo-
gue 12 and iso-butyl analogue 13 also provided no
improvement over the butyl analogue 2. Th eN-3 methyl
derivative is not shown in Table 1 since it was prepared
in the N-1 benzyl class, where it was 6-fold less potent
when compared to the corresponding N-3 butyl analo-
gue.13 Relative to the butyl analogue 2 the N-3 cyclo-
pentylmethyl analogue 14 and cyclobutylmethyl 15
provided no advantage. However the N-3 cyclopro-
In summary, the first X-ray structures of two xanthine
inhibitors bound to PEPCK and a comparison to GTP-
PEPCK are presented. We also present a number of N-3
modifications of xanthine 2. Two of these modifications,
the cyclopropylmethyl and its carboxy analogue,
provided GTP competitive inhibitors of PEPCK with
sub-micromolar IC50s.

3874
L. H. Foley et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3871–3874
Table 2. IC50 values from the enzyme assay for modifications of
xanthine 16 on the cyclopropyl ringa
N. P.; Lentz, N. L.; Meng, E. C.; Dudley, M. W.; Ogden,
A. M. L.; Demeter, D. A.; Weintraub, H. J. R.; Bey, P. J.
Med. Chem. 1990, 33, 3127. (c) Grahner, B.; Winiwarter, S.;
Lanzner, W.; Muller, C. E. J. Med. Chem. 1994, 37, 1526.
6. Gray, N. S.; Wodicka, L.; Thunnissen, A.-M. W. H.; Nor-
man, T. C.; Kwon, S.; Espinoza, F. H.; Morgan, D. O.;
Barnes, G.; LeClerc, S.; Meijer, L.; Kim, S.-H.; Lockhart,
D. J.; Schultz, P. G. Science 1998, 281, 533.
7. PDB code: 1nhx, Resolution 2.1 A, R factor 17.5%.
8. Muller, C. E. Tetrahedron Lett. 1991, 32, 6539.
9. Muller, C. E.; Shi, D.; Manning, M., Jr.; Daly, J. W. J.
Med. Chem. 1993, 36, 3341.
10. Janda, K. D.; Ashley, J. A.; Jones, T. M.; McLoed, D. A.;
Schloeder, 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.
11. 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.
Compd
R4
R5
IC50, mMb
16
22
23
24
25
26
H
Carboxyl
Methyl carboxylate
Hydroxymethyl
Hydroxymethyl
Methyl
H
H
H
0.69Æ0.29 (6)
0.56Æ0.19 (8)
3.00
H
Hydroxymethyl
H
2.27Æ0.21 (4)
2.00Æ0.42
1.82Æ0.31 (6)
a1H NMR and HRMS for compounds 16 and 22 in ref 15.
bLegend same as Table 1. Th e IC50 for compound 2 run as a standard
in these assays was 2.2–2.3 mM.
12. Muller, C. E.; Sandoval-Ramirez, J. Synthesis 1995, 1295.
13. 1-Benzyl-3-methyl-8-(N-acetyl-4-aminobenzyl) xanthine
(IC50 50 mM) in the same assay 1-benzyl-3-butyl-8-(N-acetyl-4-
aminobenzyl)xanthine (IC50 8.2 mM).
14. Wells, J. N.; Garst, J. E.; Kramer, G. L. J. Med. Chem.
1981, 24, 954.
Acknowledgements
15. Spectra. General information is given in ref 18 of the pre-
vious paper.2 All compounds described here had 1H NMR and
HRMS in agreement withthe structure. 1H NMR and HRMS
are provided below for three of the most active compounds.
N-{4-[3-Cyclopropylmethyl-1-(2-fluorobenzyl)-2,6 dioxo-2,3,6,7-
tetrahydro-1H-purin-8-ylmethyl]-phenyl} acetamide (16): 1H
NMR (DMSO-d6) d 0.42 (m, 4H), 1.24 (m, 1H), 2.02 (s, 3H), 3.85
(d, 2H), 4.00 (s, 2H), 5.12 (s, 2H), 7.04–7.27 (m, 6H), 7.50 (d, 2H),
9.89 (s, 1H ex), 13.47 (s, 1H ex); HRMS calcd for C25H24N5O3F
M+ 461.1863; obsd 461.1863.
The authors thank Ursula Kammlott for crystallization
of PEPCK with the inhibitors and Robert Crowther for
carrying out the ITC measurement of compound 20.
References and Notes
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. Foley, L. H.; Wang, P.; Dunten, P.; Ramsey, G.; Gubler,
M.-L.; Wertheimer, S. J. Bioorg. Med. Chem. Lett. 2003, 13, 3607.
3. PDB code: 1m51, Resolution 2.25 A, R factor 18.1%.
4. Dunten, P.; Belunis, C.; Crowther, R.; Hollfelder, K.;
Kammlott, U.; Levin, W.; Michel, H.; Ramsey, G. B.; Swain,
A.; Weber, D.; Wertheimer, S. J. J. Mol. Biol. 2002, 316, 257.
5. (a) van Galen, P. J. M.; Van Vlijmen, H. W.Th.; Ijzerman,
A. P.; Soudijn, W. J. Med. Chem. 1990, 33, 1708. (b) Peet,
2-[8-(4-Acetylamino-benzyl)-1-(2-fluoro-benzyl)-2,6-dioxo-
1,2,6,7-tetrahydropurin-3-ylmethyl]-cyclopropanecarboxylic
acid (22): H NMR (DMSO-d6) d 0.99 (m, 1H), 1.08 (m, 1H),
1.58 (m, 1H), 1.69 (m, 1H), 2.02 (s, 3H), 4.01 (s, 2H), 4.11 (m,
1H), 4.26 (m, 1H), 5.12 (s, 2H), 7.05–7.40 (m, 4H), 7.24 (d,
2H), 7.50 (d, 2H), 9.90 (s, 1H ex), 12.18 (br s, 1H ex), 13.48 (s,
1H ex); HRMS calcd for C26H24N5O5F M+ 506.1840; obsd
506.1826.
1