The role of phosphate in the action of thymidine phosphorylase inhibitors: Implications for the catalytic mechanism

Article abstract of DOI:10.1016/j.bmcl.2010.01.076

The design and synthesis of 5-fluoro-6-[(2-aminoimidazol-1-yl)methyl]uracil (AIFU), a potent inhibitor of thymidine phosphorylase (TP) with K_i-values of 11 nM (ecTP) and 17 nM (hTP), are described. Kinetic studies established that the type of inhibition of TP by AIFU is uncompetitive with respect to inorganic phosphate (or arsenate). The results obtained suggest that AIFU and other zwitterionic thymine analog inhibitors of TP act as transition state analogs, mimicking the anionic thymine leaving group, consistent with an S_N2-type catalytic mechanism, and anchored by their protonated side chains to the enzyme-bound phosphate by electrostatic and H-bonding interactions.

Full text of DOI:10.1016/j.bmcl.2010.01.076

Bioorganic & Medicinal Chemistry Letters 20 (2010) 1648–1651
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
The role of phosphate in the action of thymidine phosphorylase inhibitors:
Implications for the catalytic mechanism
Harsh V. Jain a, Roshni Rasheed a, Thomas I. Kalman a,b,
*
a Department of Chemistry, University at Buffalo, Buffalo, NY 14260, United States
b Department of Pharmacology & Toxicology, University at Buffalo, Buffalo, NY 14260, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
The design and synthesis of 5-fluoro-6-[(2-aminoimidazol-1-yl)methyl]uracil (AIFU), a potent inhibitor of
thymidine phosphorylase (TP) with Ki-values of 11 nM (ecTP) and 17 nM (hTP), are described. Kinetic
studies established that the type of inhibition of TP by AIFU is uncompetitive with respect to inorganic
phosphate (or arsenate). The results obtained suggest that AIFU and other zwitterionic thymine analog
inhibitors of TP act as transition state analogs, mimicking the anionic thymine leaving group, consistent
with an SN2-type catalytic mechanism, and anchored by their protonated side chains to the enzyme-
bound phosphate by electrostatic and H-bonding interactions.
Received 31 December 2009
Accepted 11 January 2010
Available online 21 January 2010
Keywords:
Thymidine phosphorylase
Transition state analog
Enzyme mechanism
Inhibition kinetics
Ó 2010 Elsevier Ltd. All rights reserved.
Thymidine phosphorylase (TP) is a multifunctional protein also
known as platelet-derived endothelial cell growth factor (PD-
ECGF) that stimulates angiogenesis, and as gliostatin, a factor con-
trolling glial cell proliferation.1,2 TP is frequently overexpressed in
many types of cancer and has prognostic value in fluoropyrimidine
therapy, since it is mainly responsible for the metabolic activation
of 5-fluorouracil prodrugs, such as capecitabine, 50-deoxy-5-fluoro-
uridine and tegafur.
fluoro-6-imidazol-1-ylmethyl-uracil (1 IFU, Ki = 51 nM; ecTP)
was designed in our laboratory as a transition state analog inhibitor
of TP (see Fig. 2) as an oxacarbenium ion mimic.7 Our results indi-
cated that IFU was bound to the enzyme in its zwitterionic form,
the fluorine contributing two orders of magnitude increase in affin-
ity, presumably due to a decrease in the pKa of the uracil ring by two
pH units.
In spite of the success of the oxacarbenium ion-like transition
state hypothesis as a basis for the design of many potent TP inhib-
itors, experimental support for an SN1-type mechanism was lack-
ing. Contrary to the prevalence of the SN1 mechanisms in
nucleoside phosphorylases, Birck and Schramm provided strong
experimental evidence for the involvement of a nearly symmetrical
SN2-like transition state in the TP-catalyzed reaction.8 In addition,
they put forward the hypothesis that the positively charged side
chain of TPI acts as a phosphate ‘chelator’ and that the ion-pair
complex may show resemblance to the SN2 transition state.8,9 In
contrast, the fact that the structure of the human TP–TPI binary
complex was obtained in the absence of phosphate,10 was taken
as an indication in a recent review that ‘phosphate is not necessary
for enzyme inhibition’ by TPI.1
Figure 1 outlines the TP-catalyzed reaction of the reversible
phosphorolysis of thymidine to thymine and 2-deoxy-
phosphate (dR-1-P). Subsequent dephosphorylation of the product
dR-1-P to 2-deoxy- -ribose in vivo triggers the pro-angiogenic re-
D-ribose-1-
D
sponse and promotes cancer metastasis. Thus, TP inhibitors may
have utility in the treatment of cancer. Indeed, a potent inhibitor
of TP, 5-chloro-6-[(2-iminopyrrolidin-1-yl)methyl]uracil (TPI)
(see Fig. 2) entered clinical trials in combination with 5-trifluoro-
methyl-20-deoxyuridine as an orally available anticancer treatment
under the name of TAS-102.3,4
Many laboratories including our own have been involved in the
design and synthesis of a variety of TP inhibitors (see Fig. 2). The
most potent inhibitors were 5-halogenated uracil derivatives with
a basic side chain at the 6-position, such as TPI. The potent inhib-
itory activity of these zwitterionic analogs bearing a positively
charged side chain appeared to confirm a hypothetical SN1-type
mechanism of catalysis involving an oxacarbenium ion transition
state (see Fig. 3), in analogy to other glycosyl bond-cleaving
enzymatic reactions.5,6 Based on the same working hypothesis, 5-
The observed high potency of the zwitterionic inhibitors is con-
sistent with either types of nucleophilic substitution mechanisms.
In both cases, the negatively charged 5-halogenated uracil rings
may resemble the anionic thymine leaving group, an integral part
of both types of transition states. However, the protonated side
chains of these inhibitors may play two fundamentally different
* Corresponding author. Tel.: +1 716 645 4184; fax: +1 716 645 6963.
E-mail address: tkalman@buffalo.edu (T.I. Kalman).
ecTP, Escherichia coli TP; hTP, human TP.
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.01.076

H. V. Jain et al. / Bioorg. Med. Chem. Lett. 20 (2010) 1648–1651
1649
Figure 1. Outline of the reaction catalyzed by TP.
Figure 2. Examples of TP inhibitors (Ki -values indicated).
roles: they can (1) either mimic the oxacarbenium ion of an SN1
transition state, or (2) interact with the phosphate nucleophile in
an SN2-type mechanism (see Fig. 3). Since AIFU may be classified
as a mechanism-based inhibitor, we sought to distinguish between
these two alternatives.
Figure 3. Hypothetical SN1 (A) and SN2 (B) transition states for the TP-catalyzed
reaction are depicted together with their zwitterionic analogs.
was used without a protecting group, the yield of AIFU was less
than 10%. Protection of the amino group of 2-aminoimidazole with
N,N-dimethylformamide dimethylacetal yielded the known form-
amidine 5,14 which was alkylated by 4 in DMF to give the desired
In order to shed further light on the catalytic mechanism, we fo-
cused our attention on the postulated interaction of the basic side
chain with phosphate. We decided to modify IFU to maximize the
interaction with Pi by replacing the imidazole (pKa = 7.18)à with the
more basic 2-aminoimidazole (pKa = 8.56)à yielding 5-fluoro-6-[(2-
aminoimidazol-1-yl)methyl]uracil (2 AIFU) (see Fig. 2). This modifi-
cation increased the potency ca. fivefold giving Ki-values of 11 nM
and 17 nM for ecTP and hTP, respectively, presumably due to the
presence of the NH2-group with additional H-bonding potential.
The potency of AIFU is similar to that of the corresponding 5-chloro
and 5-bromo analogs previously reported by Reigan et al.11 While it
is difficult to reconcile the dominance of positively charged residues
at the active site of TP with an oxacarbenium-like transition state, as
we previously pointed out,7 it is conceivable that they play a role in
the stabilization of the negative charge developing on the thymine
leaving group in the transition state, since thymine has a relatively
high pKa of 9.86 at NH-1.12 This is in line with the observed increase
in affinity of the more acidic 5-halogenated uracil-containing TP
inhibitors with an unsubstituted N-1 position (pKa = 7.7–7.9 for 5-
F-, 5-Cl-, and 5-Br-uracil). This effect is more pronounced for the
zwitterionic analogs bearing a basic side chain, which further lowers
the pKa-values by 1.9–2.0 pH units, for example, the pKa of both TPI
and AIFU is 6.05.à This also explains the weaker binding of thymidine
analogs, since substitution at N-1 prevents the all-important ioniza-
tion of the pyrimidine ring at that position.
6
15 in 65% yield. This was followed by deprotection in concentrated
hydrochloric acid at 100 °C to afford target compound 216 in 90%
yield. This is a considerable improvement of the 9–13% yields re-
ported by Reigan et al. for the direct coupling of 2-aminoimidazole
with 5-bromo- or 5-chloro-6-[(2-chloro-1-yl)methyl]uracil.11
A variety of phosphonate derivatives were reported as multi-
substrate inhibitors of TP17,18 and were found to inhibit the en-
zyme competitively with respect to phosphate. The rigid bicyclic
phosphonate 3 (see Fig. 2) was shown to be 30-fold more active
than its exo-isomer,17 clearly showing the importance of the orien-
tation of the phosphonate, which may occupy the phosphate bind-
ing site. To identify the likely location of the phosphate at the
active site, 3 was docked into the hTP structure in the closed con-
formation, using the coordinates of the previously reported crystal
structure of the human TP-TPI binary complex,10 keeping the
respective pyrimidine rings in the same orientation (see Fig. 4A).
Similarly, AIFU was docked in place of TPI to reveal the orientation
of the 2-aminoimidazole side chain. Figure 4B illustrates that in the
energy minimized conformation, the protonated side chain points
toward a cavity occupied by several water molecules, which pre-
sumably fill up the empty phosphate binding site. At the same
time, it is positioned in close proximity to the putative location
of the phosphate identified by the phosphonate group of 3, condu-
cive to charge-charge and H-bonding interactions.
The synthesis of AIFU is outlined in Scheme 1. The preparation
of 4 has been previously described.7,13 When 2-aminoimidazole
Experiments were performed using ecTP to determine the
kinetics of inhibition of the enzyme by AIFU. Enzyme activity
was monitored spectrophotometrically at 350 nm, as described,19
à
pKa values were calculated by ACD/Labs pKa DB program.

1650
H. V. Jain et al. / Bioorg. Med. Chem. Lett. 20 (2010) 1648–1651
Scheme 1. Outline of the synthesis of AIFU. Reagents and conditions: (a) DMF, DIEA (3 equiv), 60 °C, 12 h, argon; (b) concd HCl/EtOAc (1:1), 100 °C, 2 h.
(Fig. 5A) or phosphate (Fig. 5B). The nearly identical slopes of the
Dixon plots of Figure 5 clearly indicate uncompetitive type of inhi-
bition by AIFU with respect to arsenate or phosphate, as described
by the following equation:
ꢀ
ꢁ
1
vi
1
maxKi
1
Vmax
Km
½Sꢀ
¼
½Iꢀ þ
1 þ
V
Accordingly, arsenate or phosphate must bind first to the en-
zyme before AIFU can bind, to form an inactive ternary complex.
The validity of the interpretation of data obtained by using arse-
nate as a second substrate has been questioned before.20 The sim-
ilar patterns of inhibition kinetics obtained in this work justify the
use of arsenate for the study of TP inhibitors. Arsenate is generally
used only to avoid interference by the product dR-1-P formed in
the presence of orthophosphate, and to obtain more reliable kinet-
ics, since the analogous arsenate product is unstable and rapidly
degraded by hydrolysis.21
Figure 4. Ligand binding at the active site of hTP (pdb code 1UOU). Van der Waals
surfaces of docked molecules of 3 (A), and 2 (B) are shown; water molecules are
depicted as cyan-colored balls. Docking was performed using Sybyl 7.3; illustration
by RasTop 2.1.
The results of our studies described in this communication
showing the obligatory order of binding of phosphate before AIFU
suggest that the positively charged side chain of all structurally re-
lated zwitterionic inhibitors of TP may interact with the enzyme-
bound phosphate, rather than act as a mimic of an oxacarbenium
ion-like transition state and strongly support Birck and Schramm’s
hypothesis.8 Nevertheless, the importance of the involvement of an
oxacarbenium ion-like transition state as a working hypothesis in
guiding the rational design of potent TP inhibitors in a number of
laboratories, including our own, must be recognized.
With respect to future inhibitor design, it is important to con-
sider that the uncompetitive relationship of the zwitterionic TP
inhibitors with phosphate, that is, the requirement for prior phos-
phate binding, is an advantage in their predicted in vivo activity,
since the intracellular concentrations of orthophosphate are usu-
ally maintained around 1 mM,22 exceeding its Km-value for TP. In
contrast, phosphonate derivatives, like TP-64 (and most likely 3)
inhibit TP competitively with respect to orthophosphate,18 which
is to their disadvantage. The bound phosphate may permit inhibi-
tors with protonated side chains to have high affinity to the active
site occupied by several repulsive positively charged amino acid
side chains (Arg202, Lys221, His116, and Lys115). These residues
may serve to stabilize the negatively charged pyrimidine ring
resembling the transition state structure, in addition to attracting
phosphate (or arsenate) to the active site.
In summary, the results of our enzyme kinetic studies are con-
sistent with the hypothesis that the protonated side chains of zwit-
terionic inhibitors of TP, such as AIFU, interact with the enzyme-
bound phosphate (or arsenate). Validation of this hypothesis, how-
ever, will require the determination of the 3D-structure of an en-
zyme-phosphate-inhibitor ternary complex.
Figure 5. 1/v versus [I] plot of the inhibition of ecTP by AIFU. The concentrations of
arsenate (plot A) and phosphate (plot B) were: (4)1 mM; (h) 2 mM; (}) 5 mM; (s)
10 mM; 5-NO2–dUrd (0.13 mM) was used as substrate.
Acknowledgments
using a fixed concentration of 5-NO2–dUrd (0.13 mM), as chromo-
genic substrate. Rates were measured at varying concentrations of
AIFU in the presence of 1, 2, 5, and 10 mM of either arsenate
Helpful comments by Drs. John Richard and Andrew Murkin are
gratefully acknowledged.

H. V. Jain et al. / Bioorg. Med. Chem. Lett. 20 (2010) 1648–1651
1651
13. Duschinsky, R.; Pleven, E.; Heidelberger, C. J. Am. Chem. Soc. 1957, 79, 4559.
References and notes
14. Lahue, B. R.; Wan, Z.-K.; Snyder, J. K. J. Org. Chem. 2003, 68, 4345.
15. For compound 6: 1H NMR (500 MHz, D2O) d 8.02 (s, 1H), 6.96 (d, J = 2.5 Hz, 1H),
6.86 (d, J = 2.5 Hz, 1H), 4.98 (d, J = 2.1 Hz, 2H), 3.19 (d, J = 6.1 Hz, 1H), 3.06 (s,
3H), 2.94 (s, 3H). 13C NMR (75 MHz, D2O) d 159.52 (d, JC4F = 25.6 Hz), 158.21
(s), 150.60 (s), 149.40 (s), 138.69 (d, JC5F = 233.0 Hz), 133.97 (d, JC6F = 23.6 Hz),
117.83 (s), 114.26 (s), 41.11 (s), 40.41 (s), 34.69 (s). 19F NMR (376 MHz, DMSO-
d6) d ꢁ170.41 (s). MS (M+H)+ = 281.2.
1. Perez-Perez, M. J.; Priego, E. M.; Hernandez, A. I.; Camarasa, M. J.; Balzarini, J.;
Liekens, S. Mini-Rev. Med. Chem. 2005, 5, 1113.
2. Liekens, S.; Bronckaers, A.; Pérez-Pérez, M.-J.; Balzarini, J. Biochem. Pharmacol.
2007, 74, 1555.
3. Yano, S.; Kazuno, H.; Sato, T.; Suzuki, N.; Emura, T.; Wierzba, K.; Yamashita, J.;
Tada, Y.; Yamada, Y.; Fukushima, M.; Asao, T. Bioorg. Med. Chem. Lett. 2004, 12,
3443.
4. Emura, T.; Suzuki, N.; Fujioka, A.; Ohshimo, H.; Fukushima, M. Int. J. Oncol.
2005, 27, 449.
16. For compound 2: 1H NMR (500 MHz, D2O) d 6.76 (d, J = 15.1 Hz, 2H), 4.95 (s,
2H). 13C NMR (126 MHz, D2O) d 159.41 (d, J = 25.7 Hz), 150.65 (s), 146.70 (s),
139.03 (d, JC5F = 233.5 Hz), 132.33 (d, J = 23.6 Hz), 115.70 (d, J = 5.1 Hz), 113.49
(d, J = 2.7 Hz).19F NMR (376 MHz, D2O)
d -168.79 (t, J = 2.2 Hz). MS
5. Ganem, B. Acc. Chem. Res. 1996, 29, 340.
(M+H)+ = 226.2.
6. Schramm, V. L. Nucleic Acids Symp. Ser. 2003, 3, 107.
7. Kalman, T. I.; Lai, L. Nucleosides Nucleotides Nucleic Acids 2005, 24, 367.
8. Birck, M. R.; Schramm, V. L. J. Am. Chem. Soc. 2004, 126, 2447.
9. Birck, M. R.; Clinch, K.; Gainsford, G. J.; Schramm, V. L.; Tyler, P. C. Helv. Chim.
Acta 2009, 92, 823.
10. Norman, R. A.; Barry, S. T.; Bate, M.; Breed, J.; Colls, J. G.; Ernill, R. J.; Luke, R. W.;
Minshull, C. A.; McAlister, M. S.; McCall, E. J.; McMiken, H. H.; Paterson, D. S.;
Timms, D.; Tucker, J. A.; Pauptit, R. A. Structure 2004, 12, 75.
11. Reigan, P.; Edwards, P. N.; Gbaj, A.; Cole, C.; Barry, S. T.; Page, K. M.; Ashton, S.
E.; Luke, R. W.; Douglas, K. T.; Stratford, I. J.; Jaffar, M.; Bryce, R. A.; Freeman, S.
J. Med. Chem. 2005, 48, 392.
17. Allan, A. L.; Gladstone, P. L.; Price, M. L.; Hopkins, S. A.; Juarez, J. C.; Donate, F.;
Ternansky, R. J.; Shaw, D. E.; Ganem, B.; Li, Y.; Wang, W.; Ealick, S. J. Med. Chem.
2006, 49, 7807.
18. Balzarini, J.; Degrève, B.; Esteban-Gamboa, A.; Esnouf, R.; De Clercq, E.;
Engelborghs, Y.; Camarasa, M.-J.; Pérez-Pérez, M.-J. FEBS Lett. 2000, 483, 181.
19. Wataya, Y.; Santi, D. W. Anal. Biochem. 1981, 112, 96.
20. Edwards, P. N. J. Enzyme Inhib. Med. Chem. 2006, 21, 483.
21. Parks, R. E., Jr.; Agarwal, R. P., 3rd ed.. In The Enzymes; Boyer, P. D., Ed.;
Academic Press: New York, 1972; Vol. 7, pp 483–514.
22. Katz, L. A.; Swain, J. A.; Portman, M. A.; Balaban, R. S. Am. J. Physiol. Heart Circ.
Physiol. 1988, 255, H189.
12. Ganguly, S.; Kundu, K. Can. J. Chem. 1994, 72, 1120.