A multisubstrate adduct inhibitor of AICAR transformylase

Full text of DOI:10.1021/jm990323+

© Copyright 1999 by the American Chemical Society
Volume 42, Number 18
September 9, 1999
Communications to the Editor
Sch em e 1. Reaction Catalyzed by AICAR Tfase with
Natural Substrates AICAR (1) and 10-f-H4F (2); Also
Shown Is the MAI for AICAR Tfase, â-DADF
A Mu ltisu bstr a te Ad d u ct In h ibitor of
AICAR Tr a n sfor m yla se
Mark Wall, J ae Hoon Shim, and
Stephen J . Benkovic*
Department of Chemistry, 414 Wartik Laboratory,
The Pennsylvania State University,
University Park, Pennsylvania 16802
Received J une 23, 1999
AICAR Tfase (5-aminoimidazole-4-carboxamide-ribo-
nucleotide transformylase) catalyzes the transfer of the
10-formyl group of (6R,RS)-10-formyltetrahydrofolate
(10-f-H4F) to the 5-amino group of 5-aminoimidazole-
4-carboxamide ribonucleotide (AICAR) in the ninth step
of de novo purine biosynthesis, Scheme 1. Virtually all
organisms, with the exception of parasitic protozoa that
scavenge purines from their environment, use this
pathway to synthesize purines, which are essential for
cell viability.1 Rapidly dividing cancer cells, in particu-
lar, require a large amount of purines to sustain such
growth, and consequently the de novo purine biosyn-
thetic pathway is an attractive target for antineoplastic
agents.2 Moreover, AICAR Tfase requires a reduced
folate cofactor, and some of the more successful chemo-
theraputic agents developed to date, such as methotr-
exate, are folate antimetabolites. To date there are no
clinically useful inhibitors of AICAR Tfase nor is there
an available crystal structure to assist in rational
inhibitor design.
The principles of multisubstrate adduct inhibition
offer several advantages over conventional inhibitor
design.3 The combination of two substrates into a single
molecule increases the specificity of the inhibitor for the
target enzyme since it is unlikely that other enzymes
that use one of the components will recognize the
multisubstrate adduct inhibitor (MAI). One can also
expect an increase in the binding affinity due to an
entropic advantage since the MAI possesses components
of both individual substrates in a single molecule.
Evidence from earlier kinetic studies in this laboratory
indicate that the AICAR Tfase kinetic scheme follows
a sequential mechanism with direct transfer of the
formyl group, implying the formation of a ternary
complex.4,5 We have successfully used the MAI approach
in preparing a picomolar inhibitor of glycinamide ribo-
nucleotide transformylase6,7 (GAR Tfase), an enzyme
that catalyzes a similar formyl transfer from 10-f-H4F.
Here we report the synthesis and inhibitory properties
of the first MAI for AICAR Tfase, â-DADF (Scheme 1).
â-DADF incorporates most structural features of the
natural substrates with two exceptions. To reduce the
complexity of the synthesis, the reduced pterin moiety
of the natural cofactor, 10-f-H4F, was replaced with an
8-deazafolate analogue. This aromatic compound is
* To whom correspondence should be addressed. Phone: 814-865-
2882. Fax: 814-865-2973.
10.1021/jm990323+ CCC: $18.00 © 1999 American Chemical Society
Published on Web 08/14/1999

3422 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 18
Communications to the Editor
Sch em e 2a
a
Reagents and conditions: (a) 10 mol % (PhCN)2PdCl2, NEt3, CH3CN, 100 °C, 8 h; (b) NaOEt, EtOH, 25 °C, 1 h; (c) (i) dibenzyl
diisopropylphosphoramidite, tetrazole, CH2Cl2, 25 °C, 1 h, (ii) MCPBA, -40 to 0 °C, 30 min; (d) (i) 50% TFA/H2O, 25 °C, 3 h, (ii) 10 psi
H2, 10% Pd/C, CH3OH, 25 °C, 2 h, (iii) 0.1 N NaOH, 25 °C, 4 days.
Sch em e 3a
a
Reagents and conditions: (a) pivalic anhydride, pyridine, reflux, 8 h; (b) selenium dioxide, p-dioxane, 100 °C, 3 h; (c) diethyl N-(p-
aminobenzoyl)-L-glutamate, sodium triacetoxyborohydride, 1,2-dichloroethane, 25 °C, 4 h; (d) acryloyl chloride, NEt3, CH2Cl2, 25 °C, 1 h.
more resistant to oxidation, and the stereochemicalre-
quirement of 10-f-H4F at C-6 is absent in the 8-deaza-
folate analogue. This substitution is not expected to
significantly reduce the binding affinity of â-DADF since
10-formyl-8-deazafolate is an alternative substrate for
AICAR Tfase with a Km of 102 µM compared to a Km of
68 µM for 10-f-H4F.8 It has also been shown that
8-deazafolic acid inhibits L. casei AICAR Tfase with an
IC50 of 6.4 µM.9 The 5-exocyclic amino group of AICAR
is also absent in â-DADF since we were not able to find
a suitable electrophile that would react with this amine.
The decreased nucleophilicity of the exocyclic amine is
due to the electron-withdrawing 4-carboxamide, which
reduces the pKa of the 5-exocyclic amine to 2.4.10
Due to the lack of reactivity of the 5-amino group of
AICAR, we decided to explore other methods for the
coupling of AICAR and 10-f-H4F moieties. Matsuda had
reported palladium-catalyzed cross-coupling reactions
of 5-iodoimidazole nucleosides with terminal alkynes11
and methyl acrylate,12 and this prompted us to try a
similar approach. The synthesis of â-DADF was achieved
via a convergent synthesis that centered on a palladium-
catalyzed cross-coupling reaction of a suitably protected
5-iodoimidazole nucleoside (4) and a 10-acryloylfolate
derivative (8) (Scheme 2). Initially, to minimize the
number of transformations after the coupling, we had
hoped to use a nucleoside derivative that had a pro-
tected 5′-phosphate already in place (4b), but this
compound decomposed under the coupling conditions.
Therefore, we prepared compound 4a , which was readily
accessible in three steps from commercially available
AICAR. In the first step, acetonide formation was
achieved with hydrogen chloride in ethanol-acetone
according to the procedure of Robbins,13 followed by
acetylation of the 5′-OH with acetic anhydride andtri-
ethylamine. The 5-iodo group was then introduced by
diazotization with isoamyl nitrite in diiodomethane at
100 °C in a manner similar to that reported by Matsuda
to prepare 5-iodo-1-(2,3,5-tri-O-acetyl-â-D-ribofuranosyl)
imidazole-4-carboxamide.11
The synthesis of 8 is outlined in Scheme 3. Treatment
of 6-methyl-8-deazapterin14 with pivalic anhydride in
pyridine afforded 2-pivaloyl-6-methyl-8-deazapterin. The
2-pivaloate group was important because it increased
the solubility of the 8-deazapterin compounds and thus
simplified purification. The 6-methyl group was oxidized
with selenium dioxide in refluxing p-dioxane to the
aldehyde 6 with overoxidation to the acid as a side
reaction. The aldehyde was conveniently purified by
chromatography on silica gel. Compound 6 readily

Communications to the Editor
J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 18 3423
rescence, and ET is the amount of enzyme added. The
KD was determined to be 20 nM, which shows that
â-DADF binds tightly to AICAR Tfase, 103-fold more
tightly than the individual substrates.20 The inflection
point in Figure 1 is close to the concentration of â-DADF
indicating that the binding stoichiometry is 1:1. â-DADF
was competitive with both substrates precluding a
straightforward assessment of Ki; however an IC50 of
125 nM (10 nM AICAR Tfase, 25 µM AICAR, and 50
µM 10-f-H4F) indicated that â-DADF was a potent
inhibitor of AICAR Tfase.21 â-DADF was also selective
for AICAR Tfase; concentrations that completely inhibit
AICAR Tfase have no effect on GAR Tfase, an enzyme
that also uses 10-f-H4F as a cofactor.22
To date, in all of the organisms in which AICAR Tfase
has been characterized the enzyme exists as a bifunc-
tional protein that contains both AICAR Tfase and
inosine monophosphate cyclohydrolase (IMP cyclase)
activity. IMP cyclase catalyzes the cyclization of 5-formyl-
AICAR to IMP, and the question has arisen as to
whether both of these activities reside in the same active
site on the enzyme. Beardsley has prepared truncation
mutants that have only AICAR Tfase activity suggesting
that the two activities reside in different active sites.23
Our results are in agreement: â-DADF concentrations
which abolish AICAR Tfase activity have no effect on
IMP cyclase activity, and therefore the two active sites
are distinct.24
F igu r e 1. Fluorescence titration of 40 nM â-DADF with
AICAR Tfase, excitation at 278 nm and emission at 415 nm.
The line is the best fit of the data to eq 1; KD ) 20 nM.
underwent reductive amination with diethyl N-(p-ami-
nobenzoyl)-L-glutamate using sodium triacetoxyborohy-
dride in 1,2-dichloroethane to give 7. Treatment with
acryloyl chloride and triethylamine in dichloromethane
gave the desired 10-acryloylfolate 8. The position of the
acryloyl group was confirmed based on the 1H NMR,
since in compound 8 the 9-CH2 is a singlet and the 10-
NH resonance is absent, whereas in compound 7 the
9-CH2 and 10-NH were coupled appearing as a doublet
and triplet, respectively.
The palladium-catalyzed cross-coupling of 4a and 8
to give 9a was achieved using bis(benzonitrile)palladium
chloride and triethylamine in acetonitrile at 100 °C in
a sealed tube. Only one regioisomer was obtained, and
the magnitude of the coupling constant, 15.8 Hz,
indicates that the orientation of the double bond was
trans. The 5′-acetate was then selectively removed with
sodium ethoxide in ethanol, and phosphorylation of the
5′-OH was accomplished using a standard phosphora-
midite procedure15 to give 9b. The acetonide was
removed with 50% TFA/H2O, followed by debenzylation
with H2 over 10% palladium on carbon. The ethyl and
pivaloate esters were deprotected simultaneously with
0.1 N NaOH at 25 °C for 4 days to give â-DADF.16 More
concentrated NaOH solutions resulted in cleavage of the
N-10 amide bond. The structure of â-DADF was con-
firmed by 1H NMR, 31P NMR, and mass spectral data.17
The thermodynamic dissociation constant, KD, for the
E-MAI complex was determined by following the en-
hancement in fluorescence at 415 nm (excitation at 278
nm) of â-DADF upon binding to AICAR Tfase (Figure
1). Aliquots of a concentrated AICAR Tfase solution
were added to a 40 nM â-DADF solution, and measure-
ments were taken after a 2-min incubation period.18 The
data were fit to an equation that describes the fluores-
cence in measurable quantities:19
Ack n ow led gm en t. This work was supported by
PHS Grant GM24129-21 from the National Institute of
Health (S.J .B.).
Refer en ces
(1) Warren, M. S.; Mattia, K. M.; Marolewski, A. E.; Benkovic, S.
J . The transformylase enzymes of de novo purine biosynthesis.
Pure Appl. Chem. 1996, 68, 2029-2036.
(2) Elion, G. B. The purine path to chemotherapy. Science 1989,
244, 41-47.
(3) Broom, A. D. Rational design of enzyme inhibitors: multisub-
strate analogue inhibitors. J . Med. Chem. 1989, 32, 2-7.
(4) Mueller, W. T.; Benkovic, S. J . On the purification and mecha-
nism of action of 5-aminoimidazole-4-carboxamide-ribonucleotide
transformylase from chicken liver. Biochemistry 1981, 20, 344-
350.
(5) Smith, G. K.; Mueller, W. T.; Slieker, L. J .; DeBrosse, C. W.;
Benkovic, S. J . Direct transfer of one-carbon units in the
transformylations of de novo purine biosynthesis. Biochemistry
1982, 21, 1270-1278.
(6) Inglese, J .; Blatchly, R. A.; Benkovic, S. J . A multisubstrate
adduct inhibitor of a purine biosynthetic enzyme with a pico-
molar dissociation constant. J . Med. Chem. 1989, 32, 937-940.
(7) Inglese, J .; Benkovic, S. J . Multisubstrate adduct inhibitors of
glycinamide ribonucleotide transformylase: synthetic and enzyme-
assembled. Tetrahedron 1991, 47, 2351-2364.
(8) Smith, G. K.; Mueller, W. T.; Benkovic, P. A.; Benkovic, S. J .
On the cofactor specificity of glycinamide ribonucleotide and
5-aminoimidazole-4-carboxamide ribonucleotide transformylase
from chicken liver. Biochemistry 1981, 20, 1241-1245.
(9) Degraw, J . I.; Colwell, W. T.; Brown, V. H.; Sato, M.; Kisliuk, R.
L.; Gaumont, Y.; Thorndike, J .; Sirotnak, F. M. Synthesis and
biological activity of 8-deazahomofolic acid and its tetrahydro
derivative. J . Med. Chem. 1988, 31, 150-153.
F ) Fo +
(10) Yamazaki, A.; Okutsu, M. Cyclization of 5-amino-1-â-D-ribo-
furanosylimidazole-4-carboxamide (AICA-riboside): a review. J .
Heterocycl. Chem. 1978, 15, 353-358.
(F∞ - Fo){(ET + I + K ) - (E + IT + KD)2 - 4ET‚IT}
x
T
D
T
(11) Minakawa, N.; Takeda, T.; Sasaki, T.; Matsuda, A.; Ueda, T.
Nucleosides and nucleotides. 96. Synthesis and antitumor activ-
ity of 5-ethynyl-1-â-D-ribofuranosylimidazole-4-carboxamide (EI-
CAR) and its derivatives. J . Med. Chem. 1991, 34, 778-786.
(12) Minakawa, N.; Kojima, N.; Matsuda, A. Nucleosides and nucle-
otides. 138. Synthesis of 3-halo-3-deazainosines: conformational
lock with the halogen at the position of the 3-deazainosine in
anti-conformation. Heterocycles 1996, 42, 149-154.
2IT
(1)
where IT is the total inhibitor added, F∞ is the fluores-
cence of the enzyme-inhibitor complex, Fo is the
fluorescence of the inhibitor, F is the measured fluo-

3424 J ournal of Medicinal Chemistry, 1999, Vol. 42, No. 18
Communications to the Editor
(13) Srivastava, P. C.; Newman, A. R.; Matthews, T. R.; Robins, R.
K. Synthesis of 5-amino-1-(5-deoxy-â-D-ribofuranosyl)imidazole-
4-carboxamide and related 5′-deoxyimidazole ribonucleosides. J .
Med. Chem. 1975, 18, 1237-1240.
(19) Taira, K.; Benkovic, S. J . Evaluation of the importance of
hydrophobic interactions in drug binding to dihydrofolate re-
ductase. J . Med. Chem. 1988, 31, 129-137.
(20) Km for AICAR is 17 µM, and Km for 10-f-H4F is 60 µM.23
(21) Inhibition assays were performed by following the production
of H4F at 298 nm in 33 mM Tris buffer containing 25 mM KCl
at pH 7.5, 25 °C. Although â-DADF is a potent inhibitor of
AICAR Tfase, its therapeutic potential may be limited due to
its anionic nature, which most likely prohibits transport across
biological membranes.
(14) Kelley, J . L.; McLean, E. W. Synthesis and antimicrobial testing
of 2-amino-6-hydroxymethyl-4-(3H)pyrido[3,2-d]pyrimidinone. J .
Heterocycl. Chem. 1981, 18, 671-673.
(15) Yu, K.-L.; Fraser-Reid, B. A novel reagent for the synthesis of
myo-inositol phosphates: N,N-diisopropyl dibenzyl phosphora-
midite. Tetrahedron Lett. 1988, 29, 979-982.
(16) â-DADF was purified by reversed-phase HPLC on a Whatman
Partisil 10 ODS-3 9.4-mm × 50-cm column with a linear gradient
of 0-50% acetonitrile in 0.1% TFA/H2O over 30 min.
(22) E. coli GAR Tfase (5 nM) was assayed by following the produc-
tion of 5,8-dideazafolate at 295 nm in 50 mM Tris buffer, pH
7.5, 25 °C, containing 40 µM 10-f-5,8-dideazafolate and 140 µM
â-GAR. The presence of 10 µM â-DADF had no effect on the
production of 5,8-dideazafolate, a concentration that completely
inhibited 10 nM AICAR Tfase (50 µM 10-f-H4F and 25 µM
AICAR).
(23) Rayl, E. A.; Moroson, B. A.; Beardsley, G. P. The human purH
gene product, 5-aminoimidazole-4-carboxamide ribonucleotide
formyltransferase/IMP cyclohydrolase cloning, sequencing, ex-
pression, purification, kinetic analysis, and domain mapping. J .
Biol. Chem. 1996, 271, 2225-2233.
(17) 1H NMR (D2O): δ 8.34 (s, 1 H, imidazole H-2); 7.96 (d, 1 H, vinyl
H, J ) 15.9 Hz); 7.86 (d, 2 H, benzoyl H, J ) 8.6 Hz); 7.60 (q, 2
H, folate H-7,8); 7.45 (d, 2 H, benzoyl H, J ) 8.0 Hz); 6.72 (d, 1
H, vinyl H, J ) 15.8 Hz); 5.63 (d, 1 H, H-1′, J ) 7.0 Hz); 5.27
(m, 2 H, folate 9-CH2); 4.58 (m, 1 H, H-2′); 4.33 (m, 2 H, H-3′,4′);
4.07 (m, 1 H, glutamic acid R-H); 3.89 (m, 2 H, 5′-CH2); 2.32 (t,
2 H, glutamic acid γ-H’s, J ) 7.7 Hz); 2.23-1.98 (m, 2 H,
glutamic acid â-H’s). 31P NMR (D2O): δ 4.4 (s, 5′-PO4). HRMS:
calcd for C32H34N9O15P, MH+ 816.1990; found, 816.2000.
(18) Fluorescence titration assays were carried out at 25 °C, pH 7.5
in 33 mM HEPES buffer and 25 mM KCl. Enzyme was purified
from E. coli BL21(DE3) containing a plasmid that codes a His-
tagged human purH cDNA. The concentration of â-DADF was
determined by measuring the concentration of inorganic phos-
phate (Chen, P. S.; Toribara, T. Y.; Warner, H. Microdetermi-
nation of phosphorus. Anal. Chem. 1956, 28, 1756-1758) in a
solution of â-DADF that was hydrolyzed in 6 N HCl at 100 °C
in a sealed tube for 12 h.
(24) IMP cyclase (10 nM) was assayed by following the conversion of
5-formyl-AICAR to IMP at 248 nm in 33 mM Tris buffer
containing 25 mM KCl, pH 7.5, 25 °C. Concentrations of 10 µM
â-DADF had no effect on the rate of IMP production.
J M990323+