A disubstituted NAD+ analogue is a nanomolar inhibitor of trypanosomal glyceraldehyde-3-phosphate dehydrogenase

Article abstract of DOI:10.1016/S0960-894X(00)00608-9

N⁶-Naphthalenemethyl-2′-methoxybenzamido-β-NAD ⁺, a derivative of a low micromolar first-generation inhibitor of trypanosomal glyceraldehyde phosphate dehydrogenase (GAPDH), was synthesized, taking advantage of methodology for the selective phosphitylation of nucleosides. The compound was found to be a poor alternate cosubstrate for GAPDH, but an extremely potent inhibitor. Although intended for use in crystallization trials, the analogue presents possibilities for further drug design.

Full text of DOI:10.1016/S0960-894X(00)00608-9

Bioorganic & Medicinal Chemistry Letters 11 (2001) 95±98
A Disubstituted NAD+ Analogue is a Nanomolar Inhibitor of
Trypanosomal Glyceraldehyde-3-phosphate Dehydrogenase
Kevin J. Kennedy, Jerome C. Bressi and Michael H. Gelb*
Departments of Chemistry and Biochemistry, University of Washington, Seattle, WA 98195, USA
Received 13 July 2000; accepted 12 October 2000
AbstractÐN6-Naphthalenemethyl-20-methoxybenzamido-b-NAD+, a derivative of a low micromolar ®rst-generation inhibitor of
trypanosomal glyceraldehyde phosphate dehydrogenase (GAPDH), was synthesized, taking advantage of methodology for the
selective phosphitylation of nucleosides. The compound was found to be a poor alternate cosubstrate for GAPDH, but an extre-
mely potent inhibitor. Although intended for use in crystallization trials, the analogue presents possibilities for further drug design.
# 2001 Elsevier Science Ltd. All rights reserved.
Introduction
was obtained by rational design, exploiting structural
di€erences between the human and parasite enzymes
around the adenosyl moiety of the bound NAD+
cosubstrate. These di€erences are re¯ected in a 10-fold
lower anity for NAD+ measured in trypanosomal
GAPDH (Km=0.45 mM) as compared to the human
enzyme (Km=0.04 mM).4
Parasitic infections are a serious global health problem,
a€ecting millions of people annually. Speci®cally,
trypanosomiasis, those diseases caused by the haemo-
¯agellates of the genus Trypanosoma, and leishmaniasis,
the disease caused by protozoa of the related genus
Leishmania, are of major concern. However, despite the
public threat posed by these diseases, currently available
treatments are inadequate and are limited by invasive
delivery routes, toxicity, and ine€ectiveness.1
Compound 1 does not inhibit human GAPDH at its
solubility limit of 50 mM, and this selectivity is believed
to arise from binding of the 20-methoxybenzamido
group into a narrow hydrophobic cleft which is found in
trypanosomal GAPDH but occluded in the human
enzyme. In order to initiate another round of structure-
based drug design aimed at improving the potency of
this ®rst-generation inhibitor, it is necessary to obtain a
crystal structure of 1 in complex with GAPDH. How-
ever, attempts to crystallize with inhibitor 1 have thus
One way of ®nding new treatments for these diseases is
by identifying biochemical processes that are unique to
parasite physiology. One such feature that o€ers an
opportunity for exploitation is the reliance of trypano-
somes on glycolysis for energy production. Speci®cally,
the enzyme glyceraldehyde-3-phosphate dehydrogenase
(GAPDH), which catalyzes the conversion of glycer-
aldehyde-3-phosphate to 1,3-diphosphoglycerate with
concomitant reduction of NAD+, o€ers a target that
may be exploited to obtain inhibitors of trypanosomal
and leishmanial glycolysis and growth.2 Guided by this
hypothesis, earlier e€orts in this laboratory resulted in
the N6-, 20-disubstituted adenosine analogue 1, which is
a selective, low micromolar inhibitor of trypanosomatid
GAPDH, and which e€ectively blocks in vitro growth
of the bloodstream form of T. brucei (ED50=30 mM)
and T. cruzi amastigotes (ED50=20 mM).3 This compound
*Corresponding author. Tel.: +1-206-543-7142; fax: +1-206-685-
8665; e-mail: gelb@chem.washington.edu
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(00)00608-9

96
K. J. Kennedy et al. / Bioorg. Med. Chem. Lett. 11 (2001) 95±98
far proved unsuccessful, probably because it does not
induce the same conformational change as does the
binding of NAD+ to GAPDH (GAPDH without
bound NAD+ also fails to crystallize). A structure of
N6-benzyl-NAD+ bound to GAPDH has been solved, a
further indication that the full dinucleotide structure is
necessary for crystallization.3 We therefore reasoned that
coupling inhibitor 1 to b-nicotinamide mononucleotide
(b-NMN), and thereby more closely mimicking the
normal NAD+ cosubstrate, would increase the chance
of obtaining co-crystals. The synthesis of this analogue,
which extends methodology for selective phosphitylation
to a novel nucleoside substrate, is described. Also,
inhibition data is presented which show the compound to
be an extremely potent and selective GAPDH inhibitor.
substituted riboside, we expected that the disubstituted
nucleoside 1 would serve equally well as a substrate.
Therefore, following this methodology, selective phos-
phitylation of the 50-hydroxyl of 1 was accomplished
using the phosphoramidite, bis-(2-cyanoethoxy)(diiso-
propylamino)phosphine,6 activated with a substituted
tetrazole. Following oxidation, the resulting phospho-
triester, as well as unreacted starting material, were readily
puri®ed by silica gel chromatography. Deprotection of
the product gave 50-phosphorylated compound 2 (57%
based on limiting phosphoramidite, 80% based on total
riboside recovered), which was isolated as the free acid
by precipitation with 10% citric acid. Coupling of the
50-phosphate to b-NMN was accomplished by ®rst form-
ing the activated phosphoromorpholidate 3 (62%),7 and
then subsequently reacting with b-NMN in formamide
with added Lewis acid catalyst.8 The ®nal product 4 was
isolated by reverse-phase HPLC (19% isolated yield,
10% overall from 1)9 (Scheme 1).
Chemistry
The initial step of the synthesis involves the selective
phosphorylation of the 50-hydroxyl of 1. This is
commonly accomplished by employing a protection±
deprotection strategy, ®rst using a bulky protecting
group (typically dimethoxytrityl) to block the sterically
less hindered primary hydroxyl, followed by protection
of the secondary groups; the 50-hydroxyl can then be
selectively deprotected and phosphorylated. Recently
however, a useful method for the direct phosphitylation
of the primary hydroxyl group in these unprotected
nucleosides has been described.5 Although these authors
applied this chemistry to the anomers of only one mono-
Results
Compound 4 is an analogue of the NAD+ cosubstrate
normally used during catalysis by GAPDH and so
could potentially serve as an alternate cosubstrate of the
enzyme. However, when substituted for NAD+ at the Km
concentration (0.45 mM), 4 was used by L. mexicana
GAPDH at 5% of the rate of NAD+ utilization. This is
in contrast to N6-benzyl-NAD+ which was found to be
a good alternate substrate for L. mexicana GAPDH,
Scheme 1. Synthesis of disubstituted NAD+ analogue 4.

K. J. Kennedy et al. / Bioorg. Med. Chem. Lett. 11 (2001) 95±98
97
with a Km of 0.25 mM.3 Either the additional 20 sub-
stitution on the adenosine of 4 induces a binding geo-
metry such that the nicotinamide moiety is no longer
optimally aligned for hydride transfer, or the rate of
product (NADH) release is slowed by the additional
binding energy from the added adenosyl substituents.
The analogue does ®t nicely in the NAD+ binding
pocket and shows high anity for the enzyme, a fact
con®rmed by its potency as an inhibitor. These results,
in comparison to the inhibition data for the ®rst-
generation antagonist 1, are shown in Table 1.
of the negatively charged phosphate. This was expected,
as the pyrophosphate of bound NAD+ is surface
exposed and thus the charged linker is largely solvated.
This is in contrast with other nucleotide-binding
enzymes that make important electrostatic contacts with
the nucleotide 50-phosphate group, and thus bind cor-
responding nucleosides with much lower anity.10
As the phosphate substituent does not contribute much
to the increased anity of 4, it was thought that the
nicotinamide mononucleotide half of the compound
might have appreciable anity for the enzyme. How-
ever, b-NMN did not inhibit L. mexicana GAPDH
when tested as high as 500 mM. The additional 30-fold
increase in potency gained when this moiety is coupled
to 2 must thus be due to entropic contributions. That is,
although the NMN moiety itself has low anity for its
binding pocket, the binding of the N6-napthalenemethyl
and 20-methoxybenzamido substituents in the adenosine
pocket serves to lock the inhibitor in place, and allows
the NMN group to bind with little additional loss of
entropy. Presumably it would be possible to take further
advantage of the entropic contribution by coupling the
two nucleotide moieties with a more rigid linker.
Compound 4 was found to be a nanomolar inhibitor of
L. mexicana GAPDH, 100 times more potent than the
®rst generation antagonist 1. For purposes of structure-
based drug design, this is the enzyme of choice as it is
easiest to express in E. coli and provides the best dif-
fracting crystals. However, the sequence homology
between all three trypanosomatid enzymes is high, and
the NAD+ binding pocket is virtually identical for each,
with the exception of the substitution of a serine (Ser-40)
in L. mexicana GAPDH by asparagine (Asn-40) in T.
brucei/T. cruzi GAPDH. This di€erence occurs in the
hydrophobic selectivity cleft in which the 20-methoxy-
benzamido substituent is thought to bind, and this may
account for the slight di€erence in anity of 1 for the
three parasite enzymes. Compound 4 is also a potent
inhibitor of T. brucei and T. cruzi GAPDH but, con-
sistent with the presence of the amino acid substitution
in the selectivity cleft, slightly less potent than against
the L. mexicana enzyme. However, the relative di€er-
ence in potency between the three enzymes is not as
pronounced in 4 as compared to 1, presumably because
the sizeable increase in anity provided by the addition
of NMN attenuates the contribution to binding from
the selectivity cleft interaction. Finally, in accordance with
the selectivity of 20-substituted adenosines for trypano-
somal GAPDH, 2 and 4 did not inhibit mammalian
GAPDH at 100 and 500 mM, respectively.
The high molecular weight and charged nature of 4 would
seem to preclude its use as an e€ective therapeutic agent
against live parasites. As expected, when tested against
the bloodstream form of T. brucei, 4 and the phos-
phorylated analogue 2 both showed lower activity than
the lead inhibitor (Table 1). However, against T. cruzi
amastigotes both compounds were approximately 3-fold
more potent than 1. Although only a modest increase,
this result is an indication that despite their charged
nature, these analogues are able to penetrate parasitic
cells. Thus, in addition to being a tool for use in crys-
tallization trials, compound 4 provides insight towards
further drug design based upon lead inhibitor 1. Even in
the absence of a crystal structure, the potential for sub-
stitution at the 50-hydroxyl of the lead inhibitor is readily
apparent. A small set of 50-substituted analogues of 1
have previously been synthesized with no resultant
increase in anity, but these compounds do not ®ll the
NMN cavity.11 A substituent which could ®ll this
pocket connected with a linker of suitable length could
result in an analogue with equal or greater potency than
4. As noted, the phosphate moiety does not contribute
signi®cantly to binding anity, so variation of the
In order to gain insight into the binding interactions
responsible for the high anity of 4, the two joined
fragments of the analogue were also tested for GAPDH
inhibition. Compound 2, the 50-phosphate of 1, was
found to be roughly 3-fold more potent than the parent
compound against each of the three enzymes. The small
increase in potency indicates that there are not sig-
ni®cant electrostatic interactions gained from addition
a
Table 1. E€ect of analogues on trypanosomatid and mammalian GAPDH (IC50) and growth of cultured parasites (ED50
)
L. mexicana
T. brucei
T. cruzi
Rabbit muscle
IC50
3T3 Fibroblastsd
ED50
b
c
Compounds
IC50
IC50
ED50
IC50
ED50
1
2
4
6
2
25
7
0.10
Ð
Ð
Ð
30
32
63
Ð
0.008
Ð
12
5
0.10
Ð
Ð
Ð
20
6
8
Ð
Ð
0.6
>50
>100
>500
>500
Ð
>50
90
>150
Ð
Ð
0.06
>500
Ð
b-NMN
Pentamidine
Benznidazole
Ð
Ð
>50
aAll values are in mM; data for 1 is from ref 3; pentamidine and benznidazole values provided for reference.
bIC50 values determined by measuring inhibition at 3±5 inhibitor concentrations (10% error).
cED50 values determined in triplicate (15% error limit) for bloodstream T. brucei and T. cruzi amastigotes according to previous procedure.3
dNoninfected host cells used to support growth of T. cruzi amastigotes.

98
K. J. Kennedy et al. / Bioorg. Med. Chem. Lett. 11 (2001) 95±98
pyrophosphate linker, which is susceptible to cleavage
by nonspeci®c phospho-hydrolyases, would be possible.
Other groups have examined using more stable phos-
phonate12 or methylenebisphosphonate linkers13 to make
NAD+ analogues, but in this case a simple ether-type
linkage without phosphates could also be feasible. Our
expectation is that attempts to design analogues that
exploit these additional binding regions will result in
even more potent inhibitors of glycolysis and potential
treatments for the devastating diseases of trypano-
somiasis and leishmaniasis.
(0.074 mmol) of 1 and 5-ethylthiotetrazole (0.17 mmol) were
dissolved in 3 mL dry CH3CN and stirred at 10 ꢀC under
argon atmosphere. Bis-(2-cyanoethoxy)-(diisopropylamino)-
phosphine (0.45 M solution in dry CH3CN, 0.06 mmol 133 mL)
was added slowly. After complete addition the solution was
stirred at 10 ꢀC for 30 min and then tert-butyl hydroperoxide
(0.67 mmol) was added, the solution was warmed to room
temp. and stirred for an additional 20 min. Solvent was eva-
porated, the residue was diluted with CHCl3 and the product
was puri®ed by silica gel chromatography eluting with 9:1
CHCl3:EtOH. The puri®ed phosphate triester was deprotected
by stirring at room temperature in methanolic ammonium
hydroxide for 17 h. Solvent was removed to a small volume
and 2 was precipitated as its free acid by the addition of 10%
citric acid solution. The 50-phosphate (26 mg, 0.04 mmol) was
converted to the phosphoromorpholidate by dissolving with
morpholine (0.17 mmol, 15 mL) in 0.4 mL t-BuOH and 0.4 mL
H2O and heating at re¯ux. DCC (0.17 mmol, 35 mg) in 0.6 mL
t-BuOH was added slowly over 4 h, and after complete addi-
tion re¯ux continued for 16 h. The cooled solution was ®ltered,
evaporated to dryness and washed with ether. The crude
material was puri®ed by silica gel chromatography, eluting
with 7:3 CH2Cl2:MeOH (Rf 0.4) to give 3. Pyrophosphate
Acknowledgements
This research was supported by grant A144199-01A1
from the National Institutes of Health. We wish to
thank Dr. Frederick Buckner, Department of Medicine,
University of Washington, for conducting parasite
growth inhibition assays.
coupling was accomplished by dissolving
3
(16 mg,
0.023 mmol) with b-NMN (0.025 mmol, 8.3 mg) in 275 mL
formamide containing MnCl2 (0.15 mmol) and MgSO4
(0.05 mmol), and stirring at room temperature for 24 h. The
crude product was precipitated by the addition of acetone,
and the ®nal product 4 was puri®ed as the triethylammonium
salt by reverse-phase HPLC using a CH3CN:H2O (with
0.01 M triethylammonium acetate) gradient. 1H NMR
(300 MHz, CD3CN/D2O) d 1.18 (t, 9H, J=7.3 Hz), 3.05 (q,
6H, J=7.3 Hz), 3.67 (s, 3H), 4.05±4.60 (m, 9H), 5.00±5.15 (m,
3H), 5.98 (d, 1H, J=5.2 Hz), 6.23 (d, 1H, J=8.3 Hz), 7.00±
8.20 (m, 13H), 8.45 (s, 1H), 8.79 (d, 1H, J=5.2 Hz), 9.10 (d,
1H, J=9.4 Hz), 9.38 (s, 1H); 31P NMR (CD3CN/D2O, 1 M
H3PO4 external standard) d 10.97; ESI-MS (H2O) 935.24
[M H+] .
10. Erion, M. D.; Kasibhatla, S. R.; Bookser, B. C.; van
Poelje, P. D.; Reddy, M. R.; Gruber, H. E.; Appleman, J. E.
J. Am. Chem. Soc. 1999, 121, 308.
11. Bressi, J. B., unpublished results.
12. Hockova, D.; Masojidkova, M.; Holy, A. Coll. Czech.
Chem. Commun. 1996, 61, 1538.
References and Notes
1. Pepin, J.; Milford, F. Adv. Parasitol. 1994, 33, 1.
2. Hol, W. G. J.; Vellieux, F. M. D.; Verlinde, C. L. M. J.;
Wierenga, R. K.; Noble, M. E. M.; Read, R. J. In Molecular
Conformations and Biological Interactions; Belurapa, P., Rama-
seshan, S., Eds.; Indian Acad. Sci.: Bangalore, 1992; pp 215±244.
3. Aronov, A. M.; Suresh, S.; Buckner, F. S.; Van Voorhis,
W. C.; Verlinde, C. L. M. J.; Opperdoes, F. R.; Hol, W. G. J.;
Gelb, M. H. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 4273.
4. Lambeir, A.-M.; Loiseau, A. M.; Kuntz, D. A.; Vellieux, F.
M.; Michels, P. A. M.; Opperdoes, F. R. Eur. J. Biochem.
1991, 198, 429.
5. Graham, S. M.; Pope, S. C. Org. Lett. 1999, 1, 733.
6. Bannwarth, W.; Trzeciak, A. Helv. Chim. Acta 1987, 70, 175.
7. Mo€at, J. G.; Khorana, H. G. J. Am. Chem. Soc. 1961, 83,
649.
8. Lee, J.; Churchil, H.; Choi, W.-B.; Lynch, J. E.; Roberts, F.
E.; Volante, R. P.; Reider, P. J. Chem. Commun. 1999, 729.
9. The detailed synthesis of compound 4 is as follows: 40 mg
13. Hutchinson, E. J.; Taylor, B. F.; Blackburn, G. M. Chem.
Commun. 1996, 2765.