Thymidine and thymidine-5′-O-monophosphate analogues as inhibitors of Mycobacterium tuberculosis thymidylate kinase

Article abstract of DOI:10.1016/S0960-894X(03)00643-7

The affinity of a series of 2′, 3′- and 5-modified thymidine analogues for Mycobacterium tuberculosis thymidine monophosphate kinase (TMPKmt) was evaluated. The affinities of several non-phosphorylated analogues are in the same order of magnitude as those of their phosphorylated congeners. In view of drug delivery problems associated with phosphorylated compounds, these 'free' nucleosides seem more promising leads in the search of TMPKmt inhibitors as novel anti-tuberculosis agents.

Full text of DOI:10.1016/S0960-894X(03)00643-7

Bioorganic & Medicinal Chemistry Letters 13 (2003) 3045–3048
0
Thymidine and Thymidine-5 -O-monophosphate Analogues as
Inhibitors of Mycobacterium tuberculosis Thymidylate Kinase
a
a
b
c
Veerle Vanheusden, Philippe Van Rompaey, He
´
d
le
`
ne Munier-Lehmann, Sylvie Pochet,
a,
Piet Herdewijn and Serge Van Calenbergh *
aLaboratory for Medicinal Chemistry (FFW), Ghent University, Harelbekestraat 72, B-9000 Gent, Belgium
bLaboratoires de Chimie Structurale des Macromol e´ cules, Institut Pasteur, 75724 Paris Cedex 15, France
cUnite´ de Chimie Organique (URA2128), 75724 Paris Cedex 15, France
dLaboratory for Medicinal Chemistry, Rega Institute, Catholic University of Leuven, B-3000 Leuven, Belgium
Received 17 March 2003; accepted 10 April 2003
0
0
Abstract—The affinity of a series of 2 , 3 - and 5-modified thymidine analogues for Mycobacterium tuberculosis thymidine mono-
phosphate kinase (TMPKmt) was evaluated. The affinities of several non-phosphorylated analogues are in the same order of mag-
nitude as those of their phosphorylated congeners. In view of drug delivery problems associated with phosphorylated compounds,
these ‘free’ nucleosides seem more promisingleads in the search of TMPKmt inhibitors as novel anti-tuberculosis a ge nts.
#
2003 Elsevier Ltd. All rights reserved.
Recently, Mycobacterium tuberculosis thymidine mono-
phosphate kinase (TMPKmt) was put forward as an
attractive target for the design of a novel class of anti-
pyrimidine ringand Phe70 should account for a dTMP-
like orientation of the substrate analogues under inves-
tigation, thereby rendering prediction and modelling of
1ꢀ3
tuberculosis agents.
TMPK accounts for the phosphorylation of thymidine
monophosphate (dTMP) to thymidine diphosphate
(
dTDP), usingATP as a preferred phosphoryl donor. 4
Because TMPK lies at the junction of the de novo and
salvage pathways of thymidine triphosphate (dTTP)
synthesis and in view of its low (22%) sequence identity
5
with the human isozyme, it represents an attractive
target for selectively blocking mycobacterial DNA
synthesis. Recently, the X-ray structure of TMPKmt
˚
6
was solved at 1.95 A as a complex with dTMP, allow-
ing structure-based design of TMPKmt ligands (Fig. 1).
In order to establish initial structure–activity relation-
ships, a series of sugar- and base-modified nucleoside-5 -
0
O-monophosphates were examined for TMPKmt affi-
0
moiety was to preserve its hydrogen bonding and ionic
nity. The aim of conservingthe 5 -O-monophosphate
2
+
interactions with Tyr39, Arg95 and Mg . These bind-
ings, together with the stacking interaction between the
*Correspondingauthor. Tel.: +32-9-264-8124; fax: +32-9-264-8146;
e-mail: serge.vancalenbergh@rug.ac.be
Figure 1. Schematic representation of the most important amino acid
residues of TMPKmt interactingwith TMP (in black).
0
960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00643-7

3046
V. Vanheusden et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3045–3048
the interactions of the various dTMP substituents with
the enzyme more straightforward. Next, alternative
substitution patterns at the 5 -position would be
three compounds are inhibitors because any modifi-
cation, significantly perturbing the volume at the 5-
position, changes the orientation of the sugar moiety as
well as that of the a-phosphate. Surprisingly, removal
of the 5-methyl group of dTMP (dUMP) caused a
0
2
explored and combined with optimal sugar and base
modifications. These sugar- and base-modified nucleo-
tides were examined on their affinity for the enzyme
drastic decrease in affinity (K =2100 mM compared to
m
7
5
usinga reported spectrophotometric assay,
mentioned otherwise.
unless
4.5 mM for dTMP) (Table 1).
The observation that TMPKmt, as opposed to all
5
First, systematic substitution of the 5-methyl group for
various halogen atoms was explored. 5F-dUMP, 5Br-
dUMP and 5I-dUMP turned out to be substrates for
TMPKs of other species examined so far, is strongly
inhibited by AZTMP (K =10 mM) offered good pro-
i
spects for findingselective inhibitors of TMPKmt
through altering the sugar part of the dTMP scaffold.
5
TMPKmt. 5Br-dUMP (K =33 mM) showed the high-
m
0
hydrogen bonding network (denoted by the dotted lines,
est affinity, probably because a bromine occupies about
the same steric space as a methyl group. The presence of
the guanidinium group of Arg74 in the otherwise
hydrophobic pocket around the 5-position could
explain the measured affinity of the halogenated sub-
strates. It is believed that the difference in the kinetic
parameters of these compounds essentially reflects a size
effect, with the halogen atoms serving as cavity filling
At the 3 -position, the crystal structure reveals a
0
carboxyl of Asp9, which, in its turn, is linked to the
Fig. 1): the 3 -OH of dTMP interacts with the terminal
2
+
Mg -ion that is responsible for positioninga phos-
phate oxygen of TMP. Through its interaction with
0
Asp9, the 3 -azido function of AZTMP is believed to
perturb the aforementioned hydrogen bonding interplay
that is essential for catalysis, hence completely impeding
5
6
atoms. A co-crystal of 5I-dUMP with TMPKmt
showed that it, indeed, binds the phosphate acceptor
6
the phosphoryl transfer. Although an analogous inter-
action was expected upon reduction of the 3 -azido
2
0
group of AZTMP to a 3 -amine (5), this drastically
bindingsite in a very similar fashion to dTMP.
0
0
of dTMP by a 3 -F (8), on the other hand, afforded an
Since these findings confirmed the assumption that
TMPKmt can accommodate sterically larger sub-
stituents at the 5-position, 1 was synthesised to accom-
plish a favourable interaction with a water molecule
increased the K value to 235 mM. Replacingthe 3 -OH
i
0
analogue that behaves as a substrate with a Km of
30 mM.
(
W12), detected in the crystal structure. In contrast to
0
believed to render the enzyme catalytically selective for
2 -deoxynucleotides versus ribonucleotides. The ribo
the 5-halogeno-substituted substrates, this compound
behaves as a (relatively weak) inhibitor of TMPKmt.
Co-crystallisation with TMPKmt proved the predicted
The presence of Tyr103 close to the 2 -position is
0
2
interaction with W12. Other dTMP analogues with
analogue (6) of 5, however, unexpectedly showed a
much higher affinity for the enzyme (45 mM vs 235 mM),
this compound beingthe first example of a ribonucleo-
tide with good affinity for TMPKmt. Modelling indi-
cates that, in this case, Tyr103 stays somewhat further
sterically more demandingsubstituents that maintain
one polar atom (2 and 3) were found less active inhibi-
tors than 1, indicatingthat the volume of the cavity
cannot be stretched too much. It is inferred that these
0
away from the 2 -position and is, therefore, unable to
clash with the 2 -hydroxyl. Introduction of halogens at
0
0
the 2 -position (9 and 10) is better tolerated than a
1
Table 1. Kinetic parameters of TMPKmt with base-modified
nucleoside monophosphates
hydroxyl group at that position. Modelling showed that
the halogen affects the relative position of the sugar
0
which the 3 -hydroxyl usually resides. As previously
ring. As a result, the 2 -chlorine occupies the pocket, in
0
demonstrated with AZTMP, this particular domain can,
1
indeed, accommodate more voluminous substituents.
Replacingthe five-membered su ga r ringof dTMP by a
1,5-anhydrohexitol yielded a moderately active inhi-
bitor. This effect is probably due to the inability of 11 to
Compd
dTMP5
R
K
m
(mM)
V
m
(mmol/min
mgof protein)
K
i
(mM)
0
position both the phosphate and the 3 -hydroxyl groups
2
CH
H
F
Br
I
3
4.5
10.6
3.5
4.7
9.8
7.5
5
in a favourable arrangement for catalysis. With a K of
i
dUMP
2100
420
33
F-dUMP6
Br-dUMP
30 mM, the bisubstrate analogue Ap T, emerged as a
5
5
5
1
2
3
4
5
6
good inhibitor of TMPKmt. The crystal structure of the
complex of TMPKmt with Ap T, which has been
6
I-dUMP
140
5
2
2
2
2
a
CH
2
OH
110
140
2
˚
established at 2.45 A resolution, revealed an unex-
pected bindingmode for the adenine moiety of this
compound. While the thymidine residue of this bisub-
strate analogue occupies the binding pocket of dTMP,
its ADP unit, surprisingly, fits a cavity on the surface of
TMPKmt. This findingopens avenues to the desi gn of
a
a
Furan-2-yl
Thien-2-yl
Benzyl
270
28
a
The TMP kinase activity of these compounds was not determined
usingthe spectrometric test described by Blondin et al. 7 but using
m
HPLC chromatography. The K for dTMP obtained with this test is
2
40 mM.
branched molecules at the a-phosphate of Ap T, with
5

V. Vanheusden et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3045–3048
3047
0
an additional group reaching out to this cavity present
2
decreased only 2.8 times upon removal of the 5 -O-
phosphoryl. With regard to these observed similar affin-
ities of nucleosides and nucleotides and in view of the
drugdelivery problems of phosphorylated compounds,
nucleosides seem more useful leads for further drug
design.
only in TMPKmt (Table 2).
Obviously, monophosphate derivatives are unable to
enter cells and, as a consequence, can not be delivered in
such a way. Generally, nucleosides cross the cell–mem-
brane barrier in a non-phosphorylated form and are
then converted by intracellular kinases to the corre-
spondingnucleotides. In this particular case where
TMPK is the drug target, a thymidine analogue could
be activated by thymidine kinase (TK), which converts
the analogue into the corresponding monophos-
phorylated compound. However, a search of the M.
0
0
To further explore other substitution patterns at the 5 -
0
position, the 5 -hydroxymethyl was substituted for a 5 -
methyl. In 25, this caused a 42-fold decrease in affinity
compared to 6. The affinities of 20 and 19, on the con-
trary, were similar, makingassumptions about the
0
removal of the 5 -hydroxyl group unreliable. Com-
0
8
tuberculosis genome did not identify any gene coding
for a TK. This result aligns with other biochemical
studies indicatinga lack of TK activity in myco-
pound 24, with a 5 -iodo group, showed a low affinity
(K =490 mM), but it’s not clear if the 5 -iodo group or
0
the 2 -hydroxyl function accounts for this effect.
i
0
9
bacteria. At first sight, the absence of TK seems to be a
severe limitation on the use of thymidine analogues or
related compounds as anti-tuberculosis drugs. However,
the findingthat dT behaves as a competitive inhibitor of
Because 6 conveyed the impression that ribo analogues
are well tolerated by TMPKmt, two other ribonucleo-
sides (15 and 19) were examined. The affinity of 15 was
26 times lower than the affinity of dT and 19 showed to
be a poor inhibitor as well. These results indicate that
the affinity improvement resultingfrom the introduction
TMPKmt (with K in the same order of magnitude as
i
3
K of dTMP ), raises the possibility of discoveringnon-
m
phosphorylated nucleosidic inhibitors of TMPKmt.
Enhancingthe specific delivery of such inhibitors to
0
of a 2 -hydroxyl function in 5 (6) is exceptional. 16 even
0
0
bacteria within macrophages by modifying the 5 -posi-
indicates that a 2 -b-OH is better tolerated.
tion of dTMP is another option. Such modifications
may also decrease toxicity by reducinginteraction with
the host cellular TK. Replacement of the 5 -hydroxyl by
Because AZTMP is one of the best known inhibitors,
another nitrogen-containing substituent, a guanidine
group (26), was introduced at the 3 -position. Although
one would expect the delocalized positive charge of the
guanidine moiety to strengthen the interaction with
0
an azido or amino an group (13 and 14) yielded two
0
high-affinity inhibitors of TMPKmt (K values of 7 and
i
3
1
2 mM, respectively).
Asp9, the K value of 140 mM was not affirmative.
i
We investigated whether the negligible effect on
TMPKmt affinity, observed upon deletion of the phos-
phate moiety of dTMP, is parallelled in other series.
AZT, 5-BrdU and 22 exhibited similar affinities as their
nucleotide congeners. In the case of 23, the affinity
Table 3. Kinetic parameters of TMPKmt with sugar- and base-
modified thymidine analogues
Table 2. Kinetic parameters of TMPKmt with sugar-modified
nucleoside monophosphates
Compd
R1
R2
R3
R4
K
i
(mM)
ꢀ
3
4.5a
27
28
5
90
7
dTMP
dT
AZT
CH
CH
CH
Br
3
H
H
H
H
H
H
H
OH
OH
H
H
OH
OH
H
H
F
OH
OH
N
3
OPO
OH
OH
H
3
3
3
5
1
1
Br-dU
3
H
2
CH
CH
CH
CH
CH
3
3
3
3
3
OH
OH
OH
OH
OH
OH
OH
OH
OH
NHCOCH
3
Compd
R1
R2
H
R3
K
i
(mM)
3
3
N
3
NH
2
ꢀ
4.5a
3
dTMP
AZTMP
H
H
OPO
OPO
OPO
OPO
OPO
OPO
OPO
OPO
OPO
3
14
15
16
17
18
19
20
21
22
23
24
2
12
5
2ꢀ
N
3
3
10
OH
OH
OH
OH
OH
H
OH
OH
OH
I
715
238
97
1140
521
560
230
28
122
490
1900
140
5
2ꢀ
3
a
b
D4T-MP
1
Dehydro
H
OH
NH
H
Dehydro
140
235
45
120
30
2ꢀ
5
NH
NH
2
3
CF
CH CH
3
1
2ꢀ
3
6
2
2
3
1
2ꢀ
7
8
2
OH
F
OH
OH
3
F
F
1
2ꢀ
3
a
1
2ꢀ
9
Cl
F
3
19
43
CH
CH
CH
CH
CH
3
3
3
3
3
NH
F
2
1
2ꢀ
3
1
1
0
1
2
b
b
150
30
OH
Ap
5
T2
OH
OH
N
3
NH
25
26
2
H
OH
a
K
m
value.
The TMP kinase activity of these compounds was not determined
CH3
OH NHC(NH)NH2
b
7
a
usingthe spectrophotometric assay by Blondin et al.,
2
but using
K
m
value.
b
HPLC chromatography. K
m
for dTMP is 40 mM.
This hydroxyl lies at the b-side of the sugar ring.

3048
V. Vanheusden et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3045–3048
Since the efficient affinity of 1 proved that the 5-position
can accommodate intermediately sized substituents, 17
Acknowledgements
(
which has a 5-CF mimickingthe size of the 5-methyl)
3
This work was supported by grants from the FWO-
Vlaanderen, IWT, the Institut Pasteur, EEC grant BIO
CT98-0354,the Centre National de la Reserche Scienti-
fique (URA 2185 and 2128) and the Institut National de
and 18 were investigated as well. However, with Ki
values of 97 and 1140 mM, they proved to be poor inhi-
bitors of TMPKmt (Table 3).
la Sante et de la Recherche Medicale.
´ ´
In our search for high-affinity TMPKmt ligands as anti-
tuberculosis agents, most of the base- and sugar-mod-
ified dTMP-analogues (with the 5-halogeno dUMP
0
derivatives, dUMP, 3 -F-dTMP and D4T-MP as excep-
tions) proved to be inhibitors of the enzyme. However,
conversion to prodrugs is the only way to deliver phos-
phorylated compounds into cells. Since this recent work
indicates that in many cases (AZT, 5-BrdU, 22 and 23)
nucleoside analogues show affinities that approach
those of their phosphorylated counterparts, this prob-
lem could be easily circumvented by focusingon
References and Notes
1. Vanheusden, V.; Munier-Lehmann, H.; Pochet, S.; Herde-
wijn, P.; Van Calenbergh, S. Bioorg. Med. Chem. Lett. 2002,
1
2, 2695.
. Haouz, A.; Vanheusden, V.; Munier-Lehmann, H.;
2
Froeyen, M.; Herdewijn, P.; Van Calenbergh, S.; Delarue, M.
J. Biol. Chem. 2003, 278, 4963.
0
nucleosides instead of nucleotides. Concerningthe 5 -
0
azido group are good alternatives for a 5 -hydroxyl and
0
modifications we can conclude that a 5 -amino and a 5 -
3
. Pochet, S.; Dugue, L.; Douguet, D.; Labesse, G.; Munier-
Lehmann, H. Chembiochem 2002, 108.
. Anderson, E. In The Enzymes; Boyer, P. D., Ed.; Aca-
0
that further study is recommended to probe the effect of
4
0
0
0
deletion of the 5 -hydroxyl. At the 2 -position, a 2 -
hydroxyl has proven to reduce the affinity for
TMPKmt, with compound 6 as the only exception
known so far.
demic: New York, 1973; Vol. 8, p 49.
5. Munier-Lehmann, H.; Chafotte, A.; Pochet, S.; Labesse, G.
Protein Sci. 2001, 10, 1195.
6. Li de la Sierra, I.; Munier-Lehmann, H.; Gilles, A. M.;
Baˆ rzu, O.; Delarue, M. J. Mol. Biol. 2001, 311, 87.
7
Ba
. Blondin, C.; Serina, L.; Wiesmu
¨
ller, L.; Gilles, A. M.;
Future research will be directed towards findingother
high-affinity 5 -substituents that are unable to be phos-
0
phorylated by cellular kinases, thereby diminishingthe
risk for toxicity and further exploration of the 3 -posi-
ˆ
rzu, O. Anal. Biochem. 1994, 220, 219.
8
. Cole, S. T.; Brosch, R.; Parkhill, J.; Garnier, T.; Churcher,
C.; Harris, D.; Gordon, S. V.; Eiglmeier, K.; Gas, S.; Barry,
C. E., III; Tekaia, F.; Badcock, K.; Basham, D.; Brown, D.;
Chillingworth, T.; Connor, R.; Davies, R.; Devlin, K.; Felt-
well, T.; Gentles, S.; Namlin, N.; Holroyd, S.; Hornsby, T.;
Jagels, K.; Barrel, B. G. Nature 1998, 393, 537.
0
tion to improve the current inhibitors. Also, analogues
exhibitinginterestinginhibitory activity towards
`
TMPKmt will be examined for selectivity vis-a-vis the
human enzyme.
9. Saito, H.; Tomioka, H. J. Gen. Microbiol. 1984, 130, 1863.