N1-substituted thymine derivatives as mitochondrial thymidine kinase (TK-2) inhibitors

Article abstract of DOI:10.1021/jm0610550

Novel N¹-substituted thymine derivatives related to 1-[(Z)-4-(triphenylmethoxy)-2-butenyl]thymine have been synthesized and evaluated against thymidine kinase-2 (TK-2) and related nucleoside kinases [i.e., Drosophila melanogaster deoxynucleosid

Full text of DOI:10.1021/jm0610550

7766
J. Med. Chem. 2006, 49, 7766-7773
N1-Substituted Thymine Derivatives as Mitochondrial Thymidine Kinase (TK-2) Inhibitors
Ana-Isabel Herna´ndez,# Olga Familiar,# Ana Negri,§ Fa´tima Rodr´ıguez-Barrios,§ Federico Gago,§ Anna Karlsson,
Mar´ıa-Jose´ Camarasa,# Jan Balzarini,† and Mar´ıa-Jesu´s Pe´rez-Pe´rez#*
Instituto de Qu´ımica Me´dica (C.S.I.C.), Juan de la CierVa 3, E-28006 Madrid, Spain, Departamento de Farmacolog´ıa, UniVersidad de Alcala´,
E-28871 Alcala´ de Henares, Madrid, Spain, Karolinska Institute, S-141 86 Huddinge/Stockholm, Sweden, Rega Institute for Medical Research,
Katholieke UniVersiteit LeuVen, B-3000 LeuVen, Belgium
ReceiVed September 5, 2006
Novel N1-substituted thymine derivatives related to 1-[(Z)-4-(triphenylmethoxy)-2-butenyl]thymine have been
synthesized and evaluated against thymidine kinase-2 (TK-2) and related nucleoside kinases [i.e., Drosophila
melanogaster deoxynucleoside kinase (Dm-dNK) and herpes simplex virus type 1 thymidine kinase (HSV-1
TK)]. The thymine base has been tethered to a distal triphenylmethoxy moiety through a polymethylene
chain (n ) 3-8) or through a (2-ethoxy)ethyl spacer. Moreover, substitutions at position 4 of one of the
phenyl rings of the triphenylmethoxy moiety have been performed. Compounds with a hexamethylene spacer
(18, 26b, 31) displayed the highest inhibitory values against TK-2 (IC50 ) 0.3-0.5 µM). Compound 26b
competitively inhibited TK-2 with respect to thymidine and uncompetitively with respect to ATP. A rationale
for the biological data was provided by docking some representative inhibitors into a homology-based model
of human TK-2. Moreover, two of the most potent TK-2 inhibitors (18 and 26b) that also inhibit HSV-1
TK were able to reverse the cytostatic activity of 1-(â-D-arabinofuranosyl)thymine (Ara-T) and ganciclovir
in HSV-1 TK-expressing OST-TK-/HSV-1 TK+ cell cultures.
Introduction
depletion.5 Still, the mechanisms by which these nucleoside
analogues exert their mitochondrial toxicity are not fully
understood.7
Mitochondrial (mt) thymidine kinase (TK-2) belongs to the
family of mammalian deoxynucleoside kinases (dNKs) that
catalyze the phosphorylation of deoxynucleosides to their
corresponding deoxynucleoside monophosphates by γ-phos-
phoryl transfer from ATP (or other nucleoside triphosphates).1,2
Deoxynucleoside kinases are instrumental in the activation of
deoxynucleoside analogues with biological and therapeutic
properties, a large number of them being used in the treatment
of viral diseases and cancer.3 Besides the activation of nucleoside
analogues with pharmacological properties, dNKs have a
fundamental role in the salvage pathway of deoxynucleotide
synthesis. Moreover, mitochondrial DNA (mtDNA) synthesis,
which occurs rather independently of nuclear DNA synthesis,
is maintained during the whole cell cycle. In resting cells, TK-2
is the predominant, if not the exclusive, dThd-phosphorylating
enzyme, and TK-2 is suggested to play a key role in the
mitochondrial salvage pathway to provide the nucleotides for
mtDNA synthesis. Interestingly, critical point mutations in the
gene encoding TK-2 have been correlated to mtDNA disorders,
further supporting an essential role of TK-2 in mtDNA
synthesis.4,5
On the basis of its amino acid sequence, TK-2 belongs to a
large group of dNKs that includes other human enzymes such
as deoxycytidine kinase (dCK) and deoxyguanosine kinase
(dGK), together with other dNKs of different origins, such as
the multisubstrate dNK of the fruitfly, Drosophila melanogaster
(Dm-dNK), and to a lesser extent, with the thymidine kinase of
herpes simplex virus type 1 (HSV-1 TK).2 The amino acid
sequence identity between human TK-2 and Dm-dNK is about
40%, and there are no long insertions or deletions in their genes.
In 2001, the structure of Dm-dNK in complex with dCyd was
solved.8 Two other complexes of Dm-dNK with either dThd or
dTTP also have been reported.9 It is interesting to note that
although Dm-dNK shares only 10% of amino acid sequence
identity with HSV-1 TK, the core structure of these two enzymes
has a very similar overall fold. Given the sequence similarity
between TK-2 and Dm-dNK, and to a lesser extent, HSV-1 TK,
the testing of compounds has been performed in many cases
against the three enzymes. The information gained from these
experiments can also be relevant to further explore similarities
and differences among these related enzymes.
Long-term treatment with antiviral nucleoside analogues such
as zidovudine (AZT) or 2′-fluoro-5-iodo-(1-â-D-arabinofurano-
syl)uracil (FIAU) has been associated with severe mitochondrial
toxicity.6,7 Since these nucleoside analogues are phosphorylated
by TK-2, it can be speculated that their corresponding triphos-
phates may accumulate inside the mitochondria thereby com-
promising mtDNA synthesis. Several nucleoside analogues have
been linked to genetic mtDNA depletion syndromes and mtDNA
From the features described above, it can be concluded that
TK-2 is implicated in the phosphorylation of pyrimidine
nucleosides that is required for mtDNA synthesis. TK-2 could
also be involved in the mitochondrial toxicity associated with
prolonged treatment with antiviral nucleoside analogues. How-
ever, there are still many open questions regarding the real
contribution of TK-2 to these issues. Our research groups have
been involved in the identification of TK-2 inhibitors that could
become valuable tools not only to unravel the role of TK-2 in
the maintenance and homeostasis of mitochondrial dNTP pools
but also to help clarify the contribution of TK-2-catalyzed
phosphorylation of certain antiviral drugs to their mitochondrial
toxicity. Such inhibitors could also be helpful to investigate the
activity of TK-1 and TK-2 in different cell types. In 2002, we
* To whom correspondence should be addressed. Phone: 34 91 5622900
ext 411. Fax: 34 91 5644853. E-mail: mjperez@iqm.csic.es.
# Instituto de Qu´ımica Me´dica (C.S.I.C.).
§ Universidad de Alcala´.
Karolinska Institute.
† Katholieke Universiteit Leuven.
10.1021/jm0610550 CCC: $33.50 © 2006 American Chemical Society
Published on Web 12/05/2006

N1-Thymine DeriVatiVes as TK-2 Inhibitors
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 26 7767
Chart 1. Structural Formulas of Previously Described Acyclic
Nucleosides TK-2 Inhibitors
impact of these modifications on TK-2 inhibition. Compounds
bearing a less hydrophobic diethyleneglycol-based tether also
have been synthesized. Moreover, several substituents have been
incorporated at position 4 of one of the phenyl rings of the trityl
moiety to determine the role of such substitutions on TK-2
inhibition.
Chemistry
The approach followed to synthesize the tethered molecules
in which the thymine base is connected with the distal O-trityl
through different polymethylene spacers is shown in Scheme 1
and basically consisted of monotritylation on the symmetrical
diol followed by introduction of the thymine base under
Mitsunobu conditions.10 There have been different approaches
to the selective monoprotection of symmetric diols; one of the
most common procedures involves the use of a large excess of
the diol relative to the protecting reagent (typically if the diol
is inexpensive).17 Thus, reaction of an excess of the alkanediols
(4-8; n ) 3, 5-8) with trityl chloride (used as the limiting
reagent) in CH2Cl2 and in the presence of Et3N and DMAP
afforded the monotrityl alcohols 10-14 with yields that ranged
from 83 to 96%. Reaction of diethylene glycol (9) with trityl
chloride under similar conditions afforded the O-trityl derivative
15 in 50% yield. Condensation of the alcohols 10-15 with N3-
benzoylthymine18 in the presence of PS-triphenylphosphine and
diisopropyl azodicarboxylate (DIAD) in dry THF afforded the
coupling products that were immediately debenzoylated by
treatment with 1 M NaOH in a dioxane/H2O (1:1) solution. In
every case, the only condensation product detected was the
alkylated compound at N-1 of the thymine ring. Yields of the
targeted compounds (16-21) range from 60 to 74%. When
compounds 16-21 were evaluated against TK-2-catalyzed dThd
phosphorylation (see Biological Results), it was found that the
length of the tether had a major impact on enzyme inhibition;
the best results were obtained with the six-methylene compound
18, which was even more active than the prototype (Z)-2-butenyl
derivative 1. The diethylglycol derivative 21 was slightly less
active, but the less lipophilic character of this molecule (see
cLogP values in Supporting Information) prompted us to
consider it as a good candidate for further structural exploration.
Therefore, we next focused on introduction of modifications
on the trityl moiety. These modifications included replacement
of one of the phenyl rings of the trityl by either a 4-chlorophenyl
or a 4-pyridyl ring as shown in series 26-28a,b (Scheme 2).
The synthetic approach was very similar to that described in
Scheme 1 with the only exception that the corresponding trityl
chlorides had to be synthesized. (4-Chlorophenyl)diphenyl-
described the first acyclic nucleoside analogues that inhibited
TK-2, the prototype compound being 1-[(Z)-4-(triphenylmethoxy)-
2-butenyl]thymine (1) (Chart 1).10 The structure-activity rela-
tionship studies performed with a different series of analogues
of the lead compound 1 have shown the following features:11
(i) the nucleic base is crucial for the interaction of these
inhibitors with the target enzyme(s) (e.g., it has been established
that competitive inhibition by 1 is exerted with respect to
thymidine,12 suggesting that this inhibitor binds at the substrate
binding site); (ii) the spacer connecting the nucleic base and
the distal substituent has a major impact on the potency and
selectivity of the inhibitors against TK-2 and related enzymes.
Although most of the synthesized structures contain a (Z)-2-
butenyl linker, increasing the conformational freedom of the
spacer (as in compounds 2 and 3)10 is not accompanied by
augmented inhibitory potency against TK-2 [IC50 ≈ 3 µM]);
(iii) the evaluation of the effect of substituents attached at the
distal site of the compounds indicated that aromatic substituents
such as diphenylmethyl, biphenyl, and dibenzyl result in potent
TK inhibition, although the triphenylmethyl (trityl) moiety still
afforded the most pronounced inhibitory values.10,12-14 With
all this information in hand, we considered that the next
generation of compounds should keep the thymine base intact
to interact at the substrate binding site as well as an O-trityl
moiety at the distal site because this substituent has afforded
the best inhibitory values so far. On the basis of the confor-
mational flexibility of TK-2 and similar kinases,15,16 we decided
to tether the thymine base with the distal O-trityl through
polymethylene spacers of different lengths and to evaluate the
Scheme 1. Synthesis of Compounds 16-21a
a (a) TrCl, DMAP, Et3N, CH2CH2; (b) 1. N3-BzT, PS-Ph3P, DIAD, THF; 2. Dioxane/1 M NaOH (1:1).

7768 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 26
Scheme 2. Synthesis of Compounds 26-28a,ba
Herna´ndez et al.
a (a) 4-Cl-trityl chloride (a series) or 4-pyridyldiphenylmethyl chloride (b series), DMAP, Et3N, CH2CH2; (b) 1. N3-BzT, PS-Ph3P, DIAD, THF; 2.
Dioxane/1 M NaOH (1:1).
Scheme 3. Synthesis of the Pyridinium Salts 26c and 28c
carbinol was prepared by reaction of the 4-chlorobenzophenone
with phenylmagnesium bromide, following described proce-
dures,19 and was further transformed into the desired chloride
by reaction with acetyl chloride.19 Reaction of this chloride with
alcohols 6, 9, and 22, in CH2Cl2 and in the presence of Et3N
and DMAP, afforded the monosubstituted compounds (23a, 24a,
and 25a) in 63-86% yield (Scheme 2). Incorporation of the
thymine base in a similar way as described in Scheme 1,
followed by treatment with 1 N NaOH in dioxane, afforded the
N-1-substituted thymine derivatives 26a, 27a, and 28a in 55,
56, and 70% yield, respectively.
On the other hand, the commercially available diphenyl-4-
pyridylmethanol was transformed into the corresponding chlo-
ride20 and reacted with 6, 9, and 22 to afford the monosubstituted
alcohols 23b, 24b, and 25b in 84, 58 and 34% yield, respectively
(Scheme 2). Mitsunobu-type condensation with N3-benzoylthy-
mine,18 followed by basic deprotection, afforded the desired
compounds 26b, 27b, and 28b in 75, 62, and 56% yield,
respectively. Compounds 26b and 28b were further transformed
into the corresponding N-methylpyridinium salts by reaction
with methyl iodide in THF (Scheme 3) to afford compounds
26c and 28c in 73 and 47% yield, respectively.
Another modification that was considered of interest was the
introduction of a carboxamide moiety at position 4 of one of
the phenyl rings of the trityl moiety. It has been reported that
the transformation of the N-succinimidyl benzoate into the
corresponding methylbenzamide to afford the targeted com-
4-carboxy derivatives of trityls are 2-5 times more stable to
pounds 31 and 32 in 33 and 40% yield, respectively.
acidic treatment than the corresponding non-substituted ana-
logues.21 Moreover, the introduction of the carboxamide moiety
Biological Results and Discussion
at the trityl is suitable to further functionalize the compounds
through the amide moiety. On the basis of previous reports,21-23
prior to transformation of the carbinol to the chloride, the
4-(hydroxydiphenylmethyl)benzoic acid was transformed to the
corresponding N-succinimidyl benzoate. Once protected at the
carboxylic acid, the carbinol was reacted with acetyl chloride
to yield N-succinimidyl-(4-chlorodiphenylmethyl)benzoate. The
reactions between this trityl chloride and the diols 6 and 22
were performed at 0 °C, as reported in similar cases,21 to prevent
the possible formation of ester bonds between the excess of
the diol and the activated N-succinimidyl benzoate. In this way,
the monotritylated compounds 29 and 30 were obtained in 25
and 44% yield, respectively (Scheme 4). Reaction of the alcohols
29 and 30 with N3-benzoylthymine17 under Mitsunobu-type
condensation conditions, afforded the corresponding coupling
products. Further treatment with methylamine in MeOH led to
the removal of the N3-benzoyl group from thymine together with
Inhibition of [methyl-3H]dThd Phosphorylation by TK-2,
HSV-1 TK, and Dm-dNK. Compounds 16-21, 26-28(a-c),
31, and 32 were tested for their inhibitory effect against
phosphorylation of 1 µM thymidine by recombinant TK-2,
HSV-1 TK, and Dm-dNK. The results are shown in Table 1.
The lead compound 1 has also been included in the table for
comparative purposes. Evaluation of the N1-thymine-tethered-
O-trityl derivatives 16-21 clearly showed a strong relationship
between the length of the spacer and the inhibitory activity
against the tested enzymes. Short (n ) 3; 16) or long (n ) 8;
20) polymethylene linkers afforded TK-2 inhibition around 45
µM, while intermediate length linkers afforded higher inhibitory
values. In particular, a hexamethylene linker (n ) 6; 18) gave
the highest inhibitory potency with an IC50 of 0.50 µM.
Interestingly, replacement of the middle carbon by oxygen in
the five-methylene linker (21 versus 17) slightly reduced the

N1-Thymine DeriVatiVes as TK-2 Inhibitors
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 26 7769
Scheme 4. Synthesis of the Carboxamide Derivatives 31 and
on potency and selectivity against the target enzymes. In every
case, the compounds carrying a hexamethylene spacer (26a-
b, 31) afforded the best inhibitory values against TK-2, followed
by the (Z)-2-butenyl derivatives (28a,b, 32), while the (2-
ethoxy)ethyl derivatives (27a,b) were less active, as already seen
in the unsubstituted trityl series. It is also interesting to note
that the (Z)-2-butenyl derivatives (28a,b, 32) show a higher
discrimination between TK-2 and HSV-1 TK than do the
compounds with the other two spacers. The only modification
that clearly has a deleterious effect on the inhibitory potency
against the three enzymes is the introduction of the pyridinium
salt in the trityl moiety (compare compounds 26b to 26c, and
28b to 28c). To explain the much weaker potency of 26c and
28c compared to the parent pyridine derivatives 26b and 28b,
steric hindrance could not be involved on the basis of the
inhibitory potency of the 4-Cl- or 4-carboxamide substituted
derivatives. However, it might be considered that the ionic nature
of the salts 26c and 28c could be responsible for unfavorable
interactions with the enzyme. It is indeed well possible that the
trityl moiety of the TK-2 inhibitors interacts with a highly
lipophilic pocket in the enzyme that does not allow charged
(polar) entities. Alternatively, the desolvation of these charged
molecules required to efficiently interact with the enzyme, may
involve an important entropic penalty that compromises the
interaction with the target enzyme.
32a
Detailed kinetic analyses were then performed with 26b, one
of the most potent inhibitors against TK-2 described. When
tested against dThd as a variable substrate (in the presence of
saturating concentrations of ATP), 26b showed a purely
competitive inhibition of TK-2, with a Ki value of 0.29 µM,
resulting in a Ki/Km ratio of 0.32. This means that the affinity
of the TK-2 inhibitor for the enzyme is at least as good as, if
not higher than, that of the natural substrate dThd. The more
potent inhibition of TK-2 by 26b compared to that exerted by
the prototype compound 1 is also reflected in a lower Ki value
(0.29 versus 0.50 µM). When the 4-pyridyl derivative 26b was
tested against variable concentrations of ATP, which is the
cosubstrate of the TK-2-catalyzed reaction, it behaved as an
uncompetitive inhibitor of TK-2, with a Ki value of 10 µM.
Since the Km value for ATP is 19 µM, the Ki/Km ratio is 0.54.
The profile showing competitive and uncompetitive inhibition
against thymidine and ATP, respectively, is similar to that
previously observed with our prototype compound 1.12 The
uncompetitive inhibition against ATP indicates that compounds
such as 26b and 1 would only bind to TK-2 after ATP is bound
as a cosubstrate. In the light of the competitive inhibition against
dThd, it can be surmised that compounds 1 and 26b bind at the
dThd-binding site of the enzyme, but they can only do so once
ATP is already bound to the enzyme. This suggests that ATP
binding may promote a conformational change at the substrate-
binding site that allows entry and stabilization of the inhibitors.
a (a) N-Succinimidyl-(4-chlorodiphenylmethyl)benzoate, DMAP, Et3N,
CH2CH2, 0 °C; (b) 1. N3-BzT, PS-Ph3P, DIAD, THF; 2. MeNH2/MeOH,
rt.
Table 1. Inhibitory Effect of the Test Compounds on the
Phosphorylation of 1 µM [methyl-3H]dThd by TK-2, HSV-1 TK, and
Dm-dNK
IC50a (µM)
comp
TK-2
HSV-1 TK
Dm-dNK
1
16
17
18
19
20
21
26a
26b
26c
27a
27b
28a
28b
28c
31
1.5 ( 0.2
45 ( 2
2.5 ( 0.4
0.50 ( 0.01
4.7 ( 0.5
45 ( 7
9.7 ( 0.1
1.9 ( 0.7
0.47 ( 0.03
23 ( 1
5.2 ( 1.7
23 ( 0.3
2.4 ( 1.2
1.8 ( 0.1
114 ( 71
0.39 ( 0.03
1.9 ( 0.2
45 ( 1
46 ( 5
32 ( 1
3.7 ( 0.5
g50
3.3 ( 0.9
21 ( 6
15 ( 2
17 ( 10
32 ( 8
30 ( 14
56 ( 21
47 ( 1
2.7 ( 0.2
29 ( 1
45 ( 6
35 ( 2
5.7 ( 2.4
3.4 ( 0.1
144 ( 58
3.5 ( 0.1
1.6 ( 0.7
>500
17 ( 4
1.0 ( 0.9
2.0 ( 0.4
41 ( 6
2.7 ( 0.0
21 ( 0
18 ( 7
25 ( 2
>500
0.7 ( 0.4
42 ( 4
32
a 50% inhibitory concentration required to inhibit enzyme-catalyzed dThd
phosphorylation by 50%.
It may be worth mentioning that the most active TK-2
inhibitors revealed in this study are not inhibitory to cytosolic
TK-1 (data not shown). These findings suggest the lack of an
appropriate binding pocket in TK-1, and thus stress the
selectivity of the described TK-2 inhibitors in this study.
inhibitory potency against TK-2. It is worth mentioning that
the length of the linker has a major impact on inhibition of
HSV-1 TK. Whereas compound 18 (n ) 6) gave also the best
inhibitory value against HSV-1 TK (IC50 ) 3.7 µM), longer
linkers resulted in annihilation of activity (i.e., compound 20).
In contrast, the length of the linker has a much less important
impact on inhibition of Dm-dNK.
Attention was next paid to the role of substituents in one of
the phenyl rings of the triphenyl moiety, and this effect was
studied in molecules with three different spacers: a hexameth-
ylene, a (2-ethoxy)ethyl, and a (Z)-2-butenyl, the latter being
present in our lead compound 1. These series of compounds
were very helpful to confirm that the linker impinges greatly
Molecular Modeling. Since there are no crystal structures
currently available for TK-2, we endeavored to build a reliable
molecular model of this enzyme that could be used to shed light
on the experimentally obtained data. The mGenTHREADER
profile-based fold recognition method24 identified several puta-
tive structural neighbors of TK-2, most notably the human
enzymes deoxycytidine kinase (dCK) and deoxyguanosine
kinase (dGK), and the fruitfly, D. melanogaster deoxyribo-

7770 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 26
Herna´ndez et al.
Figure 2. AutoDock best-scoring poses for inhibitors 1 (C atoms in
cyan) and 18 (C atoms in pink) docked into the proposed binding site
of TK-2, which is displayed as a ribbon in olive green. Note the almost
perpendicular disposition of these inhibitors relative to ATP, and the
interactions of the trityl group with the Tyr residues just below or above
the plane delineated by these aromatic residues.
Figure 1. Schematic ribbon representation of modeled TK-2 following
deformation along the first non-translational, non-rotational normal
mode.37 The “cryptic” tunnel exposed at the interhelical region that is
proposed to be the binding site for the inhibitors’ linkers is depicted as
a semitransparent yellow surface and is seen to connect the thymidine-
binding site to the bulk solvent. Bound ATP-Mg2+ is shown for
reference only as these ligands were not explicitly taken into consid-
eration in the normal modes calculation. For greater clarity, the side
chains of Arg159 (stacking over the adenine ring of ATP), Gln77,
Tyr66, and Tyr175 are displayed as sticks. Note the varying width of
this tunnel which is narrowest at the plane delineated by these Tyr
residues.
binding site and traverses a tunnel made up of the hydrophobic
side chains of Ile26, Leu62, Met65, Tyr66, Leu76, Leu114,
Ile171, and Tyr175 (Figure 2). Depending on the length of the
spacer, the distal triaryl moiety of 1, 18, and 26b will be
positioned in a plane below (n ) 4) or above (n ) 6) the
aromatic rings of Tyr66 and Tyr175, thus allowing these rings
to establish favorable stacking and edge-to-face interactions. If
the spacer is shorter, these interactions cannot take place, and
the inhibitor is strained in the interhelical region. This results
in the decreased potency observed in going from 18 to 17 and
16. Increasing the length of the spacer results in a worsening
of these interactions and increased exposure to the solvent, both
of which also have a detrimental effect on the inhibitory activity,
as in going from 18 to 19 and 20. The model also accounts for
the experimental findings that the introduction of substituents
such as a Cl (26a, 27a, 28a) or a methylcarboxamide in the
para position of one of the phenyl rings of the triphenylmethoxy
moiety does not affect the potency because this group is facing
the solvent and therefore does not contribute to the affinity. On
the other hand, the loss of potency in 26c and 28c relative to
26b and 28b can be attributed to the unfavorable effect on the
electrostatic edge-to-face interaction of replacing the pyridine
ring with a methylpyridinium cation. With respect to the loss
in activity brought about by the replacement of a methylene
unit (26) by an ether oxygen (27), the model indicates that the
region lodging the spacer is found in the hydrophobic environ-
ment provided by Ile26 and Ile179, hence the deleterious effect
of having the polar oxygen in this position. The fact that the
longer alkyl spacers found in 19 and 20 lead to a highly
significant loss of activity against HSV-1 TK can also find an
explanation from the multiple sequence and structural align-
ment: this enzyme shows indeed a longer R3 helix that places
the N-terminus of helix R4 as a lid over the putative binding
site, thus hindering lodging of the trityl group.
nucleotide kinase (Dm-dNK), with percentages of sequence
identity of 29.3, 17.7, and 47.7%, respectively, over about 220
amino acids (Supporting Information, Figure S1). All of these
enzymes belong to the same homologous superfamily of P-loop-
containing nucleotide triphosphate hydrolases characterized
by a three-layer (RâR) sandwich architecture and a Rossmann
fold.
Examination of the normal modes for these enzymes revealed
interesting motions that coupled closure of the P-loop in the
ATP-binding site with the packing of R-helices 3, 4, 6, and 7
(data not shown). We thought that these conformational changes
could be of relevance for our studies because the kinetic analysis
data indicated that the hexamethylene derivative 26b binds at
the dThd-binding site of TK-2 only when ATP is already bound,
and also because we had previously proposed a model for
binding of 1 in which the thymine base is lodged in the dThd-
binding site and the triphenylmethoxy substituent is placed in
an interhelical hydrophobic region.14 In fact, the simulated
motions in TK-2 (Supporting Information, Figure S2) and further
exploration with CASTp25 and CAVER26 servers support the
view that a rather hydrophobic tunnel connecting the thymidine-
binding pocket with the solvent can be formed at the interface
between R-helices 3, 5, and 8, particularly when alternative
locations for the side chains of Tyr66 and Tyr175 are found
(Figure 1). Furthermore, when this region was explored with
the automated docking program AutoDock, the best-scoring
solution for 1, 18, and 26b indeed placed the thymine moiety
inside the dThd-binding cavity with the planar heteroaromatic
ring sandwiched between the phenyl ring of Phe110 on one side
and the side chains of both Trp53 and Val80 on the other side,
whereas O4 and N3 establish good hydrogen bonds with the
carboxamide group of Gln77 (Figure 2). The spacer attached
to the thymine exits this cavity at right angles from the ATP-
Evaluation of Compounds 18 and 26b in the Intact
OST-TK-/HSV-1 TK+ cells. The low levels of TK-2 compared
to the abundant cytosolic TK-1, together with its mitochondrial
localization, have been proposed as two major reasons that

N1-Thymine DeriVatiVes as TK-2 Inhibitors
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 26 7771
Conclusions
Table 2. Effect of Compounds 18 and 26b on the Cytostatic Activity of
Ara-T and GCV in OST-TK-/HSV-1 TK+ Cell Lines
On the basis of our previous results on the inhibitory activity
of 1 against TK-2-catalyzed dThd phosphorylation, a variety
of novel N1-substituted thymine derivatives were synthesized
and investigated as inhibitors of TK-2 and related enzymes
(Dm-dNK and HSV-1 TK). The length of the spacer connecting
the thymine base and the distal triphenylmethoxy moiety was
also varied, having an important impact on inhibitory activity.
Moreover, modifications were performed on one of the phenyl
rings of the trityl group including incorporation of a 4-Cl or
4-CONHCH3 substituent or replacement of one of the phenyl
rings by 4-pyridyl or a 4-pyridinium salt. Thus, by modifying
the linker and the substitution on the trityl moiety, a 4-fold gain
in potency against TK-2-catalyzed dThd phosphorylation was
achieved relative to our previous lead compound 1.
Compound 26b has been shown to be a competitive inhibitor
against TK-2 when dThd is used as a variable substrate (Ki )
0.29 µM) and an uncompetitive inhibitor when tested against
variable concentrations of ATP. The molecular model we have
built accounts for this fact as the thymine ring of the inhibitors
is proposed to occupy the thymidine-binding site in a standard
orientation, whereas the spacer would extend toward the surface
through a hydrophobic channel that is not present in the resting
state but whose existence is made apparent when an elastic
network model of the enzyme is considered. This finding
strongly suggests that the methodology employed in this work
could be useful for other targets in which “cryptic” binding sites
are exposed following a local rearrangement of the protein
backbone.29
IC50a (µM)
inhibitor (conc)
Ara-T
GCV
none
18 (10 µM)
(5 µM)
0.0062 ( 0.0015
0.15 ( 0.01
0.10 ( 0.02
0.055 ( 0.002
0.37 ( 0.00
0.24 ( 0.03
0.12 ( 0.02
0.0019 ( 0.000
0.041 ( 0.013
0.033 ( 0.000
0.018 ( 0.002
0.11 ( 0.01
(2.5 µM)
26 (10 µM)
(5 µM)
0.081 ( 0.013
0.050 ( 0.005
(2.5 µM)
a 50% inhibitory concentration required to inhibit cell proliferation by
50%. Data are the mean ((SD) of at least two to three independent
experiments.
explain why direct evaluation of TK-2 inhibitors in cell culture
is difficult. The expression of TK-2 in the cytosol of tumor cell
lines by a TK-2 gene construct has been unsuccessful so far.
However, based on the existing similarities between TK-2 and
HSV-1 TK, and since some of the TK-2 inhibitors described
herein also significantly inhibit HSV-1 TK, the effect of two
of the most potent inhibitors (18 and 26b) was evaluated in
intact HSV-1 TK-expressing OST-TK-/HSV-1 TK+ cell cul-
tures in combination with compounds that are good substrates
for HSV-1 TK, including the pyrimidine nucleoside 1-(â-D-
arabinofuranosyl)thymine (Ara-T) and the purine derivative
ganciclovir (GCV) (Table 2). The OST-TK-/HSV-1 TK+ cells
represent human osteosarcoma cells that are deficient in
cytosolic TK-1 but express HSV-1 TK in the cytosol.27 It is
well-established that when the substrates Ara-T and GCV are
added to OST-TK-/HSV-1 TK+ cell cultures, a marked cyto-
toxicity is observed with IC50 values in the low nanomolar range
(0.0062-0.0019 µM). Addition of the HSV-1 TK inhibitor 18
at a concentration of 10 µM to these cultures markedly reduced
the cytotoxic effect of Ara-T and GCV by 24- and 22-fold,
respectively. This effect is clearly dose-dependent since addition
of lower concentrations of the inhibitor 18 (5 and 2.5 µM) has,
as expected, a less pronounced effect on the IC50 values of Ara-T
and GCV. Interestingly, addition of compound 26b to the cell
cultures at a concentration of 10 µM has a more pronounced
effect on the cytotoxicities of Ara-T and GCV, which were
reduced by 60- and 58-fold, respectively, compared to the
untreated cell cultures. This “detoxifying” effect of compound
26b is also dose-dependent, as shown in Table 2. The more
pronounced effect of compound 26b compared to that of 18 in
cell culture could not be ascribed to significant differences in
the inhibition of the target enzyme HSV-1 TK as shown in the
enzymatic assay (IC50 values against HSV-1 TK of 2.0 and 3.7
µM, respectively). Therefore, it could be suggested that the
presence of the pyridine ring in the triaryl substituent in 26b
slightly reduces the LogP value of these lipophilic molecules
compared to the triphenyl derivative 18, resulting in a better
behavior (solubility and/or uptake) in the cells. Calculated LogP
values of these and other compounds in this paper are included
as Supporting information. Replacement of a phenyl by a pyridyl
is a well-established approach in bioisosterism, and recent
examples can be found in the literature.28
Among the very few molecules reported in the literature that
inhibit TK-2-catalyzed dThd phosphorylation, compounds 18,
26b, and 31 represent three of the most potent inhibitors
described so far. Moreover, compound 26b has been recently
shown to be a useful tool to evaluate the specific TK-2 activity
in human fibroblasts and has been instrumental to set up a new
assay to measure TK-2 expression in cultured cells and for the
determination of TK-2 activity in patient samples.30 Finally, a
TK-2 inhibitor may be useful to investigate the role of TK-2 in
mitochondrial homeostasis in general and in the regulation of
the mitochondrial deoxynucleotide pools in particular.
Experimental Section
Chemical Procedures. Melting points were obtained on a
Reichert-Jung Kofler apparatus and are uncorrected. Microanalyses
were obtained with a Heraeus CHN-O-RAPID instrument. Elec-
trospray mass spectra were measured on a quadrupole mass
spectrometer equipped with an electrospray source (Hewlett-
Packard, LC/MS HP 1100). 1H and 13C NMR spectra were recorded
on a Varian Gemini operating at 200 MHz (1H) and 50 MHz (13C),
respectively, on a Varian INNOVA 300 operating at 299 MHz (1H)
and 75 MHz (13C), respectively, and a Varian INNOVA-400
operating at 399 MHz (1H) and 99 MHz (13C), respectively.
Analytical TLC was performed on silica gel 60 F254 (Merck)
precoated plates (0.2 mm). Spots were detected under UV light
(254 nm) and/or by charring with phosphomolibdic acid. Separa-
tions on silica gel were performed by preparative centrifugal circular
thin layer chromatography (CCTLC) on a ChromatotronR (Kiesegel
60 PF254 gipshaltig (Merck)), layer thickness (1 or 2 mm), flow
rate (4 or 8 mL/min, respectively). Flash column chromatography
was performed with silica gel 60 (230-400 mesh) (Merck).
The dose-dependent reversal of the AraT- and GCV-related
cellular toxicity by the TK-2 (and HSV-1 TK) inhibitors thus
strongly suggest that these compounds efficiently penetrate into
the intact tumor cells and are able to reach their target enzyme
in the intracellular environment when present in the cytosol.
Experiments are underway to reveal whether the test compounds
are also able to cross the mitochondrial double membrane in
the intact cells.
Polysterene-triphenylphosphine (PS-Ph3P) was purchased from
Fluka. DIAD was purchased from Aldrich.
All experiments involving water-sensitive compounds were
conducted under scrupulously dry conditions. Triethylamine and
acetonitrile were dried by refluxing over calcium hydride. Tetrahy-

7772 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 26
Herna´ndez et al.
drofuran was dried by refluxing over sodium/benzophenone.
Anhydrous N,N′-dimethylformamide was purchased from Aldrich.
General Procedure for Monotritylation of Diols. To a stirred
solution of the corresponding diol (10 mmol) in dry CH2Cl2 (4 mL)
at 0 °C, Et3N (0.15 mL, 1.1 mmol), DMAP (5 mg, 0.04 mmol)
and trityl chloride (278 mg, 1.0 mmol) were added. The mixture
was stirred at rt for 24 h. Then, it was diluted with CH2Cl2 (20
mL) and water (10 mL). The organic phase was washed with water
(10 mL), and brine (10 mL). (The excess diol was removed in these
washing steps.) The organic layer was dried on anhydrous Na2-
SO4, filtered, and evaporated. The residue was purified by flash
column chromatography.
General Procedure for the Mitsunobu Condensation of
Alcohols with N3-Benzoylthymine. To a suspension containing
the corresponding alcohol (10-15, 23-25a,b) (0.5 mmol), PS-
Ph3P (3 mmol/g, 416 mg, 1.25 mmol), and N3-benzoylthymine (230
mg, 1.0 mmol) in dry THF (5 mL), a solution of DIAD (0.19 mL,
1.0 mmol) in dry THF (2 mL) was slowly added. The mixture was
stirred at rt overnight. The reaction was filtered, the residue was
washed with THF (2 × 5 mL), and the combined filtrates were
evaporated to dryness. The residue was dissolved in a dioxane/
NaOH 1M (1:1) mixture (9 mL) and stirred overnight. Then, EtOAc
(10 mL) and brine (5 mL) were added. The aqueous phase was
further extracted with EtOAc (3 × 10 mL). The combined organic
extracts were dried on anhydrous Na2SO4, filtered, and evaporated.
The residue was purified in the chromatotron CCTLC (hexane-
EtOAc, 1:1) to yield the target compounds.
dGK (PDB code 1jag) while retaining the connecting loop (204-
210) conformation found in dCK (PDB code 1p60). The loops
comprising amino acids 43-44 and 192-193 were modeled by
satisfaction of spatial restraints using the ModLoop server.34 Amino
acid replacement and rotamer selection were performed using the
built-in facility available in PyMOL.35 ATP-Mg2+ was incorporated
into the structure by adding the gamma phosphate to the ADP-
Mg2+ already present in the structure of human dCK so as to
reproduce the conformation of the nonhydrolyzable analogue of
ATP found in human thymidylate kinase (PDB code 1e2q).
Normal-Mode Analyses and Cavity Calculations. To probe
the flexibility of this family of enzymes, an elastic network model
was used in which all non-hydrogen protein atoms (within a cutoff
of 10 Å) were modeled as point masses and CR atoms were
connected by springs representing the interatomic force fields.36
Each protein was then analyzed as a large set of coupled harmonic
oscillators using the NOMAD-REF server.37 The conformational
changes most likely involved in ATP and inhibitor binding were
deduced by calculating the 10 lowest-frequency normal modes,
which are those with the highest amplitudes and those most often
related to large-scale structural rearrangements in proteins. Each
mode was explored in its two opposite directions, thus resulting in
two structures (“open” and “closed”) different from the initial one
within an rmsd value of 2 Å. The areas and volumes of pockets
located on the protein surface and voids buried in the protein interior
were calculated analytically by means of the CASTp server,25
whereas the CAVER server26 was employed to delineate the shape
of the tunnel connecting the dThd-binding site with the bulk solvent.
Docking of Representative Inhibitors. Compounds 1, 18, and
26b were used as ligands for the automated docking experiments.
A variety of more “open” TK-2 conformations was selected from
the normal modes calculations, and alternative positions for the
side chains of Tyr66, Gln77 and Tyr175 were found using the built-
in rotamer library implemented in PyMOL. The magnesium ion
was assigned a charge of +2, and AMBER-compatible RESP point
charges were used for the inhibitors and ATP4-, as reported
previously.14 The Lamarckian genetic algorithm38 implemented in
AutoDock 3.0.539 was used to generate docked conformations of
each inhibitor within the putative binding cavity by randomly
changing the overall orientation of the molecule as well as the
torsion angles of the different spacers. Default settings were used
except for number of runs, population size, and maximum number
of energy evaluations, which were fixed at 100, 100, and 250 000,
respectively. Rapid intra- and intermolecular energy evaluation of
each configuration was achieved by having the receptor’s atomic
affinity potentials for aliphatic and aromatic carbon, oxygen,
nitrogen, and hydrogen atoms precalculated in a three-dimensional
grid with a spacing of 0.375 Å. A distance-dependent dielectric
function was used in the computation of electrostatic interactions.
Data for compounds 10-32 are provided as Supporting Informa-
tion.
TK Assay Using [methyl-3H]dThd as the Substrate. The
activity of recombinant TK-2, HSV-1 TK, and Dm-dNK, and the
50% inhibitory concentration of test compounds were assayed in a
50-µL reaction mixture containing 50 mM Tris/HCl, pH 8.0, 2.5
mM MgCl2, 10 mM dithiothreitol, 0.5 mM CHAPS, 3 mg/mL
bovine serum albumin, 2.5 mM ATP, and 1 µM [methyl-3H]dThd
and enzyme. The samples were incubated at 37 °C for 30 min in
the presence or absence of different concentrations (5-fold dilutions)
of the test compounds. The reaction proceeded linearly during this
time period. After the incubation period, aliquots of 45 µL of the
reaction mixtures were spotted on Whatman DE-81 filter paper
disks. The filters were washed three times for 5 min each in 1 mM
ammonium formate, once for 1 min in water, and once for 5 min
in ethanol. The radioactivity was determined by scintillation
counting.
The Km values (for dThd and ATP) and the Ki values for the
inhibitor 26b using varying concentrations of dThd (ranging
between 0.4 and 5 µM) at saturating concentrations of ATP (2.5
mM) or using varying concentrations of ATP (ranging between 5
and 250 µM) at saturating concentrations of dThd (20 µM) were
determined and derived from Lineweaver-Burk plots.
Acknowledgment. A.I.H. and O.F. thank the Spanish
Ministerio de Educacio´n y Ciencia for a predoctoral fellowship.
This work has been supported by grants from the Spanish MEC
(SAF2003-07219-C02 and SAF 2000-0153-C02) (to M.J.C,
M.J.P.P., and F.G.), the EU (QLRT-2001-01004) (to M.J.C,
M.J.P.P., A.K and J.B.), and the National Foundation for Cancer
Research (to F.G.). We thank Mrs. Lizette van Berckelaer for
excellent technical assistance.
Computational Methods. Comparative Modeling of Human
TK-2. The amino acid sequence of human mitochondrial thymidine
kinase (KITM_HUMAN) was retrieved from Swiss-Prot (http://
us.expasy.org/sprot/). The mGenTHREADER profile-based fold
recognition method24 was used to predict the secondary structure
of human TK-2 and to identify putative structural neighbors, which
were found to be in agreement with the FSSP/Dali classification.31
The multiple structural alignment of the eight proteins with the
highest Dali scores (human deoxycytidine kinase [1p60], human
deoxyguanosine kinase [1jag], Drosophila melanogaster deoxyri-
bonucleotide kinase [1oe0], herpes simplex virus 1 thymidine kinase
[1p7c], equine herpes virus thymidine kinase [1p6x], yeast thymidy-
late kinase [3tmk], Escherichia coli thymidylate kinase [4tmk], and
human thymidylate kinase [1e2f]) was performed using the Mam-
moth server32 (Supporting Information, Figure S1). Because of the
greater sequence identity, Dm-dNK was used as the template for
homology modeling of TK-2 except for the predicted C-terminal
R-helix (R9) whose counterpart is missing in this crystal structure.
This R-helix was taken from the netrin-like domain of human
procollagen C-proteinase enhancer (PCOLCE) protein (PDB code
1uap)33 and superimposed on the equivalent helix found in human
Supporting Information Available: Synthesis and data for
compounds 10-15 and 16-32. Elemental analysis of compounds
16-21, 26-28a-c, 31, and 32. Calculated LogP values for
compounds 1, 18, 21, 26b, 27b, and 28b. Multiple sequence
alignment based on structural superimposition and profile-profile
comparison of human mitochondrial TK-2 and related kinases used
in the modeling work. This material is available free of charge via
the Internet at http://pubs.acs.org.
References
(1) Arner, E. S. J.; Eriksson, S. Mammalian deoxyribonucleoside kinases.
Pharmacol. Ther. 1995, 67, 155-186.

N1-Thymine DeriVatiVes as TK-2 Inhibitors
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 26 7773
(2) Eriksson, S.; Munch-Petersen, B.; Johansson, K.; Eklund, H. Structure
and function of cellular deoxyribonucleoside kinases. Cell Mol. Life
Sci. 2002, 59, 1327-1346.
(21) Shchepinov, M. S.; Chalk, R.; Southern, E. M. Trityl tags for encoding
in combinatorial synthesis. Tetrahedron 2000, 56, 2713-2724.
(22) Gildea, B. D.; Coull, J. M.; Ko¨ster, H. A versatile acid-labile linker
for modification of synthetic biomolecules. Tetrahedron Lett. 1990,
31, 7095-7098.
(23) Coull, J. M.; Gildea, B. D.; Ko¨ster, H. U.S. Patent 5,410,068, 1995.
(24) McGuffin, L. J.; Smith, R. T.; Bryson, K.; Sorensen, S. A. J., D. T.
High throughput profile-profile based fold recognition for the entire
Human proteome. BMC Bioinformatics 2006, 7, 288.
(25) Dundas, J.; Ouyang, Z.; Tseng, J.; Binkowski, A.; Turpaz, Y.; Liang,
J. CASTp: computed atlas of surface topography of proteins with
structural and topographical mapping of functionally annotated
residues. Nucleic Acid Res. 2006, 34 (Web Server issue): W116-
118. URL: http://cast.engr.uic.edu.
(26) Petrek, M.; Otyepka, M.; Banas, P.; Kosinova, P.; Koca, J.;
Damborsky, J. CAVER: a new tool to explore routes from protein
clefts, pockets and cavities. BMC Bioinformatics 2006, 7, 316.
URL: http://loschmidt.chemi.muni.cz/caver.
(3) Van Rompay, A. R.; Johansson, M.; Karlsson, A. Substrate specificity
and phosphorylation of antiviral and anticancer nucleoside analogues
by human deoxyribonucleoside kinases and ribonucleoside kinases.
Pharmacol. Ther. 2003, 100, 119-139.
(4) Saada, A.; Shaag, A.; Mandel, H.; Nevo, Y.; Eriksson, S.; Elpeleg,
O. Mutant mitochondrial thymidine kinase in mitochondrial DNA
depletion myopathy. Nat. Genet. 2001, 29, 342-344.
(5) Saada, A. Deoxyribonucleotides and disorders of mitochondrial DNA
integrity. DNA Cell Biol. 2004, 23, 797-806.
(6) Lewis, W.; Dalakas, M. C. Mitochondrial toxicity of antiviral drugs.
Nat. Med. 1995, 1, 417-422.
(7) Lewis, W.; Day, B. J.; Copeland, W. C. Mitochondrial toxicity of
NRTI antiviral drugs: An integrated cellular perspective. Nat. ReV.
Drug DiscoVery 2003, 2, 812-822.
(8) Johansson, K.; Ramaswamy, S.; Ljungcrantz, C.; Knecht, W.; Piskur,
J.; Munch-Petersen, B.; Eriksson, S.; Eklund, H. Structural basis for
substrate specificities of cellular deoxyribonucleoside kinases. Nat.
Struct. Biol. 2001, 8, 616-620.
(27) Degre`ve, B.; Johansson, M.; De Clercq, E.; Karlsson, A.; Balzarini,
J. Differential intracellular compartmentalization of herpetic thymi-
dine kinases (TKs) in TK gene-transfected tumor cells. Molecular
characterization of the nuclear localization signal of herpes simplex
virus type 1 TK. J. Virol. 1998, 72, 9535-9543.
(9) Mikkelsen, N. E.; Johansson, K.; Karlsson, A.; Knecht, W.; Andersen,
G.; Piskur, J.; Munch-Petersen, B.; Eklund, H. Structural basis for
feedback inhibition of the deoxyribonucleoside salvage pathway:
Studies of the Drosophila deoxyribonucleoside kinase. Biochemistry
2003, 42, 5706-5712.
(10) Herna´ndez, A. I.; Balzarini, J.; Karlsson, A.; Camarasa, M. J.; Pe´rez-
Pe´rez, M. J. Acyclic nucleoside analogues as novel inhibitors of
human mitochondrial thymidine kinase. J. Med. Chem. 2002, 45,
4254-4263.
(11) Pe´rez-Pe´rez, M. J.; Herna´ndez, A. I.; Priego, E. M.; Rodr´ıguez-
Barrios, F.; Gago, F.; Camarasa, M. J.; Balzarini, J. Mitochondrial
thymidine kinase inhibitors. Curr. Top. Med. Chem. 2005, 5, 1205-
1219.
(12) Balzarini, J.; Herna´ndez, A. I.; Roche, P.; Esnouf, R.; Karlsson, A.;
Camarasa, M. J.; Pe´rez-Pe´rez, M. J. Non-nucleoside inhibitors of
mitochondrial thymidine kinase (TK-2) differentially inhibit the
closely related herpes simplex virus type 1 TK and Drosophila
melanogaster multifunctional deoxynucleoside kinase. Mol. Phar-
macol. 2003, 63, 263-270.
(13) Priego, E. M.; Balzarini, J.; Karlsson, A.; Camarasa, M. J.; Pe´rez-
Pe´rez, M. J. Synthesis and evaluation of thymine-derived carboxa-
mides against mitochondrial thymidine kinase (TK-2) and related
enzymes. Bioorg. Med. Chem. 2004, 12, 5079-5090.
(14) Herna´ndez, A. I.; Balzarini, J.; Rodr´ıguez-Barrios, F.; San-Fe´lix, A.;
Karlsson, A.; Gago, F.; Camarasa, M. J.; Pe´rez-Pe´rez, M. J.
Improving the selectivity of acyclic nucleoside analogues as inhibitors
of human mitochondrial thymidine kinase: Replacement of a
triphenylmethoxy moiety with substituted amines and carboxamides.
Bioorg. Med. Chem. Lett. 2003, 13, 3027-3030.
(15) Bissantz, C.; Folkers, G.; Rognan, D. Protein-based virtual screening
of chemical databases. 1. Evaluation of different docking/scoring
combinations. J. Med. Chem. 2000, 43, 4759-4767.
(28) Pastorin, G.; Da Ros, T.; Bolcato, C.; Montopoli, C.; Moro, S.;
Cacciari, B.; Baraldi, P. G.; Varani, K.; Borea, P. A.; Spalluto, G.
Synthesis and biological studies of a new series of 5-heteroarylcar-
bamoylaminopyrazolo 4,3-e 1,2,4-triazolo 1,5-c pyrimidine s as
human A(3) adenosine receptor antagonists. Influence of the het-
eroaryl substituent on binding affinity and molecular modeling
investigations. J. Med. Chem. 2006, 49, 1720-1729.
(29) Teague, S. J. Implications of protein flexibility for drug discovery.
Nat. ReV. Drug DiscoVery 2003, 2, 527-541.
(30) Franzolin, E.; Rampazzo, C.; Pe´rez-Pe´rez, M. J.; Herna´ndez, A. I.;
Balzarini, J.; Bianchi, V. Bromovinyl-deoxyuridine: A selective
substrate for mitochondrial thymidine kinase in cell extracts. Biochem.
Biophys. Res. Commun. 2006, 344, 30-36.
(31) Holm, L.; Sander, C. Dali/FSSP classification of three-dimensional
protein folds. Nucleic Acids Res. 1997, 25, 231-234.
(32) Lupyan, D.; Leo-Macias, A.; Ortiz, A. R. A new progressive-iterative
algorithm for multiple structure alignment. Bioinformatics 2005, 21,
3255-3263. URL: http://ub.cbm.uam.es/mammoth/mult/.
(33) Liepinsh, E.; Banyai, L.; Pintacuda, G.; Trexler, M.; Patthy, L.; Otting,
G. NMR structure of the netrin-like domain (NTR) of human type I
procollagen C-proteinase enhancer defines structural consensus of
NTR domains and assesses potential proteinase inhibitory activity
and ligand binding. J. Biol. Chem. 2003, 278, 25982-25989.
(34) Fiser, A.; Sali, A. ModLoop: automated modeling of loops in protein
structures. Bioinformatics 2003, 19, 2500-2501. URL: http://
alto.compbio.ucsf.edu/modloop//modloop.html.
(35) DeLano, W. PyMOL version 0.99, DeLano Scientific LLC, South
San Francisco, California. URL: http://pymol.sourceforge.net/.
(36) Lindahl, E.; Delarue, M. Refinement of docked protein-ligand and
protein-DNA structures using low frequency normal mode amplitude
optimization. Nucleic Acids Res. 2005, 33, 4496-4506.
(37) Lindahl, E.; Azuara, C.; Koehl, P.; Delarue, M. NOMAD-Ref:
visualization, deformation and refinement of macromolecular struc-
tures based on all-atom Normal Mode Analysis. Nucleic Acid Res.
2006, 34, (Web Server issue): W52-56. URL: http://lorentz.immstr-
.pasteur.fr/nomad-ref.php.
(16) Barroso, J.; Carvalho, R.; Flatmark, T. Kinetic analysis and ligand-
induced conformational changes in dimeric and tetrameric forms of
human thymidine kinase 2. Biochemistry 2005, 44, 4886-4996.
(17) Komiotis, D.; Currie, B. L.; LeBreton, G. C.; Venton, D. L.
Preparation of monotrytilated symmetric 1,n-diols. Synth. Commun.
1993, 23, 531-534.
(18) Cruikshank, K. A.; Jiricny, J.; Reese, C. B. The benzoylation of uracil
and thymine. Tetrahedron Lett. 1984, 25, 681-684.
(19) Gibbs, L. W.; Wedegaertner, D. K. A kinetic-study of the rearrange-
ment of triarylacetonitrile oxides. J. Org. Chem. 1991, 56, 7320-
7322.
(38) Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.; Hart, W.
E.; Belew, R. K.; Olson, A. J. Automated docking using a Lamarckian
genetic algorithm and an empirical binding free energy function. J.
Comput. Chem. 1998, 19, 1639-1662.
(20) Coyle, S.; Hallett, A.; Munns, M. S.; Young, G. T. Amino-acids and
peptides 45. The protection of the thiol function of cysteine and the
imidazole-N of histidine by the diphenyl-4-pyridylmethyl group. J.
Chem. Soc. Perkin Trans. 1 1981, 522-528.
(39) http://www.scripps.edu/pub/olson-web/doc/autodock.
JM0610550