Báo cáo khoa học: Structure of the substrate complex of thymidine kinase from Ureaplasma urealyticum and investigations of possible drug targets for the enzyme pdf

Structure of the substrate complex of thymidine kinase
from Ureaplasma urealyticum and investigations of
possible drug targets for the enzyme
Urszula Kosinska
1
*, Cecilia Carnrot
2
*, Staffan Eriksson
2
, Liya Wang
2
and Hans Eklund
1
1 Department of Molecular Biology, Swedish University of Agricultural Sciences, Uppsala Biomedical Centre, Sweden
2 Department of Molecular Biosciences, Swedish University of Agricultural Sciences, Uppsala Biomedical Centre, Sweden
Two potential nucleoside kinase genes coding for a
thymidine kinase (TK) (EC 2.7.1.21) and a deoxyadeno-
sine kinase (EC 2.7.1.74) are found in all sequenced
Mollicute genomes [1–9]. Deoxyadenosine kinase from
Mycoplasma mycoides ssp. mycoides and TK from
Ureaplasma urealyticum (Uu-TK) have previously been
cloned and characterized [10,11]. U. urealyticum (also
called Ureaplasma parvum) is a human pathogen colon-
izing the urogenital tract and it is associated with
several pregnancy complications, e.g. infertility, altered
sperm motility, chorioamnionitis and pneumonia in the
neonate [12].
Bacterial TKs show moderate sequence identity
with human TK1 (hTK1) and mollicute TKs, e.g.
Uu-TK shares 29% sequence identity with hTK1. The
initial characterization demonstrated that Uu-TK also

has similar enzyme kinetic properties to hTK1, with
specificity for pyrimidine deoxynucleosides and with
dTTP serving as a feedback inhibitor [11,13]. Uu-TK is
less fastidious than hTK1 with regard to phosphate
donors, using all nucleoside triphosphates with similar
efficiency [11,13]. No genes encoding the enzymes for
the de novo pathway of deoxynucleotide biosynthesis
have been found in U. urealyticum, strongly suggesting
Keywords:
bacterial; crystallography; deoxythymidine;
nucleoside analogues; thymidine kinase
Correspondence
H. Eklund, Department of Molecular
Biology, Swedish University of Agricultural
Sciences, PO Box 590, Biomedical Centre,
S-751 24 Uppsala, Sweden
Fax: +46 18 53 69 71
Tel: +46 18 4754559
E-mail:
*Note
These authors contributed equally to this
work.
(Received 22 August 2005, revised 14
October 2005, accepted 21 October 2005)
doi:10.1111/j.1742-4658.2005.05030.x
Thymidine kinases have been found in most organisms, from viruses and
bacteria to mammals. Ureaplasma urealyticum (parvum), which belongs to
the class of cell-wall-lacking Mollicutes, has no de novo synthesis of DNA
precursors and therefore has to rely on the salvage pathway. Thus, thymi-
dine kinase (Uu-TK) is the key enzyme in dTTP synthesis. Recently the 3D

structure of Uu-TK was determined in a feedback inhibitor complex, dem-
onstrating that a lasso-like loop binds the thymidine moiety of the feed-
back inhibitor by hydrogen bonding to main-chain atoms. Here the
structure with the substrate deoxythymidine is presented. The substrate
binds similarly to the deoxythymidine part of the feedback inhibitor, and
the lasso-like loop binds the base and deoxyribose moieties as in the com-
plex determined previously. The catalytic base, Glu97, has a different posi-
tion in the substrate complex from that in the complex with the feedback
inhibitor, having moved in closer to the 5¢-OH of the substrate to form a
hydrogen bond. The phosphorylation of and inhibition by several nucleo-
side analogues were investigated and are discussed in the light of the sub-
strate binding pocket, in comparison with human TK1. Kinetic differences
between Uu-TK and human TK1 were observed that may be explained by
structural differences. The tight interaction with the substrate allows minor
substitutions at the 3 and 5 positions of the base, only fluorine substitu-
tions at the 2¢-Ara position, but larger substitutions at the 3¢ position of
the deoxyribose.
Abbreviations
AZMT, 3¢-azido-methyl-dT; Ca-TK, Clostridium acetobutylicum thymidine kinase; dNK, deoxynucleoside kinase; FCPU, 3¢-fluoro-5-cyclopropyl-
dU; FLT, 3¢-fluoro-dT; hTK, human thymidine kinase; TK, thymidine kinase; Uu-TK, Ureaplasma urealyticum thymidine kinase.
FEBS Journal 272 (2005) 6365–6372 ª 2005 The Authors Journal Compilation ª 2005 FEBS 6365
that it has to rely solely on the salvage pathway for
synthesis of DNA precursors. Thus, Uu-TK is a gate-
way for the biosynthesis of dTTP, suggesting that
Uu-TK is a good target for drug development.
Recently the 3D structures of Uu-TK and cytosolic
hTK1 were determined [14,15]. The structure of these
thymidine kinases differs significantly from the earlier
known deoxyribonucleoside kinases and form a separ-
ate structural family. Humans carry four deoxyribonu-

cleoside kinases: cytosolic TK1 and deoxycytidine
kinase, and mitochondrial TK2 and deoxyguanosine
kinase. Deoxycytidine kinase, TK2, and deoxyguano-
sine kinase form a homologous deoxynucleoside kinase
(dNK) family with similar structures. This family also
includes the herpes viral TKs and the insect multisub-
strate deoxyribonucleoside kinase showing high activity
with dT [13]. The phosphate donor binds in both fam-
ilies to an a ⁄ b domain, but this domain in the TK1
family is more similar to other ATP-binding structures
in the RecA family [14]. Furthermore, instead of the
helical part forming the substrate site and the LID
region being involved in binding of the phosphate
donor in the dNK family, there is a domain in hTK1
and Uu-TK that contains a structural zinc atom and
lasso-like loop. In the feedback inhibitor complexes of
Uu-TK and hTK1, the lasso-like loop binds the thymi-
dine moiety of the feedback inhibitor by hydrogen
bonding by main-chain atoms [14]. The structure of the
bacterial TK from Clostridium acetobutylicum (Ca-TK)
is very similar to the Uu-TK, but, in the absence of
substrate or feedback inhibitor, the substrate site is
open and the lasso-like loop is disordered [16].
In the absence of a true substrate complex for any
member of the TK1 family, we have now determined
the first structure of a member of the TK1 family in
complex with the substrate dT. The possibilities for
drug design have been investigated by enzyme kinetics
and analyzed in view of substrate binding. It appears
that a combination of substitutions at several positions

of the nucleoside can pick up the small differences
between mycoplasmic and human TK1, which suggests
the route for further advances.
Results and discussion
Overall structure
The overall tetrameric structure of Uu-TK in the sub-
strate complex is very similar to that in the complex
with the feedback inhibitor dTTP. Each subunit can
be superimposed, with rmsds for Ca atoms of
0.3–0.6 A
˚
. The main differences are located close to
the phosphate binding sites, where a flexible loop
conformation differs among the four subunits of the
tetramer and among subunits in the two complexes
(Fig. 1A). In the present substrate complex structure,
only subunit A in the tetramer has a completely visible
loop, whereas in the dTTP complex, only subunits B
and D have visible loops [14]. In all other cytosolic
TK structures determined so far, this loop is visible in
only a few subunits. For example, in one of the inde-
pendent structure determinations of hTK1, this loop is
fully visible in one of eight subunits in the asymmetric
unit [15]. It may be that this loop is involved in phos-
phate donor interactions, but no such complex has so
far been determined. In the Ca-TK structure, the prod-
uct ADP is bound at the phosphate donor site. In spite
of this, only part of the loop is ordered in one of the
subunits [16]. This conformation is similar to that
observed in the Uu-TK–dTTP complex. The phosphate

donor site in the present Uu-TK–dT structure is occu-
pied by water molecules and, in one subunit, a Tris
molecule. Although the enzyme was crystallized in the
presence of the ATP analogue adenosine 5¢-[b,c-methy-
lene]-triphosphate (p[CH
2
]ppA; AMP-PCP), there was
no density for this molecule.
Substrate binding
The substrate dT binds to the enzyme in a similar way
to the dT moiety of the feedback inhibitor dTTP
(Figs 1B and 2) and has the same interactions with the
main chain of the lasso-like loop. This loop has the
same conformation in both complexes in contrast to
Ca-TK where no substrate was present in the crystals.
There, the lasso-like loop was disordered [16]. The sub-
strate site is obviously induced by substrate binding.
The base is hydrogen-bonded to main-chain atoms: O2
to N in residue 180, N3 to O in residue 178, and O4
to N in residue 128 (Fig. 2). The methyl group is posi-
tioned in a hydrophobic pocket lined by Cb of Ser163,
Sd in Met21, and Cd1 in Leu124. The closest polar
atom is the carbonyl oxygen of residue 126. O3¢ of the
deoxyribose is hydrogen-bonded to the main-chain
amino group of Gly182.
Glu97 has different conformations in the substrate
complex and the previously determined inhibitor com-
plex. In the dT complex, O5¢ is hydrogen bonded to
Glu97, in good agreement with its role as the catalytic
base for the phosphoryl transfer reaction (Fig. 1B). In

the inhibitor complex, the phosphates of the feedback
inhibitor repel the glutamate side chain. A shift in the
catalytic base was observed when the substrate and
feedback inhibitor complex of Drosophila dNK were
compared, but the conformational change was larger
in that case [17]. In Uu-TK, only the side chain chan-
Thymidine kinase in Ureaplasma urealyticum U. Kosinska et al.
6366 FEBS Journal 272 (2005) 6365–6372 ª 2005 The Authors Journal Compilation ª 2005 FEBS
ges conformation, whereas in Drosophila dNK the shift
is accompanied by main-chain movements.
Substrate specificity, nucleoside analogues
With radiolabelled ATP and a fixed concentration
(100 lm) of various nucleoside analogues, Uu-TK
activity was measured by a TLC assay. Table 1 shows
that 5-halogenated analogues are good substrates and
have the highest activities. The iodo atom has the same
van der Waals radius as the methyl group of thymine,
and the corresponding analogue has an activity com-
parable to that of dT. The analogue with the smaller
fluorine in the 5 position has a lower activity, slightly
lower than that of dU, which has a hydrogen atom in
the 5 position. The chlorine substitution is an outlier
in the halogen substitution series, as it has the highest
activity of the halogenated analogues. There is no cor-
relation between the electronegativity of the substituent
and its activity in the phosphorylation reaction.
From these investigations, it appears that substitu-
tions at the 5 position as large as a cyclopropyl group
are tolerated, and for an ethyl group the activity is
decreased to about half of that of dT. Larger substitu-

Fig. 2. Interactions between thymidine and Uu-TK. Hydrogen bonds
are shown as dotted lines. The tight binding site for the 2¢ position
between the main chain of Lys180-Ile181 and Met21. Any substitu-
tion at the 2¢ position hinders proper closure of the lasso, and
thereby weakens substrate co-ordination.
A
B
Fig. 1. (A) The structure of one subunit of Uu-TK (yellow) with con-
formations of the flexible loop as found in Uu-TK in complex with
dT in subunit A (orange, A), Uu-TK in complex with dTTP (green, B)
and in hTK1 in complex with dTTP (grey, C). The conformation of
the loop shown in green (B) is very similar to that found in Ca-TK in
complex with ADP as well as the loops in chains C and D of the
Uu-TK–dT structure. (B) Superimposition of the nucleotide-binding
region in the substrate complex (orange) and the inhibitor complex
(olive). The side chain of the catalytic Glu97 has different positions
in the two complexes. In the substrate complex, it is in a catalyti-
cally favourable position pointing inwards the active site. In the
inhibitor complex, the side chain is repelled by the phosphates of
the inhibitor.
U. Kosinska et al. Thymidine kinase in Ureaplasma urealyticum
FEBS Journal 272 (2005) 6365–6372 ª 2005 The Authors Journal Compilation ª 2005 FEBS 6367
tions clash sterically with neighbouring residues in the
5 position binding pocket (Fig. 2). With bulkier modi-
fications, the activity decreased and no activity was
seen with, e.g. 5-(2-bromovinyl)-dU (data not shown),
which is also the case with hTK1 [18,19]. This steric
hindrance in the 5 position correlates well with the
small hydrophobic binding pocket in the enzyme.
Analogues with modifications in the N3 position

showed much lower activity than dT. The compounds
had one to three carbons added in different configura-
tions and gave 15–20% activity (Table 1). The hydro-
gen-bonding between the N3 nitrogen in dT and the
main-chain carbonyl of residue 178 is lost with alkyl
substitutions in the N3 analogues tested. A substitution
in this position will probably hinder the tight spacing of
the lasso-like loop and disturb proper binding.
The 3¢-OH is in an exposed position that can tolerate
large substitutions (Table 1). The 3¢-OH of the deoxy-
ribose forms a hydrogen bond with the main-chain
amino group of Gly182 that is lost when the hydroxy
group is replaced. Still, an electronegative substituent,
such as fluorine, retains 50% of the activity. For the
analogue with no hydroxy group in the 3¢ position,
2¢,3¢-didehydro-T, the activity was  3% of that with
dT [11]. 3¢-Modified analogues that contain polar
groups, e.g. 3¢-fluoro-dT (FLT) and 3¢-azido-dT (AZT)
can still form a hydrogen bond with the amino group
of Gly182, whereas analogues with nonpolar atoms
bound to the 3¢-carbon, such as 3¢-fluoro-methyl-dT
(FMT) and 3¢-azido-methyl-dT (AZMT), cannot form
any hydrogen bond. The latter two showed 4–7-fold
lower activity than with dT (Table 1). The correspond-
ing a form of FMT and AZMT were inactive (data not
shown). Such substitutions would clash with Met21.
Analogues with modifications at both the 3¢ and 5
position had lower activity than their corresponding
analogue with only one modification, e.g. 3¢-fluoro-
5-fluoro-dU (FFU) vs. FLT. This is probably also the

case with analogues modified at both the 3¢ and 3N
positions.
Analogues with modifications at the 2¢ position
showed the lowest activity (< 2%) of all analogues
tested, e.g. arabinosyl-dT, 2¢-difluoro-dU and 2¢-
chloro-dU (data not shown). This agrees well with the
tight binding site that is crowded on both sides of the
2¢ position (Fig. 2). The OH group in arabinosyl-dT
would interact sterically with Tyr187, which is one of
the important residues that keep the lasso in place. The
smaller fluorine in the 2¢-Ara position was accepted
and 2¢-fluoro-arabinosyl-5-iodo-dU (FIAU) and 2¢-
fluoro-arabinosyl-5-methyl-dU (FMAU) showed 44%
and 34% activity, respectively. Any substitution on the
other side, such as 2¢-difluoro-dU and 2¢-chloro-dU,
would interact with the main-chain carbonyl of residue
180 (Fig. 2).
Analogues as inhibitors
Some analogues were chosen for further analysis as
inhibitors of dT phosphorylation. The IC
50
values
are presented in Table 2. dU, 5-fluro-dU (FdU) and
Table 1. Phosphorylation of nucleoside analogues by Uu-TK and
hTK1. The values are from one experiment repeated with similar
results (< 20% variation). The specific activity with dT was set to
100% (1900 units), and 100 l
M [c-
32
P]ATP was used.

Substrate (100 l
M)
Activity (%)
ReferenceUu-TK hTK1
5-Chloro-dU (CldU)
a
122 196
5-Iodo-dU (IdU)
a
113 170
5-Fluoro-dU (FdU)
a
5-Ethyl-dU (EtdU)
b
61
50
95
80
[11,19],
[19]
dU
a
3-Methyl-dT (MeT)
c
46
21
77
43
[11,19],
[19]

3-(2-Propynyl)-dT (PropT)
c
18 21
3-Isopropyl-dT (IsoT)
c
14 17 [19]
3¢-Fluoro-dT (FLT)
b
52 30 [29]
3¢-Azido-dT (AZT)
a
3¢-Azido-methyl-dT (AZMT)
b
35
8
52
15
[11,19],
[29]
3¢-Fluoro-methyl-dT (FMT)
b
715
2¢-Fluoro-arabinosyl-5-iodo-dU (FIAU)
d
44 76 [19]
2¢-Fluoro-arabinosyl-5-methyl-dU
(FMAU)
d
34 48 [19]
3¢-Fluoro-5-cyclopropyl-dU (FCPU)

b
52 33
3¢-Azido-5-iodo-dU (AZIU)
b
22 70
3¢-Fluoro-5-fluoro-dU (FFU)
b
19 20
3¢-Fluoro-5-ethynyl-dU (FEU)
b
15 10
Source of the compounds:
a
Sigma-Aldrich;
b
N. G. Johansson,
Medivir, Stockholm, Sweden;
c
W Tjarks, College of Pharmacy, The
Ohio State University, Columbus, Ohio;
d
J Fox, Memorial Sloan
Kettering Cancer Institute, New York.
Table 2. IC
50
values of selected nucleoside analogues with Uu-TK
and hTK1. Substrate concentrations were 1 l
M dT and 2 mM ATP
for Uu-TK and 0.2 l
M dT and 2 mM ATP for hTK1.

Substrate
IC
50
(lM)
Uu-TK hTK1
dU 484 ± 24 15 ± 0.8
FdU 273 ± 13 10 ± 0.5
MeT 234 ± 12 100 ± 5
EtdU 47 ± 2 8 ± 1
FIAU 16±0.8 78±2
FLT 14 ± 0.7 3.5 ± 0.2
FCPU 11 ± 0.6 4 ± 0.2
AZMT 11 ± 0.6 62 ± 3
AZIU 6 ± 0.3 < 1
Thymidine kinase in Ureaplasma urealyticum U. Kosinska et al.
6368 FEBS Journal 272 (2005) 6365–6372 ª 2005 The Authors Journal Compilation ª 2005 FEBS
3-methyl-dT (MeT) showed low ability to inhibit
Uu-TK, with IC
50
values of 200–500 lm, using 1 lm
[
3
H]dT as substrate. 5-Ethyl-dU (EtdU) had an inter-
mediate IC
50
value (47 lm), whereas 2¢ and 3¢ ana-
logues containing fluorine or azido substitutions were
relatively efficient inhibitors (IC
50
values < 20 lm),

with 3¢-azido-5-iodo-dU (AZIU) having the lowest
IC
50
value (6 lm) (Table 2). IC
50
values for hTK1,
using 0.2 lm [
3
H]dT as substrate, were also determined
with the above analogues (Table 2). A lower dT con-
centration was used for hTK1 to compensate for the
lower K
m
value observed with hTK1 and dT (in the
presence of 2 mm ATP) [11,20]. The results with hTK1
showed a pattern, with MeT, FIAU and AZMT form-
ing a group of quite poor inhibitors with IC
50
values
of 60–100 lm. The other analogues tested with hTK1
fell into a group with relatively high capacity to inhi-
bit, and their IC
50
values ranged from 15 lm down to
below 1 lm for 3¢-azido-5-iodo-dU (AZIU), the best
inhibitor (Table 2).
FLT and 3¢-fluoro-5-cyclopropyl-dU (FCPU), with
IC
50
values of 14 and 11 lm, respectively, were the

only analogues that showed higher activity with
Uu-TK than with hTK1 (Table 1). Kinetic studies were
performed with Uu-TK and hTK1 using FLT and
FCPU as variable substrates together with 0.5 mm
[c-
32
P]ATP. These analogues had about 3.5–4-fold
higher k
cat
values than dT, but at the same time five-
fold higher K
m
values, resulting in a slightly lower effi-
ciency than dT (Table 3). In the case of hTK1, FLT
showed in this experiment higher efficiency than FCPU
and dT.
Comparison with hTK1
Overall, the relative phosphorylation rates of the ana-
logues tested with Uu-TK are equal to or lower than
the corresponding rates with hTK1. The exceptions are
FCPU and FLT, which show higher relative activity
with Uu-TK than with hTK1. FCPU has a cyclopropyl
substitution in the 5¢ position of the base, and its bet-
ter activity with Uu-TK is probably due to the slightly
larger binding pocket of Uu-TK. This enzyme has a
serine at position 163 where hTK1 has a threonine,
and the extra methyl group makes the site narrower.
FCPU and FLT also have fairly low IC
50
values,

which should be an advantage for a potential inhibitor.
However, the corresponding IC
50
values with hTK1
were still lower, and this was the case with seven out
of the nine nucleosides tested as inhibitors of dT phos-
phorylation. Only FIAU and AZMT had higher IC
50
values with hTK1 than with Uu-TK. Still, they were
not very efficiently phosphorylated by Uu-TK.
These results show that the capacity of a nucleoside
to be a good inhibitor does not directly correlate with
its rate of phosphorylation. Despite the high struc-
tural similarities between Uu-TK and hTK1, the latter
is much easier to inhibit, as shown by its lower IC
50
values. Furthermore, these two enzymes show a
10-fold difference in K
m
values for dU and dU ana-
logues [11,21]. The reasons for these differences in
function may be related to the differences around the
5 position of the substrate, where Uu-TK has a serine
and hTK1 has the more hydrophobic threonine.
The analogues investigated so far may not be
direct lead compounds for the further development
of selective Uu-TK inhibitors, but this study demon-
strates the necessity to use both a structural and
functional approach for identification of new inhibi-
tors and alternative substrates when nucleoside kinas-

es are the targets. Such inhibitors would be
desirable, as they could serve as efficient antibiotics
because U. urealyticum lacks the capacity to synthes-
ize DNA precursors de novo. The most promising
route for further drug development for Uu-TK from
this study seems to be to explore further substitu-
tions at the 5 position and 3¢ position and combina-
tions thereof.
Experimental procedures
Materials
The radiolabelled substances [
3
H]dT (25 CiÆmmol
)1
) and
[c-
32
P]ATP ( 3000 CiÆmmol
)1
) were purchased from
Amersham Biosciences (Uppsala, Sweden). Recombinant
Uu-TK and hTK1 were prepared as previously described
[11,22].
Enzyme assay
TK activity was determined by using a DE-81 filter paper
technique with [
3
H]Thd or by a phosphoryl-transfer assay
with [c-
32

P]ATP, as previously described [11]. The standard
reaction mixture contained 50 mm Tris ⁄ HCl, pH 7.6, 2 mm
Table 3. Kinetic parameters of FLT, FCPU and dT with Uu-TK and
hTK1. A fixed [c-
32
P]ATP concentration (0.5 mM) was used. k
cat
values were calculated based on a subunit molecular mass of
27.5 kDa (Uu-TK) and 25.5 kDa (hTK1), respectively.
Substrate
K
m
(lM) k
cat
(s
)1
) k
cat
⁄ K
m
(s
)1
ÆM
)1
)
Uu-TK hTK1 Uu-TK hTK1 Uu-TK hTK1
FLT 47 5 0.67 0.19 1.4 · 10
4
3.6 · 10
4

FCPU 42 8 0.80 0.09 1.9 · 10
4
1.1 · 10
4
dT 9 6 0.19 0.16 2.1 · 10
4
2.8 · 10
4
U. Kosinska et al. Thymidine kinase in Ureaplasma urealyticum
FEBS Journal 272 (2005) 6365–6372 ª 2005 The Authors Journal Compilation ª 2005 FEBS 6369
MgCl
2
,2mm ATP, 0.5 mgÆmL
)1
BSA, 5 mm dithiothreitol,
1 lm [
3
H]dT and 0.5 ng Uu-TK or 0.2 lm [
3
H]dT and 1 ng
hTK1 in a total volume of 50 lL. When the reaction was
finished, the filters were washed three times in 5 mm ammo-
nium formate and once in water. Reaction products were
then eluted with 0.5 mL 0.1 m HCl ⁄ 0.2 m KCl, and the
radioactivity was determined by liquid scintillation counting
(Beckman). The results were analysed by the SigmaPlot
Enzyme Kinetic Module version 1.1 (SPSS Science, Chi-
cago, IL, USA). The phosphoryl-transfer assay was per-
formed with 100 lm or 500 lm [c-
32

P]ATP in the same
buffer as above with variable concentrations of nucleosides
and 10 ng Uu-TK in a total volume of 25 lL. The phos-
phorylated products were separated by TLC and quantified
by phosphorimaging analysis (Fujifilm Image Gauge, ver-
sion 3.3).
One unit of kinase activity was defined as the formation
of 1 nmol deoxyribonucleoside 5¢-monophosphate per mg
protein per min.
Crystallization, data collection and refinement
The His-tag of Uu-TK was cleaved off overnight with
10 UÆmg
)1
thrombin purchased from Amersham Bio-
sciences. The protein was crystallized by the hanging drop
vapour diffusion method at 14 °C. The protein solution
consisted of 10 mgÆmL
)1
Uu-TK, 5 mm Thd and 3 mm
AMP-PCP, whereas the reservoir solution consisted of
15% poly(ethylene glycol) 3350 and 0.3 m ammonium for-
mate. A small petri dish (diameter 5.5 cm) was filled with
0.5 mL of the crystallization solution. 2 lL protein solu-
tion together with 2 lL crystallization solution was
applied to the lid of the Petri dish. Two different crystal
forms appeared after 2–5 days. Before flash-freezing in
liquid nitrogen, the crystals were swept through cryo-pro-
tecting solution consisting of 15% poly(ethylene glycol)
3350, 0.3 m ammonium formate and 20% glycerol.
Data were collected at ID14-3 ESRF, processed with

mosflm [23], and scaled with scala from the CCP4 pro-
gram suit [24]. The statistics from data reduction are pre-
sented in Table 4. The structure was solved with molrep
[25]. As search model, the tetramer of Uu-TK from the
dTTP complex, PDB code 1XMR, was used. Residues 50–
67 were omitted from each chain of the search model.
The omitted region or parts of the omitted region,
depending on the chain, could be traced with ARPwARP
[26]. After simulated annealing performed in CNS, subse-
quent refinement was performed in REFMAC5 [27]. Dur-
ing the whole refinement, NCS restraints were applied to
residues 12–49 and 69–213 for each protein chain. By the
end of the refinement, the restraints were loosened to
medium for main-chain atoms and loose for side-chain
atoms. A TLS model consisting of 15 TLS groups was
applied during the final steps of the refinement. Each
monomer was divided into four domains [the a ⁄ b domain,
the flexible loop (missing in subunit B), the lasso domain,
and the C-terminus), together creating 15 TLS groups.
Table 4 shows the statistics from data refinement. All
model building was carried out in O [28]. Each monomer
contains 223 residues, although the N-terminus and C-ter-
minus are disordered, and between 6–11 residues depend-
ing on the chain are omitted from the model. In addition,
there is a disordered region between residues 51 and 66.
Only in chain A could the whole region be traced; in
chains C and D parts of the region are modelled, and in
chain B the entire region is missing. As observed previ-
ously [14–16], this part of the structure takes different
conformations. The structure has been deposited with

PDB code 2B8T.
Acknowledgements
This work was supported by grants from the
Swedish Research Council for the Environment,
Agricultural Sciences and Spatial Planning (to L.Y.
and S.E.), the Swedish Research Council (to H.E.
and S.E.), and the Swedish Cancer Foundation (to
H.E.).
References
1 Fraser CM, Gocayne JD, White O, Adams MD, Clay-
ton RA, Fleischmann RD, et al. (1995) The minimal
gene complement of Mycoplasma genitalium. Science
270, 397–403.
Table 4. Data collection and refinement statistics. Numbers in par-
entheses refer to the outer resolution bin.
Space group P2
1
Cell dim. (A
˚
, °)a¼ 57.16
b ¼ 115.65 c ¼ 64.47
b ¼ 101.02
Content of the asymmetric unit 1 tetramer
Resolution (A
˚
) 2.00 (2.11–2.00)
Completeness (%) 99.1 (98.5)
R
meas
(%) 9.0 (48.4)

I ⁄ rI 16.4 (2.9)
Redundancy 5
No. of observed reflections 274 598
No. of unique reflections 54 312
Beam line ESRF, ID14eh3
Wavelength (A
˚
) 0.931
Temperature (K) 100
R (%) 19.5
R
free
(%) 23.5
R.m.s.d.
Bond length (A
˚
) 0.008
Bond angle (°) 1.04
Mean B-value (A
˚
2
)32.9
Thymidine kinase in Ureaplasma urealyticum U. Kosinska et al.
6370 FEBS Journal 272 (2005) 6365–6372 ª 2005 The Authors Journal Compilation ª 2005 FEBS
2 Glass JI, Lefkowitz EJ, Glass JS, Heiner CR, Chen EY
& Cassell GH (2000) The complete sequence of the
mucosal pathogen Ureaplasma urealyticum. Nature 407,
757–762.
3 Chambaud I, Heilig R, Ferris S, Barbe V, Samson D,
Galisson F, et al. (2001) The complete genome sequence

of the murine respiratory pathogen Mycoplasma pul-
monis. Nucleic Acids Res 29, 2145–2153.
4 Himmelreich R, Hilbert H, Plagens H, Pirkl E, Li BC &
Herrmann R (1996) Complete sequence analysis of the
genome of the bacterium Mycoplasma pneumoniae.
Nucleic Acids Res 24, 4420–4449.
5 Papazisi L, Gorton TS, Kutish G, Markham PF,
Browning GF, Nguyen DK, et al. (2003) The complete
genome sequence of the avian pathogen Mycoplasma
gallisepticum strain R (low). Microbiology 149, 2307–
2316.
6 Sasaki Y, Ishikawa J, Yamashita A, Oshima K, Kenri
T, Furuya K, et al. (2002) The complete genomic
sequence of Mycoplasma penetrans an intracellular
bacterial pathogen in humans. Nucleic Acids Res 30,
5293–5300.
7 Westberg J, Persson A, Holmberg A, Goesmann A,
Lundeberg J, Johansson KE, et al. (2004) The genome
sequence of Mycoplasma mycoides subsp mycoides SC
type strain PG1T the causative agent of contagious
bovine pleuropneumonia (CBPP). Genome Res 14, 221–
227.
8 Jaffe JD, Stange-Thomann N, Smith C, DeCaprio D,
Fisher S, Butler J, et al. (2004) The complete genome
and proteome of Mycoplasma mobile. Genome Res 14,
1447–1461.
9 Vasconcelos AT, Ferreira HB, Bizarro CV, Bonatto
SL, Carvalho MO, Pinto PM, et al. (2005) Swine and
poultry pathogens: the complete genome sequences of
two strains of Mycoplasma hyopneumoniae and a

strain of Mycoplasma synoviae. J Bacteriol 187 , 5568–
5577.
10 Wang L, Westberg J, Bolske G & Eriksson S (2001)
Novel deoxynucleoside-phosphorylating enzymes in
mycoplasmas: evidence for efficient utilization of deoxy-
nucleosides. Mol Microbiol 42, 1065–1073.
11 Carnrot C, Wehelie R, Eriksson S, Bolske G & Wang L
(2003) Molecular characterization of thymidine kinase
from Ureaplasma urealyticum: nucleoside analogues as
potent inhibitors of mycoplasma growth. Mol Microbiol
50, 771–780.
12 Hudson MM & Talbot MD (1997) Ureaplasma urealyti-
cum. Int J STD AIDS 8, 546–551.
13 Eriksson S, Munch-Petersen B, Johansson K &
Eklund H (2002) Structure and function of cellular
deoxyribonucleoside kinases. Cell Mol Life Sci 59,
1327–1346.
14 Welin M, Kosinska U, Mikkelsen NE, Carnrot C, Zhu
C, Wang L, et al. (2004) Structures of thymidine kinase
1 of human and mycoplasmic origin. Proc Natl Acad
Sci USA 101, 17970–17975.
15 Birringer MS, Claus MT, Folkers G, Kloer DP, Schulz
GE & Scapozza L (2005) Structure of a type II thymi-
dine kinase with bound dTTP. FEBS Lett 579, 1376–
1382.
16 Kuzin AP, Abashidze M, Forouhar F, Vorobiev SM,
Acton TB, Ma L-C, et al. (2004) X-ray structure of
Clostridium Acetobutylicum thymidine kinase with Adp.
Northeast structural genomics target Car26 PDB code
1XX6.

17 Mikkelsen NE, Johansson K, Karlsson A, Knecht W,
Andersen G, Piskur J, et al. (2003) Structural basis for
feedback inhibition of the deoxyribonucleoside salvage
pathway: studies of the Drosophila deoxyribonucleoside
kinase. Biochemistry 42, 5706–5712.
18 Eriksson S & Wang J (1995) Substrate specificities of
mitochondrial thymidine kinase and cytosolic deoxy-
cytidine kinase against 5-aryl substituted pyrimidine-
2¢-deoxyribose analogues. Nucleosides Nucleotides 14,
507–510.
19 Al-Madhoun A, Tjarks W & Eriksson S (2004) The
role of thymidine kinases in the activation of pyrimi-
dine nucleoside analogues. Mini Rev Med Chem 4,
341–350.
20 Munch-Petersen B, Tyrsted G & Cloos L (1993) Rever-
sible ATP-dependent transition between two forms of
human cytosolic thymidine kinase with different enzy-
matic properties. J Biol Chem 268, 15621–15625.
21 Munch-Petersen B, Cloos L, Tyrsted G & Eriksson S
(1991) Diverging substrate specificity of pure human
thymidine kinases 1 and 2 against antiviral dideoxy-
nucleosides. J Biol Chem 266, 9032–9038.
22 Lunato AJ, Wang J, Woollard JE, Anisuzzaman AK,
Ji W, Rong FG, et al. (1999) Synthesis of 5-(carbo-
ranylalkylmercapto)-2¢-deoxyuridines and 3-(carbo-
ranylalkyl) thymidines and their evaluation as substrates
for human thymidine kinases 1 and 2. J Med Chem 42,
3378–3389.
23 Leslie AGW (1992) Mosfilm. Joint CCP4 and ESF-
EACMB Newsletter Protein Crystallography. Daresbury

Laboratory, Warrington UK.
24 CCP4 (1994) The CCP4 suite: programs for protein
crystallography. Acta Crystallogr D Biol Crystallogr 50,
760–763.
25 Vagin A & Teplyakov A (2000) An approach to multi-
copy search in molecular replacement. Acta Crystallogr
D Biol Crystallogr 56, 1622–1624.
26 Morris RJ, Perrakis A & Lamzin VS (2003) ARP ⁄ wARP
and automatic interpretation of protein electron density
maps. Methods Enzymol 374, 229–244.
27 Murshudov GN, Vagin AA & Dodson EJ (1997)
Refinement of macromolecular structures by the maxi-
mum-likelihood method. Acta Crystallogr D Biol Crys-
tallogr 53, 240–255.
U. Kosinska et al. Thymidine kinase in Ureaplasma urealyticum
FEBS Journal 272 (2005) 6365–6372 ª 2005 The Authors Journal Compilation ª 2005 FEBS 6371
28 Jones TA, Zou JY, Cowan SW & Kjeldgaard M (1991)
Improved methods for building protein models in
electron density maps and the location of errors in these
models. Acta Crystallogr D Biol Crystallogr 47, 110–119.
29 Johansson N & Eriksson S (1996) Structure-activity
relationships for phosphorylation of nucleoside analogs
to monophosphates by nucleoside kinases. Acta Biochim
Pol 43, 143–160.
Thymidine kinase in Ureaplasma urealyticum U. Kosinska et al.
6372 FEBS Journal 272 (2005) 6365–6372 ª 2005 The Authors Journal Compilation ª 2005 FEBS