Báo cáo khoa học: Unchanged thymidine triphosphate pools and thymidine metabolism in two lines of thymidine kinase 2-mutated fibroblasts docx

Unchanged thymidine triphosphate pools and
thymidine metabolism in two lines of thymidine
kinase 2-mutated fibroblasts
Miriam Frangini
1
, Chiara Rampazzo
1
, Elisa Franzolin
1
, Mari-Carmen Lara
2
, Maya R. Vila
`
2
,
Ramon Martı
´
2
and Vera Bianchi
1
1 Department of Biology, University of Padova, Italy
2 Institut de Recerca Hospital Universitari Vall d’Hebron, Barcelona, Spain
Mitochondrial DNA depletion syndromes (MDSs)
are a group of heterogeneous mitochondrial diseases
characterized by reduced copy numbers of mtDNA,
insufficient synthesis of the mitochondrially encoded
components of the respiratory chain complexes, and
impairment of energy metabolism. MDSs are inherited
as autosomal recessive traits and present striking tissue-
specific phenotypes. The underlying genetic defects
have been identified only for a small fraction of the

reported cases [1]. Interestingly, out of nine genes
involved in MDS, four code for enzymes of deoxy-
nucleotide metabolism, namely the two mitochondrial
deoxynucleoside kinases [thymidine kinase (TK2)] [2]
and deoxyguanosine kinase [3], cytosolic thymidine
phosphorylase (TP) [4] and p53R2 [5], the stable
isoform of ribonucleotide reductase small subunit.
Whereas TP is a catabolic enzyme that degrades
thymidine and deoxyuridine and participates in the
Keywords
dTTP pool turnover; mitochondrial DNA
depletion syndrome; mitochondrial DNA
precursors; p53R2; thymidine phosphorylase
Correspondence
V. Bianchi, Department of Biology, Via Ugo
Bassi 58B, 35131 Padova, Italy
Fax: +39 0498276280
Tel: +39 0498276282
E-mail:
(Received 16 October 2008, revised 1
December 2008, accepted 11 December
2008)
doi:10.1111/j.1742-4658.2008.06853.x
Mitochondrial thymidine kinase (TK2) catalyzes the phosphorylation of
thymidine in mitochondria. Its function becomes essential for dTTP syn-
thesis in noncycling cells, where cytosolic dTTP synthesis via R1 ⁄ R2 ribo-
nucleotide reductase and thymidine kinase 1 is turned down. Mutations in
the nuclear gene for TK2 cause a fatal mtDNA depletion syndrome. Only
selected cell types are affected, suggesting that the other cells compensate
for the TK2 deficiency by adapting the enzyme network that regulates

dTTP synthesis outside S-phase. Here we looked for such metabolic adap-
tation in quiescent cultures of fibroblasts from two TK2-deficient patients
with a slow-progressing syndrome. In cell extracts, we measured the activi-
ties of TK2, deoxycytidine kinase, thymidine phosphorylase, deoxynucleo-
tidases and the amounts of the three ribonucleotide reductase subunits.
Patient cells contained 40% or 5% TK2 activity and unchanged activities
of the other enzymes. However, their mitochondrial and cytosolic dTTP
pools were unchanged, and also the overall composition of the dNTP
pools was normal. TK2-dependent phosphorylation of [
3
H]thymidine in
intact cells and the turnover of the dTTP pool showed that even the fibro-
blasts with 5% residual TK2 activity synthesized dTTP at an almost nor-
mal rate. Normal fibroblasts apparently contain more TK2 than needed to
maintain dTTP during quiescence, which would explain why TK2-mutated
fibroblasts do not manifest mtDNA depletion despite their reduced TK2
activity.
Abbreviations
BVDU, 5-bromovinyl-2-deoxyuridine; KIN109, 1-(6-[1,1-(diphenyl)-1-(4-pyridyl)methoxy]hexyl)thymine; MDS, mitochondrial DNA depletion
syndrome; TK1, thymidine kinase 1; TK2, mitochondrial thymidine kinase; TP, thymidine phosphorylase.
1104 FEBS Journal 276 (2009) 1104–1113 ª 2009 The Authors Journal compilation ª 2009 FEBS
regulation of thymidine phosphate pools, the other
three are synthetic enzymes needed for the maintenance
of dNTP pools in nonproliferating cells.
During cell growth, the main source of dNTPs for
nuclear and mtDNA replication is S-phase-specific
de novo synthesis catalyzed by the canonical R1 ⁄ R2
form of ribonucleotide reductase. Regulated proteoly-
sis of R2 in late mitosis [6] turns off R1 ⁄ R2-dependent
ribonucleotide reduction, leading to a drop in dNTP

pool sizes in G
1
and postmitotic cells. The dNTPs
required for DNA repair and mtDNA synthesis during
the whole lifespan of cells are instead produced by
salvage of deoxynucleosides by cytosolic and mito-
chondrial kinases and by R1 ⁄ p53R2-dependent ribonu-
cleotide reduction. As compared to the rate of de novo
synthesis during cell proliferation, the de novo synthesis
of dNTPs catalyzed by R1 ⁄ p53R2 amounts to only a
few per cent [7], but it is essential to supply precursors
for mtDNA. In humans, genetic inactivation of p53R2
causes a very severe form of MDS that affects multiple
organs, leading to death within a few weeks after
birth [5].
The broad spectrum of affected tissues in p53R2-
mutated individuals [5] suggests a general, fundamental
function of p53R2 for dNTP synthesis in nonprolifer-
ating cells. Conversely, mutations in either of the two
mitochondrial deoxynucleoside kinases have more
restricted effects. Deficiency of deoxyguanosine kinase
causes a hepatocerebral MDS [3], and TK2 deficiency
was originally discovered to cause acute fatal mito-
chondrial myopathy [2].
The phenotypic spectrum of TK2 deficiency has
turned out to be wider than the myopathy observed
originally [8], and also the recently described mouse
models of TK2 deficiency show mtDNA depletion in
multiple organs [9,10]. A further variable is the speed
of the MDS progression. In most cases of TK2 defi-

ciency, the disease has early onset and rapid develop-
ment, with death occurring during early childhood. In
two cases of TK2 deficiency, a severe reduction of
TK2 enzymatic activity was associated with late onset
and slow progression of the myopathy and long sur-
vival [11–14]. The two patients were both compound
heterozygotes, harbored two different pairs of TK2
mutations, i.e. T77M ⁄ R161K [13] and R152G ⁄ K171del
[12] (according to the latest GenPept entry,
NP_004605.3, the mutations are T150M ⁄ R234K and
R225G ⁄ K244del), and presented contrasting and dis-
tinct features of myopathic damage. Although the
overall condition of the patients deteriorated with
time, in both cases an apparent negative selection of
abnormal muscle fibers suggested that compensatory
molecular mechanisms restored respiratory chain func-
tions or mtDNA levels in the surviving fibers. In one
case, compensatory adaptations of nucleotide meta-
bolism were hypothesized [12].
In nonproliferating fibroblasts, the dTTP pool is
maintained by a network of interlocked synthetic and
catabolic enzyme activities. TK2 operates in parallel
with R1 ⁄ p53R2 in the synthesis of thymidine nucleo-
tides, and mitochondrial and cytosolic deoxynucleotid-
ases and cytosolic TP are active in their degradation
[15]. The balance between the competing enzyme activ-
ities sets the level of dTTP in cytosolic and mitochon-
drial pools. The availability of thymidine in the
extracellular milieu is an additional factor, enhancing
the influence of salvage relative to de novo synthesis

[15]. A rapid exchange of nucleotides across the mito-
chondrial inner membrane maintains the cytosolic and
mitochondrial compartments in equilibrium.
We were interested in investigating whether in the
fibroblasts of the two TK2 patients the operation of
the network was altered to compensate for the reduced
TK2 activity. In extracts from quiescent cultures of
patient and control fibroblasts, we measured the
amount of TK2 and other enzymes of the network
from their catalytic activity, by western blotting or,
indirectly, by mRNA quantification. We then deter-
mined in situ activities from the phosphorylation of
[
3
H]thymidine in intact cells. Surprisingly, although the
activity of TK2 in extracts from quiescent patient cells
was only 5–40% of that of control cells, the in situ
activity of the enzyme was hardly affected, with no
changes being seen in the composition of the dNTP
pools. We detected no major change in the expression
of other enzymes involved in dTTP regulation. These
observations suggest that TK2 activity in wild-type
fibroblasts largely exceeds the basic requirements for
the maintenance of mitochondrial dTTP.
Results
Low TK2 activity in extracts of patient fibroblasts
Cycling cells are unaffected by loss of TK2 activity,
because the bulk of mitochondrial dTTP is produced
de novo in the cytosol by ribonucleotide reduction and
by thymidine kinase 1 (TK1)-catalyzed salvage of thy-

midine [15]. Therefore, we perfomed all experiments
with quiescent fibroblasts where, in the absence of
S-phase-specific de novo synthesis, TK2 contributes to
the maintenance of the dTTP pool together with the
R1 ⁄ p53R2 variant of ribonucleotide reductase.
We first compared the activity of TK2 in protein
extracts of fibroblasts from patients Pa [12] and
Pb [13] and from two control lines (Table 1). We
M. Frangini et al. Thymidine metabolism in TK2-mutated fibroblasts
FEBS Journal 276 (2009) 1104–1113 ª 2009 The Authors Journal compilation ª 2009 FEBS 1105
employed both a specific assay based on the phosphor-
ylation of 5-bromovinyl-2-deoxyuridine (BVDU), a
deoxyuridine analog that is a good substrate for TK2
but not for TK1 [16], and the phosphorylation of
thymidine, used in earlier determinations of TK2 in
the same patient cells [12,13]. A specific inhibitor of
TK2 [1-(6-[1,1-(diphenyl)-1-(4-pyridyl)methoxy]hexyl)
thymine (KIN109)] [16] strongly inhibited thymidine
phosphorylation, demonstrating that the reaction
depended on TK2 in all cell lines (Table 1). The assays
with the two substrates gave concordant results, show-
ing lower TK2 activity in Pa extracts (about 5% of the
control) than in Pb extracts (40% of the control). Con-
sidering that the mutated enzymes might be unstable
during preparation or storage, we tested, by the
BVDU phosphorylation assay, whether the TK2
activity of Pa and control extracts was enhanced by
glycerol, 15 mm MgCl
2
or increasing concentrations

(0.2–5 mm) of ATP, and whether freezing decreased
the activity of the cell lysate. None of these modifica-
tions changed the specific activity of TK2, which in
Pa extracts remained below 10% of the control value
(not shown).
Unchanged dNTP pools in patient fibroblasts
We next compared the size of the dTTP pools in quies-
cent cultures of control and TK2-mutated fibroblasts.
No significant difference was observed. The total cellu-
lar pool contained 1.72 ± 0.2 pmol dTTP per 10
6
cells
in the controls, 1.67 ± 0.3 pmol dTTP per 10
6
cells in
Pa cells, and 1.33 ± 0.2 pmol dTTP per 10
6
cells in
Pb cells. Also, the mitochondrial dTTP pools were
similar in control and patient fibroblasts: 0.11 ±
0.03 pmol per 10
6
cells and 0.08 ± 0.03 pmol dTTP
per 10
6
cells respectively. These data differ from the
only published report on dTTP pools in human TK2-
deficient fibroblasts, where the sizes of the mitochon-
drial dTTP pools of two mutant lines were 50% and
30% smaller than that of the controls, and also the

dCTP pool was reduced, albeit not significantly, caus-
ing an imbalance of the dNTP pools as compared to
the controls [17]. We determined the amounts of all
four dNTPs in the cytosolic and mitochondrial pools
of the two patient lines and of two additional controls.
The relative sizes of the pools were those observed pre-
viously in quiescent skin and lung fibroblasts [18], and
we found no dNTP pool imbalance in the patient cells
(not shown). Thus, the strong reduction of TK2 activ-
ity in Pa and Pb fibroblasts did not lead to detectable
modifications of mitochondrial and cytosolic dTTP
during 10 days of quiescence in culture.
There is no increase of dNTP de novo synthesis
in patient fibroblasts
We reasoned that in the patient fibroblasts, the
R1 ⁄ p53R2-dependent de novo pathway might be over-
expressed to compensate for the TK2 defect. We there-
fore determined the mRNA levels of the three
ribonucleotide reductase subunits and of p53, the tran-
scription factor that controls the expression of p53R2,
during proliferation and after 10 days of quiescence
(Table 2). As compared with proliferating cells, quies-
cent cells contained between 20% and 50% of R1
mRNA, whereas R2 mRNA decreased to 10%. Both
p53 and p53R2 mRNAs increased during quiescence.
Table 1. TK2 activity in cell extracts from quiescent cultures of
control and TK2-mutated fibroblasts. Two enzyme assays specific
for TK2 were used, as detailed in Experimental procedures. In one
assay, TK2 activity was measured using [
3

H]BVDU, a TK2-specific
substrate [16]. In the second assay, the phosphorylation of [
3
H]thy-
midine was measured in the absence or presence of 100 l
M
KIN109, a TK2-specific inhibitor [16]. The more than 80% inhibition
of thymidine phosphorylation caused by KIN109 indicates that TK2
is the kinase involved in the reaction. Pa and Pb, TK2-mutated cells.
TK2 enzyme activity is expressed in picomol of substrate phosphor-
ylated ⁄ min of incubation ⁄ mg of protein. Values are means of dupli-
cate determinations from three to six separate experiments ±
standard deviation of the mean.
TK2 activity (pmolÆmin
)1
Æmg
)1
protein)
Cell line
[
3
H]BVDU
(0.2 l
M)
[
3
H]Thymidine (1 lM)
)KIN109 +KIN109
Controls 5.2 ± 1.2 12.1 ± 5.8 0.4 ± 0.2
Pa 0.2 ± 0.1 0.9 ± 0.3 0.2 ± 0.1

Pb 2.1 ± 0.4 4.8 ± 1.8 0.3 ± 0.1
Table 2. Comparison of expression of the three ribonucleotide
reductase subunits and p53 in quiescent and cycling cultures of
control and TK2-mutated fibroblasts. The levels of the mRNAs for
ribonucleotide reductase subunits R1, R2 and p53R2 and for p53
were evaluated by real-time RT-PCR on cDNAs prepared from total
RNA extracted from proliferating and quiescent fibroblasts. Cyclo-
philin A was the internal control used for normalizations. The level
of each mRNA is expressed as fold increase relative to that mea-
sured in cycling cells. Data are means of measurements from three
different experiments run in triplicate ± standard deviation of the
mean. Pa and Pb, TK2-mutated lines.
mRNA fold increase (quiescent ⁄ cycling cells)
Cell line R1 R2 p53R2 p53
Control 0.35 ± 0.06 0.06 ± 0.01 2.82 ± 0.77 3.40 ± 0.59
Pa 0.18 ± 0.03 0.10 ± 0.03 5.10 ± 0.53 8.53 ± 1.57
Pb 0.47 ± 0.04 0.12 ± 0.03 3.78 ± 0.23 5.47 ± 0.90
Thymidine metabolism in TK2-mutated fibroblasts M. Frangini et al.
1106 FEBS Journal 276 (2009) 1104–1113 ª 2009 The Authors Journal compilation ª 2009 FEBS
Whereas the changes of ribonucleotide reductase
subunits were similar in all lines, the induction of p53
mRNA was more variable, with the largest increase
being seen in Pa fibroblasts. At the protein level,
western blotting experiments (Fig. 1) confirmed the
large increase of p53 in Pa cells. For p53R2, we found
no clear variations between patient and control cells.
Thus, the R1 ⁄ p53R2 form of ribonucleotide reduc-
tase was present at comparable levels in patient and
control fibroblasts during quiescence. To assess
whether, in the former case, ribonucleotide reduction

contributed to the synthesis of dTTP more than in the
controls, we used hydroxyurea, a specific inhibitor of
ribonucleotide reductase. In all cell lines, 2 h of treat-
ment with 3 mm hydroxyurea induced, at most, a 20%
decrease of dTTP, with no preferential effect on the
TK2-mutated fibroblasts (not shown).
De novo synthesis of dTMP occurs mostly by deami-
nation of dCMP followed by methylation of dUMP
[19]. dCMP is derived from dephosphorylation of
dCDP produced by ribonucleotide reductase and from
phosphorylation of deoxycytidine by deoxycytidine
kinase. We found no difference in the activity of
deoxycytidine kinase in extracts from patient and
control cells, ruling out the possibility that enhanced
deoxycytidine salvage contributed to the synthesis of
dTTP in the TK2-mutated cells (not shown).
There is no downregulation of deoxynucleotide
catabolism in patient fibroblasts
The size of dNTP pools results from the interplay
between synthesis and degradation of their components
[20]. Pa and Pb fibroblasts maintained apparently nor-
mal dTTP pools without upregulating the de novo
pathway. A possible mechanism might be the down-
regulation of catabolic enzymes such as TP or the two
deoxynucleotidases, cytosolic deoxynucleotidase and
mitochondrial deoxynucleotidase [21]. In protein
extracts from quiescent cultures of the two control and
two TK2-mutated lines, we measured the activity of
TP and the combined activity of the two deoxynucleo-
tidases (Table 3). Whereas deoxynucleotidase activity

was identical in all lines, TP levels were more variable,
but the range of variation was the same in patient and
control cells. Thus, reduced breakdown of deoxynucle-
otides or thymidine did not account for the mainte-
nance of dTTP in patient fibroblasts.
Thymidine metabolism in TK2-mutated
fibroblasts
To measure the in situ activity of the metabolic path-
way that leads, in intact cells, to the synthesis of
dTTP, we incubated quiescent TK2-mutated and con-
trol fibroblasts with 25 nm [
3
H]thymidine for 5 and
20 min, and measured changes in the size of the dTTP
Fig. 1. Western blot analysis of ribonucleotide reductase subunits
and p53 in extracts from quiescent and cycling cultures of control
and TK2-mutated fibroblasts. We examined by immunoblotting the
abundance of p53R2, R1, R2 and p53 in extracts of control (Ca, Cb)
and TK2-mutated (Pa, Pb) fibroblasts in proliferating cultures (P) and
after 10 days of quiescence (Q). The relative amount of each pro-
tein was normalized using b-actin as loading control. Q ⁄ P, ratio
between levels of each protein in quiescent and proliferating cul-
tures of each cell line. The blot for p53 in Cb extracts was obtained
from a separate electrophoresis run.
Table 3. TP and deoxynucleotidase activity in cell extracts from
quiescent cultures of control and TK2-mutated fibroblasts. TP acti-
vity was assayed with 1 m
M thymidine as substrate, and total
(mitochondrial + cytosolic) deoxynucleotidase activity with 5 m
M

[
3
H]dUMP. Enzyme activities are expressed as nanomol of product
formed ⁄ hours of incubation ⁄ mg of protein. Data are mean values
of three different experiments ± standard deviation.
Subject
TP activity
(nmol thymineÆh
)1
Æmg
)1
)
Deoxynucleotidase
activity (nmol
deoxyuridineÆh
)1
Æmg
)1
)
Ca 273 ± 52 2340 ± 240
Cb 587 ± 125 2160 ± 540
Pa 244 ± 87 2160 ± 300
Pb 731 ± 67 2340 ± 240
M. Frangini et al. Thymidine metabolism in TK2-mutated fibroblasts
FEBS Journal 276 (2009) 1104–1113 ª 2009 The Authors Journal compilation ª 2009 FEBS 1107
pool and its specific radioactivity (Fig. 2A,B). As
found earlier in quiescent skin fibroblasts [15], the
addition of thymidine even at such a low concentration
caused an immediate, albeit modest, expansion of the
dTTP pool, with only minor differences being seen

between control and patient cells (Fig. 2A). During the
incubation with 25 nm [
3
H]thymidine, the presence of
1 lm BVDU prevented the expansion of the pool with
the same efficiency in Ca and Pa cells (Fig. 2A), con-
firming the involvement of TK2 in the phosphorylation
of thymidine in both cell lines inferred above (Table 1)
from the inhibition of thymidine phosphorylation
in vitro caused by KIN109.
The specific radioactivity of the dTTP pool reflects
the efficiency with which the salvage of extracellular
thymidine competes with the endogenous de novo
synthesis of dTTP [15,22]. The [
3
H]thymidine supplied
in the medium had a specific radioactivity of
20 000 c.p.m.Æpmol
)1
. After 20 min of incubation, the
dTTP pool had reached a specific radioactivity of
about 7000 c.p.m.Æpmol
)1
in Ca and Pb cells, and
about 5000 in Cb and Pa cells (Fig. 2B), indicating
that one-third of the dTTP of Ca and Pb cells and
between one-quarter and one-fifth of the dTTP of Cb
and Pa cells was derived from salvage of extracellular
thymidine. BVDU decreased the specific radioactivity
of dTTP after 20 min of incubation to 75% in Ca cells

and to 25% in Pa cells (Fig. 2B). As BVDU inhibited
the expansion of the dTTP pool in the two lines to the
same extent (Fig. 2A), the difference in specific radio-
activity suggests that TK2 competed less efficiently
with the de novo synthesis by p53R2 in the TK2-
mutated cells.
To determine the turnover of the dTTP pool, we
performed a pulse–chase experiment with Ca and Pa
fibroblasts. We incubated the cultures with 0.1 lm
[
3
H]thymidine for a total of 90 min (pulse). In the
chase, after 60 min we shifted one-half of the cultures
to medium containing 0.1 lm nonradioactive thymi-
dine and the second half to medium without thymi-
dine. At 5, 15 and 30 min after medium change, we
analyzed, in both series of cultures, the size of the
dTTP pool and its specific radioactivity (Fig. 3). The
pool size doubled during the first 60 min of pulse
(Fig. 3A) and returned to the prepulse value during
the 30 min chase in both control and TK2-mutated
cells in the absence of thymidine. When thymidine was
present during the chase, the decline of the dTTP pool
was only transitory. Under both conditions of chase,
the specific radioactivity of dTTP declined progres-
sively, indicating a turnover of the pool (Fig. 3B). In
the presence of extracellular nonradioactive thymidine,
the half-life was about 14 min in Ca fibroblasts and
18 min in Pa fibroblasts, similar to values observed
earlier in normal lung fibroblasts [7]. In thymidine-free

medium, the half-life of dTTP-specific radioactivity
exceeded 30 min. The slow decay reflected the de novo
synthesis of unlabeled dTTP.
From the changes of the specific radioactivity and
the concentration of dTTP during the chase, we could
calculate the rate of resynthesis of dTTP from thymi-
dine by the procedure described earlier in similar
pulse–chase experiments with TK2-proficient lung
fibroblasts [7]. During the first 15 min of chase and in
A
B
Fig. 2. In situ phosphorylation of [
3
H]thymidine by control and TK2-
mutated fibroblasts. Quiescent cultures of control (Ca and Cb) and
TK2-mutated (Pa and Pb) fibroblasts were incubated for 0 min
(open bars), 5 min (gray bars) and 20 min (black bars) with 25 n
M
[
3
H]thymidine at a specific radioactivity of 20 000 c.p.m.Æpmol
)1
.To
confirm the involvement of TK2 in the in situ phosphorylation of
thymidine, Ca and Pa lines were also incubated in the presence of
1 l
M BVDU added 15 min before [
3
H]thymidine, and the incubation
was continued for 5 min (dotted bars) or 20 min (striped bars). (A)

Size of the total cellular dTTP pool (pmol dTTP per 10
6
cells). (B)
Specific radioactivity of dTTP in the cellular pool (c.p.m.Æ pmol
)1
).
Data are means of determinations from five separate experiments.
Bars indicate standard deviations.
Thymidine metabolism in TK2-mutated fibroblasts M. Frangini et al.
1108 FEBS Journal 276 (2009) 1104–1113 ª 2009 The Authors Journal compilation ª 2009 FEBS
the absence of added thymidine, dTTP was synthesized
at a rate of 0.045 pmolÆ10
)6
cellsÆmin
)1
in Ca cells and
0.054 pmolÆ10
)6
cellsÆmin
)1
in Pa cells. When thymi-
dine was present in the medium, the corresponding
rates were 0.066 and 0.090 pmolÆmin
)1
, respectively.
The higher rates observed in the presence of thymidine
reflect the salvage activity of TK2.
These data confirm once more that the synthesis of
dTTP in the wild-type and TK2-mutated fibroblasts
occurs with virtually the same efficiency, although the

in vitro determination of TK2 activity in cell extracts
shows a 15-fold difference (Table 1). The assay in
Table 1 was performed with 1 lm [
3
H]thymidine. To
better compare the in vivo rate of TK2 activity shown
in Fig. 3 with that measured in vitro , we repeated the
TK2 assay using [
3
H]thymidine at 0.1 lm, i.e. the con-
centration employed in the pulse–chase experiment of
Fig. 3, and expressed the measured enzyme activity in
pmolÆ10
)6
cells. We obtained values of 0.5 pmolÆ min
)1
with Ca extracts and 0.02 pmolÆmin
)1
with Pa extracts,
the same difference between patient and control fibro-
blasts that was observed with higher concentrations
of substrate in Table 1. However, the in vitro value of
TK2 activity in the control extract was 10-fold higher
than the rate of dTTP synthesis measured during the
chase in vivo, whereas this did not occur with the
patient fibroblasts. This remarkable difference suggests
that TK2-proficient fibroblasts contain an excess of
TK2 enzyme whose potential is not fully employed
under normal conditions in intact cells.
Discussion

Thymidine kinases catalyze the rate-limiting step of the
salvage pathway that converts thymidine to dTTP. The
cell contains two such enzymes, the S-phase-specific
cytoplasmic TK1 and mitochondrial TK2. Both kinases
are encoded by nuclear genes present in all cells of the
body. However, the mtDNA depletion caused by
mutations of the TK2 gene appears only in selected
types of cells, especially in skeletal muscle [2]. The
consequent mtDNA depletion syndrome is a severe
disease, generally characterized by early onset and
death in infancy. All the known cases of the syndrome
are caused by mutations that impair but do not com-
pletely abolish TK2 enzyme activity, suggesting that
Fig. 3. Analysis of dTTP pool turnover in
control and TK2-mutated fibroblasts by a
pulse–chase experiment with [
3
H]thymidine.
Quiescent cultures of control (Ca, squares)
and TK2-mutated (Pa, triangles) fibroblasts
were incubated for 90 min with 100 n
M
[
3
H]thymidine. After a 60 min pulse (black
lines), the chase (dotted lines) was started
by shifting part of the cultures to fresh med-
ium without thymidine (open symbols) or
with 100 n
M nonradioactive thymidine

(closed symbols) for 5, 15 and 30 min.
(A) Size of the total cellular dTTP pool
(pmol dTTP ⁄ 10
6
cells). (B) Specific radio-
activity of the dTTP pool (c.p.m. Æ pmol
)1
).
M. Frangini et al. Thymidine metabolism in TK2-mutated fibroblasts
FEBS Journal 276 (2009) 1104–1113 ª 2009 The Authors Journal compilation ª 2009 FEBS 1109
the latter condition is incompatible with life in
humans. In the mouse, complete TK2 knockout leads
to death within 2–4 weeks after birth [9]. The same
reduction of lifespan was observed in mice expressing
a mutated but partially functional TK2 [10].
The fibroblasts analyzed in the present investigation
were derived from patients affected by an unusual
form of TK2 deficiency, characterized by late onset of
the syndrome, relatively slow progression, and long
survival [11–14]. Phenotypic changes during the devel-
opment of the disease suggested the operation of some
undefined molecular mechanism partially compensating
for the metabolic defect.
Skin fibroblasts are, as a rule, unaffected by gene
mutations causing mtDNA depletions in other organs.
They seemed to be a promising cellular system in
which to investigate how cells can cope with a strongly
reduced activity of TK2. The enzyme becomes relevant
for the maintenance of the mitochondrial dTTP pool
when cells have left the cell cycle and the synthesis of

dNTPs in the cytosol is strongly decreased [7,15] due
to proteasome-dependent degradation during mitosis
of TK1 and R2, the S-phase-specific small subunit of
ribonucleotide reductase. Fibroblasts can be main-
tained in a quiescent state in culture for extended times
[7,23], and have been used as in vitro models of TP or
TK2 deficiency [7,15,23].
Only one report exists on the mitochondrial dNTP
pools in TK2-mutated human fibroblasts [17]. The cells
were from two patients homozygous for TK2 mutations
different from those present in the fibroblasts examined
here. In that case, both patient lines contained smaller
mitochondrial dTTP and dCTP pools relative to the
controls, and presented a moderate imbalance of the
overall mitochondrial dNTP pool. No analysis of thymi-
dine metabolism was performed in those cells.
The two mutated fibroblast lines examined here
carry two distinct heterozygote genotypes [12,13] that
produce a TK2 with strongly reduced activity as deter-
mined in whole cell extracts (Table 1). TK2 activity in
extracts of the same patients’ fibroblasts has been
reported previously [12,13]. However, enzyme activity
was determined by an assay based on the competition
between thymidine and deoxycytidine, in one case in
whole cell extracts [13], and in the other [12] in
mitochondrial extracts. Although it is not possible to
compare directly these data with ours, the < 10%
residual TK2 activity that we detected with thymidine
as substrate in Pa extracts (Table 1) agrees with the
1% residual activity measured in mitochondrial

extracts from the same cell line in [12], although the
absolute values differed. In [13], the range of TK2
activity in the controls was identical to ours, but the
activity in patient extracts (corresponding to our Pb
extracts) was 0.5 pmolÆmg
)1
proteinÆmin
)1
, 10-fold
lower than here (Table 1).
The recombinant TK2 proteins encoded by the two
mutated alleles of patient Pb were produced in bacteria
and characterized by Wang et al. [13]. Both mutated
forms of the enzyme showed increased K
m
and reduced
V
max
values, resulting in a 98% reduction of activity
as compared with the recombinant wild-type enzyme
[13], which appeared to correlate directly with the
development of disease in the patient. Our analysis of
the in situ activity of the compound heterozygous
mutant enzyme in the patient fibroblasts provides a
different picture, with no difference between Pb and
control fibroblasts in the phosphorylation of radioac-
tive thymidine (Fig. 2).
Despite their decreased TK2 enzymatic activity, in
both TK2-mutated lines we did not detect any signifi-
cant change in dNTP pool composition either in the

cytosol or in the mitochondria. This normal pool pheno-
type did not result from a metabolic adaptation of the
enzymatic network that regulates the dTTP pools in qui-
escent fibroblasts [15]. We did not find an increased
dependence of the mutated cells on R1 ⁄ p53R2-depen-
dent de novo synthesis, and nor did we observe downre-
gulation of the catabolic enzymes directly involved in
dTTP regulation, i.e. the cytosolic and mitochondrial
deoxynucleotidases and TP (Table 3). When we com-
pared the dynamics of the dTTP pool in mutated and
control fibroblasts by incubating the cells with radio-
active thymidine, in striking contrast to the low TK2
activity measured in cell extracts, the mutated cells phos-
phorylated thymidine as efficiently as the wild-type cells.
This was particularly unexpected for the Pa line, which
apparently contained less than 10% of the control TK2
activity. We concentrated on this cell line, and com-
pared, in a pulse–chase experiment, the turnover of the
dTTP pool with that of the Ca control line, which
expresses a similar level of TP (Table 3). This experi-
ment revealed that the control cells in vivo synthesized
dTTP with the same efficiency as the mutant cells, i.e. at
a rate 10-fold lower than that observed in an in vitro
assay for the phosphorylation of thymidine by TK2,
which is the rate-limiting reaction in the conversion of
thymidine to dTTP. Thus, in the control quiescent cells,
most of the TK2 present is not active, possibly due to
feedback inhibition by dTTP and dCTP [24].
The expression of TK2 increases when cultured fibro-
blasts become quiescent [15], but the enzyme does not

work at its full potential. The present data show that a
small fraction of such potential TK2 activity is
sufficient to maintain the dTTP pools of wild-type
fibroblasts. In the mutants, however, the in vivo and
Thymidine metabolism in TK2-mutated fibroblasts M. Frangini et al.
1110 FEBS Journal 276 (2009) 1104–1113 ª 2009 The Authors Journal compilation ª 2009 FEBS
in vitro rates of TK2 activity coincide, and by fully
exploiting their TK2 complement, the cells preserve
their dTTP pools, which might explain the lack of
mtDNA depletion in these cells. How the mutated
enzymes escape from the inhibitory mechanism operat-
ing on the wild-type enzyme remains to be established.
We were unable to detect feedback inhibition of TK2
by dTTP and dCTP in assays with crude cell extracts.
The TK2-mutated fibroblasts described by Saada
et al. [17] had normal mtDNA content and cytochrome
oxidase activity, and yet were reported to contain lower
pyrimidine dNTP pools in mitochondria. This pheno-
type contrasts with our present finding, and might be
linked to the kind of TK2 mutations involved.
We earlier silenced TK2 by small interfering RNA
transfection of normal skin fibroblasts (the Cb line
used here), obtaining a small (10%) reduction of the
dTTP pool. When thymidine was added to the med-
ium, however, the silenced cells were not able to sal-
vage the nucleoside as efficiently as control cells [15].
The reduction of TK2 activity in the silenced cells in
those experiments was comparable to that found here
in the Pa cells, but it affected the dTTP pool size, in
contrast to the present findings. We believe that this

may depend on the multifactorial control of dNTP
metabolism. A complex network of enzymes modulates
the pools. Interindividual variations in specific compo-
nents of the network between control and affected lines
may mask the effects of single enzyme mutations. The
differences in TP activities detected among the four
fibroblast lines shown in Table 3 exemplify this point.
How do our present data relate to the tissue-specific
phenotype of TK2-associated mtDNA depletion? Skel-
etal muscle is known to contain low levels of TK2
activity [25,26]. Indeed, when myoblasts differentiate
into myotubes, there is no induction of TK2 expres-
sion (C. Rampazzo, unpublished observations). Thus,
it is possible that the physiological level of TK2 in
muscle cells is very close to the minimum required for
pool maintenance, and that when the enzyme is par-
tially inactivated by gene mutations its activity may
fall below the required threshold.
Experimental procedures
Cell lines and cell growth
The TK2-mutated fibroblast lines were derived from
two patients with different compound heterozygous geno-
types, i.e. R152G ⁄ K171del in the case of Pa [12] and
T77M ⁄ R161K in the case of Pb [13]. The cell lines were
obtained by M. R. Vila
`
from skin biopsy specimens taken
from patients at ages 14 years (Pa) and 10 years (Pb). The
age-matched control fibroblast lines were available in the
Padova laboratory. Cells were grown in DMEM with 10%

heat-inactivated fetal bovine serum and antibiotics. Conflu-
ent, contact-inhibited cultures were shifted to fresh medium
with 0.1% serum, and maintained in a quiescent state for
10 days before the experiments. Cells were periodically
checked for mycoplasma contamination by a PCR-based
method (Minerva Biolabs GmbH, Berlin, Germany). We
determined cell numbers with a Coulter counter, and cell
cycle distribution by flow cytometry.
Enzymatic assays
We prepared whole cell extracts as described previously [16],
by adding protease inhibitor mixture to the lysis buffer. The
supernatants were aliquoted and stored at )80 °C until use.
We measured protein concentration by the colorimetric pro-
cedure of Bradford [27], with BSA as standard. All enzyme
assays were done with two different aliquots of extracts to
check for proportionality. We assayed TK2 activity with
0.2 lm [
3
H]BVDU (Moravek Biochemicals and Radiochemi-
cals, Brea, CA, USA) as the substrate [16] in the presence of
50 lm 5-bromouracil (Sigma Aldrich, St Louis, MO, USA)
to inhibit TP, or with 1 lm [
3
H]thymidine (Perkin-Elmer Life
Sciences, Waltham, MA, USA) in the absence or presence of
100 lm KIN109, a specific TK2 inhibitor [16]. We deter-
mined TP activity according to Martı
´
et al. [28], with the
modifications detailed in [15], and total deoxynucleotidase

(cytoplasmic deoxynucleotidase + mitochondrial deoxynu-
cleotidase) activity with 5 mm [
3
H]dUMP as substrate, as
described in [29]. We expressed TK2 activity as pmol prod-
uctÆmin
)1
Æmg
)1
protein, and TP and deoxynucleotidase
activities as nmolÆh
)1
Æmg
)1
protein.
mRNA expression analysis by real-time RT-PCR
To quantify mRNA expression of ribonucleotide reductase
subunits and p53, we performed real-time RT-PCR using
the Applied Biosystems 7500 Real Time PCR System
(Applied Biosystems Inc., Foster City, CA, USA). Total
RNA was obtained from cycling and quiescent cultures of
the four cell lines. Cells were treated with RNase-free
DNase, and the RNA was quantified by spectrophotometry.
Two micrograms of RNA were reversed transcribed using
the High-Capacity cDNA Archive Kit (Applied Biosystems
Inc.), following the manufacturer’s instructions. A 5 lL
aliquot of cDNA diluted 1 : 50 was mixed with the TaqMan
Universal PCR Master Mix and the following Gene Expres-
sion Assays Taqman probes: p53 (TP53, Hs00153349), ribo-
nucleotide reductase R1 subunit (RRM1, Hs00168784_m1),

ribonucleotide reductase R2 subunit (RRM2, Hs
Hs00357247_g1), and the p53-dependent subunit 2 of ribo-
nucleotide reductase (RRM2B, Hs00153085_m1). The rela-
tive quantity of mRNA was normalized using cyclophilin A
(PPIA, Hs99999904_m1) as endogenous control. The PCR
M. Frangini et al. Thymidine metabolism in TK2-mutated fibroblasts
FEBS Journal 276 (2009) 1104–1113 ª 2009 The Authors Journal compilation ª 2009 FEBS 1111
cycle consisted of an initial step of 50 °C for 2 min, fol-
lowed by 10 min at 95 °C, and 40 repetitions of two-step
cycles of 50 °C for 15 s and 60 °C for 1 min. All assays
were performed at least in triplicate. PCR data were pro-
cessed by genamp 7500 SDS software.
Immunoblotting
The cell protein lysates were prepared and quantified as
described in [23]. The conditions for gel electrophoresis and
immunoblotting and the antibodies employed were as
detailed in [15].
Isotope experiments and dNTP pool analyses
The procedures for the isotope experiments and the separa-
tion of cytosolic and mitochondrial dNTP pools were
detailed previously [22]. Before starting the incubation with
[
3
H]thymidine (20 000 c.p.m.Æpmol
)1
), we substituted the
medium with fresh medium with 0.1% dialyzed serum and
left the cultures to equilibrate for 1–2 h. All manipulations
were performed in a 37 °C room to avoid thermal shocks.
Incubations with 25 nm [

3
H]thymidine were performed for
5 and 20 min. In one protocol, 1 lm BVDU (Sigma) was
added 15 min before the radioactive nucleoside. For the
pulse–chase experiment with 100 nm [
3
H]thymidine, we
applied the protocol described in [7]. In all experiments,
[
3
H]thymidine incubations were ended by moving the cells
on ice to a cold room, and total dNTP pools were extracted
with ice-cold 60% methanol for 1 h from the cells still
attached to the plates [18]. Cytosolic and mitochondrial
dNTPs were separated by differential centrifugation of cell
homogenates [22] and extracted with 60% methanol. The
sizes of dNTP pools and the specific radioactivity of dTTP
were determined with a DNA polymerase-based assay
[22,30] with the modifications reported in [22], incubating
two different aliquots of pool extracts for 1 h at 37 °Cina
reaction mix containing 0.25 lm [
32
P]dATP and 0.2 units of
Klenow enzyme.
Acknowledgements
This work was supported by grants from Italian
Telethon (Grant GGP05001), AIRC, the Italian
Association for Cancer Research, and the Cariparo
Foundation to V. Bianchi, and from the Spanish Insti-
tuto de Salud Carlos III (PI 06 ⁄ 0735 and CP 04 ⁄ 0240

to R. Martı
´
and PI 04 ⁄ 0415 to M. R. Vila
`
).
References
1 Copeland WC (2008) Inherited mitochondrial
diseases of DNA replication. Annu Rev Med 59,
131–146.
2 Saada A, Shaag A, Mandel H, Nevo Y, Eriksson S &
Elpeleg O (2001) Mutant mitochondrial thymidine
kinase in mitochondrial DNA depletion myopathy. Nat
Genet 29, 342–344.
3 Mandel H, Szargel R, Labay V, Elpeleg O, Saada A,
Shalata A, Anbinder Y, Berkowitz D, Hartman C,
Barak M et al. (2001) The deoxyguanosine kinase gene
is mutated in individuals with depleted hepatocerebral
mitochondrial DNA. Nat Genet 29, 337–341.
4 Nishino I, Spinazzola A & Hirano M (1999) Thymidine
phosphorylase gene mutations in MNGIE, a human
mitochondrial disorder. Science 283, 689–692.
5 Bourdon A, Minai L, Serre V, Jais JP, Sarzi E, Aubert
S, Chretien D, de Lonlay P, Paquis-Flucklinger V,
Arakawa H et al. (2007) Mutation of RRM2B, encod-
ing p53-controlled ribonucleotide reductase (p53R2),
causes severe mitochondrial DNA depletion. Nat Genet
39, 776–780.
6 Chabes AL, Pfleger CM, Kirschner MW & Thelander L
(2003) Mouse ribonucleotide reductase R2 protein: a new
target for anaphase-promoting complex-Cdh1-mediated

proteolysis. Proc Natl Acad Sci USA 100, 3925–3929.
7 Pontarin G, Ferraro P, Valentino ML, Hirano M, Rei-
chard P & Bianchi V (2006) Mitochondrial DNA deple-
tion and thymidine phosphate pool dynamics in a
cellular model of mitochondrial neurogastrointestinal
encephalomyopathy. J Biol Chem 281, 22720–22728.
8 Oskoui M, Davidzon G, Pascual J, Erazo R, Gurgel-
Giannetti J, Krishna S, Bonilla E, De Vivo DC, Shan-
ske S & DiMauro S (2006) Clinical spectrum of mito-
chondrial DNA depletion due to mutations in the
thymidine kinase 2 gene. Arch Neurol 63, 1122–1126.
9 Zhou X, Solaroli N, Bjerke M, Stewart JB, Rozell B,
Johansson M & Karlsson A (2008) Progressive loss of
mitochondrial DNA in thymidine kinase 2-deficient
mice. Hum Mol Genet 17, 2329–2335.
10 Akman HO, Dorado B, Lo
´
pez LC, Garcı
´
a-Cazorla A,
Vila
`
MR, Tanabe LM, Dauer WT, Bonilla E, Tanji K
& Hirano M (2008) Thymidine kinase 2 (H126N) knoc-
kin mice show the essential role of balanced deoxy-
nucleotide pools for mitochondrial DNA maintenance.
Hum Mol Genet 17, 2433–2440.
11 Barthe
´
le

´
my C, Ogier de Baulny H, Diaz J, Cheval MA,
Frachon P, Romero N, Goutieres F, Fardeau M &
Lombe
`
s A (2001) Late-onset mitochondrial DNA deple-
tion: DNA copy number, multiple deletions, and com-
pensation. Ann Neurol 49 , 607–617.
12 Vila
`
MR, Segovia-Silvestre T, Gamez J, Marina A, Naini
AB, Meseguer A, Lombe
`
s A, Bonilla E, Di Mauro S, Hir-
ano M et al. (2003) Reversion of mtDNA depletion in a
patient with TK2 deficiency. Neurology 60, 1203–1205.
13 Wang L, Limongelli A, Vila MR, Carrara F, Zeviani M
& Eriksson S (2005) Molecular insights into mitochon-
drial DNA depletion syndrome in two patients with
Thymidine metabolism in TK2-mutated fibroblasts M. Frangini et al.
1112 FEBS Journal 276 (2009) 1104–1113 ª 2009 The Authors Journal compilation ª 2009 FEBS
novel mutations in the doexyguanosine kinase and thy-
midine kinase 2 genes. Mol Genet Metab 84, 75–82.
14 Vila
`
MR, Villaroya J, Garcia-Arumi E, Castellote A,
Meseguer A, Hirano M & Roig M (2008) Selective mus-
cle fiber loss and molecular compensation in mitochon-
drial myopathy due to TK2 deficiency. J Neurol Sci
267, 137–141.

15 Rampazzo C, Fabris S, Franzolin E, Crovatto K, Fran-
gini M & Bianchi V (2007) Mitochondrial thymidine
kinase and the enzymatic network regulating thymidine
triphosphate pools in cultured human cells. J Biol Chem
282, 34758–34769.
16 Franzolin E, Rampazzo C, Perez-Perez MJ, Hernandez
AI, Balzarini J & Bianchi V (2006) Bromovinyl-deoxy-
uridine: a selective substrate for mitochondrial thymi-
dine kinase in cell extracts. Biochem Biophys Res
Commun 344, 30–36.
17 Saada A, Ben-Shalom E, Zyslin R, Miller C, Mandel H
& Elpeleg O (2003) Mitochondrial deoxyribonucleoside
triphosphate pools in thymidine kinase 2 deficiency.
Biochem Biophys Res Commun 310, 963–966.
18 Ferraro P, Pontarin G, Crocco L, Fabris S, Reichard P
& Bianchi V (2005) Mitochondrial deoxynucleotide
pools in quiescent fibroblasts: a possible model for
mitochondrial neurogastrointestinal encephalomyopathy
(MNGIE). J Biol Chem 280, 24472–24480.
19 Bianchi V, Pontis E & Reichard P (1987) Regulation of
pyrimidine deoxyribonucleotide metabolism by sub-
strate cycles in dCMP-deaminase deficient V79 hamster
cells. Mol Cell Biol 7, 4218–4224.
20 Reichard P (1988) Interactions between deoxyribonucleo-
tide and DNA synthesis. Annu Rev Biochem 57, 349–374.
21 Bianchi V & Spychala J (2003) Mammalian 5¢-nucleo-
tidases. J Biol Chem 278, 46195–46198.
22 Pontarin G, Gallinaro L, Ferraro P, Reichard P &
Bianchi V (2003) Origins of mitochondrial thymidine
triphosphate: dynamic relations to cytosolic pools.

Proc Natl Acad Sci USA 100, 12159–12164.
23 Pontarin G, Ferraro P, Ha
˚
kansson P, Thelander L,
Reichard P & Bianchi V (2007) p53R2-dependent
ribonucleotide reduction provides deoxyribonucleotides
in quiescent human fibroblasts in the absence of
induced DNA damage. J Biol Chem 282, 16820–
16828.
24 Wang L, Saada A & Eriksson S (2003) Kinetic proper-
ties of mutant human thymidine kinase 2 suggest a
mechanism for mitochondrial DNA depletion myopa-
thy. J Biol Chem 278, 6963–6968.
25 Saada A, Shaag A & Elpeleg O (2003) mtDNA deple-
tion myopathy: elucidation of the tissue specificity in
the mitochondrial thymidine kinase (TK2) deficiency.
Mol Genet Metab 79, 1–5.
26 Rylova SN, Mirzaee S, Albertioni F & Eriksson S
(2007) Expression of deoxynucleoside kinases and
5¢-nucleotidases in mouse tissues: implications for mito-
chondrial toxicity. Biochem Pharmacol 74, 169–175.
27 Bradford MM (1976) A rapid and sensitive method for
the quantitation of microgram quantities of protein
utilizing the principle of protein-dye binding. Anal
Biochem 72, 248–254.
28 Martı
´
R, Spinazzola A, Tadesse S, Nishino I, Nishigaki
Y & Hirano M (2004) Definitive diagnosis of mitochon-
drial neurogastrointestinal encephalomyopathy by bio-

chemical assays. Clin Chem 50, 120–124.
29 Rampazzo C, Mazzon C, Reichard P & Bianchi V
(2002) 5
¢-Nucleotidases: specific assays for five different
enzymes in cell extracts. Biochem Biophys Res Commun
293, 258–263.
30 Sherman PA & Fyfe JA (1989) Enzymatic assay for
deoxyribonucleoside triphosphates using synthetic
oligonucleotides as template primers. Anal Biochem 180,
222–226.
M. Frangini et al. Thymidine metabolism in TK2-mutated fibroblasts
FEBS Journal 276 (2009) 1104–1113 ª 2009 The Authors Journal compilation ª 2009 FEBS 1113