Báo cáo khoa học: Evidence of a new phosphoryl transfer system in nucleotide metabolism doc

Evidence of a new phosphoryl transfer system in
nucleotide metabolism
Daniela Vannoni
1,
*, Roberto Leoncini
1,
*, Stefania Giglioni
1
, Neri Niccolai
2
, Ottavia Spiga
2
, Emilia
Aceto
1
and Enrico Marinello
1
1 Department of Internal Medicine, Endocrine-Metabolic Sciences and Biochemistry, University of Siena, Italy
2 Department of Molecular Biology, University of Siena, Italy
Purine nucleotides are precursors of nucleic acids and
participate in many metabolic pathways as substrates,
coenzymes and energy sources. Much attention has
been focused on the roles and intracellular levels of
AMP, ADP and ATP in various tissues under normal
and pathological conditions [1–3]. Their relationships
to genetic diseases, blood disorders, drugs, tumours
and other pathologies have been studied extensively
[4–6].
Although ATP formation occurs via several well-
known mechanisms, ADP is thought to be formed
only from ATP, either by the adenylate kinase reaction

or by ATPase-mediated hydrolysis. The possibility that
ADP can be formed by de novo synthesis from
low-energy precursors has never been investigated
fully. Even less studied is the possibility that ADP
might be synthesized from low-energy precursors under
special conditions, such as ischaemia and hypoxia, in
which massive depletion of ATP and elevation of
AMP are known to occur [7–10].
In this article, it is shown that, under physiological
conditions, ADP may be formed in mammalian tissues
by the disproportionation of AMP, consistent with an
AMP–AMP phosphotransferase reaction. This reaction
is carried out by enzymes of purine metabolism which,
under specific cellular conditions, associate in a biolog-
ical network and cooperate in a reaction not reported
previously.
Keywords
adenosine deaminase; adenosine kinase;
adenylate kinase; ADP; ATP
Correspondence
R. Leoncini, Department of Internal
Medicine, Endocrine-Metabolic Sciences
and Biochemistry, Via A. Moro 2,
53100 Siena, Italy
Fax: +39 0577 234285
Tel: +39 0577 234287
E-mail:
*These authors contributed equally to this
work
(Received 16 July 2008, revised 24

September 2008, accepted 5 November
2008)
doi:10.1111/j.1742-4658.2008.06779.x
Crude rat liver extract showed AMP–AMP phosphotransferase activity
which, on purification, was ascribed to a novel interaction between adeny-
late kinase, also known as myokinase (EC 2.7.4.3), and adenosine kinase
(EC 2.7.1.20). The activity was duplicated using the same enzymes purified
from recombinant sources. The reaction requires physical contact between
myokinase and adenosine kinase, and the net reaction is aided by the pres-
ence of adenosine deaminase (EC 3.5.4.4), which fills the gap in the energy
balance of the phosphoryl transfer and shifts the equilibrium towards ADP
and inosine synthesis. The proposed mechanism involves the association of
adenosine kinase and myokinase through non-covalent, transient interac-
tions that induce slight conformational changes in the active site of
myokinase, bringing two already bound molecules of AMP together for
phosphoryl transfer to form ADP. The proposed mechanism suggests a
physiological role for the enzymes and for the AMP–AMP phosphotrans-
ferase reaction under conditions of extreme energy drain (such as hypoxia
or temporary anoxia, as in cancer tissues) when the enzymes cannot display
their conventional activity because of substrate deficiency.
Abbreviations
ADA, adenosine deaminase; AdK, adenosine kinase; CE, capillary electrophoresis; MD, molecular dynamics; MK, myokinase.
FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS 271
These enzymes are well known for their conven-
tional reactions:
(a) Adenylate kinase, also called myokinase (MK),
which catalyses the reversible reaction:
AMP þ ATP $ ADP
(b) Adenosine kinase (AdK), which catalyses the
irreversible reaction:

Adenosine þ ATP ! AMP þ ADP
(c) Adenosine deaminase (ADA), which catalyses the
irreversible reaction:
Adenosine þ H
2
O ! inosine þ NH
3
The AMP–AMP phosphotransferase reaction is:
AMP þ AMP ! ADP þ inosine þ NH
3
ADP and inosine were identified by various meth-
ods, including HPLC and diode array analysis; the
enzymes responsible for the reaction were identified
through a complex purification procedure. Our find-
ings demonstrate that MK and AdK carried out the
reaction, and ADA enhanced the rate. ATP was
formed only in the presence of ADA, and at longer
incubation times, when the ADP concentration passed
the threshold necessary to allow the initiation of the
MK reaction and the AMP concentration decreased
below the inhibitory concentration for MK. In this
study, we investigated the complex mechanisms under-
lying this reaction and its physiological role in cell
metabolism.
Results
AMP–AMP phosphotransferase reaction
Dialysed rat liver supernatant was incubated with
AMP and Mg
2+
; HPLC analysis revealed the forma-

tion of two products with retention times of 1.3 and
7.1 min, which were identified as inosine and ADP,
respectively (Fig. 1). Using purified [
32
P]AMP as a sub-
strate, ADP formation was consistent with the transfer
of
32
P between two molecules of [
32
P]AMP to form
[
32
P]ADP[a,bP]. Thus, starting from [
32
P]AMP with a
specific radioactivity of 50 500 ± 1517 d.p.m.Ænmol
)1
(mean of five experiments), we obtained ADP with a
specific radioactivity of 108 125 ± 4325 d.p.m.Ænmol
)1
,
double that of the starting AMP. Hydrolysis of the
product yielded two moles of
32
P
i
per mole of ADP.
These findings indicate that supernatants starting
from low-energy precursors catalyse the AMP–AMP

phosphotransferase reaction:
AMP þ AMP ! ADP þ inosine þ NH
3
ð1Þ
resulting from reactions (2) and (3):
AMP þ AMP ! ADP þ adenosine ð2Þ
Adenosine þ H
2
O ! inosine þ NH
3
ðADA reactionÞð3Þ
ADA converts adenosine to inosine and ammonia in
stoichiometric amounts with respect to ADP.
Enzyme purification and identification by mass
spectrometry
Protein purification demonstrated that ADP forma-
tion occurred via the activities of MK and AdK, with
the cooperation of ADA. The crude supernatant and
P90d fraction showed AMP–AMP phosphotransferase
activity, but any further chromatography led to the
loss of activity, presumably because the two different
proteins were separated from one another. Therefore,
we purified each protein individually by an appro-
priate procedure. AMP–AMP phosphotransferase
activity was restored every time we combined the two
separate protein preparations. ADA was also isolated.
When ADA was added to the assay mixture, AMP–
AMP phosphotransferase activity was greatly
enhanced.
The purifications were performed as reported in

Table 1. The final SDS-PAGE showed a single band
for each protein preparation.
Fractions X2 and Y2 were identified as MK and
AdK, respectively, by electrospray mass spectrometry.
The X2 fraction spectra revealed a component with a
molecular mass of 26 232.5 ± 0.5 Da (Fig. 2). Twenty
signals ranging in mass from m ⁄ z 780.7 to m ⁄ z 1726.8
were selected and entered into the non-redundant
National Center for Biotechnology Information data-
base using pro found software. The query returned a
Fig. 1. Identification of ADP and inosine in the reaction mixture.
The figure shows the typical HPLC pattern of an assay mixture
incubated with crude extract of rat liver (0.5 mg) for 50 min.
Numbered peaks: 1, inosine; 2, AMP; 3, ADP.
Formation of ADP from AMP D. Vannoni et al.
272 FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS
significant match with the protein MK (isoenzyme 2)
from Rattus norvegicus. The mass signals used in the
search procedure accounted for 72% of the entire MK
sequence, confirming identification. The electrospray
mass spectra of Y2 revealed a component with a
molecular mass of 38 344.2 ± 2.1 Da (Fig. 3). Twenty
signals ranging in mass from m ⁄ z 1235.94 to m ⁄ z
3316.66 were selected, and the query returned a highly
significant match with the protein AdK from R. nor-
vegicus. The mass signals used in the search procedure
accounted for 80% of the entire AdK sequence,
confirming identification. Mass analysis of fractions
X2 and Y2 confirmed the purity of the two purified
proteins (Figs 2 and 3).

AMP–AMP phosphotransferase activity by
recombinant enzymes
One microgram of Escherichia coli-expressed human
recombinant AdK, mixed with commercial recombi-
nant MK and ADA, exhibited AMP–AMP phospho-
transferase activity. Commercial preparations of
purified MK and ADA from several sources produced
the same result as the purified rat liver enzymes (AdK
was not commercially available).
General properties of the reaction
AMP–AMP phosphotransferase activity was detectable
in different rat tissues (muscle, brain, spleen) and was
absolutely specific for AMP. The formation of ADP
and inosine progressed over time and no products were
formed when the protein extract was denatured by
heat or acid. The time course of the reaction was linear
for at least 60 min when only MK and AdK were
present. When ADA was added, the shape of the curve
remained as before for the first 15 min; thereafter, the
reaction proceeded linearly for 50 min, at a rate 20
times greater than that for MK and AdK alone
(Fig. 4).
ADP and inosine formation varied according to the
incubation temperature, pH, AMP concentration and
Mg
2+
concentration. Maximum activity was found at
pH 6.5–6.8, and the activity decreased sharply at pH
values above 7.5 and below 5.5. The K
m

value for
AMP was 0.8 mm (Fig. 5A). Mg
2+
was essential for
the AMP–AMP phosphotransferase reaction, with the
optimal concentration in the range 0.8–1.5 mm. The
apparent K
m
value of Mg
2+
was 0.35 mm (Fig. 5B).
Certain ions added to the incubation mixture inhibited
(NH
þ
4
,Li
+
and SO
2À
4
) or had no effect (Ca
2+
and
Table 1. Purification of rat liver X2 and Y2 proteins. The two proteins responsible for AMP–AMP phosphotransferase activity in rat liver
were purified. The procedure started with a common trunk and had three steps: supernatant production, ammonium sulfate precipitation
and dialysis. During these steps, the two proteins were not separated and phosphotransferase activity was present in all fractions. After
DE-52 chromatography, the purification process diverged: non-retained fractions were utilized for the purification of protein X (X1 or X2;
finally identified as MK) and part of the retained fractions was used for the purification of protein Y (Y1 or Y2; finally identified as AdK).
AMP–AMP phosphotransferase activity was only detected when suitable amounts (0.003–0.5 mg) of fractions from each purification branch
were pooled (i.e. X1 + Y1, X2 + Y2).Total AMP–AMP phosphotransferase activity is expressed in IU, which corresponds to the number of

micromoles of ADP or ADP + ATP formed per minute. ND, AMP–AMP phosphotransferase activity was not detectable.
Common trunk
Step
Supernatant
P90d
DE-52
Retained (R)
X protein
X1 Blue Sepharose
X2 Superdex 75
Y protein
Y1 AMP Sepharose
X1 + Y1
Y2 + Y2
Y2 Superdex 75
Not retained (NR)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
D. Vannoni et al. Formation of ADP from AMP
FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS 273

PO
2À
3
) on the reaction. Zn
2+
completely eliminated
phosphotransferase activity.
The amounts and ratios of enzymes in the assay
mixture reflected those reported to be present in rat
liver, namely MK @ 0.45 IU and AdK @ 0.015 IU,
with an MK : AdK ratio of approximately 30 [11,12].
In our assay mixture, we used 1–3 lg of pure protein,
corresponding to approximately 0.21–0.63 IU MK and
0.006–0.018 IU AdK, with a ratio of about 10–30.
When ADA was present in the assay mixture, ATP
was formed when the incubation time exceeded
15–20 min, and increased with time. ATP, with a
retention time of 9.2 min on the HPLC chromatogram,
was identified using the same criteria as for ADP.
Using [
32
P]AMP as a substrate, the ATP formed had a
specific radioactivity of 146 338 ± 7316 d.p.m.Ænmol
)1
,
three times that of AMP. ATP was formed by the con-
ventional MK reaction, when the ADP concentration
in the mixture passed the threshold of 0.05 mm (data
not shown); at lower ADP concentrations, MK activity
was inefficient because of a lack of substrate. At longer

incubation times, when the ADP concentration
exceeded the K
m
value for MK (0.3 mm) and the AMP
concentration fell below the inhibitory concentration
(K
i
of AMP for MK, 2.13 mm), MK exerted its con-
ventional activity, eventually reaching equilibrium [12].
The K
i
value of AMP for MK was not influenced by
the presence of AdK.
Reversibility of the reactions
We considered the reversibility of the reactions:
AMP þ AMP !
ðAdK;MK;ADAÞ
ADP þ inosine þ NH
3
ð1Þ
AMP þ AMP !
ðAdK;MKÞ
ADP þ adenosine ð2Þ
Reaction (1) was irreversible and reaction (2) also
appeared to be irreversible; in the presence of ADP
and adenosine alone, MK activity was absolutely
A
B
Fig. 2. Electrospray ionization mass spectra of the X2 fraction identified as MK. (A) Gaussian-type distribution of multiply charged ions. (B)
The m ⁄ z spectrum converted to a molecular mass profile by maximum entropy processing. The mass profile is dominated by a single com-

ponent, showing the purity grade of the protein. The molecular mass of the sample was calculated by the processing software associated
with the mass spectrometer to a mass accuracy of 0.01%, and can be read directly on the graph.
Formation of ADP from AMP D. Vannoni et al.
274 FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS
predominant, and so equimolar amounts of AMP and
ATP were formed by the very rapid conventional MK
reaction [see reaction (a) in the introductory section].
Moreover, if we added 0.2 lCi of [
14
C]adenosine to
the incubation mixture, no [
14
C]AMP was formed.
Reaction mechanism studies – cooperation
between AdK and MK
Micro-equilibrium dialysis experiments
No reaction products were formed when AdK and
MK were separated by a dialysis membrane during the
reaction, regardless of the presence of ADA. The
AMP–AMP phosphotransferase reaction could only be
detected when AdK and MK were incubated in the
same chamber, in which case the presence of ADA
only affected the rate of the reaction.
Detection of reaction intermediates by HPLC, capillary
electrophoresis (CE) and NMR analysis
HPLC, CE and NMR analysis of the incubation mix-
ture at several points during the incubation indicated
that no intermediate products were formed. In all cases,
we observed a decrease in AMP concentration over
time and an increase in adenosine or inosine, ADP and

ATP concentrations; no other products were detected.
Enzyme–phosphate intermediate trapping
When enzyme–phosphate intermediate trapping experi-
ments were performed, no spot was detected on the
autoradiography slide, indicating direct transfer of
the phosphoryl group from one AMP molecule to
another, without the formation of an intermediate
A
B
Fig. 3. Electrospray ionization mass spectra of the Y2 fraction identified as AdK. (A) Gaussian-type distribution of multiply charged ions. (B)
The m ⁄ z spectrum converted to a molecular mass profile using maximum entropy processing. The mass profile is dominated by a single
component, showing the purity grade of the protein. The molecular mass of the sample was calculated by the processing software associ-
ated with the mass spectrometer to a mass accuracy of 0.01%, and can be read directly on the graph.
D. Vannoni et al. Formation of ADP from AMP
FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS 275
phosphoenzyme species. When vanadate was added to
the reaction mixture, no inhibition of AMP–AMP phos-
photransferase was observed. Moreover, the addition of
[
14
C]nucleoside (adenosine, guanosine, inosine) did not
promote the formation of [
14
C]AMP, [
14
C]GMP or
[
14
C]IMP. The formation of ADP was unchanged.
Gel filtration and SDS-PAGE

Assay mixtures incubated for 0 and 60 min were sub-
mitted to gel filtration. The resulting chromatograms
showed no differences in retention times, indicating
that no stable protein complex was formed. Moreover,
when the same samples were resolved by SDS-PAGE,
no bands were observed at a molecular mass higher
than AdK, ruling out the formation of covalent bonds
between the proteins.
Docking simulation studies
We performed a molecular dynamics (MD) simulation
of the interaction between AMP and AdK or MK
using autodock. This procedure suggested that, from
an energy point of view, each protein had two binding
sites, either of which could be occupied by AMP.
gromacs was then used to optimize the molecular
models with energy minimizations followed by a 1 ns
MD simulation. Using the MD trajectory, possible
changes in AMP position and in the network of bonds
between the AMP molecules and binding pockets were
examined. In the case of rat MK, one of the two
bound AMP molecules maintained the same location
and orientation during all MD runs, whereas the other
molecule changed position and approached the first
AMP molecule. Indeed, the distance between these two
molecules was 8.19 A
˚
at the beginning of the docking
simulation, and 5.61 A
˚
at the end of the simulation

(Fig. 6B), close to the 4.5 A
˚
distance between the natu-
ral ligands AMP and ATP (Fig. 6A). With regard to
AdK, the MD simulation showed that the two AMP
molecules were bound in positions very distant from
each other (Fig. 7).
Kinetic analysis
Kinetic experiments were performed in the absence
of ADA. AMP–AMP phosphotransferase activity
9
A
B
6
nmolesnmoles
3
0
0
200
150
100
50
0
10 20 30
Minutes
Minutes
40 50 60
0102030405060
Fig. 4. Time course of ADP formation by the AMP–AMP phospho-
transferase reaction. (A) 0.38 IU of purified rat liver MK and

0.012 IU of AdK. The amount of ADP produced (nmol) is shown on
the ordinate. (B) 0.38 IU of purified rat liver MK, 0.012 IU of AdK
and 0.9 IU of ADA. The amount of ADP + ATP formed (nmol) is
shown on the ordinate.
Fig. 5. (A) Direct and double reciprocal plot of initial velocities with
variable AMP concentration (0.0–6.0 m
M) at a constant Mg
2+
concentration (1.0 mM). (B) Direct plot of the initial velocities with
variable Mg
2+
concentration (0.1–1.4 mM) and constant AMP
concentration (4.0 m
M). Values are the mean ± standard deviation
of five experiments. Assay mixtures contained 0.15 IU MK,
0.004 IU AdK and 0.3 IU ADA. The amount of ADP + ATP formed
(nmolÆh) is shown on the ordinate.
Formation of ADP from AMP D. Vannoni et al.
276 FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS
gradually increased to a plateau, following a typical
hyperbolic curve (Fig. 8), and reached a maximum
saturating value when the concentration of AdK
exceeded 2 nmol (AdK : MK ratio > 2000). The K
d
value was 1.496 lm, as determined by the Scatchard
equation [13].
Inhibition experiments
The results of the inhibition assays are presented in
Table 2. Each inhibitor completely blocked the
activity of its respective enzyme, i.e. Ap

5
A inhibited
MK and A134974 inhibited AdK. Neither inhibitor
interfered with the progress of the other reaction.
AMP–AMP phosphotransferase activity was only
affected by the presence of Ap
5
A, when it fell to
zero. A134974 had only a slight effect on the
AMP–AMP phosphotransferase reaction (< 10%
inhibition).
AMP–AMP phosphotransferase reaction in human
colorectal mucosa from cancer patients
Table 3 shows the activities of AdK, MK, ADA and
AMP–AMP phosphotransferase in normal and cancer-
ous human colorectal mucosa. The MK activity did
not vary, but the AdK and ADA activities were signifi-
cantly elevated (P < 0.0001) in cancer tissue with
respect to the surrounding normal mucosa. AMP–
AMP phosphotransferase activity was only detectable
in tumour tissue.
A
B
Fig. 6. (A) Backbone superimposition of human (green) and rat
(red) MK. The natural ligands (AMP and ATP) and the two AMP
molecules (red) are shown in bold. (B) Backbone superimposition of
rat MK before (magenta) and after (green) the MD run. Note the
reduction in the distance between the two ligands.
Fig. 7. Backbone superimposition of human (blue) and rat
(magenta) AdK. The natural ligands (ADA and ATP, blue) and the

two AMP molecules (magenta) are shown in bold.
Fig. 8. Direct plot of AMP–AMP phosphotransferase activity (as
percentage of total activity) in mixtures containing various amounts
of AdK (0.1–4.0 nmol in a final volume of 0.25 mL) and a fixed
amount of MK (0.0008 nmol) in the absence of ADA. Inset:
Scatchard plot.
D. Vannoni et al. Formation of ADP from AMP
FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS 277
Discussion
Cooperation between the three enzymes
Our data demonstrate that the AMP–AMP phospho-
transferase reaction occurs by cooperation between
MK and AdK specifically, and is enhanced by ADA;
none of the three enzymes could be substituted by
others.
ADA strongly accelerates the AMP–AMP phospho-
transferase reaction, carrying out the coupled reaction
(3) (adenosine fi inosine + NH
3
). ADA does not
participate directly in ADP synthesis, but its coupled
reaction enables the subtraction of a reaction product,
helping to drive the reaction forwards. Moreover, the
exergonic formation of inosine and ammonia from
adenosine fills the gap in the energy balance of
phosphoryl transfer (DG° ranges from )4to)10 -
kcalÆmol
)1
) [14]. ADA has a K
eq

value of approxi-
mately 10
5
[15], a low K
m
value and a high efficiency
close to the diffusion-limited rate [16], and is found in
most mammalian tissues [17]. It follows that the
AMP–AMP phosphotransferase reaction always bene-
fits from the presence of ADA in vivo.
The AMP–AMP phosphotransferase reaction is irre-
versible. Any attempt at producing AMP by incubat-
ing ADP, inosine and NH
3
together is ineffective. In
the absence of ADA, the addition of [
14
C]adenosine to
the mixture does not produce [
14
C]AMP.
Demonstrating the association between AdK and
MK was the key to understanding the mechanism of
AMP–AMP phosphotransferase. We obtained the
following results.
(a) No free intermediates were formed during any
incubation experiment (that is, no transient dinucleo-
tide compound), in contrast with the NADase reaction
[18]. HPLC, CE and NMR analysis produced no
evidence of a reaction intermediate.

(b) The existence of an enzyme–phosphate interme-
diate, as is formed with 5¢-nucleotidase [19], was ruled
out. No direct evidence of an enzyme–phosphate com-
plex was found. Experiments using vanadate or the
addition of nucleoside also yielded negative results. We
found no evidence of AdK–phosphate or MK–phos-
phate complex formation in the literature.
(c) Microdialysis demonstrated that MK and AdK
must be able to associate and work together. When
they were physically separated, no ADP was formed.
(d) Electrophoresis and gel filtration experiments
excluded covalent or similarly strong interactions.
(e) In silico simulations suggested that MK contains
the active site of the AMP–AMP phosphotransferase
reaction, but raised the issue of how the two AMP
molecules achieve the correct distance for interaction,
and the role of AdK in the reaction.
(f) Inhibition experiments confirmed the role of MK
in the reaction. Ap
5
A inhibited MK and AMP–AMP
phosphotransferase activities, whereas A134974 only
inhibited AdK activity, indicating that the active site
of AdK is not essential for the AMP–AMP phospho-
transferase reaction.
(g) Kinetic experiments (Fig. 8) demonstrated that
interactions occurred between the two proteins. When
we fixed the amount of MK at a low concentration
and increased the concentration of AdK, we obtained
a hyperbolic curve with a saturation trend resembling

protein–protein interaction. Indeed, isothermal curves,
such as the enzyme–substrate curves of Michaelis–
Menten, hormone–receptor curves and antigen–anti-
body affinity curves, all represent an association
between two different molecules and, in the last two
cases, between two different proteins [20]. In all of
these curves, the ordinate values indicate the amount
of dimer formed. In the case of Michaelis–Menten
plots, the value of V is related to the enzyme–substrate
complex; in the case of hormone–receptor curves, the
ratio B ⁄ B
max
represents the amount of receptor joined
Table 2. Rat liver MK, AdK and AMP–AMP phosphotransferase
specific activities, expressed as lmolÆ(min mg)
)1
(means ± stan-
dard deviation of eight experiments), were tested in the presence
or absence of 0.25 l
M Ap
5
A (specific inhibitor of MK) or 0.1 nM
A134974 (specific inhibitor of AdK). The mixtures contained 0.2 IU
MK, 0.02 IU AdK and 2 IU adenosine deaminase (the latter only in
the AMP–AMP phosphotransferase assay).
Assay Specific activity
MK activity 62.15 ± 1.59
MK activity in the presence of Ap
5
A0

MK activity in the presence of A134974 61.42 ± 1.18
AdK activity 5.42 ± 0.3
AdK activity in the presence of A134974 0
AdK activity in the presence of Ap
5
A 5.46 ± 0.3
Table 3. MK, AdK, ADA and AMP–AMP phosphotransferase activi-
ties in normal and human colorectal cancer mucosa were assayed.
Partially purified protein preparations (20, 40, 80 and 100 lg,
respectively) were incubated. The activities of AdK, ADA and AMP–
AMP phosphotransferase are expressed as lmolÆ(h mg)
)1
, and are
the mean activity ± standard deviation from 10 different patients.
*P < 0.0001.
Activity Normal mucosa Tumour mucosa
AdK 0.010 ± 0.004 0.021 ± 0.009*
MK 25.89 ± 9.72 22.41 ± 7.81
ADA 0.441 ± 0.156 0.657 ± 0.253*
Formation of ADP from AMP D. Vannoni et al.
278 FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS
to the hormone with respect to the maximum complex
possible. In our case, the ordinate shows the formation
of ADP, which represents the number of dimers
formed between the two proteins. In all of these cases,
the association is caused by non-covalent interactions,
such as hydrogen or ionic binding, hydrophobic inter-
actions or van der Waals’ interactions.
We conclude that AdK and MK associate through
transient interactions that induce a slight conforma-

tional change in the active site of MK, thereby bring-
ing two already bound molecules of AMP together at
the correct distance for interaction and phosphoryl
transfer from one molecule to the other, ultimately
forming ADP.
Physiological role
Our data support the conclusion that the AMP–AMP
phosphotransferase reaction is physiologically impor-
tant. The reaction occurs in homogenates and crude
supernatants at pH values close to 7. The apparent K
m
value for the substrate is close to its intracellular con-
centration in rat liver [21–23], and is of the same order
of magnitude as that of many enzymes involved in
nucleotide metabolism [24–28]. The AdK, MK and
ADA concentrations used in these experiments coin-
cide with the concentrations found in rat liver tissue
[11,12].
Comparing the rates of the AMP–AMP phospho-
transferase, AdK and MK reactions is inappropriate
because the activities, rates and efficiencies of the
three reactions differ in vitro and in vivo according to
the concentrations of their natural substrates (AMP,
ADP, ATP), which vary continuously in different
situations.
Under physiological conditions, the AMP–AMP
phosphotransferase reaction may contribute to the fine
regulation of ADP levels. Its importance may be
greater under situations associated with ADP and
ATP deficiency, or increased requirements, such as

prolonged physical exertion (when ATP is dramatically
reduced), fructose-induced hyperuricaemia with ATP
depletion [29] and severe nucleotide depletion, as in
rheumatoid arthritis [30] and during cell division. The
reaction may play a specific role during transient
ischaemia, anoxia or after reperfusion. The behaviours
of AMP, ADP and ATP under such conditions have
been studied extensively and are similar in the liver [1],
heart and brain [10,31]. During ischaemia, levels of
ATP and ADP decrease [1,6,10,12,32], whereas those
of AMP increase sharply, reaching up to 2 lmol AMPÆ
g
)1
tissue [32]. During prolonged ischaemia, ATP
levels decrease to < 10% [1,6,10,32], and ADP levels
decrease to 25–50% of their respective basal concentra-
tions [6,32], whereas AMP levels increase by more than
20 times [32]. ADP levels are presumably sustained by
continuous regeneration, which is unlikely to occur
through the classical MK reaction because, under such
conditions, the levels of ATP are too low and the
levels of AMP are too high to permit classical MK
activity. Tamura et al. [33] reported that liver MK was
inhibited by AMP concentrations above 0.5 mm and
that the physiological significance of the data were
unclear, as this high concentration greatly exceeds the
concentration of AMP in rat liver (0.1 mm). In this sit-
uation, MK is inhibited and the action of AMP–AMP
phosphotransferase prevails, regenerating ADP. Gly-
colysis and oxidative phosphorylation can regenerate

ATP from ADP and P
i
; cooperation between these
processes and AMP–AMP phosphotransferase will
sustain ADP levels and regenerate ATP.
The experiments on human cancer mucosa produced
interesting results. Hypoxia is a well-known feature of
locally advanced solid tumours [34], and induces major
adaptive responses, such as the production of angio-
genic cytokines that promote vascularization and over-
expression of the hypoxia-inducible factor-1a gene,
which is a classical feature of tumour tissues [35] and
was confirmed in our specimens (data not reported). In
tumour tissue, AMP levels are more than five times
higher than in normal mucosa, whereas the ATP con-
centration is more than 10 times lower. In contrast,
ADP levels remain almost the same (data not
reported). Moreover, AMP–AMP phosphotransferase
activity was not detectable in normal mucosa, but was
substantial in tumour tissue. In the same specimens,
MK activity did not vary, whereas ADA and, espe-
cially, AdK activities increased. Therefore, in hypoxic
tissue, the enzyme ratios reach the correct value for
the AMP–AMP phosphotransferase reaction. The
importance of the reaction in tumour tissues will be
the subject of future research.
We conclude that, under specific conditions, ADP
and ATP may be salvaged and restored through the
concerted effects of classical ATP formation pathways
and the AMP–AMP phosphotransferase network. The

latter reaction may represent a compensatory mecha-
nism for maintaining and increasing ADP to the levels
necessary for the restoration of ATP, with the simulta-
neous re-utilization of AMP. The existence of ATP
turnover pathways has been suggested by other
authors in studies on MK in creatine kinase-deficient
transgenic hearts [36], specifically the existence of alter-
native pathways for phospho transfer in the myocar-
dium. Our experiments indicate that the cooperative
action of different proteins may provide fine regulation
D. Vannoni et al. Formation of ADP from AMP
FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS 279
of metabolic reactions, a poorly understood phenome-
non and an avenue of research that should be explored
further.
Experimental procedures
Materials
Male Wistar rats (body weight, 250 g; 9 weeks of age) were
purchased from Harlan Company (S. Pietro al Natisone,
Udine, Italy). Nucleosides, nucleotides, bases, enzymes,
analytes and the ATP Bioluminescent Assay Kit (FLAA-
1KT) were obtained from Sigma Life Science (Milan, Italy).
SDS-PAGE reagents and protein assay kits were procured
from Bio-Rad Laboratories s.r.l. (Milan, Italy). Chromato-
graphic supports, radioactive compounds and low-molecu-
lar-mass protein molecular mass marker kits were obtained
from GE Healthcare Europe GmbH (Milan, Italy). N-Hy-
droxysuccinimide, 1-ethyl-3-(3-dimethylaminopropyl) and
ethanolamine hydrochloride were purchased from Affinity
Sensor (Cambridge, UK). Aquasafe 300 plus was obtained

from Zinsser Analytic (Frankfurt, Germany). HPLC-grade
trifluoroacetic acid was obtained from Carlo Erba Reagenti
SpA (Rodano, Italy). Recombinant enzymes were pur-
chased from ABNOVA Corporation (Taipei, Taiwan). All
other chemicals and HPLC solvents were of analytical
grade and were acquired from Merck KGaA (Darmstadt,
Germany) and J T Baker Italia (Milan, Italy).
Enzyme assays and identification of reaction
products
Enzyme assays
The AMP–AMP phosphotransferase assay mixture con-
tained 4.0 mm cold AMP or 640 kBq [
32
P]AMP, 50 mm
Bistris (pH 6.5), 1.0 mm MgCl
2
, up to 0.5 mg of dialysed
crude rat liver supernatant or 1–3 lg of purified fraction
X2 (corresponding to 0.19–0.57 IU of pure MK), fraction
Y2 (corresponding to 0.006–0.018 IU of pure AdK) and
ADA (0.9 IU) in a final volume of 0.25 mL. Incubations
were performed at 37 °C for 15–50 min and stopped with
perchloric acid (neutralized with KOH) or 0.01 mm EDTA;
aliquots were processed by HPLC according to Webster
and Whaun [37]. The ammonium content was assayed using
the hypochlorite–phenol Berthelot reaction, according to
Imler et al. [38].
One International Unit (IU) of MK, AdK, ADA or
AMP–AMP phosphotransferase was defined as the amount
of enzyme that produced 1 lmolÆmin

)1
of reaction product.
We considered ADP (or ADP + ATP) to be the product
formed by the AMP–AMP phosphotransferase reaction.
AMP and all other substrates used in the assay mixture
were purified by HPLC. Traces of ATP in the substrates
and enzyme preparations were excluded using the ATP
Bioluminescent Assay Kit CLS II (Roche Diagnostics
GmbH, Mannheim, Germany; Sirius Luminometer-
Berthold GmbH, Pforzheim, Germany). AdK and ADA
were assayed according to Tavernier et al.[39]. MK was
assayed according to Zhang et al. [40].
HPLC analysis
We used a Perkin-Elmer (Monza, Italy) 1020LC Plus system
equipped with a ready-to-use prepacked column (Hypersil
SAX 5 lm, 150 · 4.6 mm; Alltech Italia s.r.l., Segrate, Italy)
washed with 5.0 mm ammonium phosphate (pH 2.9). Elution
was achieved with 0.5 m ammonium phosphate buffer
(pH 4.8) at a flow rate of 1.5 mLÆmin
)1
, using a linear gradi-
ent from 0% to 100% in 10 min. The lower limit of detection
of the method was 100 pmol.
When [
32
P]AMP was used, the ADP and ATP peaks
were collected and mixed with 10 mL Aquasafe 300 Plus
emulsifying scintillator, and the radioactivity was measured
using a Packard Model 1500 TriCarb b-counter (Hewlett
Packard, Monza, Italy).

Inosine and ADP were identified by multiple means: (a)
by determining the retention times and adding an internal
standard; (b) by determining the ultraviolet (UV) spectra in
the 210–350 nm range by diode array analysis with a Per-
kin-Elmer 235C detector system in line; and (c) by acid
hydrolysis of the presumed ADP peak and identification of
the products. ADP identification was confirmed by testing
its ability to act as a substrate for pyruvate kinase
(EC 2.7.1.40) [41]. After incubation with pyruvate kinase,
the HPLC chromatogram of the mixture revealed the disap-
pearance of ADP and the formation of an ATP peak,
which was identified in the chromatogram using the same
criteria as for ADP and inosine.
CE analysis
All of the above compounds (nucleosides and nucleotides)
were also analysed by CE using a Bio-Rad Biofocus 3000
apparatus equipped with a variable wavelength UV detec-
tor. The assays were performed in an uncoated silica capil-
lary (40 cm · 75 lm) with the following operating
conditions: 20 mm borate buffer, pH 10, 12 kV and 10 s;
hydrostatic load at 25 °C; 254 nm. The compounds were
identified by comparing their retention times with those of
known internal standards.
Enzyme purification and identification by mass
spectrometry
The AMP–AMP phosphotransferase reaction occurred only
when two or three different enzymes were combined, which
were purified according to the following procedures
(Table 1).
Formation of ADP from AMP D. Vannoni et al.

280 FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS
(a) Common trunk. Rat livers were obtained from ani-
mals housed at 20 °C, fed a standard laboratory diet and
killed by decapitation after a 12 h fast. The livers were
homogenized (15%) in buffer A (50 mm Tris ⁄ HCl, pH 7.5,
containing 0.1 mm dithiothreitol) and 0.25 m sucrose, and
centrifuged at 240 000 g for 60 min. The supernatant was
precipitated with ammonium sulfate (65–90% saturation)
and dialysed overnight (P90d).
(b) X2 fraction purification. The P90d fraction was applied
to a DE-52 cellulose column (5 · 6.5 cm; flow rate, 60 mLÆh
)1
)
and washed with three column volumes of buffer A.
Non-retained proteins were concentrated and washed with
buffer B (10 mm potassium phosphate, pH 7.0, with 0.1 mm
dithiothreitol) using centrifugal filters (CentriconPlus 20:
Fraction X; Millipore SpA, Vimodrone, Italy), and applied
to a Blue-Sepharose HiTrap column (5 mL) equilibrated with
buffer B and eluted using a step gradient with buffer B plus
KCl at a flow rate of 3 mLÆh
)1
. Fractions eluted with 0.5 m
KCl were pooled, washed several times with buffer C (10 mm
Tris ⁄ HCl, pH 7.5, with 0.1 mm dithiothreitol and 0.15 m
KCl), concentrated (fraction X1), submitted to gel filtration
chromatography using a prepacked Superdex 75 HR 10 ⁄ 30
column, and eluted with buffer C at a flow rate of 30 mLÆh
)1
.

The pure final protein was named X2.
(c) Y2 fraction purification. A DE-52 column was devel-
oped with 10 column volumes of a linear gradient (0–100%)
of buffer A spiked with 0.1 m KCl. The eluted and concen-
trated fractions were named fraction Y and applied to an
AMP-Sepharose column (1.6 · 2.5 cm), equilibrated with
buffer B at a flow rate of 3 mLÆh
)1
and washed with buffer B
containing 5 mm adenosine. The eluted proteins were
washed several times with buffer C, concentrated (fraction
Y1) and submitted to gel filtration chromatography as
described for X2. The pure final protein was named Y2.
(d) ADA purification. ADA was purified from P90d
according to Schrader et al. [42], with slight modifications
(data not shown).
All of the chromatographic procedures were performed
using a high-performance AKTA Purifier System (GE
Healthcare) and a unicorn Control System (Version 5.10)
as the control platform.
Total proteins were assayed according to Bradford [43]
using BSA as a standard. The purity of the fractions was
monitored by 12% SDS-PAGE under non-reducing condi-
tions [44], and the gels were developed with silver nitrate
stain. X2 and Y2 proteins were identified by mass spectro-
metry, performed at the International Centre for Mass
Spectrometry Services, University of Naples, or at the Centre
for Structural Determinations, University of Siena. A 200 lg
portion of X2 or Y2 fractions was desalted by RP-HPLC on
a Phenomenex C4 column (25 · 0.46 cm, 5 lm), and eluted

with 0.1% trifluoroacetic acid (solvent A) and 0.07% trifluo-
roacetic acid in 95% acetonitrile (solvent B) using a linear
gradient of 20–95% solvent B over 15 min. The elution
was monitored at 220 nm, and the fraction was collected
manually and injected directly into the electrospray mass
spectrometry apparatus. Analysis was performed using a ZQ
single quadrupole mass spectrometer equipped with an
electrospray ion source (Micromass, Manchester, UK) and a
Harvard syringe pump, with a flow rate of 5 lLÆmin
)1
.
Spectra were recorded by scanning the quadrupole (10 sÆper
scan). Data were acquired and processed using masslynx
software (Micromass). Mass scale calibration was performed
with multiply charged ions from a separate injection of horse
heart myoglobin (average molecular mass, 16 951.5 Da).
Trypsin digestion of the proteins was performed overnight in
0.4% ammonium bicarbonate (pH 8.5) at 37 °C using an
enzyme to substrate ratio of 1 : 50 (w ⁄ w). Trypsin–peptide
mixtures were analysed with a Voyager DE-PRO MALDI-
TOF mass spectrometer (Applied Biosystems, Monza, Italy)
equipped with a Reflectron analyser. The mass range was
calibrated using a mixture of five standard peptides provided
by the manufacturer. A 1.0 lL sample (approximately) was
applied to a sample slide and mixed with 1.0 lLofa
10 mgÆmL
)1
la-cyano-4-hydroxycinnamic acid solution in
acetonitrile with 0.2% trifluoroacetic acid (70 : 30, v ⁄ v)
before air drying. Peptide mass values recorded in the

MALDI spectra were used for a database search employing
profound 4.10.5 software downloaded from the web (http://
prowl.rockefeller.edu/).
Unknown proteins were identified by determining the
probability score associated with each putative candidate.
Expression of AdK in recombinant E. coli
An aliquot of human AdK clone 911, obtained from the
human liver cDNA kZAP library in a pET-24b vector
(kindly donated by J. Spychala, University of North Caro-
lina), was used to express AdK in E. coli BL 21[DE3] cells
Induction was performed for 4–6 h in the presence of 1 mm
isopropyl thio-b-d-galactoside. Recombinant AdK protein
was obtained from the cells by lysing the bacterial pellet in
10 volumes of 50 mm Tris ⁄ HCl (pH 7.5) containing 1 mm
EDTA, 1 mm dithiothreitol and 1 mgÆmL
)1
lysozyme, using
three freeze–thaw cycles in liquid nitrogen. The lysate was
ultracentrifuged and the supernatant was purified as
described previously [45].
Reaction mechanism studies – cooperation
between AdK and MK
Micro-equilibrium dialysis experiments
The micro-equilibrium dialysis apparatus (Micro-Equilib-
rium Dialyser, Harvard Apparatus Ltd., Edenbridge, UK)
consisted of two chambers for small samples of equivalent
volume (0.1 mL), separated by a cellulose acetate membrane
(5 kDa cut-off) impermeable to proteins. Each chamber
contained AMP, Bistris and MgCl
2

, as reported for the stan-
dard assay mixture, in a final volume of 0.1 mL. Suitable
D. Vannoni et al. Formation of ADP from AMP
FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS 281
amounts of AdK and MK were placed in the same chamber,
or one in either chamber separated by the membrane, in the
presence or absence of ADA. In all cases, the apparatus was
incubated at 37 °C for 1, 24 or 36 h. Reactions in both cham-
bers were stopped with 0.01 mm EDTA and analysed by
HPLC.
Detection of intermediates by HPLC, CE and NMR
analysis
The presence of intermediate reaction products was deter-
mined by HPLC and CE, as described previously, and by
31
P NMR spectra, obtained using an FT spectrometer
(model AMX-400, Bruker BioSpin, Milan, Italy) operating
at 161.923 MHz. In these experiments, AMP–AMP phos-
photransferase assay mixtures (2 mL) spiked with
2
H
2
O
were placed in 10 mm diameter tubes and incubated over-
night inside the magnet at 37 °C. The spectra were collected
as an accumulation of 128 scans with a 0.5 s repetition
time. Analysis was performed on an SGI data station, and
acquisition and processing parameters were identical in all
experiments. Spectra were obtained with 5 Hz line broaden-
ing to improve the signal-to-noise ratio.

Enzyme–phosphate intermediate trapping experiments
We used the procedure described by Baiocchi et al. [19] to
study cytosolic 5¢-nucleotidase phosphotransferase. The
reaction mixture contained 0.12 nmol pure MK, 0.14 nmol
pure AdK, 0.11 nmol pure ADA, 1.5 mm [
32
P]AMP
(3 CiÆmmol
)1
), 2.5 mm AMP, 50 mm Bistris ⁄ HCl buffer
(pH 6.5) and 1 mm MgCl
2
in a final volume of 50 lL. After
15 and 30 min at 37 °C, the reaction was stopped by the
addition of 12 lL 20% glycerol ⁄ 4% SDS in 125 mm
Tris ⁄ HCl (pH 7.4). The samples were loaded onto a 12%
SDS-PAGE gel, as described previously. The proteins were
electroblotted onto a nitrocellulose membrane (0.45 mm)
for 1 h at 100 V. After 4 days at )80 °C,
32
P-labelled
proteins were identified by autoradiography.
Another approach used to ascertain the formation of phos-
phoryl enzyme intermediates consisted of spiking the reaction
with vanadate, which inhibits the enzymes that catalyse phos-
phoryl transfer through the formation of phosphoryl enzyme
intermediates [46], or with nucleosides, such as [
14
C]adeno-
sine, [

14
C]guanosine or [
14
C]inosine. If covalent phospho-
enzyme intermediates (enzyme–phosphate–nucleoside) are
formed in the principal reaction, an exchange reaction of
nucleotide fi nucleoside is possible; then, a new nucleotide
is released and the principal reaction is inhibited [47].
Gel filtration and SDS-PAGE
Gel filtration experiments were performed using a pre-
packed Superdex 75 HR 10 ⁄ 30 column equilibrated and
eluted with 10 mm potassium phosphate buffer (pH 7.0)
and 0.15 m NaCl at a flow rate of 0.5 mLÆmin
)1
. In some
experiments, 0.1 mm AMP was added to the eluent
buffer. Twenty micrograms of pure AdK, MK and ADA
were incubated for 0 and 60 min at 37 °C and loaded on
to the column. Under these conditions, proteins and
reaction products were separated and detected in the
same run. SDS-PAGE was performed as described
previously.
Docking simulation studies
blast software ( />was used to select suitable templates for homology model-
ling of AdK and MK structures. The modelling procedure
was performed using deepview v.3.5 [48] and structure vali-
dation was obtained with procheck v.3.5.4 [49]. To find
the best templates, we submitted the sequence of two pro-
teins (SwissProt ID code Q5EBC5 for AdK and Q642G1
for MK) ( to blast. The corre-

sponding human proteins, exhibiting 92% and 90% iden-
tity, respectively, were used as templates by retrieving their
structures from the Protein Data Bank (http://
www.rcsb.org/pdb/): ID code 3AdK for AdK [50] and PDB
ID code 1BX4 for MK [51].
The initial docking simulation of AMP into the protein
active sites was performed by manually placing the mole-
cule at different locations. autodock software v.2.4 [52]
was then used to obtain the conformation with the lowest
potential energy. The protein complexes of rat AdK and
MK were placed at the centre of a rectangular box that
was filled with 8450 and 14 203 simple point charge water
molecules, respectively, and one Cl
)
and four Na
+
ions, to
obtain a net charge of zero. To relax the configuration of
the solvent, a steep descent minimization was performed.
All simulations were run using the gromacs v.3.2 bimole-
cular package [53]. The following protocol was used for the
MD simulation. To equilibrate the system, the temperature
was increased stepwise from 50 K to the target temperature
of 300 K in a 40 ps run. This was followed by a 1 ns pro-
duction run at constant temperature. A trajectory frame
dumped every picosecond. The protein and water were sep-
arately coupled to a temperature bath with a relaxation
constant of 0.5 ps. The pressure was maintained at
101.325 kPa by coupling to a pressure bath with a relaxa-
tion constant of 0.5 ps.

Kinetic experiments
In the kinetic experiments, various amounts of pure AdK
(0.4–1.6 lm) were incubated with a fixed amount of MK
(0.0032 lm); ADA was not added to the assay mixture.
The K
d
value was determined according to the Scatchard
equation.
Formation of ADP from AMP D. Vannoni et al.
282 FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS
Inhibition experiments
Three series of experiments were performed in which we
tested MK, AdK or AMP–AMP phosphotransferase activ-
ity with or without the addition of 0.25 mm Ap
5
A
(P
1
,P
2
-diadenosine-5¢-pentaphosphate), a specific competi-
tive inhibitor of MK [54], or 0.1 nm A134974, a specific
competitive inhibitor of AdK [55], or both.
AMP–AMP phosphotransferase reaction in human
colorectal mucosa from cancer patients
Tissue was obtained from surgically resected colon speci-
mens from 10 patients with cancer (eight men, two women;
age, 46–74 years) hospitalized at the Department of Gen-
eral Surgery, University of Siena. The clinical histories of
the patients were known; no patient had received chemo-

therapy or radiation therapy prior to surgery. Samples of
tumour tissue and normal mucosa located more than 10 cm
from the tumour were obtained from all patients. The fresh
surgical biopsies (70 mg) were immediately frozen in liquid
nitrogen and stored at )80 °C until assay. After thawing,
the samples were rinsed of residual blood with cold 10 mm
Hepes (pH 7) spiked with 0.1 mm MgCl
2
, homogenized in
the same buffer using an Ultra-Turrax blender (Ika Werke
GrbH, Staufen, Germany) and centrifuged at 65 000 g for
30 min in a Beckman L7-65 ultracentrifuge (Beckman
Coulter SpA, Cassina de Pecchi, Italy) at 4 °C. The clear
supernatant was partially purified on a CM Sepharose
column (1.6 cm · 1 cm), equilibrated with the same buffer
and eluted at a flow rate of 0.1 mLÆmin
)1
to remove any
interfering contaminants. The non-retained fraction
contained MK and AdK activities. ADA was eluted using
a linear gradient with the same buffer spiked with 2 m
NaCl in five column volumes.
MK, AdK, ADA and AMP–AMP phosphotransferase
activities were assayed using the mixtures reported previ-
ously for rat liver, incubating 20, 40, 80 and 100 lg, respec-
tively, of partially purified protein preparations.
Statistical analysis
Data were analysed by the Mann–Whitney U-test using
prism 4.0 by GraphPad Software Inc. (San Diego, CA,
USA). Differences yielding a P value of < 0.05 were con-

sidered to be significant. Kinetic data (K
m
and K
d
) were
evaluated using the same software.
Acknowledgements
We thank Professor J. Spychala, University of North
Carolina, for the human AK clone 911, and Paul
Kretchmer PhD, Managing Director San Francisco
Edit (), for assistance in editing
the manuscript. This work was supported by grants
from the University of Siena and Italian Ministry of
Education.
References
1 Vincent MF, Van der Berghe G & Hers HG (1982) The
pathway of adenine nucleotide catabolism and its
control in isolated rat hepatocytes subjected to anoxia.
Biochem J 202, 117–123.
2 Makarewicz W (1998) The role of radiation therapy in
the treatment of recurrent carcinoma of the vulva after
surgery. Adv Exp Biol 431, 351–357.
3 Meghji P, Middleton KM & Newby A (1998) Absolute
rates of adenosine formation during ischaemia in rat
and pigeon hearts. Biochem J 249, 695–703.
4 Curto R, Voit EO & Cascante M (1998) Analysis of
abnormalities in purine metabolism leading to gout and
to neurological dysfunctions in man. Biochem J 329,
477–487.
5 Weber G (1983) Enzymes of purine metabolism in

cancer. Clin Biochem 16, 57–63.
6 Wie S, Reillaudou M, Apiou F, Peyre H, Petridis F &
Lucciconi C (1999) Purine metabolism in two human
melanoma cell lines: relation to proliferation and differ-
entiation. Melanoma Res 9, 351–359.
7 Hardie DG & Carling D (1997) The AMP-activated
protein kinase-fuel gauge of the mammalian cell? Eur J
Biochem 246, 259–273.
8 Rauch U, Schulze K, Witzenbicher B & Schultheiss HP
(1994) Alteration of the cytosolic–mitochondrial distri-
bution of high-energy phosphates during global myocar-
dial ischemia may contribute to early contractile failure.
Circ Res 75, 760–769.
9 Hayashi K, Ochiai T, Ishinoda Y, Okamoto T, Maruy-
ama T, Tsuda K & Tsubouchi H (1997) Relationship
between cellular ATP content and cellular functions of
primary cultured rat hepatocytes in hypoxia. J Gastro-
enterol Hepatol 12, 249–256.
10 Bak MI & Ingwall JS (1998) Regulation of cardiac
AMP-specific 5¢-nucleotidase during ischemia mediates
ATP resynthesis on reflow. Am J Phys 274, 992–1001.
11 Arch JRS & Newsholme EA (1978) Activities and some
properties of 5¢-nucleotidase, adenosine kinase and
adenosine deaminase in tissues from vertebrates and
invertebrates in relation to the control of the concentra-
tion and the physiological role of adenosine. Biochem J
174, 965–977.
12 Sapico V, Litwack G & Criss WE (1972) Purification
of rat liver adenylate kinase isozyme II and compari-
son with isozyme 3. Biochim Biophys Acta 258, 436–

445.
13 Scatchard G (1942) Equilibrium thermodynamics and
biological chemistry. Science 95, 27–32.
D. Vannoni et al. Formation of ADP from AMP
FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS 283
14 Kurz LC, Moix L, Riley MC & Frieden C (1992) The
rate of formation of transition-state analogues in the
active site of adenosine deaminase is encounter-con-
trolled: implications for the mechanism. Biochemistry
31, 39–48.
15 Alberty RA (2000) Calculating apparent equilibrium
constants of enzyme-catalyzed reactions at pH 7.
Biochem Educ 28, 12–17.
16 Hunt SW & Hoffee PA (1982) Adenosine deaminase
from deoxycoformycin-sensitive and resistant rat hepa-
toma cells. Purification and characterization. J Biol
Chem 257, 14239–14244.
17 Ma PF & Fisher JR (1969) Comparative studies of
mammalian adenosine deaminases. Some distinctive
properties in higher mammals. Comp Biochem Physiol
31, 771–781.
18 Basile G, Taglialatela-Scafati O, Damonte G, Armirotti
A, Bruzzone S, Guida L, Franco L, Usai C, Fattorusso
E, De Flora A et al. (2005) ADP-ribosyl cyclases gener-
ate two unusual adenine homodinucleotides with cyto-
toxic activity on mammalian cells. Proc Natl Acad Sci
USA 102, 14509–14514.
19 Baiocchi C, Pesi R, Camici M, Itoh R & Tozzi MG
(1996) Mechanism of the reaction catalysed by cytosolic
5¢-nucleotidase ⁄ phosphotransferase: formation of a

phosphorylated intermediate. Biochem J 317, 797–801.
20 Tyler JM, Reinhardt BN & Branton D (1980) Associa-
tions of erythrocyte membrane proteins. Binding of
purified bands 2.1 and 4.1 to spectrin. J Biol Chem 255,
7034–7039.
21 Jackson RC, Lui MS, Boritzki J, Morris HP & Weber
G (1980) Purine and pyrimidine nucleotide patterns of
normal, differentiating, and regenerating liver and of
hepatomas in rats. Cancer Res 40, 1286–1291.
22 Itakura M, Maeda N, Tsuchiya M & Yamashita K
(1986) Increased rate of de novo purine synthesis and its
mechanism in regenerating rat liver. Am J Physiol 251,
G585–G590.
23 Reiss PD, Zuurendonk PF & Veech RL (1984) Mea-
surement of tissue purine, pyrimidine, and other nucleo-
tides by radial compression high-performance liquid
chromatography. Anal Biochem 14, 162–171.
24 Spychala J, Madrid-Marina V & Fox IH (1988) High
K
m
soluble 5¢-nucleotidase from human placenta. Prop-
erties and allosteric regulation by IMP and ATP. J Biol
Chem 263, 18759–18765.
25 Zalkin H (1983) Structure, function, and regulation of
amidophosphoribosyl transferase from prokaryotes. Adv
Enzyme Reg 21, 225–237.
26 Kelley MJ & Carman GM (1987) Purification and
characterization of CDP-diacylglycerol synthase from
Saccharomyces cerevisiae. J Biol Chem 262, 14563–
14570.

27 Ipata PL & Cerignani G (1978) Cytosine and cytidine
deaminase from yeast. Methods Enzymol 51, 394–401.
28 Rowe PB, MacCairns E, Madsen G, Sauer D & Elliott
H (1978) De novo purine synthesis in avian liver.
Co-purification of the enzymes and properties of the
pathway. J Biol Chem 253, 7711–7722.
29 Perheentupa J & Raivio K (1967) Fructose-induced
hyperuricaemia. Lancet, 2, 528–531.
30 Marinello E, Carlucci F, Tabucchi A, Leoncini R,
Pizzichini M & Pagani R (1994) The purine nucleotide
content of lymphocytes from patients with rheumatoid
arthritis. Int J Chin Pharmacol Res 14, 57–63.
31 Barsotti C, Tozzi MG & Ipata PL (2002) Purine and
pyrimidine salvage in whole rat brain. Utilization of
ATP-derived ribose-1-phosphate and 5-phosphoribosyl-
1-pyrophosphate generated in experiments with dialyzed
cell-free extracts. J Biol Chem 277, 9865–9869.
32 Bontemps F, Vincent MF & Van der Berghe G (1993)
Mechanisms of elevation of adenosine levels in anoxic
hepatocytes. Biochem J 290, 671–677.
33 Tamura T, Shiraki H & Nagakawa H (1980) Purifica-
tion and characterization of adenylate kinase isozymes
from rat muscle and liver. Biochim Biophys Acta, 612,
56–66.
34 Vaupel P & Mayer A (2007) Hypoxia in cancer: signifi-
cance and impact on clinical outcome. Cancer Metasta-
sis Rev 26, 225–239.
35 Semenza GL (2007) Hypoxia and cancer. Cancer Metas-
tasis Rev 26, 223–224.
36 Dzeja PP, Vitkevicius KT, Redfield MM, Burnett JC &

Terzic A (1999) Reduced activity of enzymes coupling
ATP-generating with ATP-consuming processes in the
failing myocardium. Circ Res 84, 1137–1143.
37 Webster HK & Whaun JM (1981) Interaction of
hypoxia and aging in the heart: analysis of high energy
phosphate content. J Chromatogr 209, 283–292.
38 Imler M, Frick A, Schlienger JL & Stahl A (1979) An
automated microassay for blood ammonia. J Clin Chem
Clin Biochem 17, 247–250.
39 Tavernier M, Skladanowski AC, De Abreu RA & de
Jong JW (1995) Kinetics of adenylate metabolism in
human and rat myocardium. Biochim Biophys Acta
1244, 351–356.
40 Zhang YL, Zhou JM & Tsou CL (1993) Inactivation
precedes conformation change during thermal denatur-
ation of adenylate kinase. Biochim Biophys Acta
1164,
61–67.
41 Bucher T & Pfleiderer G (1955) Pyruvate kinase from
muscle. In Methods in Enzymology, Vol. 1 (Colowick
SP & Kaplan NO, eds), pp. 435–440. Academic Press
Inc., NY.
42 Schrader WP, Andrew RS & Pollara B (1976) Purifica-
tion of human erythrocyte adenosine deaminase by
affinity column chromatography. J Biol Chem 251,
4026–4032.
43 Bradford MM (1976) A rapid and sensitive method for
the quantitation of microgram quantities of protein
Formation of ADP from AMP D. Vannoni et al.
284 FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS

utilizing the principle of protein–dye binding. Anal
Biochem 72, 248–254.
44 Laemmli UK (1970) Cleavage of structural proteins
during the assembly of the head of bacteriophage.
Nature 227, 680–685.
45 Spychala J, Datta NS, Takabayashi K, Datta M, Fox
IH, Gribbing T & Mitchell BS (1996) Cloning of human
adenosine kinase cDNA: sequence similarity to
microbial ribokinases and fructokinases. Proc Natl Acad
Sci 93, 1232–1237.
46 Mimoni M, Bontemps F & Van den Berghe G (1994)
Kinetic studies of rat liver adenosine kinase. Explana-
tion of exchange reaction between adenosine and AMP.
J Biol Chem 269 , 17820–17825.
47 Workn Y & Newby CA (1993) Phosphorylation of
adenosine in anoxic hepatocytes by an exchange
reaction catalysed by adenosine kinase. Biochem J 205,
503–510.
48 Guex N & Peitsch MC (1997) SWISS-MODEL and the
Swiss-PdbViewer: an environment for comparative
protein modeling. Electrophoresis 18, 2714–2723.
49 Laskowski RA, Moss DS & Thornton JM (1993) Main-
chain bond lengths and bond angles in protein struc-
tures. J Mol Biol 231, 1049–1067.
50 Dreusicke D, Karplus PA & Schulz GE (1988) Refined
structure of porcine cytosolic adenylate kinase at 2.1 A
˚
resolution. J Mol Biol 199, 359–371.
51 Mathews II, Erion MD & Ealick SE (1998) Structure
of human adenosine kinase at 1.5 A

˚
resolution.
Biochemistry 37, 15607–15620.
52 Morris GM, Goodsell DS, Huey R & Olson AJ (1996)
Distributed automated docking of flexible ligands to
proteins: parallel applications of AutoDock 2.4. J Com-
put Aided Mol Des, 10, 293–304.
53 Van Der Spoel D, Lindahl E, Hess B, Groenhof G,
Mark AE & Berendsen HJ (2005) GROMACS:
fast, flexible, and free. J Comput Chem 26, 1701–
1718.
54 Sheng XR, Li X & Pan XM (1999) An iso-random Bi
Bi mechanism for adenylate kinase. J Biol Chem 274,
22238–22242.
55 Zhu CZ, Mikusa J, Chu KL, Cowart M, Kowaluk A,
Jarvis MF & McGaraughty S (2001) A-134974: a novel
adenosine kinase inhibitor, relieves tactile allodynia via
spinal sites of action in peripheral nerve injured rats.
Brain Res 905, 104–110.
D. Vannoni et al. Formation of ADP from AMP
FEBS Journal 276 (2009) 271–285 ª 2008 The Authors Journal compilation ª 2008 FEBS 285