Nucleotide promiscuity of 3-phosphoglycerate kinase is in focus: Implications for the design of better anti-HIV analogues

Article abstract of DOI:10.1039/c1mb05051f

The wide specificity of 3-phosphoglycerate kinase (PGK) towards its nucleotide substrate is a property that allows contribution of this enzyme to the effective phosphorylation (i.e. activation) of nucleotide-based pro-drugs against HIV. Here, the structural basis of the nucleotide-PGK interaction is characterised in comparison to other kinases, namely pyruvate kinase (PK) and creatine kinase (CK), by enzyme kinetic analysis and structural modelling (docking) studies. The results provided evidence for favouring the purine vs. pyrimidine base containing nucleotides for PGK rather than for PK or CK. This is due to the exceptional ability of PGK in forming the hydrophobic contacts of the nucleotide rings that assures the appropriate positioning of the connected phosphate-chain for catalysis. As for the d-/l-configurations of the nucleotides, the l-forms (both purine and pyrimidine) are well accepted by PGK rather than either by PK or CK. Here again the dominance of the hydrophobic interactions of the l-form of pyrimidines with PGK is underlined in comparison with those of PK or CK. Furthermore, for the l-forms, the absence of the ribose OH-groups with PGK is better tolerated for the purine than for the pyrimidine containing compounds. On the other hand, the positioning of the phosphate-chain is an even more important term for PGK in the case of both purines and pyrimidines with an l-configuration, as deduced from the present kinetic studies with various nucleotide-site mutants of PGK. These characteristics of the kinase-nucleotide interactions can provide a guideline for designing new drugs.

Full text of DOI:10.1039/c1mb05051f

View Article Online / Journal Homepage / Table of Contents for this issue
www.molecularbiosystems.org
Volume 7 | Number 6 | 1 June 2011 | Pages 1781–2088
ISSN 1742-206X
PAPER
Mária Vas et al.
Nucleotide promiscuity of 3-phosphoglycerate kinase is in focus: implications for the design of better anti-HIV analogues

Dynamic ArticVleiewLiAnrktiscle Online
Molecular
BioSystems
Cite this: Mol. BioSyst., 2011, 7, 1863–1873
www.rsc.org/molecularbiosystems
PAPER
Nucleotide promiscuity of 3-phosphoglycerate kinase is in focus:
implications for the design of better anti-HIV analoguesw
a
c
b
b
a
´
Andrea Varga,* Laurent Chaloin, Gyula Sagi, Robert Sendula, Eva Graczer,
´
Karoly Liliom, Peter Zavodszky, Corinne Lionne and Maria Vas*
´
´
a
a
c
a
´
´
´
´
Received 7th February 2011, Accepted 30th March 2011
DOI: 10.1039/c1mb05051f
The wide specificity of 3-phosphoglycerate kinase (PGK) towards its nucleotide substrate is a
property that allows contribution of this enzyme to the effective phosphorylation (i.e. activation)
of nucleotide-based pro-drugs against HIV. Here, the structural basis of the nucleotide-PGK
interaction is characterised in comparison to other kinases, namely pyruvate kinase (PK) and
creatine kinase (CK), by enzyme kinetic analysis and structural modelling (docking) studies. The
results provided evidence for favouring the purine vs. pyrimidine base containing nucleotides for
PGK rather than for PK or CK. This is due to the exceptional ability of PGK in forming the
hydrophobic contacts of the nucleotide rings that assures the appropriate positioning of the
connected phosphate-chain for catalysis. As for the D-/L-configurations of the nucleotides, the
L-forms (both purine and pyrimidine) are well accepted by PGK rather than either by PK or CK.
Here again the dominance of the hydrophobic interactions of the L-form of pyrimidines with
PGK is underlined in comparison with those of PK or CK. Furthermore, for the L-forms, the
absence of the ribose OH-groups with PGK is better tolerated for the purine than for the
pyrimidine containing compounds. On the other hand, the positioning of the phosphate-chain is
an even more important term for PGK in the case of both purines and pyrimidines with an
L-configuration, as deduced from the present kinetic studies with various nucleotide-site mutants
of PGK. These characteristics of the kinase-nucleotide interactions can provide a guideline for
designing new drugs.
However, such detailed understanding of non-physiological
Introduction
actions of PGK, as outlined below, is still lacking.
3-Phosphoglycerate kinase (PGK), one of the key enzyme of
glycolysis, catalyses a phospho-transfer from 1,3-bisphospho-
glycerate (1,3-bPG) to MgADP, yielding 3-phosphoglycerate
(3-PG) and MgATP. This enzyme has been extensively studied
by our laboratory in terms of both structure1–5 and function.6–9
The immense knowledge of enzymological and structural data
accumulated about PGK offered the possibility to get a closer
insight into the mechanism of its domain-motion assisted
functioning.3,10,11 Site directed mutagenesis of individual
amino acid residues in and around the molecular hinge-regions
approved the importance of domain closure in the catalysis.12,13
It was demonstrated recently that PGK has an important
role in the activation of some antiviral (anti-HIV) L-nucleoside
analogue prodrugs.14,15 These prodrugs have to be administrated
to the patient in the form of a nucleoside to be able to cross the
cellular membrane. Therefore, they require three sequential
steps of phosphorylation within the cell in order to reach their
triphosphate form which possesses biological activity. Since an
elevated intracellular concentration of the nucleotide triphos-
phate form correlates with the more pronounced inhibition of
HIV reverse transcriptase (HIV-RT),16 the understanding
of the phosphorylation pathway of individual nucleoside
analogues is required in order to design new structurally
modified nucleosides with a more efficient activation.
a Institute of Enzymology, Biological Research Center, Hungarian
Academy of Sciences, P. O. Box 7, H-1518 Budapest, Hungary.
E-mail: varga@enzim.hu, vas@enzim.hu; Fax: 36-1-466-5465;
Tel: 36-1-279-3152
The first two phosphorylation steps of nucleosides are better
clarified (reviewed in Deville-Bonne et al.17). The first one is
catalysed by deoxycytidine kinase,18–20 thymidine kinase 2,21
or deoxyguanosine kinase.22 Uridine/cytidine monophosphate
kinase23,24 is responsible for the second phosphorylation step.
However, the third phosphorylation step is of potential
importance, since this step is the bottleneck for the production
of triphosphorylated nucleoside analogues into the cell.
b Institute of Biomolecular Chemistry, Chemical Research Center,
Hungarian Academy of Sciences, 1025 Budapest, Hungary
c
´
Centre d’etudes d’agents Pathogenes et Biotechnologies pour la Sante
(CPBS), UMR 5236, CNRS-Universite Montpellier Sud de France,
`
´
´
1919 route de Mende, 34293 Montpellier cedex 5, France
w Electronic supplementary information (ESI) available. See DOI:
10.1039/c1mb05051f
c
This journal is The Royal Society of Chemistry 2011
Mol. BioSyst., 2011, 7, 1863–1873 1863

View Article Online
This step can be catalysed by PGK or other enzymes,
which may also have potential role (cf. the recent review by
Deville-Bonne et al.17). These are nucleoside diphosphate
kinase (NDPK), creatine kinase (CK) and pyruvate kinase
(PK). NDPK catalyses a reversible transfer of a phospho-group
between natural D-nucleoside di- and triphosphates, with little
base specificity.25 Nevertheless, the presence of the ribose
30-hydroxyl group is crucial for NDPK activity.26 This hydroxyl
group is often absent in the prodrug molecules, therefore the
other three kinases CK, PK and PGK, with low substrate
specificities, may be preferable as potential activating enzymes.
Although no reasonable kinetic measurements are published,
binding data indicate relatively good interactions of ADP-analogues
lacking the 30 hydroxyl group either to CK or PK.27 As for
PGK, dephosphorylation of 2030-ddATP has also been noted.15
L-nucleoside analogues in most cases are found to be less
toxic, because these analogues are not recognised by the host
cell polymerases as substrates, but well recognised by HIV-
RT.28 L-analogues of nucleoside-diphosphates were shown to
be better phosphorylated by PGK compared to other kinases,
and these compounds are often more effective against HIV than
their D-counterpart.29 Currently eight dideoxynucleoside
analogues targeting HIV-RT are approved by the US Food and
Drug Administration, two of them are in the L-configuration.
The results presented to date about the nucleotide substrate
specificity of PGK are limited and there are some uncertainties
due to the different methodology applied and/or to some
misleading presentation of the data.14 Furthermore, investigations
are mainly restricted to the dephosphorylation reaction.15,30
Studies of the physiologically more relevant phosphorylation
reactions by comparing D- and L-ADP substrates have been
initiated in our group with human PGK (hPGK).31,32
Moreover, using X-ray crystallography, efforts have been
made to describe the binding modes of L-ADP and L-CDP,
in comparison to their D-counterparts.33 The structural data
were consistent with the substrate-like binding of the L-analogues.
Here we complete the investigations with these L-enantiomers by
systematic comparison of their interactions with various single
mutants made at the nucleotide binding site of human PGK.
Besides, the substrate specificity of hPGK is comprehensively
tested with nucleotides possessing purine (ADP, GDP, 20-dADP)
and pyrimidine (CDP, UDP, 20-dTDP) bases in comparison with
two known kinases, CK and PK. The activity data towards the
various nucleotides are interpreted on the basis of the structural
characteristics of their binding to these enzymes. The possible
binding modes are determined by molecular docking into the
recently published 1.6 A resolution structure of hPGK,34 as well
as to the known X-ray structures of CK35 and PK.36 On these
bases, implications for designing better anti-HIV drugs are
considered.
starting material according to Visser (yield: 98.8%).37 20,30,50-
Tri-O-benzoyl-L-cytidine (2) was prepared from 1.38 Contrary
to the cited description, we found that 2.5 equiv. of TMSOTf
(cf. Abbreviations) and 7 h stirring at 95 1C were necessary to
drive the reaction to completion (yield: 91.0%). L-Cytidine (3)
was obtained as a white precipitate with a yield of 94.6% after
stirring of 2 in 6.5 M NH3/MeOH solution. L-Cytidine-50-
monophosphate (4) was prepared according to the well-known
Yoshikawa method,39 but purification of the product was
modified since, after the usual anion-exchange chromatography
on
a DEAE-Sephadex A-25 column, the product still
contained significant amounts of inorganic salts. Therefore,
the solid residue thus obtained was dissolved in water, the
pH was adjusted to about 2 by addition of 1 M HCl, and this
acidic solution was applied to a Dowex 50W X8 H+ column
that was first washed with water to elute the inorganic
components (HCl, H3PO4), then with concentrated NH4OH
solution to elute the pure L-CMP as an ammonium salt. The
UV active NH4OH fractions were combined and finally
evaporated twice with EtOH. Starting from 274 mg (1.13 mmol)
of 3, we isolated 297 mg (0.92 mmol) of 4 as a white hygroscopic
solid that was free of inorganic salts, giving a yield of 81.4%
(Rf (D) = 0.33, [a]2D0 = ꢀ8.5 ꢁ 11 c = 1, H2O). The spectral
data were identical with those reported in the literature,33 but
our yield was considerably higher, probably due to this
improved purification method which seems to be applicable
for desalting of all nucleoside monophosphates containing
amino groups on the base moiety. L-Cytidine-50-diphosphate
(5) was synthesized from 4 as described by Gondeau et al.33
After the last purification step (RP-HPLC, elution with water),
60 mg of pure L-CDP was obtained. N6,20,30,50-Tetrabenzoyl-
L-adenosine (6) and L-adenosine (7) were synthesized according
to the literature38 without any modifications. Preparations of
L-AMP (8) and L-ADP (9) were executed in an analogous way
to that described previously for 4 and 5. Physical and spectral
data of these compounds were identical with those reported in
the literature.40
The Mg-nucleotides were formed by addition of MgCl2
(Sigma) to each of the nucleotides. NAD, NADH, NADP,
creatine, creatine phosphate, phosphoenolpyruvate and
glucose were from Sigma. 1,3-bPG was prepared from glycer-
aldehyde 3-phosphate (Sigma) according to the Negelein
method41 with the modifications described by Furfine and
Velick.42 Isopropyl b-D-thiogalactopyranoside (Fermentas),
chloramphenicol, ampicillin and kanamycin (Sigma) were
used for the fermentation. All other chemicals were reagent
grade commercial preparations.
Enzymes
Human PGK (hPGK) was prepared as described earlier.8
Human CK was purified with some modifications, as
described in the paper by Hahn et al.43 The enzyme was prepared
from inclusion bodies using 7 M urea as a solubilizing agent.
After solubilization, the protein was renaturated in 50 mM
Tris, 0.5 mM NaCl pH 8.3 containing 5 mM MET. During
this process, about half of the total amount of the enzyme was
precipitated. The soluble fraction was dialysed against the
usual buffer for Ni-NTA separation and purified according to
Materials and methods
Chemicals
All nucleotides, except L-ADP and L-CDP, were from Sigma.
L-ADP and L-CDP were prepared as described by Gondeau
et al.33 with a slight modification. Briefly, 1-O-Acetyl-2,3,5-tri-
O-benzoyl-L-ribofuranose (1) was prepared from L-ribose as a
c
1864 Mol. BioSyst., 2011, 7, 1863–1873
This journal is The Royal Society of Chemistry 2011

View Article Online
the manufacturer’s (Qiagen) instruction. Rabbit muscle PK
was purchased from Sigma. The auxiliary enzymes hexokinase
(HK), glucose-6-phosphate dehydrogenese (G6PDH) (Sigma)
and glyceraldehyde 3-phosphate dehydrogenase (GAPDH)
nyi44) were used for
+
(prepared according to Elodi and Szore
´
PGK as a function of saturation by nucleotides, using 5,50-
dithiobis(2-nitrobenzoic acid) (DTNB) as a modifying agent.5
Modelling of the different nucleotides binding to the kinases
¨
Molecular modelling of the different nucleotides (either D- or
L-enantiomers) were carried out using Biopolymer implemented
in Insight II (Accelrys, Inc.). Indeed, both enantiomers were
constructed from ADP, CDP, GDP, UDP and TDP. The
potential energy of the nucleotides was minimized using
Discover and the consistent valence force field in two phases,
1000 steps of steepest descent followed by 5000 steps of
conjugate gradient with a tolerance of 0.001 kcal molꢀ1 A.
The following crystal structures of the different kinases were
selected from the PDB according to ligands and transition state
analogues already present: 2WZB (ADP/3-PG/MgF3 bound
to human PGK),34 3GR4 (ADP-bound PK from human)36
and 1VRP (ADP/Nitrate/(diaminomethyl-methyl-amino)-acetic
acid bound to CK from Torpedo californica).35 Addition of the
hydrogen atoms and deletion of one ADP or ATP molecule
was achieved with GOLD (Genetic Optimization for Ligand
Docking, CCDC software limited) prior to the docking of the
nucleotide enantiomers with Gold 5.0 program. Briefly,
50 genetic algorithms were performed with a radius of 15 A
around the selected atoms as a target for docking (backbone
nitrogen of Pro 338 for hPGK, backbone oxygen of Thr 322
for CK and Leu 74 (a-carbon) for PK). The water molecules
present in the crystal structures were retained. Goldscore was
used as scoring function, and the different docking poses were
analyzed by the clustering method (complete linkage) from the
rmsd matrix of ranking solutions. For the best ranked docking
poses, the potential energy was minimized in three steps using
the CHARMM (Chemistry at HARvard Macromolecular
Mechanics47) program (500 steps of steepest descent (SD),
followed by 5000 steps of a conjugate gradient with a tolerance
of 0.01 kcal molꢀ1 A without applying any constraint).
For visualizing and analysing the molecular details of the
nucleotide binding, the INSIGHT II (Biosym/MSI, San diego,
CA, USA) software was used. The distance limits for H-bonds,
hydrophobic and ionic interactions were taken to be 3.5, 4.8
and 4.5 A, respectively.
the coupled enzyme activity measurements.
Kinetic measurements
All experiments were carried out in 50 mM Tris, pH 7.5,
containing 1 mM DTT at 20 1C. For the D-ADP substrate, the
activities of the investigated kinases have been checked using
the appropriate coupled enzyme systems.
Briefly, for determination of the activity of CK, the produced
ATP was reacted with glucose catalysed by HK, and then the
product glucose-6-phosphate was oxidized by glucose-6-
phosphate dehydrogenase. During this latter reaction, NADP
was reduced to NADPH, which served as a spectrophoto-
metric signal, detectable at 340 nm.8
The PGK assay mixture for testing various nucleotides as
substrates was prepared as described by Varga et al.31 Briefly,
the coupled enzyme, GAPDH, produces the substrate 1,3-bPG
for PGK which is accompanied by the production of NADH.
Under the action of PGK, 1,3-bPG reacts with the nucleotide
substrate applied. The consumption of 1,3-bPG leads to
further production of NADH, i.e. further increase in the
spectrophotometric signal, due to shifting of the equilibrium
of the GAPDH reaction. In this case, the HK-G6PDH
coupled system is not applicable because of the strict nucleo-
tide specificity of HK.
The CK assay mixture contained 10 mM phosphocreatine,
10 mM DTT, 5 mM MgCl2 and the nucleotide diphosphates at
varying concentrations.45 The parameters of the HK-G6PDH
coupled enzyme system were the same as described by
Flachner et al.8 For the phosphorylation of L-ADP, CK
activity was measured using PGK and GAPDH as a coupled
enzyme system. Briefly, the product L-ATP was dephosphorylated
by PGK and the produced 1,3-bPG was reacted in the coupled
GAPDH reaction, where the consumption of NADH served
as an appropriate spectrophotometric signal.6
The PK assay mixture contained 2 mM phoshoenolpyruvate,
100 mM KCl (which increases the activity of PK by about
4-fold), 20 mM MgCl2 and the nucleotides at varying
concentrations. Here, the product pyruvate reacted with
NADH catalysed by lactate dehydrogenase (LDH), and the
decrease in NADH concentration served as a spectrophoto-
metric signal, followed at 340 nm.46
Results and discussion
Comparison of the nucleotide specificities of phosphoglycerate
kinase, creatine kinase, pyruvate kinase and nucleoside
diphosphate kinase
For the coupled enzyme systems, as usual, the concentration
of the auxiliary enzyme (e.g. GAPDH for PGK or LDH for PK)
was set so as to not limit the investigated catalysed reaction.
Among these kinases, the data are mainly about the nucleotide
specificity of NDPK (Table 1). As known from the literature,
NDPK has little base specificity.25 The 30-OH-group of the
nucleotide is very important in the catalysis of NDPK. Indeed,
the catalytic efficiency of NDPK with D-30-dTDP is negligible
(0.003%) compared to that with D-20-dTDP26 and the
relatively good activity with D-20-dADP also seems to be
consistent with these findings. Regarding that the 30-OH-group
is missing in most anti-HIV analogues, we have to reconsider
the possible participation of other kinases in the activation
(phosphorylation) process of these nucleotide compounds.
One of these enzymes is PGK, which was reported to
Binding studies
All experiments were carried out with human PGK and its
mutants in 50 mM Tris, pH 7.5, containing 1 mM DTT at
20 1C. Determination of the Kd values for D- and L-ADP and
CDP was determined either by the method using the protein
bound 1-anilinonaphthalene-8-sulfonic acid (ANS) as a signal,13
or by detecting the reactivity of the two reactive thiol-groups of
c
This journal is The Royal Society of Chemistry 2011
Mol. BioSyst., 2011, 7, 1863–1873 1865

View Article Online
Table 1 Comparison of the catalytic efficiences (kcat/Km, Mꢀ1 sꢀ1) of different kinases with natural nucleoside diphosphate substrates. All data refers
to subunit concentrations of the enzymes
Nucleotides
aNDPK
PGK
PK
eCK
Purine
D-ADP
1.8 ꢂ 107
4.5 ꢂ 106
2.9 ꢂ 107
1.3 ꢂ 106
3.8 ꢂ 106
2.6 ꢂ 106
(62%)
(16%)
(100%)
(4.5%)
(13%)
(9%)
b,c9.0 ꢂ 106
2.9 ꢂ 105
9.1 ꢂ 106
c1.7 ꢂ 102
4.6 ꢂ 103
3.6 ꢂ 102
(95%)
(3.2%)
(100%)
(0.002%)
(0.05%)
(0.004%)
d4.9 ꢂ 105
d3.4 ꢂ 104
d9.3 ꢂ 104
d9.5 ꢂ 103
d2.0 ꢂ 104
136
(100%)
(6.9%)
(19%)
(1.9%)
(4.1%)
(0.03%)
f5.6 ꢂ 106
1.4 ꢂ 106
7.9 ꢂ 103
3.3 ꢂ 103
1.4 ꢂ 103
n.d.
(100%)
(25%)
(0.14%)
(0.06%)
(0.025%)
D-20-dADP
D-GDP
Pyrimidine
D-CDP
D-UDP
D-TDP (20d)
a
Schaertl et al.25 b Varga et al.31 c Data are in accordance with Gallois-Montbrun et al.15 d Data are in accordance with Plowman and
Krall.46 e James and Morrison.45 f A value of 6.6 ꢂ 106 was obtained by us.
phosphorylate L-nucleoside analogues.31,33 Therefore, we have
made an attempt (see next section) to establish the detailed
structural basis of the nucleotide specificity/promiscuity of this
enzyme. Apart from this, a systematic comparison of the
functions of PGK, PK and CK towards natural nucleotide
diphosphates with various bases are also needed, since in the
literature no appropriate comprehensive data are available.
Therefore, in Table 1 we present new measurements together
with the previously published data. Considering the substrate
specificity, the nucleotides with purine base (especially GDP)
are much better phosphorylated by PGK compared to either
PK or CK. On the contrary, in the case of nucleotides with a
pyrimidine base (such as CDP), PK and CK seem to be much
better candidates than PGK. Modelling studies of D-CDP
binding to these kinases, indeed, seem to be consistent with
this finding. The number (Table 2) and the detailed list of the
atomic contacts in Supplementary Table 1, ESI,w have
revealed that the strength and number of the hydrophobic
contacts with the nucleotide rings are extremely important
only for PGK (e.g. the hydrophobic contacts of Pro 338 with
D-CDP are greatly reduced compared to those with D-ADP).
These contacts possibly assure the appropriate positioning of
the connected phosphate-chain for catalysis. In contrast, the
mode of interaction of the purine vs. pyrimidine nucleotide
rings with the other kinases, such as PK and CK (although
differences are also notable, cf. Table 2 and Supplementary
Table 1, ESIw) is seemingly less important in respect to
catalysis (Table 1). This selectivity between purines and
pyrimidines among the three kinases, however, is much smaller
for UDP. Furthermore, it is notable that PK possesses the
lowest nucleotide specificity among the three kinases. Our new
data with D-20-dTDP, however, provides an exception rather
than following this rule (see Table 1), which can be only partly
due to the absence of the 20-OH-group of ribose. On the other
hand, CK is found to be a bit more specific, than PK,45
although it can tolerate the absence of ribose hydroxyls more
than any other kinases, as mentioned in the Introduction,27
and also approved by the activity data with D-20-dADP.45
Since no data have been available for comparison among
PGK, PK and CK in respect of their action on L-nucleotides,
we have carried out comparative kinetic and docking studies
using L-ADP and L-CDP. Complete kinetic characterisation
could not be carried out in all cases, therefore activities of the
three enzymes have been compared only at a single concentration
(2.5 mM) of either D- or L-forms of these nucleotide substrates
(Table 3). Thus, the observed reduction in the activity values
(v) with the D- vs. L-forms can be equally due to the decrease
of kcat and/or increase of Km. The data clearly demonstrates
that the catalytic efficiency of PGK on the nucleotides with
L-configuration is much better (by many orders of magnitude)
than that of either PK or CK (Table 3). These findings are
entirely consistent with the modelling study when L-ADP and
L-CDP were docked to each of the three investigated kinases in
comparison to their D-counterparts (Fig. 1). This figure
illustrates only those atomic contacts that are differently
formed during interaction with the respective D- and L-nucleotides.
It is immediately noticeable that for PGK there is a smaller
number of differences between interactions of these two
counterparts of both ADP and CDP (Fig. 1A and B), while
more extensive differences exist between the D- and L-forms in the
case of both PK (Fig. 1C and D) and CK (Fig. 1E and F). In
addition, for PK, the positioning of the base is drastically
different between both enantiomers, as seen for L-ADP and
L-CDP (Fig. 1C and D). It is notable, however, that the
number of the atomic contacts (Table 2) does not always
demonstrate the differences in their nature. Yet, it is evident
Table 2 Number of the atomic contacts of various nucleotides bound to
PGK, PK and CK as determined by docking to the respective crystal
structures
Various structural parts of the nucleotide
Phosphate-chain Ribose-ring Ring of the base
Kinases Nucleotides
PGKa D-ADP
L-ADP
10
11
5+33b
6+35
0+33
5+24
5+32
7+33
5+31
3+20
0+19
0+21
3+18
0+15
1+21
3+28
1+23
4+23
5+22
2+14
3+41
3+37
0+38
3+35
1+30
0+34
1+34
5+46
2+37
5+30
5+19
1+22
2+28
2+39
3+39
4+30
4+29
4+20
D-20-dADP 12
D-GDP
D-CDP
L-CDP
D-UDP
10
10
8
9
D-20-dTDP 10
PKa
CKa
D-ADP
L-ADP
D-dADP
D-CDP
L-CDP
D-ADPa
L-ADP
8
8
9
10
11
24
21
D-20-dADP 21
D-CDP
L-CDP
15
15
a
X-Ray structures (pdb) used for docking: 2WZB (PGK), 3GR4 (PK)
b
as well as 1VRP (CK). Where two different values are given, the first
one refers to the number of H-bonds plus ionic interactions, while the
second one refers to the number of hydrophobic contacts.
c
1866 Mol. BioSyst., 2011, 7, 1863–1873
This journal is The Royal Society of Chemistry 2011

View Article Online
Table 3 Comparison of the enantioselectivity of different kinases with ADP or CDP substrates. Catalytic constants (kcat, sꢀ1) and reaction velocities
(v, sꢀ1) refer to subunit concentrations of the enzymes. Reaction velocities (v) have been determined using constant (2.5 mM) nucleotides
PGK
PK
CK
kcat (sꢀ1
)
v (sꢀ1
)
v (%)
kcat (sꢀ1
)
v (sꢀ1
)
v (%)
kcat (sꢀ1
)
v (sꢀ1
)
v (%)
Nucleotides
D-ADP
L-ADP
D-CDP
L-CDP
1085
685
2.8
1035
618
0.363
31.9
100
151
n.d.
48
66
100
345
338
2.8
6.3
n.d.
100
0.83
1.9
59.7
0.035
3.1
0.045
10.7
0.019
0.0298
7.1
0.0126
n.d.
a26.4
n.d.
58
n.d.
n.d.
a
James and Morrison.45
Fig. 1 Pairwise comparison of binding of D- and L-nucleotides to the different kinases. Highest score docking poses obtained with D-ADP and L-ADP
(panels A, C and E) and with D-CDP and L-CDP (panels B, D and F) after docking onto the three kinases, hPGK, PK and CK. The substrates are
colored as indicated on the top and represented in ball and stick model, while the interacting residues of each kinase are in black stick models. The
dashed lines (coloured as the nucleotides) show only those atomic contacts (hydrogen bonds, electrostatic and van der Waals) that are differently
formed in the complexes with the two compared enantiomers bound to the enzyme. The limiting distances of the various types of contacts are given in
A-s in Materials and Methods.
from Table 2 that the total number of contacts of L-CDP rings
with PGK is significantly higher than those for D-CDP, in line
with the finding that L-CDP is acting as a better substrate
for PGK.
At this point, it may be worth discussing the question
concerning the reality of the docking results. The protein
models for docking are always held as rigid structures,
although it is known that generally large changes in the
c
This journal is The Royal Society of Chemistry 2011
Mol. BioSyst., 2011, 7, 1863–1873 1867

View Article Online
structure occur upon substrate binding. Yet, the initial inter-
actions with substrates are always very crucial in respect to the
further elementary steps of catalysis, as indicated by fast
kinetic studies with hPGK as an example.32,48,49 Since each
kinase has its own mechanism for the binding and the
phosphorylation of nucleotides, probably with a different
time-course from kinase to kinase, using molecular dynamics
simulation it would have been extremely difficult and time-
consuming to carry out the experiments on the appropriate
(different) time-scale in each case. Therefore, here we restricted
ourselves to do simple docking studies with the known active
(closed) conformational states of these kinases and in each
case we used energy minimization after the docking steps.
This means that 20-dTDP can somehow compensate the loss of
its hydroxyl group through its base interactions (cf. differences
in the interaction of the base of UDP and 20-d-TDP, Fig. 2F).
However, a change in the C4-position of uracil (e.g. replace-
ment of the oxo- by an amino-group in case of cytosine) leads
to decrease in kcat of about one order of magnitude and in
addition, an increase in the Km, as observed in the case of CDP
(Table 4). The structural basis of this type of selection of PGK
was investigated by modelling. Docking of these nucleotides to
hPGK (cf. Fig. 2D–F) may provide an explanation consistent
with the above-mentioned importance of the hydrophobic
interactions of the nucleotide rings with Pro 338. Thus, correct
positioning of the base directs both rings towards Pro 338
together with the ribose hydroxyls towards Glu 343 and
this is manifested further in the optimal orientation of the
phosphate-chain for the catalysis.
Nucleotide base promiscuity of wild-type PGK: steady-state
kinetic, binding and modelling studies
In contrast to our kinetic data in Table 4, Krishnan and
coworkers50 found that the Km values of the nucleotides even
with purine and pyrimidine bases are very similar. The reason
of these different findings might be due to the different
methodology used by this group for activity measurements.
As it was also noted by Gallois-Montbrun and coworkers,15
Cheng’s laboratory50 used an HPLC method to detect the
amount of produced triphosphate during the enzyme reaction.
In these cases, the product can accumulate and inhibit the
enzyme reaction leading to lower velocity and lower Km values
for pyrimidine nucleotides compared to those obtained by
either Gallois-Montbrun and coworkers or the authors of this
paper. Furthermore, our binding studies have provided
additional evidence in contrast to Krishnan and coworkers’
data.50 Namely, using the fluorescent dye ANS, only the
binding of the purine-base containing nucleotides was
detectable. However, no any change could be detected in the
fluorescence of PGK-bound ANS upon binding of the nucleo-
tides with pyrimidine-base. This means that in terms of the
binding of the base, the purine and pyrimidine nucleotides
behave differently, i.e. their binding modes to PGK are
different. This observation is entirely consistent with their
different Kd values, listed in Table 4.
In order to characterise in more details the dependence of
nucleotide specificity of PGK on the chemical structure of the
nucleotide base, more exhaustive kinetic measurements of
phosphorylation of ADP, GDP, CDP, UDP and 20-dTDP,
as well as binding studies with PGK, were carried out and
compared (Table 4). PGK binds more strongly, as well as
phosphorylates more effectively, the nucleotides containing
purine bases (ADP, GDP) compared to the ones with pyrimidine
bases (CDP, 20-dTDP, UDP). Not only the catalytic velocities
(kcat), but also the catalytic interactions with substrates (Km),
are very similar for ADP and GDP. Docking of D-GDP into
the active site of hPGK provided a consistent picture (Fig. 2B).
In the case of the pyrimidine base containing nucleotides,
UDP can be phosphorylated much better than either CDP or
20-dTDP (Table 4). The similarity in the structure of the bases
of UDP and 20-dTDP is manifested only in their interactions
with the enzyme in terms of both Kd and Km values. The
difference in their kcat values can be attributed to the missing
20-OH-group of 20-dTDP which was reported to be important,
but not essential, in the binding of the nucleotides to PGK.13 It
is a bit surprising that there is no difference in their Km values
(if we compare the Km values of D-ADP and D-20-dADP there
is about a 10-fold increase due to the loss of the 20-OH-group).
Table 4 Importance of different residues for the high purine specificity and the low enentioselectivity of PGK. Catalytic and binding properties of
wild-type hPGK and its mutants with various nucleotides are compared. The experimental error of the determinations are ꢁ5–10%
Nucleotides
kcat (sꢀ1
)
Km (mM)
kcat/Km (Mꢀ1 sꢀ1
)
Kd (mM)
a
Wild-type hPGK
Purine
D-ADP
1085
685
438
914
2.8
58
0.12
0.27
1.5
0.10
16.8
2.1
5.2
9.0 ꢂ 106
2.5 ꢂ 106
2.9 ꢂ 105
9.1 ꢂ 106
167
0.024
0.071
0.26
0.011
3.5
1.74
2.7
a
L-ADP
D-20-dADP
D-GDP
D-CDP
L-CDP
Pyrimidine
2.8 ꢂ 104
D-UDP
D-20-dTDP
D-ADP
L-ADP
L-CDP
D-ADP
L-ADP
L-CDP
24
2.3
4635
355
6.5
4.3
E343A mutant
N336A mutant
K219A mutant
b41 (3.8%)
b40 (5.8%)
0.53 (0.9%)
11 (1.0%)
2.3 (0.33%)
0.10 (0.17%)
2.3 (0.2%)
3.65 (0.5%)
>0.014
b1.24 (10ꢂ)
b1.35 (5ꢂ)
5.37 (2.6ꢂ)
0.68 (5.7ꢂ)
1.26 (4.7ꢂ)
9.58 (4.7ꢂ)
2.34 (20ꢂ)
2.34 (8.7ꢂ)
>2.34
b3.3 ꢂ 104
b3.0 ꢂ 104
98
c0.57 (24ꢂ)
0.71 (10ꢂ)
>1.21 (0.7ꢂ)
c0.38 (16ꢂ)
0.49 (7ꢂ)
>0.71 (0.4ꢂ)
1.96 (82ꢂ)
0.47 (6.6ꢂ)
2.44 (1.4ꢂ)
1.6 ꢂ 104
1.8 ꢂ 103
10.4
D-ADP
L-ADP
L-CDP
983
1560
4.2
Varga et al.31 b Gondeau et al.33 c Szabo
´
et al.13
a
c
This journal is The Royal Society of Chemistry 2011
1868 Mol. BioSyst., 2011, 7, 1863–1873

View Article Online
Fig. 2 Nucleotide base promiscuity to hPGK. Docking results obtained with purine base nucleotides, D-ADP (black, panels A, B, C and D),
D-GDP (blue, panel B), D-dADP (pink, panel C) and pyrimidine base nucleotides, D-CDP (green, panels D and E), D-UDP (orange, panels E and F),
D-20-dTDP (cyan, panel F). All ligands are represented in ball and stick model and hPGK residues in stick models. In the pairwise comparison the
dashed lines (coloured as the nucleotides) illustrate only those atomic contacts (hydrogen bonds, electrostatic and van der Waals) that are
differently formed during interaction with the protein. The limiting distances of the various types of contacts are given in A (see Materials and
Methods section).
Effective phosphorylation of L-ADP and L-CDP by hPGK:
kinetic, binding and modelling studies
published both kcat and Km values separately for L-CTP, while
for D-CTP they have reported only the ratio (kcat/Km), i.e. the
catalytic efficiency values.15 These differences are also reflected,
although in a much smaller extent, in the Kd values of D- and
L-CDP binding (Table 4). Thus, we can conclude that the
catalytic interactions of L-CDP with PGK are more favourable
and the position of the b-phosphate is optimised better for
acceptance of the transferred phospho-group. Although at face
value this finding is in accordance with the conclusions drawn
by Gondeau and coworkers from the crystal structures of the
ternary complexes of hPGK with D-CDP and 3-PG, as well as
with L-CDP and 3-PG,33 the real structural explanation
might be missing since these structures represent inactive
conformations with open relative domain positions.
The phosphorylation reaction of L-ADP by hPGK has been
characterised earlier. There was around a 2.5-fold increase in
Km and approximately a 1.5-fold decrease in kcat compared to
the values obtained with the natural substrate D-ADP.31 This is
not the case for the pairs of D-CDP and L-CDP, where L-CDP is
phosphorylated 21-fold faster by PGK than D-CDP and its Km
value is lower by one order of magnitude compared to its
D-counterpart (Table 4). These values resulted in approximately
a 150-fold difference in the catalytic efficiency in favour of
L-CDP relative to D-CDP. Gallois-Montbrun and coworkers
found similar effects for D-CTP and L-CTP, however they have
c
This journal is The Royal Society of Chemistry 2011
Mol. BioSyst., 2011, 7, 1863–1873 1869

View Article Online
Fig. 3 Illustration of the open and closed conformations of hPGK (A) and location of the mutated nucleotide binding residues in the closed
conformation of hPGK (B). The Ca traces of the open conformation (black, pdb: 2XE854) complexed with 3-PG and the ATP analogue AMPꢀ-PCP
as well as the closed conformation (red, pdb: 2WZB34) complexed with 3-PG, ADP and the transition state analogue (TSA) MgF3 are
superimposed according to the inner b-strands of the N-terminal domain in (A). All the ligands are shown in stick models with the corresponding
colour of protein conformation. (B) The three side-chains of residues in the nucleotide binding site that were mutated into alanine in this work are
shown in stick models (coloured by atom types).
Fig. 3A illustrates the difference between the open and
closed conformations of hPGK. In the open form the two
reacting substrates are positioned far away from each other,
thus occurrence of catalysis is prevented. In contrast, the
closed conformation assures close proximity of the reactive
groups required by the phospho-transfer between the bound
substrates. Therefore the question arises whether the crystallo-
graphically determined binding mode of L-nucleotides is close
to that existing in the catalytically active closed conformation
and which amino acid residues are essential for the productive
interaction with these analogues. These questions have been
addressed here by modelling studies of L-ADP and L-CDP
binding to the recently determined closed conformation of
wild-type hPGK34 (cf. Table 2 and supplementary Table 1A, ESIw)
as well as by kinetic and binding studies (cf. below) on various
nucleotide-site mutants of hPGK with these nucleotides.
As discussed above, from modelling of L-ADP and L-CDP
binding to the wild-type hPGK and from the comparison
with their D-counterparts (Fig. 1A and B and Table 2 and
supplementary Table 1A, ESIw) it follows that while the
important hydrophobic interactions of the base and ribose
rings are only slightly modified in L-ADP, those of L-CDP are
more dominant compared to D-CDP, although both CDPs
possess a considerably smaller number of interactions relative
to those of ADPs.
respectively.13 Since in the previous31,32 and the present work
with the non-natural nucleotides a different method applying a
different coupled enzyme system had to be used (see explanation in
Materials and methods), the new data in Table 4 are some-
what, but not essentially, different from the above-listed
values. It was shown previously that with the mutant
E343A, the catalytic processing and interactions of D- and
L-ADP became similar,33 which was approved here in terms of
their kcat, Km and Kd values (Table 4). From this observation it
was suggested and confirmed here that the different catalytic
processing of D- and L-ADP by wild-type PGK is at least in
part due to the differences in their E343-ribose hydrogen bond
strength. This means that in the catalytic complex with
L-ADP, the interaction with E343 is slightly weaker compared
to D-ADP, in accordance with the crystal structures possessing
open domain conformations.33 In fact, the significance of the
interactions of ribose hydroxyls of D-ADP in formation of the
Michaelis complex is also supported by the increased Km value
of D-20-d-ADP (Table 4), in agreement with the present
modelling of D-20-dADP into the active site of wild-type
hPGK (Fig. 2C). On the other hand, modelling of L-ADP into
the crystal structure with closed domain conformation could
not reveal significant differences, at least not in respect to
ligation of the ribose OH-groups (Fig. 1A, Table 2 and
supplemenary Table 1A, ESIw). This picture is entirely
consistent with the relatively good kcat value of L-ADP compared
to D-ADP. Thus, taken together, in case of purine base-
containing nucleotides with L-configuration, the absence of
ribose hydroxyl groups should not be detrimental in respect to
their phosphorylation.
Insight into the structural basis of the effective phosphorylation
of L-nucleotides by PGK from kinetic and binding studies with
various nucleotide-site mutants
A more direct insight can be obtained about the optimal
structural interactions of the various nucleotide substrates
with PGK from the activity measurements of the nucleotide-
site mutants E343A, N336A and K219A towards D- and
L-ADP as well as D- and L-CDP as substrates. Positions of
these residues in the molecule of hPGK are illustrated in
Fig. 3B. Using the natural substrate D-ADP, these residues
were found to be important both in the formation of the
catalytic complex and in the turnover of the enzyme: the Km
values of D-ADP were 0.12, 0.91, 0.42 and 1.42 mM for the
wild-type, E343A, N336A and K219A mutant enzymes,
Here, we also tested two other mutants to see whether K219
and N336 can contribute to the substrate-like binding of
L-ADP. K219 participates in the binding of the a-phosphate and
the g-phosphate of an analogue of MgATP in its catalytically
competent closed conformation of TmPGK.51 Moreover,
K219 is involved in the stabilization and neutralization of
the transition state analogue bound in the closed crystal
structure of hPGK.34 Thus, K219 can be considered as an
important catalytic residue, in agreement with the dramatic
activity loss observed upon its mutation into alanine.13
N336 links the b-phosphate to the interdomain helix 13 through
c
1870 Mol. BioSyst., 2011, 7, 1863–1873
This journal is The Royal Society of Chemistry 2011

View Article Online
an H-bond formation with G37152 and these interactions possibly
contribute to the catalytically important domain motions.13
Here we investigated whether the above-mentioned inter-
actions are also crucial for the phosphorylation of L-ADP.
Based on the experimental data (see Table 4), in the case of
K219A, the catalytic processing (both in terms of kcat and
Km values) was found to be similar for both D- and L-ADP.
Thus, as it was found with E343A, the data obtained with the
mutant K219A shows that, in case of the wild-type enzyme,
there should be a difference in the strength of interactions of
D- and L-ADP with the residue K219, being weaker for L-ADP
because this difference is lost in the mutant K219A. Besides the
kinetic constants, the same is reflected by changes in the ratio
of Kd values of the nucleotides upon mutation of K219. This
may mean that the positioning of the phosphate-chain of
L-ADP through the interaction with K219 is less important
(compared to its D-counterpart) for catalysis. However, this is
not the case for N336A. Here, in the kcat values there is
approximately a 5-fold difference between D- and L-ADP
(see Fig. 4A), which means that the interaction of N336 with
the phosphate-chain of L-ADP is more important compared to
D-ADP in order to reach effective phosphorylation. However,
changes in Km and especially in the Kd values (Table 4) rather
reflect the importance of the interaction of Asn 336 with the
phosphate-chain of both D- and L-ADPs. Such conclusions
could not be derived from the previously determined X-ray
structures of the ADP 3-PG PGK ternary complexes, obtained
with both D- and L-ADPs,33 where the distances between the
b-phosphate of D- and L-ADP from the side-chain of N336
were found to be as large as about 7 and 10 A, respectively,
indicating the absence of any atomic interactions in either
case, which may be due to the fact that these open crystal
structures did not represent catalytically relevant conformations.
Furthermore, helix 13, that was shown to interact with N336
(through G371), and to be involved in the allosteric transmission
of the nucleotide effect, is not well resolved in these X-ray
structures. However, when L-ADP was docked into the active
site of the closed conformation of hPGK, well-defined inter-
actions of its phosphates with N336 are indeed confirmed
(supplementary Table 1A, ESIw).
In the case of L-CDP, Fig. 4B compares its catalytic
processing with the wild-type as well as the mutants E343A,
K219A and N336A hPGKs. There is about a 100-fold
decrease in kcat with E343A compared to the wild-type enzyme
(Table 4). Comparison of this decrease to the 15-fold decrease in
the kcat of L-ADP for the same mutant, one can conclude that
binding of the ribose hydroxyls of L-CDP to E343 is much more
important (compared to either L- or D-ADP) in respect to effective
phosphorylation. On the basis of the even larger decrease in kcat of
L-CDP obtained with either N336A or K219A mutants (Fig. 3B),
one can conclude that positioning of the phosphate-chain of
L-CDP is even more important in term of its phosphorylation
than positioning of its ribose hydroxyls by E343. This conclusion
is in agreement with the picture obtained above by modelling of
D- and L-CDP binding to hPGK (Fig. 1B and suppl. Table 1A,
ESIw). Thus, the order of decrease in kcat of the enzyme forms
(wild-type, E343A, N336A and K219A) is the same for the
different nucleotides, but in the case of L-CDP the effects are even
larger (see Table 4). Concerning the Km and Kd values, there is a
Fig. 4 Effect of mutations at the nucleotide site of hPGK on the steady
state kinetics measured with the enantioselective pairs of both ADP and
CDP. Catalytic processing of D-ADP (m) and L-ADP (&) by 169 nM
mutant N336A (A) as well as of L-CDP by 25.3 nM wild-type hPGK
(K) and its mutants E343A (J), N336A (m) and K219A (n) in
concentrations of 3.5, 20.3 and 72.7 mM, respectively (B).
general tendency of increases for L-CDP, which is more or less
correlated with the decrease of its kcat values (Table 4).
It is notable, that L-CDP can be considered as a model
compound for the anti-HIV drug, L-3TC (known L-SddC in
another publication, which is the drug lamivudine used in
anti-HIV treatment). This analogue contains a cytosine base
without the OH-groups of ribose. Gallois-Montbrun et al.15
and Krishnan et al.14 tested the phosphorylation of the
diphosphates of L-3TC by PGK and obtained the values 800
and 4000 Mꢀ1 ꢀ1, respectively for kcat/Km. These values
s
indicate a further decrease of at least one magnitude compared
to the catalytic efficiency obtained by us with L-CDP (Table 4).
For L-FMAUDP, which is a thymine base containing anti-
HIV drug, possessing the 20-OH-group of ribose, Krishnan
et al. detected a kcat/Km 1.3 ꢂ 106 Mꢀ1 sꢀ1 efficiency, which
decreased to 4.8 ꢂ 104 Mꢀ1 sꢀ1 if they replaced E343 with
alanine. Thus, these data underline the importance of the
interaction(s) of ribose OH-group(s) with the side-chain of
E343 of PGK in the case of pyrimidine base containing drug
molecules. Our investigations, shown above with L-CDP, have
led to similar conclusions. This fact emphasizes the relevance
of experimental approaches with model compounds with the
purpose of designing new anti-HIV drugs.
c
This journal is The Royal Society of Chemistry 2011
Mol. BioSyst., 2011, 7, 1863–1873 1871

View Article Online
types of chemical modifications of the nucleotide base also
have to be considered as a means of structural variation.
Among the kinases, the structural characteristics of the
nucleotide binding pocket of PGK seems to be advantageous
in this respect, since a water filled cavity behind the bound
nucleotide base is observed by crystallography.33 Thus, it is
advisable to test any newly designed drug in the phosphoryl-
ation reaction by PGK.
General conclusion
The data presented in this paper allow us to establish some
general principles that should be taken into account in the
course of designing new nucleoside-based anti-HIV drugs. We
have to emphasize that our investigations were not dealing
with the function of HIV reverse transcriptase itself,
but restricted to the phosphorylation process leading to
production of the nucleoside triphosphates, which represent
the active form of drugs. This step is as important as the
reaction catalysed by the transcriptase, because the phosphoryl-
ation step produces the substrate for this last reaction.
Thereby, incorporation of the drug will result in termination
of the viral DNA chain elongation.
List of abbreviations
ANS
1-anilinonaphthalene-8-sulfonic acid
1,3-bisphosphoglycerate
1,3-bPG
CK
creatine kinase (EC 2.7.3.2)
dithiotreitol
Based on substrate specificity of the compared kinases,
among the nucleotides with D-configuration the purine base
containing ones are much better phosphorylated by either
PGK, PK or CK, compared to the nucleotides with pyrimidine
bases. In the case of D-nucleotides with a pyrimidine base,
mainly PK (and CK) contributes to their phosphorylation. This
means that from the nucleotides with a natural D-configuration,
it is more favourable to choose the ones with a purine base
(either adenine or guanine) for drug design, because this way
the three kinases can act together, leading to a significantly
increased cellular level of the total amount of the active
phosphorylated drug.
DTT
DTNB
HIV-RT
5,50-dithiobis(2-nitrobenzoic acid)
human immunodeficiency virus reverse
transcriptase
HK
hexokinase (EC 2.7.1.1)
GAPDH
D-glyceraldehyde-3-phosphate
dehydrogenase (EC 1.2.1.12)
glucose-6-phosphate dehydrogenase
(EC 1.1.1.49)
G6PDH
LDH
D-lactate dehydrogenase (EC 1.1.1.28)
mercaptoethanol
MET
NDPK
3-PG
nucleoside diphosphate kinase (EC 2.7.4.6)
3-phosphoglycerate
In the case of the nucleotides with an L-configuration,
however, the catalytic efficiency of PGK is much better
(by many orders than that of either PK or CK). Since almost
in all cases the L-nucleoside analogues feature a more
advantageous toxicity profile than their D-counterparts, it is
quite important to take into account this property and favour
the usage of L-analogues in HIV therapy, in spite of the fact
that up to now only a single kinase (PGK) has been recognised
to be able to phosphorylate them effectively. Investigations
with several mutants of human PGK clearly showed the
favourable phosphorylation of L-nucleotides with purine
bases. In addition, the absence of the interaction of the ribose
hydroxyl with PGK is better tolerated in the case of purine-
compared to pyrimidine-containing compounds. Since the
30-hydroxyl group of the ribose is important for DNA chain
elongation, it has to be absent in the anti-HIV drugs. Thus, the
more insensitive the kinase for the lack of hydroxyls, the more
favourable phosphorylation can occur. This can be fulfilled in
case of the L-purine analogues with PGK as a catalyst.
In agreement, there is a potential drug in the literature—the
L-form of b-2,6-diaminopurine dioxolane (L-DAPD) is a
potent inhibitor against HIV (EC50 is 14 nM).53
PK
pyruvate kinase (EC 2.7.1.40)
3-phosphoglycerate kinase (EC 2.7.2.3)
human PGK
PGK
hPGK
TMSOTf
Trimethylsilyl trifluoromethanesulfonate
Acknowledgements
This work was supported by the grant OTKA (NK 77978) of
the Hungarian National Research Fund and by the National
Development Agency KMOP-1.1.2-07/1-2008-0003 grant.
Travelling support by NKTH/OMFB (No. F-31/2009) and
Egide (PHC Balaton No. 22295 QE) are gratefully acknowl-
edged. The authors are indebted to Mrs. E. Braun (Institute of
Biomolecular Chemistry, Chemical Research Center,
Budapest, Hungary) for her valuable technical assistance in
the chemical preparative work. The plasmid of CK was a
generous gift from M. J. Hahn, Department of Molecular
Cell Biology, Sungkyunkwan University School of Medicine
(Suwon, Korea).
References
In summary, the analogues with purine-bases are potentially
better phosphorylated by kinases tested here in the case of
either D- or L-configurations. Based solely on phosphorylation,
the recommendation is to design either adenine or guanine-
base containing drugs, because kinases can better tolerate any
changes (e.g. the absence of 30-OH) in the ribose in these cases,
being essential for chain termination. Either D- or L-form can
potentially be used, this should be decided based on the
toxicity profile of the appropriate analogue-pair.
1 K. Harlos, M. Vas and C. C. F. Blake, Proteins: Struct., Funct.,
Bioinf., 1992, 12, 133–144.
2 A. May, M. Vas, K. Harlos and C. C. F. Blake, Proteins: Struct.,
Funct., Bioinf., 1996, 24, 292–303.
3 A. N. Szila
2001, 306, 499–511.
4 Z. Kovari, B. Flachner, G. Na
2002, 41, 8796–8806.
5 B. Flachner, Z. Kova
G. Naray-Szabo and M. Vas, Biochemistry, 2004, 43, 3436–3449.
´
gyi, M. Ghosh, E. Garman and M. Vas, J. Mol. Biol.,
´
ray-Szabo and M. Vas, Biochemistry,
´ ´
´
ri, A. Varga, Z. Gugolya, F. Vonderviszt,
´
´
6 P. Tompa, P. T. Hong and M. Vas, Eur. J. Biochem., 1986, 154,
643–649.
7 A. N. Szilagyi and M. Vas, Biochemistry, 1998, 37, 8551–8563.
´
Because new drugs are always needed due to the frequently
developed resistance against them, the possibilities of various
c
1872 Mol. BioSyst., 2011, 7, 1863–1873
This journal is The Royal Society of Chemistry 2011

View Article Online
8 B. Flachner, A. Varga, J. Szabo
P. Zavodszky and M. Vas, Biochemistry, 2005, 44, 16853–16865.
9 A. Varga, J. Szabo, B. Flachner, Z. Gugolya, F. Vonderviszt,
P. Zavodszky and M. Vas, FEBS Lett., 2009, 583, 3660–3664.
10 A. Varga, B. Flachner, E. Gra
M. Vas, FEBS J., 2005, 272, 1867–1885.
´
, L. Barna, I. Hajdu
´
, G. Gyimesi,
34 M. J. Cliff, M. W. Bowler, A. Varga, J. P. Marston, J. Szabo,
´
´
A. M. Hounslow, N. J. Baxter, G. M. Blackburn, M. Vas and
J. P. Waltho, J. Am. Chem. Soc., 2010, 132, 6507–6516.
35 S. D. Lahiri, P. F. Wang, P. C. Babbitt, M. J. McLeish,
G. L. Kenyon and K. N. Allen, Biochemistry, 2002, 41,
13861–13867.
36 A. Allali-Hassani, G. A. Wasney, I. Chau, B. S. Hong,
G. Senisterra, P. Loppnau, Z. Shi, J. Moult, A. M. Edwards,
C. H. Arrowsmith, H. W. Park, M. Schapira and M. Vedadi,
Biochem. J., 2009, 424, 15–26.
37 G. M. Visser, J. van Westrenen, C. A. A. van Boeckel and J. H.
van Boom, Recl. Trav. Chim. Pays-Bas, 1986, 105, 528–537.
38 H. Vorbruggen, I. M. Lagoja and P. Herdewijn, Curr. Protoc.
Nucleic Acid Chem., 2007, ch. 1, Unit 1 13.
39 M. Yoshikawa, T. Kato and T. Takenishi, Tetrahedron Lett., 1967,
50, 5065–5068.
´
40 J. He, B. Roy, C. Perigaud, O. B. Kashlan and B. S. Cooperman,
´
´
´
´
czer, S. Osvath, A. N. Szilagyi and
´ ´
´
11 A. Varga, B. Flachner, P. Konarev, E. Gra
D. Svergun, P. Zavodszky and M. Vas, FEBS Lett., 2006, 580,
2698–2706.
12 J. Szabo, A. Varga, B. Flachner, P. V. Konarev, D. I. Svergun,
P. Zavodszky and M. Vas, FEBS Lett., 2008, 582, 1335–1340.
13 J. Szabo, A. Varga, B. Flachner, P. V. Konarev, D. I. Svergun,
P. Zavodszky and M. Vas, Biochemistry, 2008, 47, 6735–6744.
14 P. Krishnan, Q. Fu, W. Lam, J. Y. Liou, G. Dutschman and
Y. C. Cheng, J. Biol. Chem., 2002, 277, 5453–5459.
15 S. Gallois-Montbrun, A. Faraj, E. Seclaman, J. P. Sommadossi,
D. Deville-Bonne and M. Veron, Biochem. Pharmacol., 2004, 68,
1749–1756.
16 D. S. Stein and K. H. Moore, Pharmacotherapy, 2001, 21, 11–34.
17 D. Deville-Bonne, C. El Amri, P. Meyer, Y. Chen, L. A.
Agrofoglio and J. Janin, Antiviral Res., 2010, 86, 101–120.
18 G. Maury, Antivir. Chem. Chemother, 2000, 11, 165–189.
19 E. Sabini, S. Hazra, M. Konrad, S. K. Burley and A. Lavie, Nucleic
Acids Res., 2007, 35, 186–192.
´
czer, J. Szabo
´
,
´
´
´
´
´
FEBS J., 2005, 272, 1236–1242.
41 E. Negelein and H. Bromel, Biochem. Z., 1939, 301, 135.
¨
42 C. S. Furfine and S. F. Velick, J. Biol. Chem., 1965, 240, 844–855.
43 H. S. Hahn, Y. D. Park, J. R. Lee, K. H. Park, T. J. Kim,
J. M. Yang and M. J. Hahn, J. Protein Chem., 2003, 22, 563–570.
+
44 P. Elodi and E. Szorenyi, Acta Physiol. Acad. Sci. Hung., 1956, 9,
´
¨
339–350.
20 E. Sabini, S. Hazra, M. Konrad and A. Lavie, J. Med. Chem.,
2007, 50, 3004–3014.
21 M. P. Sandrini and J. Piskur, Trends Biochem. Sci., 2005, 30, 225–228.
22 K. Johansson, S. Ramaswamy, C. Ljungcrantz, W. Knecht,
J. Piskur, B. Munch-Petersen, S. Eriksson and H. Eklund,
Nat. Struct. Biol., 2001, 8, 616–620.
23 J. Y. Liou, G. E. Dutschman, W. Lam, Z. Jiang and Y. C. Cheng,
Cancer Res., 2002, 62, 1624–1631.
24 J. A. Alexandre, B. Roy, D. Topalis, S. Pochet, C. Perigaud and
´
D. Deville-Bonne, Nucleic Acids Res., 2007.
25 S. Schaertl, M. Konrad and M. A. Geeves, J. Biol. Chem., 1998,
273, 5662–5669.
45 E. James and J. F. Morrison, J. Biol. Chem., 1966, 241, 4758–4770.
46 K. M. Plowman and A. R. Krall, Biochemistry, 1965, 4, 2809–2814.
47 B. R. Brooks, C. L. Brooks, 3rd, A. D. Mackerell, Jr., L. Nilsson,
R. J. Petrella, B. Roux, Y. Won, G. Archontis, C. Bartels,
S. Boresch, A. Caflisch, L. Caves, Q. Cui, A. R. Dinner,
M. Feig, S. Fischer, J. Gao, M. Hodoscek, W. Im, K. Kuczera,
T. Lazaridis, J. Ma, V. Ovchinnikov, E. Paci, R. W. Pastor,
C. B. Post, J. Z. Pu, M. Schaefer, B. Tidor, R. M. Venable,
H. L. Woodcock, X. Wu, W. Yang, D. M. York and
M. Karplus, J. Comput. Chem., 2009, 30, 1545–1614.
48 A. Geerlof, F. Travers, T. Barman and C. Lionne, Biochemistry,
2005, 44, 14948–14955.
´
49 A. Varga, C. Lionne, P. Lallemand, J. Szabo, N. Adamek,
C. Valentin, M. Vas, T. Barman and L. Chaloin, Biochemistry,
2009, 48, 6998–7008.
50 P. Krishnan, J. Y. Liou and Y. C. Cheng, J. Biol. Chem., 2002, 277,
31593–31600.
51 G. Auerbach, R. Huber, M. Grattinger, K. Zaiss, H. Schurig,
¨
26 P. Gonin, Y. Xu, L. Milon, S. Dabernat, M. Morr, R. Kumar, M. L.
Lacombe, J. Janin and I. Lascu, Biochemistry, 1999, 38, 7265–7272.
27 C. K. Wennefors, M. I. Dobrikov, Z. Xu, P. Li and B. R. Shaw,
Bioorg. Chem., 2008, 36, 169–177.
28 K. S. Anderson, Biochim. Biophys. Acta, 2002, 1587, 296–299.
and G. Gosselin, Antiviral Res., 2006, 71, 276–281.
29 C. Mathe
´
30 W. K. Krietsch and T. Bucher, Eur. J. Biochem., 1970, 17, 568–575.
R. Jaenicke and U. Jacob, Structure, 1997, 5, 1475–1483.
52 G. J. Davies, S. J. Gamblin, J. A. Littlechild, Z. Dauter,
K. S. Wilson and H. C. Watson, Acta Crystallogr., Sect. D: Biol.
Crystallogr., 1994, D50, 202–209.
¨
31 A. Varga, J. Szabo
´
, B. Flachner, B. Roy, P. Konarev, D. Svergun,
rigaud, T. Barman, C. Lionne and M. Vas,
Biochem. Biophys. Res. Commun., 2008, 366, 994–1000.
P. Zavodszky, C. Pe
´
´
32 C. Gondeau, L. Chaloin, A. Varga, B. Roy, P. Lallemand,
53 G. Gumina, Y. Chioi and C. K. Chu, in Antiviral Nucleosides:
Chiral Synthesis and Chemotherapy, ed. C. K. Chu, Elsevier B.V.,
2003, pp. 1–76.
´
54 L. Zerrad, A. Merli, G. F. Schroder, A. Varga, E. Graczer,
¨
P. Pernot, A. Round, M. Vas and M. W. Bowler, J. Biol. Chem.,
C. Pe
2008, 47, 3462–3473.
33 C. Gondeau, L. Chaloin, P. Lallemand, B. Roy, C. Pe
´
rigaud, T. Barman, M. Vas and C. Lionne, Biochemistry,
´
rigaud,
T. Barman, A. Varga, M. Vas, C. Lionne and S. T. Arold,
Nucleic Acids Res., 2008, 36, 3620–3629.
2011, DOI: 10.1074/jbc.M110.206813, in press.
c
This journal is The Royal Society of Chemistry 2011
Mol. BioSyst., 2011, 7, 1863–1873 1873