Novel diadenosine polyphosphate analogs with
oxymethylene bridges replacing oxygen in the
polyphosphate chain
Potential substrates and/or inhibitors of Ap
4
A hydrolases
Andrzej Guranowski
1
,El
_
zbieta Starzyn
´
ska
1
, Małgorzata Pietrowska-Borek
2
, Dominik Rejman
3,
* and
George M Blackburn
3
1 Department of Biochemistry and Biotechnology, University of Life Sciences, Poznan
´
, Poland
2 Department of Plant Physiology, University of Life Sciences, Poznan
´
, Poland
3 Department of Molecular Biology and Biotechnology, University of Sheffield, UK
Dinucleoside 5¢,5¢¢¢-P
1
,P

Tel: +48 61 8487201
E-mail:
G. M. Blackburn, Department of Molecular
Biology and Biotechnology, Sheffield
University, Sheffield S10 2TN, UK
Fax: +44 1142222800
Tel: +44 1142229462
E-mail: g.m.blackburn@sheffield.ac.uk
*Present address
Institute of Organic Chemistry and
Biochemistry AS CR, v.v.i., Prague, Czech
Republic
(Received 24 June 2008, revised 3
December 2008, accepted 30 December
2008)
doi:10.1111/j.1742-4658.2009.06882.x
Dinucleoside polyphosphates (Np
n
N¢s; where N and N¢ are nucleosides
and n = 3–6 phosphate residues) are naturally occurring compounds that
may act as signaling molecules. One of the most successful approaches to
understand their biological functions has been through the use of Np
n
N¢
analogs. Here, we present the results of studies using novel diadenosine
polyphosphate analogs, with an oxymethylene group replacing one or two
bridging oxygen(s) in the polyphosphate chain. These have been tested as
potential substrates and/or inhibitors of the symmetrically acting Ap
4
A

OpOpA (6). The eukaryotic asymmetrical Ap
4
A hydrolases
degrade two compounds, 3 and 5 , as anticipated in their design. Analog 3
was cleaved to AMP (pA) and b,c-methyleneoxy-ATP (pOCH
2
pOpA),
whereas hydrolysis of analog 5 gave two molecules of a,b-oxymethylene
ADP (pCH
2
OpA). The relative rates of hydrolysis of these analogs were
estimated. Some of the novel nucleotides were moderately good inhibitors
of the asymmetrical hydrolases, having K
i
values within the range of the
K
m
for Ap
4
A. By contrast, none of the six analogs were good substrates or
inhibitors of the bacterial symmetrical Ap
4
A hydrolase.
Abbreviations
DCC, dicyclohexylcarbodiimide; MCPBA, 4-chloroperoxybenzoic acid; NEP, 2-chloro-5,5-dimethyl-2-oxido-1,3,2-dioxaphosphinane.
1546 FEBS Journal 276 (2009) 1546–1553 ª 2009 The Authors Journal compilation ª 2009 FEBS
nonspecific enzymes exist that degrade these dinucleo-
tides to mononucleotides [12]. Ap
3
A and Ap

4
A hydrolase from Caenorhabditis elegans, was
used to determine the 3D structure of the enzyme–sub-
strate complex [15]. Some nondegradable analogs
appeared to be extremely strong inhibitors of the
Ap
4
A hydrolases; two adenosine-5¢-O-phosphorothioy-
lated pentaerythritols are strong inhibitors of the (sym-
metrical) Ap
4
A hydrolase from Escherichia coli (with
K
i
values of 0.04 and 0.08 lm) [16], and methylene
analogues of adenosine 5¢-tetraphosphate (p
4
A)
strongly inhibited the asymmetrically acting Ap
4
A
hydrolases with K
i
values in the nanomolar range [17].
Finally, potential medical application has been demon-
strated for AppCHClppA, a competitive inhibitor of
ADP-induced platelet aggregation, which plays a
central role in arterial thrombosis and plaque forma-
tion [18], and for [
18

4
A. As observed for a series of bb¢-
substituted Ap
4
A analogs, their efficiencies as
substrates of the Ap
4
A hydrolase from Artemia
increased in direct proportion to increasing electroneg-
ativity [22]. Guranowski et al. [21] found that those
compounds were not substrates of the symmetrically
acting Ap
4
A hydrolase from E. coli, but later work by
McLennan et al. [22] reported that AppCH
2
ppA,
AppCF
2
ppA and AppCHFppA underwent slow
hydrolysis using their preparation of bacterial enzyme,
with 25-, 50- and 125-fold reduced rates, respectively,
compared with that of Ap
4
A hydrolysis.
In this report we describe, first, the chemical synthe-
sis of new Ap
n
A analogs with a methyleneoxy or an
oxymethylene bridge that substitutes for one or two

nonisosteric (the P–P distance is one atom longer), yet
isoelectronic (charge identical), in comparison with
natural Ap
n
As. To answer this question, we performed
studies on the interaction of the enzymes with the
aforementioned oxymethylene analogs of Ap
n
A. When
analyzing the reaction mixtures in the TLC system that
separates each of the analogs tested, as a potential sub-
strate, from possible reaction products, we found that
none of the six new Ap
n
A analogs was a substrate of
the symmetrically acting Ap
4
A hydrolase. Each analog
(0.5 mm) was incubated at 30 °C in 0.05 mL of the
reaction mixture containing 50 mm Hepes/KOH
(pH 7.6), 0.02 mm dithiothreitol and 5 mm MgCl
2
, for
up to 16 h with an amount of enzyme sufficient to
achieve complete cleavage of 0.5 mm Ap
4
Ain
< 15 min. This result is consistent with previously
published results [20–23], which established that the
hydrolase from E. coli shows almost no cleavage of

A hydrolases, compounds 3 and 5 were
readily hydrolyzed. This was demonstrated both for
the human and the plant enzymes, and the reaction
products were clearly identified by comparing them
with AMP and synthetic oxymethylene analogs of
ADP or ATP. In addition to TLC analysis, we also
used an HPLC system (see the example of elution pro-
files in Figs 2A,B) that effectively separated potential
substrates from possible products and thus could be
used to estimate the relative velocities of the hydrolysis
reactions (Table 1). The asymmetric analog 3 was first
hydrolyzed by both asymmetric hydrolases to AMP
and the bc-methyleneoxy-ATP (32) (Fig. 3A), and then
the latter, relatively unstable, nucleotide hydrolyzed
spontaneously to give a second AMP.
An alternative cleavage of analog 3 to AMP and
bc-oxymethylene-ATP (18) was also observed. For the
human asymmetric hydrolase this mode of cleavage
was approximately six times less frequent than the
dominant mode and in the case of the lupin enzyme it
was over 20 times slower. Such slower cleavage to give
18 could arise either from weaker binding of 3 in the
active site of the hydrolase in the reverse orientation
(Fig. 3B) or from a reduced rate of cleavage. While
A
B
Fig. 2. Time course of ApOpCH
2
OpOpA hydrolysis catalyzed by
narrow-leaved lupin Ap

, pH 6.0 (solvent A); solvent
A/methanol (9 : 1, v/v) (solvent B): 0–9 min, 0% B; 9–15 min, 25%
B; 15–17.5 min, 90% B; 17.5–19 min, 100% B; 19–23 min, 100%
B and 23–35 min, 0% B. Profiles in (B) show standards run under
identical conditions.
Oxymethylene diadenosine polyphosphate analogs A. Guranowski et al.
1548 FEBS Journal 276 (2009) 1546–1553 ª 2009 The Authors Journal compilation ª 2009 FEBS
the pK
a
values for the ATP analogs released (32 and
18) have not yet been determined, it is reasonable to
assume that a pK
a
value of 4 for 32 is similar to that
of ATP (ca. 7.1), whereas that for 18 will be similar to
that of bc-methylene-ATP (ca. 8.2. [27]). The asym-
metrical pyrophosphohydrolase from Artemia is known
to exhibit a strong dependence on the rate of cleavage
on the pK
a
of the leaving group (Brønsted coefficient
0.5 [22]). A similar b-leaving group-dependence for the
human and lupin enzymes studied here would lead to
a reduction in rate of about 10-fold for the formation
of 18 relative to that of 32. Thus, the present kinetic
results do not provide any evidence for differential rec-
ognition of the alternative orientations on the P-O-C-P
bridge for these two enzymes.
The enzymatic hydrolysis of symmetrical analog 5,
by both human and plant asymmetric hydrolases,

recognize dinucleoside triphosphates as substrates.
Thus, it was to be expected that the oxymethylene ana-
logs of Ap
3
A – compounds 1 and 2 – would not be
degraded. The absence of any detectable hydrolysis of
compounds 4 and 6 suggests that the enzymes tolerate
neither a -CH
2
-P
a
- sequence, which occurs in 4, nor a
-CH
2
-P
c
-CH
2
- sequence, as in 6. Apparently, ‘the
frameshift’ is unable to accommodate two oxymethyl-
ene inserts, as occurs in 6.
Finally, we investigated whether the novel Ap
n
A
analogs inhibit Ap
4
A hydrolysis catalyzed by the
asymmetrically acting Ap
4
A hydrolases. Only analogs

hydrolase from
human
Narrow-leaved
lupin
Ap
fl
pppA 1 1
ApCH
2
OpOCH
2
pA (1)0 0
ApOCH
2
pCH
2
OpA (2)0 0
Ap
fl
OpCH
2
OpO
fl
pA (3) 0.48 0.92
ApCH
2
OpOpOCH
2
pA (4)0 0
ApOCH

tations of the P-O-C-P bridge as a surrogate for
pyrophosphate in nucleotides. First, exactly as
expected, none of the three enzymes can cleave the
P–O bond in the P-O-CH
2
-P linkage. Second, the
asymmetric cleaving enzymes accept the P-O-C-P
bridge in the position adjacent to the P-O-P cleavage
locus in either orientation. Third, hindrance of normal
P-O-P cleavage can lead to a frameshift response, even
though this involves a three-atom shift, but only for
one orientation of the P-O-C-P insert. Lastly, the
asymmetric hydrolases accept the P-O-C-P inserts as
competitive inhibitors, whereas the bacterial symmetri-
cal hydrolase does not. Thus, these novel compounds
will be tools of specific application for studies on the
metabolism of dinucleoside polyphosphates and on
Ap
4
A-degrading enzymes and they also merit further
attention for the investigation of nucleotide metabolic
pathways. Kindred studies on the full range of ATP
analogs containing an oxymethylene bridge will be
reported in due course.
Experimental procedures
Enzymes
Homogeneous recombinant asymmetrically acting human
Ap
4
A hydrolase (EC 3.6.1.17) [24] was kindly donated by

which was developed in dioxane/ammonia/water (6 : 1 : 4,
v/v/v).
Enzyme assays
Estimation of the reaction rates and calculation of the K
i
values for the analogs with the use of radiolabeled Ap
4
A
were performed as described previously [16]. Relative rates
of the hydrolysis of dinucleotide substrates and analogs
were estimated by the use of HPLC on a reverse-phase col-
umn (for details see the legend to Fig. 2a) and were based
on peak-area analysis.
Synthesis of oxymethylene and methyleneoxy
analogs of ADP, ATP and Ap
n
A
ADP, ATP and Ap
n
A analogs with one -OCH
2
- or -CH
2
O-
group that substitutes for a bridging oxygen in adenosine or
diadenosine oligophosphates have not been synthesized pre-
viously. The tripolyphosphate analog, pOCH
2
pCH
2

3
A (ApCH
2
Op-
OCH
2
pA) (1), ab,a¢b-bis(oxymethylene)Ap
3
A (ApOCH
2
pC-
H
2
OpA) (2), bb¢-methyleneoxy-Ap
4
A (ApOpCH
2
OpOpA)
(3), ab,a¢b¢-bis(methyleneoxy)Ap
4
A (ApCH
2
OpOpOCH
2
pA)
(4), ab,a¢b¢-bis(oxymethylene)Ap
4
A (ApOCH
2
pOpCH

King BF (2001) Diadenosine polyphosphates as extra-
cellular signal molecules. Drug Dev Res 52, 260–273.
Oxymethylene diadenosine polyphosphate analogs A. Guranowski et al.
1550 FEBS Journal 276 (2009) 1546–1553 ª 2009 The Authors Journal compilation ª 2009 FEBS
3 McLennan AG, Barnes LD, Blackburn GM, Brenner
Ch, Guranowski A, Miller AD, Rovira JM, Rotlla
´
nP,
Soria B, Tanner JA et al. (2001) Recent progress in the
study of the intracellular functions of diadenosine
polyphosphates. Drug Dev Res 52, 249–259.
4 McLennan AG (2000) Dinucleoside polyphosphates –
friend or foe? Pharmacol Ther 87, 73–89.
5 Zamecnik PC, Stephenson ML, Janeway CM & Rande-
rath K (1966) Enzymatic synthesis of diadenosine tetra-
phosphate and diadenosine triphosphate with a purified
lysyl-sRNA synthetase. Biochem Biophys Res Commun
24, 91–97.
6 Sillero A & Gu
¨
nther Sillero MA (2000) Synthesis of
dinucleoside polyphosphates catalyzed by firefly lucifer-
ase and several ligases. Pharmacol Ther 87, 91–102.
7 Fontes R, Gu
¨
nther Sillero MA & Sillero A (1998) Acyl
coenzyme A synthetase from Pseudomonas fragi cata-
lyzes the synthesis of adenosine 5¢-polyphosphates and
dinucleoside polyphosphates. J Bacteriol 180, 3152–
3158.

4
U and
p
5
U) and uridine(5¢)polyphospho(5¢)nucleosides
(Up
n
Ns) can be synthesized by UTP:glucose-1-phos-
phate uridylyltransferase from Saccharomyces cerevisiae.
FEBS Lett 561, 83–88.
12 Guranowski A (2000) Specific and nonspecific
enzymes involved in the catabolism of mononucleoside
and dinucleoside polyphosphates. Pharmacol Ther 87,
117–139.
13 Guranowski A (2003) Analogs of diadenosine tetra-
phosphate (Ap
4
A). Acta Biochim Polon 50, 947–972.
14 Lazewska D & Guranowski A (1990) P
a
-Chiral phosp-
horothioate analogues of bis(5¢-adenosyl)tetraphosphate
(Ap
4
A); their enzymatic synthesis and degradation.
Nucleic Acids Res 18, 6083–6088.
15 Bailey S, Sedelnikova S, Blackburn GM, Abdelghany
HM, Baker PJ, McLennan AG & Rafferty JB (2002)
The crystal structure of diadenosine tetraphosphate
hydrolase from Caenorhabditis elegans in free and

,P
4
-tetraphosphate (Ap
4
A)
and positron-emission tomography. Proc Natl Acad Sci
USA 103, 15992–15996.
20 Guranowski A, Biryukov A, Tarussova NB, Khomutov
RM & Jakubowski H (1987) Phosphonate analogues of
diadenosine 5¢,5¢¢¢-P
1
,P
4
-tetraphosphate as substrates
or inhibitors of prokaryotic and eucaryotic enzymes
degrading dinucleoside tetraphosphates. Biochemistry
26, 3425–3429.
21 Guranowski A, Starzyn
´
ska E, Taylor GE & Blackburn
GM (1989) Studies on some specific Ap
4
A-degrading
enzymes with the use of various methylene analogues of
P
1
,P
4
-bis(5¢,5¢¢¢-adenosyl) tetraphosphate. Biochem
J 262, 241–244.

4
-tetraphosphate
A. Guranowski et al. Oxymethylene diadenosine polyphosphate analogs
FEBS Journal 276 (2009) 1546–1553 ª 2009 The Authors Journal compilation ª 2009 FEBS 1551
in prokaryotes; purification and properties of diadeno-
sine 5¢,5¢¢¢-P
1
,P
4
-tetraphosphate (symmetrical) pyro-
phosphohydrolase from Escherichia coli K12. J Biol
Chem 258, 14784–14789.
27 Blackburn GM, Kent DA & Kolkmann F (1981) Three
new bc-methylene analogues of adenosine triphosphate.
JCS Chem Commun 1188–1190.
28 Walkowiak B, Baraniak J, Cierniewski Cz S & Stec W
(2002) Inhibition of ADP-triggered blood platelet aggre-
gation by diadenosine polyphosphate analogues. Bioorg
Med Chem Lett 12, 1959–1962.
29 Rejman D, Olesiak M, Chen LQ, Patterson SE, Wilson
D, Jayaram HN, Hedstrom L & Pankiewicz K (2006)
Novel methylenephosphophosphonate analogues of
mycophenolic adenine dinucleotide. Inhibition of
inosine monophosphate dehydrogenase. J Med Chem
49, 5018–5022.
30 Rejman D, Masojı
´
dkova
´
M & Rosenberg I (2004)

OpOCH
2
pA (1) Monobenzyl
phosphonate 8 [29] was esterified with tetrabenzoylade-
nosine 7 [29] using either 2-chloro-5,5-dimethyl-
2-oxido-1,3,2-dioxaphosphinane (NEP)/methoxypyri-
dine-N-oxide/pyridine system [30–32] or Mitsunobu
conditions (Scheme 1). The dimethoxytrityl (DMTr)
group of phosphonate 9 was removed with acetic acid
giving compound 10. Phosphoramidite generated by the
reaction of phosphonate 10 with benzyloxybis(diisopro-
pylamino)phosphine [33] reacted with a second molecule
of 10 to produce the fully protected symmetrical Ap
3
A
analog 11. Target compound 1 was obtained by two-
step deprotection and DEAE-Sephadex column chroma-
tography using a linear gradient of TEAB in water.
Benzyl esters were removed by catalytic hydrogenation
followed by aqueous ammonia treatment to remove
benzoyl protecting groups.
Scheme S2. ApOCH
2
pCH
2
OpA (2) Tetrabenzoyl aden-
osine 7 was converted into phosphoramidite 12 by
reaction with benzyloxybis(diisopropylamino)phos-
phine [33] (Scheme 2). Phosphoramidite 12 underwent
reaction with benzyl bis(hydroxymethane)phosphinate

2
pOpCH
2
OpA (5) Phosphorami-
dite 12 was reacted with dibenzyl phosphonate 22
using tetrazole catalysis and, after MCPBA oxidation,
afforded compound 23 (Scheme 6). ADP analogue 24,
obtained by catalytic hydrogenation of 23, was dimer-
ized using DCC in pyridine giving, after DEAE-Sepha-
dex column chromatography, target Ap
4
A analog 5.
Scheme S6. ApOpOCH
2
pCH
2
OpOpA (6) Benzyl phos-
phinate 13, after treatment with bis-benzyloxy-(diiso-
propylamino)phosphine [33] using tetrazole catalysis
and MCPBA oxidation, gave compound 25 (Scheme 5).
Catalytic hydrogenation of 25 gave bis(hydroxymethyl-
enephosphinic acid) phosphate 26 which underwent
condensation with morpholidate 19 to give, after
DEAE-Sephadex column purification, the target Ap
5
A
analogue 6.
Scheme S7. pOCH
2
pOpA (32) Bis(2-cyanoethyloxy)(di-

described above employed two main synthetic
approaches. Phosphoramidite condensations appeared
as the ideal method and gave excellent yields. Phosp-
horomorpholidate condensation proved to be an alter-
native method and gave moderate to good yields.
While DCC couplings appeared useful, they gave lower
yields. Using the combination of base-labile benzoyl
and hydrogenolytically-removable benzyl groups
proved to be compatible with rather unstable poly-
phosphate products. The structures of all compounds
prepared were established by a combination of
1
H and
31
P NMR and high resolution mass spectroscopy (data
not shown).
This supplementary material can be found in the
online version of this article.
Please note: Wiley-Blackwell is not responsible for
the content or functionality of any supplementary
materials supplied by the authors. Any queries (other
than missing material) should be directed to the corre-
sponding author for the article.
A. Guranowski et al. Oxymethylene diadenosine polyphosphate analogs
FEBS Journal 276 (2009) 1546–1553 ª 2009 The Authors Journal compilation ª 2009 FEBS 1553