Bioreversible protection of nucleoside diphosphates

Article abstract of DOI:10.1002/anie.200803100

(Figure Presented) Going incognito: A new prodrug approach has been developed to facilitate the diffusion of highly polar polyphosphorylated nucleosides across cell membranes (see scheme). Inside the cell, the masking groups on the nucleoside diphosphate

Full text of DOI:10.1002/anie.200803100

Angewandte
Chemie
DOI: 10.1002/anie.200803100
Nucleoside Diphosphate Prodrugs
Bioreversible Protection of Nucleoside Diphosphates**
Henning Jacob Jessen, Tilmann Schulz, Jan Balzarini, and Chris Meier*
Nucleoside analogues are applied widely in antiviral and
antitumor therapy. A severe limitation of these compounds is
that they must undergo biotransformation into the corre-
sponding active nucleoside triphosphates (NTPs) by the
stepwise addition of phosphate groups by kinases.[1] If this
activation proceeds insufficiently, the antiviral or antitumor
activity of the nucleoside analogues can possibly be enhanced
by using prodrugs of the phosphorylated metabolites. This
approach bypasses nucleoside kinases, which are responsible
for the often inefficient activation of the analogues, and hence
leads to an increase in the intracellular levels of active
metabolites. Drug resistance can also be overcome in some
cases, and the spectrum of application of a nucleoside may be
broadened further to cover multiple viruses.[2] Moreoever,
masked nucleotides are able to penetrate cell membranes in
their intact form as a result of their high lipophilicity and are
therefore not prone to degradation by nonspecific plasma
phosphatases. For these reasons, various prodrugs for nucleo-
side monophosphates (NMPs) have been designed, for
example, in cycloSal, phosphoramidate, bis(S-acylthioethyl),
or bis(pivaloxymethyl) approaches.[3] However, the design of
nucleoside diphosphate (NDP) or triphosphate (NTP) pro-
drugs has only been addressed very rarely. The lack of
research towards the development of such prodrugs is
remarkable, because it is known, for example, that 3’-azido-
3’-deoxythymidine (AZT), the first approved nucleosidic
anti-HIV drug, is only phosphorylated very slowly to the
diphosphate AZTDP by thymidylate kinase.[4] The resulting
accumulation of AZT monophosphate (AZTMP) provokes
severe side effects.[5]
phosphate diesters based on glycerides were reported by
Hostetler and co-workers.[6] However, these compounds do
not serve as NDP prodrugs, but instead release the corre-
sponding NMPs through cleavage of the pyrophosphate
group. Another approach was described by Huynh-Dinh
and co-workers,[7] who attached different acyl moieties to the
b phosphate group of the pyrophosphate unit. This method
relies on a faster cleavage of the mixed anhydride bond than
the cleavage of the phosphate anhydride bond. The concept
was proven in hydrolysis studies in an aqueous buffer.
However, undesired decomposition occurred in biological
media (RPMI culture medium).[8]
We first attempted to apply the cycloSal approach to the
lipophilic modification of NDPs. It had been shown in
extensive studies that the cycloSal system enhances the
antiviral activity of some nucleoside analogues considerably
when used to mask NMPs.[3b] After the preparation of several
cycloSal nucleoside diphosphates (cycloSal-NDPs), we ana-
lyzed the hydrolysis pathways by 31P NMR spectroscopy
(Scheme 1; paths a and b). We found that the compounds did
not release NDPs effectively, as a result of the initial chemical
Scheme 1. Hydrolysis of cycloSal NDP prodrugs by paths a and b.
ONucl=nucleoside.
The reason for the difficulty in masking NDPs lipophili-
cally lies in the inherent instability of the phosphate
anhydride bond. This bond is only stable kinetically as a
result of the negative charges, which prevent nucleophilic
attack at the phosphate moieties. Several nucleoside pyro-
activation step. The hydrolytic cleavage of the pyrophosphate
bond dominated (Scheme 1, path b), with the predominant
release of the corresponding NMP and cycloSal phosphate.
Only small amounts of the NDP were detected.
To circumvent these problems, we investigated an enzy-
matically activated type of prodrug:[9] bis(4-acyloxybenzyl)-
nucleoside diphosphates (BAB-NDPs, 1). The general struc-
ture of these prodrugs and the proposed mechanism of
hydrolysis are shown in Scheme 2. To gain insight into the
behavior of these potential NDP prodrugs, we prepared
different derivatives by varying the nucleoside and the acyl
moiety. In this context, we were interested in modulating their
stability and polarity. The removal of the masking units
should be initialized by hydrolysis of the acyl ester bond,
either by pH-dependent chemical hydrolysis or by enzyme-
catalyzed hydrolysis.
This initial cleavage results in an inversion of the polarity
of the substituent (an acceptor is transformed into a donor)
and thus to the destabilization of the benzyl phosphate ester
bond. Through 1,4-elimination and hydrolysis, the prodrug
decomposes to give 4-hydroxybenzylalcohol and the mono-
[*] Dr. H. J. Jessen, T. Schulz, Prof. Dr. C. Meier
Organic Chemistry, Department of Chemistry
Faculty of Science, University of Hamburg
Martin-Luther-King-Platz 6, 20146 Hamburg (Germany)
Fax: (+49)40-42838-2495
E-mail: chris.meier@chemie.uni-hamburg.de
Homepage: http://www.chemie.uni-hamburg.de/oc/meier/
Prof. Dr. J. Balzarini
Rega Institute for Medical Research, Katholieke Universiteit Leuven
Minderbroedersstraat 10, 3000 Leuven (Belgium)
[**] We thank Leen Ingels and Lizette van Berckelaer for excellent
technical assistance. The research was supported by the University
of Hamburg (C.M.) and by a grant from the K.U.Leuven (GOA no.
05/19; J.B.)
Supporting information for this article, including experimental
details, is available on the WWW under http://dx.doi.org/10.1002/
anie.200803100.
Angew. Chem. Int. Ed. 2008, 47, 8719 –8722
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
8719

Communications
the purification procedures, so that compounds could
be prepared in yields ranging from 46 to 65%.
Next, we determined the stability of compounds
1a–h in different media (phosphate buffer (PBS,
pH 7.3), citrate/HCl buffer (pH 2.0), T-lymphocyte
(CEM/0) cell extracts, 20% human plasma, and
RPMI culture medium). We also investigated the
mechanism of hydrolysis by ion-pair reversed-phase
HPLC analysis and 31P NMR spectroscopic studies.
The obtained data confirmed the mechanism postu-
lated in Scheme 2. For example, BAB-AZTDP (1e)
underwent hydrolysis via the singly masked inter-
mediate AB-AZTDP (2), which was isolated and
characterized, to give AZTDP almost exclusively at
physiological pH (PBS, pH 7.3; see the Supporting
Information). The determined half-lives of BAB-
AZTDP (1e) in PBS for this process based entirely on
chemical hydrolysis were 17 h for the degradation of the first
masking group to give AB-AZTDP (2) and 110 h for the
cleavage of the second masking group to give AZTDP. Only a
very small amount ( ꢀ 5%) of AZT monophosphate, which
results from the cleavage of the anhydride bond of BAB-
AZTDP (1e), was observed. The incubation of AB-AZTDP
(2) in PBS did not lead to the formation of AZT mono-
phosphate; thus, the anhydride bond in this compound is not
prone to hydrolysis. In line with our expectation, the chemical
stability of the compounds increased if branched alkyl
residues were used in the acyl moiety. For example, the
half-life of BIB-AZTDP (1 f) in PBS is considerably higher
(39 h) than that of 1e (Table 1). Similarly high stability was
Scheme 2. Structure of BAB-NDP prodrugs and proposed mechanism of enzy-
matic hydrolysis.
substituted NDP (AB-NDP; e.g. 2, Scheme 2). Repetition of
this cleavage mechanism should then lead to formation of the
NDP.
In the mechanism shown in Scheme 2, no reaction takes
place at the phosphate anhydride bond. Thus, possible side
reactions are minimized. Moreover, we showed in preliminary
studies that the a-phosphate atom has to remain unmasked to
prevent rapid cleavage of the anhydride bond (data not
shown).
The compounds were synthesized by dicyanoimidazole-
mediated coupling of para-acyloxybenzylphosphoramidites
with bis(tetra-n-butylammonium)nucleoside monophos-
phates and subsequent oxidation with tert-butyl hydroper-
oxide (TBHP; see the Supporting Information). The products
were purified by chromatography on RP-18 silica gel by
gradient elution with water/methanol mixtures. First, 2’,3’-
dideoxy-2’,3’-didehydrothymidine (d4T) and AZT were
chosen as antivirally active nucleoside analogues. The deriv-
atives synthesized and nonoptimized yields are summarized in
Scheme 3. During the course of the project, the yields were
increased considerably, in particular through optimization of
Table 1: Half-lives (t1/2 in h) of 1a and 1c–h in different media.
pH 7.3[a]
CEM/0[b]
plasma[c]
RPMI/FCS[d]
1a
1c
1d
1e
1 f
10
63
82
17
39
75
82
0.05
1
7
0.02
0.3
1
n.d.[e]
12
13
5
10
24
4
n.d.
3
n.d.
2
2.5
3
1
1g
1h
0.5
[a] Phosphate buffer (25 mm). [b] Human T-lymphocyte cell extract
(pH 6.9). [c] Human plasma (20%) diluted withPBS (80%, 50 m m,
pH 6.8). [d] RPMI incubation medium with10% heat-inactivated fetal
calf serum (FCS). [e] Not determined.
observed in citrate/HCl buffer (pH 2.0). In contrast, the
incubation of BAB-AZTDP (1e) in CEM/0 cell extracts led
to a dramatic acceleration of the cleavage reaction, as the
phenyl ester moiety was now hydrolyzed enzymatically. The
first hydrolysis to form AB-AZTDP (2) was 500 times faster
(t1/2 = 2 min) than the corresponding chemical hydrolysis. A
half-life of 3 min was observed for the second hydrolysis step
to give AZTDP (> 95%), which corresponds to a 2500-fold
acceleration with respect to the chemical hydrolysis. This
result is quite astonishing, as the enzymatic cleavage of the
second masking unit in BAB-nucleotide prodrugs was sig-
Scheme 3. Combination of the nucleosides and acyl moieties in
prodrugs 1a–h. BAB=bis(4-acetoxybenzyl), BPB=bis(4-pivaloyloxy-
benzyl), BIB=bis(4-isobutyryloxybenzyl), BOB=bis(4-octanoyloxyben-
zyl), BBB=bis(4-benzoyloxybenzyl).
8720 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 8719 –8722

Angewandte
Chemie
nificantly slower than the first step.[9c] The HPLC traces in
Figure 1show the selective cleavage of compound 1e after
incubation in CEM/0 cell extracts. Identical results were
stable in human plasma than in cell extracts, in which the
esterase activity is significantly higher.
We determined the half-lives of the compounds in the
incubation medium used in anti-HIV tests (RPMI/FCS). We
found a destabilization of all compounds in this medium
relative to their stability in phosphate buffer. We then
evaluated compounds 1a–h for their ability to inhibit the
replication of HIV in T-lymphocyte cells. For this purpose, a
suspension of wild-type CEM cells was infected with HIV-1or
HIV-2, and a mutant thymidine kinase deficient CEM cell
culture was infected only with HIV-2. The infected cell
suspensions (100 mL per well) were transferred to a 96-well
microtiter plate and mixed with solutions of the test
compounds at the appropriate dilution (100 mL). After 4–
5 days, giant-cell formation was recorded microscopically in
the HIV-infected cell cultures. Although d4T derivatives 1a–d
were found to be less stable in RPMI/FCS culture medium
than in phosphate buffer, compounds 1c and 1d retained anti-
HIVactivity in the thymidine kinase deficient cell line (CEM/
TKÀ; Table 2). This result is a first proof of the ability of these
Figure 1. HPLC traces for BAB-AZTDP (1e) after incubation in CEM/0
cell extracts for the time shown (0–90 min).
Table 2: Antiviral activity of 1a–d against HIV-1 and HIV-2.
EC50 [mm][a]
CEM/0[b]
HIV-1
EC50 [mm]
CEM/0
HIV-2
EC50 [mm]
CEM/TKÀ[c]
HIV-2
CC50 [mm][d]
CEM/0
obtained for BAB-d4TDP (1a). To the best of our knowledge,
these compounds are the first NDP prodrugs to be described
that exhibit high chemical stability and undergo fast and
highly selective enzymatic cleavage in a cell extract to deliver
NDPs.
1a
1b
1c
1d
d4T
1.05
0.78
0.90
0.20
0.40
1.5
21.0
14
2.6
0.85
70
>125
66Æ1
81Æ1
36Æ5
>100
0.90
1.30
0.30
0.3
Extremely rapid and highly selective degradation in cell
extracts is crucial to the success of this method, as undesired
decomposition of the phosphate anhydride bond is thus
suppressed. In contrast to the design of NMP prodrugs (e.g.
the cycloSal approach), we believe that only concepts based
on fast enzyme-catalyzed cleavage at least in the first
activating step without the involvement of the phosphate
anhydride bond will be successful for the design of NDP
prodrugs. As already mentioned, we were able to increase the
chemical stability of the compounds by introducing branching
in the acyl moiety. However, this modification also enhanced
the enzymatic stability of the compounds (compare data for
1e–g, Table 1).
Strikingly, an increase in enzymatic stability led to a
decrease in the amount of NDP released. In the case of BIB-
AZTDP (1 f; t1/2 = 20 min, first hydrolysis), the amount of
AZTDP released dropped to 60% in the CEM/0 cell extract.
BOB-d4TDP (1c; t1/2 = 60 min, first hydrolysis) only released
about 35% d4TDP in the CEM/0 cell extract. Derivative 1g,
which contains pivaloyl ester groups, released no AZTDP, as
the intermediate PB-AZTDP (analogous to 2) was not
degraded by the esterase. In this case, we observed complete
cleavage of the phosphate anhydride bond with the formation
of various products. The following requirements for the
development of an efficient diphosphate prodrug can be
derived from these results: Despite the desired high chemical
stability, the masks must be removed in biological media as
fast as possible; otherwise, hydrolysis of the pyrophosphate
competes effectively. Half-lives of 1–30 min enable the
formation of large amounts of NDP. Although the compounds
lose some stability in human plasma, they are much more
[a] Antiviral activity in T lymphocytes: 50% effective concentration
(values shown are the mean of two or three independent experiments).
[b] Wild-type T lymphocytes. [c] Thymidine kinase deficient T lympho-
cytes. [d] Cytostatic concentration: CC50 is the concentration of the
compound required to inhibit CEM/0 cell proliferation by 50%.
compounds to diffuse across cell membranes and release
biologically active metabolites intracellularly. We think that
compounds 1a and 1b failed to bypass TK for different
reasons. Although 1a released d4TDP efficiently in the other
experiments described, it may be too polar to penetrate the
cell membrane efficiently and is thus degraded in RPMI/FCS.
The lack of activity of compound 1b may result from an
inability to be fully degraded to unmasked d4TDP, as only the
first masking unit was cleaved in CEM/0 extracts. None of the
compounds examined displayed marked cytotoxicity.
In summary, our study has revealed the requirements for
the bioreversible protection of nucleoside diphosphates as
prodrugs. In principle, the BAB concept presented herein
should be applicable to other nucleoside analogues. We are
currently optimizing of the properties of different target
structures with respect to hydrolysis and lipophilicity.
Received: June 27, 2008
Published online: October 2, 2008
Keywords: antiviral agents · drug delivery · medicinal chemistry ·
nucleotides · prodrugs
.
Angew. Chem. Int. Ed. 2008, 47, 8719 –8722
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8721

Communications
Ilsley, C. Frohlick, R. Street, E. T. Hall, R. D. Kuchta, P. Melanco,
J. Biol. Chem. 1995, 270, 22836 – 22841.
[1] a) J. Balzarini, P. Herdewijn, E. De Clercq, J. Biol. Chem. 1989,
264, 6127 – 6133; b) J. Balzarini, Pharm. World Sci. 1994, 16, 1 1 3 –
126.
[6] a) K. Y. Hostetler, L. M. Stuhmiller, H. B. M. Lenting, H. van
den Bosch, D. D. Richman, J. Biol. Chem. 1990, 265, 6112 – 6117;
b) G. M. T. van Wijk, K. Y. Hostetler, H. van den Bosch, Biochim.
Biophys. Acta Lipids Lipid Metab. 1991, 1084, 307 – 310.
[7] a) A. Kreimeyer, J. Ughetto-Monfrin, A. Namane, T. Huynh-
Dinh, Tetrahedron Lett. 1996, 37, 8739 – 8742; b) D. Bonnaffꢀ, B.
Dupraz, J. Ughetto-Monfrin, A. Namane, T. Huynh-Dinh,
Tetrahedron Lett. 1995, 36, 531– 534; c) A. Kreimeyer, F.
Andrꢀ, C. Gouyette, T. Huynh-Dinh, Angew. Chem. 1998, 110,
3013 – 3016; Angew. Chem. Int. Ed. 1998, 37, 2853 – 2855.
[8] D. Bonnaffꢀ, B. Dupraz, J. Ughetto-Monfrin, A. Namane, Y.
Henin, T. Huynh-Dinh, J. Org. Chem. 1996, 61, 895 – 902.
[9] The bis(AB) approach was used to mask nucleoside mono-
phosphates, but was not pursued further: a) A. Routledge, I.
Walker, S. Freeman, A. Hay, N. Mahmood, Nucleosides Nucleo-
tides 1995, 14, 1545 – 1558; b) S. Freeman, K. C. Ross, Prog. Med.
Chem. 1997, 34, 111 – 147; c) C. Meier, U. Muus, J. Renze, L.
Naesens, E. De Clercq, J. Balzarini, Antiviral Chem. Chemother.
2002, 13, 101 – 114.
[2] a) C. Meier, T. Knispel, E. De Clercq, J. Balzarini, J. Med. Chem.
1999, 42, 1604 – 1614; b) C. Meier, A. Lomp, A. Meerbach, P.
Wutzler, J. Med. Chem. 2002, 45, 5157 – 5172; c) D. Cahard, C.
McGuigan, J. Balzarini, Mini-Rev. Med. Chem. 2004, 4, 371– 381,
and references therein.
[3] a) X. Tan, C. K. Chu, F. D. Boudinot, Adv. Drug Delivery Rev.
1999, 39, 117 – 151; b) C. Meier, Eur. J. Org. Chem. 2006, 1081 –
1102; c) C. R. Wagner, V. V. Iyer, E. J. McIntee, Med. Res. Rev.
2000, 20, 417 – 451; d) R. J. Jones, N. Bischofsberger, Antiviral Res.
1995, 27, 1– 17, and references therein.
[4] a) A. Lavie, I. Schlichting, I. R. Vetter, M. Konrad, J. Reinstein,
R. S. Goody, Nat. Med. 1997, 3, 922 – 924; b) A. Lavie, I. R.
Vetter, M. Konrad, R. S. Goody, J. Reinstein, I. Schlichting, Nat.
Struct. Biol. 1997, 4, 601– 604.
[5] a) J. P. Sommadossi, R. Carlisle, Z. Zhou, Mol. Pharmacol. 1989,
36, 9 – 14; b) J. A. Harrington, J. E. Reardon, T. Spector, Anti-
microb. Agents Chemother. 1993, 37, 918 – 920; c) J. P. Yan, D. D.
8722
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 8719 –8722