Potential lipophilic nucleotide prodrugs: Synthesis, hydrolysis, and antiretroviral activity of AZT and d4T acyl nucleotides

Article abstract of DOI:10.1021/jo951354p

Three general methods for the synthesis of acyl nucleotides (mono-, di-, and triphosphates) have been developed and applied to different HIV inhibitors. These new types of compounds, where a fatty acid moiety is linked to the nucleotide phosphate chain by an acyl phosphate bond, were designed as lipophilic prodrugs of HIV inhibitors metabolites. Acyl nucleoside monophosphates 1a,b were prepared by acylation of the corresponding nucleoside monophosphates. Acyl nucleoside diphosphates 2a-c and 3a,b were synthesized directly from the free nucleosides using DCC activation of acyl pyrophosphates. Acyl nucleoside triphosphates 4a-c and 5a were obtained using phosphoromorpholidate chemistry and acyl pyrophosphates as nucleophiles. Hydrolysis of acyl nucleotides liberated the corresponding nucleotides by selective cleavage of the acyl phosphate bond, with half lives ranging from 51 to 185 h at 37°C in triethylammonium acetate buffer pH 7.0. Their antiretroviral activity, measured by the inhibition of cytopathogenicity and reverse transcriptase activity in the cultures supernatants, did not reveal any differences between an acyl nucleotide and its corresponding nucleotide. These results are explained in term of rapid aminolysis of the acyl phosphate bond in culture media.

Full text of DOI:10.1021/jo951354p

J . Org. Chem. 1996, 61, 895-902
895
P oten tia l Lip op h ilic Nu cleotid e P r od r u gs: Syn th esis, Hyd r olysis,
a n d An tir etr ovir a l Activity of AZT a n d d 4T Acyl Nu cleotid es
David Bonnaffe´,†,§ Bernadette Dupraz,† J oe¨l Ughetto-Monfrin,† Abdelkader Namane,†
Yvette Henin,‡ and Tam Huynh Dinh†,*
Unite´ de Chimie Organique, URA CNRS 487, Unite´ d’Oncologie Virale URA CNRS 1157,
Institut Pasteur, 28 rue du Docteur Roux 75724 Paris Cedex 15, France
Received J uly 24, 1995X
Three general methods for the synthesis of acyl nucleotides (mono-, di-, and triphosphates) have
been developed and applied to different HIV inhibitors. These new types of compounds, where a
fatty acid moiety is linked to the nucleotide phosphate chain by an acyl phosphate bond, were
designed as lipophilic prodrugs of HIV inhibitors metabolites. Acyl nucleoside monophosphates
1a ,b were prepared by acylation of the corresponding nucleoside monophosphates. Acyl nucleoside
diphosphates 2a -c and 3a ,b were synthesized directly from the free nucleosides using DCC
activation of acyl pyrophosphates. Acyl nucleoside triphosphates 4a -c and 5a were obtained using
phosphoromorpholidate chemistry and acyl pyrophosphates as nucleophiles. Hydrolysis of acyl
nucleotides liberated the corresponding nucleotides by selective cleavage of the acyl phosphate bond,
with half lives ranging from 51 to 185 h at 37 °C in triethylammonium acetate buffer pH 7.0.
Their antiretroviral activity, measured by the inhibition of cytopathogenicity and reverse tran-
scriptase activity in the cultures supernatants, did not reveal any differences between an acyl
nucleotide and its corresponding nucleotide. These results are explained in term of rapid aminolysis
of the acyl phosphate bond in culture media.
Although new therapies, such as antisense thiooligo-
nucleotides, protease inhibitors, or cytokine based im-
mune system restoration, are currently under clinical
trials,1 nucleoside analogs that inhibit the human im-
munodeficency virus reverse transcriptase (HIV RT) are
the only drugs widely used to inhibit HIV replication in
vivo. There is still an urgent need for new compounds
which would be active from the beginning of the infection,
since recent findings have shown that HIV is far from
being inactive during the asymptomatic part of the
infection but replicate at a fast rate in lymph nodes and
peripheral blood cells.2 It has been shown that the
activity of some nucleoside analogs could be enhanced
by preparing prodrugs able to deliver nucleoside analogs
5′-monophosphates (NMPs) in infected cells.3 However,
the metabolism of a nucleosidic HIV RT inhibitor in-
volves, in fact, three kinase steps leading to the nucleo-
side analog 5′-triphosphate which is the active HIV RT
inhibitor.4 Prodrugs of nucleoside 5′-di or 5′-triphos-
phates (NDPs or NTPs) would be more interesting since
they would bypass more metabolic steps. Moreover,
severe side effects, such as peripheral neuropathies and
bone marrow toxicity, which result from the accumulation
of the NMP analog4,5 may be avoided.
To our knowledge no NDP nor NTP lipophilic prodrug
has ever been described. Cationic DABCO derived car-
riers, able to complex nucleotides, have been reported by
Li et al.,6 and are effective for liquid membrane nucleo-
tide transport. But, when used with liposomes, these
compounds act like detergents and break the bilayer
structure. We have recently introduced a new type of
compound: acyl nucleotides 1-5, as potential nucleotide
lipophilic prodrugs.7 In such compounds the lipophilic
acyl moiety should allow passive diffusion of the charged
nucleotide through cells membranes, while the mixed
carboxylic phosphoric anhydride is expected to be hydro-
lyzed before the symmetrical phosphoric anhydride,8 thus
allowing the liberation of the corresponding free nucleo-
tides 6-8 (Scheme 1). In order to test these hypothesis
we have developed three different synthetic approaches
leading to the preparation of myristoyl NMPs 1a ,b, acyl
NDPs 2, 3 and acyl NTPs 4, 5 (Table 1). We took
advantage of the bivalent properties of the phosphate
group which is a good nucleophile but which can also
easily be activated to become a good electrophile: in this
work, DCC was alternatively used to activate the acyl
or the phosphate moiety, whereas acyl pyrophosphates
were alternatively used as nucleophiles or as electro-
philes. Acyl NMPs were prepared using DCC activation
of the acyl moiety, followed by displacement of dicyclo-
hexylurea by the NMPs bis(tetrabutylammonium) salts
(Scheme 2). Acyl NDPs were directly prepared from acyl
pyrophosphates and free nucleosides. In this reaction
† Unite´ de Chimie Organique, Institut Pasteur.
‡ Unite´ d’Oncologie Virale, Institut Pasteur. Present address: Rhone-
Poulenc Rorer, 94400 Vitry sur Seine.
§ Present address: Laboratoire de Chimie Organique Multifonc-
tionnelle, URA CNRS 462, Bat. 420, Universite´ Paris Sud, 91405 Orsay
Cedex.
X Abstract published in Advance ACS Abstracts, J anuary 15, 1996.
(1) J ohnston, M. I.; Hoth, D. F. Science 1993, 260, 1286.
(2) Pantaleo, G.; Graziosi, C.; Demarest, J . F.; Cohen, O. J .;
Vaccarezza, M.; Gantt, K.; Muro-Cacho, C.; Fauci, A. S. Immunol. Rev.
1994,140, 105. Emilie, D.; Fior, R.; Llorente, L.; Marfaing-Kora, A.;
Peuchmaur, M.; Devergne, O.; J arrousse, B.; Wudenes, J .; Boue, F.;
Galanaud, P. Immunol. Rev. 1994, 140, 5. Wei, X.; Ghosh, S. K.; Taylor,
M. E.; J ohnson, V. A.; Emini, E. A.; Deutsch, P.; Lifson, J . D.;
Bonhoeffer, S.; Nowak, M. A.; Han, B. H.; Saag, M. S.; Shaw, G. M.
Nature 1995, 373, 117. Ho, D. D.; Neumann, A. U.; Perleson, A. S.;
Chen, W.; Leonars, J . M.; Markowitz, M. Nature 1995, 373, 123.
(3) Huynh Dinh, T. Curr. Opin. Invest. Drugs 1993, 2, 905.
(4) Furman, P. A.; Fyfe, J . A.; St. Clair, M. H.; Weinhold, K.; Rideout,
J . L.; Freeman, G.A.; Lehrman, S. N.; Bolognesi, D. P.; Broder, S.;
Mitsuya, H.; Barry, D. W. Proc. Natl. Acad. Sci. U.S.A. 1986, 83, 8333.
(5) Izuta, S.; Saneyoshi, M.; Sakurai, T.; Suzuki, M.; Kojima, K.;
Yosida, S. Biochem. Biophys. Res. Commun. 1991, 179, 776.
(6) Li, T.; Krasne, S. J .; Persson, B.; Kaback, R.; Dietrich, F. J . Org.
Chem. 1993, 58, 380.
(7) Bonnaffe´, D.; Dupraz, B.; Ughetto-Monfrin, J .; Namane, A.;
Huynh Dinh, T. Tetrahedron Lett. 1995, 36, 531.
(8) Hydrolysis of acetyl phosphate into acetyl and phosphate re-
quires 3-6 kcal less than that of ATP to ADP (see: Biol. Chem.;
Mahler, H. R., Cordes, E. H., Eds.; Harper Int.: New York, 1963; p
201).
0022-3263/96/1961-0895$12.00/0 © 1996 American Chemical Society

896 J . Org. Chem., Vol. 61, No. 3, 1996
Bonnaffe´ et al.
Sch em e 1a
Ta ble 1: Su bstitu en ts a n d P h osp h a te Con ten t for Acyl
Nu cleotid es 1-5, Nu cleotid es 6-8, a n d
P h osp h or om or p h olid a tes 17
Sch em e 2
compd
X
n
R1
N3
CdC
N3
CdC
OAc
N3
CdC
N3
CdC
OH
N3
N3
CdC
N3
R2
1a
1b
2a
2b
2c
3a
3b
4a
4b
4c
5a
6a
6b
7a
7b
8a
8b
17a
17b
17c
CH2
CH2
CH2
CH2
CH2
O
1
1
2
2
2
2
2
3
3
3
3
1
1
2
2
3
3
1
1
1
H
CdC
H
CdC
H
H
CdC
H
CdC
H
H
H
CdC
H
CdC
H
CdC
H
CdC
H
O
CH2
CH2
CH2
O
-
-
-
-
CdC
N3
CdC
N3
CdC
OH
of nucleotides. Although we were mostly interested in
the preparation of acyl NDPs or NTPs, we first studied
the acylation of 5′-monophosphate nucleoside analogs and
then tried to apply it to nucleosides 5′-di- or 5′-triphos-
phates. Some acyl AMPs have already been prepared in
the early 1960s, using DCC11 a,b or ethyl chloroformate11c
activation of the carboxylic acids moiety, but their
characterization mainly relied on chemical, biochemical,
and electrophoretical properties, and no NMR data are
available for such compounds. We chose to test this
reaction on thymine-containing nucleoside analogs such
as zidovudine (14a ) (AZT) and stavudine (16b) (d4T), in
order to avoid problem that could arise from base
-
-
-
-
-
DCC was used to activate acyl pyrophosphates, generat-
ing an electrophilic phosphorus. Formation of the awaited
acyl NDP bond resulted from a nucleophilic attack, at
the phosphatidylurea phosphorus, by the nucleoside 5′-
hydroxyl (Scheme 3). In the preparation of acyl NTPs,
acyl pyrophosphates were used as nucleophiles in order
to displace morpholine from various phosphoromorpholi-
dates9 electrophilic phosphorus (Scheme 4).
acylation.
Nucleoside 5′-monophosphate analogs,
AZTMP (6a ) and d4TMP (6b), were prepared according
to Tener procedure,12 their spectroscopic data being
consistent with those published in the literature.13 Nu-
cleotide analogs 5′-monophosphates 6a ,b were used as
tetrabutylammonium salts and acylated with myristic
acid (2 equiv) in methylene chloride, using DCC (2 eq)
as activator (Scheme 3). In order to obtain reasonable
reaction rates catalytic 4-(dimethylamino)pyridine (DMAP)
was added to the reaction mixture. Longer reaction time,
compared with the acylation of inorganic pyrophosphate,7
are needed for these reactions to go to completion, this
being mainly explained by lower reagent concentrations.
Although these reactions were nearly quantitative by
TLC analysis, isolated yields are relatively low: Myr-
AZTMP (1a ) and Myr-d4TMP (1b) were obtained, re-
spectively, in 26 and 13% isolated yields.14 It should be
noted that, whatever quenching was used, great extent
We chose to study the effect of two different lipophilic
fatty acyl moieties: myristic acid (9) was used as a model,
while 13-oxamyristic acid (10) (13-OMA) was used not
only for its lipophilicity, but also for its ability to inhibit
HIV replication by perturbing N-myristoylation of GAG
polyprotein precursor, thus preventing virus assembly.10
Hydrolyses kinetics and in vitro antiretroviral activities
of acyl nucleotides 1-5 are also examined in this paper.
Resu lts a n d Discu ssion
Acyla tion of Nu cleosid es 5′-Mon o-, Di- a n d Tr i-
p h osp h a te 6-8. We have recently described a new
protocol for the acylation of inorganic pyrophosphate 11
which allows the preparation of acyl pyrophosphates 12
and 13 with good yields and purity (Scheme 5).7 In order
to prepare acyl nucleotides 1-5, we investigated the
possibility to extend this type of reaction to the acylation
(11) (a) Talbert, P. T.; Huennekens, F. M. J . Am. Chem. Soc. 1956,
78, 4671. (b) Berg, P. J . Biol. Chem. 1958, 233, 608. (c) Michelson, A.
M.; Letters, R. Biochim. Biophys. Acta 1964, 80, 242.
(9) (a) Moffat, J . G.; Khorana, H. G. J . Am. Chem. Soc. 1961, 83,
649. (b) Moffat, J . G. Can. J . Chem. 1964, 42, 599. (c) Hostetler, K. Y.;
Stuhmiller, L. M.; Lenting, H. B. M.; van der Bosch, H.; Richman, D.
D. J . Biol. Chem. 1990, 265, 6112.
(10) Bryant, M. L.; Heuckeroth, R. O.; Kimata, J . T.; Ratner, L.;
Gordon, J . I. Proc. Natl. Acad. Sci. U.S.A. 1989, 86, 8655. Devadas,
B.; Kish, re N. S.; Adams, S. P.; Gordon, J . I. Bioorg. Med. Chem. Lett.
1993, 3, 779.
(12) Tener, G. M. J . Am. Chem. Soc. 1961, 83, 159.
(13) (a) Ma, Q.; Bathurst, I. C.; Barr, P. J .; Kenyon, G. L. J . Med.
Chem. 1992, 35, 1938. (b) Mansuri, M. M.; Starret, J . E., J r.; Ghazzouli,
I.; Hitchcock, M. J . M.; Sterzycky, R. Z.; Brankovan, V.; Lin, T.; August,
E. M.; Prusoff, W. H.; Sommadossi, J . P.; Martin, J . C. J . Med. Chem.
1989, 32, 461.

Potential Lipophilic Nucleotide Prodrugs
J . Org. Chem., Vol. 61, No. 3, 1996 897
Sch em e 3
phosphorus is generally 8-10 ppm more shielded than
a free pyrophosphate phosphorus. 31P NMR signals of
acyl NMP 1a ,b are in accordance with these findings:
phosphorus in 1a ,b resonate 8 ppm higher field than the
parent NMPs 6a ,b signals.15 The unambiguous proof for
the presence of an acyl-phosphate bond in 1a ,b is given
by the occurrence of a 9.0 Hz coupling constant between
the acyl carbonyl carbon and nucleotide phosphorus.
P r ep a r a tion of Acyl NDP s 2 a n d 3. It has been
reported that NDPs may be prepared by a SN2 displace-
ment of a 5′-iodo or 5′-tosyl nucleoside by a pyrophos-
phate salt,16 but, when using acyl pyrophosphate as
nucleophile, this protocol failed leading only to degrada-
tion products. Carbodiimide activation of cyanoethyl
phosphate is a well known method to prepare NMPs from
the corresponding nucleoside:12 a phosphatidylurea in-
termediate is generated in situ by the reaction of DCC
with cyanoethylphosphate, while the nucleoside 5′-hy-
droxyl group acts as a nucleophile which attacks the
electrophilic phosphorus. This procedure is not used to
prepare NDPs, since symmetrical dinucleotides are
awaited. With acyl pyrophosphates, we thought that the
acyl moiety could be considered as a protective group for
one of the pyrophosphate phosphorus and that it should
be possible to prepare acyl NDPs directly from a nucleo-
side and an acyl pyrophosphate using DCC activation.
This reaction was tested on 3′-acetyl thymidine 14c12 and
then applied to different HIV inhibitors: 3′-azido-3′-
deoxythymidine (AZT) (14a ) and 3′-deoxy-2′,3′-dide-
hydrothymidine (d4T) (14b) (Scheme 3). In preliminary
experiments, we established that the reaction rate was
much higher when 1 equiv of (()-camphorsulfonic acid
was added to the reaction mixture. When a proton source
was omitted, for example when tris(tetrabutylammo-
nium) acyl pyrophosphate was used alone, dicyclohexyl-
urea anion accumulated and prevented protonation of the
phosphatidylurea intermediate, thus lowering the reac-
tion rate. In order to avoid the addition of a strong acid
we found it more convenient to exchange the tetrabutyl-
ammonium coutercations of 12 and 13 and to use tris-
(tributylammonium) acyl pyrophosphates 15 and 16 as
phosphorylating agents. In a typical experiment the free
nucleosides were dissolved in anhydrous THF, 15 and
16 (4 equiv) and DCC (4 equiv) were added over 72 h.
The new compounds acyl NDPs 2a -c and 3a ,b were
obtained in moderate 16-24% isolated yields.14 Acyl-
ation of the 5′ free hydroxyl of the nucleosides was
observed as a side reaction; this was not surprising since
acyl pyrophosphates are mixed carboxylic phosphoric
anhydrides. Depending on the nucleoside, 20-40% of the
starting material could be recovered after Zemplen
deacylation.
Sch em e 4
Sch em e 5
of degradation occurred at this step. When trying to
extend this reaction to NDPs 7 or NTPs 8, degradation
on quenching occurred to such an extent that only traces
of acyl NDPs or NTPs could be recovered after purifica-
tion.
These new acyl NMP 1a ,b were fully characterized,
as sodium or triethylammonium salts, by FAB+ mass
spectra, elemental analysis,14 and H, 31P, and 13C NMR.
1
The characteristic NMR data of these new compounds
are in accordance with the proposed structures. For acyl
AZTDPs 2a ,3a , H5-H5′ signals are deshielded from 0.3
ppm in respect to AZT; this deshielding effect of phos-
phorylation is only 0.07 ppm for acyl d4TDPs 2b,3b. The
presence of an acyl phosphate bond is inferred from a
We will focus here on some characteristic NMR features
of acyl NMP. H5-H5′ signals are deshielded approxi-
mately 0.1-0.3 ppm, in respect to the parent nucleoside,
this being in agreement with a 5′-phosphorylated nucleo-
tide. In the course of our studies on pyrophosphate
acylation,7 we have already observed that an acylated
(15) The phosphorus chemical shift greatly depends on the solvent
and the pH; in order to compare 31P chemical shifts, acylated and
unacylated nucleotide were mixed and 31P NMR spectra of this mixture
were recorded in D2O.
(16) Dixit, V. M.; Poulter, C. D. Tetrahedron Lett. 1984, 25, 4055.
(17) The preparation of this compound was not optimized, nor was
it fully characterized. However, 1H, and 31P NMR and FAB+ mass data
are given in the experimental part, since 4c trilithium salt is the only
acyl NTP we obtained showing a PR-H5-5′ coupling constant, thus
allowing the unambiguous assignment of the PR 31P chemical shift.
(14) The purity of acyl nucleotides 1-5 was checked by reverse phase
HPLC and typically ranged between 98-100% for acyl nucleotide in
the triethylammonium form, and 90-95% for products in the sodium
form, after exchange on Dowex AG50 WX8. Recently we found that
precipitation of the acyl nucleotide triethylammonium salt, dissolved
in the minimum methanol at 0 °C, by addition of a 1 M NaI acetone
solution followed by extensive washing with cold acetone, allowed the
isolation of the acyl nucleotide sodium salt without detectable decom-
position. However, these products slowly decomposed even at -18 °C.

898 J . Org. Chem., Vol. 61, No. 3, 1996
Bonnaffe´ et al.
Ta ble 2. An tivir a l Activity on HIV1 of th e Acyl
9.0-9.5 Hz coupling constant between the acyl carbonyl
carbon and PR. By comparison of the 31P chemical shifts
of acyl NDPs with free NDPs15 we concluded that the acyl
chain is linked to Pâ: this phosphorus resonates around
-19 ppm, this being 9-10 ppm higher field than an
unacylated NDP’s Pâ; moreover, this phosphorus presents
no coupling constant with H5-H5′. The other phosphorus,
PR, is coupled with H5-H5′ (J P-H ≈ 5 Hz) and resonates
around -11.5 ppm, only 0.5 ppm higher field than a
nonacylated NDP’s PR. These results are in accord-
ance with our previous studies on acyl pyrophosphates:
the acylated phosphorus is generally 8-10 ppm more
shielded than the nonacylated one, which resonates only
0.2-0.3 ppm higher field than free pyrophosphate phos-
phorus.7
P h osp h a te Ser ies of AZT
MTT ED50 (nM) RT ED50 (nM)
SI
CD50
(µM)
6
10
6
10
6
10
days
days
days
days days days
1a
2a
3a
4a
5a
6a
7a
8a
>20
>50
>50
>50
>50
>20
>20
>20
10
30
-
24
100
20
20
10
20
150
100
467
100
647
60
170
100
100
<9
9
8
<9
8
<9
10
<9
8
700 2000 133
700 1667 500
400
500 2083 500
400 500 77
- 107
500 1000 333
1000 1000 118
900 2000 200
300 2500 500
14a (AZT) >50
Ta ble 3. An tivir a l Activity on HIV1 of th e Acyl
P h osp h a te Ser ies of D4T
P r ep a r a tion of Acyl NTP s 4 a n d 5. In order to
prepare acyl NTPs, we turned to classical NTP chemis-
try: the nucleophilic displacement of a phosphoromor-
pholidate by a pyrophosphate salt.9a,b Once more, acyl
pyrophosphates proved to be valuable intermediates in
the synthesis of the previously unavailable acyl NTPs 4
and 5 (Scheme 4). In this reaction, the acyl pyrophos-
phate was used as nucleophile to displace morpholine
from the electrophilic phosphoromorpholidate phospho-
rus. The acyl pyrophosphates were used in the tri-
butylammonium form, and (()-camphorsulfonic acid (1
equiv) was added to catalyze the reaction and protonate
the liberated morpholine, thus avoiding aminolysis of the
labile acyl phosphate bond. It should be noted that when
the proton source is omitted, only nucleoside triphosphate
could be recovered at the end of the reaction. Thymidine
phosphoromorpholidate9a,b 17c was used as a model and
the reaction was optimized with AZT and d4T phos-
phoromorpholidates 17a 9c and 17b, leading to the acyl
NTPs 4a -c and 5a in 7-20% isolated yields.14
NMR data collected for the acyl NTPs are in agreement
with the proposed structure and show that Pγ, and not
PR nor Pâ, was acylated. H5-H5′ signals for AZT deriva-
tives are deshielded 0.3 ppm in respect to AZT, this effect
being much lower for acyl d4TTP 4b. The occurrence of
an acyl phosphate bond is inferred from a 9.5 Hz J C-P
coupling constant determined on the carbonyl carbon
signal, while Pγ resonates 9-11 ppm higher field than a
nonacylated NTP Pγ.15 The chemical shift of PR, coupled
with H5-H5′ (J P-H ) 5.9 Hz for 4c in DMSO-d6), is only
slightly affected by acylation (-0.06 to -0.37 ppm).15 The
characteristic triphosphate Pâ signal (triplet J P-P ≈ 18
Hz) appears around -23 ppm, only 0.5-1.5 ppm higher
field than a nonacylated NTP Pâ.15 It should be noted
that line broadening was observed when 4a ,b sodium salt
spectra were recorded in DMSO-d6, and this effect
increased with concentration and was attributed to
cluster formation. This effect also occurred with acyl
NDPs disodium salts, although not to such an extent.
Line broadening did not occurred when 4c lithium salt
was recorded in DMSO-d6, or when D2O was used as
solvent.
MTT ED50 (nM) RT ED50 (nM)
SI
CD50
(µM)
6
10
6
10
6
10
days
days
days
days days days
1b
2b
4b
6b
7b
8b
>20
>50
>20
>20
>20
>20
190
190
220
280
160
220
9
540
600
530
710
600
760
400
100
100
100
100
100
100
10
1000
1500
1000
1000
1000
1000
105 37
263 83
91 38
71 28
125 33
91 26
14b (D4T) >10
600 1111 25
supernatant (RT ED50).18 We present in full details in
this paper data obtained at days 6 and 10, postinfection
by HIV1 (Tables 2 and 3).19
The first parameter we examined for myristoyl nucleo-
tides in the AZT series (14a , 1a , 2a , 4a , 6a , 7a , 8a ) and
the d4T series (14b, 1b, 2b, 4b, 6b, 7b, 8b) was the 50%
cytotoxic dose (CD50 ) concentration required to reduce
the viability of uninfected cells by 50%). Acyl and free
nucleotides were devoid of toxicity for CEM4 cells at
concentration up to 50 µM, alike AZT (>50 µM) and d4T
(> 10 µM). ED50 at day 6 postinfection for nucleotide
compounds in the AZT series (Table 2) laid in the same
order range as the free nucleoside (10-30 nM), whereas
in the d4T series (Table 3), nucleotides and acyl nucleo-
tides are 1 log less active than the free nucleoside. When
comparing activities of acyl nucleotides with their parent
nucleotide, the differences observed are in the error range
of the activity determination. We concluded that no
antiretroviral activity differences could be detected in
these assays between acyl nucleotides and their corre-
sponding nucleotides either 6 or 10 days after infection.
These results were confirmed by the assays on the
inhibition of HIV1 RT production and clearly illustrated
by the selectivity index: no significant differences were
detected between acyl nucleotides and their correspond-
ing nucleotides.
In a second series of experiments, antiretroviral activ-
ity of 13-OMA-AZTDP (3a ) and 13-OMA-AZTTP (5a ) was
compared to the activity of Myr-AZTDP (2a ) and Myr-
AZTTP (4a ). The grafting of the anti-HIV fatty acid
analog on the nucleotides did not improve their anti-
retroviral activity.
The same results are obtained on HIV2 (data not
shown). In conclusion, myristoyl or free nucleotides are
as cytotoxic as the corresponding nucleoside in this
standard test. Other experiments are in progress on
An tir etr ovir a l Activities of Acyl Nu cleotid es 1-5.
The antiretroviral activities were measured on CEM 4 T
cell line infected with HIV1 LAI or HIV2 Rod cell free
supernatants. Two parameters were studied, at days 6
and 10 postinfection, to evaluate the antiretroviral activ-
ity of the compounds: inhibition of HIV induced cyto-
pathic effect (ED50), using the MTT cell viability assay
and the inhibition of the HIV RT production in culture
(18) Henin, Y.; Gouyette, C.; Schwartz, O.; Debouzy, J . C.; Neumann,
J . M.; Huynh Dinh, T. J . Med. Chem. 1991, 34, 1830.
(19) It should be noted that the data obtained with cells infected by
HIV2 confirm the data presented here and are not described in this
paper for clarity purpose.

Potential Lipophilic Nucleotide Prodrugs
J . Org. Chem., Vol. 61, No. 3, 1996 899
F igu r e 2. Degradation kinetics of Myr-AZTDP (2a ) (b), and
apparition of AZTDP (7a ) ([), AZTMP (6a ) (O), and AZT (14a )
(2) in TEAA buffer 10 mM, pH 7.0, at 37 °C.
F igu r e 1. Degradation kinetics of Myr-AZTMP (1a ) (b), and
apparition of AZTMP (6a ) (O), and AZT (14a) (2) in TEAA
buffer 10 mM, pH 7.0, at 37 °C.
peripheral blood lymphocytes with measurements after
longer periods of infection.
Hyd r olysis of Acyl Nu cleotid es 1-5. The basic
hypothesis on which lies this work is the preferential
hydrolysis of the acylphosphate bond over the polyphos-
phate bond in acyl nucleotides8 (Scheme 1). In order to
test this hypothesis we followed the hydrolysis kinetics
of acyl nucleotides 1-5 at 37 °C in a 10 mM triethyl-
ammonium acetate (TEAA) buffer at physiological pH.
The disappearance of the acyl nucleotide and the appari-
tion of the nucleoside mono-, di-, and triphosphates of
AZT (14a ) or d4T (14b) were monitored by HPLC, using
authentic nucleotide samples for calibration.13,20 As
predicted from the hydrolysis free enthalpy of acetyl
phosphate, and ATP (into ADP and phosphate),8 the
results corroborated our hypothesis: acyl nucleotides are
cleanly hydrolyzed into their corresponding nucleotides.
Hydrolysis of Myr-AZTMP (1a ) gave the corresponding
nucleotide AZTMP (6a ) with a half-life of 51 h (Figure
1), the latter being further degraded into AZT. The same
experiment performed on Myr-AZTDP (2a ) showed that
this product was exclusively hydrolyzed into AZTDP (7a ),
with a half-life of 100 h (Figure 2), this nucleotide being
then further degraded into AZTMP and finally AZT.
Similarly (Figure 3), hydrolysis of Myr-AZTTP (4a )
liberated only AZTTP (8a ) (half-life 150 h), which is then
further hydrolyzed into AZTDP, AZTMP, and finally AZT.
Similar results were also obtained in the d4T series 1b,
2b, 4b, and with the 13-OMA AZT acyl nucleotides 3a
and 5a .21
F igu r e 3. Degradation kinetics of Myr-AZTTP (4a ) (b), and
apparition of AZTTP (8a ) (0), AZTDP (7a ) ([), AZTMP (6a )
(O) and AZT (14a ) (2) in TEAA buffer 10 mM, pH 7.0, at 37
°C.
of acyl nucleotides into the cells during the 6-10 days
cell culture. Unfortunately, the antiviral activity did not
reveal any differences between acyl nucleotides 1-5 and
their corresponding free nucleotides 6-8. As already
mentioned, acyl nucleotides are mixed anhydrides, and
some of us have shown that acyl pyrophosphates may be
used as water soluble fatty acyl donors for the acylation
of lysine ꢀ-amino groups in proteins.22 These experiments
suggested that acyl nucleotides may be prone to rapid
aminolysis in RPMI culture media containing aminated
compounds and proteins. We undertook comparative
hydrolysis kinetics on acyl nucleotides 2a and 4a , in
On the basis of these hydrolysis experiments, we were
rather confident in the ability of acyl nucleotides to
enhance nucleotide transmembrane diffusion, since half-
lives of 2 to 7 days should be sufficient to allow diffusion
(20) (a) Kedar, P. S.; Abbots, J .; Kovacs T.; Lesiak, K.; Torrence, P.;
Wilson, S. H. Biochemistry 1990, 29, 3603. (b) Sergheraert, C.; Pierlot,
C.; Tarton, A.; He´nin, Y.; Lemaˆıtre, M. J . Med. Chem. 1993, 36, 826.
(21) Bonnaffe´, D.; Dupraz, B.; Ughetto-Monfrin, J .; Namane, A.;
Huynh Dinh, T. Proceedings of the 11th Round Table on Nucleosides,
Nucleotides and their Biological Applications; Leuven, Belgium, 1994.
Nucleosides Nucleotides 1995, 14, 783.
(22) Heveker, N.; Bonnaffe´, D.; Ulmann, A. J . Biol. Chem, 1994,
32844.

900 J . Org. Chem., Vol. 61, No. 3, 1996
Bonnaffe´ et al.
on silica gel 60 F254, eluting with acetonitrile/200 mM am-
monium bicarbonate/AcOEt (3/1/1) ternary mixture. Detection
was performed using UV light, anisaldehyde-sulfuric acid for
the nucleoside moiety, and a modified Ditmer and Lester
reagent for phosphate revelation.23 Flash column chromatog-
raphies were performed at 4 °C on reverse phase C18 silica
gel, using a 0-70% water-acetonitrile gradient. Preparative
HPLC were performed using a C18 reverse phase radial pack
cartridge and a 10-60% 10 mM TEAA pH 7.0 acetonitrile
gradient. NMR spectra were recorded at 300 MHz for 1H, 75
MHz for 13C, and 121.5 MHz for 31P. 1H and 13C chemical shifts
are given in ppm (δ) relative to TMS as internal standard in
CDCl3 and external standard in D2O. 31P chemical shifts are
given in ppm (δ) relative to 85% H3PO4 as external standard.
Quint and sext are the abbreviations, respectively, used for
quintuplet and sextuplet. DEPT 135 sequence was used for
the attribution of 13C signals, and 2D 1H-13C heteronuclear
COSY for the attribution of d4T 1H and 13C signals. FAB mass
spectra were recorded using a glycerol or glycerol/thioglycerol
(50/50) matrix. 3′-Azido-3′-deoxy-5′-phosphoromorpholidate-
thymidine 17a and 3′-deoxy-2′,3′-didehydro-5′-phosphoromor-
pholidate-thymidine 17b were prepared using Moffat
procedure9a-c and used without further purification. Hydrolysis
kinetics were performed in duplicate at a 1 mg/mL acyl
nucleotide concentration in 10 mM TEAA, pH 7.0, or RPMI
culture medium (RPMI 1640, Life Technologies, with 10%
foetal calf serum and 1% glutamine). HPLC was carried out
using a 5 µm C18 column (150 × 4.6 mm) and eluting with
TEAA 10 mM pH 7.0/acetonitrile gradients: 6 min isocratic
elution with 4.8% acetonitrile, followed by a linear gradient
from 4.8 to 95% over 14 min. The detection was performed at
265 nm using a diode-array detector.
Tetr a d eca n oyl 3′-a zid o-3′-d eoxy-5′-th ym id ylp h osp h a te
(1a ): 200 mg of AZTMP (6a )12,13a disodium salt (0.51 mmol)
was exchanged on Dowex AG 50WX8 200 (NBu4+) and lyo-
philized. The resulting powder was dissolved in 10 mL of
anhydrous methylene chloride and 232 mg of myristic acid
(1.02 mmol), 210 mg of DCC (1.02 mmol), and 5 mg of DMAP
were added. The mixture was kept for 72 h at room temper-
ature under magnetic stirring, time after which no more
starting material could be detected by TLC. The mixture was
diluted with 5 mL of toluene and extracted three times with
10 mL of H2O at 0 °C, using centrifugation to facilitate
decantation. The aqueous phases were pooled together and
evaporated under oil pump vacuum. Flash chromatography,
lyophilization, and exchange on Dowex AG 50 WX8 200 (Na+)14
gave 68 mg of 1a sodium salt (23%): 1H NMR for 1a
tetrabutylammonium salt (CDCl3) δ 7.91 (br s, 1H), 7.76 (br
s, 1 H), 6.15 (t, 1 H, J ) 8 Hz), 4.39-4.30 (m, 1 H), 4.07-3.98
(m, 2 H), 3.91-3.83 (m, 1 H), 3.25-3.03 (m, 8 H), 2.31-2.09
(m, 4 H), 1.81 (s, 3 H), 1.56-1.19 (m, 10 H), 1.25 (sext, 8 H, J
) 7.3 Hz), 1.06 (br s, 20 H), 0.81 (t, 12 H, J ) 7.3 Hz), 0.70 (t,
3 H, J ) 6.7 Hz); 13C NMR 1a sodium salt (D2O) δ 172.1 (d,
J C-P ) 9.1 Hz), 166.7, 151.7, 137.9, 112.5, 85.6, 84.0 (d, J C-P
F igu r e 4. (a) Hydrolysis at RT of Myr-AZTDP (2a ) in RPMI
culture medium (b) or TEAA buffer (O). Apparition of AZTDP
(7a ) by hydrolysis of Myr-AZTDP (2a ) in RPMI medium (2).
(b) Hydrolysis at RT of Myr-AZTTP (4a ) in RPMI culture
medium (b) or TEAA buffer (O). Apparition of AZTTP (8a ) by
hydrolysis of Myr-AZTTP (4a ) in RPMI medium (2).
RPMI and TEAA media at room temperature. As shown
by the apparition curves of AZTDP (Figure 4a) and
AZTTP (Figure 4b), acyl nucleotides are also selectively
cleaved in RPMI medium into their corresponding nucleo-
tide. However, their half lives are dramatically lower
than in TEAA (Figure 4a,b): 1.7 h versus 114 h for Myr-
AZTDP (2a ) and 1.4 h versus 156 h for Myr-AZTTP (4a ).
This rapid aminolysis could explain the lack of anti-
retroviral activity enhancement of nucleotide acylation:
the half-lives of acyl nucleotides in RPMI medium is too
low to allow transmembrane diffusion, and thus it is
impossible to detect any antiretroviral activity differences
between the nucleotide and the acylated prodrug.
In conclusion, we have described three general methods
allowing the preparation of acyl nucleosides mono-, di-,
and triphosphates. As awaited, the hydrolysis of such
compounds liberated the corresponding nucleotide by
selective cleavage of the acyl phosphate bond. Hydrolysis
experiments in TEAA showed that the half-lifes of acyl
nucleotides were compatible with their use in aqueous
media. However, preliminary in vitro anti HIV activity
data did not allowed us to ascertain the hypothesis of a
nucleotide transmembrane diffusion acceleration by the
lipophilic acyl chain, since our compounds are readily
aminolyzed in RPMI culture media.
) 8.9 Hz), 65.1 (d, J C-P ) 4.4 Hz), 61.5, 37.0, 36.0 (d, J C-P
)
6.1 Hz), 32.7, 30.6-29.9, 25.3, 23.4, 14.6, 12.4; 31P NMR
(CDCl3) δ -10.14 (br t, J P-H ) 4.3 Hz); MS FAB+ for
C24H39N5O8PNa m/ z 602.6 (M + Na)+. Anal. Calcd for
C24H39N5O8PNa‚H2O: C, 48.24; H, 6.92; N, 11.72. Found: C,
48.52; H, 6.56; N, 11.72.
Tet r a d eca n oyl
3′-d eoxy-2′,3′-d id eh yd r o-5′-t h ym id -
in ylp h osp h a te (1b): d4TMP (6b)12,13b disodium salt (178 mg,
0.51 mmol) was treated as described for the preparation of 1a ,
giving 70 mg of 1b sodium salt (26%): 1H NMR for 1b
tetrabutylammonium salt (CDCl3) δ 8.12 (br s, 1 H), 7.82 (br
s, 1 H), 7.00 (quint, 1 H, J ) 1.6 Hz), 6.39 (dt, 1 H, J ) 1.6,
6.1 Hz), 5.72 (dt, 1 H, J ) 1.6, 5.8 Hz), 4.95 (br s, 1 H), 4.32-
4.21 (m, 1 H), 4.21-4.09 (m, 1 H), 3.42-3.20 (m, 8 H), 2.29 (t,
2 H, J ) 7.5 Hz), 1.97 (d, 3 H, J ) 1.0 Hz), 1.72-1.49 (m, 10
H), 1.41 (sext, 8 H, J ) 7.5 Hz), 1.42 (br s, 20 H), 0.97 (t, 12
1
H, J ) 7.5), 0.86 (t, 3 H, J ) 6.7 Hz); H NMR for 1b sodium
Exp er im en ta l Section
salt (DMSO-d6) δ 11.28 (br s, 1 H), 7.73 (br s, 1 H), 6.94 (quint,
1 H, J ) 1.5 Hz), 6.36 (dt, 1 H, J ) 1.5, 6.1 Hz), 5.91 (dt, 1 H,
J ) 1.5, 5.8 Hz), 4.87 (br s, 1 H), 4.09-3.98 (m, 1 H), 3.93-
All moisture sensitive reactions were performed under
nitrogen using oven-dried glassware. Solvents were dried and
distilled prior to use: THF from sodium/benzophenone and
CH2Cl2 from phosphorus pentoxide. Reactions were monitored
(23) Ryu, E. K.; MacCoss, M. J . Lipid Res. 1979, 20, 561.

Potential Lipophilic Nucleotide Prodrugs
J . Org. Chem., Vol. 61, No. 3, 1996 901
3.69 (m, 1 H), 2.29 (t, 2 H, J ) 7.3 Hz), 1.82 (s, 3 H), 1.56-
1.39 (m, 2 H, CH2 â), 1.26 (br s, 20 H), 0.89 (t, 3 H, J ) 6.5
Hz); 13C NMR (D2O) δ 172.0 (d, J C-P ) 9.4 Hz), 166.8, 152.5,
138.7, 135.0, 126.6, 111.9, 90.5, 86.7 (d, J C-P ) 8.7 Hz), 67.5
(d, J C-P ) 4.0 Hz), 35.9 (d, J C-P ) 5.6 Hz), 32.8, 30.7-29.8,
Tetr a d eca n oyl 3′-O-a cetyl-5′-th ym id in yld ip h osp h a te
(2c): 3′-O-acetyl-5′-thymidine (14c) (284 mg, 1 mmol) was
treated with 15 and DCC as described for the preparation of
2a , giving 170 mg of 2c disodium salt (24%);24 1H NMR
(DMSO-d6) δ 11.29 (br s, 1 H), 7.87 (br s), 6.21 (dd, 1 H, J )
5.7, 9.3 Hz), 5.21 (d, 1 H, J ) 5.5 Hz), 4.11-4.03 (br s, 1H),
4.0,1-3.83 (m, 2H), 2.31 (ddd, 1 H, J ) 5.5, 9.3, 13.7 Hz), 2.28
(t, 2H, J ) 7.3 Hz), 2.16 (dd, 1H, J ) 5.7, 13.7 Hz), 2.05 (s, 3
H), 1.81 (s, 3 H), 1.52-1.36 (m, 2 H), 1.21 (br s, 20 H), 0.84 (t,
3H, J ) 7.6 Hz); 13C NMR (DMSO-d6) δ 170.8 (d, J C-P ) 9.0
Hz), 170.1, 164.0, 150.8, 136.4, 110.5, 83.8, 83.4 (d, J C-P ) 8.5
Hz), 75.9, 65.0 (d, J C-P ) 5.0 Hz), 36.4, 35.1 (d, J C-P ) 4.5
Hz), 31.5, 29.3-28.6, 24.4, 22.3, 21.0, 14.1 , 12.3; 31P NMR
(DMSO-d6) δ -8.11 (br dt, 1 P, J P-P ) 19.1 Hz, J C-P ) 7.5
Hz), -15.98 (d, 1 P, J P-P ) 19.1 Hz); MS FAB+ for C26H42N2-
25.3, 23.5, 14.7, 12.5; 31P NMR (CDCl3) δ -10.00 (t, J P-H
)
5.3 Hz); 31P NMR (DMSO-d6) δ -6.00 (t, J P-H ) 6.7 Hz). MS
FAB- for C24H38O8N2PNa m/ z 513 (M - Na)-.
Tetr a d eca n oyl (3′-a zid o-3′-d eoxy-5′-th ym id in yld ip h os-
p h a te (2a ): 2.5 mL of a 0.4 M methylene chloride stock
solution of tris(tributylammonium) myristoyl pyrophosphate
(15) (1 mmol) was added to 267 mg of AZT (14a ) (1 mmol).
The mixture was evaporated in vacuo and redissolved in 5 mL
of anhydrous THF, and then 1.03 g of DCC (5 mmol) was
added. The mixture was kept 24 h at rt. Myristoyl pyrophos-
phate 15 (0.4 M solution) (2.5 mL, 1 mmol) and 500 mg of DCC
(2.5 mmol) were mixed, evaporated, redissolved in 2.5 mL of
THF and added to the reaction mixture. This reagent addition
was repeated every 24 h until no more AZT could be detected
by TLC (generally 72 h reaction, 3 equiv 15 and 10 equiv of
DCC). In order to destroy excess DCC, the reaction mixture
was diluted with 50 mL of anhydrous THF and 900 mg of
oxalic acid (10 mmol), dissolved in 50 mL of THF, were added
dropwise over 4 h, checking regularly that the pH did not drop
under 3. THF was evaporated in vacuo and the residue
dissolved in 20 mL of CH2Cl2. The solution was kept overnight
at -10 °C and then filtered. The solvent was evaporated in
vacuo, and toluene and 100 mM carbonate buffer, pH 7.0 (10
mL each), were added at 0 °C. The mixture was decanted by
centrifugation and the organic phase24 extracted two times
with 5 mL of water. The aqueous phases were filtered and
directly purified using reverse phase flash chromatography
and preparative HPLC, as described in the general methods
section, and finally desalted on Sephadex G10 at 4 °C. The
fractions containing the product were lyophilized and condi-
tioned in the sodium form on Dowex AG50 WX8 200 (Na+),14
giving 153 mg 2a disodium salt (23%): 1H NMR (DMSO-d6) δ
11.12 (br s, 1 H), 7.87 (br d, 1 H, J ) 1.0 Hz), 6.16 (dd, 1 H, J
) 6.2, 7.9 Hz), 4.69-4.61 (m, 1 H), 4.01-3.95 (br s, 1H), 3.96-
3.80 (m, 2H), 2.50-2.35 (m, partly in DMSO signal, J ) 14.0
Hz), 2.31 (t, 2H, J ) 7.3 Hz), 2.22 (ddd, 1H, J ) 3.0, 6.2, 14.0
Hz), 1.82 (s, 3 H), 1.54-1.38 (m, 2 H), 1.24 (br s, 20 H), 0.86
O
13P2Na2 m/ z 722 (M + Na)+, 699.9 (M + H)+. Anal. Calcd
for C26H42N2O13P2Na2‚2H2O: C, 42.51; H, 6.31; N, 3.81; P, 8.43.
Found: C, 42.61; H, 5.93; N, 3.85; P, 7.71.
13-Oxa tetr a d eca n oyl 3′-a zid o-3′-d eoxy-5′-th ym id in yl-
d ip h osp h a te (3a ): AZT (14a ) (134 mg, 0.5 mmol), 13-
oxatetradecanoyl pyrophosphate (16) and DCC were mixed in
2.5 mL of THF as described for 2a , giving 68 mg of 3a disodium
salt (20%);24 1H NMR (D2O) δ 7.80 (s, 1 H), 6.29 (t, 1 H, J )
6.7 Hz), 4.63-4.50 (m, 1 H), 4.30-4.10 (m, 3 H), 3.48 (t, 2 H,
J ) 6.7 Hz), 3.34 (s, 3 H), 2.48 (t, 2 H, J ) 5.9 Hz), 2.42 (t, 2
H, J ) 7.4 Hz), 1.94 (s, 3 H), 1.65-1.50 (m, 4 H), 1.35-1.15
(m, 14 H); 13C NMR (D2O) δ 172.8 (d, J C-P ) 9.4 Hz), 167.0,
152.1, 137.9, 112.4, 85.4, 83.6 (d, J C-P ) 9.2 Hz), 73.4, 66.2 (d,
J C-P ) 5.6 Hz), 61.5, 58.2, 37.1, 35.5 (d, J C-P ) 6.2 Hz), 33.5,
29.2-28.7, 25.8, 24.5, 12.3; 31P NMR (D2O) δ -11.48 (br d,
1P, J P-P ) 21.3 Hz), -19.21 (d, 1P, J P-P ) 21.3 Hz); MS FAB+
for C23H37N5O12P2Na2 m/ z 706 (M + Na)+, 684 (M + H)+.
13-Oxa tetr a d eca n oyl 3′-d eoxy-2′,3′-d id eh yd r o-5′-th y-
m id in yld ip h osp h a te (3b): d4T (14b) (112 mg, 0.5 mmol)
was treated with 16 and DCC as described for 3a , giving 60
mg of 3b disodium salt (15%);24 1H NMR for 3b bis(triethyl-
ammonium) salt (D2O) δ 7.69 (d, 1 H, J ) 1 Hz), 6.94 (q, 1 H,
J ) 1.6 Hz), 6.51 (dt, 1 H, J ) 1.6, 6.1 Hz), 5.93 (dt, 1 H, J )
1.6, 5.8 Hz), 5.09 (br s, 1 H), 4.20-4.07 (m, 2 H), 3.47 (t, 2 H,
J ) 6.7 Hz), 3.33 (s, 3 H), 3.19 (q, 18 H, J ) 7.5 Hz), 2.40 (t,
2 H, J ) 7.4 Hz), 1.90 (d, 3 H, J ) 1.0 Hz), 1.22-1.52 (m, 4
H), 1.27 (t, 27 H, J ) 7.5 Hz), 1.23 (br s, 14 H); 13C NMR (D2O)
δ 172.5 (d, J ) 9.6 Hz), 167.3, 152.7, 139.0, 134.9, 126.1, 112.2,
90.5, 86.6 (d, J C-P ) 8.4 Hz), 73.5, 67.2 (d, J C-P ) 6.1 Hz),
58.6, 47.3, 35.6 (d, J C-P ) 6.2 Hz), 29.6-28.8, 25.9, 24.6, 12.6,
8.9; 31P NMR (D2O) δ -11.41 (br d, 1P, J ) 23.0 Hz), -19.33
(t, 3H, J ) 6.6 Hz); 13C NMR (DMSO-d6) δ 170.7 (d, J C-P
)
9.0 Hz), 164.0, 150.8, 136.4, 110.3, 83.7, 83.2 (d, J C-P ) 8.5
Hz), 64.7 (d, J C-P ) 7.5 Hz), 61.3, 35.9, 34.9, 33.6, 29.0-28.4,
24.4, 24.1, 22.0, 13.9, 12.1; 31P NMR (DMSO-d6) δ -8.38 (dt,
+
(d, 1P, J ) 23.0 Hz); MS FAB+ for C23H36N2O12P2 2HNEt3
1 P, J P-P ) 20.6 Hz, J P-H ) 6.0 Hz), -16.10 (d, 1 P, J P-P
)
20.6 Hz); 31P NMR (D2O) δ -11.02 (br s 50 Hz, 1 P), -18.88
(br s, 50 Hz, 1 P); MS FAB+ for C24H39N5O11P2Na2 m/ z 704
(M + Na)+. Anal. Calcd for C24H39N5O11P2Na2‚H2O: C, 41.21;
H, 5.91; N, 10.01; P, 8.86. Found: C, 40.92; H, 5.11; N, 9.50;
P, 8.13.
m/ z 799.9 (M + H)+.
Tetr a d eca n oyl 3′-a zid o-3′-d eoxy-5′-th ym id in yltr ip h os-
p h a te (4a ): 212 mg of AZT phosphoromorpholidate (17a )9c
(0.3 mmol) and 1.5 mL of 0.4 M methylene chloride solution
of tris(tributylammonium) myristoyl pyrophosphate (15) (0.6
mmol) were evaporated and redissolved in 5 mL of anhydrous
THF; 209 mg of camphorsulfonic acid was then added (0.9
mmol, 1.5 equiv in respect to pyrophosphate 15). The mixture
was kept 4 h at rt under magnetic stirring. The solution was
then cooled to 0 °C, diluted with 10 mL of toluene, and
extracted with three times 5 mL of TEAA, 1 M, pH 7.0. The
aqueous phases were pooled together, lyophilized, and purified
as described for acyl NDPs, giving 48 mg of 4a trisodium salt
(20%);13 1H NMR (DMSO-d6) δ 11.28 (br s, 1 H), 7.85 (br s, 1
H), 6.20 (t, 1 H, J ) 7.1 Hz), 4.72-4.61 (m, 1 H), 4.12-3.92
(br s, 3H), 2.38 (t, 2H, J ) 7.0 Hz), 2.35-2.30 (m, 1H), 2.30-
2.17 (m, 1 H), 1.87 (s, 3 H), 1.60-1.45 (m, 2 H), 1.28 (br s, 20
Tetr a d eca n oyl 3′-d eoxy-2′,3′-d id eh yd r o-5′-th ym id in yl-
d ip h osp h a te (2b): d4T (14b) (224 mg, 1 mmol) was treated
with 15 and DCC as described for the preparation of 2a , giving
102 mg of 2b disodium salt (16%);24 1H NMR (DMSO-d6) δ
11.13 (br s, 1H), 7.57 (d, 1 H, J ) 1.0 Hz), 6.70 (m, 1 H, J )
1.6 Hz), 6.31 (br d, 1 H, J ) 5.9 Hz), 5.74 (br d, 1 H, J ) 5.6
Hz), 4.74 (br s, 1H), 3.94-3.82 (m, 1 H), 3.77-3.65 (m, 1 H),
2.18 (t, 2 H, J ) 7.4 Hz), 1.69 (br s, 3 H), 1.42-1.28 (m, 2 H),
1.25 (br s, 20 H), 0.73 (t, 3 H, J ) 6.6 Hz); 13C NMR (DMSO-
d6) δ 170.7 (d, J C-P ) 9.0 Hz), 164.0, 150.8, 136.8, 134.8, 125.9,
109.8, 88.7, 85.7 (d, J C-P ) 9.0 Hz), 65.4 (d, J C-P ) 4.5 Hz),
35.0 (d, J C-P ) 4.5 Hz), 31.4, 29.1-28.5, 24.3, 22.2, 14.0, 12.0;
31P NMR (DMSO-d6) δ -8.31 (dt, 1 P, J P-P ) 20.7 Hz, J C-P
)
H), 0.90 (t, 3H, J ) 6.8 Hz); 13C NMR (D2O) δ 173.1 (d, J C-P
)
4.8 Hz), -16.10 (d, 1 P, J P-P ) 20.7 Hz); 31P NMR (D2O) δ
-11.99 (dt, 1 P, J P-P ) 20.5 Hz, J C-P ) 4.8 Hz), -19.02 (d, 1
P, J P-P ) 20.5 Hz); MS FAB+ for C24H38N2O11P2Na2 m/ z 662
(M + Na)+, 639 (M + H)+. Anal. Calcd for C24H38N2-
9.9 Hz), 167.3, 152.4, 138.2, 112.7, 85.9, 84.0 (d, J C-P ) 8.3
Hz), 66.7 (d, J C-P ) 7.4 Hz), 62.1, 37.4, 35.8, 32.3, 31.2, 29.8-
29.2, 24.8, 23.1, 14.4, 12.6; 31P NMR (DMSO-d6) δ -6 to -7
(m, 1P), -14.8 to -15.8 (m, 1P), -15.8 to -16.8 (m, 1P); 31P
NMR (D2O, proton decoupled) δ -11.03 (d, 1 P, J ) 18.6 Hz),
-19.04 (d, 1 P, J ) 18.6 Hz), -22.69 (t, 1P, J ) 18.6 Hz); MS
FAB+ for C24H39N5O14P3Na3 m/ z 806.6 (M + Na)+, 784.8 (M
+ H)+. Anal. Calcd for C24H39N5O14P3Na3‚2.5H2O: C, 34.79;
H, 5.35; N, 8.45; P, 11.22. Found: C, 34.71; H, 5.15; N, 7.49;
P, 10.83.
O
11P2Na2‚2H2O: C, 42.72; H, 6.28; N, 4.15; P, 9.19. Found:
C, 42.33; H, 5.89; N, 3.85; P, 10.31.
(24) The organic layer contains mainly 5′-acyl nucleoside. After
evaporation, Zemplen deacylation, and silica gel flash chromatography,
20-40% starting nucleoside could be recovered.

902 J . Org. Chem., Vol. 61, No. 3, 1996
Bonnaffe´ et al.
Tetr a d eca n oyl 3′-d eoxy-2′,3′-d id eh yd r o-5′-th ym id in yl-
tr ip h osp h a te (4b): d4T phosphoromorpholidate (17b) (179
mg, 0.3 mmol), acyl pyrophosphate 15, and camphorsulfonic
acid were mixed as described for 4a , giving 25 mg of 4b
trisodium salt (11%); 1H NMR (DMSO-d6) δ 11.13 (s, 1H), 7.59
(d, 1 H, J ) 1.0 Hz), 6.78 (s, 1 H), 6.46 (d, 1 H, J ) 5.5 Hz),
5.81 (br d, 1 H, J ) 5.5 Hz), 4.44 (br s, 1H), 4.05 (br s, 1 H),
3.88 (br s, 1 H), 2.29 (t, 2 H, J ) 6.8 Hz), 1.76 (br s, 3 H),
1.40-1.24 (m, 2 H), 1.19 (br s, 20 H), 0.81 (t, 3 H, J ) 6.8 Hz);
31P NMR (DMSO-d6) δ -6.10 (br s, 1 P), -15.03 (br s, 1 P),
-16.03 (br s, 1 P); MS FAB+ for C24H38N2O14P3Na3 m/ z 741
(M + H)+. Anal. Calcd for C24H38N2O14P3Na3‚3H2O: C, 36.26;
H, 5.58; N, 3.53. Found: C, 35.71; H, 5.05; N, 3.48.
J ) 7.3 Hz), 0.86 (t, 3 H, J ) 6.7 Hz); 13C NMR δ 170.4 (d,
J C-P ) 7.8 Hz), 58.4, 35.3, 31.7, 30.1-29.0, 24.5, 23.9, 22.5,
19.5, 14.0, 13.6; 31P NMR δ -10.53 (d, 1P, J P-H ) 18.5 Hz),
-18.04 (br d, 1P, J P-H ) 18.5 Hz); 31P NMR (D2O) δ -10.44
(br d, 1P, J P-H ) 15.8 Hz), -19.44 (br d, 1P, J P-H ) 15.8 Hz);
MS FAB+ for C14H28O8P2(NBu4)2 m/ e 871 (M + H)+. Anal.
Calcd for C14H27O8P2Na3: C, 37.02; H, 5.99; P, 13.64. Found:
C, 36.96; H, 6.03; P, 13.63.
Tr is(tetr a bu tyla m m on iu m ) 13-oxa tetr a d eca n oyl p yr o-
p h osp h a te (13): tris(tetrabutylammonium) pyrophosphate
(11) (4.78 g, 5 mmol) 13-oxamyristic acid10 (10) (4.60 g, 20
mmol) and DCC (4.12 g, 20 mmol) were treated and purified
as previously described,7 giving 4.12 g 13 (74%); 1H NMR
(CDCl3) δ 3.48-3.23 (m, 29 H25), 2.46 (t, 2 H, J ) 7.4 Hz),
1.76-1.57 (m, 26 H), 1.47 (sext, 24 H, J ) 7.3 Hz), 1.41-1.21
(br s, 18 H), 1.00 (t, 36 H, J ) 7.3 Hz); 13C NMR δ 170.2 (d,
J C-P ) 8.4 Hz), 73.1, 58.7, 58.6, 35.5 (d, J C-P ) 5.0 Hz), 29.7-
29.0, 26.2, 25.2, 24.6, 24.1, 20.2, 19.7, 13.8; 31P NMR δ -11.16
(d, 1P, J P-H ) 17.9 Hz), -19.13 (br d, 1P, J P-H ) 17.9 Hz).
Tr is(t r ib u t yla m m on iu m ) t et r a d eca n oyl p yr op h os-
p h a t e (15) a n d t r is(t r ib u t yla m m on iu m ) 13-oxa t et r a -
d eca n oyl p yr op h osp h a te (16): Dowex AG50 WX8 200 was
conditioned in the tributylammonium form as follows: tri-
butylamine was added to a suspension of Dowex AG50 WX8
200 (H+) under magnetic stirring in 95/5 MeOH/H2O until the
pH remained basic. The resin was filtered off and extensively
washed with methanol and then water; before use the resin
was washed with acetonitrile. Acetonitrile solutions of 12 and
13 were passed through columns of 10 mol equiv Dowex AG
50 WX8 200 (NBu3H+). After evaporation under vacuo, 0.4
M methylene chloride stock solution was prepared without
further purification. It should be noted that 12, 13, 15, and
16 methylene chloride solution are stable for months when
kept at -20 °C.
Tetr a d eca n oyl 5′-th ym id in yl tr ip h osp h a te (4c):17 thy-
midine phosphoromorpholidate (17c)9a,b (70 mg, 0.1 mmol) was
treated with 15 and camphorsulfonic acid as described for 4a ,
excepted for the final ion exchange which was performed on
Dowex AG50 WX8 200 lithium form, giving 5.4 mg of 4c
trilithium salt (8%): 1H NMR (DMSO-d6) δ 7.68 (d, 1 H, J )
0.8 Hz), 6.10 (t, 1 H, J ) 7.5 Hz), 5.73 (d, 1 H, J ) 4.2 Hz),
4.39-4.28 (m, 1 H), 4.03-3.84 (m, 2 H), 3.84-3.74 (m, 1 H),
2.28 (t, 1 H, J ) 7.5 Hz), 2.14-1.89 (m, 2 H), 1.75 (s, 3 H),
1.50-1.31 (m, 2 H), 1,31-0.97 (m, 20 H), 0.78 (t, 3 H, J ) 6.7
Hz); 31P NMR (DMSO-d6) δ -8.06 (dt, 1 P, J C-P ) 5.9 Hz, J P-P
) 18.4 Hz), -16.31 (d, 1 P, J P-P ) 18.4 Hz), -17.72 (t, 1 P
J P-P ) 18.4 Hz); MS FAB+ for C24H41N2O15P3Na2 (4c disodium
salt) m/ z 737 (M + H)+.
13-Oxa tetr a d eca n oyl 3′-a zid o-3′-d eoxy-5′-th ym id in yl-
tr ip h osp h a te (5a ): AZT phosphoromorpholidate (17a ) (354
mg, 0.5 mmol) was treated with 13-oxamyristoyl pyrophos-
phate (16) and camphorsulfonic acid as described for 4a , giving
25 mg of 5a triethylammonium salt (5%): 1H NMR (D2O) δ
7.80 (d, 1 H, J ) 1.0 Hz), 6.28 (t, 1 H, J ) 7.2 Hz), 4.60 (m, 1
H), 4.21 (m, 3 H), 3.47 (t, 2 H, J ) 7.4 Hz), 3.33 (s, 3 H), 3.19
(q, 18 H, J ) 7.3 Hz), 2.47 (m, 2 H), 2.44 (t, 2 H, J ) 7.9 Hz),
1.93 (br s, 3 H), 1.58 (m, 4 H), 1.26 (t, 41 H, J ) 7.3 Hz); 13C
NMR (D2O) δ 173.0 (d, J C-P ) 9.4 Hz), 167.2, 152.4, 138.1,
112.7, 85.6, 83.8 (d, J C-P ) 9.4 Hz), 73.6, 66.5 (d, J C-P ) 5.0
Hz), 61.9, 58.3, 47.4, 37.1, 35.6 (d, J ) 6.1 Hz), 29.4-28.9, 25.9,
24.6, 12.5, 9.0; 31P NMR (D2O) δ -11.30 (br d, 1 P, J ) 20.5
Hz) -19.30 (d, 1 P, J ) 19.3 Hz), -23.21 (br t, 1 P); MS FAB+
for C23H38N5O15P3(HNEt3)2 (5a diethylammonium salt) m/ z
922.9 (M + H)+.
An tivir a l Activity. The assays were performed as de-
scribed in ref 18.
Ack n ow led gm en t. We thank ANRS (Agence Na-
tionale pour la Recherche sur le SIDA) for a postdoctoral
fellowship (D.B.), V. Huteau for the synthesis of d4T,
and S. R. Sarfati for fruitful discussions.
Su p p or tin g In for m a tion Ava ila ble: 1H and 31P NMR
spectra of 1a , 1b, 2a , 2b, 2c, 3a , 3b, 4a , 4b, 4c, 5a , 12, and
13 (26 pages). This material is contained in libraries on
microfiche, immediately follows this article in the microfilm
version of the journal, and can be ordered from the ACS; see
any current masthead page for ordering information.
Tr is(tetr a bu tyla m m on iu m ) tetr a d eca n oyl p yr op h os-
p h a te (12): tris(tetrabutylammonium) pyrophosphate (11) (18
g, 20 mmol), myristic acid (9) (18.3 g, 80 mmol), and DCC (16.5
g, 80 mmol) were treated and purified as previously described,7
giving 15.4-17.6 g of 12 (63-70%); 1H NMR (CDCl3) δ 3.45-
3.23 (m, 19 H)25 2.53 (t, 2 H, J ) 7.5 Hz), 1.64-1.53 (m, 21 H),
1.45 (sext, 19 H, J ) 7.3 Hz), 1.22 (br s, 20 H), 0.97 (t, 29 H,
J O951354P
(25) The tetrabutylammonium countercation content of acyl pyro-
phosphates 12 and 13 may vary from two to three depending on the
sample. Yields for each preparation were calculated from 1H and 31P
NMR spectra.